INDUSTRIAL AND ENGINEERING CHEMISTRY
May 1954
The fact that Pease and Munro (Sf)did not detect propionaldehyde as a product of propane oxidation is accounted for by the postulated reactions of the n-propyl radical.
+
T L - C ~ H ~ '02 + TL-C~H~OO' n-CaH;OO'
+
[C€LCH,CHO]
+ OH
LCaH6 + CO
(9) (10)
(11)
The infrared analysis of the paraffin gas from run 71 disclosed that it contained approximate11 1% of ethane, an amount sufficient to account for the extent of Reactions 10 and 11. An alternative scheme for the disappearance of the propyl peroxide radical is that of Bell, Raley. Rust, Seubold, and T'aughan (a),who postulate that it reacts with an alkyl radical or another peroxy radical to form an alkoxy radical, rather than decomposing as suggested by Lewis and Von Elbe. The formation of formaldehyde can be accounted for by the oxidation of methyl radicals formed in Reaction 1 and in the decomposition of acetaldehyde. CHI'
+ On -+
HCHO
+ OH
Methane probably results from the reaction of methyl radicals with propane, or another hydrocarbon. CH3'
+ C3H8 -,CHI + CnH,'
NEGATIVETEMPERATURE COEFFICIENT.A negative temperat u r e coefficient for the reaction rate, noted earlier by Pease ($9) and Kooijman (bo), was also found in this work, and must be explained by a valid oxidation mechanism. Using a 5.6 mole ratio of propane to oxygen, the amount of oxygen reacting in 4.8 to 4.9 seconds increased from 2% a t 350" C. to a maximum of 18% a t 375" C., declined to a minimum of 5% a t 425' C., and then increased to 10% a t 450" C. and to 97% at 475" C. Judging from the changes in product composition, as illustrated in the figures, the shift in reaction with increasing temperature is an
1007
increase in Reaction 2 and a decrease in Reactions 5 and 9. A decrease in the over-all reaction rate with temperature would occur if the products of Reaction 2, propylene and perhydroxy radicals, were less active in continuing the chain than the products of Reactions 5 and 9, which are aldehydes, methoxy radicals, and hydroxy radicals. This appears to be true, since propylene is known to inhibit free radical reactions ( l 7 ) ! while aldehydes participate readily. Likewise the perhydroxy radical is considered to be less reactive than the hydroxy radical ($8). I t has also been postulated that the negative temperature coefficient region is caused by an increase in temperature decreasing the amount of acetaldehyde react,ing by a branched-chain path (35). SUMMARY
Under the conditions of this study the results indicate that a key reaction occurs between the propyl radical and oxygen, which may follow two alternate paths, one forming oxygenated organic products and the other forming propylene and theperhydroxy radical. The perhydroxj- radical reacts with propane, or another hydrocarbon, to form hydrogen peroxide and a free radical. The subsequent disappearance of hydrogen peroxide during the reaction is probably decomposition on the reactor surface, forming water and oxygen. The reactions which form the oxygenated organic products appear inadequate to account for the quantities of water and hydrogen peroxide which are formed. It is necessary to include in the reaction mechanism a reaction, such as Reaction 2, which will connect the olefin production with that of water. The negative temperature coefficient of the reaction rate is attributed to the relative react,ivities of the products formed in the two alternate reactions of the propyl radical with oxygen. With an increase in temperature, the reaction forming propylene increases relative to that forming oxygenated organic species, and since the products of the propylene-forming reaction are lees reactive, this results in a decrease in the over-all rate in the region 375" to 425' C.
(Partial Oxidation of Propune) SEPARATION OF HYDROGEN PEROXIDE FROM THE PRODUCTS CHARLES N. SATTERFIELD, ROBERT E. WILSON1, THEODORE W. STEIN, AND DALE 0. COOPER2 Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Mass.
T
H E industrial potentialities of the production of hydrogen peroxide by partial oxidation of hydrocarbons will depend to a large extent upon the ease with which it can be separated from the other species present. A major problem is caused by the fact that in the liquid phase hydrogen peroxide reacts with the aldehydes present to form organic peroxides, Although the ratio of aldehyde to peroxide in the product gas can be varied somewhat by varying the reaction conditions, substantial amounts of aldehydes are always formed under conditions yielding hydrogen peroxide. The reaction proceeds in two consecutive, reversible steps: RCHO
+ H~0n
RCH(0H)OOH (1) monohydroxyalkyl hydroperoxide
1 Present address, Research Laboratories Division, General Motors C o r p . . Detroit, Mich. 1 Present address, Wright-Patterson Airbnse, Dayton, Ohio.
RCH(0H)OOH
+ RCHO +dihydroxydialkyl RCH(OH)OOCH(OH)R peroxide
(2)
The equilibrium constants for Reactions 1 and 2 have been determined by Dunicz, Perrin, and Style (IO)at 25" C. for mixtures of formaldehyde and hydrogen peroxide and by Kooijman and Ghijsen (21) a t 0" C. for mixtures of formaldehyde and hydrogen peroxide, and acetaldehyde and hydrogen peroxide. The equilibrium constants reported ($1) indicate, for example, that if the initial mole ratio of aldehyde to peroxide in a 15 weight % aqueous hydrogen peroxide solution exceeds 1.5 to 1, nearly all of the peroxide will be present as organic peroxide. Thus, to attain high percentage recovery of hydrogen peroxide, it is necessary either to remove i t before it can react with the aldehydes, or else to use techniques causing substantially complete reversal of Reactions 1 and 2.
INDUSTRIAL AND ENGINEERING CHEMISTRY
1008
The reaction rate constants for Reactions 1 and 2 were evaluated by Dunicz, Perrin, and Style ( I O ) for formaldehyde-hydrogen peroxide mixtures. Rate const,ants have also been reported for Reaction 1 for the three lowest members of the aldehyde homologous series (34). These show that the rate increases in the order: formaldehyde, acetaldehyde, propionaldehyde. Calculations show that even formaldehyde and hydrogen peroxide react rapidly when present in a solution having the typical concentrations obtained from partial oxidation of propane. As an exaniple, in a 13 weight % hydrogen peroxide solution containing initially a formaldehyde-hydrogen peroxide mole ratio of I, the uncombined formaldehyde will decrease to approximately one half its initial value a t the end of 2 minutes' reaction a t room teniperature. The above figure applies to formaldehyde hydrate, in which form almost all of the forinaldehyde exists in moderately dilut,e aqueous solutions. Holyever, t,he formaldehyde formed in the combustion procese is presumably unhydrated and therefore might react even more rapidly with hydrogen peroxide, depending upon the rate of this reaction relative to that of the hydration yeaction of formaldehyde. Unfortunately, no studies appear to have been published on the rate of the hydration reaction. The separation is furt,her coniplicat,ed by the fact that t,he dihydroxydimethyl peroxide formed h y the reaction of formaldehyde and hydrogen peroxide decomposes readily, especially in basic solution.
CH,(OH)OOCHiOH e 2HCOOH
+ 13,
(3)
The reaction is unimolecular \vit,h a rate constant of 1.86 X 10-4 min-' a t 25" C. in neutral solut,ion. The activat,ion energy is 24.9 kcal. per gram-mole ( I O ) . ,4t 25" C. the half life is therefore approximately 5400 minutes, while a t 100" C. t,he half life is only 1 or 2 minutes. Dihydroxydiethyl peroxide, formed from the acetaldehyde-hydrogen peroxide reaction, however. is much more stable, as shown by the data in Table I, vhich were obt,ained by allowing an acetaldehyde-hydrogen peroxide solut,ion and a formaldehyde-hydrogen peroxide solution to each stand for a number of days a t room temperature.
Vol. 46, No. 5
water here is consequently much less than 1. Therefore an appreciable selective condensation of hydrogen peroxide and water might be expected to occur only if the coiidensate could be removed from the condenser before liquid-phase reactions of formaldehyde wit,h water and hydrogen peroxide can occur to a11 appreciable extent. The feasibility of this method was studied by making sis runs in which the hot product gases from a reactor were passed through two surface condensers in series, the first being held a t some fixed temperature in the range of -10" to +SO" C. f o r each run, and the second being held at a constant temperaturc of -35" C. for all runs ( 7 ) . The follo\\-ing reactor condition8 were' maintained for each run: i n k t gas temperature, 472" C.; propane-oxygen mole ratio, 8 to I ; residence time, 5 seconds. The fraction of the hydrogen peroxide formed which was recovered in the first condenser constituted 95% to 100% of the total iormed, a t all condensing temperatures studied. However, most of the aldehyde formed also appeared in the first condenser, even a t t h c highest temperatures studied. For example, a t a condenser temperature of 50" C., 95% of the hydrogen peroxide formed W H S removed but also 75yoof the aldehydes, producing a condensatc containing 23 weight % hydrogen peroxide, 13 weight % aldehyde as formaldehyde, 63 weight, % water, and lese than 1% or methanol. These results suggest that t'he condensate vas not removed from the condenser sufficiently rapidly. Therefor(. additional runs were made using a short condenser and also, i i i some cases, adding steam or water t o the condenser to wash o u i the condensate rapidly. illthough the residence time of tllc. rondensate was est'imated to be less than 1 second in these runs, i!o subetantial improvement in separation was obtained, UISTILLATIOlV O F FORMALDEHYDE-HYDROGEN PER0XII)E SOLL'TIOJVS
.-le stated above, the partial pressure of formaldehyde above its iicpeous solutions is very low at the temperatures wliich occur 011 wcuuni distillation, because of the formation of formaltlrh>~tlc hydrate and formaldehyde polymers. iit the temperatures of tlistillation a t slightly above at,mospheric pressure and higher, formaldehyde becomes the more volatile component, presumably because of reversal of the dehydration and polymerization aqueous fornialdcreactions (59). For example, ii 8.82 weight hyde solution a t 20 mm. of mercury1 a t approximately 20" C., is in equilibrium with a vapor containing 0.46 weight % forinaldehyde, Tvhile the same solution a t 08' C. forms a vapor containing 6.93 weight % formaldehyde. Overhoff has reported (97) that the addition of a large excesF: of methanol aids in the remov:il of formaldehyde from an aqueous solution initially containing formaldehyde and hydrogen peroxide. The methanol wa? prrsumed to have reacted with iornialdehyde to form methylal, which, being relatively volatile, distilled off. The reaction proceeds in two con~rcutive,reversible s t e p :
r0
.Icetaldehyde Hydrogen peroxide = ""
lormaldeliydf
I ~ ~ ~ e ; p e r & d=c 1 ' 21 Initial HzOr = 8.24of initial HZOZ 100.0 99.8 95.5
8i.O 73.3
so
1
46.3 22.1
12.5 5.8
Time. I l s y s
Initial HZOZ= 5.57 wt. % 70 of initial HzOi
.__
0 0 2s I
2 3 2 8.17 6.12 1 6 !i
23.0 32.0
100.0 99, G 99.8 99.1 98.8 99.1
99.8
99.3 99.4
+ CHaOH e methyl formaldehyde CHg(OH)OCI& hemiacet,sl CH,(OH)OCHa + CHSOH CH1(OCHa), + H20 methylal
HCHO Three methods of separation are briefly considered here: (1) fractional condensation of oxidation products, (2) removal of organic components from the condensate by distillation, and (3) conversion of the peroxide in the condensate to calcium peroxide, which can be readily converted t o h.vdrogen peroxide. FRACTIOIVAL CONDENSATION
Harris ( 1 4 ) has suggested that if the gaseous products of propane oxidation were passed through a condenser maintained a t about 40" C., a selective condensation of the hydrogen peroxide would occur, but quoted no data. The vapor-liquid equilibrium data for aqueous formaldehvde solutions in this temperature range show abnormally IQK vapor pressures of formaldehyde because it exists in the liquid phaEe almost exclusively in the hydrated or polymerized forms. The relative volatility of formaldehyde to
$
(4)
(5)
If the equilibria for these reactions lie very far to the right, thc amount of formaldehyde combined with hydrogen peroxide is reduced appreciably. Although Rpactions 4 and 5 have been investigated (1, I f ) , no true eyuilibrium constants are knowri. Several tests of this method were therefore made by batch diptillations without rectification (91). Formaldehyde was chosen for study rather than acetaldehyde; bccause the hydroxyalkyl peroxides 13-hichit forms with hydrogen peroxide are less stable than the equivalent compounds from aaetaldehyde, thereby offering a more. ere test of the practicability of a distillation process. The dist te was analyzed for formaldehyde a t the end of all runs and for acetal a t the end of one run.
INDUSTRIAL AND ENGINEERING CHEMISTRY
May 1954
The hydrogen peroxide was determined by the hydrogen iodideglacial acetic acid method. The formaldehyde was determined using the alkali hydrogen peroxide method described by Walker (39). Methylal \vas calculated as the difference between formaldehyde and the sum of methylal plus formaldehyde, the latter sum being determined by the alkali hydrogen peroxide method after any acetal present had first been hydrolyzed to formaldehyde and methanol. The solutions which were distilled all had the following initial compositions: hydrogen peroxide, 46.5 grams; formaldehyde, 26.0 grams; methanol, 218 grams; sulfuric acid, 32.5 grams; water, 159 grams. The sulfuric acid was added as a catalyst for acetal formation and the solution was allowed to stand for several hours before distillation in order for equilibrium to be attained. Distillations were performed a t various pressures between i60 and 15 mm. of mercury, with corresponding average bottoms temperatures of about 73" to 22" C. The distillation rates varied from 2 to 5 ml. per minute. At the higher temperatures, when 50% of the sample had been volatilized a substantial fraction of the hydrogen peroxide had disappeared and 20 to 32% of the initial formaldehyde was still present in the bottoms. At the lowest temperature, 16% of the hydrogen peroxide had disappeared a t the mid-point but most of the formaldehyde was still present in the bottoms. Distillation a t atmospheric pressure of a solution containing initially 423 grams of water rather than 159 grams of methanol, other constituent amounts remaining unchanged, resulted in the removal of 99% of the formaldehyde a t the mid-point of the distillation, but two thirds of the peroxide decomposed. Here the higher boiling point increased the relative volatility of formaldehyde but also increased the rate of decomposition of the peroxides. Better results could possibly be obtained in a continuous distillation operation designed to minimize the residence time of the peroxide solution in the still. The analysis of the distillate from one run revealed no methylal within an experimental error of 4% of the aldehyde present. Thus, contrary to the claim in the patent by Overhoff ( 2 7 ) , i t appears that methylal is not formed here and that the benefits gained by adding methanol result mainly from a reduced distillation temperature and partly from an increase in the formaldehyde relative volatility rather than from methylal formation. I t would appear that distilling the aqueous formaldehydehydrogen peroxide solution a t subatmospheric pressures without added methanol should be almost as effective. The distillation of such solutions can be very hazardous if a large concentration of diluent, either water or methanol, is not present. For example, in another laboratory, a violent explosion recently occurred when the following experiment was made.
A solution was prepared by mixing 28 ml. of methanol, 9.5 grams of 37% formaldehyde, 14 grams of 35% hydrogen peroxide, and 2 ml. of 50% sulfuric acid. This solution was distilled under reduced pressure, the pressure being gradually lowered from 137 to 31 mm. of mercury as the distillation progressed to keep the bottom temperature a t about 40" C. After about 60% of t,he material had been volatilized overhead, a violent explosion occurred. Similar experiments h4d been carried out previously without, event. It was suggested that possibly some peroxyformic acid had been formed, which is very sensitive thermally. PRECIPITATION OF PEROXIDE A S CALCIUM PEROXIDE
Kooijman (20) has set forth a method for converting hydroxyalkyl peroxides, such a8 are formed by Reactions l and 2, to alkaline earth peroxides, which precipitate and can be filtered from the solution and then converted to hydrogen peroxide by treatment with acid. Both barium hydroxide and calcium hydroxide form relatively insoluble peroxides. Ca(OH)2
+ H202
-.L
CaOz
+ 2Hz0
(6)
1009
Calcium hydroxide is of particular interest, however, because of its cheapness and because of the possibility of recycling the calcium compounds in an industrial process. For example, once the calcium peroxide has been filtered from the solution, it can be made into a slurry and the hydrogen peroxide regenerated by introducing carbon dioxide. CaOz
+ COZ+ HzO
.-f
CaCOj
+ HzOp
(7)
The calcium carbonate could be filtered from the slurry and then calcined to calcium oxide, which after slaking is ready for re-use in the precipitation step, Reaction 6. Although barium hydroxide could be used in a similar manner, the calcination of barium carbonate is subst'antially more difficult than that of calcium carbonate. A patent has very recently been issued to Cook (6) on the use of alkaline earth metal hydroxides in such a process. Cook describes an example in which a lime slurry a t 4' C. was added in slight excess to a 1 to 1 mole ratio mixture of formaldehyde and hydrogen peroxide a t 0" C., containing a 1.5iM concentration of peroxide. By very rapid mixing, filtering, and washing, 96.8y0 of the starting peroxide was converted to calcium peroxide. Subsequent treatment with carbon dioxide gas resulted in recovery as hydrogen peroxide of 89.8% of the starting peroxide. Similar studies were made by the authors before the issuance of the above patent, using mixtures of hydrogen peroxide with formaldehyde or acetaldehyde in about the same concentrations as quoted above and contacting them with lime slurries under various conditions. I t was found that in contact with the lime slurry, mixtures of hydrogen peroxide and formaldehyde were far more unstable than hydrogen peroxide alone or equivalent mixtures with acetaldehyde. Also, the formaldehyde-hydrogen peroxide solution decomposes far more rapidly in contact with a lime slurry than when mixed with a sodium hydroxide solution, indicating that surface decomposition is the most rapid reaction occurring. It is particularly important to operate a t temperatures near 0" C., since the decompositions are exothermic and can lead to a self-accelerating reaction. Recoveries of over 90% can be readily obtained from acetaldehyde-hydrogen peroxide solutions by operation near 0" C., even if the solutions are in contact with lime slurry for several hours. With formaldehydehydrogen peroxide solutions, hoa-ever, rapid manipulation is required for high recovery. For a given procedure, the per cent recovery drops as the formaldehyde-hydrogen peroxide ratio increases. Thus, the per cent recovery of calcium peroxide from the condensed product mill vary with the composition. Practically all of the peroxide not recovered is destroyed by decomposition-Le., the peroxide content of the filtrate is always very small. SUMMbRY
Of three methods which were st,udied for the separation of the oxidation products, neither fractional condensation nor distillation in the presence of excess methanol was found to give satisfactory results and the latter process can be hazardous. A high per cent recovery can be obtained by precipitation of calcium peroxide and its conversion to hydrogen peroxide. ACKNOWLEDGMENT
The authors wish to acknowledge the contributions of Leon Lapidus, who assisted in the original design of the propane oxidation equipment and performed the preliminary work which delineated the experimental regions of interest. Thev are indebted to the General Motors Corp. for financial aid in the form of a fellowship to one of the authors (R. E. W.) and for the detailed quantitative infrared analysis of exit gas samples. Financial support was also received from the Office of Naval Research under Contract N5ori-07819, NR-092-008.
1010
INDUSTRIAL AND ENGINEERING CHEMISTRY LITERATURE CITED
ddams, E . W., and Adkins, H., J . Am. C h e n ~ .Soe., 47, 1358 (1925).
Bell, E. R., Raley, J. H., Rust, F. F.,Seubold, F.H.. Jr., and Vaughan, K.E., Discussions Faradav Snc., 1951, No. 10, 242. Burke, 0 . W., Starr. C. E., and Tueniniler. F. D., “Light Hydrocarbon .Inalysis.” pp. 134, 223-31. Yew Tork, Reinhold Publishing Corp.. 1981. Chernyak, S . Ya., and Shtern. T-. Ya., Doklady A k a d . S a u k S.S.S.R., i8,91 (1951). Cook, G. A. (to Linde Air Pioducts Co.), U.5. Patent 2,416,156 (Feb. 18. 1947). Cobk, G. i. (to Union Carbide & Carbon Carp.), I b i d . , 2,614,907 (Oct. 21, 1952). Cooper, D. O., 8.31.thesis, llassachusetts Institute of Tcchnology, 1962. Dickey, F. H., et al., IND.ENG.CHXM., 41, 1673 (1949). Duke, F. R., IXD.ESG. C m s f . , A 1 y . k ~ .ED.,17, 572 (1948!. Dunics, B. L., Perrin. D. D., and Style, D. IT7.. G., Trans. Faraday Soc., 47,1210 (1951). Evans, M. G., Hush, S . S., and r r i , S., Quart. Rev. (London), 6,186 (1952). Fischer, E., and Giebe, G . , Ber., 30, 305 (1897). Geib, K. H., and Harteck, P., 2. p h y s i k . Chem., AIi0, 1 (1934). Harris, C . R. (to E. I. du Pont de Semours 8; Co.), U. S. I’atent’2,533,581(Dee. 12, 1950). Harris, E. J., Trans. Faladall Soc., 44,764 (1948). Harris, E. J., and Egerton, A. C.. Chem. Rers., 21, 287 (1937). Hinshelwood, C. N.,“The Kinetics of Chemical Change,” p. 97, London, Oxford University Press, 1947. Jost, W., tr. by Croft, H. O., “Explosion and Combustion Processes in Gases,” pp. 317, 326, New Tork, RIcGraw-Hill Book Co., 1946. Kahler, E. J.,et al., IKD.Ex.CHEM.,43, 2777 (1951). Kooijman, P. L., Rec. tias. chin.,66, 5 , 217 (1947). Kooijman, P. L., and Ghijsen, W. L., Ibid., 66, 205 (1947). Lacomble, A. E. (to Shell Development Co.), U. 9. Patent 2,376,257 (May 15, 1945).
Vol. 46, No. 5
Lewis, B., and \‘on Elbe, G., “Combustion, Flames, and Ex-. plosion of Gases,” pp. 28, 30. 101, 119, 121, 130, New York, Academic Press, Inc., 1961. AIcDonald, G. E., and Schalla, R. L., Satl. Advisory Comm. Aeronaut,., NACA RM E52G17 (Aug. 28, 1952) hlclane, C. K., J . Chern. Phus., 17,379 (1Y49). and Thornes, L. S.,J . Chem. Soc., 1937, 1656. Newitt, D. M., Overhoff, J. (to N. V. de Bataafsche Petroleum Riaatschappij), Dutch Patent 61.465 (dug. 16, 1948). Paneth, F., and Lautsch, TTr,, Ber., 64, 2708 (1931). Pease, R. N., J . A m . Chem. Soc., 51, 1839 (1929). Ibid.,57,229G (1935). Pease, R. IT,, and LIunro. W. P., Ibid., 56, 2034 (1934). Rice, F. O., and Evering, B. L., Ibid.. 55,3898 (1933). cKee, F. S., and Quagliano, Rodebush, W.H., Iieizer, C . R J. V., J . Am. Chem. Soc., 69, 538 (1947). Satterfield, C. S., and Case, L. C., ISD.EXG.CHEM.,in press. Shtern. V. Ya, and Antonovski, P. L., Doklady &ad. A97acde S.S.S.R., 78,303 (1951). Steacie, E.W. R., ”Atomic and Free Radical Reactions,” A. C.S. Monograph 102, pp. 284. 318, 339-44, 369, New York, Reinhold Publishing Corp., 1946. Stein, T., S.M. thesis, Massachusetts Institute of Technology, 1952. Sewarc. h i . , Discussions Faraday Soc., S o . 10, 143 (1961). Walker, 3. F., “Formaldehyde,” 2nd ed., A.C.S. Monograph 120, New Tork. Reinhold Publishing Corp., 1953. Walsh, A. D., J . Chem. Soc., 1948,331.
RECEIVED for review April 27, 1953.
ACCEPTEDJanuary 29, 1954. IIaterial supplementary t o this article has been deposited as Document No. 4198 with tlie AD1 duxiliary Publications Project, Photoduplication Service, Library of Congress, Washington 25, D. C. A copy may be secured by citing the document number and by remitting $1.25 for photoprints or $1.25 for 35-mm1.microfilm. Advance payment is required. Make checks or money orders payable to Chief, Photoduplication Service, 1.ibrary of Congress.
Pressure Limits of Flame Propagation Qf Propane-Air Mixtures INFLUENCE OF WALL QUENCHING FRANK E. BELLES AND DOROTHY AI. SIMOX Sational Advisory Committee f o r Aeronautics, Cleveland, Ohio
ROBERT C. WEAST Case InstitzLte of Technology, Cleveland, Ohio
T
HE accelerated pace of combustion research in recent years
has produced a large amount of data on the so-called fundamental properties of combustion. The interpretation of these results, however, has been made more difficult by their very quantity and by the fact that, until recently (fO), there has been little understanding of tlie possible relat,ions of the combustion phenomena to one another. This difficulty is quite apart from the basic one-that, owing to lack of knowledge of the most intimate, molecular-scale processes of combustion, no universally applicable theory of combustion has as yet appeared. It seemed to the authors t,hat one way in vhich combustion knowledge could be partly systematized x-as through a more careful study of the pressure limits of flame propagation. It has long been knoxyn that there are concentration limits of flammability for the process of combustion (9)-that is, the mixture of fuel and oxidant may contain too much (rich limit) or too little (lean limit’) fuel to burn. Many workers have investigated the concentration limits at atmospheric pressure (2); a few have determined the
effect of reduced pressure on these limits and have found that the concentration range of flammability narrows as the pressure is reduced, until at some critical pressure the rich and lean limits converge (3, 4, I f , 12). The reeult is the familiar U-shapcd curve of pressure limit against fuel concentration? in which the uprights of the U correspond to the concentration limits of flammability. These pressure limits have been measured and reported as a fundamental property of combustion, but scattered references in the literature, summarized by Friedman and Johnston (Or), present some evidence that the limit,s are possibly governed by quenching effects. From this point of view, then, t,he pressure limit is not a property of the combustible mixturc alone, but is due t o effects of t,he confining walls on the propagation of the flame. The aim of the work described herein, therefore, was to systematize part of the combustion data by making a quantitntivc? connection between preesure limits and wall quenching. Tile method of at,tack \\-as to measure the pressure limits of flame