Single-Pulse shock tube studies of the decomposition of ethoxy compounds

Article abstract of DOI:10.1021/jp970653a

Tetraethyl orthocarbonate (TEOC), diethyl carbonate (DEC), and diethoxymethane (DEM) have been decomposed in single-pulse shock tube experiments. TEOC decomposes to give DEC, ethylene, and ethanol as organic products, while DEC results in only the latter two. In both cases the ethylene to ethanol ratio is equal to 1, and the mechanisms appear to involve molecular eliminations. The rate expressions for the initial processes are the following: k(TEOC → products) = 10^13.91±0.14 exp((-27 529 ± 348)K/T) s^-1, T = 1005-1180 K; k(DEC → C₂H₅OCO₂H + C₂H₄) = 10^13.03±0.11 exp((-23 290 ± 26 7 K/T) s^-1, T = 955-1095 K. The listed uncertainties are one standard deviation. DEM decomposes more slowly than the other two compounds. For each DEM reacted, 1.2 ethylene and 0.5 ethanol are produced. Methane and ethane are also observed as products. The mechanism is postulated to involve both molecular and bond-breaking channels. It is concluded that ethanol arises only through the molecular channel, and on this basis the following rate expressions have been derived: k(DEM → products) = 10^15.93±0.15 exp((-36 179 ± 403 K/T) s^-1, T = 1150-1260 K; k(DEM → ethanol + products) = 10^15.07±0.45 exp((-34 517 ± 1090 K/T) s^-1, k(DEM → ethyl + OCH₂OC₂H₅) = 10^16.32±0.45 exp((-38 214 ± 1160 K/T) s^-1. The results are compared with those dealing with the stability of analogous ethers, esters, and silicon compounds. For carbon compounds the addition of ethoxy groups to the compound destabilizes the molecule. It is further concluded that rate data on the molecular decompositions of ethoxy carbon compounds cannot be easily extrapolated to silicon-containing species.

Full text of DOI:10.1021/jp970653a

5494
J. Phys. Chem. A 1997, 101, 5494-5499
Single-Pulse Shock Tube Studies of the Decomposition of Ethoxy Compounds
Ju1rgen Herzler, Jeffrey A. Manion,* and Wing Tsang
National Institute of Standards and Technology, Physical and Chemical Properties DiVision,
Gaithersburg, Maryland 20899
ReceiVed: February 24, 1997; In Final Form: May 16, 1997X
Tetraethyl orthocarbonate (TEOC), diethyl carbonate (DEC), and diethoxymethane (DEM) have been
decomposed in single-pulse shock tube experiments. TEOC decomposes to give DEC, ethylene, and ethanol
as organic products, while DEC results in only the latter two. In both cases the ethylene to ethanol ratio is
equal to 1, and the mechanisms appear to involve molecular eliminations. The rate expressions for the initial
processes are the following: k(TEOC f products) ) 1013.91(0.14 exp((-27 529 ( 348)K/T) s-1, T ) 1005-
1180 K; k(DEC f C2H5OCO2H + C2H4) ) 1013.03(0.11 exp((-23 290 ( 26 7 K/T) s-1, T ) 955-1095 K.
The listed uncertainties are one standard deviation. DEM decomposes more slowly than the other two
compounds. For each DEM reacted, 1.2 ethylene and 0.5 ethanol are produced. Methane and ethane are
also observed as products. The mechanism is postulated to involve both molecular and bond-breaking channels.
It is concluded that ethanol arises only through the molecular channel, and on this basis the following rate
expressions have been derived: k(DEM f products) ) 1015.93(0.15 exp((-36 179 ( 403 K/T) s-1, T ) 1150-
1260 K; k(DEM f ethanol + products) ) 1015.07(0.45 exp((-34 517 ( 1090 K/T) s-1; k(DEM f ethyl +
OCH2OC2H5) ) 1016.32(0.45 exp((-38 214 ( 1160 K/T) s-1. The results are compared with those dealing
with the stability of analogous ethers, esters, and silicon compounds. For carbon compounds the addition of
ethoxy groups to the compound destabilizes the molecule. It is further concluded that rate data on the molecular
decompositions of ethoxy carbon compounds cannot be easily extrapolated to silicon-containing species.
Introduction
but has been invoked by Chu and co-workers12 to explain some
aspects of tetraethoxysilane (TEOS) decomposition. Even for
TEOS, however, our recent studies11 suggest that the four-center
process is the only important molecular channel.
This paper deals with the thermal stability of organic
compounds containing adjacent ethoxy groups. Three com-
pounds, tetraethyl orthocarbonate (TEOC), diethyl carbonate
(DEC), and diethoxymethane (DEM) have been selected for
study. Earlier work1-11 has shown that the primary pathway
for the decomposition of organic compounds containing ethoxy
or other larger alkoxy groups is through molecular reactions
leading directly to stable compounds. Currently, except through
analogies, there is no basis for predicting the decomposi-
tion rate expressions of such processes. Thus the thermal
stability characteristics of such compounds are very uncertain.
This is in contrast to the situation for bond-breaking reactions
where fairly good estimates can be made from the thermo-
chemistry.
Table 1 contains a summary of earlier work on a number of
compounds that have similarities to those being studied here or
are used in the analysis of the data. The general decomposition
mechanisms are illustrated in Figure 1. Much work has been
done to establish the decomposition mechanism of the carbon-
ates (see e.g. ref 4 and citations therein). It is now generally
accepted that for carbonates containing a â hydrogen the only
important initial process is a six-center retro-ene reaction (Figure
1a). Likewise alcohols and ethers are well known to undergo
a four-center 1,2-elimination to generate an alkene and either
water or the respective alcohol (Figure 1b). We are unaware
of any previous work on the decomposition mechanisms of
diethoxy or tetraethoxy compounds. It is noteworthy that six-
center processes involving the transfer of hydrogen have only
been firmly established in cases where the starting material
contains a π-bond. Also included in Figure 1 is a six-center
transition state for the instance where there is no double bond
in the parent compound. This type of reaction has less precedent
The reactions of TEOC or DEM have not been studied to
our knowledge, but there are three lower temperature studies
(550-710 K) on DEC decomposition.1-3 They all agree that
equal amounts of ethylene, ethanol, and carbon dioxide are
formed. The rate coefficients of Bigley and Wren2 are in quite
good agreement with those of Cross et al.,3 whereas the values
of Gordon and Norris1 are about a factor of 5 to 10 higher. The
data in Table 1 show that DEC and ethyl methyl carbonate
decompose much faster than ethyl acetate, which indicates that
the substitution of the methyl group by an additional ethoxy or
methoxy destabilizes the molecule by an appreciable factor. In
contrast, methoxy for ethoxy substitution leads to rate constants
that are roughly a factor of 2 smaller. Since this is about equal
to the change in the reaction degeneracy, the nature of the
nonreacting alkoxy group seems to have little effect on the rate
constant. All of the results are consistent with the usual4
assumption that the initial product is a hydroxy-alkoxy
compound which rapidly decomposes to form the alcohol.
Figure 2 contains some relevant bond dissociation energies
(BDEs) for DEM,13,14 where we adopt the usual convention that
the BDE is equal to the enthalpy of reaction at 300 K. Values
for the stronger bonds should be fairly accurate since they are
based on either known values of the radicals and parent
compounds or the assumption that γ-substitution of O for CH2
should have minimal effect on the relevant bond energies. To
derive a value for the weakest bond, the enthalpy of formation
of the OCH2OC2H5 radical needs to be known. We are unaware
of any experimental enthalpy values for this type of radical and
have assumed that the O-atom substitution for methylene at the
â-position will have negligible effects. This may be subject to
some error.
X Abstract published in AdVance ACS Abstracts, July 1, 1997.
S1089-5639(97)00653-1 This article not subject to U.S. Copyright. Published 1997 by the American Chemical Society

Decomposition of Ethoxy Compounds
J. Phys. Chem. A, Vol. 101, No. 30, 1997 5495
TABLE 1: Rate Expressions for Some Decomposition Processes Relevant to the Present Study
substance reaction
diethyl carbonate
(C2H5O)2CdO f C2H4 + C2H5OCO2H
rate expression (s-1), temp (K)
ref
7.9 × 1013 exp(-23 154 K/T), 573-647
1.6 × 1013 exp(-23 350 K/T), 554-594
4.0 × 1013 exp(-23 750 K/T), 663-708
1.1 × 1013 exp(-23 454 K/T), 584-683
1.1 × 1013 exp(-23 290 K/T), 955-1095
5.0 × 1013 exp(-23 154 K/T), 573-647
5.2 × 1011 exp(-21 890 K/T), 582-667
1.6 × 1012 exp(-22 545 K/T), 626-682
4.0 × 1012 exp(-24 155 K/T), 772-1156
1.0 × 1013 exp(-31 300 K/T), 753-798
8.0 × 1013 exp(-33 230 K/T), 783-823
7.8 × 1014 exp(-33 162 K/T), 1025-1200
4.0 × 1014 exp(-33 300 K/T), 920-1175
2.0 × 1015 exp(-34 900 K/T), 1164-1278
1
2
2
3
TWa
1
ethyl methyl carbonate
C2H5OC(O)OCH3 f C2H4 + CH3OCO2H
3
4
5,6
ethyl acetate
CH3CO2C2H5 f C2H4 + CH3CO2H
diethyl ether
C2H5OC2H5 f C2H4 + C2H5OH
t-butanol
t-C4H9OH f i-C4H8 + H2O
tetraethoxysilane
7
8
9
10
11
(C2H5O)4Si f C2H4 + (C2H5O)3SiOH
a This work.
mechanistic consequence is that it will be impossible to form
ethanol from these or any of the other radical fragments
(CH3CH2O, CH2OH, etc.) that may be decomposition products
in these shock tube experiments. This is because the only
straightforward way of forming ethanol from a radical fragment
would involve a radical recombination. At the high tempera-
tures and low radical concentrations of our experiments such
bimolecular reactions cannot possibly compete with the facile
unimolecular decompositions of the radicals.
Recently, there has been considerable interest in the decom-
position of TEOS11 since it is widely used in the semiconductor
industry for the chemical vapor deposition of silicon dioxide
and a report on its mechanism of decomposition has appeared.6
The present study will yield some interesting information on
the effect of silicon for carbon substitution. In view of the
increasing interest in silicon compounds for materials synthesis
and processing applications and the large volume of existing
data on the thermal decomposition of organic compounds, such
information may prove to be extremely useful. It should also
be noted that many chemical agents contain alkoxy groups
attached to phosphorus. The destruction of such compounds is
currently an important issue. Here again the thermal stability
characteristics are very important.
Figure 1. Some possible concerted processes in the decomposition of
ethoxy compounds: (a) retro-ene decomposition of ethyl acetate, (b)
1,2-elimination of water from ethanol, and (c) six-center elimination
as suggested by Chu et al.,12 for TEOS decomposition.
The single-pulse shock tube method which was used in this
study is a very suitable technique for determining the mecha-
nisms and rate constants for the decomposition of large organic
molecules. The short reaction time assures that surface
processes cannot occur and radical chain reactions can be
completely eliminated through the use of large quantities of an
inhibitor. The use of gas chromatography allows us to work
with dilute concentrations, thus eliminating bimolecular reactions
and allowing us to isolate the primary unimolecular processes.
True thermal unimolecular rate constants are being measured.
The use of an internal standard to determine the reaction
temperature leads to the determination of precise rate constants
and expressions, since these values are now constrained
primarily by the accuracy of concentration determinations by
gas chromatography (GC).
Figure 2. Estimated bond dissociation energies for DEM in kJ/mol.
The enthalpies of reaction at 300 K (BDE) need not be equal
to the measured activation energies for bond breaking as
determined from experiments near 1100 K. Experience with
alkanes indicates that if the BDE values are indeed correct, the
actual experimental activation energies can be as much as 20
kJ/mol lower.15 This is due to the relation between enthalpy
and activation energy ∆H ) ∆E + RT (assuming no activation
energy for the reverse combination process) and the temperature
dependence of the bond energy due to heat capacity effects.
Although the activation energies for bond cleavage reactions
are large, the high A-factors for bond-breaking reactions (in
the 1016-1017 s-1 range) can make such processes competitive
with the molecular channels at the high temperatures used in
these shock tube experiments.
Experimental Section16
Our experiments were carried out in a heated single-pulse
shock tube designed to permit us to study relatively nonvolatile
compounds. The shock tube and the associated gas handling
and preparation system were all maintained at 108 °C. Analysis
of the reactants and products was by gas chromatography. The
heavier components were determined using a 530 µm i.d. 30 m
carbowax capillary column in the programmed temperature
Cleavage of the weakest bond in DEM (Figure 2) leads to
ethyl and the OCH2OC2H5 radical. Both species can readily
decompose through low-energy â-fission reactions. A key

5496 J. Phys. Chem. A, Vol. 101, No. 30, 1997
Herzler et al.
inhibitor
1% 135TMBb
1% 135TMB
1% 135TMB
TABLE 2: Mixtures Used in the Present Studya
mode, while the lighter species were analyzed on a 2 m × 1
mm i.d. HayeSep S column followed by a 1 m × 1 mm i.d.
column packed with HayeSep N. Mesitylene (1,3,5-trimethyl-
benzene) was used as the chemical inhibitor. It is thermally
stable under our conditions and inhibits chain processes through
the reactions:
substrate
standard
340 ppm TEOC
410 ppm DEC
250 ppm DEM
410 ppm ethyl acetate
410 ppm 2-chloropropane
125 ppm t-butanol
a Argon was the diluent for all mixtures. b 135TMB ) 1,3,5-
trimethylbenzene.
H + (CH3)3C6H3 f CH3 + (CH3)2C6H4
X + (CH3)3C6H3 f XH + (CH3)2C6H3CH2
X + (CH3)2C6H3CH2 f (CH3)2C6H3CH2X
from Aldrich and 4-methylcyclohexene was from Wiley Organ-
ics. Except for vigorous degassing and removing of ethanol
from the ethoxy compounds by the use of 0.2 nm molecular
sieves, they have been used without further purification. The
argon from Matheson was of Ultrapure Grade.
The substituted benzyl radicals produced by hydrogen abstrac-
tion are thermally stable under the present reaction conditions.
Their resonance stabilization renders them highly unreactive to
stable species. They have few options other than radical
recombination reactions, thus rapidly terminating any chain
processes.
Results
The mixtures used in these studies are given in Table 2. The
reaction pressures ranged from 200 to 400 kPa (2-4 bar). No
pressure effect was observed for any of the compounds. For
two of the three substrates, TEOC and DEC, the decomposition
processes are very clean in the sense that they yield a few
specific products, and excellent mass balances could be
achieved. Both TEOC and DEC lead to the formation of equal
amounts of ethylene and ethanol, while TEOC produces DEC
as an additional product. The net molar yield of DEC per
reacted TEOC ranged from 0.68 at the lowest temperature to
0.03 at the highest. At all temperatures the sum of DEC,
ethanol, and ethylene accounted for the total reacted TEOC.
These observations make it apparent the mechanism involves
the process TEOC f DEC + C2H4 + C2H5OH, followed by
decomposition of DEC. While the overall reaction seems clear,
it is uncertain if DEC is formed directly from TEOC in a single
step (e.g. Figure 1C) or whether there is a very unstable
triethoxycarbinol intermediate:
Standard Reactions. As internal temperature standards we
used the well-known decompositions of ethyl acetate to ethylene
and acetic acid, of 2-chloropropane to propylene and HCl, and
of t-butanol to isobutene and water. Ideally a single temperature
standard would be used, but the choices were dictated both by
the temperature ranges of the decompositions and the necessity
of separating the components in the GC analyses. To best
compare the relative rates it is obviously desirable to use a self-
consistent set of rate coefficients for the standard reactions. In
the case of TEOC and DEC it is therefore necessary to establish
the relative rates of the ethyl acetate and 2-chloropropane
decompositions. This was accomplished by using the absolute
rate parameters for the ethyl acetate decomposition from ref 5
and then using the measured relative rates k(ethyl acetate)/k(2-
bromopropane)17 and k(2-bromopropane)/k(2-chloropropane)18
to establish a consistent value for 2-chloropropane. This leads
to a slightly larger rate constant than originally19 reported for
2-chloropropane, but is consistent with other relative rate studies
from this laboratory.15 The adopted rate parameters are
TEOC f C2H4 + (C2H5O)3COH
(C2H5O)3COH f DEC + C2H5OH (fast)
In the case of DEC our data support previous1-3 suggestions
that the overall process is DEC f CO2 + C2H4 + C2H5OH.
As indicated in the Introduction, the reaction is believed4 to
proceed in two steps, a retro-ene reaction followed by rapid
decomposition of the alkoxy acid:Å
k(ethyl acetate f CH3CO2H + C2H4) )
4.0 × 1012 exp(-24 155K/T) s-1
k(2-chloropropane f propylene + HCl) )
5.2 × 1013 exp(-25 413K/T) s-1
DEC f C2H4 + C2H5OCO2H
The above compounds decompose too rapidly to serve as
effective standards for DEM, which reacted much more slowly
than TEOC or DEC. In the experiments with DEM, t-butanol
was selected as the standard. The rate parameters k(t-butanol
f isobutene + H2O) ) 6.0 × 1014 exp(-33 640 K/T) were
chosen as discussed in ref 11. At the higher temperatures of
our study, a second channel, t-butanol f CH3 + (CH3)2COH
f acetone + H + CH3, becomes important enough that it must
be taken into account. From the measured t-butanol, isobutene,
and acetone concentrations, the average reaction temperature
can be calculated by the relation
C2H5OCO2H f CO2 + C2H5OH (fast)
If the alkoxy acid is indeed an intermediate, the 1:1 ratios of
ethanol to ethylene establish that it is fully decomposed under
our conditions.
Arrhenius plots of the rate constants for decomposition of
the TEOC and DEC can be found in Figure 3. Rate constants
for DEC can be derived both from the direct studies of this
compound and from the TEOC experiments where DEC is an
intermediate. The results are in excellent agreement which
strongly supports our postulated mechanism for TEOC decom-
position.
ln([t-butanol]i/[t-butanol]f) ) τ(1 + [acetone]f/[isobutene]f)
(6.0 × 1014 exp(-33 640 K/T))
where the subscripts i and f refer to initial and final concentra-
tions, τ is the residence time of 500 µs, T is the reaction
temperature, and the rate expression is that for the isobutene
channel.
The rate expressions for the two decomposition processes are
k(TEOC f products) )
1013.91(0.14 exp(-27 529 + 348 K/T) s-1 (1005-1180 K)
k(DEC f C2H4 + C2H5OCO2H) )
1013.03(0.11 exp(-23 290 + 267 K/T)s-1 (955-1095 K)
Chemicals. The 2-chloropropane, t-butanol, mesitylene,
tetraethyl orthocarbonate, and diethyl carbonate were purchased
The small uncertainties given above are one standard deviation

Decomposition of Ethoxy Compounds
J. Phys. Chem. A, Vol. 101, No. 30, 1997 5497
Figure 3. Arrhenius plots for the decomposition of TEOC (O) and
DEC (0). Also included in the plot are DEC decomposition rate
constants calculated from experiments with TEOC where DEC is an
intermediate (2).
Figure 4. Arrhenius plots for the overall decomposition of DEM (4),
the process that leads to the production of ethanol (b), and the difference
between rate constants for DEM destruction and ethanol formation (0),
interpreted in terms of bond breaking.
and reflect only the experimental scatter. They do not include
possible systematic errors or uncertainties in the rate parameters
of the standard reaction. Taking these into account leads to
more realistic uncertainties of a factor of 2 in the A-factor, 500
K in the activation energy, and 30% in absolute rate. The self-
consistency of these results, in particular the excellent mass
balances, is very important for validating the results on DEM,
where the situation is not as unambiguous.
the initial reaction is a four-center elimination of ethylene, and
is followed by the rapid elimination of ethanol from the ethoxy
carbinol:
DEM f C2H4 + C2H5OCH2OH
C2H5OCH2OH f C2H5OH + CH2dO (fast)
DEM decomposed much more slowly than TEOC or DEC
and its product spectrum was more complex. It was found that
for every DEM destroyed, approximately 1.2 molecules of
ethylene and 0.5 molecule of ethanol were formed. The ethylene
and ethanol thus account for about 85% percent of the available
ethoxy groups. Significant amounts of methane and ethane were
also found and account for much of the missing mass. It should
be noted that with a flame ionization detector we are unable to
detect compounds such as CO2, CO, and formaldehyde. The
ratio of ethylene to ethanol product increased slightly at higher
temperatures.
Both possibilities lead to the formation of a single ethanol from
DEM. As long as any ethoxy carbinol intermediate is fully
decomposed, we can use ethanol production as a marker for
the molecular channel(s). Subtraction of this rate constant from
that for total DEM disappearance leads to a rate constant that
we associate with bond breaking. On the basis of Figure 1,
fission of the C2H5-O bond is the most important bond fission
channel. The results are shown in Figure 4 and lead to the
following rate expressions.
The nonstoichiometric ratio of ethylene to ethanol, as well
as the observation of methane and ethane as products, suggests
that DEM decomposition is more complex than the decomposi-
tions of TEOC and DEC. Probably at least two parallel paths
are important. Since the mechanism is uncertain it is necessary
to determine decomposition rate constants from the disappear-
ance of starting material rather than from appearance of product.
Because at low conversions we are taking the difference of two
large numbers, this method is less accurate than that using
product formation. However, for TEOC and DEC the results
using product formation and destruction of the starting material
were essentially the same, which gives us confidence that the
procedure is still satisfactory.
The following represents our attempts at determining the
relative contributions of the various pathways for DEM. The
possible unimolecular destruction modes are molecular reactions
and bond fission reactions. Ethylene can be produced via both
pathways, so this product cannot be used to follow a specific
channel. As indicated in the introduction, ethanol cannot readily
arise from radical intermediates, so its formation must be the
result of a molecular process. Two molecular mechanisms seem
most likely. One possibility is a six-center reaction leading
directly to ethanol and ethylene:
k(DEM f products) )
1015.93(0.15 exp(-36 179 ( 403 K/T) s-1
T ) 1150-1260 K
k(DEM f C2H4 + C2H5OCH2OH) )
1015.07(0.45 exp((-34 517 ( 1090)K/T) s-1
k(DEM f C2H5 + OCH2OC2H5) )
1016.32(0.45 exp((-38 214 ( 1160)K)/T] s-1
In the above we have postulated the molecular channel that
ultimately leads to ethanol to be a four-center process, but it
should be realized that the only basis for this assignment is by
analogy with more established systems. The quoted uncertain-
ties in our rate expressions are one standard deviation. For the
overall process, the total uncertainty is similar to those of the
other compounds, about a factor of 2 in the A-factor, 500 K in
the activation energy, and 30% in absolute rate. A meaningful
estimate of the uncertainties for the rate parameters of the
specific channels is more difficult to make since this aspect
depends heavily on the correctness of our mechanistic assump-
tions. In particular is the assumption that the yield of ethanol
from the initial molecular channel is 100% over the entire
temperature range. At higher temperatures the mass balance
can be used as a check, but at low temperatures we are dealing
with differences of large numbers and small inconsistencies
would be difficult to detect. An error of about 10% at the low
end of our temperature range would reduce the A-factor for the
An alternative with more precedent is a sequential process where

5498 J. Phys. Chem. A, Vol. 101, No. 30, 1997
Herzler et al.
Figure 6. Possible decomposition pathways for the oxygen-containing
radical formed by fission of the weakest bond in DEM.
they are dependent on the postulated mechanism. In this regard
the A-factor for the molecular channel is larger than expected
when compared with the decompositions of TEOC, DEC, or
the more similar diethyl ether. One possible scenario is that
the initial molecular channel is indeed a 1,2-molecular elimina-
tion to form ethylene and ethoxy carbinol, but that this latter
compound is not fully decomposed at the lower end of our
temperature range. While our mass balance is such that very
little of the carbinol could be unreacted, the data are subject to
the greatest uncertainty at the lowest extents of reaction. If even
10% of the ethoxy carbinol remained at the lowest temperatures
of our experiments, the activation energy for the molecular
channel would be reduced by about 10 kJ/mol and the A-factor
would be more in line with expectations. Such an explanation
would of course imply that the particular carbinol formed from
DEM is more stable than those inferred in the decompositions
of DEC and TEOC. The likelihood of this is difficult to judge
given the complete absence of quantitative data on the decom-
positions of carbinols.
Figure 5. Literature rate constants for the decompositions of DEC
and ethyl methyl carbonate. For easier comparison with DEC, the rate
constants for ethyl methyl carbonate have been multiplied by a factor
of 2 to adjust for the reaction degeneracy. Results for DEC: b, ref 1;
2 and f, ref 2; 9, ref 3; dashed line, extrapolated results from the
present work. Results for ethyl methyl carbonate: 0, ref 1; O, ref 3;
3, ref 4.
molecular channel by about a factor of 3 and decrease the
activation energy by about 10 kJ/mol. Conversely, for the bond
fission channel the A-factor would be increased by about a factor
of 5 and the activation energy by nearly 20 kJ/mol.
Discussion
Decompositions of TEOC and DEC. The clean product
spectrum and excellent mass balances found in the decomposi-
tions of TEOC and DEC indicate that these processes are
molecular in nature. For both substrates we noted the absence
of significant quantities of methane, ethane, and the xylenes.
The latter is the expected product from H-atom attack on the
inhibitor mesitylene, while the first two compounds can arise
from methyl radicals that are formed. Clearly, bond breaking,
which would inevitably lead to the formation of such radicals,
cannot be of any consequence. This deduction is further
confirmed by the finding of pre-exponential factors in the range
of 1013 s-1. Bond fission processes would be expected to yield
much larger A-factors in the 1016-1017 s-1 range. A compari-
son of our results for DEC with literature data can be found in
Figure 5. It can be seen that extrapolation of our results by 5
orders of magnitude lands us among the results of three earlier
works.
Nevertheless, our data do not permit us to differentiate
between a number of the possible concerted mechanisms. While
the mechanism for DEC decomposition seems well-established
to be the retro-ene reaction, the situation for TEOC is less clear.
An interesting question is whether TEOC decomposes in a six-
center process resulting in the simultaneous release of ethylene
and ethanol or if the mechanism involves 1,2-elimination of
ethylene followed by rapid decomposition of the triethoxy-
carbinol to give ethanol. In the absence of a π bond, we are
unaware of any well-established precedent for a six-center
complex fission involving the transfer of an H-atom. While there
are a few molecular decompositions, e.g. 1,3,5-trioxane f 3
H2CdO,20 which apparently proceed via a six-center transition
state in the absence of a π bond, no atoms are transferred and
such reactions appear to be substantially different from those
considered here. In this respect the 1,2-process seems more
likely for the TEOC system. If this is the case, the very
satisfactory material balances and 1:1 ratios of ethylene to
ethanol demonstrate that the ethoxy carbinol is extremely
unstable under our conditions.
The derived activation energy for bond breaking in DEM (317
kJ/mol) is much lower than that we have estimated (Figure 2),
assuming that â-substitution of O for methylene has no effects.
While the A-factor is at the lower end of the range that one
would expect for bond breaking, it is close enough that our
postulated mechanism appears to be correct. As noted earlier,
the weakest bond in DEM is the ethyl-O bond, and the rate
parameters should be associated with this bond. Taken at face
value, the data indicate that substitution of an oxygen for a
methylene group â to the bond being broken decreases the bond
energy by about 20 kJ/mol. However, since the rate parameters
for this channel were derived by difference, it should be
emphasized that this value is strongly dependent on our
assumption that the ethoxy carbinol is fully decomposed. An
error of about 10% in the rate constant for the molecular channel
at the lowest temperatures of our study would increase the
A-factor for bond breaking to about 1 × 1017
and lead to an
activation energy of about 335 kJ/mol. Because of this, the
derived bond energy should be treated with caution.
Cleavage of the weakest bond in DEM (Figure 2) leads to
ethyl and the OCH2OC2H5 radical. Ethyl can be expected to
decompose to ethylene through fission of a â C-H bond.
Possible pathways for the breakdown of the oxygenated radical
can be found in Figure 6. Although experimental data do not
exist, three channels are possible. The one involving â C-O
bond fission is energetically reasonable and is consistent with
known pathways in analogous compounds. It leads to formation
of formaldehyde and methyl radical. One could conceivably
imagine the formation of ethanol through a 1,1 elimination from
the OCH2OC2H5 radical. There is, however, no precedent for
such a process, and one would expect it to have a barrier
substantially higher than that for â C-O bond fission. The
third path involves isomerization of the initial radical to HOCH2-
OCH2CH2 prior to decomposition. In larger alkyl radicals, such
1,5 processes can be important although relatively few rate data
exist. Because HOCH2OCH2CH2 will break down to ethylene,
Decomposition of DEM. The rate expressions for the two
channels in DEM decomposition are subject to the caveat that

Decomposition of Ethoxy Compounds
J. Phys. Chem. A, Vol. 101, No. 30, 1997 5499
consequences of this study. What is less clear is if the rate
differences reflect a relative stabilization of the four-center
transition state or if there is an actual change in the mechanism.
For instance could the six-center process be important in TEOC,
despite the absence of a π system in the parent compound?
Resolution of this question must await further studies.
Comparison with Ethoxy Silicon Compounds. TEOS is
the silicon analog of TEOC where the central carbon atom has
been replaced by silicon. Our results11 indicate that the
molecular elimination reaction of TEOS is a four-center process
and, at 1150 K, is a factor of 20 slower than the reaction of
TEOC. The derived activation energy for TEOS is nearly 60
kJ/mol higher. It is further interesting that for ethoxy silicon
compounds the rate constant for elimination was unaffected by
the number of ethoxy groups on the central silicon atom,11
except for changes associated with the reaction path degeneracy.
This contrasts with the carbon compounds where large rate
differences are observed. The reasons for the shift in behavior
are uncertain. One might speculate that the relative charge-
accepting properties of Si and C are important and possibly the
large difference in the O-C and O-Si bond lengths. Irrespec-
tive of the explanation, it is apparent that rate data from carbon
systems are not directly transferable to the silicon compounds.
Further, there are presently no methods of predicting the rate
constants in lieu of experimental data.
Figure 7. Comparison of present results on DEM (1) and TEOC (O)
with data on TEOS (4).
formaldehyde, and OH, this has an important consequence since
it results in two ethylenes (one each from ethyl and HOCH2-
OCH2CH2) from bond fission in DEM rather than one if
isomerization is unimportant.
On the basis of our product spectrum and postulated mech-
anism we are able to draw some conclusions as to the fate of
OCH2OC2H5. Our data show that for every DEM destroyed,
approximately 1.2 molecules of ethylene and 0.5 molecule of
ethanol are formed. From the amount of ethanol we can
conclude that the molecular channel accounts for about 50% of
reaction with the balance presumably due to bond fission.
Further, since the molecular channel leads to a 1:1 ratio of
alcohol to alkene, the amount of ethylene from the bond fission
channel is 1.2 - 0.5 ) 0.7 per reacted DEM. Of this amount,
0.5 unit of the ethylene must result from decomposition of the
ethyl radical produced in the initial bond fission and the
remaining 0.2 from decomposition of OCH2OC2H5. As shown
in Figure 6, this can only occur if the OCH2OC2H5 radical
undergoes a hydrogen shift prior to â C-O bond fission. Our
data indicate that the branching ratio between isomerization and
â-bond fission is about 0.4. In a more general sense, the above
analysis shows that attempts to model such systems will require
that isomerization reactions are properly considered.
References and Notes
(1) Gordon, A. S.; Norris, W. P. J. Phys. Chem. 1965, 69, 3013.
(2) Bigley, D. B.; Wren, C. M. J. Chem. Soc., Perkin Trans. II 1972,
926.
(3) Cross, J. T. D.; Hunter, R.; Stimson, R. Aust. J. Chem. 1976, 29,
1477.
(4) Taylor, R. J. Chem. Soc., Perkin Trans. II 1983, 291.
(5) Beadle, P. C.; Golden, D. M.; Benson, S. W. Int. J. Chem. Kinet.
1972, 4, 265.
(6) Benson, S. W.; O’Neal, E. Unimolecular Reactions, NSRDS-21;
US Government Printing Office: Washington, DC, 1971.
(7) Foucault, J.-F.; Martin, R. J. Chim. Phys. 1978, 75, 132.
(8) Seres, I.; Huhn, P. Magy. Kem. Foly. 1975, 81, 120.
(9) Newman, C. G.; O’Neal, H. E.; Ring, M. A.; Leska, F.; Shipley,
N. Int. J. Chem. Kinet. 1979, 11, 1167.
(10) Lewis, D.; Keil, M.; Sarr, M. J. Am. Chem. Soc. 1974, 96, 4390.
(11) Herzler, J.; Manion, J. A.; Tsang, W. J. Phys. Chem., following
paper in this issue.
Relative Rates of Ethoxy Carbon Compounds. Figure 7
shows that rate constants for TEOC are a factor of 20 to 30
larger than those of DEM in the temperature range under
consideration. To compare the rates of the molecular channels,
the overall rates should be corrected for the number of ethoxy
groups and it should be recalled that for DEM the molecular
channel accounted for only about 50% of the overall destruction
rate. After these adjustments, the rate difference between what
we have postulated as mechanistically similar processes is still
a factor of 20-30. This is equivalent to a change in the
activation energy by about 33 kJ/mol. If the data on diethyl
ether7,8 are extrapolated to our temperatures, it is seen that the
rate constants for the molecular channel vary by more than 2
orders of magnitude and follow the order diethyl ether < DEM
< TEOC. It is thus apparent that addition of ethoxy groups to
the central carbon atom destabilizes the molecule by an
appreciable amount. This is one of the most important
(12) Chu, J. C. S.; Breslin, J.; Wang, N. S.; Lin, M. C. Mater. Lett.
1991, 12, 179.
(13) Pedley, J. B.; Naylor, R. D.; Kirby, S. P. Thermochemical Data
for Organic Compounds; Chapman and Hall: London, 1977.
(14) DeMore, W. B.; Sander, S. P.; Golden, D. M.; Hampson, R. F.;
Kurylo, M. J.; Howard, C. J.; Ravishankara, A. R.; Kolb, C. E.; Molina,
M. J. Chemical Kinetics and Photochemical Data for Use in Stratospheric
Modeling, NASA JPL Publicaton 94-26; Jet Propulsion Laboratory,
California Institute of Technology, Pasadena, CA, December 15, 1994.
(15) Tsang, W. in Shock WaVes in Chemistry; Lifshitz, A., Ed.; Marcel
Dekker: New York, 1981, p 59.
(16) Certain commercial materials and equipment are identified in this
paper in order to specify adequately the experimental procedure. In no case
does such identification imply recommendation or endorsement by the
National Institute of Standards and Technology, nor does it imply that the
material or equipment is necessarily the best available for the purpose.
(17) Gutman, D.; Braun, W.; Tsang, W. J. Chem. Phys. 1977, 67, 4291.
(18) Tsang, W. Int. J. Chem. Kinet. 1970, 2, 311.
(19) Tsang, W. J. Chem. Phys. 1964, 41, 2487.
(20) Hochgreb, S.; Dryer, F. L. J. Phys. Chem. 1992, 96, 295.