INDUSTRIAL AND E N G I N E E R I N G CHEMISTRY
November 1949
0
I/ + H,O +R-C-OH
*
+ HBr
(24)
All of these steps, excepting reaction 24 which simply represents the hydrolysis of an acyl bromide, have been proposed previously in the discussion. Verification of the formation of acyl bromide was obtained from a liquid phase oxidation of benzyl bromide wherein benzoyl bromide was isolated as an intermediate ( I ) . All of the equations, 11 through 24, dealing with the oxidations of primary carbon-hydrogen bonds may be summarized by the following equation: H ~~
I
2R-C-H
I
0
HBr
+ 302 +2R-
e
+ 2H20
-OH
2-phenyl-2-hydroperoxypropane is, by analogy with isobutane, to be expected from the oxidation of cumene (2-phenylpropane); actually phenol and acetone are obtained, probably by degradation of the predicted product. Similar cleavages have been revealed by the investigations on other hydrocarbons. I n support of these mechanisms of hydrogen bromide-catalyzed oxidations, a t least one example of every postulated nonfree radical intermediate or end product has been isolated or identified a t some point during these studies. These compounds acid, are: hydroperoxide, ROOH; carbonylic, R,R,C=O; RCOOH; alcohol, ROH; acid bromide, RBrC=O; and bromide, RBr. ACKNOWLEDGMENT
The authors wish to thank A. E. Lacomble, chairman of the board of Shell Development Company, for his encouragement and criticism of their efforts. To many of their colleagues in the company they are indebted for the benefit of helpful discussions.
(25)
H and it is seen that under ideal conditions there would be no loss of hydrogen bromide which then would function strictly as a catalyst. I n actual operation this condition has been closely approximated. The formation of by-product alkyl bromides, particularly wmonobromides, is not harmful because, as shown, these compounds themselves are oxidized to acid. The fifth paper in this series covers the oxidations of some aromatic compounds; i t also briefly treats some liquid phase oxidations. Because of greater structural complexities, the oxidations discussed therein are less clean-cut than those of the simpler aliphatic hydrocarbons. However, the processes apparently conform to the theories developed for the simpler cases. Thus, toluene, by reason of its primary carbon-hydrogen bonds, is converted in good yield to benzoic acid, while ethyl benzene with its methylenic group gives the anticipated acetophenone. Further, the studies on the aromatic compounds reveal a general reaction, peroxide degradation. As an example of this, the hydroperoxide,
2597
LITERATURE CITED
Barnett, B., Bell, E. R., Dickey, F. H., Rust, F. F., and Vsughan W. E.. IND. ENQ.CHEM..4 1 . 2 6 1 2 (1949). Bell, E. R., Dickey, F. H.,‘ Raley, J. ‘H., Rust, F. F., and Vaughan, W. E., Ibid., p. 2597. Bell, E. R., Irish, G. E., Raley, J. H., Rust, F. F., and Vaughan, W. E., Ibid., p. 2609.
Harris, E. J., Proc. Roy. SOC.(London), 173A, 126 (1936). Jost, W., “Explosions- und Verbrennungsvorganpe - in Gasen,” pp. 416-21, Berlin, Julius Springer, 1939. Milas, N. A., and Surgenor, D. M., J . Am. Chem. Soc., 68, 205 (1946).
Nawrocki, P. J., Raley, J. H., Rust, F. F., and Vaughan, W. E., IND. ENG.CHEM.,4 1 , 2 6 0 4 (1949).
Reiche, A., “Alkylperoxyde und Ozonide,” Leipzig, Steinkopff, 1931.
Rust, F. F., and Vaughan, W. E., J . Org. Chem., 7, 491 (1942). Vaughan, W. E., and Rult, F. F., U. S. Patent 2,403,771 (July 9, 1946). RECEIVED August 17, 1948. Presented in part a t the meeting of the Gordon Research Conferences of the American Association for the Advancement of Soienoe, Colby Junior College, New London, N. H., June 1948.
(Oxidation of Hydrocarbons Catalyzed by Hydrogen Bromide)
OXIDATION OF BRANCHED-CHAIN COMPOUNDS E. R. BELL, F. H. DICKEY, J. H. RALEY, F. F. RUST, AND W. E. VAUGHAN Shell Development Company, Emeryville, Calif. a
T h e hydrogen bromide-catalyzed oxidation of isobutane and related branched-chain compounds gives high yields of organic peroxides as primary products. The mechanism of the reaction appears to be a free radical chain, in which the catalyst functions both in initiation and in propagation.
H
YDROGEN bromide-catalyzed oxidation of the lower hydrocarbons (1%)is unique in its high degree of selectivity and its low order of carbon-carbon bond rupture. The oxidation of lower branched-chain hydrocarbons (19)is of particular interest because i t represents the first isolation in major amount of organic peroxides formed by the reaction of paraffin hydrocarbons with oxygen. The mechanism of this peroxide synthesis, as well as of some of the side reactions, is suggestive of steps occurring during noncatalyzed hydrocarbon oxidation.
EXPERIMENTAL
APPARATUS.The hydrocarbon oxidations summarized in the first paper have for the most part been carried out in a Pyrex glass coil reactor of 25-mm. inside diameter and 2940-m1. capacity, immersed in a thermostated oil bath regulated to =t0.5’ C. The reactants, if gases, were metered through ordinary differential flowmeters. I n other cases liquid reactants flowing from a constant-pressured reservoir passed through a rotameter and a glass regulating stopcock into a vaporizing unit. To minimize explosion hazards, oxygen was introduced through one arm and the other gases through the other arm of a W-shaped preheater. Mixing occurred in the middle ascending tube and the mixture was then introduced directly into the top of the coil reactor. The effluent material was collected in various ways, depending upon the nature of the products and other materials to be handled. The only modification in the reactor system for operation at pressures other than atmospheric was the installation on the exit line of a commercial regulator loaded for from 5- t o 50pound gage pressure. I n the series of experiments where reduced
I N D U S T R I A L AND E N G I N E E R I N G CHEMISTRY
2598
% tiBr i n Feed
Figure 1.
Isobutane Oxidation, Effect of
Hydrogen Bromide Concentration o n Oxygen Consumption
Total flow, 608 cc./min.; oxygenlisobutane
-
1
pressure was desired, the low pressure side of the regulator was connected to a vacuum pump. I n certain experiments the reaction mixture was sampled a t various points along it's path by means of side tubulations sealed onto one of the reactor coils. These ports also could be utilized for injection of additional reactants, or catalyst, or certain sensitizing agents. Smaller scale exp1orator.v experiments mere carried out in an all-glass, vapor-jacketed coil of 15-mm. inside diameter tubing and approximately 450-ml. capacity. A static system was used for certain exploratory work arid in experimenh designed to study the influence of surface. The reactor, which was connected t,o a mercury manometer, was a cylindrical vessel of 2.5-cm. inside diameter and 144-ml. capacity (or 4.5-cm. inside diameter and 400-ml. capacity) surrounded by a vapor jacket in which suitable liquids could be refluxed. Through stopcocks and connecting lines t,he reactor could be evacuated either by a mercury diffusion pump or, when a gas sample was desired, by a 200-ml. Toepler pump. A trap cooled by dry ice or liquid nitrogen was connected in series with the latter pump in order to remove condensable material. During certain photochemical studies the static apparatus was modified by insertion of a 1-em. inside diameter clear quartz well connected to the Pyrex glass by a graded seal. A small quartz mercury arc lamp was placcd in this well. REACTASTS.The oxygen R-RS taken from commercial cylinders. The organic chemicals were all from commercial sources and in some instances were further purified by careful distillation. CATALYST,The hydrogen bromide catalyst was prepared by passing a gaseous mixture of the elements containing a slight excess of the halogen through a quartz tube heated to redness. The unreacted bromine was removed by preferential adsorption on activated charcoal, and the hydrogen bromide after passage through calcium chloride mas condensed in shinless steel (18-8) cylinders chilled in dry ice. A typical analysis of gas withdrawn from one of the vessels was 99.4% hydrogen bromide, 0.35% nitrogen, 0.22y0hydrogen, and 0.03% oxygen. OXIDATION OF ISOBUTANE
The oxidation of isobutane is in many ways the simplest reaction to be discussed in this series. The important products are tertbutyl hydroperoxide, di-tert-butyl peroxide, and tert-butyl alcohol. High conversions (85 to 90% per pass), low operating temperatures (155" to 165' C.), a small amount of catalyst ( 2 to 4% hydrogen bromide), and relatively low quantities of degradation products are salient operating characteristics. (In the absence of the catalyst no detectable oxidation occurs under the operating conditions.) Increase in pressure is particularly beneficial; it results in increased conversion, a reduction in catalyst requirements, and retardation of degradation reactions leading to oxides of carbon. Operation at higher pressures also tends to offset inhibiting action of the surface of the reactor.
Vol. 41, No. 11
Experimental work has been principally concerned with the reaction of equimolar proportions of oxygen and hydrocarbon. Higher oxygen concentrations have been arbitrarily eliminated in deference to the increased explosion hazard and, more importantly, the two reagents under good operating conditions are consumed in almost stoichiometric ( 1 to 1)proportions. An appraisal of the reaction can be macle from balances of the reactants such as are presented in Tables I and 11. These data were obtained with 4 and 20% catalyst input, respectively, and thus illustrate the effect of catalyst on product composition. Of interest are the yields of the principal products, the two peroxides and the alcohol, in relationship to one another. The higher catalyst concentration (20%) results in a greater yield of both di-tert-butyl peroxide and tert-butyl alcohol as compared to that of hydroperoxide. At the higher hydrogen bromide concentration the yield of by-products, especially bromo compounds, is increased. Analytical Procedure. The crude, two-phase liquid product was collected in a mixture with water and the effluent gases and vapors partially condensed in traps cooled with dry ice. The uncondensable gases were collected in an aspirator bottle for Orsat analysis. For the purpose of material balances the products were determined as follows: Repeated extraction with water effectively separated the ditert-butyl peroxide and alkyl bromides from the water-soluble tert-butyl hydroperoxide and tert-butyl alcohol. When the dialkyl peroxide contained substantial quantities of water-insoluble impurities, a further purification was obtained by azeotropic distillation with added tert-butyl alcohol and water. (The ternary, which contains 44.070 peroxide, 49.3% alcohol, and 6.7% water, boils at 77" C.) Purification of the peroxide was achieved by washing the distillate with water and 30% sulfuric acid. The amount of tert-butyl hydroperoxide was determined by iodometric titration. Since bromoacetone also oxidizes acidified potassium iodide, the amount of this halide was deducted from the indicated amount of peroxide. Ketone was determined by means of hydroxylamine hydrochloride. If acetone, hydroperoxide, and bromoacetone were
TABLE I. ISOBUTANE OXIDATION AT 163" C.a [Conditions: flows (oo./min.) b : iso-C4HioC, 288: 0 2 , 288; HBr, 24. Threeliter reactor] Mol&/100 Moles Consumed of Products Iso-C4Hio 0 2 HBr twt-Butyl hydroperoxide 69.5 70.3 ... Di-tert-butyl peroxide 6.0 8.1 ... tert-Butyl alcohol 9.8 9.7 Alkyl bromides (est.) 4.7 ibb dcetone Trace Trace ... Carbon dioxide 0.9 0.9 Carbon monoxide 1.2 1.2 ... Water 24.0 ... a 48% hydrogen bromide recovered, 887c isohiitane consumed, 87% oxygen consumed, 3.37, input isobutane unaccounted for, and 3.5% input oxygen unaccounted for. b I n all statements of flow rates in this and the succecdine papers, cc./ min. means cubic centiMeters of vapor per minute at room tomperalure and the operating pressure. 0 Pupity, 9cJ.77c.
...
...
TABLE11. ISOBUTANE OXIDATION AT 158" C." [Conditions: flows (cc./min.): iso-CdHio, 240; 0 2 , 240; HBr, 1201 Moles/100 Moles Consumed of IsO-C4Hio 02 HBr Products tert-Butyl hydroperoxide 13.9 13.5 ... Di-tart-butyl peroxide 20.0 19.5 ... tert-Butyl alcohol 31.4 30.5 ... tert-Butyl bromide 5.0 11.2 Dibromoisobutane 6.1 ... 13.7 Acetone 1.1 1.1 ... Bromoaoetone 1.9 1.9 4.3 Carbon monoxide 1.4 1.4 ... Carbon dioxide 0.6 0.5 ... 0.6 ... Isobutylene Ethane 4.4 ... ... 79.6 ... Water Bromine 20.5 23% hydrogen bromide recovered, 92% isobutane consumed, 91% oxygen consumed, 0% input isobutane unaccounted for, 9% input oxygen unaccounted for, and 13% input hydrogen bromide unaccounted for.
...
...
... ...
...
INDUSTRIAL AND ENGINEERING CHEMISTRY
November 1949
*
1
present, the total ketone and total oxidizing power were measured, the acetone distilled off, and the residual bromoacetone determined. this procedure supplied all the data necessary for an estimate of the proportions of the three components. (tertButyl hydroperoxide interferes somewhat with the ketone analysis but the effects are minimized if the peroxide solution is dilute.) The tert-butyl alcohol was concentrated by distillation of the aqueous extract and was treated with concentrated hydrochloric acid to convert the alcohol to tert-butyl chloride. Measurement of the chloride phase then provided a rough estimate of the original alcohol. The alkyl bromide determinations were the most uncertain of all. I n those cases where the in u t of hydrogen bromide was of the order of 4%, the alkyl bromife value was assumed equal to the difference between input and recovered hydrogen bromide. Where larger amounts were obtainable, estimates were based on bromine analysis, on separation of mono- and dibromides by microdistillation, and on refractive indexes. Bromine appeared in variable amounts in the product and it was found desirable to eliminate it by introducing small amounts of propylene or other olefin into the effluent product stream.
IOC
I
I
145
150
2599
I --0320tm! !
8C 0
E
6C 0
d
8 4c
2c
C
160
I55
165
170
T,'C.
Identification and Properties of the Peroxide. Di-tert-butyl peroxide has been identified by the following determinations : Actual Theoretical Molecular weight (cryoscopic, benzene) 147 146 6 5 . 0 and 6 5 . 5 85.7 Carbon % 12.3 12.3 Hydrolen % Active ox&en, % (concd. HI at 60' C.) 11.0 11.0 Molar refractivity 43.4 43. la a Computed from derived values for -0-0from known constants of related peroxides. The value used as a mean is 3.96 or 1.98 for the contribution of each peroxide oxygen atom to the molar refractivity (9). The other values used-2.148 for carbon and 1.100 for hydrogen-were taken from reference ( 5 ) .
A comparison of the characteristic physical constants of ditert-butyl peroxide as obtained by the authors with those of Milas and Surgenor ( 6 )is given: Authors' Boiling point, C. Freezing point, O C.
Milas and Surgenor's
111 -40.0
-109-109.2 18
1.3890 0.796
1.3872 0.793
Di-tert-butyl peroxide is discussed a t some length in the literature (9,16). tert-Butyl hydroperoxide has been characterized by the following determinations on material from the reactor analyzing 95% on the basis of active oxygen (iodometric procedure using a 0.1-gram sample in 50 ml. of isopropyl alcohol, 2 ml. of glacial acetic acid, and 2 ml. of saturated aqueous potassium iodide boiled for 2 minutes): Carbon, % Hydrogen, % Active oxygen, % Boiling point (18 mm.),
C.
Actual
Theoretical
54.4 10.8 16.9 37.0
53.3 11.1 17.8 38-38.5
Heat of Reaction. The heat of combustion of tert-butyl hydroperoxide is 654.0 f 0.3 kg.-cal. (15" C.)per mole CdHeOOH (liquid, 25" C.). This datum, together with Rossini's (10) value for isobutane, 686.3,and the calculated heat of vaporization of 11.5, leads to a value of AH2g8 = -21 kg.-cal. for the reaction: CaHio(g)
+ 02(g)
----t
C4HeOOH
(d
The vapor pressure of the hydroperoxide is given by the equation: loglep(mm.) = 9.3441
-
2507.3 .
Figure 2. Isobutane Oxidation, Effect of Pressure and Temperature on Oxygen Consumption Flows (cc./rnin.):
01,288; iso-CdXla, 288; HBr, 24
The results indicate that hydrogen bromide is being lost simultaneously with the formation of oxygenated product. With the exhaustion of the catalyst, the oxidation ceases. PRESSURE. Studies on the effect of pressure covering the range 0.5 t o 2 .atmospheres absolute and temperatures from 143 " to 168" C. are presented graphically in Figure 2. The runs a t 2 atmospheres could not be extended t o lower temperatures probably because of an alteration of the reactor surface by an organic deposit. The effects of the vessel surface on the reaction are discussed in a followingsection. A reduction in the amounts of degradation products-carbon dioxide, carbon monoxide, and olefin-is effected by increased pressure (Table 111). At 2 atmospheres of total pressure, approximately as great a consumption of oxygen (85 t o 87%) has been obtained with 2% hydrogen bromide as was previously obtainable a t 1 atmosphere with a 4% catalyst concentration. Appreciable reaction (62%) is obtainable even a t 158" C. with only 1%hydrogen halide. Mechanisms of Product Formation. The isolation of tertbutyl hydroperoxide as a principal oxidation product is of considerable scientific interest. It appears that this is the first time that a pure organic peroxide has been obtained as a primary oxidation product of a paraffin hydrocarbon. The finding is probably significant to an understanding of the fundamentals underlying combustion processes in general and controlled hydrocarbon oxidations in particular.
+ O2 Br + (HO2.. ....1 ) + Br e (CH&C- + HBr + (CHa)sCOO(CH&C- + (CH&COO- + HBr + (CH&COOH + Br HBr
--3
(1)
0 2
(3)
(CH&CH
(4)
Reaction 1 is similar to that used to explain the initiation of the peroxide-catalyzed "abnormal" addition of hydrogen
TABLE111. EFFECT OF PRESSURE ON ISOBUTAN~ OXIDATION [Conditions: bath temperature: J58' C.: flows (oc./min.): 01,288; isoU H l o , 2 8 8 ; HBr, 241 Per Cent Consumed Oxygen Going to Gaseous By-products Pressure Products 0.5 atm. 1 atm. 1.5 atm. 2 atrn.O 0.0 0.8 0.0 0.3 Carbon dioxide 0.9 1.0 2.5 2.5 Carbon monoxide 1 . 8 0 . 5 0 . 2 ( 5 ) 0.0(2) Olefin 35.5 76.0 85.0 90.5 % ' Oxygen reacted a 3% hydrogen bromide used in this experiment. One molecule of water is assumed to be formed per molecule of olefin. ~
Effects of Operating Variables. TEMPERATURE AND CATALYST CONCENTRATION. The effect of catalyst concentration upon the rate of oxidation is illustrated in Figure 1 for two temperatures, 158" and 165' C. The apparent absence of a temperature coefficient for the reaction is probably not real; rather, the curves are probably influenced b y the rate of conversion of hydrogen bromide t o catalytically inactive alkyl bromides.
(2)
I
2600
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 41, No. 11
mine, resulting from reaction 5 or from other reactions, decomposes hydroperoxide in the liquid phase, Equation 6.
+ ROOH --+ ROH + Brp + H20 ROOH + BrP + HzO +ROH + 2HBr + O2 2HBr
(5) (6)
Di-tert-butyl peroxide (14) is readily formed by condensation of tert-butyl hydroperoxide with tert-butyl alcohol in liquid phase in the presence of strong acids [see also (S)]. Reaction 7 explains the appearance of the ketone and its bromo derivatives among the products of reaction. These thermal decomposition reactions of the peroxides and the reactions of the free radicals produced thereby have been discussed elsewhere (5, '7, 11). C4HsOOCaHs --+ CzHo
-
I 0
I
I
10 cc/mln
20 te:t
30
40
50
Butyl Alcohol Vapor
Figure 3. Isobutarie Oxidation, Retardation of Oxygen Consumption by tert-Butyl Alcohol Flows (cc./min.): 0 2 , 267; iso-CdHlo, 267; HBr, 6 6 ; tert-but,l alcohol, T. Temperature, 158' C.
(e).
bromide to olefins Essentially, it is the chain-initiating step b j which bromine atoms are first formed. The high specificity of the attack of the bromine atom on the tertiary hydrogen atom (Equation 2) is striking; the product balances show that little attack occurs on any of the nine primary hydrogen atoms. The tert-butyl radical associates with oxygen to form a peroxy radical, Equation 3, which saturates its free valence and becomes a stable molecule by the action of hydrogen bromide as a hydrogen donor, Equation 4. By the latter process a bromine atom is generated and the chain continued. Equation 4 is the fundamental reaction which stabilizes the reactive peroxy radicals and which distinguishes hydrogen bromide-catalyzed oxidations from the ordinary hydrocarbon oxidations. Termination could occur by association of radicals in the vapor phase or by their destruction or adsorption a t surfaces. Equations 1 to 4 are thought to be basic for all of the hydrogen bromide-catalyzed oxidations and will be referred t o in the following papers on other types of compounds. Thc formations of fed-butyl alcohol and the greater part of the di-tert-butyl peroxide evidently occur as a result of subsequent rcactions involving the hydroperoxide and hydrogen bromide. This is demonstrated by the following experiments: Steam, te&butyl hydroperoxide, and hydrogen bromide, the latter in slight molar excess, were passed through a clean Pyrex tube a t 168" C. The principal products were alkyl bromides and di-tert-butyl peroxide. There were only minor amounts of aldehyde and tert-butyl alcohol, and apparently no alkyl hydroperoxide remained. tert-Butyl hydroperoxide with steam alone did not decompose noticeably a t this temperature. This pair of experiments indicates that dialkyl peroxide originates in the reactor while only a minor amount of tert-butyl alcohol is obtained under the same conditions. Evidence that the major part of the tert-butyl alcohol must be formed outside the reaction zone is given by the fact that the alcohol adversely affects the oxidation. This was provcd by inclusion of varying amounts of tert-butyl alcohol in the input vapors. In Figure 3 the alcohol input (as vapor) in the reaction mixturc is plottcd against the volume of reacted gases. This latter factor is roughly proportional to the oxygen consumed. These data indicate clearly that an amount of alcohol such as is usually found in the oxidation product would be an effiL'#lent retardant if present in the reaction zone. This does not eliminate the possibility of alcohol formation in the reaction zone but under these conditions the compound is probably converted to dialkpl peroxide a t such a rate that the concentration of the alcohol is never critically high. Although alcohol is formed by the reduction of the hydroperoxide by hydrogen bromide, there is usually more alcohol present in the product than can be accounted for on the basis of available hydrogen bromide. This is due to the fact that bro-
+ 2CHaCOCHa
(7)
Surface Effects. In the flow experiments i t became apparent that the reaction was surfacc-sensitive. This was manifested hp a decrease in conversions and simultaneous accumulation of carbonaceous deposits. Removal by ordinary methods (such as acid treatment or baking) gave variable responses ranging from improvement to permanent deactivation. Study of these effects in the static reactor has clarified this problem. PYREXGLASS SURFACE.A plot of the pressure decrease against the time for a typical experiment in a clean Pyrex vessel is shown by curve 1 of Figure 4. The initially rapid oxidation is seen t o decelerate abruptly when far short of completion and t o be surpassed in rate by some pressure-increasing reaction. So extensive is this retardation that virtually no oxygen is consumed after the maximum in the curve is attained. This behavior persists at higher temperatures (156" C.) and in the presence of nitrogen added in an attempt to reduce the effect of the surface. The pressure increasing reaction may be ascribed to a surface-catalyzed decomposition of tert-butyl hydroperoxide since this compound, which is ordinarily the principal product, is obtained in very low yield. The presence in the reaction product of acetone and an alcohol, known decomposition products of the hydroperoxide, confirms this view. It is the accumulation of organic decomposition products on the reactor wall that ultimately causes reaction to cease. Evacuation of the vessel (at 145" C.) does not entirely remove the wall deposits, as shown bv the lack of appreciable reaction when a fresh mixture is admitted. Even baking in air a t approximately 575 C. does not return the surface to its original condition (curves 2 and 3 of Figure 4),and the amount of reaction becomes smaller after each successive baking. These results simulate the gradual deterioration of the O
10
Time
Imin )
Figure 4. Isobutane Oxidation, Effect of Baking Pretreatment on Pyrex Reactor Reaction temperature, 145a C.; partinl preesurea : iso-C~Hio,140 =t3 mm.; 0 2 , 1 4 0 * 3 mm.; HBr, 20 * 1.5 mm. Reactor baked in air at approximately 575O C. between runs; numerals refer to order of experiments
November 1949
INDUSTRIAL AND ENGINEERING CHEMISTRY
larger scale flow reactors. The concept of surface poisoning by accumulation of degradation products suggested the possibility of coating the Pyrex vessel with a material less active in catalyzing the hydroperoxide decomposition.
2601
initiator. The validity of the latter concept could be demonstrated by providing other sources of reaction centers. I n addition t o boric acid and silica gel, other coating materials were investigated. Certain of these may be grouped, as, they were active for reaction initiation but catalyzed decomposition of the product; in this group are sodium tetraborate, sodium bromide, and titanium dioxide. CHAIN INITIATION AT THE SURFACE. Evidence for surface initiation is afforded by the accelerating effect of ultraviolet light on the oxidation in a reactor coated with silica gel and by comparison of this effect with the acceleration observed in a reactor coated with boric acid. It is known that hydrogen bromide is dissociated into its atoms by radiation of less than 2800 d. For these experiments the reactor was equipped with a quartz thimble
.
125
100
E
.5
x 75
f
Time ( m i n )
p'
Figure 5. Isobutane Oxidation, Behavior of Reactor Coated with Boric Acid
0'
;so 0'
Reaction temperature 1 4 5 O C.; partial pressures: iso-C~Hio i 4 0 * 3 mm.; 0, 140 * 3 mm.; HU;, 20 * 1.5 mm. NA treatment of reactor hetwcen runs. Numerals refer to order of experiments
BORICACID COATING. This study of surface effects led t o the discovery that undesirable decomposition of product can be minimized by treating the Pyrex vessel with boric acid. The coating may be applied by the simple process of rinsing the reactor with a 2 to 5% aqueous solution of boric acid, draining, and evacuating the vessel a t 145" C. There seems to be no advantage t o higher temperature treatment (350" C.) although the resultant coating, presumably boric oxide, functions equally well. Figure 5 shows the smooth, continuous reaction curves typical of the oxidation in a boric acid-coated vessel. I n contrast with an uncoated reactor, the hydroperoxide yields are 75 to 85% (based on consumed oxygen) and the catalyst recovery is much higher. This suppression of product decomposition is reflected in the superior reproducibility obtained with the coated vessel (compare Figures 4 and 5 ) . The behavior of "soda" (or "soft") glass is very similar to that of Pyrex in that such surface is active in reaction-chain initiation but, like Pyrex, it effectively catalyzes the decomposition of tert-butyl hydroperoxide. Furthermore, a coating of borfc acid on the %oda" glass is equally effective in preventing this decomposition. SILICAGEL COATING. Although silica gel proved to be an unsatisfactory coating material, its use has provided data on reaction-chain initiation. This coating was applied by rinsing the Pyrex vessel with an aqueous 10% by weight suspension of silica hydrogel (containing a little hydrochloric acid), then sealing the reactor t o the apparatus and evacuating a t 145' C. As may be seen from the results in Table IV, the silica gel almost completely inhibits the oxidation; data for uncoated and boric acidcoated Pyrex reactors are included for comparison. To be noted also is the high hydrogen bromide recovery in the silica gel-coated vessel, because this result shows that lack of reaction is not due t o insufficient catalyst. The strong inhibition suggests, therefore, that this surface is either a remarkably efficient chain terminator or, mere likely, a very inefficient chain
25
20
IO
30
3
T i m e (minutes)
Figure 6. Isobutane Oxidation, Photochemical Acceleration in Coated Reactors Reaction temperature 145' C. 8 partial 140 * 3 mm.; 0% HrBsL)ures: iso-C~H~a. 40 3 mm.; HBr, 20 A 1.5 mm. 1: boric acid coated photo reaction; 2 boric acid coated, &ark reaction; 9, silica: gel coated, photo reactions 4, silica gel coated, dark reaction
*
T A B LIV. ~
EFFECTOF COATINGON OXIDATION OF ISOBUTANE (Temperature, 145' C . )
Coating Silica eel None Boric acid
TABLEv.
Reaction Time, Min. 33 18 46
HBr Recovery, % 85 9 46
Maximum Pressure Decrease, Mm. 0.6 26 124
PHOTOCHEMICAL ACCELERATION IN S I L I C A AND BORICACID-COATED REACTORS
GEL-
(Temperatiire: 145' C.; partial pressures: iso-C~H~o, 140 mm.; 0 1 , 140 mm.; HBr, 20 mm.) Silica Gel-Coated Boric Acid-Coated Reactor Reactor "Dark" Photo "Dark" Photo exgeriexpsriexperiexperiment Iwnt ment ment Reaction time min. 25 20 38 18 Maximum pkessure decrease, mm. 6 38 117 109 Oxygen oonsymed, % 15 52 92 93 Hydroperoxide yield, 44 76 64 % of consumed 02 Rate of pressure decrease, mm. min. l.lb 4.3 7.1 10.3 P+otoohemicai' rate increment, mm./ min. 3.2 3.2 (I
Peroxide product too small for titration. A maximum value maintained for only 3 minutes.
INDUSTRIAL AND ENGINEERING CHEMISTRY
2602
which housed a 100-watt mercury arc. A metered air stream cooled the arc and thus assured more nearly constant light intensity. The data &relisted in Table V and plotted in Figure 6. Of primary interest is the presence of a moderately rapid photo reaction in the silica gel-coated vessel, thus showing that oxidation will proceed if reaction centers are supplied by some external source. Furthermore, if the respective velocities of the "dark" reactions in silica gel- and boric acid-coated vessels are a measure only of the chain-interrupting efficiencies of the two surfaces, there should exist a comparable difference between the increments in rate produced by irradiation in the two vessels. I n reality, however, the observed increments are the same (Table V). Therefore, the difference in the activities of the two surfaces must be ascribed largely to the superior chain-initiating property of the boric acid. EFFECT OF SURFACE-VOLUME RATIO. The flow apparatus used for these experiments consisted of coiled 8-mm. inside diameter tubing mounted in an oil bath. The total reactor length was 55 meters and the volume about 3 liters which gives the same residence times as in the 25-mm. (inside diameter) helical reactors ordinarily used in the oxidations, but a much higher surfacevolume ratio. The inferior behavior of the small bore reactor is evident from Table VI. The oxidation is not only slower in the smaller tubing but also produces much larger quantities of degradation products. Degradation increases in importance at higher operating temperatures. Operation at higher pressure (2atmospheres absolute) results in only slight improvement.
OF 8-MM. AND 25-MM. (I.D.) PYREX TABLE VI. COMPARISON FLOW REACTORS FOR ISOBUTANE OXIDATION
iso-CIHIo, 3!4; Ot, 250; HBr, 36. Contact time, approximately 3 minutes] 8-Mm. Reactor 25-Mm. Reactor Temperature, C. 160 175 150 155 Oxygen consumed, cc./min. 87 174 164 216 tart-Butyl hydroperoxide produced, cc./min. 4.4 5.1 110 169 1 1 Ketone produced, cc./min. 21 83 Ketone-tort-butyl hydroperoxide ratio 4.8 16.4 0.01 0.01
[Conditions:
TABLEVII.
A
(CH& CHICHI
+ O2
-
HBr
Products
(CHl)&HaCHs
and the mechanism appears to be strictly analogous to that of isobutane (Equations 1-4). Efforts to find methyl isopropyl ketone were unsuccessful, a result which serves to eniphasize the greater reactivity of tertiary hydrogen atoms as compared with secondary. The mechanism of ketone formation in these hydrogen bromide-catalyzed o x i d e
...
...
... ...
69.0
.. .. ..
for
e' Purity w+%.
tions is discussed in the following paper under the reactions of the straight-chain paraffins which yicld this type of compound as the principal product. In the balance of reactants (Table VII) the ratio of total ketone to acetic acid is greater than unity probably because of the known decomposition of di-tert-amyl peroxide to acet.one and methyl ethyl ketone (7, 8, 11). I n other experiments with smaller inputs of hydrogen bromide, less di-tert-amyl peroxide is formed and the ratio more closely approaches unity. Di-terkamyl peroxide (14) has been identified by the following analyses: Molecular weight (cryoscopic, benzene) Carbon yo Hydroien % Equivalen't weight (ooncd. HI a t 60' C.)
Actual 177 69.4 12.8 86.5
Theory 174 68.9 12.7 87
A comparison of the characteristic physical constants of di&t-amyl peroxide as obtained by the authors with those of Milas and Surgenor ( 7 ) is given:
n%?
(8)
iso-CsHnb, 276; 0 2 , 276; HBr. 461 Moles/100 Moles Consumed of Iso-CBHiz 0 2 NBr 14.0 13.7 12.6 12.4 26.1 ... 26.6 8.5 8.6 5.2 ... 5.3 20.0 5.9 5.9 13.8 .., 4.1 ... 30.6 9.0 1.7 1.8 3.3 3.3 1.3 3.9
a 22% Hydrogen bromide recovered, 44.5% isopentane consumed, ,46% oxygen consumed, 2.9% input isopentane unaccounted for, 5.0% Input oxygen unaccounted for, and 4.1% input hydrogen bromide unaccounted
Boiling point, C. Melting point, C.
OOH
O
teyt-Amyl hydroperoxide Di-tert-amyl peroxide tert-Amyl alcohol Acetic acid Ketone Bromoketone tert-Amyl bromide tert-Amylene dibromide Carbon dioxide Carbon monoxide Propylene Hydrocarbons (carbon No. 2.8) Water
OXIDATION OF O T H E R BRANCHED-CHAIN COMPOUNDS
H
ISOPENTANE OXIDATIOX AT 165 C."
[Conditions: flows (cc./min.):
flow rates (cc./min.):
The hydrogen bromide-catalyzed oxidations of isopentane, 2,3-dimethylbutane, isobutyl chloride, and isobutyl bromide have also been investigated. I n addition, the oxidation of isopropyl chloride is included in the following discussion because the mechanism thereof can best be interpreted in conjunction a i t h that of the branched-chain compounds. Oxidation of Isopentane. The major products of the isopentane oxidation a t 165" C., as expected by analogy with the isobutane reaction, are tert-amyl hydroperoxide, tert-amyl alcohol, and di-tert-amyl peroxide (Table VII). In addition, however, a reaction involving scission of the bond between the secondary and tertiary carbon atoms of the molecule, (CH&CH . CHzC H , is indicated by the production of acetone and acetic acid. This "peroxide degradation" reaction, as mentioned in the first paper, is also indicated in the oxidations of other compounds (1). The main primary reaction is represented as follows:
Vol. 4,1, NO. 11
dZo
Authors' 159.0 (extrapolated) " 55 1.4085 0.808
-
Milas and Surgenor's
..,
...
1.4095 0.821
1741.7 0
Log,,p(mm.)
=
7.3816
- t ( " C.) -I- 228'
terl-Amyl hydroperoxide was identified by being converted to di-tert-am)l peroxide at room temperature with an excess of an equimolar sclution of tert-amyl alcohol and 65% sulfuric acid. The identity of the alcohol in the product was established by conversion to lert-amyl chloride with concentrated hydrochloric acid (boiling point, 86.8" C.; n? 1.4050; literature values: boiling point, 86" C.; ng, 1.4050). The acetono was isolated and the benzaldehyde derivative prepared: determined melting point, 110" to 111"C.; literature value for dibenzalacetone, 111" t o 112' C. The sodium salt of the organic acid was recrystallized from alcohol and the p-toluidine derivative prepared; determined melting point, 146" to 147" C. (aceto-p-toluide melting point, 148.2" C.). Oxidation of 2,3-Dimethylbutane. 2,3-Diniethylbutane vapor, oxygen ( a t rates of 276 ml. per minute each), and hydrogen bromide (at 46 ml. per minute) were passed through the reactor at 170' C. Degradation reactions were substantial and accounted for the major fraction of the product. Acetone, isopropyl alcohol, and acetic acid were the principal products, with substantial amounts of bromoketone and an unidentified peroxide also present. The peroxide is probably 2,3-dimethyl-2-hydroperoxybutane. 2,3-Dibromo-2,3-dimethylbutanehas also been isolated and its identification confirmed by melting point (165" t o 166' C.;
INDUSTRIAL AND ENGINEERING CHEMISTRY
November 1949
literature value 166' to 168" C,), and bromine content (determined, 64.2%; theory, 65.5%). Acetone and isopropyl alcohol are probably formed by a reaction of peroxide degradation, as follows : CH3 CH, I I
I
H$
L
CHSCOCH,
.
H
/ (9)
Some dimethyl isopropyl carbinol might be expected; however, it has not been found. Oxidation of Isobutyl Chloride. Chloro-tert-butyl hydroperoxide, a new compound, has been prepared by oxidation of isobutyl chloride. By analogy with isobutane and isopentane oxidations, chloro-tert-butyl hydroperoxide, isobutylene chlorohydrin, and bis (chloro-tert-butyl) peroxide are the expected products. The first two of these compounds have been isolated from the oxidation. The balance given in Table VI11 was obtained on the reactants a t 170O C. Because organic chlorides and bromides are present and their separate determinations are difficult, no attempt was made to obtain a hydrogen bromide balance.
TABLE VIII.
bromoketone, and bromohydroperoxide. Under the experimental conditions used for the exploratory runs (Table I X ) , bromohydrin is the principal product and only a relatively small amount of the expected peroxide was found. This may be taken to indicate either comparative ease of reduction by hydrogen bromide or thermal decomposition in a liquid phase resulting from condensation within the reactor.
TABLF, IX. ISOBUTYL BROMIDE OXIDATION AT 160" C."
-.I
+ (CHa)&HOH
2603
ISOBUTYL CHLORIDEOXIDATION AT 170" C."
[Conditions: flows (cc./min.): iso-CdHoC1, 275; 0 2 , 275; HBr, 451 Molea/100 Moles Consumed of Products Iso-CtH&l 0% Chloro-tert-butyl hydroperoxide 41.2 36.4 Isobutvlene chlorohvdrin 38.9 33.8 4.7 4.0 8.7 7.5 5.2 ... 1.3 1.2 7.2 6.3 Water ... 45.7 " 4 8 % isobutyl chloride consumed 60Y oxygen consumed 0 % input isobutyl chloride unaccounted for, ahd 5.8% input oxygen dnacoounted for.
The most successful method of concentrating chloro-tert-butyl hydroperoxide has been water extraction of the organic material followed by extraction with terl-butyl chloride. The terhbutyl chloride was distilled off and the remaining material distilled under'vacuum. A fraction was collected which analyzed 91.6% chloro-tert-butyl hydroperoxide on the basis of oxidizing power; boiling point 51 O to 61 O C. a t 2 mm.; ,':n 1.4452. The chloro-tert-butyl hydroperoxide was converted to chlorodi-ler2-butyl peroxide by reaction with a 35% excess of tert-butyl alcohol in an equimolar quantity of 65% sulfuric acid at room temperature. The purest chloro-di-led-butyl peroxide (ng, 1.4210; freezing point, -31" C.) was prepared from waterextracted hydroperoxide. An elemental analysis gave carbon 52.8%, hydrogen 9.4%, chlorine 19.7(5)%; theory, carbon 53.1%, hydrogen 9.4y0,chlorine 19.6(5)%. Isobutyric acid was isolated as the sodium salt and the p toluide prepared: melting point of derivative 104.1 O to 104.7"C.; melting point of N-p-tolylisobutyramide, 104" to 105" C. The value for haloketones was derived from the difference in oxidizing power of the sample before and after treating with excess base, which serves t o hydrolyze haloketones and thus destroys their oxidizing power. The chlorohydrin was determined from the amount of halide ion formed following treatment of the aqueous sample with excess base. This value less the amount derived from haloketones should give the amount of chlorohydrin present. Oxidation of Isobutyl Bromide. Paralleling the oxidation of the chloride, isobutyl bromide is converted to bromohydrin,
[Conditions: flows (cc./min.): iso-CdH~Br,275; Oa, 275; HBr, 501 Moles/100 Moles Consumed of Products Iso-CdHsBr 0 7 Bromo-tert-butyl hydroperoxide 5.4 4.7 Isobutylene bromohydrin 63.4 55.0 Alkyl bromidesb 15.7 ... Bromoketone 9.0 7.8 Water 98.2 " 50% hydrogen bromide recovered, 557' isobutyl bromide consumed 61% oxygen oonsumsd, 4% input isobutyl tromide unaccounted for, and 9% input oxygen unaocounted for. b Not isobutyl bromide.
...
Approximately one half the input catalyst w a s recovered and about the same proportion of the two reactants was consumed. The bromohydrin was cpnverted toisobutylene oxide which boiled a t 49" t o 51' C. (literature value, 50" C.) and had an index :n = 1.3718 (literature value, 1.3732). The ketone in the product is believed to be bromoacetone because of its limited water solubility and extremely lachrymal character. T h e bromohydrin, the ketone, and the peroxide were extracted from the product with water. The organic phase was stripped of unreacted bromide and the residue, which had a density of 1.7,was assumed t o be dibromides. The ketone and peroxide were determined by conventional titrations, and bromohydrin content calculated on the hydrolyzable bromine titer less the ketone value. Organic acid, dialkyl peroxide, and free bromine appeared t o be absent. Separation of the small amount of hydroperoxide from the bromohydrin, although difficult, was effected by extracting the product with a large volume of water and then re-extracting the aqueous phase with a limited amount of ether. I n this way a small sample was prepared which analyzed 30% ' bromo-tertbutyl hydroperoxide on the basis of oxidizing power. It was made to react with an excess of an equimolar mixture of tert-butyl alcohol and 65% sulfuric acid and after 72 hours a water-insoluble liquid separated. This appears to be bromo-di-tert-butyl peroxide, an analysis of which follows: Actual 224 43.9 7.8 35.3
Molecular weight Carbon % Hydrogkn, %
Bromine, %
Theoretical 225 42.7 7.6 35.5
Oxidation of Isopropyl Chloride. Although isopropyl chloride is not a branched-chain hydrocarbon, it does contain a carbon atom bonded t o a single hydrogen atom, and its prinripal products of reaction in the presence of oxygen and hydrogen bromidenamely, acetone, acetic acid, and 2-chloro-2-bromopropane-are best explained by analogy with those derived from branched-chain compounds. Isopropyl chloride is also capable of sensitizing the oxidation of certain paraffin hydrocarbons (see following paper) and for this reason the mechanism of its oxidation possesses additional interest. Following the mechanism for the branched-chain hydrocarbons, the primary reactions in the oxidation of isopropyl chloride are probably as follows:
FORACETONE FORMATION: CHaCHClCHa
+ Br eCH8A CICHa + HBr 00I
(10)
INDUSTRIAL AND ENGINEERING CHEMISTRY
2604
00-
I
CHsCCICHa
[
?OH ] + HBr +[CHtCCICHl + Br
(12)
]+
?OH
CH&C1CH3 2HBr + (an over-all reaction for the reduction of hydroperoxide to alcohol) (13)
Vol. 41, No. 11
TABLE x. ISOPROPYL CHLORIDE OXIDATION AT 160’
c.“
[Conditions: flows (cc./min.): iso-CaH~Cl,185; 0 2 , 185; Nz,186; HBr, 461 Moles/100 Moles Consumed of Products 02 EBrb Iso-CaHiC1 Acetone 28.6 26.8 ... Bromoacetone 12.0 11.3 23.Zb Acetic acid 25.5 23.8 2-Chloro-2-bromopropane 0.8 ... 18.Sb Propylene 12.0 ... ... Carbon dioxide 14.1 13.2 Carbon monoxide 27.8 26.1 ..* Water ... 59.2 ... 48% isopropyl chloride consumed, 51% oxygen consumed, 3.2% input isopropyl chloride unaccounted for, and 0.6% input oxygen unaccounted for. b Moles/lOO moles of HBr input.
...
...
[
-+- CHsCOCH8 + HC1
CHs!:lCHJ
(14)
FOR Z-CIiLoRo-2-RRo\rorHoP~sE F ~ R V A T I OTlle S : reaction involves the same chloroisopropyl radical produced in Equation 10:
I CHsCCICHs + Brz --+-CH3CBrCICH3
+ Br
The near equality of the molar amounts of acetic acid and carbon monoxide (see Table X) suggests a common origin and the oxidation leading to this result is expressible as follows: CHaCHClCHs
LITERATURE CITED
(15)
HBr + 202 --+ CHICOOH + CO + H20 + HC1 (16)
Barnett, B., Bell, E. R., Dickey, F. H., Rust, F. F., and Vaughan W. E., IND. E m . CHEW., 41,2612 (1949).
Dickey, F. H., Raley, J. H., Rust, F. F., Treseder, R. S., and Vaughan, W. E., I b i d . , 41, 1673 (1949). Htlckel, W., “Theoretisohe Grundlagen der Organischen Chemie,” Vol. 11,p. 86, Leipzig, Akad. Verlags., 1935. Mayo, F. R.,and Walling, C., Chem. Revs., 27, 372 (1940). Milas, N. A., and Harris, 8 . A,, J . Am. Chem. SOC.,6 0 , 2434 (1938).
However, the steps leading to this end result are obscure and no comparable phenomena have been observed with other compounds. Attempts were made to find methyl chloride and methyl alcohol, but results were negative or at best inconclusive. It would be predicted from Equation 13 that an increase in the hydrogen bromide input should increase the acetone yield a t the expense of that of the acetic acid and actually such a trend is observed. That methyl radicals are formed in the reaction is indicated by the considerable amount of methane found when operating with higher hydrogen bromide inputs.
Milas, N. A., and Surgenor, D. M., I b i d . , 68, 205 (1946). Ibid.,p. 643. Raley, J. H., Rust, F. F., and Vaughan, W. E., Zhid., 70, 88
+ HBr --+ CH4 + Br
Vaughan, TI7. E., and Rust, F. F., C. S. Patent 2,395,523 (Feb.
CH,
(17)
With lower hydrogen bromide inputs, as under conditions of the balance run (Table X), oxygen would be expected to compete more successfully for methyl radicals and actualIy somewhat larger amounts of carbon monoxide are found. This apparent ability of isopropyl chloride to generate methyl radicals under oxidizing conditions may be responsible for its sensitization of hydrogen bromide-catalyzed oxidations.
(1948).
Rieche, A., “Alkylperoxyde und Ozonide,” Leipzig, Steinkopff
I
1931.
Rossini, F. D., J. Research Natl. Bur. Standards, 15, 357 (1935). Rust, F . F., Seubold, F. H., and Vaughan, W. E., J . A m . Chem. Soc., 70, 95 (1948). Rust, F. F., and Vaughan, Wr.E., IND. Eiw. CHEM.,41, 2596 (1949). 26, 1946).
Zhid., 2,403,771 (July 9, 1946). Wiles, Q. T., Bishop, E. T., Devlin, P. A., Hopper, F. C., Schroeder, C. W., and Vaughan, W. E., IND.ENG.CHEM.,41, 1678 (1949).
RICEWEOAugust 17, 1948. Presented in p a r t a t the meeting of the Gordon Research Conferences of the American Association for the Advancement of of Science, Colby Junior College, New London, N. H., June 1948.
(Oxidation of Hydrocarbons Catalyzed by Hydrogen Bromide)
OXIDATION OF STRAIGHT-CHAIN COMPOUNDS P. J. NAYXOCICI, J. H. RALEY, F. F. RUST, AND W. E. VAUGHAN Shell Development Company, Emeryville, Calv. T h e hydrogen bromide-catalyzed oxidation of propane, butane, and related compounds containing secondary carbon atoms produces ketones in high yield. The reaction-a free radical chain probably involving intermediate hydroperoxide formation-is profoundly affected by the inclusion of certain organic compounds which serve as chain branching agents-for example, isopropyl chloride-or chain initiators-for exampIe, ditert-butyl peroxide,
S DISCUSSED in the preceding papers of this series (1, 7 ) the hydrogen bromide-catalyzed oxidation of branched chain hydrocarbons produces organic hydroperoxides through attack at the tertiary position. The analogous oxidation of straight-chain compounds, in which a secondary carbon atom is involved, results in the formation of ketones ( 5 ) . For example,
the hydrogen bromide-catalyzed oxidation of propane produces acetone in yields as high as 7575, based on consumed hydrocarbon, under conditions where 75% of the latter is reacted. Despite the difference in final product, the two oxidations are believed t o follow similar mechanisms, the distinction being the relative behaviors of tertiary and secondary alkyl hydroperoxides. Thus, for the propane oxidation the chain may be represented as follows and the relationship to Equations 1 to 4 of the preceding paper is to be noted: HBr
+
CHaCH&H3
0 2
+Br
+ (HO,. . . . .?)
I + Br ---+ HBr + CH3CHCHs
(1) (2)
02-
CHJ!X3CH8
+
I
0 2
-+-CHdCHCH3
(3)
