212
INDUSTRIAL AND ENGINEERING CHEMISTRY
As would be expected, the lower molecular weight alkyl diaryl phosphate esters are more readily volatilized from plasticized compositions than the higher molecular weight members. The alkyl ditolyl esters are less volatile than the corresponding alkyl diphenyl compounds. As was noted in the thermal stability of these phosphate esters per se, the alkyl ditolyl esters impart greater heat stability to plasticized compositions than the alkyl diphenyl esters. With respect to the low temperature properties of plasticized compositions, the alkyl diphenyl phosphate esters are more efficient than the alkyl ditolyl phosphates, the normal alkyl derivatives of a homologous series imparting generally the greatest efficiency. The apparent similarities between the low temperature properties of the plasticized compositions and the viscosity characteristics of the phosphate esters should be noted. The commercial usefulness of the triaryl and the trialkyl orthophosphate esters is well known. A comparison of the properties of these two classes with the properties of the alkyl diaryl phosphates suggests many similar utilizations. CONCLUSIONS
Maiiy of the chemical and physical properties of the alkyl diaryl phosphates have been described in this paper. Considering that they are esters of an inorganic acid, they exhibit good stability and other excellent characteristics in comparison with esters in general. The alkyl diaryl phosphates are chemically intermediate between trialkyl and triaryl phosphates and their
Vol. 46, No. 1
properties are also intermediate. Alkyl diaryl phosphates are useful as plasticizers, especially for polyvinyl chloride and polyvinyl copolymers. The economic value of these compounds is the substitution of lower cost aryl groups for essentially functional equivalent alkyl groups. ACKNOWLEDGMENT
The authors are indebted to H. I. Anthes, D. B. Guthrie, N. L. Jennings, and W. S. Knowles for the preparation of many of the esters. J. K. Craver supplied the plasticizer application information. LITERATURE CITED
(1) Bried, E., Kidder, H. F., Murphy, C. AI., and Zisman, W. A., IND. ENG.CHEM.,39,484 (1947). (2) Gamrath, H. R., U. S.Patent 2,504,121 (April 18, 1950). (3) Gamrath, H.R.,and Craver, J. K., Ibid., 2,557,089 (June 19, 1951); 2,596,140 (May 13, 1952); 2,596,141 (May 13, 1952). (4) Graves, G., Ibid., 2,005,619 (June 18, 1935). (5) Modern PEastics, 29, No. 2, 107 (1951). (6) Natelson, S., and Zuckerman, J. L., IND.ENQ. CHEM.,ANAL. ED.,17, 739 (1945). (7) Vogeli, F., Ann., 69, 190 (1849). (8) Williamson and Scrugham, Ann., 92,316 (1854). (9) Zelger, G. E., U. S. Patent 1,685,443 (Sept. 25, 1928). RECEIVED for review June 4, 1953. ACCEPTED September 12, 1963. Presented before the Division of Paint, Plastics, and Printing I n k Chemistry a t the 123rd Meeting of the AMERICAN CHEVICAL SOCIETY,Loa Bngeles, Calif.
Reaction Processes Leading to the Spontaneous Ignition of Hydrocarbons CHARLES E. FRANK' AND ANGUS U. BLACKHAM2 Applied Science Research Laboratory, University of Cincinnati, Cincinnati, Ohio
A
CONSIDERABLE amount of data is available showing the formation of peroxides, aldehydes, ketones, and other products by the partial oxidation of hydrocarbons (IO). However, because of the extreme complexity of the reaction, there is a lack of quantitative information concerning variation in the nature and amounts of these intermediates with hydrocarbon structure and with reaction conditions. Earlier studies in this laboratory have shown the effect of hydrocarbon structure on ease of spontaneous ignition (4). The present work was initiated to uncover fundamental information regarding the intermediates leading to spontaneous ignition, and the variation in these intermediates with different representative hydrocarbons and a t different stages of reaction. Results along similar lines have been reported recently by Malmberg (11). APPARATUS
The apparatus used comprises essentially three stages: the hydrocarbon-oxygen mixing tube, the oxidation chamber, and a series of traps. Oxygen from a cylinder was dried by bubbling through sulfuric acid, passed through a flowmeter, and bubbled into the hydrocarbon, which was held a t a constant temperature to secure a constant fuel-oxygen ratio. The fuel-oxygen mixture entered the oxidation chamber through a glass tube heated slightly higher 1 Present address, Research Division, National Distillers Products Ccrp., Cinoinnati, Ohio. 4 Present address, Rrigham Young University, Provo. Utah.
than the vapor bath to prevent any condensation. This tube contained a plug of glass wool to inhibit the propagation of the reaction from the oxidation chamber back to the fuel reservoir. Despite this precaution, an explosion occurred in the fuel reservoir on two occasions; accordingly, the apparatus should be shielded to minimize this hazard. In the case of isobutane, a second flowmeter was used instead of the hydrocarbon-oxygen bubbler. All oxidations employed a stoichiometric mixture of oxygen and . . hydrocarbon. Reaction has been carried out either in a single glass tube 1/18 inch in inside diameter, or in a bundle of six such tubes employed to obtain larger amounts of oroduct. With the sinele tube, thermal equiliubrium and contaEt were obtained by inseking the glass tube in a drilled steel rod which fitted into a 12-inch combustion furnace mounted vertically; the multiple reaction tubes were inserted in a single glass tube filled with iron fillings. The various contact times and temperatures are listed in the tabulations of experimental results. The temperature distribution throughout the reaction tube was measured a t various temperature settings a t zero gas flow rate. The characteristic curves obtained are shown in Figure 1. Another series of readings was made a t 550' C. and a t several gas flow rates. As the only appreciable difference in readings occurred within the first inch of the furnace length, the temperature distribution throughout the furnace was considered essentially unchanged within the range of gas flow rates employed. The use of the l/ls-inch tube throughout this work is a major factor in obtaining reaction control. This is in accord with previous observations regarding the effect of reaction chamber dimensions on the course of a vapor phase, free-radical reaction.
I N D U S T R I A L A N D E N G I N EZE R I N G C H E M I S T R Y
January 1954
In general, a high surface-volume ratio tends to suppress chainbranching, explosive reaction by supplying a “third body” to absorb energy from the system. The critical effect of tube size in this work has been demonstrated in subsequent experiments (6) in which a ‘/c-inch tube was substituted for the l/le-inch tube. In the ’/a-inch tube, violent explosions occurred a t the higher temperatures where the ‘/l&-inchtube had permitted easy reaction control; a t the lower temperatures, however, results were qualitatively the same as those obtained in the 1/16-inchtube. No study was made of possible effects of “conditioning” the tube before reaction; as each reaction occurred over a considerable period, it is not believed that previous conditioning would have had any appreciable effect on the experiments reported.
COO
-OUTSIDE THERMOCOUPLE READING:
600‘
500
P 400
a U
k!
t
300
200
too
0
TOP
I
2
3
4
5
6
7
8
9
IO
II
12 I3 BOTTOM
LENGTH IN INCHES ALONG THE REACTION TUBE
Figure 1. Temperature Distribution in Reaction Tube
The third stage consisted of a series of two dry ice traps and a 2,4-dinitrophenylhydrazine solution through which the gas was bubbled after leaving the traps. Substantially all the condensable products collected in the first dry ice trap. However, inall cases where considerable reaction occurred, an appreciable portion of the carbonyl fraction also was collected in the dinitrophenylhydrazine trap. The exit gas was checked in some runs for carbon monoxide, carbon dioxide, hydrogen, oxygen, and volatile hydrocarbons by means of a Burrell Industro gas analyzer. However, in view of the large amounts of unreacted oxygen present, this determination was not sufficiently sensitive to give meaningful results. In a number of the n-heptane runs, persistent “fogs” were obtained which could not be condensed in cold traps or in solvent8 (8). A quantity of this fog was collected and allowed to stand overnight in a stoppered flaek. An oily film was deposited on the walls of the flask; no attempt was made to identify this material. TEST PROCEDURE
All the hydrocarbons used in this study were obtained from the Phillips Petroleum Co. and were of a 99 mole % purity. A t the conclusion of a reaction, the walls of the inlet tubes and traps were coated with a solid icelike substance. At room temperature this melted to an iso-octan&mmiscible liquid exhibiting strong peroxide and carbonyl tests. This material contained most of the hydrogen peroxide and formaldehyde, undoubtedly partly as the carbonyl-peroxide addition product, HOOCHzOH (8). The water present in this phase was not determined. Major components of the hydrocarbon layer were olefins, organic peroxides, and carbonyl compounds. The following procedure was employed for separating the various reaction products.
213
Water was added to the reaction mixture to dissolve the icelike phase. This water solution was then transferred to a volumetric flask and the hydrocarbon phase was washed with more water, About six such extractions removed the water-soluble components from the hydrocarbon phase. The washings were collected in a volumetric flask and diluted to a standard volume, and aliquot portions were analyzed for hydrogen peroxide, total peroxide, total carbonyl, formaldehyde, and acid content. The hydrocarbon phase was analyzed for total peroxide and total olefin, the latter on a distilled sample. The olefin products were separated only in the case of iso-octane (2,2,4trimethylpentane). ANALYTICAL METHODS
Total peroxides were determined iodometrically. A piece of dry ice was added to 10 ml. of glacial acetic acid in the titration flask to sweep out the air. Excess potassium iodide solution and an aliquot portion of the product were then added; the flask was stoppered loosely and allowed to stand one-half hour in the dark. The liberated iodine was titrated with 0.1N thiosulfate, Hydrogen peroxide was determined colorimetrically by the procedure of Eisenberg (9), using titanium sulfate reagent. The transmittance a t 420 mp was measured with a spectrophotometer, and the hydrogen peroxide content read from a standard curve prepared with known samples of hydrogen peroxide, The difference between the total peroxide and hydrogen peroxide determinations was assumed a meamre of the organic peroxides. Total carbonyl content was obtained by precipitation of the 2,4-dinitrophenylhydrazonesusing the reagent solution described by Shriner and Fuson (16). The precipitate was filtered, washed, and transferred to a weighing bottle after dissolving in ether. The ether was removed, and the product dried under reduced pressure and weighed. The chromatographic procedure of Roberts and Green (14) was investigated as a means of separating the 2,4-dinitrophenylhydrazones. This involves the development of the chromatogram on a silicic acid-Super-Cel column with a 4% solution of ether in petroleum ether. However, difficulty was encountered in attempting to identify the various bands on the column; several recrystallizations usually were necessary and the amounts of material were small. Accordingly, the chromatographic method as used so far in this work served mainly as an indication of the number of carbonyl compounds present and of their relative amounts. The following is a typical analysis of carbonyl fraction obtained by the oxidation of iso-octane: Fraction 1 2 3 4
wt
Mi: 3.6 6.1 12.6 29.4
M.P., O C. 188 166 124-134 156-156
Fraction 4 (most strongly adsorbed) was shown to be largely the formaldehyde derivative (melting point 166’ C.); fraction 3 was largely the acetone derivative (melting point 124” C.), Fractions 1 and 2 indicate the presence of other carbonyl compounds in smaller amounts. When such mixtures are chromatographed, the precipitate of mixed dinitrophenylhydrazones usually contains a small amount of the reagent itself which tends to crystallize with the derivatives. As the reagent is more strongly adsorbed than any of the derivatives, it is removed most conveniently by chromatographing through a short columnLe., of sufficient length to adsorb the reagent only. In most of the work, the aldehyde component of the total carbonyl was determined by precipitation of the methone (5,5dimethyldihydroresorcinol) derivative (17). In all cases this aldehyde fraction was found to be largely formaldehyde; this was determined by melting point and by digestion with acetic acid, which permits the separation of the methone derivative of formaldehyde from those of other aldehydes (17). Miscellaneous analytical procedures used on occasion for approximating aldehyde content, or as a qualitative check for acetone, comprised the bisulfite titration ( 7 ) and the acetone-onitrobenzaldehyde reaction to yield indigo (1). Total olefins were determined throughout most of this work by the catalytic hydrogenation procedure (16) rather than the
I N D U S T R I A L A N D E.N G I N E E R I N G C H E M I S T R Y
214
TABLEI. ANALYSISOF ISO-OCTANE-OXYGEN REACTIONMIXTURES" Number 1 2 a 4 6 6 7
Reaction Conditions Temperature C . 550 550 550 550 0.04 0.08 0.16 Contact tim;. sec. 0.24 Analyses b Hydrogen peroxide 0.51 1.6 5.5 7.4 Other peroxides 0.05 0.16 0.23 0.28 Olefin total 0.8 3.5 14 25 0.8 1.8 7 9 Olefin (Ca) 1.1 3.5 6.3 Carbonyl totalc Trace Formaldehyde Trace 0.6 1.3 2.1 Acid 0.33 a Moles h drocarbon : moles oxygen = 0.05. 6 Millimogs product per mole iso-octane passed through reactior. tube. 0 Assuming average molecular weight of carbonyl compounds -44.
....
....
....
450 0.24
450 0.68
350 0.24
89.0 3.0 150 64 254 97 48
0.5 -0.5 2 -2 Trace Trace
14.7 1.4 46 26 35 12 6.5
Trace Trace
bromate-bromide titration. This was necessary, because bromine substitution is an appreciable factor with the more highly branched hydrocarbons. Hydrogenation of the product from isobutane was not attempted, however, because of its extreme volatility; in this case a cold iso-octane solution of the hydrocarbon fraction was titrated with a freshly standardized solution of bromine in carbon tetrachloride.
25
r----l
0
.05
.IO
.I5
.20
.25
CONTACT T I M E IN SECONDS
Figure 2. Compositionof Series of Iso-octaneOxygen Reaction Mixtures (550" C.) 1. Total olefin CSolefin
2. 3.
Hydrogen peroxide
4. Total carbonyl 5. Formaldehyde 6. Acid
Identification of the olefin fraction was attempted only with the iso-octane oxidation products. In this case, isobutylene and diisobutylene were found to be the major olefin constituents. The identity of other olefins which may be present was not established, but it is probable that there are traces in the C? range. In addition, with one of the earlier iso-octane oxidations, a polymeric material was observed about the exit end of the tube and on the walls of the receiver. The reactor tube in this case extended some 6 inches below the end of the furnace, and apparently this polymeric material resulted from further reaction of the isobutylene and radical fragments before passing into the receivers cooled by dry ice-acetone. In subsequent reactions, the reactor tube extended only about 2 inches below the end of the furnace; the formation of this polymeric material has not been observed under these conditions. Isobutylene and diisobutylene were identified in one of the typical iso-octane reaction products (obtained a t 550' C.. contact time of 0.23 second) as follows: The temperature of the reaction mixture waa allowed to rise from -70" C. to room temperature; a small amount of volatile
....
material which was distilled from the mixture was collected in a cold trap. Water was added to dissolve any carbonyl compound, and the volatile material allowed to distill into a second trap containing bromine. Two samples of pure isobutylene also were brominated: (1) with an aqueous solution of bromine, (2) with pure bromine. Ref r a c t i v e indices taken on these three samples indicated that the olefin fraction contains an appreciable quantity of isobutylene.
5
550 0.68
....
....
.... .
.
.
Vol. 46, No. 1
I
....
Dibromides
n
Isobutylene (1) with bromine water (2) with uure bromine Sample .
1.5158 1.5157 1.6154
The reaction mixture was fractionated, and i t was observed that the fraction coming over a t the boiling point of iso-octane contained a definite olefin concentration (by bromine titration). Dry hydrogen bromide was passed into the remaining portion of this fraction. The iso-octane was removed by a simple vacuum distillation, leaving a small amount of a bromo derivative. This derivative was compared with a derivative obtained in a similar fashion from a known diisobutylene-iso-octane mixture. The refractive index of the standard was 1.4550 and for the sample 1.4565. The conclusion made on the basis of these data is that one or more Cs olefins are present in the reaction mixture, and that this Cs olefin fraction contains the components of diisobutylene. Acids were determined by titration with 0.01 or 0.lN sodium hydroxide, using phenolphthalein indicator. ISO-OCTANE. Table I and Figure 2 summarize the more significant data obtained on the oxidation of iso-octane a t several temperatures and contact times, and a t a single molar ratio of oxygen to hydrocarbon of 12 to 1. The major reaction products a t very short contact times are hydrogen peroxide, isobutylene, and diisobutylene. The small diameter of the reaction tube is undoubtedly as important a factor leading to this specificity of reaction as is the short contact time. Thus, the greatly enhanced wall effect under these conditions would be expected to suppress chain-branching reactions effectively and prevent their ascendancy over the simpler process leading to hydrogen peroxide and olefins. These data might suggest that the major initial raactions under these conditions comprise the following:
C C-
b-C-C-C
A h
+ Or+
C-
c
-C-C-C
h
+ .OOH
C
C
I C-C-C-C-C 1
8A
I
+ .OOH
+
I I
C-C-C-C-C
c
c +
c-
c
hI -c-C=C+N. I c
C
c-A-c-c-c c1 c!
+ HOOH
C
C
I c-c-c-c-c I I c c
!
c
C
+
c- h . + c-c=c I h c
This mechanism starts with an attack a t the tertiary hydrogen atom of isc-octane, just as is commonly proposed. There are
F
January 1954
INDUSTRIAL AND ENGINEERING CHEMISTRY
215
three reasons why the authors wish to propose an alternative mechanism to account for these and subsequent oxidation products: (1) the presence of seventeen 1" and 2" hydrogens in the molecule, (2) the known shielding effect of nine hydrogen atoms on a position which is five atoms removed (here, the 3" H) ( I S ) , and ( 3 ) failure to isolate any appreciable quantity of hydroperoxide. From a compilation of the available data on the combustion of paraffins, Boord et a2. ( 2 ) have shown that the approximate relative reactivity of 1', 2 O, and 3' C-H bonds is 1 :3 . 6 :6.9. The total reactivity of the various C-H bonds in iso-octane then could be estimated as follows:
shown to supplement the amounts of major reaction products obtained. The above conclusions are admittedly tentative, and of necessity based on considerable speculation a.e well as fact. However, a t this stage of the work, they appear to fit the data a t hand better than any other. Despite the initial dominance of the reactions leading to hydrogen peroxide and olefins, the formation of carbonyl compounds is not completely suppressed even in the earliest stages of reaction, This indicates that eome of the iso-octyl (and Ledbutyl) radicals undergo oxidation as well as cleavage and dehydrogenation. I t is significant that formaldehyde has been identified as a major component of the carbonyl products even in the very earliest stages. In the overI11 IV all reaction, it is also signifiI I1 cant that, while formaldehyde C C C c and acetone are the major carc-6-c-q-c c-+-+c c-4- -c-c-c. c-A-c-c-c bony l compounds, chromatoc c c c (5 g r a p h i c s e p a r a t i o n of t h e No. of C-H bonds 1 2 6 9 phenylhydrazones has shown 7 3.6 1 1 Reactivity of C-H bond 7 7 Total reactivity of bonds 6 9 the presence of small amounte of other such products. This tabulation would predict approximately the same con-4s the contact time increases from 0.25 to 0.68 second (550' C.) tribution of each of the four radicals listed to the total oxidation the formation of hydrogen peroxide and olefin no longer domiproducts. The general agreement of these relative reactivity nates the reaction. This abrupt rise in the amounts of carratios throughout the field of free-radical chemistry is too conbonyl compounds indicates that a larger number of the simple sistent to permit formulation of a combustion mechanism based initial radicals can no longer escape reaction with oxygen molesolely on attack at the 3' C-H bond. cules, and accordingly, proceed through a series of reactions such The above reasoning would predict that only about 25% of as those in the tabulation above. Finally, the formation of conthe reaction products arise from radical I. However, another siderable quantities of acidic products shows a third series of aspect of iso-octane's structure should be considered. Newman reactions now coming into prominence, presumably by the ( I S ) points out numerous examples where the umbrella effect of oxidation of carbonyl compounds to acids. [Garner and Petty a tert-butyl group markedly inhibits reaction a t the sixth position ( B ) , however, suggest that acids may be formed without carbonyl from the nine hydrogen atoms of the krt-butyl group. While intermediates. ] At these longer contact times, the large amounts this "rule of six" was formulated most specifically for addition of olefins now present in the reaction mixture may be expected reactions, it would appear a general phenomenon that "the to participate in further reactions and thereby contribute to the greater the number of atoms in the six position, the greater the complexity of the product. R-HEPTANE.Results obtained from the oxidation of m-Rep steric hindrance.', The authors therefore are led to propose that attack a t the tertiary C-H bond in iso-octane actually tane are summarized in Table 11. The much greater ease of accounts for much less than the 2570 of the reaction which would oxidative attack and much greater difficulty of reaction control be predicted on a normal reactivity basis. h similar explanacompared with iso-octane are apparent from a comparison of tion for the lack of reactivity of iso-octane in liquid phase oxidaTables I and 11. Thus, iso-octane in 0.08-second contact time tion has recently been suggested (18). This would explain failat 550' C. gives less reaction than does n-heptane in 0.04-second ure to detect appreciable quantities of organic peroxide, which contact time a t 350' C. Also, raising the reaction temperature 50' with n-heptane results in a most marked increase in extent would be expected if reaction occurred at the tertiary C-H bond. If the above is true, then attack a t the other positions Radical I11 IV in the iso-octane molecule I5 must be able to account for *I F the major products obtained: c-d-6-c-c c--oc--oc. c-d-c-c-cl isobutylene, diisobutylene, formaldehyde, and acetone. Ca olefinic products C None This is believed to be feasible C+cT on the basis of the tabulation presented herewith. *Most of the diisobutylene C6 olefinic products (directly None None C from initial radical) would be expected to arise (May get isobutylene if odd electron shifta to terc=b 4- c & -. cfrom radical 11,and probably tiary position) somewhat less from radical c=c-c 111; the major amount of isobutylene would come from radical IV. Less directly, all HCHO ECHO + Some of possible oxygen-containC 0 C three radicals can give rise Q ing products from Cs radical I II t o isobutylene, formaldehyde, (through ROO.) C-b-0. HC-C-C c-c-c-C-Q. .o-6-c-c-c I I and acetone (for possible reC action mechanisms, see 9). Some of possible oxygen-containCHaCOCHs I In addition, other products ing products from subsequent radicah CHs. 06-C. + CHaCHO CHaCOCHi f .C-C-C are obtained in smaller Y amounts. Any r a d i c a l I etc. formed can, of course, be etc. stc
C ! b
I
s
A &
& c :
&
A
d-TC1
"P
"
+
"4
+
'
P
b
t., b J.
1A
INDUSTRIAL AND ENGINEERING CHEMISTRY
216
of reaction, the aldehyde fraction, for example, increasing more than 100-fold. All these results are consistent with the view (2) that all the secondary C-H bonds in a normal paraffin are attacked with substantially the same ease, and that once initiated rapid chain-branching reaction ensues. After these threshold conditions are passed (Reaction 2, Table 11), changes in the extent and type of reaction over the next 100’ temperature rise are relatively slight.
....
-
The deep-seated nature of the degradation involved from almost the very outset of n-heptane oxidation (in contrast to that of iso-octane) is most apparent from the formation of large amounts of carbonyl compounds. These were a much more complex mixture than those obtained from iso-octane, and were considerably more difficult to isolate quantitatively. Thus, extraction of the hydrocarbon phase with water failed to remove all the carbonyl compounds, suggesting the presence of ketones and aldehydes containing fouE or more carbon atoms (6). Under conditions of Reaction 5, Table 11,the ratio of aldehyde to ketone was indicated to be approximately 5 to 2 by determining total carbonyl as the phenylhydrazones and total aldehyde as the methone derivatives. Treatment of the methone derivatives of the aldehydes with acetic acid ( 1 7 ) gave a molar ratio of formaldehyde to other aldehydes of approximately 6 to 1. Chromatographic separation of the dinitrophenylhydrazones of the water-soluble carbonyl compounds gave a molar ratio of formaldehyde to total carbonyl of 5 to 1. While it is not possible to relate these ratios directly, all evidence points to the formation of formaldehyde as the major carbonyl product.
TABLE111. ANALYSISO F
2,2,8-TRIMETHYLHEXANE-OXYGES
REACTION MIXTURES~
Reaction Number
1
2
Conditions 450 420 Temperature C. 0.04 0.08 Contact timi. see. Analyses b 6.8 0.03 Hydrogen peroxide 6.7 2.0 Other peroxides 16 0.4 Olefin 12 Trace Total carbon 1 -2.1 None FormaldehyA 3.7 None Acid moles oxygen = 0.07. a Moles hydrocarbon b Millimoles product’ per mole 2,2,5-trimethylhexane passed through reaction tube.
The dominance of organic peroxides over hydrogen peroxide a t 350’, and the marked increase in hydrogen peroxide formation a t higher temperatures, are believed significant indications as to the reaction mechanisms involved. At 350°, the reactivity of the oxygen molecule is slight, so that the amount of hydrogen peroxide formed by the initiation reactions ( l a and lb) is relatively small:
RCHR’
+ .OOH
(la)
+ HOOH
(1b)
+
RCH,R~ .OOH -+ RCHR’
On the other hand, the stability of the RCHR’ and RCH( OO.)R’ radicals is adequate to permit fairly efficient chain propagation, Equations 2 and 3:
T.4BIJ3 11. -4NALYSIS O F n-HEPTANE-OXYGEN REACTION MIXTURES~ Reaction Number 1 2 3 4 5 Conditions Temperature,°C. 350 400 450 500 350 0 04 0.04 0.04 0.04 0.08 Contact time, see. Analysesb Hydrogen peroxide 0.3 63 79 86 37 9.1 11.3 2.1 1.0 10.4 Other peroxides 58 78 93 -70 Olefin 2.9 Total carbonyl .... .... -360 Aldehydes0 2 13 360 360 sio’ 250 Acid 4.0 57 60 40 34 a Moles hydrocarbon : moles oxygen 0.09. Ir Millimoles product er mole n-heptane passed through reaction tube. C Obtained with metgone reagent, assuming derivative mixture t o have approximately molecular weight of formaldehyde derivative (292).
-
+- o2
RCH~R’
Vol. 46, No. 1
0 0
+
R ~ H R I 0, .+
RCHR’ H
b
0
0
RCHR’
+ RCH~R’
-+
0
RCHR’
+R~HR’
(3 )
From 400’ to 500’ C. the reactivity of oxygen molecule has increased sufficiently (see 11) to permit the formation of hydrogen peroxide in much larger amounts (Equations l a and lb). Since, at the same time, the thermal stability of the RCHR’ and RCH(OO.)R’ radicals will have decreased, Reactions 2 and 3 leading to organic peroxide now are of relatively minor importance because of the ascendancy of the decomposition reactions (4 and 5 ) :
RCHCH~R’-+ RCH = CHR’
0 0 RCHCHtR’
-+
1 7 + i
+ H.
+
R-C-CHgR’ .OH {RCHO R’CHZO.
J.
R‘ .
(4)
(5)
+ CHgO
This view of the initial stages of the reaction is consistent not only with the data of Table 11, but also with the more recent concepts of hydrocarbon oxidation (9). Exploratory reactions on two additional hydrocarbons, isobutane and 2,2,5-trimethylhexane, lend support to the above generalizations. 2,2,6-TRIYETHYLHEXANE. The oxidation Of 2,2,&trirnethylhexane is of particular interest for comparison with that of isooctane. As expected, this hydrocarbon lies between iso-octane and n-heptane in ease of oxidative attack. As shown in Table 111, it is less oxidized in 0.04 second at 450” than is n-heptane at 350” C. 2,2,5-Trimethylhexane resembles n-heptane in that organic peroxide is dominant over hydrogen peroxide in the initial reaction products. On the other hand, it resembles iso-octane in that the yield of olefins exceeds the yield of carbonyl compounds through an appreciable period of the initial reaction (compare Reaction 3, Table I, Reaction 2, Table 111, and Reactions 1 and 2, Table 11). Like iso-octane, isobutane yields mainly hydroISOBUTAXE. gen peroxide and olefin in the early stages of reaction (Table IV). This is consistent with the structural similarities of these two hydrocarbons, and with the fact that both require high temperatures for appreciable oxidative attack. These similarities are apparent from a comparison of Reactions 2 and 3 in Table I with Reactions 1 and 2 in Table IV. While the data on bobutane are not so complete as with iso-octane, certain differences also are indicated. One of these is the somewhat greater ease of oxidative attack of iso-octane. which is in line with its higher molecular weight and, accordingly, greater number of C-H bonds. Another is the higher ratio of hydrogen peroxide t o organic peroxide obtained with iso-octane.
.
INDUSTRIAL AND ENGINEERING CHEMISTRY
January 1954
TABLE IT. ANALTSIS
OF
ISOBUTANE-OXYGEN REACTION
RIIX-
TURES“
Reaction Number 1 2 Conditions 550 550 Temperature. ’ C. 0.16 0.08 Contact time, sec. Analyses a 1.2 3.6 Hydrogen peroxide 0.6 0.8 Other ueroxides Olefin-2.5 -5.6 Total carbon -0 2 -0.8 Forrnaldehyc$J -0 1 -0.4 Scid .... .... 0 Moles hydrocarbon : moles oxygen = 0.15. b Millimoles p r o d u c t per iirolr iyobutane passed through reaction tube TABLE
1T:mp., C.
IT. HYDROGEX PEROXIDE-ROOHRATIO 0.5
1.O
2.0
C CACCC
Ad A
CCC
I1202,fROOH Ratio
24
..
..
...
., .
.. ..
.. ..
...
..
..
,
.
...
, .
.
10
450
,,
.
..
1.6
.,.
580
,..
..
2
4
450
0.015
..
,
1.2
.
, , ,
, ,
’
100
.
560
,,
.
Extent of Reactiona IO 20 50
5.0
,
10
C 2ACCCC
. l bL
b
500 ., .. ... ... 350 , , . .. ,, ,,, Indicated b y 8um of olefin, carbonyl, and acid.
CCCCCCC 0
.
.,
.
0:633
HYDROGEN PEROXIDE FORRIATIOX. Observations regarding hydrogen peroxide formation are not only of theoretical interest. but also of potential commercial importance. For this reason, the data on hydrogen peroxide formation from all four hydrocarbons studied are listed in Table V at several temperatures and extents of reaction. MJhile the information is not directly comparable, a sufficient number of points are available to indicate certain trends. These trends fit in with the general concepts already outlined regarding the effect of reaction temperature on the facility with which the oxygen molecule can attack the hydrocarbon directly, and on the stability of the alkyl radicals formed. Thus, hydrogen peroxide is obtained most efficiently-Le., in highest percentage yield-from iso-octane, apparently because it combines difficulty of oxidative attack with poor thermal stability of the alkyl radicals initially formed. Isobutane is slightly more difficult to attack, but givee slightly more organic peroxide, apparently because of a somewhat greater stability of the alkyl radicals formed, .4ccordingly, a t optimum conversions-Le., percentage less than 20% of the hydrocarbon reacting-the yield of hydrogen peroxide from the four hydrocarbons investigated lies in the order of iso-octane > isobutane > 2,2,5-trimethylhexane > n-heptane. SUMiMARY OF RESULTS
Analysis of the products obtained by the vapor phase oxidation of iso-octane. n-heptane, 2,2,5-trimethylhexane, and isobutane a t temperatures of 350” to 550’ C. and contact times of 0.04 to 0.68 second, and in a ’/is-inch diameter reactor has led to the following conclusions regarding the initial reaction processes. 1. If a hydrocarbon is attacked a t temperatures when the alk 1 and peroxy radicals have adequate thermal stability, the hysoperoxide can be obtained as the major product (Reaction 1, Table 11). 2. If the hydrocarbon is attacked at higher temperatures where the peroxy radical is very unstable and the alkyl is still fair1 stable, carbonyl compounds can be obtained as the major y-od;ob, with olefins of secondary importance (Reactions 2 to 5, able 11).
217
3. If the hydrocarbon is attacked at temperatures so high that decomposition of the alkyl radical tends to occur before the peroxy radical can form, olefins and hydrogen peroxide will comprise the major products (Reactions 1 to 4, Table I). 4. The structure of the hydrocarbon influences not only the ease of oxidative attack but also the thermal stability of the alkyl and peroxy radicals formed. Aecordingly, the extent to which the initial reaction follows course 1, 2, or 3 will depend upon structural factors as well as temperature. 6 . Major products obtained a t successive stages in the oxidation of iso-octane comprise hydrogen peroxide and olefins, and formaldehyde and acetone. Acids come into prominence only after relatively large amounts of carbonyl compounds have formed. 6. The failure of iso-octane to yield appreciable quantities of organic peroxide, together with other considerations. leads to the oroaosal that the initial oxidative‘attack occurs fargely a t the 200 500 primary and secondary C-H bonds rather than a t the tertiary C-H bond. , .. 30 7 . Major products obtained at successive stages in the oxidation of n-heptane comprise ( a ) or.. . .. . ganic peroxides and ( b ) carbonyl compounds ,.. ,. , (mrtinly formaldehyde). Stage ( b ) develops with much great.er ease and rapidity than in the case of iso-octane. 6. In resistance to oxidative attack isobu... .. , tane > iso-octane 2,2,5-trimethylhexane > n-heptane. 9. I n line with general conclusions 1 to 4, the potential efficiency of hydrogen peroxide for3’7 mation from these hydrocarbons is indicated to be: iso-octane > isobutane > 2,2,5-trimethylhexane > n-heptane. ACKNOWLEDGMENT
This work was done under National Advisory Committee for Aeronautics Contract NAw-6116; the authors are grateful to the rational Advisory Committee for Aeronautics for the sponsorship of this research and for permission to publish these results, LITERATURE CITED
Adanis, C. A., and Nicholls, J. R., Analyst, 54, 2 (1929). Boord, C. E., Greenlee, K. W., and Derfer, J. M , Divisions of Petroleum and Gas and Fuel Chemistry, Symposium on Conibustion Chemistry, 119th Meeting, AM. CHEM.SOC., Cleveland, Ohio, 1951, p. 171. Eisenberg, G. M., IND.ENO.CHEM.,ANAL.ED., 15, 327 (1943). Frank, C. E., and Blaokham, A. U., IND.ENO.CHEM.,44, 862 (1952). Frank, C. E., and Swarts, D. E., unpublished information. 47, 877 Garner, G. H., and Petty, D. S.,Trans. Faraday SOC., (1951). Goldman, F. H., and Yagoda, H., IND. ENC.CHEM.,ANAL.ED., 15,377 (1943). Kahler, E. J., Bearse, A. E., and Stoner, G. G., IKD.EXO.CHEM., 43,2777 (1951). Lewis, B., and Von Elbe, G., Divisions of Petroleum and Gas and Fuel Chemistry, Symposium on Combustion Chemistry, 119th Meeting, AM. CHEM.SOC., Cleveland, Ohio, 1951,p. 47. Lewis, B., and Von Elbe, G., “Combustion, Flames and Explosions of Gases,” pp. 172-86, New Pork, Academic Press, 1951. in press. Malmberg, Earl, et al., J . Am. Ckem. SOC., Minkoff, G. J., Discussions Faraday Soc., 2, 147 (1947). Newman, M. S., J . Am. Chem. Soc., 72,4783 (1950). Roberts, J. D., and Green, C., Anal. Chem., 18, 335 (1946). Shriner, R. L., and Fuson, R. C., “Systematic Identification of Organic Compounds,” p. 97, New York, John Wiley & Sons, 1948. Siggia, S., “Quantitative Organic Analysis via Functional Groups,” p. 37, New York, John Wiley & Sons, 1949. Walker, J. F., “Formaldehyde,” A.C.S. Monograph 98, pp. 2646, New York, Reinhold Publishing Gorp., 1944. (18) Wibaut, J. P., and Strand, A., Proc. Koninkl. Ned. A k a d . Wetenschap., 54B,229 (1951). RECEIVEDfor review M a y 11, 1953.
ACCEPTED September 15, 1953.
