`
`OLEFIN OXIDATION
`
`By L. BATEMAN
`
`(THE BRITISH RUBBER PRODUCERS' RESEARCH ASSOCIATION,
`WELWYN GARDEN CITY, HERTS.)
`
`THE interaction of olefins with molecular oxygen is not only a subject of
`widespread industrial importance, but is one of the most thoroughly under(cid:173)
`stood chemical processes. This Review attempts to give a broad picture
`of the main mechanistic features. An earlier article 1 has reviewed the
`historical background and has given details of the method of approach and
`of the earlier kinetic data which were largely responsible for opening up
`this field.
`In Section (l) we present the generally accepted chain mechanism and
`the simpler rate expressions which are often obeyed. Certain quantita(cid:173)
`tive comparisons of olefinic reactivity derivable on this basis are then dis(cid:173)
`cussed. Our main concern, however, is to show how comparatively small
`changes in certain mechanistic details can give rise to substantial differences
`in the observed kinetics-so much so that a profound change in mechanism
`might be imagined. Section (2) deals with rate measurements under non(cid:173)
`stationary state conditions designed to determine the propagation- and
`termination-rate constants separately, and emphasises the inherent limita(cid:173)
`tions to accuracy which oxidation systems present in this respect.
`Sections (3), (4), and (5) are concerned with the initiation of the oxidation
`chain, and the part played in this by the hydroperoxide which is the primary
`reaction product. This behaviour of the hydroperoxide is responsible for
`the autocatalytic character of the oxidations, and the complexity and
`environmental sensitivity of its decomposition serves to complicate the
`kinetics of the oxidations as u. whole. Attention is drawn to circumstances
`where the fraction of hydroperoxide undergoing decomposition is large
`compared with that being formed, so that the character and kinetics of the
`process are greatly altered despite the same fundamental reactions being
`involved. Section (6) .describes efforts to analyse the initiation process
`quantitatively in order that tho number of oxidation chains being started
`under given conditions can be specified.
`Under mild conditions of oxidation, the chain is long and the fraction of
`the hydroperoxide which decomposes to initiate fresh chains is very small.
`The overall yield of hydroperoxide should thus be nearly quantitative.
`In
`Section (7) serious discrepancies are interpreted in terms of the dual reac(cid:173)
`tivity of peroxy.radicals towards olefinH, the consequence being that the
`measured rate conHtants are composite qnan1 ities relating to both hydrogen
`t'xtraction and double-bond u.ddition.
`The allylic radicals formed on removal of an o:.-methylenic hydrogen
`a.tom from an olefin are mesomeric and hence the derived product may
`1 Bolland, Quart. Reviews, 1949, 3, 1.
`147
`
`Page 1 of 21
`
`Noven Ex. 1020
`
`
`
`148
`
`QUARTERLY REVIEWS
`
`View Article Online
`
`consist of allylic isomerides. The behaviour of mono-olefins and 1 : 4-
`diolefins in this respect is discussed in Section ( 8).
`
`Termination :
`
`0
`
`1. General Kinetic Behaviour
`The following reaction scheme, where RH represents the olefin with an
`<X-methylenic hydrogen atom H, ri is the rate of chain initiation, and the
`k's are the velocity coefficients of the reactions indicated, accounts for the
`reaction characteristics with remarkable comprehensiveness.
`Initiation :
`Propagation :
`
`r 1
`Production of R· or R0 2• radicals
`R· + 0 2
`--+ R0 2•
`k 2
`R0 2• + RH --+ R0 2H + R·
`k 3
`2R· ~}Non-initiating or k 4
`k-
`R· R0 2 • ~ -propagating
`k;
`2R02• ~ products
`The more obvious of these are: (i) high yields of the hydroperoxide,
`R02H (cf. p. 162) ; (ii) catalysis by light and by free-radical producing
`substances, indicating the free-radical nature of the reaction ; 2 (iii) quantum
`yields greater than one and a proportionality between rate and the square
`root of the light intensity in photo-oxidations, indicating a chain reaction
`with mutual destruction of two chain carriers in the termination step ; 3
`(iv) a parallelism between oxidisability and the relative ease of rupture of
`the C-H bond in RH, indicating the importance of a hydrogen-exchange
`reaction such as (3) (cf. p. 149) ; (v) the formation of conjugated-diene
`hydro peroxides from 1 : 4-dienes, in agreement with the generation of
`mesomeric R· radicals as in (2) (cf. p. 164); and (vi) the marked retardation
`in rate produced by phenolic compounds (among others), which interfere
`with the propagation process by providing an alternative and easier reaction
`for the R02• radicals that does not liberate a radical equivalent to R•. 4
`2 = k4k6 , the above mechanism yields the rate
`It being assumed that k5
`equation (for long chains) 5
`ik k -![RH]
`k2k6-l[02]
`kak~=l[RH] + k2k6 -1[02]
`r = ri a a
`where r is the overall rate of oxidation and the square brackets signify
`concentration terms.
`Oxidisability at "High" Oxygen Pressures.-"\Vhen reaction (2) is so
`much faster than (3) that [R] ~ [R02], termination can be assumed to occur
`entirely by reaction (6), and equation (1) simplifies to
`(2)
`r co = r1lk3k6 -i[RH]
`Equation (2) accurately expresses the observed kinetics for most olefins at
`
`(1)
`
`2 Bateman and Bolland, Proc. XIth International Congress of Pure and Applied
`Chern., 1947.
`3 Bateman and Gee, Proc. Roy. Soc., 1948, A, 195, 376.
`'Bolland and ten Have, Trans. Faraday Soc., 1947, 43, 201; Discuss. Faraday
`Soc., 1947, 2, 252.
`6 Bolland, Proc. Roy. Soc., 1946, A, 186, 218.
`
`Page 2 of 21
`
`Noven Ex. 1020
`
`
`
`BATEMAN ; OLEFIN OXIDATION
`
`View Article Online
`149
`
`oxygen pressures greater than 100 mm. (the "high" pressure region)-an
`interesting exception being discussed later. At constant r1, estimates of
`k3k6 -i are thus obtained from r 1()/[RH] and these measure the relative
`reactivities of different olefins in reaction (3), since k6 is not very sensitive
`to changes in R (seep. 153). Bolland 6 has in this way developed a correla(cid:173)
`tion between olefinic structure and oxidisability. Referring to propene,
`CH3·CH:CH2, at 45°, he concludes that:
`(a)
`(c)
`(b)
`(i) Replacement of one or two hydrogen atoms at (a) and/or (c) by
`alkyl groups increases k3 by 3·3n, where n is the total number of substituents;
`similar replacement at (b) is without effect.
`(ii) Replacement of a hydrogen atom at (a) by a phenyl group increases
`k3 23-fold.
`(iii) Replacement of a hydrogen atom at (a) by an alk-1-enyl group
`increases k3 107 -fold.
`(iv) The value of k3 appropriate to an cx-methylenic group contained in a
`cyclic structure is 1·7 times that of the group contained in an analogous
`acyclic structure.
`These rules relate to broad variations, as implied by the assumed
`equivalency of different alkyl groups. The assumptions that k 6 is invariable
`and that benzoyl peroxide (used as a standard initiator) initiates throughout
`with equal efficiency also introduce second-order uncertainties. More
`serious discrepancies occur in special cases. Thus 2 : 4-dimethylpent-2-ene
`is at least 10 times less reactive than would be predicted, 7 presumably
`because of steric hindrance at 0 3-behaviour simulated in a saturated
`
`0
`
`TABLE l
`
`Olefin, RH ~>
`
`Ea b
`
`~Ha Ea/04
`
`~(AH3)
`
`*
`CH 2:CH•CH3 •
`*
`CH 2:CH·CH2Alk .
`*
`CHAlk:CH•CH 2Alk
`*
`CAlk 2:CH ·CH2Alk
`*
`CH 2:CH·CH2Ph
`*
`CHAlk:CH ·CH 2·CH:CHAlk .
`*
`CH:CH•CH 2
`I
`I
`Alk-Alk
`
`13·5
`
`ll·s
`
`10·5
`
`9
`
`10
`
`6
`
`9·5
`
`34
`
`29
`
`26
`
`23
`
`25
`
`15
`
`24
`
`0
`
`5
`
`8
`
`11
`
`9
`
`19
`
`10
`
`I
`
`I
`
`I
`
`All values are in kcal.jmole
`Reactive oc-methylenic group
`indicated by an asterisk.
`7> Calculated from
`!E1 + jE6, where E 0 designates the overall activation energy of oxidation-;
`l!J3 = E 0 -
`catalysed by benzoyl peroxide. E 6 is taken a<3 zero and E1 as 30 kcal.jmole (Bolland's
`published values are based on E 1 = 31 kcal. /mole).
`--- '" ---
`11 Bolland, Trans. Faraday Soc., 1950, 46, 358.
`
`7 Morris, unpublished result.
`
`Page 3 of 21
`
`Noven Ex. 1020
`
`
`
`View Article Online
`
`150
`
`QUARTERLY REVIEWS
`
`In general,
`hydrocarbon by the inertne::,s of 2: 2: 4-trimethylpentane.8
`howevl"r, Bolland's rules rationalise the behaviour of different olefins and
`of differl"nt allylic systems in the same olefin. For example, in the isoprenic
`unit ·OH2·C(OH3):0H·CH2·, the relative Cl-methylenic activity at the three
`(y)
`(1)
`(x)
`positions, x : y : z, is approximately I : 3 : II.
`The numerical factors given for 45° become smaller at higher tempera(cid:173)
`tures, since increased reactivity partly reflects a lower activation energy (E)
`for reaction ( 3). 6 A vE-ra go values of E 8 for the systems considered in
`(i)-(iv) above are given in Table I. E 3 may be related to the corresponding
`heat of reaction, t:.H 3 , 6 whose variation, tl.(M 3 ), from olefin to olefin ex(cid:173)
`presses differences in resonance energy and other stabilising influences in
`the different allylic radicals.
`Oxygen-pressure Dependence.-Decreasing the oxygen pressure reduces
`the overall rate of oxidation only when reaction (2) is not incomparably
`faster than reaction (3), i.e., when [R·] is not negligible compared with
`[R0 2·]. The pressure at which this condition prevails depends on the
`reactivity of the olefin- the lower the reactivity, the slower is reaction (3),
`and the lower the value of [0 2] necessary to reduce the rate of reaction (2)
`accordingly. This effect may be enhanced by the reactivity of R in reaction
`
`(a)
`
`___ _ ___...o---
`
`.-o-------
`
`oo~--~--2~~~--~--~~--~--~6~~~~~~
`Oxygen pressure (mm)
`FIG. 1
`Variation of the rates of oxidaMon of (a) hexadec-1-ene (45°), (b) ethyl linolenate (45°)
`and (c) 2 : 6-dimethylhepta-2 : 5-diene (25°) with oxygen presaure.
`
`(2) being qualitatively the inverse of that of RH in reaction (3), although
`the quasi radical-radical nature of (2) renders it far less responsive than (3}
`to changes in R. Some rate-pressure dependences are illustrated in Fig. I.
`Hexadec-1-ene shows no depl"ndence above I mm., but with increasing
`olefin reactivity the pressure at which the rate becomes insc:>nsitive also
`increases. With the intensely reactive 2 : 6-dimethylhepta-2 : !5-diene, the
`rate at atmospheric pressure is well below r co· Ar:;; only reaction (3) of
`the propagation and termination steps has an appreciable temperature
`
`s W1baut and Strang, Pt·oc. K. Ned. Akad. Wet., 1951, 54, B, 229.
`
`Page 4 of 21
`
`Noven Ex. 1020
`
`
`
`BATEMAN: OLEFIN OXIDATION
`
`Yie\v Article Online
`151
`coefficient, the dependence of rate on oxygen pressure extends to higher
`pressures at higher temperatures, as exemplified in Fig. 2.
`Generalised Rate Equation.-For some olefins, equation (1) expresses
`exactly the observed kinetics over the whole range of oxygen pressures
`where accurate measurements are possible (down to about 1 mm.).
`In
`general, however, deviations are found which arise from departures from
`.,-t
`2 = k4k0 (p. 148), and which vary from olefin to
`~ the assumption that k6
`2
`olefin in an intelligible manner-in extreme cases being sufficient to modify
`::;::;;
`the oxidation kinetics at pressures near atmospheric.
`
`C':l
`
`0
`
`0 o~----~2o~--~40~----~~----~~~----,.=~o
`Oxygen pressure (mm)
`FIG. 2
`Influence of temperature on the oxygen-pressut·e dependence of ethyl Zinolenate at (a) 25°,
`(b) 35°, and (c) 45°.
`
`•
`
`The completely general form of the rate equation (for long chains) is: 9
`r- 2 = r1- 1(k2- 2k4[02]- 2 + 2k2 - 1k3 - 1k5(RH]-1[02]-1 + k3 - 2k6[RH]- 2)
`(3)
`or, alternatively, by combination of (3) and (2):
`(r OC)/r)2 = 1 + 2cpk2 -lk4lkalc6 -l[RH] [02]-1 + k2 - 2k4k3 2k6 - 1[RH] 2[02]-2 • (4)
`where cp = k 4-ik6k6-l and r 00 is defined by equation (2). Equation (4)
`requires a plot of (r 00/r) against [02]-1 to be linear if 4> = l [as assumed in
`deriving (I)], concave to the latter axis if cp < I, and convex if cp > I.
`Examples of all three conditions are known.11 From the slope and ordinate
`intercept of the plot of { (r 00/r) 2 - 1 }[02] against [02]-1 (see equation 4),
`cp and the composite coefficients k2k4 -1 and k3k6 -1 can be determined. The
`data listed in Table 2 show that the large variations in k3k6 -1 with olefinic
`structure are not paralleled by any of comparable magnitude in k2k, -!.
`8 Bateman, Gee, Morris, and Watson, Discuss. Faraday Soc., 19lH, 10, 250.
`
`Page 5 of 21
`
`Noven Ex. 1020
`
`
`
`152
`
`QUARTERLY REVIEWS
`
`View Article Online
`
`TABLE 2*
`
`Olefin
`
`10'k3k6 - *< 45°)
`(mole-! ti sec. -i)
`
`10-8k1k, -t{45°)
`(mole- i 1.* sec.-!)
`
`Methyl oleate
`Phyteno
`Digeranyl.
`Ethyllinoleate
`Ethyl linolenate
`2 : 6-Dimethylhepta-2 : 5-diene
`
`..
`
`1·53
`1·07
`3·82
`20·7
`41·4
`130 (25°)
`
`0·5
`0·4
`0·9
`1·6
`1·3
`1·2 (25°)
`
`"'
`0·3
`1·0
`3·1
`2·5
`3·3
`6·5
`
`* Absolute comparison requires the composite and individual rate constants in this
`and the following Tables to be multiplied by factors of eiB p and eB.P• respectively,
`where eB.P denotes the initiating efficiency of benzoyl peroxide (see p. 161). As this
`quantity has been determined only for a few olefins and is variable, it is preferable
`here to base all the data on the value, e.a.r = 1.
`
`~ Values.-Two points concerning the values of 4> may be noted : (i) they
`are all rather small compared with some of the large values found for the
`E-quivalent quantity for cross-termination in copolymerisations; and (ii)
`they increase with the reactivity of RH. These features probably have a
`common link in reflecting a large diminution in the resonance and polarity
`properties of the group R on relay through the 0-0 bond of the R02•
`In copolymerisations, the analogous if> values relate to the inter(cid:173)
`radical.
`play of structural effects in substituted alkyl radicals only; in the oxidations,
`the R02 • radical is essentially a common factor from system to system and
`tends to depress in reaction (5) any variation in reactivity in R· which is
`fully manifest in reaction (4).
`Influence of ~ on the Kinetic Form.-As the reactivity of RH increases,
`two factors enhance the kinetic importance of the R· radicals : (i) the lessen(cid:173)
`ing of the difference between k2 and k3 ; and (ii) the increase in cfo. The
`practical repercussions are strikingly illustrated by comparing the variation
`in the relative importance of reactions (4), (5), and (6) at different oxygen
`pressures for different olefins (Fig. 3).10 The displacement, broadening, and
`intensification of the R· + R0 2• curve on passing from phytene to
`2 : 6-dimethylhepta-2 : 5-diene leads to such marked kinetic differences as
`to suggest that the oxidation mechanisms are fundamentally different.
`Even at pressures near atmospheric, equation (2) does not apply even
`approximately to the heptadiene; the rate is neither directly proportional
`to [RH] nor independent of [0 2]. At pressures higher than 100 mm.,
`termination by reaction (4} is negligible and the dependence of rate on
`[RH] is then given by equation (3) without the term k 2 - 2k4[0 2]-2 .
`The ability of the oxidation mechanism to account in so detailed and
`rational a manner for the kinetic behaviour of olefins of widely varying
`reactivity establishes its formal correctness. As discussed later (p. 162),
`it is sometimes necessary to modify or supplement the scheme given on
`p. 148 in order to obtain consistency with product data.
`
`1o Bateman and Morris, Trans. Faraday Soc., 1953, 49, 1026.
`
`Page 6 of 21
`
`Noven Ex. 1020
`
`
`
`BATEMAN: OLEFIN OXIDATION
`
`View Article Online
`153
`
`(c}
`
`(b)
`
`(a.)
`
`80
`
`60
`
`40
`
`20
`
`/
`
`0
`""" ~
`'-
`~
`.;:.so-
`t::l
`~
`~
`~60
`
`...... .s .a
`
`~40
`§
`.;::.
`~
`~20
`ct
`
`0
`
`80
`
`60
`
`4D
`
`..t
`0
`0l
`
`'2 et:l
`
`~
`
`('j
`s:::
`0
`E
`0
`::l
`0 ;.,
`c:
`!.;l
`@
`t
`"§
`-B
`~
`,.0
`"1:1
`(\.)
`"' et:l
`..s::
`~ :::: c..
`..t
`•n
`0\
`>·i
`1;:;
`::::
`c
`....,
`et:l
`
`0
`c:
`0
`"1:1
`(\.)
`..s::
`.~
`:0 ;::s
`c..
`
`o~~~~~~~~~~~--._~~~
`1
`10
`Oxygen pressure (mm)
`FIG. 3
`Termination characteristics of the oxidation of (a) phytene (45°), (b) ethyl linoleate (45°),
`and (c) 2 : 6-dimethylhepta-2 : 5-diene (25°) at various oxygen pressures.
`
`2. Individual Rate Coefficients of the Propagation and Termination
`Reactions
`For all chain reactions, measurements under stationary-state conditions
`permit only composite velocity coefficients (such as k3k6-l) to be determined.
`
`Page 7 of 21
`
`Noven Ex. 1020
`
`
`
`154
`
`QUARTERLY REVIEWS
`
`View Article Online
`
`Non-stationary-state measurements with oxidising olefins have been made
`by the rotating-secto~ technique n, 12 and by following directly the photo(cid:173)
`chemical pre- and after-effects, 9 , 12 • 13 • 14 and absolute values of the several
`propagation and termination constants derived. However, severo limits to
`accuracy are imposed by certain inherent complications, which are common
`to similar measurements in all gas-liquid systems and deserve to be more
`widely known.
`The principle of the photochemical pre- and after-effects is expressed
`If the oxidation is followed in the dark (rate = rD) and then
`in Fig. 4.
`the light is switched on, a time interval elapses before the uniform
`
`Ttme
`
`Darlr Light
`
`(/J)
`
`Time
`
`(«-)
`
`L~t ~lr
`
`FIG. 4
`Definition of (a) the rate decay intercept, Id, and (b) the rate growth intm·cept, Ir..
`
`light rate (rL) is established, i.e., while the increased concentration of
`chain carriers builds up. The inverse occurs when the light is switched off.
`The intercepts lg and Jd represent amounts of oxygen absorbed during
`the non-stationary state conditions, and can be shown to be defined by
`Jd = a ln{(rL + rD)/2rn} and lg = a In {2rL/(rL + rn)}, where a is a
`complex quantity containing the propagation and termination constants in
`different ratios from those in the stationary rate equations (under "high"
`pressure conditions, a reduces to k3k6 - 1[RH]). The important complicating
`factor is that the oxygen concentration in the solution does not remain
`constant during the change from rn to rL. As oxygen is continually being
`removed by reaction, the actual value of [02] is always lower than the satura(cid:173)
`tion value. The extent of this difference depends on the speed by which
`
`11 Bateman and Gee, Proc. Roy. Soc., 1948, A, 195, 391.
`12 Bamford and Dewar, ibid., 1949, A, 198, 252.
`n Bateman and Gee, Trans. Faraday Soc., 1951, 47, 155.
`u Bateman, Bolland, and Gee, ibid., p. 274.
`
`0
`t:
`
`~
`IJ.)
`...t:
`.~ :g
`
`0..
`
`Page 8 of 21
`
`Noven Ex. 1020
`
`
`
`BATEMAN; OLlllFIN OXIDATION
`
`View Article Online
`1.55
`
`the oxygen can be replenished from the gas phase by agitation. The inter(cid:173)
`cepts actually measured are not in fact I g and I d but are given by 13
`I'll = /d + (rL -
`rn)/ks and I's = lg + (rL- rn)/k~o, where k8 represents
`the shaking efficiency.
`In principle, therefore, the change in oxygen
`concentration can be compensated for automatically by evaluating
`I'g (= Id- lg =a ln{(rL + rn) 2/4rLrn}).
`In practice, the term
`1'11 -
`(rL- rn)/k8 , while often negligible compared with [02], is large compared
`with Id and Ig. For example, under favourable experimental conditions
`with ethyllinoleate at 15°, an oxygen pressure of 550 mm., and a shaking
`frequency of 650 per minute, the values of 10 6I'd, l0 6I'g, and 10 6(rL- 1'n)/k.,
`were 27, 19, and 16 molejl., respectively.
`In the " low " pressure region, where rL and rn themselves vary with
`changes in [0 2], an exceedingly complicated situation exists,14 and deriva(cid:173)
`tions of the relevant constants are subject to much greater uncertainty.
`Fairly reliable estimates of the several constants for ethyllinoleate and
`digeranyl are given in Table 3. The values of k3 and k 6 are believed to be
`numerically significant, those of k2 , k4 , and k 5 express the order of magnitude.
`
`TABLE 3. Velocity coe.fficients at 25° (mole-1 l. sec. - 1)
`
`Ethyllinoleate
`Digeranyl
`
`10-•k,
`
`10
`1
`
`k,
`
`50
`3
`
`10-•~;.
`
`I0-1k 4
`
`10-•k.
`
`20
`1
`
`50
`10
`
`20
`10
`
`in circum(cid:173)
`involves measurements
`technique
`The rotating-sector
`stances where changes from rL to rn to rL occur in rapid succession. The
`terms thus cancel out automatically. Even under high
`(rL- rn)/k8
`pressure conditions (as above, a complex situation prevails at "low"
`pressures), the advantage which this confers has not been realised owing
`to a lack of sensitivity in other respects, but practical improvements to
`remedy this appear feasible and worth developing.
`
`3. Autocatalysis and Hydroperoxide Decomposition
`Benzoyl peroxide and azoisobutyronitrile undergo unimolecular thermal
`dissociation into free radicals and .catalyse the oxidation of ole:fins propor(cid:173)
`tionally to the square root of their concentration. This affords critical
`evidence, in conjunction with photocatalysis (p. 148), for the form of r1 in
`equation (3) and its simplified versions.
`Bimolecular Bydroperoxide Decomposition.-The autocatalytic character
`of the oxidation is illustrated in Fig. 5. The overall rate is proportional
`to the hydroperoxide produced during the earlier stages of the reaction,
`and thus from equation (3) r1 cc [R02H] 2, i.e., chain initiation ensues from
`a bimolecular decomposition of the hydroperoxide. This result was unex(cid:173)
`pected when first encountered because saturated and arylated hydroperoxides
`had previously been said to undergo a unimolecular primary scission. The
`self-consistency of the kinetic data on oxidation catalysis and a direct study
`L
`
`Page 9 of 21
`
`Noven Ex. 1020
`
`
`
`156
`
`QUARTERLY REVIEWS
`
`View Atiicle Online
`
`3
`
`(,)
`
`1 •
`
`~
`... ~2
`......
`..!!
`0
`~
`J-1
`~
`
`s
`
`2
`
`10
`
`w
`5
`~
`[0-t]aosorbed (mole/l)
`FIG. 5
`Autoa-idaNon of <'yclohexene at 45° and 728 mm.
`
`n
`
`of the decomposition of an olefinic hydroperoxide,15 however, combine to
`establish its validity. As mentioned later, differences concerning the order
`of peroxide decomposition may not be antagonistic.
`].,or the above catalytic form in the "high" pressure region, we have
`(5)
`r = {ek"[R02H]2}lk3k6-1[RH]
`where k" is the bimolecular velocity coefficient for the hydroperoxide
`decomposition and e represents the efficiency with which the liberated
`radieals produce R· or R02• radicals. From benzoyl-catalysed oxidations,
`
`0
`
`Olefin
`
`Allyl benzene
`Oct-1-ene .
`
`Methyl oleate .
`cycloHexene
`Ethyllinoleate
`
`4-Methylhept-3-ene
`1-Methylcyclohexene .
`
`1 : 3 : 5-Trimethylcyclohexelw
`Dicyclohex-2-enyl .
`
`Squalene
`Digeranyl .
`
`TABLE 4
`
`Hydroperoxide·type, %
`
`pnm.
`
`Bee.
`
`tert.
`-- - - -
`100
`30
`70
`
`100
`100
`100
`
`70
`70
`
`15
`25
`
`30
`30
`
`85
`75
`
`.-.100
`, .•• ,J 00
`
`104kake -i *
`(molc-i 1.! sec.
`
`108ck" •
`(mole- 1 1. sec.-')
`
`l)
`
`14·4
`3·6
`
`21·5
`37·0
`278·
`
`32·4
`6iH
`
`150
`200
`
`39·7
`49·4
`
`0·28
`0·29
`
`0·46
`0·54
`0·47
`
`1·72
`1·14
`
`3·25
`2·48
`
`2·97
`2·62
`
`15 Bateman and (Mrs:) Hughes, J., 1952, 4594.
`
`Page 10 of 21
`
`Noven Ex. 1020
`
`
`
`View Article Online
`157
`BATEMAN : OLEFIN OXIDATION
`and determination of the appropriate value of e for this system, 16 k3k6 -l can
`In Table 4, 17 the olefins are grouped
`be determined and thus (ek") from (5).
`according to the structural type of hydroperoxide involved, as deduced
`from the relative susceptibility of non-equivaJent <X-methylenic 0-H bonds
`to attack by R02• radicals (p. 149) and from the tendency of allylic systems
`to form isomeric products (p. 166).
`No parallelism is apparent between the differences in ek'' and the
`oxidisabiliti~s of the olefin (k3k6 -l), but a clear correlation exists with
`hydroperoxide type in the sense ek"1nim. : ek"sec. : ek"te1t. = 1 : 2 : 14. For
`reasons unknown, this order is the reverse of the commonly rc<'ogniHed
`stability of analogous saturated hydroperoxides.
`
`16
`
`2
`
`z 4
`6
`10 [02]ahsorbed(molejl.)
`FIG. 6
`Auto:ddation of (a) tetralin at 75° and 180 mm. and (b) 1-methylcyclohexene at 65° and
`350 mtn. at low extents of oxidation.
`
`8
`
`10
`
`Unimolecular Hydroperoxide Decomposition.-A curious feature of the
`plots of r against [02]absorbed (such as in Fig. 5) is that extrapolation of
`the linear portion to [02]absorbed = 0 gives a small but real intercept on the
`r-axis. This was first thought to represent the rate of the direct reaction
`between the olefin and oxygen (RH + 0 2 -+ ).1
`In fact, the basis of
`performing the extrapolation has proved fallacious. The true behaviour
`is shown for two olefins in Fig. 6. The curvature towards the origin in
`the very early stages of the reaction denotes catalysis of the form
`roc [R02H]l, instead of the commonly observed roc [R02H]. This in turn
`implies that the hydroperoxide at low concentrations ( < I0- 2 mole/!. in
`the temperature range studied) yields radicals by a first-order decomposition,
`which is superseded by a second-order decomposition at a higher concen(cid:173)
`tration. This unique change in decomposition order with concentration
`16 Bateman and Morris, Trana. Faraday Soc., 1952, 48, 1149.
`17 Morris, Ph.D. Thesis, London, 1952.
`
`Page 11 of 21
`
`Noven Ex. 1020
`
`
`
`158
`
`QUARTERLY REVIEWS
`
`appears to be associated with the state of molecular association of the
`hydroperoxide :
`
`View Article Online
`
`H
`2(R0 2H) ~ (RO·O ..... H•0 2R)
`
`unimol./
`
`bimol.
`"""
`decomp.
`~decomp.
`RO· + H 20 + R0 2•
`2(RO· + ·OH)
`Infra-red spectroscopy provides clear-cut evidence that the .inteuse
`associatiou in the neat hydroperoxide becomes progressively less with
`dilution.15 In general, conditions conducive to a low degree of association
`would be expected to emphasise the first-order decomposition. Consisteutly,
`the catalytic form r' a:: [R02H]~ persists to higher concentrations at higher
`temperatures, 17 and the addition of more strongly bonding substances than
`the peroxide itself suffices to change the observed catalysis from r a:: [R0 2H]
`to r a:: [R02H]l,15 , 17 The effect of temperature is significant in providing
`a probable explauation of the differences in hydroperoxide decomposition
`reported by different workers (of. p. 155). The inference from the oxidation
`kinetics of a bimolecular mechanism relates to temperatures lower than
`about 80°, while the direct decompositions have mostly been studied at
`above 130° where the first-order dissociation will be greatly favoured.
`
`TABLE 5
`
`Olefin
`
`Temp.
`
`107e'k'
`(sec.-1)
`
`Oleftn
`
`Temp.
`
`107e'k'
`(sec. - 1)
`
`0
`
`Allyl benzene
`cycloHexene .
`Ethyllinoleate
`1-Methylcyclohex-1-ene
`"
`1 : 3 : 5-Trimethyl-
`cyclohex -1-ene
`
`75°
`55
`55
`45
`65
`65
`
`2·9
`0·1 6
`0·20
`0·45
`3·9
`3·7
`
`Dicyclohex-2-enyl
`"
`2-Methyloct-2-ene .
`2-Methyl-4-phenyl-
`but-2-ene
`Digeranyl
`
`45°
`65
`55
`55
`
`45
`
`0·7
`6·9
`0·50
`1·6
`
`0·5,
`
`The quantity e'k', analogous to ek", can likewise be determined for the
`first-order hydroperoxide initiation process (Table 5).17 For the limited
`data available, no well-defined correlation with hydroperoxide type as in
`the case of ek" can be recognised, but the predominantly tertiary derivatives
`again seem to be the more reactive.
`
`4. The Direct Reaction between an Olefin and Oxygen
`A natural consequence of the free-radical character of oxygen is that it
`should display in some measure the reactivity of R02• radicals towards
`olefins.
`It is actually so much less reactive that direct olefin-oxygen inter(cid:173)
`action (RH + 0 2 --+) has so far proved impossible to measure. As described
`in section (3), instead of being able to define this rate relatively easily by
`extrapolating the plot of r against [02]absorbed (as l!,ig. 5) to [02]absorbed = 0
`(e.g., 6 x l0- 6 mole I.-1 sec.-1 for tetralin at 75°), the true value is so many
`times smaller that it is difficult to observe (see Fig. 6). The absorption of
`
`Page 12 of 21
`
`Noven Ex. 1020
`
`
`
`BATEMAN : OLEFIN OXIDATION
`
`View Article Online
`159
`as little as 0·02-0·05 ml. (N.T.P.) of oxygen per ml. produces a degree of
`hydroperoxide catalysis sufficient to obscure any possible initiation by
`direct olefin-oxygen interaction.17 ' 18 The latter clearly cannot be an
`observable component of the overall oxidation reaction at moderate tem(cid:173)
`peratures. The initiation step as a whole accounts for only Ijnth of the
`total products, where n is the chain length, and hydroperoxide decomposition
`accounts for nearly all of this fraction.
`Obviously no examination of the oxidation product can hope to provide
`information on any non-hydroperoxidic initiation.
`In fact it would appear
`that the only means of obtaining critical evidence on the direct olefin-oxygen
`reaction is to study the system in the presence of a highly efficient inhibitor
`which would prevent any primary peroxy-intermediate from becoming a
`hydroperoxide (therefore not a phenolic-type inhibitor).
`Hydroperoxide initiation is likewise predominant in photochemical
`oxidations. The formation of a very small amount of hydroperoxide
`has been shown quantitatively to establish R02H + hv~ rather than
`RH + hv~ as the primary activation process. 3
`5. Metallic-ion Catalysis
`The intense activity of certain metallic compound~ (notably those of
`iron, cobalt, nickel, copper, and manganese) as oxidation catalysts is a
`matter of immense technological interest. The consequences can be both
`highly undesirable and advantageous. Thus the comparatively rapid
`oxidative deterioration induced in rubber or lubricating oils calls for strict
`preventive measures; while the use of cobalt compounds a<s "driers'' to
`promote the rapid oxidative hardening of unsaturated esters is all-important
`in paint technology. Although knowledge of how these compounds act
`remains obscure in many details, the general picture is fairly clear. Of
`particular interest in the present context are certain distinctive kinetic
`characteristics.
`The active metals are those having two or more valency states, clearly
`suggesting that an oxidation-reduction process is involved. They function
`via their ions, as is evident from the industrial practice of using the so-called
`sequestering (complex-forming) agents (e.g. ethylenediaminetetra-acetic acid)
`to counteract metallic contamination-the metal is converted from an ionic
`into a chelated form and thereby rendered innocuous. Obtaining of a suit(cid:173)
`able homogeneous reaction system for mechanistic studies is thus a difficulty.
`The solvent employed so far has been acetic acid, which is a catalyst for
`hydroperoxide decomposition and therefore might be expected to create
`confusion in any direct comparison of results with those obtained for oxida(cid:173)
`tion in hydrocarbon solvents. Whether this is so or not remains to be
`proved, but fortunately the catalysis by active metal salts is so great that
`the reaction can be studied under conditions where oxidation in acetic acid
`alone is negligible.
`Working with cobaltous acetate in acetic acid, Bawn and his
`
`18 Bateman, (Mrs.) Hughes, and Morris, Discuss. Faraday Soc., 1953, 14, 190.
`
`0
`
`Page 13 of 21
`
`Noven Ex. 1020
`
`
`
`160
`
`QUARTERLY REVIEWS
`
`View Article Online
`
`co-workers 19 • 20 conclude that the rate-determining initiation process is
`R0 2H + MS+ ~ R0 2• + MH + H+ ... r1(M)
`where 1\.P+ represents a complex tervalent cobaltic ion. The cobaltous
`ion produced is immediately reconverted into MS+ by the much faster
`reaction:
`R0 2H + M2+ ~ RO· + Ma+ + OR-
`The sum of these consecutive reactions is seen to be exactly the bimolecular
`decomposition pattern proposed for the hydroperoxide by itself:
`2R02H~R02 • + RO• + H 20
`cf. p. 158, and the truly catalytic role of the metal salt is readily apparent.
`Oxidation subsequently proceeds by the ordinary mechanism, i.e., in(cid:173)
`In
`volving reactions (2), (3), and (6) under the conditions employed.
`conformity with this, the reaction shows autocatalysis in the earlier stages
`and hydroperoxide is steadily formed. However, since the catalyst pro(cid:173)
`motes the decomposition of the hydroperoxide so strongly, r1(M) will rapidly
`increase, the chain length {r jr1(M)} will decrease correspondingly, and we
`should expect the reaction soon to lose its chain character. For these
`circumstances, the formation and decomposition of the hydroperoxide
`become equal. This can arise, of course, independently of the mode of
`initiation, and the generalised kinetic changes and their detection experi(cid:173)
`mentally have been discussed in detail by Tobolsky and his co-workers. 21
`Under these conditions, the products of hydroperoxide decomposition and
`of inter-radical reactions, essentially (6), form a major part of the total
`products.
`The maximum limiting rate of oxidation at a given temperature when the
`chain length tends to one follows from eqn. (2) as rum.= k3
`2k6 - 1(RH]2.