transition state involving a six-membered ring, since an
oxygen atom occupies the required position in the chain and
therefore no such abstractable H-atom is available. Surpris-
ingly, the observed product 3-hydroxybutyl formate indicates
that the isomerization reaction proceeds through a transition
state involving a seven-membered ring. This proposed
transition state includes the oxygen atom of the ether linkage
in the cyclic structure. With respect to the ring strain in
cycloalkanes, a 1,6-H-shift is expected to be slower than a
1,5-H-shift, but the large amounts of 3-hydroxybutyl formate
product observed strongly indicates that this reaction path
is the main channel for alkoxy radical 9. The other possible
isomerization reaction of alkoxy radical 9 is a 1,4-H-atom via
a five-membered ring transition state, not including the
oxygen atom of the ether linkage. This process would be
predicted to lead mainly to the formation of ethylene glycol
monoformate and the radical 13, which is proposed (24) to
form mainly acetaldehyde and formaldehyde via decomposi-
tion of the intermediary â-hydroxy alkoxy radical 14. The
low initial yields of acetaldehyde (<5% of 2-butoxyethanol
removed) that were observed indicate that this alternative
isomerization route for alkoxy radical 9 is of limited impor-
tance and possibly does not occur. In competition with the
isomerization reactions, alkoxy radical 9 may also react with
oxygen to form 2-(3-oxo-butoxy)ethanol or decompose to
form acetaldehyde and alkoxy radical 15 or alternatively
decompose to form a methyl radical and 3-(2-hydroxy ethoxy)-
propionaldehyde. A mass spectrum of the latter compound
was reported by Espinosa et al. (25), but no equivalent mass
spectrum was found in our analyses. As no major unknown
peaks were found in the present study, we conclude that
alkoxy radical 9 reacts mainly by the 1,6-H-atom shift
proposed above.
occurrence of isomerization reactions in alkoxy radicals with
the structure RCH(O•)OR comes from the high yields of 2-
and 3-hydroxybutyl formates. As these products could have
arisen from the isomerization reactions of alkoxy radicals 8
and 9, leading subsequently to alkoxy radicals 11 and 16, one
might expect by analogy with alkoxy radicals in alkanes (11,
27) a second fast isomerization step, abstracting the reactive
H-atoms R to the hydroxy group, which would lead to ketone
products. It appears, however, that no second isomerization
step occurs, but that the intermediate alkoxy radicals 11 and
16 again mainly undergo unimolecular decomposition.
Secondary Reactions of OH Radicals with Products. The
rate coefficients of OH radicals with butoxyacetaldehyde and
2-propyl-1,3-dioxolane were measured since they are ex-
pected to be of the same order of magnitude as that of their
precursor 2-butoxyethanol and hence their loss via OH radical
reaction is important. The structure activity relation (SAR)
prediction (16) for rate coefficients with OH radicals over-
estimates the values of cyclic polyethers and several hetero
atom-substituted acetaldehydes. The measured rate coef-
ficients for 2-propyl-1,3-dioxolane of 10.8 × 10-12 cm3
molecule-1 s-1 may be compared to that reported for
2-methyl-1,3-dioxolane, measured under the same condi-
tions, of 9.4 × 10-12 cm3 molecule-1 s-1 (3). The enhance-
ment of rate coefficient for 2-propyl-1,3-dioxolane is ex-
plained by the presence of two additional secondary CH2
groups. The result for butoxyacetaldehyde, 20.6 × 10-12 cm3
molecule-1 s-1, may be compared to that for ethoxyacetal-
dehyde reported previously, i.e., 16.6 × 10-12 cm3 molecule-1
s-1 (3). Both aldehydes react only slightly faster than the
structurally related alcohols 2-butoxyethanol (19.4 × 10-12
cm3 molecule-1 s-1 (2)) and 2-ethoxyethanol (14.5 × 10-12
cm3 molecule-1 s-1 (2)).
All-in-all the proposed mechanism gives a reasonably
quantitative account of the observed product distribution.
The absence of major unknown peaks in our analyses and
the fact that the observed products, in conjunction with the
proposed mechanism, account for about 110% of the 2-bu-
toxyethanol removed indicate that all the main routes are
accounted for in the degradation of this hydroxy ether. Clearly
the relative high errors in the reported product yields as well
as the uncertain errors for the compounds with estimated
yields does not rule out the possibility of additional minor
pathways. One such additional source of error seems likely
to involve the loss of the low vapor pressure products on the
Teflon walls and on aerosols formed in the system. These
loss processes were investigated by irradiating a typical bag
mixture and following the decay of the concentrations of the
products in the dark over a period of 3 h. The compounds
containing hydroxy groups showed losses of 4-9%/ h. For
ethylene glycol monoformate, the compound that showed
the largest loss, these processes gave rise to a reduction of
7-9% of the concentrations measured at the end of a typical
experiment. No additional loss due to photolyses of the
products butyl formate, propyl nitrate, 2-propyl-1,3-diox-
olane, butoxyacetaldehyde, and 2-butoxyethanol was ob-
served under the irradiation conditions of the experiments.
An interesting feature of the observed product distribution
is the high yield of 3-hydroxybutyl formate. The OH reactivity
at the CH2 site that gives rise to this product is about 4 times
higher than a CH2 group in an alkane. This fact implies that
the CH2 group in γ-position relative to the ether linkage is
activated toward OH attack. As previously reported (2), the
rate coefficients of OH-radical reactions with a series glycol
ethers from 2-methoxyethanol to 2-butoxyethanol increase
more per carbon atom in the alkyl chain than those for the
corresponding alkanes. A similar effect has also been
observed for rate coefficients of the OH reactions with the
series of linear ethers, alcohols, ketones, and esters (28-30,
15). For the linear esters (15), this effect was tentatively
ascribed to the formation of a complex between the OH radical
and the carbonyl group, prior to H-atom abstraction, and
this proposed mechanism was supported by the negative and
discontinuous temperature dependence of the overall rate
coefficient of some of these esters with the OH radical. We
are presently undertaking theoretical calculations of the
stability of such proposed complexes to try to establish the
feasibility of such a mechanism.
Acknowledgments
The authors thank the Schweizerische Nationalfonds zur
Fo¨ rderung der wissenschaftlichen Forschung for financial
support and Dr. Marc Suter of EAWAG for carrying out the
high-resolution mass spectroscopic measurements.
In keeping with ethers in general, the most reactive sites
in glycol ethers towards OH attack are the CH
2 groups adjacent
to the secondary O-atom center (26). It is therefore important
to know whether the resulting alkoxy radicals RCH(O•)OR
undergo isomerization reactions via H-atom shifts in cyclic
transition states. In this study, none of the observed products
indicates an isomerization reaction for the postulated alkoxy
radicals 6 and 7. Since high yields of butyl formate and
ethylene glycol monoformates were detected and no major
unknown products were observed, it appears that the alkoxy
radicals 6 and 7 mostly undergo unimolecular decomposition
reactions, despite the fact that the expected isomerization
reactions would involve H-abstraction at reactive RCH2R or
RCH2(OH) carbon centers (4). Further evidence for the non-
Literature Cited
(1) ECETOC. The Toxicology of Glycol Ethers and its Relevance to
Man; Technical Report No. 64; ECETOC: Brussels, 1995.
(2) Stemmler, K.; Kinnison, D. J.; Kerr, J. A. J. Phys. Chem. 1996, 100,
2114-2116.
(3) Stemmler, K.; Mengon, W.; Kerr, J. A. Environ. Sci. Technol. 1996,
30, 3385-3391.
(4) Atkinson, R. Atmos. Environ. 1990, 24A, 1-41.
(5) Krausz, F. Ann.Chim. 1949, 12, 810-831.
9
VOL. 31, NO. 5, 1997 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 1 5 0 3