R. R. Baker1 McGill University Montreal, Canada and D. A. Yorke2 The University Hull, England
I
Theories on the Slow, Gas-Phase Oxidation of Hydrocarbons
The gas-phase oxidation of hydrocarbons and their derivatives has been actively studied since the beginning of this century, and while there is now agreement on many aspects of such reactions, the detailed mechanism is still far from complete. This is hardly surprising since these systems are, in general, extremely complex and markedly sensitive to the experimental conditions. This has resulted in the oxidation systems showing many apparently diverse effects, and also a general lack of reproducibility. Depending on the temperature and pressure, a mixture of a hydrocarbon and oxygen shows regions of slow reaction, explosive reaction (true ignition), and cool flames (in which the temperature and pressure momentarily increase and light is emitted from the system). All of these types of reaction are usually preceded by induction periods, occasionally as long as one hour. The slow oxidation usually shows the remarkable behavior of possessing a region of negative-temperature coefficient between about 300 and 450°C, where the rate of reaction actually decreases with increasing temperature (Fig. 1).
Early Chain Theories
Prior to about 1960 (9) researchers appear to have been mainly concerned with establishing the nature of the degenerate-branching intermediate, and with an explanation of the unusual, rather than the normal, features of hydrocarbon oxidation. Few workers had attempted to establish the nature of the main chain propagation steps, and consequently there arose the curious situation where the details of the oxidations were being discussed before the major reactions had been established. An important cause of this was the absence of suitable techniques for the separation, identification, and quantitative estimation of the many products. Considerable advances in the study of hydrocarbon oxidation have occur~edin the last decade or so, as a result of the advent of gas and paper chromatography. Particularly with gas chromatography, it is now possible to determine quickly and quantitatively products formed a t all stages of the reaction, and especially during the very early stages. The two current theories of hydrocarbon oxidation, the alkene theory of Knox, and the alkylperoxy-radical-isomerization theory due mainly to Fish, have both resulted from detailed analytical studies of the initial products of the reaction, and both attempt to account for the major products as well as for the kinetic features. The developments which led to the emergence of these two theories will now be considered. The Alkylperoxy-Radical-lromerization Theory
temperature
(OC)
Figure 1. The influence of temperature on the moximum rate. 60 mm Hg propane and 1 2 0 mm Hg oxygen (from ref. I11 b y courtesy of the Royal Society).
The aim of this review is to present the current theories on hydrocarbon oxidation in the slow reaction region. I n general terms, the kinetic mechanism of hydrocarbon oxidation can be described as a chain reaction with degenerate branching (2-8), i.e., a chain reaction in which radicals can also he produced from intermediate products of the reaction. The chief problem is in giving a precise chemical definition to the kinetic mechanism. Present address: G r o u ~Research and Develo~rnentCentre.
In the last ten years the results of many investigations (l(r26) have suggested that many of the products of hydrocarbon oxidation are formed by rearrangements of alkylperoxy radicals. Thus in general terms the overall mechanism is represented as RH R.
Initiation P~opagation
RH
O2 = R' + HOn' ++isomeriaation 01 = ROO'
+ radical + (radicalH)
ROO' = product
+ radical
=
R.
The scheme will also include degenerate branching and termination reactions. Fish (17, 30, 21) has greatly extended the theory, and explains the formation of the products of hydrocarbon oxidation in terms of the rearrangements of the alkylperoxy radical, and the different routes of decomposition of the hydroperoxy radical so formed. The basic modes of rearrangement and decomposition he proposes are summarized below: (1) Intramolecular hydrogen abstraction, giving the rearranged radical Volume 49, Number 5, May 1972
/
351
(2) Decomposition of the rearranged radical
oxirans
As the temperature is increased, intramolecular hydrogen abstraction (which is assumed to be unimolecular and with a high activation energy) becomes increasingly important and competes with the formation of alkyl monohydroperoxide. As decomposition of the rearranged peroxy radical is not a branching process, unlike the decomposition of the monohydroperoxide, an increase in the temperature will therefore result in a reduction in the rate of branching. Thus the rate of the overall oxidation will fall as the temperature is increased.
(i)loea of OH (iil CCfisaion
R
\ C=CHCH&H2R" /
+
+
.OH
alkenes
RCCHR'
1' 0
+
1+"
Process (1) can also occur with internal hydrogen abstraction from the B or y carbon atoms, leading to oxetans, oxolan derivatives, and carbonyl compounds. (3) Isomerization by group transfer, followed by decomposition of the rearranged radical RR'COORv
0-0 fiasion
----t RR'CO
+
R"0' carbonyl alkoxy compounds radical
The alkylperoxy-radical-isomerization-theory is very attractive from the point of view of explaining, in qualitative terms, the products of alkane oxidation. For instance, Fish has used it to account for the formation of over forty different products from the oxidation of 2methylpentane (19). The high yield of hexenes formed above 350°C was attributed to the increasing importance of reaction (1) as the temperature was increased, and reactions (1) and (2) were envisaged as being the main propagating reactions a t these high temperatures.
However, a sudden increase in the importance of reaction (1) as the temperature is increased implies that it has significant activation energy, whereas reactions of this type have low activation energies, of the order of 20 k~mol-?(27). Fish has also used the isomerization theory of alkylperoxy radicals to account for the existence of a nega tive-temperature coefficient during the slow combustion of hydrocarbons (17,21). In the low-temperature region where the oxidation is quite selective, Fish (1 7, 19) suggests that degenerate branching results from the alkyl monohydroperoxide, produced by a linear chain involving intermolecular hydrogen abstraction by the alkylperoxy radical RO,.
- RH
/
-R,OOH
ROOH
RO.
+
.OH (branching)
Journol of Chemical Education
prcduct
An essential part of this mechanism is that OH is a main propagating radical in the negative-temperature coefficient region. Hydrogen abstraction by OH is always exothermic, with a low activation energy (28, 29) and so OH will always unselectively attack an alkane molecule. Alkane oxidation in the negativetemperature coefficient region is unseleetive (IQ), which gives weight to the theory. In the high-temperature region, the oxidation rate increases with temperature because branching is reestablished by the decomposition of dihydroperoxides which are formed by the further addition of oxygen. However, the supposition that the further addition of oxygen to the rearranged radical occurs predominantly a t high temperatures is unsatisfactory, since it implies that the oxygen-addition reaction has a significant activation energy, and this is most improbable. The Alkene Theory
There is considerable evidence (7,14,24,50-40) that during the oxidation of the lower alkanes (C&L tO C4HIo)between 300 and 500°C, a t least 70% of the alkane lost during the initial stages appears, wherever possible, as the conjugate alkene. This fact appears to have been unobserved, or neglected, by most workers prior to 1963, probably because product analyses were only carried out a t relatively late stages in the reaction. Knox (41, 42) suggested that the alkene (AB) was formed in reaction (3). R. On = AB H02' (3)
+
R'
+ OS
-
+
oxygenated product
(4)
I n contrast with Fish's views, that the competition of reactions (3) and (4) means that production of the alkene is only favored a t high temperatures, Knox believes that any scheme for alkane oxidation, even at temperatures as low as 300°C, must not only involve the formation of the alkene as the major product, but also involve the reactions of the alkene under the conditions of the oxidation. Aldehydes, ketones, and, to a lesser extent, oxirans are major products of alkene oxidation (52, 55, 45-49) and this probably explains the luge yields of these products during alkane oxidation, after a significant reaction time. Knox's results (33) indicate that the minor products of hydrocarbon oxidation are formed heterogeneously
tR. 352
hydrogen
I
R1'CH2CH2
RCOCH,R' carbonyl compounds with rearranged skeletons
-
intramolecular
HOz
R'
RRIR"COO.
-
R. %.
from alkylperoxy radicals, and the system is represented by the reactions R. + 0, = AB 80%
+ HO,'
(5)
R. + 0,
20%
= ROO' aall ROO' = minor products
(6) (7)
It might be supposed that the main propagating chain in such a systemis RH
R. + Oz = AB + HOz' = R.
+ HOS.
(5) (8)
+ HIOZ
Since most hydrogen abstractions by H 0 2 radicals are endothermic (50), the selectivity of HOz radical abstractions should be high. Knox (41, 51) has shown that there is a marked decrease in selectivity between the earliest stages of alkane oxidation and the intermediate stages, and he concludes that a selective radical (H02) removes the alkane initially, while a more reactive and unselective radical (OH) is important later. Since the alkene-formingreaction (5) must occur throughout the oxidation, there must he some process which converts the H 0 2radical into an OH radical, and Knox postulates (dl,&) the following scheme
+
= 'ABOOH
AB H01' .ABOOH OZ= .00ABOOH H02' = HOOABOOH =
+
+
(9)
.00ABOOH HOOABOOH 0 9 2 0 H A 0 BO
(10) (11) (12)
20H 4- A 0 f BO
(13)
+
+
+
Overall reaction 2HO2' + AB
=
AB is the conjugate alkene of the parent alkane RH, and A 0 and BO are carbonyl compounds or oxirans. An important feature of the scheme is that once the overall oxidation process is underway, it proceeds by a propagating chain that involves molecular iutermediates (AB). Degenerate branching can occur by oxidation of the carbouyl compounds A 0 and BO, giving radicals RCHO
+ On
=
RCO'
+ HOz'
(14)
or from the decomposition into radicals of alkyl hydroperoxides, which are not formed by disproportionation of H o p and R02. Below about 400°C termination is effected by disproportionation of two HO2 radicals. H01
+ HOI. = HtOz +
01
about 370°C for most alkanes under typical oxidation conditions (41,52). Knox has explained the existence of the region of negative-temperature coefficient in the following manner. At the temperature where the efficiency of the conversion begins to decrease, further increase in temperature will cause the relative concentration of OH radicals to decrease continuously, and so the relative proportion of alkane abstraction by OH radicals will decrease. Thus, the overall reaction rate also decreases because alkane abstraction by H 0 2 is much slower than that by OH radicals. Above the temperature where there is absolutely no conversion of HOz to OH, the overall rate will increase gradually with increasing temperature due to the normal increase in the velocity constant for HO, abstraction with temperature. The increase in overall rate with temperature will become more marked above about 450°C when homogeneous decomposition of HzOzinto OH radicals occurs, and abstraction by OH radicals is established once again. Conclusion
The two current theories of hydrocarbon oxidation explain the formation of products and the kinetic features of oxidation reactions in different ways, and it is not possible a t the present time to decide which theory is the more generally applicable. Both theories have their virtues, both have their disadvantages, both have evidence to support them, and both have been criticized. Knox believes that the activation energies for the formation of oxygenated products via isomerization reactions of R 0 2radicals are high (53), probably of the order of 140 kJ mole-' (54), and so their occurrence below about 500°C will be negligible relative to other reactions of ROp radicals (reaction with RH, heterogeneous reaction, etc.). Further, Cartlidge and Tipper (I$) have shown that between 300 and 400°C the prcdominant reaction of the rearranged peroxy radical is the further addition of oxygen, and not decomposition. However, kinetic evidence has more recently been obtained which indicates that at 480°C, the rearranged 2,2-dimethylpropylperoxy radical, (CH3)2C(CH2')-
(15)
The complete mechanism in the low-temperature region is summarized in Figure 2. The efficiency of conversion of H02 into OH by this mechanism is dependent on the point of equilibrium of reversible reaction (9), and on the relative rates of reactions (10) and (15). When the rate of reaction (10) is greater than that of reaction (15), the rate of removal of ABOOH will be greater than the rate of removal of HOz radicals by reaction (15). The majority of HOP radicals will then be converted into OH radicals. The strength of the AB-OOH bond is only about 50-60 kJ mole-' (41,5$), and therefore, as the temperature is increased, the stability of the ABOOH radical will rapidly decrease. Thus the point of equilibrium of reaction (9) is shifted to the left and so effective conversion becomes impossible above a certain temperature, and this temperature has been estimated to be
f
Initiation
I
1
ROOH
+
+
.OH -+
R(h
7
ROOH
RO.
wall destruction
'1 "2
subsidiary chain
Ox
- -
,0RH
~
R.
I
RO.
+
.OH (branching)
+
minor products
3 chain termination
+
\\,
AB
initial chain
'I
....
I
.OH
I)"
I
maim propagation chain
+
-
ckrbonyl compounds
1" branching
Figure 2. The dksne theory mechanism for the low-temperature oxidation of alkanes.
Volume
49, Number 5, May 1972 / 353
(CHzOOH). can either decornoose or add on another . oxygen moikcule (55). The alkene theory is not, however, free from criticism, since there is evidence (66-58) that for higher alkanes (higher analogs than CIHIQ),the conjugate alkene is in fact less reactive than the parent alkane, which would suggest -- that it olavs no oart in the mechanism of alkane oxidation.& Further, a shock tube study of the oxidation of pentane a t 298 to 435°C (59) has shown that no trace of pen-2-ene was formed, even in the first 1% of reaction. However, Knox (65) now considers that the alkene theory is in fact limited to the lower alkanes. Fish (19, 20) has attempted to reconcile Knox's alkene theory with his own alkylperoxy-radical-isomerization theory, since alkyl hydroperoxy radicals are produced in both schemes. He has combined the two theories into a single comprehensive mechanism, which is shown schematically in Figure 3. Fish believes that Initiation
1
ROOH
R
.ABOOH
%
-+
RO.
+
.OH (branching)
oxygen ring product orcarbonyl compound
+ alkene
RH Figure 3 . The combined olkons theory and alkylpemxy-radical-isomerizotion theory mechanism for the low-temperoture oxidation of ohones.
this reconciliation will hold if icn > k ~ and , kc > kn [RH], and he shows that these two conditions are met. It is likely that this scheme is now a good approximation to the mechanism of many alkane oxidation systems. Recently, Gray (60-65) has put forward a purely mathematical approach of hydrocarbon oxidation which may influence future developments in the field. However, these mathematical theories do not specify the chemical nature of the reactions involved, and considerable uncertainty about these reactions still exists. There is a general sparsity of reliable quantitative information on the reactions of alkyl radicals, particularly with oxygen, while the reactions of alkanes with radicals important in oxidation systems, such as OH (28, $9) and HOz (64), have only very recently been put on a quantitative basis. Clearly, this information is necessary if meaningful reaction schemes are to be proposed for alkane oxidation. Once all of this information is assembled, we shall be able to turn to the next, and perhaps most exciting, stage in the development of hydrocarbon oxidation, when the proposed mechanisms can be tested quantitatively against experimental observations, by computer simulation, using known values for the rates of the elementary reactions. 354
/
Journal o f Chemical Education
The authors would like to thank Professor R. R. Baldwin and R, W, Walker, of the University of Hull, England, for much discussion, criticism, and advice during the initial stagesof the production of this review, Literature Cited (1) SeAanrs, M., A N D Hmsxlnwoon, Srn C. N., Pmo. Roy. Soc. Sar. A, 276,324 (1963). ( 2 ) SEMENOV, N. N.. Themica1 Kinetics and Chain Reactions." Oxford university press, 1935, Chapters 2 and 3. v .. N.. "Some Pmbisms of Chemical Kinetios and Reao(3) S ~ r r ~ o N tivity," Pergamon Presa, London, 1959, Yol. 2, Chapter9. (4) DAINTON, F. S.. "Chain Reactions: An Introduction." (2nd ed.). Methuen's Chemical Monographs, Methuen and c o . Ltd.. London, 1966. Chapter 5. a . J., A N D T I P P E ~C. , F. H.. "Chemistry of Combustion (5) M ~ ~ x o r G. Reaotions." Butterrvorths. London. 1962, p. 103-107. (6) KNOT J. H., Seventh Symposium (International) on Combustion. Butterworths. London. 1959, p. 122. ( 7 ) KNOX.J. H., Tram Forodav Soc., 55,1362 (1959). H.. Q u ~RCU. . (London),11,313 (1957). ( 8 ) TIPPER,C.F. ( 9 ) Sx~enl*.V . Ya., "The Gas Phase Oxidation of Hydrooarbond' ( T ~ o n s lolo?: MULL IN^, M . F.), Pergamon Presa, London. 1964, Chapters (1959). (11) CARTLIDOE. J.. A N D TIPPER, C. F. H., PVOC.Chew& SO
