C O M B U S T I O N A N D F L A M E 5 2 : 2 5 - 3 5 (1983) 25

Chemical Kinetics of the Oxidation of Methyl Tert-Butyi Ether (MTBE)

J. C. BROCARD, F. BARONNET, and H. E. O'NEAL*

Ddpartment de Chimie Physique des Rdactions, INPL-ENSIC 1, rue Grandville 54042 Nancy, France

An experimentalinvestigationin a conventionalstatic apparatusof the oxidationof equim01ecularmethyltea-

butylether (MTBE)-oxygen mixtures has shown that MTBE oxidizes far less readily than alkanes like
neopentane or 2,2-dimethylhotane. In the investigatedpressure (up to 1000 Tort) and temperature ranges (300-
500"C) we have not observed any cool flame formation; the mixtures either react slowly or ignite. Such
behaviour is in line with the well-known antiknock effect of MTBE since this ether is already largely used as
an octane booster in commercial gasolines, especiallyin Germanyand in the UnitedStates.
An evaluationof the rate constantsof the possible elementarysteps by the methods of thermochemicalkinetics
has allowed us to propose a reaction scheme which is in satisfactoryagreementwith our experimentalresults. It
is shownthat the reactionof MTBE-oxygen mixturesaround450°C does not reallypresentthe characteristicsof a
genuineoxidation reaction and appears more as a thermal degradation. This reaction chiefly gives
formaldehydeand rather unreactive t-C4 H9" radicalswhich react with oxygento give i-C4H8 and HO2 ". The
HO2" radicalscannothe efficientpropagating radicals in oxidationreactions.
In the presence of 2% added HBr, the ignitionlimits of equimolecularmixtures of MTBE and 02 are
considerably lowered (~500C), and with 10% added HBr, the usual cool flame behaviorof
hydrocarbonoxidationsare similarly observed for MTBE/O: mixtures.

INTRODUCTION
The expanding use of MTBE (methyl tert - butyl ether, 2 -
methoxy, 2-methyl propane) as a gasoline component
suggested to us a fundamental investi-gation of the
oxidation of this compound, aiming at a better
understanding of its rather high octane number .(octane
blending value: 115 to 136 RON [1 ] ). MTBE properties
are well adapted to its use in gasoline and it is easily
manufactured b y a sim-ple reaction between isobutene and
methanol. The present capacity of MTBE units is over
1,300,000 tons per year in the United States and 600,000
tons per year in Europe [2].
We had already investigated the pyrolysis of MTBE [3]
and found in agreement with Daly and
*Visiting professor. Current address: Department of
Chemistry, San Diego State University, San Diego, CA
92182.
Wentrup [4] and with Choo et al. [5] that the re-action is a
four-center elimination, very sensitive to wall effects.
EXPERIMENTAL PROCEDURE
MTBE was purchased from Fluka (purum, 98%) or from
Merck (99%). Equimolecular mixtures MTBE/oxygen were
admitted in an evacuated cylindrical reaction vessel made
of Pyrex glass (volume 350 ml) incorporated in a
conventional static reaction system. Temperatures ranged
from 300 to 500 ° and experiments were performed at
pressures up to 1 arm b y ufing an injection device similar
to those used in gas chromatographs and also proposed by
Lucquin and coworkers [6] in cool flame investigations.
The reaction was fol-lowed by means of a 0 - 200 Torr
pressure trans-ducer (SEDEME PR 200) and a time gauge
thermo-couple positioned at the center o f the reaction yes-

CopyrightI~) 1983by The CombustionInstitute

Published by ElsevierSciencePublishingCo., Inc. O010-2180/83/503.00
52 VanderbiltAvenue, New York, NY 10017
26
sel. The thermocouple, made of fine chromel-alumel
wires (80/an) , was not catalytic, in agree-ment with
Ben-Aim [7].
Reaction products were usually analyzed by gas
chromatography: formaldehyde was analyzed by
colorimetry, water by using a reactive column o f
C2Ca and H202 by polarography.
EXPERIMENTAL RESULTS
Pressure-TemperatureDiagrams
The oxidation of MTBE is much slower than that of
alkanes of similar structure. By measuring the
consumption of reactants, we have observed that, in
the case o f equimolecular mixtures, the rate o f
oxidation o f MTBE at 450°C was approximately the
same as that of 2,2-dimethylbutane at ca. 300°C.
J . C . BROCARD ET AL.
The low reactivity of MTBE, as compared to that
o f alkanes, is also apparent from a pressure-
temperature diagram. Thus, in the oxidation of
neopentane (2,2-dimethylpropane), a cool flame
region was observed by Fish [8] in the usual ranges o
f pressure and temperature with the mixture 1 neo
C5Hi2 + 2 0 2 .
Our experimental results (slow reaction-cool
flame limit) in the case o f equimolecular neopen-
tane-oxygen mixtures are quite similar to those of
Fish (Fig. 1). This comparison also shows that our
apparatus gives experimental results which are close
to those o f other authors.
On the contrary, in the case of MTBE, we do not
observe any cool flame region, even when our
investigation is extended with the injection tech-
nique towards higher temperatures and pressures. We
obtain only
Fig. 1. Pressure-temperaturediagramof neopentaneoxidation: - - , 1 ; 2
neopentane-oxygen mixtures (from Fish [ 8 ] ) ; - - - , 1:1 neopentane-
oxygenmixtures (present work).
OXIDATION OF METHYL TERT-BUTYL ETHER 27

12s
24s
Fig. 2. Temperature (AT) and pressure rise (zkP) prof'~es in the oxidation of equi-
molecular MTBE-O 2 mixtures: , zXT; - - - , ,aP; A , slow reaction (T = 470°C, total
pressure = 168 Torr); B, ignition (T - 462°C, total pressure = 192 Torr).

liquid phase; its vaporization tends to cool down

1. a slow reaction (Fig. 2A)
2. or an ignition (Fig. 2B).
The beginning o f the A T versus time curve is
not significant because MTBE is injected in the
the reaction vessel, which can explain the discrep-
ancy between its initial and final baselines in the A
T-time profiles. Our experimental pressure tem-
perature diagram is shown in Fig. 3.

T °C

500, \ O,. O

400.

---.<..o
Slow Combustion
~ 0 ~

300.
P (roar)
>
0 ' 2~o ' 400' ' 6~o ' 8~0 ' 1000'

Fig. 3. Pressure-temperature diagram of ignition limits for equimolecuiar

MTBE-O 2 mixtures.
28 J . C . BROCARD ET AL.

500~ T°C

450_ o/

~I~~"° ~ MTBE + 2 °/oHBr

e ~ ~ O ~ 0 ...~._

400_
"~NxMTBE +10~o HBr"

I \
Slow combustion ]
350_ ignition

1 £lames

300_
•~•
__ -m.
100 200 3 0
P (tort)
Fig. 4. Influenceof HBr upon the ignition limits of equimolecularMTBE--O2 mixtures.

The absence o f cool flames under our experi-
mental conditions is quite consistent with the anti-
knock effect of MTBE, considering the connec-tions
between cool flames and two-stage ignition put
forward by Fish, Affleck, and coworkers [9] and also
the rfiathematical model for hydrocarbon
autoignition proposed by Quinn and coworkers
[10] . It is worth noting that these results do not mean
that cool flames cannot be observed under more
severe conditions, for instance, at higher pres-sures.
Such an experimental investigation, however, would
require a different reactor. The absence o f a negative
temperature coefficient for the slow reac-tion
confirms the previous experimental results.
However, we have observed a very different pic-
ture when the oxidation of MTBE is investigated in
the presence o f relatively small amounts o f HBr
(Fig. 4). In the presence o f 2% added HBr, the slow
reaction/ignition limit is lowered b y approxi-mately
50°C. When we add 10% HBr, this limit is again
lowered and we observe cool flames (A T -~ 50°C)
for initial total pressure around 200 Torr and at initial
temperatures around 300°C (i.e., in the usual pressure
and temperature ranges where cool flames are
observed with alkanes and many other organic
substances, such as ketones and esters).
Reaction Products
We have not tried to detect and to analyze all the
reaction products, however we have measured the
amounts o f the major products and the con-sumption
of the reactants: oxygen and MTBE. Our
OXIDATION OF METHYL TERT-BUTYL ETHER 29

experimental results are plotted in Fig. 5. They show I P (for'r) P

that [... .,ion produds] [MTBE]A/ 54

1. i-C4Hs and CH20 are the major primary

8. H20/
products;
2. the slope of the curve accounting for MTBE
52
consumption is nearly equal to that o f i-C4Hs
formation.
We have also observed that the yields o f metha-
6 /J ic<H/

nol are approximately the same as in the pyrolysis "'-* MTBE / / 50

reaction and that there is an excess o f isobutene over
formaldehyde. Despite the scattering of the
experimental points, it appears that the difference
(isobutene) - (formaldehyde) is o f the same order of
magnitude as the methanol yields. If we com-pare
these results with those obtained in the case of the
pure MTBE pyrolysis reaction under the same
experimental conditions [3], it appears that all o f the
methanol formed can be accounted for by the 4-
center molecular reaction (i.e. the pyroly-sis
reaction):
MTBE (C5H120) = i-C4H a +CHaOH,
k ~- 1013.6e-59,0001RT S-1 .
Therefore, the formaldehyde and excess isobutene
yields are measures o f the more complex oxidation
processes induced by the presence o f O 2.
INTERPRETATION AND DISCUSSION
The experimental results on the thermal reactions o f
MTBE in the presence and absence o f O z can be
rationalized in terms o f the reactions shown in Fig.
6. The thermochemistries o f the pertinent species are
given in Table 1, and thermochemical kinetic
estimates o f the relevant rate constants are given in
Table 2. The major aspects o f the system can be
understood in terms o f just a few "domi-nant"
reactions, namely, the isomerization and de-
composition reaction channels (1), (2), ( - 2 ) , (3) o f
the two radicals (R. = CHaOC(CHa)2CHz.; R' . = t-
BuOCH2") generated by H-abstraction from the
MTBE reactant. This follows from the fact that at
reaction temperatures the oxygen-trap-
/Zo "
, ~ .48
d'\
0 2 4 6 8 10
t, (mi,.)
Fig. 5. Slow oxidation of equimolecular MTBE-O2
mix-tures; product yields and reactant consumption
versus time T = 453°C; total pressure = 100 Torr.
ping reactions (4), (5) are an order of magnitude or
more slower than these dominant reactions. Thus the
reaction channels (6), ( - 6 ) , (8); (10), ( - i 0 ) , (11);
and (7), ( - 7 ) , ( 9 ) a r e processes that would be
more important in oxidations o f MTBE carried out at
lower temperatures. However, they are not very
important in the reaction o f MTBE/Oz mixtures at
temperatures employed in this study (i.e., T =
726°K).
Returning to the dominant reactions, we see that
R" and R' . radicals (Table 1) can interconvert via a
five-center transition state. This isomerization is
comparable in rate to the R' . decomposition via
reaction (3) to t-Bu. + CH 2 = 0 (and eventually to
isobutene and HO z .), and several times faster than
the decomposition via reaction (1) o f R- to CHsO" +
isobutene. Thus the estimated reaction
to products o f R" radicals via channels (1), (8), (3),
and (11) are [ignoring terms in the denominators
of the radical steady state expressions which are
 30 J . C . BROCARD ET AL.

CH20 + *OH CH20 + HO 2"

c.3, t ?Ha t
(~Id2=COCH 3 ÷ "CH 2 0 0 H CH 2 = CCH2OOPI + CH30 . CH2=C(CH3) 2 ÷ .OCH2OOH

(91

CH 3 c%
CMZ-I?-OCH3 CH3--C--O-- CH2--O--OH
• CH 2
O

C H 3 ) 2 C = CH • HO 2"

II0/
(l$omerizatlon;
size or transitio n (I $ omerization ;
c.2o • (c.sh ~c.2oo. siZe or transition
stat e ring : 6)
i r a t e rln9 : 7)

1(11)
~"~, c% CH 3
I
• CH 3 - C - O C H 2" CH3--C-O--CH2--O--O.
CH2OOH
(1o)
c . 3 - c - oc . 3
(( - Io )
CHZ
.O-O/

*I[+02] /[.o2]

CH 3 (2) CH 3
I
CHs-- IC--OCH s I
(-2) c.,

('1) [ D e c o m p o s i l i o . . . . . po,;,ion (3)

CH 3 CH
I
CHS- ~ ÷ OCN 3 CH3 -- C . + CH20
I
CH 2 CH 3

HO 2" + i-C4H 8
[+%1
Fig. 6. Schematic mechanism of MTBE oxidation.

less than 10% o f the dominant term(s)]

Radical generated Reaction channel Rate expression1 Rate (numerical)
CHaO. (1) k I (R.) 1.7 X 107 (R.)
CH30. (8) k8K 6 (R.) 2.2 X 105 (R.)
HO2. (3) k 2 (R.) 5.5 X 107 (R.)
 HO2. (11) klo (R.) 2.2 X 105 (R,)

k 5 (R.) (02)
I Note that at the steady state (RO2-)= = 1.01 ( R . ) .
k8K 6 + k l o + k _ 5
OXIDATION OF M ETHYL TERT - BUTYL ETHER 31

TABLE 1

Species AHf (kcal/mole) a ~o (cal/mole °K)a

t-BuOCH3 -69.0 85.8

CH2=C(CH3)2 -3.8 70.0
CH20 -26.0 52.3
R., CHaOC(CHa)2CH 2. -23.1 89.7 d
R'., t-BuOCH2. -27.1 87.4 c
CHaOC(CH3)2CH2OO. -52.2 105.7 b
CH3OC (CH3) (CH2OOH)CH2 • -41.7 111.8 a
•CH2OOH 9.9 69.1 c
CH2=C(CHa)OCHa -34.3 81.0
CH30. 3.5 55.0 e
CH2 =C(CHa)CH2OOH - 20.9 92.0
(CH3)2(CH2OOH)COCH 2. -45.8 110.4 c
t-BuOCH2OO. -60.2 103.1b
HOOCH2OC(CHa)2CH 2. -49.8 110.0 d
HO2. 3.0 54.4 b
t-Bu, 8.7 72.1f
CH302. 4.2 64.4 b
CH3OH -48.0 57.3

a Heats of formation and entropies (all at 298°K) were calculated from group additivity relations [14].
b Based on DH° (ROO-H) = 87.6 kcal/mole [15].
c Based on DH°(ROCH2-H) = 94.0 kcal/mole [13, 14].
d Based on DH°(RCH2-H) = 98.0 kcal/mole [14].
e Based on DH°(RO-H) = 104.0 kcal/mole [14].
f Based on DH°(t-Bu-H) = 92.0 kcal/mole [ 14].

These numbers suggest that only about 24% of the R. CHaO" + 02 ( l a ) C H 2 0 + HO2",
radicals formed in propagation reactions de-compose to the
kl3 ~ 108.5 X e - 4 0 0 0 / R T M- 1 s-l,
"active" chain carrier CHaO . radi-cal. The remainder
produce the relatively inert HO2" radical. In addition, the CHaO- + MTBE t14)
> t--BuOCH2. + CHaOH ,
chain propagation ability of CHaO . is moderated in our
k14 --- 10s.a × e-5000/nTM-~ s-1.
system b y its metathesis reaction with 02 [reaction (13)] .
In a similar fashion, the rate constants of Table 2 indicate
The latter competes more or less equally with that the R' . radicals react almost exclu-sively to HO 2 • and
H-abstraction [reaction (14)] : isobutene . Thus,

Radical generated Reaction channel Rate expression Rate (numerical)

HO2" (3) k 3 (R'.) 2.7 × 108 (R'.)
HO2 • (9) k7K 4 (02) (R'.) 1.8 X 105 (R'.)
k _ 2 k I (R'.)
CH30. (1) 1.3 × 106 (R'.)
k2+kI

Here, more than 99% of the product radicals are the bonds o f the t - butyl group: 94 kcal/mole for the former
"inert" HO 2 • radicals. [13, 1 4 ] ; 9 8 kcal/mole for the latter [14] .
The (C-H) b o n d in the methoxy group is sig-nificantly This means that radical attack on MTBE will favor
weaker (by 4 kcal/mole) than the (C-H) production o f R " radicals over R .. In fact, if
 32 J . C . B R O C A R D ET AL .

TABLE 2

Thermochemical Kinetic Rate Constant Estimatesa

E (kcal/ E (kcal/
Reactions A mole) k (726°K, s- 1 ) Reactions A mole) k (726°K, s--l)

(1) 1014.7 s- 1 24.8 1.7 X 107 (7) 1012.o s--1 26.4 1.1 X 104
(-1) 109.3 M--1 s--1 2.0 5.0 × 108 (-7) 1010.5 s--1 16.0 4.6 X 105
(2) 1012.5 s- 1 15.8 5.5 × 107 (8) 1015.1 s--1 26.3 1.5 X 107
(-2) 1013.o s- 1 20.8 5.4 × 106 (-8) 109.3 M- I s- I 2.0 5.0 X 108
(3) 1014"4 s- 1 19.8 2.7 × 108 (9) 1015.1 s- 1 26.3 1.5 x 107
(-3) 108.5 M--1 s--1 10.0 2.9 X 105 (-9) 109.3 M- I s- 1 2.0 5.0 X 108
(4) 109.5 M- 1 s - 10.0 109.5 (02) = 106.5 (10) 1011.8 s--1 21.4 2.2 x 105
(-4) 1014'5 s- 1 30.6 1.8 × 105 ( - 10) 1011.3 s- 1 16.0 3.0 X 106
(5) 109.5 M--1 s--10.0 109.5 (02) = 106.5 (I 1) 1014.4 s- 1 19.8 2.7 X 108
(-5) 1014.8 s- 1 27.8 2.6 x 106 ( - 11) 106.5 M- 1 s- 1 10.0 2.9 X 105
(6) 1013'4 s- 1 21.0 1.2 × 107 (12) 1014.3 s- 1 32.7 2.9 x 104
(-6) 1012"1s- 1 10.6 8.1 x 108 ( - 12) 107.5 M--1 s--1 15.4 6.9 x 102

a Reactions( - 1), ( - 8 ) , ( - 9 ) are RO. additions (Mar- A-factors for reactions (2), (6), (7), (10) were obtained from the
kownikoff) to isobutenes. These strong electrophiles should add approximate relation,
with quite low activation energies [12], and with "looser" than
normal tight transition states [14]. Thermochemical kinetic
estimates for the "normal" A-factor of addition give A -~ 108.3
A =- X r p d X exp s- 1 ,
M- 1 s- 1 , and we have set these values higher by a factor of 10 h R
(A 1 = A _ 8 = A _ 9 = 109.3 M- 1 s--I). Reactions ( - 3 ) and ( -
1 1 ) are t-butyl-type radical additions to the oxygen side of where rpd = reaction path degeneracy = number of ab-
CH2=O. While addition to the carbon of the carbonyl is facile stractable H atoms, and Anirot ~ = change in the number of
(Eae t -~ 2 kcal) [12], addition to the oxygen is much more internal rotations as reactant moves to transition state. The
difficult. We estimate E _ 13 -~ E_11 -~ 10.0 kcal. This is reverse reaction A-factors were then calculated from the
consistent with the experi-mental activation [16] energy and reaction entropies. For example,
enthalpy [14] of the methoxymethyl radical decomposition:
ekT f(2 - 5 + 1)3.5_1
A2=- X 3 X exp - 1012.5s -- l;
h R
f
A _ 2 = (1012.5 X e2.3/R) = 1013.0 s- 1 .
CH3OCH 2. ~- >CH 3 . + C H 2 0 , AH ° =9.4 kcal; b
Activation energies for the isomerization reactions were set
Ef = 19 kcal/mole; E b = 9.6 kcal/mole. equal to the activation energies for H-abstraction of the
analogous radical + substrate bimolecular reaction plus the
The t-Bu. radical additions to CH20 have the same strain energy of the cyclic transition state. The latter were taken
thermochemistry, and it is reasonable to assume similar to be 1.0 kcal (6-center), 6.3 kcal (5-center), and 6.4 kcal (7-
kinetics. center). To illustrate,
A-factors and activation energies of the decomposition
reactions (1), (3), (8), (9), (11), (12)were obtained from their E 2 = Eaet(C2H 5. + CH3OCH3) + Estrain(5.eenter)
back Arrhenius parameters and their thermochemis-tries. = 9.5 + 6.3 = 15.8 kcal/mole.

The intramolecular H-abstraction, radical isomeriza-tion Activation energies of RO 2. abstraction of H were taken to be
reactions proceed via 5-center (2), ( - 2 ) , 6-center (6), ( - 6 ) , 14 kcal (tertiary H), 17 keal (secondary H), and 20 kcal
and 7-center (10), (-10), (7), ( - 7 ) , transition states. (primary H) in accord with suggestions by Benson and Nangia [
11 ].
OXIDATIO N OF METHYL TERT - BUTYL ETHE R
the propagating radical is HO 2 ., R' . will be favored
over R" by a factor o f about ( 1 / 3 ) e 4 ° ° ° l a w = 5.4 z,
a at 453°C . It follows, then, that the relatively low
reactivity o f MTBE to induce free radical de-
composition is directly due to the near quantita-tive ( ' 9
5 % ) productio n o f HO 2. radicals in its product
formation reactions. And, as noted b y Nangia and
Benson [ 11 ] a, HO z . radicals c a n n o t
t-BuOt-Bu
R" + t-BuOt-Bu ~- RH + t-BuOC(CHs)2CH 2.
chain t- BuOC(CHa)2 CH 2 •
t-BuO.
(chain carrier is R . -- CH a .);
CHaOCH3
chain [ R . + CH aOCH a
/ CHsOCH 2"
(chain carrier is R- = CH a .).
Thus thermal degradation o f bot h ethers pro-duce
methyl radicals which are goo d chain carriers at high
temperatures in the absence or presence o f oxygen (i.e.,
H-abstractions b y methyl radicals are relatively fast, and
at high temperatures oxygen is no t effective trap for CH
a • because o f the reversi-bility o f the trapping reaction,
CH a - + 0 2
C H a 0 2 . ). The symmetric ethers, therefore,
thermally degrade at muc h faster rates than the
unsymmetric MTBE 4. It follows that MTBE is a good
antiknock additive because it serves as a
2 The back activation energies for R. and R'. with H202 are
undoubtedly very similar, hence Eact for R. + H202
should be larger than that for R'. by the reaction endo-
thermicity difference; the factor 1/3 arises from reac-tion
path degeneracy: 9H to give R. and 3H to give
R'%
a The RO2-H bond dissociation energy is relatively low
(~87.6 kcal/mole); hence almost all H-abstraction reac-
tions of.HO2" are endotherrnic, high activation energy
processes. Thus Benson estimatesEae t (HO2 ° abstraction
of tertiary H) ~, 14 kcal [11]. This suggests Eaet ~ 17 kcal
(secondary H), and Eae t ;~ 20 kcal (primary H).
4 We have observed that, conl~ary to that of MTBE, an
addition of di-tert butyl ether does not increase the in-
duction period of cool flames in alkane oxidation [17].
33
propagate fast chain reactions.
It is interesting to compare the free radical chain
reaction pathways o f the symmetric ethers, di-tert-butyl
ether (t-BuOt-Bu) and dimethyl ether (CHaOCHa), with
that o f the mixed ether o f this study. The radicals
produced by H-abstraction from these symmetric ethers
and their subsequent decomposition products are
t-BuO. + CHg=C(CHa) 2
CHaCOCH a + CH a •
> RH + C H a 0 C H
2. C H 2 0 + CH a-
chain transfer agent which essentially metathe-sizes
"active" alkyl radicals to "inactive" HO 2-radicals.
Specifically, the antiknock character o f MTBE can be
traced directly to three kinetic and thermochemical
properties:
1. it has relatively weak C - H bonds on the m e t h
o x y methyl, which directs H-abstraction re-actions to
that site;
2. decomposition o f the radical produced b y H-
abstraction from the m e t h o x y group is signifi-cantly
faster than all competing processes, and produces the
rather inactive t-Bu • radical which in 02 generates the
even less active HO2" radical;
3. isomerization o f the radical produced b y H-
abstraction from the t - butyl group (to the radi-cal
formed b y H-abstraction from the m e t h o x y group) is
significantly faster than its decomposi-tion to the
"active" C H s O . radical.
Our results in the presence o f HBr are also
con-sistent with the proposed inhibition
mechanism as well as with the important kinetic
role o f the t-Bu-and HO2 • radicals. Thus, as
noted b y several authors [ 18, 19] HBr accelerates
reaction rates b y replacing the less active free
radicals (e.g., HO2", t - Bu . )
34
with the relatively reactive Br. atoms, e.g., reac-
tion (15):
HO z • + HBr ~ H 2 0 2 + Br' . (15)
The net result is the propagation o f long chain
processes, sustained in part b y Br- atoms, and
maintained at lower temperatures where the equi-
librium position o f reaction (5) favors RO2"
species. Oxidation pathways are thereby promoted,
and these can lead to OH . production, degenerate
chain branching, and the production o f cool flames
(as observed).
CONCLUSION
We believe that the previous mechanism is con-sistent
not only with the slow oxidation rate and the absence o f
cool flames but also with the in-hibiting influence o f
MTBE o n the oxidation o f other organic compounds,
such as alkanes, under usual conditions o f cool flame
formation (T -~ 300°C, pressure around 100 Torr). We
have shown that MTBE increases the induction period o
f cool flame formation in the oxidation o f n-pentane
[20] . An organic c o m p o u n d should fulfil the fol-
lowing conditions to inhibit an oxidation reaction o f
another organic substance:
1. It should be relatively stable thermally (i.e., it
should no t initiate chains as a result o f its ow n
decomposition).
2. It should be as or more reactive than the ma-jor
substitute toward free radical induced reactions. This
provides the mechanistic entry to overall reac-tion
inhibition.
3. It should react to produce relatively inert
(nonchain propagating) free radicals.
These conditions are fulfilled in the case o f well-
known inhibitors such as toluene or N-methyl aniline;
the unreactive radicals derived from these inhibitors are
readily obtained b y hydrogen ab-straction; these
substances are also thermally stable.
In the case o f MTBE, the formation o f unreac-tive
species is less straightforward, but we have shown that
the elementary steps transform nearly all radical species
derived from this ether into HO2" radicals which are not
reactive enough to
J . C . BROCARD ET AL.
propagate chain reactions. The oxidation o f MTBE
is also a rather difficult reaction, which suggests that
this ether cannot initiate radical chain reactions.
We think that it is possible to extend the basic
principle o f this interpretation to the antiknock ef-
fect o f MTBE in m o t o r engines, even if the ele-
mentar y steps can be somewhat different because
o f the different experimental conditions (especially
m u c h higher pressures). MTBE on its ow n is an
interesting gasoline component, as it does not
readily oxidize. It also seems able to produce
rather unreactive species which slow down the
propagation sequences responsible for autoig-
nition, since engine knoc k is a two-stage ignition
preceded and controlled b y a cool flame [9] .
REFERENCES
1. Stinson, S. C., Chem. Eng. News: 35 (1979).
2. Chauvel, A., Rev. Inst. Ft. Pdtrole 36:685 (1981).
3. Brocard, J. C., and Baronnet, F.,Oxid. Comm. 1:321
(1980).
4. Daly, N. J., and Wentrup,C . ,Aust . J. Chem. 21:2711
(1978).
5. Choo, K. Y., Golden, D. M., and Benson, S. W . ,lnt .
J. Chem. Kinetics 6:631 (1974).
6. Delfosse, L., Baillet, C., and Lucquin, M., C.R.
Acad. Sc. 272C:1446 (1971).
7. Karbassian, A., Ben-Aim, R. I., Chamboux, J., and
Viossat, V . ,Rev . Inst. Ft. Pdtrole 36:801 (1981).
8. Fish, A., Combust. Flame 13:23 (1969).
9. Affleck, W. S., and Fish, A., Combust. Flame 12:243
(1968). Fish, A., Read, I. A., Affleck, W. S., and
Haskeil, W. W., Combust. Flame 13:39 (1969). Fish,
A., Proc. Roy . Soc. A293:378 (1966); A298:204
(1967).
10. Halstead, M. P., Kirsch, L. J., Prothero, A., and
Quinn, C. P., Proc. Roy. Soc. A346:515 (1975).
11. Benson, S. W., and S. Nangia Prakash, Accounts o f
Chem. Research 12:223 (1979).
12. Choo, K. Y., and Benson, S. W.,Int. J. Chem. Kinetics
13:833 (1981).
13. Cruickshank, F. R., and Benson, S. W., Int. J. Chem.
Kinetics 1:381 (1969).
14. Benson, S. W., Thermochemical Kinetics, 2nd ed.,
John Wiley, New York, 1976.
15. Weissman, M., Shum, L. G. S., Heneghen, S. P., and
Benson, S. W.,J. o f Phys. Chem. 85:2863 (1981).
16. Marcus, R.A.,et al.,J. Chem. Phys. 16:987 (1948).
17. Baronnet, F., unpublished results.
18. CuUis, C. F., Fish, A., and Ward, R. B., Proc. Roy .
Soe. A276:527 (1963). Bastow, A. W., and Cullis,
C. F., Symposium on the Mechanisms o f Hydrocar-
O X I D A T I O N O F METHYL TERT - BUTY L ETHE R
bon Reactions, Siofok (Hungary), June 5-7, 1973,
pp. 765-778.
19. Hardacre, A., Skirrow, G., and Tipper, C. F. H.,
Combust. Flame 7:100 (1963); 9:53 (1965).
20. Baronnet, F., Brocard, J. C., Niclause, M., Ahmed, A.,
Vichnievsky, R., and Charpenet, L., Communica-
35
tion at the Colloque International Berthelot, Vieille,
Mallard, Le Chatelier, Universit6 de Bordeaux I
(France) July 20-24, 1981, Book of preprints, Vol. II,
p. 425.
Received 26 July 1982; revised 5 January 1983