(PREFLAME COMBUSTION OF HYDROCARBONS) Identification o f 8-Dicarbonyl Compounds M. R. BARUSCH,
H. W. CRANDALL, J. Q. PAYNE, AND
J. R. THOMAS
California Research Corp., Richmond, Calif.
T
HOilIAS and Crandall ( 6 ) have established the potential importance to the cool flame reaction of the compound which absorbs light a t 2600 A. The present paper concerns its identification. Il-hile investigating the slow oxidation of n-butane by means oi ultraviolet absorption spectroscopy, Egerton and Pidgeon (3) observed the growth of a strong band at about this wave length, indicating the appearance of an unknown reaction intermediate Further investigations by Ubbelohde (71, who termed this intermediate compound X, and by Egerton and Young (4)failed to idwtify the compound responsible for the absorption band, although the former author eupiessed the opinion that the conipound probably resulted from decomposition of a peroxide and that its structqe might resemble that of ascorbic acid. Egerton and Young stated that this material was probably a complicated peroxide possessing a ring structure which might arise from polymerization or condensation reactions between aldeh] des and alkyl peroxides or hydrogen peroxide. Ubbelohde ('7) reported that compound X, which absoibed a t 2600 A, did not appear unless the hydrocarbon giving iise to the cool flame contained a t least a four-carbon chain. The fact that the material being studied in the present work likewise showed both of these chaiacteristics left little doubt that thc authors were working with Ubbelohde's compound X. Identification of the agents absoibing at 2600 8.nas made by investigating the spectra of a variety of conjugated systems of the appropriate molecular weight, since the 2600 A. band indicated such a structure. 2,4Pentanedionc gave an almost exact match of the absorption shown by the product from n-pentane oyidation. Firm proof of the identity nas established by concentrating the appropriate fraction from the cool flame product of ?&-pentaneand converting it to the copper enolate. The copper enolate from the n-pentane cool flame product was shown by x-ray diffractionto be identical nith a sample prepared from known 2,4-pentanedione. Application of the above technique to the identification of the cool flame product from n-butane nas not practicable because of the instability of butanal-3-one. Honever, synthetic butanal-3one prepared and kept in low concentration in a dry nonaqueous solvent gave an absorption spectrum quite similar to that of 2,4pentanedione and to the 2600 A. intermediate band produced in the cool flame from n-butane. By analogy to the case of 2,4pentanedione from n-pentane established above, the source of the 2600 A. band absorption from n-butane can be attributed to butanol-3-one. Some corroboration of the assignment was obtained by showing that the rate of deconiposition by water of an iso-octane solution of known butanal-3-one matched the rate of decomposition similarly induced by adding 7%ater to the isooctane solution of the cool flame product of n-butane. The production of butanal-3-one from n-butane suggests that wine pentanal-3-one might be produced in addition to the 2,4pentanedione from the cool flame combustion of n-pentane. This absorption band. P-ketoaldehyde should contribute to the 2600 9. However, it would not be as stable as the @-diketoneand probably for this reason was not isolated from the combustion product. Since all 8-diketones and p-ketoaldehydes will probably show similar absoiption spectra, it is expected that the 2600 A. absorption band observed during the combustion of higher paraffiic hydrocarbons is due to one or both of these types of structures. The compounds 1,3-propanediol and 2-methyl-l,3-propanediol,
possible intermediates in propane and isobutane oxidation, respectively, do not appear to hare been prepared. The failure to observe a 2600 A. absorption band during the combustion of propane and isobutane may be due to the instability of thcsc compounds. EXPERIMENTAL
2,~PENT.lh-EDIOS'iC1N A comparison of thp ultraviolet absorption spectra of 2,4-pentanedione in the vapor phase and in iso-octane solution with the cool flame combustion products of n-pentane is presented in Figure 1. The solid curves represent the pure vapors in both cases and the broken lines represent the iso-octane solutions. The data obtained for thc solution of 2,4-pentanedione are in agreement with those of Rasmussen, Tunnicliff, and Brattain (6). The combustion product vapors were produced from equal mole mixtures of n-pentane and oxygen a t 276" C. in the quartz cell of the spectrogmph.as desciibed by Thomas and Crandall(6). rLTRAVI0LET SPECTRAL E V I D E N C E O F
TKY COalBUSTION PRODUCT O F n-PENTANE.
.
3 01
I _ -
I
I
I
V
I
0.1 2000
2200
2400
2600
2800
3000
3200
3400
3600
3800
4
IO
W A V E LFNGTH A
Figure 1.
Spectra of 2,4-Pentanedione and Combustion Product of n-Pentane A. Combustion product in iro-octane E.
2,4-Pentanedione vapors
D.
Combustion product vapors
C. 2,4-Pentanedione in iao-octane
The combustion product which was dissolved in iso-octane wa8 obtained from the organic phase of the condensate produced in B flow system of the type described by Baruech and Payne ( I ) . The reaction was carried out at 304" i.3" C. employing 150 mm. CJf pentane and 610 mm. of air. Inspection of Figure 1shows that the peaks of the spectra of the synthetic and combustion vapors as well as the corresponding iso-octane solutions are in good agreement. ISOLATIOlV O F 2,4-PEhTTSPSEDIOUE FROM THE C O O L rL4M.E PRODUCTS OF %-PENTANE. The cool flame flow systemmentioned above was used to collect a large sample of combustion product The reaction was again carried out at 304" f 3" C. The npentane was introduced by carburetion with air maintaining the carburetor at 0" C. The flow rate was adjusted so that 0.5 ml. of liquid pentane was consumed per minute. The condensate 13 as cooled in a dry-ice trap. It consisted of approximately one part aqueous phase to five parts organic phase. Spectroscopic
2764
INDUSTRIAL AND ENGINEERING CHEMISTRY
December 1951
data indicated a concentration of about 0.015% P-diketone in the organic phase. From about 1900 ml. of n-pentane approximately 1100 ml. of organic phase were recovered. This was concentrated t o 125 ml. by evaporation a t atmospheric ressure. During this operation the maximum temperature reacxed by the bottoms was 50" C. This material was concentrated to 20 ml. by distillation at 4 mm. Spectrochemical data indicated a concentration. of 0.7% 2,4pentanedione in the final 20 ml. of concentrate. Ten milliliters
WAVE LENGTH A .
Figure 2.
Spectra of Butanal-3-one and Combustion Product of n-Butane A.
8.
C. D.
Combustion product in iso-octane Butanal-3-one vapors Butanal-%one in iso-&ne Combustion product vapon
of this concentrate was agitated with 10 ml. of saturated aqueous
cupric acetate. After 10 minutes 0.08 gram of the enolate was collected. This material was purified b recrystallization from benzene, A comparison of the x-ray digraction pattern of this material with a known sample of the copper enolate of 2,4-pentanedione showed the materials to be identical. Carbon-hydrogen analysis of the enolate isolated from the coxxibustion product was as follows: Calculated for CioHiaOcC~i Found
*
Carbon 45 9 46 0
Hydrogen 5 4 5 5
SPECTRAL EVIDENCE OF BuTANAL-~-oNE. Figure 2 presents curves of the ultraviolet absorption spectra of the cool flame prdducts of n-butane and of synthetic butanal-3-one. The vapor phase-combustion product was obtained by reaction of an equal mole mixture of n-butane and oxygen at 280" C. and 460 mm. The combustion product in iso-octane was obtained from the reaction of a stoichiometric mixture of n-butane and air at atmospheric pressure and 343" f 3" C. Since pure li uid butanalS o n e is unstable, the sodium salt was prepared by t i e method of Claisen (8). A solution of the dicarbonyl compound wa8 obtained by decomposition of the salt with 10% sulfuric acid in the presence of iso-octane. The maxima of the curves in the spectra from the combustion product as shown in Fi ure 2 are in agreement with the maxima obtained from the synt%etic butanal-Sone solution. WATER-INDUCED INSTABILITY O F BUTANAL3-ONE. A Solution of butanal-3-one in iso-octane was prepared to give the same optical density at 2675 A. as a sample of the combustion product of n-butane dissolved in iso-octane. Both solutions were vigorously agitated with 5% by volume of water. The organic phases were separated and the optical density measured again. Both solutions showed considerable but approximately equal dro s in optical density. After storage overnight neither had signiAant absorption at 2675 A. DISCUSSION
One of the most interesting observations made in these investigations is that of the remarkable decrease in the concentration of the p-dtcarbonyl compounds during the cool flame reaction (6). A priori it is not certain that the concentration of a critical intermediate behaves in this manner. However, the nature of the cool flame phenomena and, in particular, the feature of flame
2765
periodicity may find explanation in an intermediate whose concentration increases to a critical point resulting in the occurrence of the flame and consequent destruction of the intermediate. The decrease in p-dicarbonyl concentration with occurrence of the cool flame is real and not a temperature effect on the keto-enol equilibrium. The keto-enol tautomerism is reversible and if it were responsible for the rapid decrease in the apparent 8-dicarbonyl concentration indicated in the spectrograms, the amount of 0-dicarbonyl should return to the original value shortly after passage of the cool flame. Experimental data on the chemical and physical behavior of 8-dicarbonyl compounds showed that the keto-enol tautomerism is responsible for only a small portion of the decrease in the 2600 A. absorption band. While the heat evolution during cool flame combustion is not great, the flame zone is a t a temperature 100" to 200" C. higher than that of the reaction cell, and consequently the optical density is influenced to some extent. Although the partial recovery of the optical density following the initial abrupt drop may be due, at least in part, to a shift in the tautomeric equilibrium, the major decrease must be due to a decrease in the gross concentrations of the compounds. The partial recovery of the optical density following the cool flame could also be accounted for by a thermally induced decomposition of the B-diketone's precursor. The i n t e r p r e t a t i o n that the major decrease in 8-dicarbonyl concentration was not due to the keto-enol tautonierism was verified by adding 2,4-pentanedione to isobutane which gives no 2600 A. absorption band and by followingits behavior during the cool flame combustion of the primary fuel. A t pressuresof theorder of lmm. of m e r c u r y t h e 2 , 4 pentanedione was relatively stable a t these elevated temperatures in both the absence and p r e s e n c e of o x y g e n . The time dependence b e h a v i o r of 2,4-pent a n e d i o n e during the Figure 3. Optical Densit at cool flame combustion of 2600 A. as a Function of {me i s o b u t a n e g i v e n in Cool flame combustion ot lsobubne plus Figure 3 definitely P,4-~entanQdlone shows that nonreversible chemical destruction of the compound takes place. Similar results were obtained with butanal-3-one. The amount of p-dicarbonyl compounds produced in the reaction is of interest. In typical cool flame oxidations of n-butane or n-pentane the maximum absorption intensities observed at 2600 A. corresponded to concentrations of p-dicarbonyl compounds in the order of magnitude of 1% of the hydrocarbons charged. At temperaturea below the cool flame zone as much as 10% of the initial n-pentane present was found to be converted to the p-dicarbonyl compound. If it is assumed that the rate of destruction of the 6-dicarbonyl compounds is the same during the induction period preceding the cool flame as that of 2,4pentanedione shown in Figure 3, it can be estimated that as much as 10 to 15% of the initial n-pentane charged waa converted to a @-dicarbonyla t some time during the reaction. As only a portion of the fuel is oxidied, it seems apparent that a considerable fraction of it must go through a p-dicarbonyl structure.
2166
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
Vol. 43, No. 12
SUMMARY
LITERATURE CITED
The combustion intermediates which give rise to the 2600 A. absorption band observed during the Oxidation oi most hydrocarbons and referred to by Ubbelohde ( 7 ) as compound X were identified ‘as p-dicarbonyl compounds. The concentrations of these compounds decrease during cool flames whereas the concentrations of all the other intermediates observed increase during cool flames. Representative p-dicarbonyl compounds established were butanal-3-one from n-butane and 2,4-pentanedione from n-pentane. It was shown that a considerable fraction of the oxidized hydrocarbon goes through a p-dicarbonyl structure at some time during its oxidation.
(1) Barusch, R.1. It., and Payne, J. Q., IND.ENG.CHEM.,43, 2329 (1951). (2) Claisen, L., and Meyerowvita,L., E’er., 22,3273 (1889). (3) Egerton, A, C., and Pidgeon, L. M., Proc. R o y . SOC.( L o n d o n ) , A142, 26 (1933). (4) Egerton, A. C., and Young, G. H. S.,Trans. Faraday Soc., 44, 750 (1948). ( 5 ) Rasmussen, R. S.,Tunnicliff, D. D., and Brattain, R. R., J . Am. Chem. Soc., 71, 1068 (1949). (6) Thomas, J. R., and Crandall, H. W., IND. ENG.CHEM.,43, 2761 (1951). (7) Ubbelohde, A. R., Pioc. Roy. Soc. ( L o n d o n ) , A152, 378 (1935). RECEIVED July 9. 1951.
(PREFLAME COMBUSTION OF HYDROCARBONS) Possible lntermediates to p-Dicar6onyl formation M. R. BARUSCH, J. T. NEU, J. Q. PAYNE, AND J. R. THOMAS California Research Corp., Richmond, Calif.
H E paper by Barusch, Crandall, Payne, and Thomas ( 2 ) has established that a relatively large amount of hydrocarbon oxidized in preflame reactions passes through a @-dicarbonyl intermediate. The rate of formation and consumption of &dicarbonyl compounds during the cool flame processes suggests thalt these compounds might be important intermediates in the preflame oxidation of hydrocarbons. This report is concerned with possible precursors to the formation of @-dicarbonyl compounds and with some speculation on a possible mechanism by which these materials are formed. The likelihood of a single multimolecular collision giving rise to p-dicarbonyl compounds is negligible and consequently it is concluded that the formation of these compounds occurs in a series of steps. As intermediates to the p-dicarbofyls one might reasonably expect free radicals-e.g., CH~CH~CHCHS .OH, , .OOH, etc.-rather unstable compounds such as peroxides, and possibly some comparatively stable compounds, such as aldehydes or ketones. This paper is concerned chiefly with possible stable intermediates. In this work a variety of conceivable intermediates in the preflame combustion of n-butane were evaluated for their tendencies to produce p-dicarbonyl compounds.
T
EXPERIMENTAL
The equipment employed was essentially the same experimental arrangement described by Thomas and Crandall (8) with the exception that some of the experiments were carried out with a gas handling system which was mercury-free. I n the latter equipment an oil diffusion pump was substituted for the mercury vapor pump. Also a modified sensitive altimeter and a therniocouple gage replaced the mercury manometer and McLeod gage, Control experiments with n-butane showed no obvious difference between the system containing mercury vapor and the mercury-free apparatus. Three methods were used for determining the likelihood of a given compound being an intermediate in the formation of the p-dicarbonyl compounds from n-butane. The most direct method was to measure the tendency of the pure compound to produce p-dicarbonyls in preflame reactions by subjecting mixtures of oxygen and the compound in question to preflame reactions. The majority of the data reported herein was obtained in this manner. The second method, used in some cases, was to substitute a small amount of the compound to be studied for a part of the nbutane to determine the amount of p-dicarbonyl formed quanti-
tatively. This method nas not generally employed since the addition of compounds to the n-butane so changed the nature of its oxidation-e.g., shortened or lengthened the induction period, promoted or inhibited the cool flame-that interpretations of the amount of p-dicarbonyl formed were subject to considerable uncertainty. In the third method, the compound to be investigated was oxidized with a “carrier” hydrocarbon which had similar characteristics to n-butane but did not itself produce @-dicarbonyls. If p-dicarbonyls were formed, they would presumably be produced by the compound rather than the carrier. Propane and isobutane are such possible carriers. However, experiments of this nature using propane were unsuccessful. A mixture of 20 mm. pressure of n-butane Trith 180 mm. of propane and 200 mm of oxygen did not produce the characteristic p-dicarbonyl band on oxidation; consequently this approach was rejected. Kone of the above three methods was ideal. The first one proved to be the best and nas generally applied but it had the following disadvantages: 1. The conditions under \vhich the euperimental iniyture exhibits preflame reactions did not necessarily overlap the conditions under which the butane-oxygen mixture reacts. In some cases higher temperatures were required to obtain preflame reactions and in others lower temperatures were employed in ordei to avoid hot flames. 2. The experimental arrangement employed restricted the usable pressures to the vapor pressures of the compounds a t room temperature. Thus, the pressure of the compound being investigated would be, in general, less than what could be produced if.it were an intermediate. 3. In high concentration some of the compounds exhibited absorption or formed products which had absorption that overlapped the band of p-dicarbonyl compounds. In such caws it was difficult to ascertain the presence of p-dicarbonyl compounds.
In employing methods one and two, the spectra were recorded by two techniques, separate experiments being necessary for each determination. With the first technique, the gases were introduced into the cell and successive rapid scans were made of the region from 4000 to 2000 A. Each scan required about 45 seconds. This was suitable for following changes in the optical density where these changes were slow and extended over a period of a t least 2 minutes. When a cool flame occurred, rapid changes in the absorption were caused. This effect was evident when the tracing of the absorption was made a t one wave length rather than over the whole spectrum. The second technique, used in some cases, was to follow the change of absorption of a given wave length with time during an oxidation reaction.
*