A laser flash photolysis study on 2-azido-N,N-diethylbenzylamine - The reactivity of iminoquinone methides in solution

Article abstract of DOI:10.1002/1099-0690(200107)2001:13<2463::AID-EJOC2463>3.0.CO;2-6

Photolysis of 2-azido-N,N-diethylbenzylamine (1) in acetonitrile solution in the presence of various quenchers yields products of common nitrene chemistry (that is, derived from a didehydroazepine or a triplet nitrene), together with products arising from trapping of stereoisomeric iminoquinone methides. These intermediates are formed in a monophotonic process, by dediazotation of the precursor combined with a 1,4-hydrogen shift. The reactivity of the iminoquinone methides (IQM) towards a variety of quenchers, including dienophiles and various nucleophiles, has been explored, using acetonitrile, DMF, and n-hexane as solvents. IQM reacts efficiently with electron deficient dienophiles [k_q(maleic anhydride) = 2.4·10⁷ M^-1 s^-1], but not with simple olefins such as cyclohexene. IQM shows low to medium reactivity towards simple primary or secondary amines [k_q(n-propylamine) = 6.4·10⁴ M^-1 s^-1], medium reactivity with sterically unhindered simple alcohols [k_q(methanol) = 4.6·10⁶ M^-1 s^-1], and high to very high reactivity towards thiols [k_q(n-dodecanethiol) = 1.1·10⁹ M^-1 s^-1], diols, and triols [k_q(glycerol) = 1.5·10⁸ M^-1 s^-1], and sugars [k_q(D-glucose) = 1.0·10⁹ M^-1 s^-1]. IQM also reacts rapidly with cytosine: k_q(cytosine) = 3.4·10⁸ M^-1 s^-1, while the reaction with adenine or thymine is less efficient [k_q(adenine) = 6.4·10⁷ M^-1 s^-1, k_q(thymine) = 7.7·10⁶ M^-1 s^-1].

Full text of DOI:10.1002/1099-0690(200107)2001:13<2463::AID-EJOC2463>3.0.CO;2-6

FULL PAPER
A Laser Flash Photolysis Study on 2-Azido-N,N-diethylbenzylamine ؊
The Reactivity of Iminoquinone Methides in Solution
Götz Bucher[a]
Dedicated to the memory of Professor Lennart Eberson
Keywords: Cycloadditions / Iminoquinone methides / Laser spectroscopy / Nitrenes / Quinoid compounds
Photolysis of 2-azido-N,N-diethylbenzylamine (1) in aceto- ride) = 2.4·107
M
−1
s−1], but not with simple olefins such as
nitrile solution in the presence of various quenchers yields
products of common nitrene chemistry (that is, derived from
a didehydroazepine or a triplet nitrene), together with prod- 6.4·104
cyclohexene. IQM shows low to medium reactivity towards
simple primary or secondary amines [k
q
(n-propylamine) =
−1
−1
M
s ], medium reactivity with sterically unhindered
(methanol) = 4.6·106
M
−1
s
−1
], and high to
ucts arising from trapping of stereoisomeric iminoquinone
simple alcohols [k
q
methides. These intermediates are formed in a monophotonic very high reactivity towards thiols [k
q
(n-dodecanethiol) =
9
−1 −1
8
−1
process, by dediazotation of the precursor combined with a
,4-hydrogen shift. The reactivity of the iminoquinone me-
thides (IQM) towards a variety of quenchers, including dien- reacts rapidly with cytosine: k
ophiles and various nucleophiles, has been explored, using while the reaction with adenine or thymine is less efficient
1.1·10
M
s
], diols, and triols [k
q
(glycerol) = 1.5·10
M
−1
-glucose) = 1.0·109
−1 −1
1
s
], and sugars [k
q
(D
M
s ]. IQM also
(cytosine) = 3.4·108
M
−1
s ,
−1
q
(adenine) = 6.4·107
M
−1
s
−1
, k
(thymine) = 7.7·106
M
−1 −1
s ].
acetonitrile, DMF, and n-hexane as solvents. IQM reacts effi-
[k
q
q
q
ciently with electron deficient dienophiles [k (maleic anhyd-
Introduction
The preceding contribution[1] describes the photochem-
istry of matrix-isolated aryl azides bearing π-donor sub-
stituents in ortho-benzylic positions. It was found that al-
most any π-donor in this position will enable a photochem-
ical 1,4-hydrogen shift to take place, yielding iminoquinone
methides. Quinone methides and iminoquinone methides
are a current focus of research. A number of papers have
dealt with their use as reactive dienes in preparative organic
[
2Ϫ7]
chemistry.
Other work has been devoted to stabilizing
these highly reactive species by complexation to transition
[
8Ϫ11]
metals,
and to elucidating their properties and reactiv-
[12Ϫ14]
ity by using time-resolved techniques,
spectroscopy,
photoelectron
[
15,16]
[17Ϫ19]
or matrix isolation spectroscopy.
Quinone methides have received attention for their role in
[20]
the metabolism of a number of preservatives,
for their
potential use as anti-cancer agents and as active component
[
21,22]
of insecticides,
efficiently.
and for their ability to alkylate DNA
The reactions between parent o-quinone
methide and DNA bases have been investigated in detail by
[
21,23]
[
23]
Pande et al. In their study, o-quinone methide was liber-
ated from a silylated 2-bromomethylphenol by fluoride-in-
duced desilylation followed by bromide elimination. It was
found, if the free bases were used as targets, that o-quinone
methide reacted preferentially with cytosine, while, if duplex
DNA was employed, reaction with dC was suppressed. The
relative reactivity determined was 6200:520:210 for dC/dG/
dA (dT trapping was not observed) if deoxynucleosides
were used, with 310:38:8.4 for single-strand DNA, and
1
.7:19:1.0 for duplex DNA.[ In order to obtain absolute
23]
kinetic data on this interesting system, an efficient photo-
chemical precursor for o-quinone methide would be re-
[
a]
Lehrstuhl für Organische Chemie II, Ruhr-Universität
Bochum,
Universitätsstrasse 150, 44801 Bochum, Germany
E-Mail: goetz.bucher@orch.ruhr-uni-bochum.de
quired. The hydroxybenzyl alcohols used by Wan et
al.,[
12,13,24,25]
however, require excitation with λexc ϭ 248 or
Eur. J. Org. Chem. 2001, 2463Ϫ2475
 WILEY-VCH Verlag GmbH, 69451 Weinheim, 2001
1434Ϫ193X/01/0713Ϫ2463 $ 17.50ϩ.50/0
2463

G. Bucher
FULL PAPER
2
66 nm, which precludes quenching studies with DNA ba-
ses. Aryl azides, on the other hand, may be conveniently
excited using the 3rd harmonic of an Nd-YAG laser at
λexc ϭ 355 nm. The observation of a highly efficient 1,4-
hydrogen shift in 2-[(dialkylamino)methyl]phenylnitrenes, as
described in the preceding contribution,[ offers the prom-
ise that, at least in this case, laser flash photolysis of the
corresponding azides should result in the detection of imin-
oquinone methides. The work reported here was under-
taken with this aim in mind. It describes studies on the pho-
tochemistry of 2-azido-N,N-diethylbenzylamine (1). This
precursor was shown in the preceding contribution[ to un-
dergo a highly efficient 1,4-hydrogen shift when photolyzed
under matrix isolation conditions.
1]
1]
Scheme 1
Results
Steady-state photolysis (λ ϭ 320 nm, 30 min) of 1 in
1
CH CN/HNEt (9:1) with H NMR spectroscopic analysis,
3
2
Product Studies
yielded a mixture consisting mainly of the methyleneazep-
ine 9 and unchanged precursor. Methyleneazepine 9 had
previously been synthesized in this laboratory[ and was
available as a reference sample. It is formed by elimination
of diethylamine from azepine 10, which in turn is formed
from didehydroazepine 5 (Scheme 2). Aminal 11, the prod-
uct expected from trapping of iminoquinone methides by
diethylamine, was not found to be a major constituent of
the mixture. Aminal 11 would have to show a characteristic
The primary products of the photolysis of 1 at very low
temperature could be unequivocally identified as sa-3 ϩ aa-
27]
3
, by IR spectroscopy on matrix-isolated samples.[1] Dide-
hydroazepine 5, on the other hand, was only formed in mi-
nute amounts. Low-temperature photochemistry of 1 in ar-
gon thus represents a very clean reaction, yielding a well-
defined set of products. In contrast to this, photolysis of 1
in solution at ambient temperature gave rise to complex
mixtures.
Three quenchers Ϫ water, diethylamine, and acrylonitrile
Ϫ were employed to test the photochemistry of 1. The
photolysis of 1 was performed using acetonitrile as solvent,
and both steady-state and pulsed-laser excitation were used.
For product identification, several compounds suspected as
possible photoproducts were synthesized independently. If
a reference compound was not available, only tentative
product assignments were made, on the basis of the MS
decay patterns.
1
methine H resonance (singlet) between δ ഠ 3.3 and
28,29]
4
.4,[
and this was not detected as a major peak in the
1
H NMR spectrum of the mixture.
GC-MS analysis of the products formed by steady-state
photolysis (λ ϭ 320 nm, 30 min) of 1 in acetonitrile solution
showed that only a small amount (Ͻ 15%) of volatile prod-
ucts had been formed. As the main product (Ͻ 5%), an as
yet unidentified species with m/z ϭ 150 was formed.[ The
product mixture also contained small amounts (ca. 1%) of
26]
2
-aminobenzaldehyde (6). Addition of 1 or 5% water to the
solvent resulted in a significantly increased yield (8% or
1
3
1%, respectively) of 6.
If a pulsed laser (Nd-YAG, 355 nm or XeCl-excimer,
08 nm) was used for irradiation of a sample of 1 in
CH CN/1% H O, the composition of the product mixture
3
2
did not change significantly. As a new product, 2,1-benzi-
soxazole (8) (1%) was identified, in addition to 6% of 2-
aminobenzaldehyde (6), 2% of the aniline 7, and three dif-
Scheme 2
If the analysis of the photolyzed sample (solvent: 1% di-
ferent unidentified products (Σ ϭ 13%), with molecular ethylamine in acetonitrile) was performed by GC/MS, how-
masses of M ϭ 176 (twice) or 150. The products formed by ever, small amounts (1%) of a compound with m/z ϭ 249
photolysis of 1 in acetonitrile or in acetonitrile/water are were detected in addition to methyleneazepine 9 (25%), 2-
amino-N,N-diethylbenzylamine (7)[ (2%), 2-aminobenzal-
30]
given in Scheme 1.
2464
Eur. J. Org. Chem. 2001, 2463Ϫ2475

A Laser Flash Photolysis Study on 2-Azido-N,N-diethylbenzylamine
dehyde (6) (4%), 2-amino-N,N-diethylbenzamide (12) (2%),
FULL PAPER
Pulsed-laser (355 or 308 nm) irradiation of solutions of 1
enamine 13 (2%, tentative assignment based on MS decay in acetonitrile containing diethylamine produced results
pattern), and a multitude of other, unidentified products that did not differ substantially from those obtained using
present in small concentrations. Peaks with m/z ϭ 249 steady-state irradiation. Again, methyleneazepine 9 was de-
correspond to products of reactions between the nitrene Ϫ tected as the main volatile product. As an additional
or rearranged intermediates derived from it Ϫ and di- photoproduct, small amounts of 2,1-benzisoxazole (8, 1%)
ethylamine. Three products with m/z ϭ 249 can be con- were detected. Steady-state photolysis of 1 in acetonitrile
ceived of, corresponding to diethylamine trapping of singlet containing 1% acrylonitrile yielded 4% of 2-aminobenzal-
nitrene, iminoquinone methide, or didehydroazepine. dehyde (6), 1% of quinoline (15), 2% of the aniline 7, and
Scheme 3 depicts the products formed upon photolysis of 1 a total of 7% (1% ϩ 2% ϩ 3% ϩ 1%) of compounds with
in acetonitrile in the presence of diethylamine.
m/z ϭ 229, in addition to a total of 16% of various uniden-
tified products. A molecular mass of 229 is to be expected
for the products of trapping of iminoquinone methides or
2
-(diethylaminomethyl)phenylnitrene with acrylonitrile
(
Scheme 4).
Scheme 3
Scheme 4
Among the four products with m/z ϭ 229, there are two
subsets of two compounds, each exhibiting very similar
mass spectra. This observation might point towards the
formation of a mixture of the four isomeric cyanotetrahyd-
roquinolines depicted in Scheme 4, as the mass spectra of
the cis and trans isomers of each of the two regioisomers
may be anticipated to be similar. The formation of small
amounts of quinoline (15) as a side product would support
this notion, as quinoline can be formed from 16Ϫ19 by
elimination of both diethylamine and HCN. As a caveat it
has to be noted that in DielsϪAlder reactions between elec-
tron-rich dienes and electron-deficient olefins, the outcome
is predominantly such that the substituents (here diethylam-
ino and cyano) end up located ortho (or para, if the diene
is substituted in 3-position) to each other. Thus, it would
appear very surprising, that an iminoquinone methide 3
should show so little selectivity.
Given the fact that methyleneazepine 9 is formed as a
major product, the elimination of diethylamine from azep-
ine 10 must occur readily. For that reason, it appears un-
likely that this compound would survive the high temper-
atures in the injector block of the gas chromatograph. In-
sertion of a singlet nitrene into an NϪH bond of a second-
ary amine, producing hydrazine derivatives such as 14, has
been observed before.[
31]
This reaction, however, seems to
be limited to highly reactive nitrenes activated by electron-
withdrawing substituents such as nitro groups[31] or fluorine
atoms, in which cases CϪH insertion and addition to Cϭ
C double bonds has been observed.[
32Ϫ34]
Hence, aminal 11
appears to be the most likely candidate. The structure is in
agreement with the mass spectrum.[35]
Laser Flash Photolysis: Transient Identification
Laser flash photolysis (λexc ϭ 266 or 355 nm) of 1 in
acetonitrile gave a transient spectrum (transient X, see Fig-
ure 1), which was not affected by purging the solution with
oxygen. Apart from the absolute transient intensity, there
was no difference between data recorded using 266- or 355-
nm excitation. A plot of transient intensity vs. laser power
(
λexc ϭ 266 nm) reveals a linear power dependence (Fig-
ure 2), which means that the transient observed is formed
Eur. J. Org. Chem. 2001, 2463Ϫ2475
2465

G. Bucher
FULL PAPER
in a monophotonic process. Figure 1 shows that λmax
shifted from an initial 450 nm to 410 nm within 10 µs. After
1
50 µs, λmax was 370 nm. Fitting of the transient trace of
transient X (λ ϭ 450 nm) with a first-order rate law gave a
lifetime of τ ϭ 33 µs. However, close examination of the
fitting revealed a systematic deviation in the residuals (Fig-
ure 3). This observation, together with the time-dependent
λmax, suggests more complex transient behavior. Figure 3
suggests, that at least three different transient species con-
tribute to the decay measured at λ ϭ 450 nm. Furthermore,
a first-order growth (τ ϭ 1.4 µs) was detected at λ ϭ
3
70 nm (Figure 4).
Figure 3. Plot of the residuals of a trace fitting; trace obtained by
LFP (266 nm) of 1 in ACN, λ ϭ 460 nm; bottom: first-order fitting
(
monoexponential); center: 2 ϫ first-order fitting (biexponential);
top: 3 ϫ first-order fitting (triexponential)
Figure 1. Transient spectrum obtained after LFP (355 nm) of 1 in
acetonitrile, room temp., Ar purge; black dots: 2.2 µs after LFP;
light dots: 10.7 µs after LFP; black triangles: 34 µs after LFP; light
triangles: 150 µs after LFP
Figure 4. Transient trace, monitored at λ ϭ 360 nm after LFP
(
266 nm) of 1 in ACN
ilar to the behavior detected when using acetonitrile. Again,
a transient with λmax ϭ 450 nm was detected, with a life-
time τ of the order of 30 µs (Figure 5).
The transient spectrum of transient X obtained by LFP
of 1 is very similar to the UV/Vis spectrum assigned to im-
inoquinone methide sa-3, matrix-isolated in Ar.[ For sa-3,
the reactivity of a reactive diene, undergoing facile
DielsϪAlder cycloadditions and addition of nucleophiles,
was anticipated. Quenching experiments were therefore per-
formed, using dienophiles such as 2-chloroacrylonitrile
1]
Figure 2. Plot of the initial transient intensity at λ ϭ 450 nm, ob-
served after LFP (266 nm) of 1 in ACN, versus the laser power (CAN) and nucleophiles such as methanol. Indeed, the 450-
(
measured at the place of the sample); the straight line represents
nm transient was quenched both by methanol and by CAN.
Hence an assignment of transient X to sa-3, possibly plus
a linear fit to the experimental data points
Some LFP experiments were performed with DMF as another stereoisomer of 3, appears to be a plausible
solvent. The transient spectra and lifetimes were virtually working hypothesis.
identical to those obtained in acetonitrile, while the quench-
The traces recorded at λmon ϭ 450 nm reveal the presence
ing rate constants may differ (see below). If cyclohexane or of more than one transient species. In order to identify the
n-hexane were used as solvent, the transient behavior de- transients observed, it was therefore necessary to deconvo-
tected upon LFP (λexc ϭ 355 nm) of 1 was again very sim- lute the superposition of transient traces observed. This
2466
Eur. J. Org. Chem. 2001, 2463Ϫ2475

A Laser Flash Photolysis Study on 2-Azido-N,N-diethylbenzylamine
FULL PAPER
ics, a standardized technique was employed to obtain con-
sistent and reproducible results. At low quencher concentra-
tions, traces were fitted according to first-order kinetics.
Upon addition of quencher, the decay rate constants for the
two faster components increased, reducing the quality of
the first-order fit, and allowing a biexponential fit (which
is very difficult if both kdecay are similar). This point was
4
Ϫ1
usually reached when k1st(450) ϭ 6·10 s . As the rate con-
stant kslow for decay of the slow component 5 proved to be
independent of quencher concentration (with the exception
of amines used as quenchers: see above), a typical value of
4
Ϫ1
kslow ϭ 2·10 s was used in fitting the traces, thereby redu-
cing the number of parameters to be fitted. Starting from
5
Ϫ1
kfast ϭ 6·10 s , the traces could again be fitted according
to first-order kinetics, as the slow component was practic-
ally stable in this time regime.
Figure 5. Transient spectra, measured during different time win-
The quenching plots measured in this study can be di-
dark dots: 3 µs after LFP; light dots: 11 µs after LFP; black tri- vided into two categories. Quenching plots obtained when
dows after LFP (355 nm) of 1 in cyclohexane purged with argon;
angles: 30 µs after LFP
employing non-hydroxy quenchers (that is, dienophiles and
amines) proved to be linear (Figure 6). If the quencher con-
could partially be achieved by selective quenching of some
tained at least one hydroxy functionality, on the other hand,
of the transients. While approximate fitting of the traces
according to first-order kinetics was possible at low
the plots were nonlinear, consisting of an initial section in
which kdecay was almost independent of quencher concen-
quencher concentrations, further addition of quencher led
tration, and a second, linear part, in which efficient quench-
to traces that could be much better fitted using biexponen-
ing was observed. Figure 7 shows quenching plots obtained
tial kinetics. In typical quenching experiments (using meth-
using water and [D ]water. The concentration needed for
2
anol as quencher, for example), two components of the
transient signal monitored at λ ϭ 450 nm were quenched,
the onset of efficient quenching was dependent on the na-
ture of the quencher employed. It is interesting that deuter-
while one slow component was not affected. The decay rate
ium oxide reproducibly required a slightly higher concentra-
constant of this slow third component was an invariant
tion than water, while the slope of the linear section was
4
Ϫ1
kslow ϭ 2·10 s . The identity of the slow component can-
identical in both isotopomers (Figure 7). No such observa-
tion was made when comparing the quenching plots ob-
tained using CH OH vs. CH OD. In this case, however, the
not be deduced from comparison with matrix data; how-
ever, a broad λ
at λ Ͻ 400 nm is characteristic of didehy-
max
36]
3
3
[
droazepines. The possibility that the slow component can
be attributed to such a seven-membered ring is supported
by the observation that it is quenched by diethylamine. In
the case of this quencher, the lifetime of the slow compon-
ent proved not to be independent of quencher concentra-
slopes of the linear sections differed by a factor of two (Fig-
ure 8).
[
37]
tion. The rate constant for the reaction between the slow
4
component and diethylamine was obtained as k ϭ 5.8·10
q
Ϫ1 Ϫ1
, which is in a typical range.[36] One possible struc-

s
ture for the slow component is therefore didehydroazepine
5
.
As far as the growth at λ ϭ 370 nm is concerned, a clear
assignment cannot be made. The rate constant of the 370-
nm growth is nearly independent of quencher concentra-
tion; addition of 7% triethylamine merely changed it from
5
Ϫ1
5 Ϫ1
an initial k ϭ 7.1·10 s to k ϭ 7.5·10 s . Addition of 1%
thiophenol, however, resulted in efficient quenching. One
possible interpretation might be the formation of trans-2,2Ј-
bis(diethylaminomethyl)azobenzene by dimerization of two
molecules of triplet 2-(diethylaminomethyl)phenylnitrene.
Figure 6. Plot of the decay rate constant after LFP (266 nm) of 1
in acetonitrile, monitored at λ ϭ 450 nm, versus the concentration
of n-propylamine added; the line represents the linear fitting of
the data
Laser Flash Photolysis: Quenching Studies in Polar
Solvents
The reactivity of X was examined using a series of quen-
chers, including dienophiles and a variety of different nucle-
Another detail deserving mention is the observation that
ophiles. As the decay at λ ϭ 450 nm obeyed complex kinet- the addition of small concentrations of relatively unreactive
Eur. J. Org. Chem. 2001, 2463Ϫ2475
2467

G. Bucher
FULL PAPER
Figure 9. Transient traces, recorded after LFP (266 nm) of 1 in
ACN; lower trace: in pure acetonitrile; upper trace: 3.2% tert-butyl
alcohol added; inset: plot of the decay rate constant, monitored at
λ ϭ 450 nm, vs. [tBuOH]
Figure 7. Plots of the decay rate constant at λ ϭ 450 nm, LFP
(
266 nm) of 1 in acetonitrile, versus quencher concentration; black
dots: quenching with water; light dots: quenching with deuterium
oxide
Figure 10. Plots of the decay rate constant at λ ϭ 450 nm, LFP
(355 nm) of 1 in DMF, versus quencher concentration; light dots:
quenching with cytosine; black dots: quenching with adenine; black
triangles: quenching with thymine; inset: plot of the decay rate con-
stant after LFP (266 nm) of 1 in acetonitrile, monitored at λ ϭ
Figure 8. Plots of the decay rate constant at λ ϭ 450 nm, LFP
(
266 nm) of 1 in acetonitrile, versus quencher concentration; black
dots: quenching with MeOH; light dots: quenching with MeOD
4
50 nm, versus the concentration of thiophenol added
hydroxy quenchers such as 2-propanol or tert-butyl alcohol
led to a small, but reproducible decrease in kdecay (see Fig-
ure 9). With phenol as quencher, the quenching plot ob-
tained was curved downward, reaching a plateau at
(
3) Deuterium kinetic isotope effects are small (meth-
18
anol) or not detectable (water). No O-isotope effect was
observed for water quenching.
[
phenol] ϭ 0.03 . Similar effects were observed with thiols
(
4) Reactions with methanol are significantly more rapid
and the DNA bases as quenchers. Here the plateau was
reached very early ([RϪSH] Ͼ 0.2 m, [cytosine] Ͼ 1 m)
in cyclohexane and acetonitrile than in DMF.
(
Figure 10). The implications of these observations are dis-
Quenching Studies Using Cyclohexane as Solvent
cussed below. The pseudo first-order rate constants ob-
tained in the quenching studies are summarized in Table 1.
Some conclusions from Table 1:
The reactivity of transient X was also examined using
cyclohexane as solvent. Here, the kinetics of the transient
(
1) X is quenched by electron-deficient dienophiles such decay, monitored at λ ϭ 450 nm, are no longer triexponen-
as maleic anhydride, but not by simple olefins such as cyclo- tial, but can be fitted in terms of a biexponential decay.
hexene or tetramethylethylene. This made it possible to obtain individual quenching rate
(
2) Particularly efficient quenching is found with thiols, constants for both species absorbing at λ ϭ 450 nm. In the
7
or if the quencher has two or more hydroxy groups in close case of methanol quenching, k was determined as 6.9·10
proximity. Introduction of steric hindrance reduces the
quenching rate constant.
fast
Ϫ1 Ϫ1
5
Ϫ1 Ϫ1

s
(Figure 11) and kslow ϭ 1.4·10 
s
. The num-
ber for kslow may not be significant, as kdecay,slow varied only
2468
Eur. J. Org. Chem. 2001, 2463Ϫ2475

A Laser Flash Photolysis Study on 2-Azido-N,N-diethylbenzylamine
FULL PAPER
Table 1. Second order rate constants for reactions between
transient X (λmon ϭ 450 nm) and various quenchers
4
Ϫ1 Ϫ1 [a]
Quencher
-Chloroacrylonitrile
Rate constant (10 
s )
2
1400
Maleic anhydride
Cyclohexene
Tetramethylethylene
Acetonitrile
2400
no quenching
no quenching
Ͻ 0.15[
no quenching
460, 95 , 850
b]
Acetone
[c]
[d]
f]
Methanol
[
e]
[D
1
]Methanol
230
180
33
Ethanol
Isopropyl alcohol
tert-Butyl alcohol
Water
no quenching[
66
[
[
D
2
]Water
68
68
1
8
[g]
O]Water
1400[
2300
2900
1900
15000
100000
29
110
6.4
130
510
h]
Phenol
Ethylene glycol
,3-Propanediol
trans-1,2-Cyclohexanediol
Glycerol
Figure 11. Plot of the rate constant of the fast component of the
decay at λ ϭ 450 nm, observed after LFP (266 nm) of 1 in cyclo-
hexane, versus methanol concentration
1
-Glucose
Calculations
Triethylamine
Diethylamine
n-Propylamine
Ethylenediamine
In order to obtain further evidence for the hypothesis
that transient X might be a mixture of stereoisomers of im-
inoquinone methide 3, transition state geometries for some
representative reactions of stereoisomers of 4 were optim-
ized using DFT calculations [B3LYP/6-31G(d); 4 was
chosen rather than 3 in order to reduce computation time].
The energies are summarized in Table 2. As reference
points, a number of complexes formed from isomers of 4
and the corresponding quencher molecules were optimized.
Some conclusions from Table 2:
2-Aminoethanol
Aniline
Pyridine
Pyridine N-oxide
Adenine
Cytosine
Thymine
Thiophenol
n-Dodecanethiol
14
0.37
780
6400[
c] [i]
[c] [j]
34000
770[
c] [k]
120000[
l]
l]
l]
[
110000
1
,3-Dimercaptopropane
130000[
3
(1) In DielsϪAlder reactions, sa-4 is predicted to show a
prominent preference for electron-deficient dienophiles
Oxygen ( O
2
)
no quenching
[
a]
Measured at room temp. in acetonitrile, unless otherwise noted. such as acrylonitrile. The calculations predict a preferential
Because of the multiexponential decay, the accuracy of the numbers
given is generally limited to Ϯ10%. Very low rate constants, such
as measured for pyridine quenching, are less accurate (Ϯ20%). If
ortho orientation of the substituents in the product.
2) With simple olefins, reaction rates are predicted to be
considerably lower. Additional alkyl substituents are pre-
(
the quenching plot is strongly curved as in footnotes[
d,fϪi]
, the accu-
racy is estimated to Ϯ30%. The rate constants given refer to the dicted to lead to an increase in the barrier.
bulk concentration of quencher. No distinction is made between
(
3) The reactivity of the iminoquinone methides is pre-
reactions with monomeric quenchers, and reactions in which clus-
dicted to depend sensitively on the orientation of the imino
hydrogen atom and the dialkylamino functionality. In reac-
In cyclohexane, room tions with ethylene, the anti-anti isomer aa-4 is predicted to
[
b]
ters of quencher may be involved. Ϫ
kdecay (X) divided by the
molarity of pure acetonitrile. This value therefore represents an up-
In DMF, room temp. Ϫ
temp., first order fit of the traces. Ϫ CH
[
c]
[d]
per limit. Ϫ
[e]
[f]
3
OD. Ϫ See Figure 8. show the highest reactivity. According to preliminary AM1
[
g]
[h]
Ϫ
90% isotopic enrichment. Ϫ Quenching plot curved down-
calculations, this statement also holds if as-4 is included in
the consideration.
(
ward (see text). This value corresponds to the slope of the plot at
[
i]
[
phenol] Ͻ 0.03 . Ϫ Quenching plot curved downward (see text).
4) In the TS of the reaction between sa-4 and ethylene,
This value corresponds to the slope of the plot at [adenine] Ͻ 3 m.
[
j]
Ϫ
Quenching plot curved downward (see text). This value corre- both the new CϪC bond (r ϭ 235.7 pm) and the new CϪN
[k]
sponds to the slope at [cytosine] Ͻ 1 m. Ϫ
Quenching plot
bond (r ϭ 214.1 pm) are formed to a comparable degree,
indicating, that the reaction is both concerted and syn-
chronous. Similar geometries were computed for the TSs of
other IQM ϩ olefin reactions.
curved downward (see text). This value corresponds to the slope at
thymine] Ͻ 16 m. Ϫ [ Quenching plot curved downward (see
l]
[
text). This value corresponds to the slope at [RϪSH] Ͻ 0.2 m.
(
5) The TS of the reaction between aa-4 and water was
4
Ϫ1
4 Ϫ1
between 4·10 s and 6·10 s within the limited concen- also optimized at the B3LYP/6-31G(d) level of theory. In
tration range available for methanol in cyclohexane. First the TS, the new NϪH bond is more fully developed than
order fitting of the traces gave a value between these figures the new CϪO bond [r(NϪH) ϭ 130.5 pm, r(HϪO) ϭ 119.1
6
Ϫ1 Ϫ1
(
k ϭ 8.5·10 
s ). This value is higher than that ob- pm, r(CϪO) ϭ 213.5 pm]. It is characterized by one mode
tained when using acetonitrile or DMF as solvent.
with negative force constant, which may be described as the
Eur. J. Org. Chem. 2001, 2463Ϫ2475
2469

G. Bucher
FULL PAPER
Table 2. Reactions of iminoquinone methides; calculated energies of reactants, reactant complexes, and transition states; activation ener-
gies and enthalpies of complex formation
Substrate Quencher
Energy of
separate
Energy of
precursor
Energy of
TS [H]
Barrier
[kcal/mol] [kcal/mol]
∆H(complex) Dipole
moment
Dipole
moment
reactants [H] complex [H]
(complex) [D] (TS) [D]
sa-4
aa-4
ss-4
Ethylene
Ethylene
Ethylene
Ϫ537.967596 Ϫ537.970756
Ϫ537.939573 19.6
Ϫ2.0
Ϫ
Ϫ
6.027
Ϫ
Ϫ
4.513
2.678
4.123
4.122
7.278
6.170
3.438
4.359
3.371
6.244
5.602
Ϫ538.219781 not calculated Ϫ538.198401 15.4[
a]
a]
Ϫ538.213891 not calculated Ϫ538.182500 21.7[
sa-4
sa-4
sa-4
aa-4
aa-4
aa-4
aa-4
aa-4
Tetramethylethylene Ϫ695.484327 Ϫ695.486128
Ϫ695.442498 28.4
Ϫ630.184455 20.1
Ϫ630.200122 10.3
Ϫ536.045137 10.2
Ϫ612.237014 7.9
Ϫ575.351688 9.0
Ϫ689.602043 4.4
Ϫ516.170127 15.9
Ϫ0.8
Ϫ2.8
Ϫ2.8
Ϫ12.3
Ϫ18.2[
Ϫ13.1
Ϫ13.2
Ϫ7.3
6.238
9.788
9.788
3.953
4.264
3.911
3.910
2.773
[
b]
Acrylonitrile
Ϫ630.212071 Ϫ630.216468
Ϫ630.212071 Ϫ630.216468
Ϫ536.041281 Ϫ536.065314
Ϫ612.220592 Ϫ612.249587
Ϫ575.346752 Ϫ575.370719
Ϫ689.587951 Ϫ689.609028
Ϫ515.950129 Ϫ515.961708
Acrylonitrile[
Water
c]
d]
Water dimer
Methanol
Ethylene glycol
Ammonia
[
a]
[b]
Assuming a ∆H(complex) of Ϫ2.0 kcal/mol as in the case of sa-4. Ϫ TS leading to 4-(dimethylamino)tetrahydroquinoline-2-carbonitr-
ile, cyano group exo. Ϫ [ TS leading to 4-(dimethylamino)tetrahydroquinoline-3-carbonitrile, cyano group exo. Ϫ Reaction enthalpy
c]
[d]
for aa-4 ϩ water dimer Ǟ complex.
migration of a proton from water to the imino nitrogen vibrational spectra [all calculations B3LYP/6-31G(d)].
atom of aa-4. The calculated activation energy is 10.2 kcal/ Table 3 gives the relevant energies, and Figure 12 shows the
mol (in this system, the TS energy has to be compared with results in graphic form. The zero-point was set as the energy
the energy of the complex [aa-4 ϫ H O], which is calculated of the triplet nitrene plus 18.3 kcal/mol, which is the singlet-
2
to be lower in energy than aa-4 ϩ H O by 12.3 kcal/mol). triplet energy gap determined experimentally for parent
2
A similar situation is encountered when the reaction be- phenylnitrene.[
38,39]
The triplet energy of sa-4 is obtained as
tween aa-4 and methanol is calculated. Here, the barrier ∆ES-T ϭ 25.5 kcal/mol [(U)B3LYP/6-31G(d)].
starting from [aa-4 ϫ MeOH] is predicted to amount to 9.0
kcal/mol, slightly less than the activation energy calculated
for water as quencher.
Discussion
(
6) The barrier for the reaction between sa-4 and ethylene
Identification of Transient X
glycol is calculated to be significantly smaller than the bar-
rier for the reaction with methanol or water. In the trans-
Iminoquinone methides sa-3 and aa-3 are formed photo-
ition state, r(CϪO) is 241.4 pm, while r(OϪH) is calculated chemically (λexc ϭ 320 nm) from azide 1, matrix-isolated in
as 129.3 pm and r(NϪH) as 121.1 pm. Unlike the reaction Ar at T ϭ 10 K.[ While the data obtained with matrix
between sa-4 and water or methanol, the reaction between isolation spectroscopy thus draw a clear picture, the situ-
sa-4 and ethylene glycol predominantly has the character of ation is definitely less obvious when looking at the product
a protonation of the imino group. The second hydroxy studies, LFP data, and calculations presented in this contri-
group of ethylene glycol forms a hydrogen bond with the bution. Certainly, a significant part of the photochemistry
1]
attacking hydroxy functionality [r(O···H) ϭ 210.7 pm].
7) The barrier calculated for reaction with dimeric water singlet nitrene chemistry, eventually yielding methyl-
is calculated to be significantly smaller than the barrier for eneazepine 9. This statement is supported both by the re-
reaction with monomeric water. sults obtained by LFP and by the product studies. Most
8) For the reaction with ammonia, a significant barrier probably, some of the products obtained from irradiation
of 1 runs via the well-established pathways of conventional
(
(
of 15.9 kcal/mol is predicted.
of 1, such as the aniline 7, derive from triplet nitrene chem-
The relative DielsϪAlder reactivities of the dialkylamino- istry. There is also some evidence, from the LFP studies,
substituted iminoquinone methides predicted by DFT are that a triplet nitrene may be involved, as the growth at λ ϭ
in qualitative accord with the experimental findings, where 370 nm may be assigned to formation of an azo dye. Many
high reactivity towards electron-deficient dienophiles such of the products identified (such as 2-aminobenzaldehyde),
as maleic anhydride was observed, and no reactivity to- however, cannot be formed by these conventional routes.
wards olefins such as cyclohexene or tetramethylethylene. The reactivity pattern found with transient X (high reactiv-
As far as the reactions with nucleophiles are concerned, the ity towards electron-deficient dienophiles and nucleophiles)
relative reactivities calculated for the reactions between aa- limits the choice of possible structures to a number of react-
4
and water or methanol are in agreement with the experi- ive molecules. These are iminoquinone methides 3, 2-(di-
mental findings (methanol being more reactive). ethylaminomethyl)(phenyl)nitrene (20) (singlet or triplet),
Additionally, the geometries of a number of relevant sta- and (2-aminophenyl)(diethylamino)carbene (21). Scheme 5
tionary points on the C H N surface were optimized and depicts the pathways by which the intermediates should be
9
12
2
characterized as minima or transition states by calculating formed.
2470
Eur. J. Org. Chem. 2001, 2463Ϫ2475

A Laser Flash Photolysis Study on 2-Azido-N,N-diethylbenzylamine
FULL PAPER
Table 3. Energies (in Hartrees) and dipole moments of stationary points relevant to this study
Stationary point
Energy [H]
Energy ϩ ZPC [H]
Dipole moment [D]
sa-4
Ϫ459.627618
Ϫ459.633110
Ϫ459.626510
Ϫ459.632777
Ϫ459.586720
Ϫ459.576891
Ϫ459.585313
Ϫ459.593840
Ϫ459.613308
Ϫ459.593662
Ϫ459.583425
Ϫ538.256879
Ϫ538.032909
Ϫ459.431370
Ϫ459.436668
Ϫ459.430473
Ϫ459.436514
Ϫ459.393392
Ϫ459.383755
Ϫ459.392397
Ϫ459.399773
Ϫ459.417478
Ϫ459.402446
Ϫ459.390683
Ϫ538.003088
Ϫ537.778861
6.122
4.153
5.036
3.261
5.755
9.951
4.982
6.630
2.006
3.289
4.014
6.377
3.798
aa-4
ss-4
as-4
TS aa-4/sa-4
TS sa-4/ss-4
TS ss-4/as-4
TS aa-4/as-4
Carbene 22
TS aa-4/carbene 22
sa-4 triplet
sa-3
aa-3
Figure 12. Schematic representation showing the relative energies of important stationary points and calculated activation energies on
the C surface
9 12 2
H N
Nitrene 20 can be ruled out as a possible candidate. Sing-
The iminoquinone methides 3 represent intermediates
let nitrenes are known to be very short-lived, with a lifetime that should be quenchable by dienophiles and by nucleo-
[
40Ϫ44]
in the ps range.
Triplet nitrenes, on the other hand, philes. Reaction between 3 and water should eventually
react sluggishly in intermolecular reactions, preferentially yield 2-aminobenzaldehyde (6), which is indeed formed in
decaying through H abstraction or dimerization.[ In this significant amounts if 1 is photolyzed in wet acetonitrile.
study, a transient growth (λ ϭ 370 nm) has tentatively been With acrylonitrile as quencher, 4-(diethylamino)tetrahydro-
assigned to dimerization of triplet nitrene 20, yielding an quinoline-3-carbonitriles should be formed, and these are
azo dye. This growth, however, does not correlate with the in fact probably detected. Further support for the observa-
45]
decay of transient X, which represents a further argument tion of 3 comes from the fact that the UV/Vis spectrum
of sa-3/aa-3 in argon matrix[ and the transient spectrum
1]
against an assignment of X as triplet nitrene 20.
The case against carbene 21 is not as evident. This species observed immediately after the laser pulse are very similar.
would have to be formed by transfer of a second hydrogen It thus appears very likely that transient X can indeed be
atom to the imine nitrogen atom. The second 1,4-H shift assigned as a mixture of probably two stereoisomers of
required for the formation of carbene 22 (R ϭ NMe in- neutral iminoquinone methide 3. These intermediates are
2
stead of NEt ) is calculated by B3LYP/6-31G(d) ϩ ZPE therefore, in fact, surprisingly short-lived species (τ ഠ 30
2
46]
[12,13]
[
MP2/6-31G(d) ϩ ZPE gives similar results] to be endo- µs),[ much shorter lived than o-quinone methide,
thermic by 11.9 kcal/mol, requiring an activation energy of and even shorter lived than didehydroazepine 5. This was
E ϭ 24.3 kcal/mol (Figure 12, Table 3). For this reason, also observed in the low-temperature studies presented pre-
a
viously:[ If an organic glass containing both the didehy-
1]
formation of carbene 21 can be ruled out.
Eur. J. Org. Chem. 2001, 2463Ϫ2475
2471

G. Bucher
FULL PAPER
the four products with m/z ϭ 229. However, the mechanistic
scenario still requires clarification.
Curved Quenching Plots
In the quenching studies, curved quenching plots were
very often observed. Upon addition of small concentrations
of hydroxy quenchers such as methanol or water, the life-
time of the iminoquinone methide mixture hardly changed.
Starting from a certain threshold of quencher concentra-
tion, a linear rise in kdecay was observed (quenching plot
curved upward, see Figure 7). In certain cases, such as
quenching by phenol, thiols, or DNA bases, a plateau re-
gion was reached (quenching plot curved downward, see
Figure 10). If relatively unreactive hydroxy quenchers such
as 2-propanol or tert-butyl alcohol were employed, a de-
crease in kdecay was observed upon addition of relatively
small concentrations of quencher (see Figure 9).
The quenching plots could reflect several factors com-
plicating the system studied: complex formation between
quencher and iminoquinone methide, complex formation
between azide and quencher, and dimerization or oligo-
merization of the quenchers themselves.
Quenching Plots involving Hydroxy Quenchers
Alcohols dimerize or oligomerize in aprotic solvents. If
the alcohol is small, the dimer is usually more reactive than
the monomer, producing quenching plots that are curved
upward. Such quenching plots have frequently been ob-
served for reactions between carbenes and alcohols.[
In a study on the reactions between silyloxycarbenes and
alcohols, conducted in acetonitrile, Kirmse, Guth, and
Steenken pointed out, that in this particular polar solvent,
Scheme 5
droazepine 5 and iminoquinone methides sa-3 ϩ aa-3 is
thawed, sa-3 and aa-3 disappear before 5.
48Ϫ51]
Product Studies
A number of open questions remain to be discussed. As methanol oligomerization plays no role up to methanol
far as the product studies are concerned, the following ob- concentrations of 0.4 .[ In the case of reactions between
52]
servations deserve comment:
silyloxycarbenes and weakly acidic alcohols, the quenching
(1) Small amounts of 2,1-benzisoxazole (8) are formed plots were curved upward. For these reactions, the authors
upon pulsed-laser irradiation of 1. This compound is prob- suggested a mechanism in which one molecule of alcohol
ably formed by photolysis of 2-azidobenzaldehyde. 2-Azido- forms a complex with the carbene, while the second molec-
benzaldehyde is in turn formed in three steps from the rad- ule of alcohol then assists in the conversion of the
[
47]
[52]
ical cation of 1.
The observation that 8 is detected in carbeneϪalcohol complex to the final ether product. Ac-
small yield only if 1 is irradiated using a pulsed laser indic- cording to Kirmse et al., more acidic alcohols such as 2-
ates that direct photoionization of 1 requires at least two chloroethanol do not require activation in a ternary car-
photons.
bene/alcohol dimer complex. The downward-curved
(
2) Four isomeric products with m/z ϭ 229 are formed quenching plots observed for these quenchers thus have to
upon photolysis of 1 in the presence of acrylonitrile. Only be due to alcohol association.[
cis- and trans-4-diethylamino-1,2,3,4-tetrahydroquinoline- Most of the peculiarities found with the quenching plots
-carbonitrile, however, can be expected to be formed in a presented in this study can be explained in terms of the
52]
3
concerted [4 ϩ 2] cycloaddition, as the pathways yielding model presented by Kirmse, Guth, and Steenken. Accord-
cis- or trans-4-diethylamino-1,2,3,4-tetrahydroquinoline-2- ing to the calculations presented above (Table 2), aa-4 forms
carbonitrile are calculated to have a prohibitively high bar- complexes with water or methanol, which gives rise to a
rier (see Table 2). At this point, it should be noted that non- significant stabilization (∆Hcomplex ϭ Ϫ 13.05 kcal/mol for
regiospecific cycloadditions have been observed before for aa-4 ϩ methanol in the gas phase) of the iminoquinone
cycloadditions involving iminoquinone methides. Ito et al.[3] methide. If the complexing alcohol is bulky, such as tert-
obtained a mixture of both regioisomeric tetrahydroquinol- butyl alcohol, complex formation may in fact result in a
ine carbonitriles upon generation of N-benzyl-o-iminoqui- kinetic stabilization of 3, since dimerization of the imin-
none methide in the presence of acrylonitrile. Given this oquinone methide/tert-butyl alcohol complex will be im-
precedent, it seems reasonable to assign structures 16Ϫ19 to peded relative to the uncomplexed iminoquinone methide.
2472
Eur. J. Org. Chem. 2001, 2463Ϫ2475

A Laser Flash Photolysis Study on 2-Azido-N,N-diethylbenzylamine
FULL PAPER
This explains the decrease in kdecay observed upon addition methides generated in this work. If the alcohol already con-
of small to medium concentrations of tert-butyl alcohol or tains a second hydroxy functionality to catalyze the reac-
2
-propanol. If the alcohol concentration is increased fur- tion, no second alcohol molecule is required, and the ac-
ther, product formation through the ternary iminoquinone tivation entropy is more favorable. The calculated geometry
methide/alcohol dimer complex becomes more significant, of the transition state of the reaction between sa-4 and
resulting in an increase in kdecay (see Figure 9). In quench- ethylene glycol indicates the presence of a hydrogen bond
ing plots with deuterium oxide, the onset of efficient formed by the second hydroxy group Ϫ which is not directly
quenching requires a higher quencher concentration than involved in the reaction Ϫ and the attacking hydroxy group.
in quenching plots obtained using water as quencher (Fig- In the reaction with 1,3-propanediol, this hydrogen bond
ure 7). This can be explained in terms of slightly different would be part of a six-membered ring, which is more favor-
dimerization (and probably complex formation) properties able than the five-membered ring hydrogen bond present in
in the two isotopomers, due to subtle entropic effects.[
53]
the reaction with ethylene glycol. The hydrogen bond most
The idea that water dimers react more efficiently than probably increases the acidity of the attacking alcohol. A
monomeric water with 3 is confirmed by the calculations similar mechanism may operate for compounds containing
presented above (barrier for reaction between aa-4 and one hydroxy group and one amino functionality (2-amino-
water: 10.2 kcal/mol; for reaction with dimeric water: 7.9 ethanol) or two amino functionalities (ethylenediamine).
kcal/mol, see Table 2).
Finally, the extremely high reactivity of the iminoquinone
methides 3 towards thiols is to be expected on the basis of
the weak SϪH bond, and the high acidity and nucleophilic-
Quenching Plots Involving DNA Bases and Thiols
ity of these quenchers. The NϪH bond in aniline is also
relatively weak, it is a good nucleophile, and yet the rate
constant of its reaction with 3 is lower by four orders of
magnitude. The crucial difference probably lies in the very
Figure 10 shows quenching plots measured for the reac-
tions between 3 and DNA bases and thiophenol. DNA ba-
ses are well known to form dimeric complexes. Dimeric
DNA bases can be expected to react more slowly with 3
than the free bases, which should result in downward
curved quenching plots.
Dimer formation is unlikely to be of any importance in
the case of dilute solutions of thiols in acetonitrile. At least
here, the downward curvature and plateau formation ob-
served in the quenching plot shown in Figure 10 are in all
likelihood not due to any kind of association phenomenon.
The only explanation that can be offered is protonation of
azide 1 by the relatively acidic thiols. This would probably
result in completely different transient phenomena at high
thiol concentrations.
different acidities; the pK of unprotonated aniline (in
a
[
54]
DMSO) is 30.6, while thiophenol in the same solvent has
[54]
a pK of 10.3.
Thus, thiols probably act primarily as
Brönsted acids, first protonating the basic imine moiety in
. In reactions with good nucleophiles that are very weak
a
3
acids, the quencher has to act primarily as a nucleophile,
attacking at the sterically congested methylene site of 3. The
order of reactivity found with n-propylamine vs. di-
4
Ϫ1 Ϫ1
6
Ϫ1
ethylamine vs. triethylamine (6.4·10 
s
vs. 1.1·10 
Ϫ1
5
Ϫ1 Ϫ1
s
vs. 2.9·10 
s ) reflects the increase in nucleophilic-
ity upon going from a primary amine to a secondary amine,
while the reactivity of the tertiary amine is reduced by steric
constraints. In the reaction of the tertiary amine, a zwitter-
ion is the logical primary product.
Reactions of Iminoquinone Methides
The iminoquinone methides show pronounced reactivity
towards thiols, compounds containing at least two close hy-
droxy functionalities, cytosine, and adenine. If one com-
pares the figures obtained for cytosine, adenine, and thym-
ine quenching (Table 1) with the relative reactivities for re-
Why are the iminoquinone methides presented in this
work such extraordinarily reactive compounds? The reactiv-
ity of dienes is a function of the HOMOϪLUMO gap. This
gap is calculated for sa-4 [B3LYP/6-31G(d)] as 3.31 eV,
compared with 3.59 eV calculated for o-quinone methide,
3.91 eV calculated for cyclopentadienone, and 5.49 eV cal-
culated for cyclopentadiene. The very low calculated triplet
energy of 25.5 kcal/mol for sa-4 is in agreement with this
picture. Furthermore, the stereoisomers of 3 and 4 are polar
molecules (Table 3). The calculated dipole moments are of
the order of 3.8 D (aa-3) or 6.4 D (sa-3). Reactions with
electrophiles (acids) or nucleophiles should be facilitated by
this dipolar character.
actions between deoxynucleosides and o-quinone methide
as published by Pande et al.,[
23]
one finds that the relative
reactivities determined in this work (C/A/T ϭ 44:8:1) paral-
lel the results published for reactions of o-quinone methide
(dC/dG/dA ϭ 30:2.5:1; no reaction found with dT). Over-
all, the iminoquinone methides 3 are less selective than o-
quinone methide, which is in agreement with the much
shorter lifetime of 3. Given the very high reactivity of 3
towards polyhydroxy quenchers, reaction of 3 with the
sugar backbone of DNA would probably compete with
The high reactivity of 3 holds promise that azidobenzyl-
DNA alkylation. What makes diols, glycerol, and glucose amines might find application in photoaffinity labelling. To
so reactive towards 3? In reactions between (silyloxy)carb- make the reaction useful for preparative chemistry, the com-
enes and simple alcohols, a second alcohol molecule cata- peting ring expansion of the singlet nitrene to didehy-
lyzes the conversion of a carbeneϪalcohol complex to the droazepine 5 has to be suppressed, which might be achieved
[
52]
ether product. A similar mechanism may be invoked for by introduction of additional substituents in the 6-position.
the analogous reactions undergone by the iminoquinone Studies directed towards this goal are in progress.
Eur. J. Org. Chem. 2001, 2463Ϫ2475
2473

G. Bucher
FULL PAPER
[
55]
Conclusion
Guanine proved to be too sparingly soluble even in DMF. Deu-
1
8
terium and O kinetic isotope effects were determined by measure-
ment of the rate constants for both labeled and unlabeled quencher
on a single afternoon, keeping all experimental parameters exactly
the same for both isotopomers. Ϫ Azide 1 was obtained as de-
scribed in the previous contribution. Ϫ Calculations were per-
formed using Gaussian 98 or Titan software, using the hybrid
All optimized structures were characterized as
minima or transition states by calculating vibrational spectra. All
calculated reaction barriers given are corrected for zero point vibra-
Photolysis of 2-azido-N,N-diethylbenzylamine in aceto-
nitrile (or DMF, n-hexane) solution results in formation of
stereoisomeric iminoquinone methides, which can be
trapped with water to yield up to 10% of 2-aminobenzal-
dehyde (6). In addition, the formation of significant
amounts of methyleneazepine 9 in the presence of di-
[1]
[56]
[57]
[58]
B3LYP method.
ethylamine
is
indicative
of
typical
singlet
nitreneϪdidehydroazepine ring-expansion chemistry. The tional energy.
iminoquinone methides were characterized by time-resolved
laser flash photolysis. They show high reactivity towards
Acknowledgments
electron-deficient dienophiles, but do not react measurably
with simple olefins. In reactions with nucleophiles, particu- The author thanks W. Sander for supporting this work. Financial
larly high reactivity was found towards thiols, polyhydroxyl- support by the Deutsche Forschungsgemeinschaft (SFB 452) and
ated compounds (glycerol, glucose), and cytosine, while the Fonds der Chemischen Industrie is gratefully acknowledged.
simple alcohols, amines, and thymine quenched the imin-
oquinone methides less efficiently.
[
[
[
1]
2]
3]
G. Bucher, Eur. J. Org. Chem. 2001, 0000Ϫ0000, preceding pa-
per.
H. Steinhagen, E. J. Corey, Angew. Chem. 1999, 111,
2
054Ϫ2056; Angew. Chem. Int. Ed. 1999, 38, 1928Ϫ1931.
Experimental Section
Y. Ito, E. Nakajo, T. Saegusa, Synth. Commun. 1986, 16,
1
073Ϫ1080.
Product Studies: In a typical experiment, 1 (10 mg) was dissolved
in solvent (40 mL) and placed in a quartz vessel. Oxygen was re-
moved by purging the sample with Ar for 30 min. The sample was
then irradiated for 15 min with λexc ϭ 320 nm, using a Hg low-
pressure lamp with filter (Gräntzel, Germany). After removal of
the solvent, 1-bromonaphthalene (10 mg) was added as external
[4]
[5]
K. Wojchiechowski, Tetrahedron 1993, 49, 10017Ϫ10026.
M. Lancaster, D. J. H. Smith, Chem. Commun. 1980, 471Ϫ472.
K. Wojchiechowski, Tetrahedron 1993, 49, 7277Ϫ7286.
K. Wojchiechowski, Pol. J. Chem. 1997, 71, 1375Ϫ1400.
A. Vigalok, D. Milstein, J. Am. Chem. Soc. 1997, 119,
[
[
[
6]
7]
8]
7
873Ϫ7874.
A. Vigalok, L. J. W. Shimon, D. Milstein, J. Am. Chem. Soc.
998, 120, 477Ϫ483.
[9]
2 2
standard, the sample was dissolved in ca. 1 mL of CH Cl , and
1
analyzed by GC/MS (HewlettϪPackard, column: OV1, 29.5 m). If
a reference sample of a given product was available, its peak integ-
ral was calibrated against 1-bromonaphthalene by injecting 1:1
mixtures with 1-bromonaphthalene. In several experiments, pulsed
lasers [355 nm from the Nd-YAG laser used in the LFP studies
[10]
O. Rabin, A. Vigalok, D. Milstein, J. Am. Chem. Soc. 1998,
20, 7119Ϫ7120.
1
[11]
O. Rabin, A. Vigalok, D. Milstein, Chem. Eur. J. 2000, 6,
4
54Ϫ462.
[
[
[
12] [12a]
L. Diao, C. Yang, P. Wan, J. Am. Chem. Soc. 1995, 117,
5369Ϫ5370. [
12b]
B. Barker, L. Diao, P. Wan, J. Photochem.
(130 mJ/pulse), and 308 nm from a Lambda-Physik EMG 100 ex-
cimer laser operated with He/Xe/HCl, 80 mJ/pulse] were used as
excitation light sources. Ϫ Laser flash photolysis (LFP): A standard
LFP set-up was used. The apparatus consisted of a Spectra-Physics
Quanta-Ray Lab 130 Nd-YAG laser operated at 1 Hz repetition
rate and 266 nm (50 mJ/pulse, 10 ns, used whenever possible) or
Photobiol. A 1997, 104, 91Ϫ96.
13] [13a]
Y. Chiang, A. J. Kresge, Y. Zhu, J. Am. Chem. Soc. 2000,
22, 9854Ϫ9855. [
13b]
E. Modica, R. Zanaletti, M. Freccero,
1
M. Mella, J. Org. Chem. 2001, 66, 41Ϫ52.
P. Wan, B. Barker, L. Diao, M. Fischer, Y. Shi, C. Yang, Can.
J. Chem. 1996, 74, 465Ϫ475.
G. Pfister-Gouillouzo, F. Gracian, A. Senio, M. Letulle, J.-L.
Ripoll, Tetrahedron Lett. 1992, 33, 5753Ϫ5756.
A. Chrostowska, F. Gracian, J. M. Sotiropoulos, G. Pfister-
Gouillouzo, K. Wojchiechowski, Eur. J. Org. Chem. 2000,
313Ϫ318.
14]
355 nm (130 mJ/pulse, 10 ns, used if the quencher absorbed at λ ϭ
[15]
266 nm), a pulsed Xe high-pressure lamp (Müller, Germany), and
[16]
a Spex Minimate monochromator coupled to a photoelectron mul-
tiplier tube. The sample was irradiated at a 90° angle relative to
the monitoring beam. To avoid depletion of starting material and
product build-up, a flow system was used. Data acquisition was
performed using a LeCroy 9361 digital oscilloscope. The entire sys-
tem was programmed using LabView software. Solutions of 1 were
[
[
17]
18]
J. Morawietz, W. Sander, M. Träubel, J. Org. Chem. 1995, 60,
6
368Ϫ6378.
W. Sander, J. Morawietz, Tetrahedron Lett. 1993, 34,
1
913Ϫ1916.
Ϫ5
ca. 4Ϫ6·10  in acetonitrile (spectroscopic grade, used as re-
[19]
[20]
H. Tomioka, Pure Appl. Chem. 1997, 69, 837Ϫ840.
M. A. Lewis, D. Graff Yoerg, J. N. Bolton, J. A. Thompson,
Chem. Res. Toxicol. 1996, 9, 1368Ϫ1374.
H. W. Moore, R. Czerniak, Med. Res. Rev. 1981, 1, 249Ϫ280.
A. A. L. Gunatilaka, Triterpenoid Quinonemethides and Related
Compounds (Celastroloids), in: Progress in the Chemistry of
Organic Natural Products (Eds.: W. Herz, G. W. Kirby, R. E.
Moore, W. Steglich, Ch. Tamm), Springer, Vienna, New York,
ceived) for 266-nm excitation and 3 m in acetonitrile (or DMF)
for 355-nm excitation. They were purged with argon for 30 min
before starting the experiment. The alkenes and amines used as
quenchers were freshly distilled immediately prior to use. With the
exceptions of glycerol, phenol, 1,2-cyclohexanediol, pyridine N-ox-
ide, glucose, maleic anhydride, and the DNA bases, the quenchers
were injected into the flow tank neat. Glycerol was injected as a 0.4
[
[
21]
22]
1
996, vol. 67, chapter 1, p. 1Ϫ123.

solution in THF, trans-1,2-cyclohexanediol as a 0.2  solution in
[23]
P. Pande, J. Shearer, J. Yang, W. A. Greenberg, S. E. Rokita, J.
Am. Chem. Soc. 1999, 121, 6773Ϫ6779.
CH Cl , pyridine N-oxide, phenol, and maleic anhydride as 1 
2
2
solutions in acetonitrile, and glucose as a 1  solution in water. The
DNA bases were insufficiently soluble in acetonitrile. Experiments
involving them were conducted using DMF as solvent, preparing
the stock solution separately for each point on a quenching plot.
[24]
[25]
Y. Shi, P. Wan, Chem. Commun. 1997, 273Ϫ274.
Y. Shi, P. Wan, Chem. Commun. 1995, 1217Ϫ1218.
A number of possible products with m ϭ 150 were synthesized,
among them 2-amino-N-ethylbenzylamine, 2-(ethylamino)-
[26]
2474
Eur. J. Org. Chem. 2001, 2463Ϫ2475

A Laser Flash Photolysis Study on 2-Azido-N,N-diethylbenzylamine
FULL PAPER
[
[
44]
45]
benzylamine, and N-ethyl-N-methyl-o-phenylenediamine. None
of the compounds was identical to the product formed in the
photolyses.
N. P. Gritsan, D. Tigelaar, M. S. Platz, J. Phys. Chem. A 1999,
103, 4465Ϫ4469.
G. B. Schuster, M. S. Platz, Photochemistry of Phenyl Azide, in:
Advances in Photochemistry (Eds.: D. H. Volman, G. Ham-
mond, D. C. Neckers), John Wiley & Sons Inc., 1992, vol. 17,
p. 69Ϫ143.
Heinzelmann has reported on the formation of simple imin-
oquinone methides by acid-induced tautomerization of 2-ami-
nobenzaldimines. In the light of the results presented in this
study, this assignment has to be incorrect. Most probably, the
compounds wrongly identified as neutral iminoquinone me-
thides in fact are the conjugate acids of the iminoquinone me-
thides, which are the 2-aminobenzaldiminium cations; W. Hein-
zelmann, Helv. Chim. Acta 1978, 61, 618Ϫ625.
P. Subramanian, J. Rajaram, V. Ramakrishnan, J. C. Kuriacose,
Indian J. Chem. 1991, 30A, 913Ϫ919.
D. Griller, M. T. H. Liu, J. C. Scaiano, J. Am. Chem. Soc. 1982,
104, 5549Ϫ5551.
R. S. Sheridan, R. A. Moss, B. K. Wilk, S. Shen, M. Wlostow-
ski, M. A. Kesselmayer, R. Subramanian, G. Kmiecik-Lawry-
nowicz, K. Krogh-Jespersen, J. Am. Chem. Soc. 1988, 110,
7563Ϫ7564.
R. A. Moss, S. Shen, L. M. Hadel, G. Kmiecik-Lawrynowicz,
J. Wlostowska, K. Krogh-Jespersen, J. Am. Chem. Soc. 1987,
109, 4341Ϫ4349.
[
27] [27a]
G. Bucher, H. G. Korth, Angew. Chem. 1999, 111,
09Ϫ211; Angew. Chem. Int. Ed. 1999, 38, 215Ϫ218. Ϫ [27b] G.
Bucher, Angew. Chem. 1999, 111, 1247; Angew. Chem. Int. Ed.
999, 38, 1192.
2
[
46]
1
[
[
28]
29]
G. I. Kokorev, J. Gen. Chem. USSR 1990, 60, 408Ϫ409; Zh.
Obshch. Khim. 1990, 60, 469Ϫ471.
C. Betschart, B. Schmidt, D. Seebach, Helv. Chim. Acta 1988,
7
1, 1999Ϫ2021.
[
[
30]
31]
D. M. Hall, E. E. Turner, J. Chem. Soc. 1945, 694Ϫ699.
T-Y. Liang, G. B. Schuster, J. Am. Chem. Soc. 1987, 109,
7
803Ϫ7810.
[
[
32]
33]
[47]
H. Zhai, M. S. Platz, J. Phys. Org. Chem. 1996, 10, 22Ϫ26.
E. Leyva, M. J. T. Young, M. S. Platz, J. Am. Chem. Soc. 1986,
[
[
48]
49]
1
08, 8307Ϫ8309.
[
[
34]
35]
M. J. T. Young, M. S. Platz, J. Org. Chem. 1991, 56,
403Ϫ6406.
The compound with m/z ϭ 249 shows the following decay pat-
6
ϩ
ϩ
tern: m/z ϭ 249 [M ] (ca. 10%), 176 /175 [M Ϫ HNEt
H] (ca. 30%), 161 [176 Ϫ NH] (ca. 25%), 147 [175 Ϫ C
100%). The base peak with m/z ϭ 147 corresponds to the 2-
2
, Ϫ
H ]
2 4
[
[
[
50]
51]
52]
(
amino-N-ethylbenzonitrilium cation. The very low yield of 11
may be due to hydrolysis of this rather labile compound,
yielding 2-aminobenzaldehyde (6). Several attempts were made
to synthesize 11 independently. While it was possible to obtain
an aminal from 2-nitrobenzaldehyde by refluxing the aldehyde
in neat diethylamine for 48 h, all attempts to reduce this com-
pound to 11 failed. Under all conditions tried (catalytic hydro-
genation, various catalysts and solvents, presence of di-
ethylamine), the aminal functionality was reduced before the
nitro group. 2-Aminobenzaldehyde (6), on the other hand, did
not react with diethylamine, polymerizing instead.
X.-M. Du, H. Fan, J. L. Goodman, M. A. Kesselmayer, K.
Krogh-Jespersen, J. A. LaVilla, R. A. Moss, S. Shen, R. S.
Sheridan, J. Am. Chem. Soc. 1990, 112, 1920Ϫ1926.
W. Kirmse, M. Guth, S. Steenken, J. Am. Chem. Soc. 1996,
118, 10838Ϫ10849.
G. W. Robinson, J. Phys. Chem. 1991, 95, 10386Ϫ10391.
F. G. Bordwell, J.-P. Cheng, G.-Z. Ji, V. Satish, X. Zhang, J.
Am. Chem. Soc. 1991, 113, 9790Ϫ9795.
[
[
53]
54]
[
[
55]
56]
R. A. Kenley, S. E. Jackson, Y. Shunko, J. S. Winterle, J.
Pharm. Sci. 1986, 75, 648Ϫ653.
[
[
36]
37]
B. A. DeGraff, D. W. Gillespie, R. J. Sundberg, J. Am. Chem.
Soc. 1974, 96, 7491.
In the case of n-propylamine quenching, all traces could be
fitted to first-order kinetics. Here, quenching of X and the
M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M.
A. Robb, J. R. Cheeseman, V. G. Zakrzewski, Jr., M. A.
Montgomery, R. E. Stratmann, J. C. Burant, S. Dapprich, J. M.
Millam, A. D. Daniels, K. N. Kudin, M. C. Strain, O. Farkas, J.
Tomasi, V. Barone, M. Cossi, R. Cammi, B. Menucci, C. Po-
melli, C. Adamo, S. Clifford, J. Ochterski, G. A. Petersson, P.
Y. Ayala, Q. Cui, K. Morokuma, D. K. Malick, A. D. Rabuck,
K. Raghavachari, J. B. Foresman, J. Ciolowski, J. V. Ortiz, B.
B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaroni, R.
Gomperts, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham,
C. Y. Peng, A. Nanayakkara, C. Gonzalez, M. Challacombe,
P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, J. L. Andres,
C. Gonzalez, M. Head-Gordon, E. S. Replogle, J. A. Pople,
Gaussian 98, Revision A.6, Gaussian Inc., Pittsburgh, PA.,
1998.
‘
‘slow’’ transient obviously takes place with a very similar rate
4
Ϫ1 Ϫ1
constant. Again, a rate constant of the order of 6·10 
s
is consistent with a assignment of the slow component to dide-
hydroazepine 5.
M. J. Travers, D. C. Cowles, E. P. Clifford, G. B. Ellison, J. Am.
Chem. Soc. 1992, 114, 8699Ϫ8701.
R. N. McDonald, S. J. Davidson, J. Am. Chem. Soc. 1993,
[
[
[
[
[
[
38]
39]
40]
41]
42]
43]
1
15, 10857Ϫ10862.
N. P. Gritsan, T. Yuzawa, M. S. Platz, J. Am. Chem. Soc. 1997,
19, 5059Ϫ5060.
R. Born, C. Burda, P. Senn, J. Wirz, J. Am. Chem. Soc. 1997,
19, 5061Ϫ5062.
1
1
[57]
N. P. Gritsan, Z. Zhu, C. M. Hadad, M. S. Platz, J. Am. Chem.
Soc. 1999, 121, 1202Ϫ1207.
N. P. Gritsan, A. D. Gudmundsdottir, D. Tigelaar, M. S. Platz,
J. Phys. Chem. A 1999, 103, 3458Ϫ3461.
Titan Vers. 1.0.1, Schrödinger, Inc., Portland, 1999.
A. D. Becke, J. Chem. Phys. 1993, 98, 5648Ϫ5652.
Received February 1, 2001
[
58]
[O01047]
Eur. J. Org. Chem. 2001, 2463Ϫ2475
2475