Photocycloaddition and ortho-hydrogen abstraction reactions of methyl arylglyoxylates: Structure dependent reactivities

Article abstract of DOI:10.1039/a901092k

In photocycloaddition reactions between methyl arylglyoxylates and certain cyclo-1,3-dienes, the diastereochemical outcome of the photoproducts is shown to depend on the steric demand of the aryl group in the glyoxylate. Exclusive endo-aryloxetanes were p

Full text of DOI:10.1039/a901092k

View Article Online / Journal Homepage / Table of Contents for this issue
Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Photocycloaddition and ortho-hydrogen abstraction reactions of
methyl arylglyoxylates: structure dependent reactivities†
Shengkui Hu‡ and Douglas C. Neckers*
Center for Photochemical Sciences,1 Bowling Green State University, Bowling Green,
Ohio 43403-0213, USA
Received (in Gainesville, FL) 3rd February 1999, Accepted 14th May 1999
In photocycloaddition reactions between methyl arylglyoxylates and certain cyclo-1,3-dienes, the diastereochemical
outcome of the photoproducts is shown to depend on the steric demand of the aryl group in the glyoxylate.
Exclusive endo-aryloxetanes were produced with bulky arylglyoxylates while no significant stereoselectivity was
observed with glyoxylates containing less demanding aryl groups. This observation is rationalized by considering
the stability of possible conformers of the intermediate 1,4-biradical at the instance of intersystem crossing.
When these ortho-substituted phenylglyoxylates were irradiated in benzene, different reaction patterns were
observed with different substituents on the aryl rings. The rates of ortho-hydrogen abstractions vary significantly
among the individual compounds. Photoproducts thus resulting are dependent on the substrate structures.
Introduction
Intermolecular photocycloaddition reactions between triplet
excited carbonyl compounds and alkenes are of recent research
interest from both mechanistic and synthetic points of view.2
In a series of studies, Griesbeck et al. demonstrated that the
diastereoselectivity in the resulting oxetane products depends
on both the size of the α-substituent of the carbonyl com-
pound and the substituent pattern as well as the flexibility of
the alkene molecules.3 The diastereoselectivity in the photo-
products was traced to the relative stability of the possible con-
formers adopted by the triplet 1,4-biradical intermediate at
the point of intersystem crossing.3 The cycloaddition reaction
between alkyl phenylglyoxylates and various alkenes had been
previously studied.4,5 Different alkyl esters of phenylglyoxylic
acid were shown to react with certain alkenes and the diastereo-
chemistry of the photoproducts depended on the size of the
ester alkyl group of the α-keto ester.5,6 Herein, we report that
the reaction outcome from the cycloaddition reactions between
methyl arylglyoxylates and some cyclo-1,3-dienes is strongly
dependent on the aryl group in the ester molecules, i.e. the
diastereochemistry of the resulting bicyclic oxetanes is deter-
Scheme 1
mined by the steric demand of the aryl group in the glyoxylate.
Similar reactions between other triplet carbonyl compounds
and furan were recently reported by Griesbeck’s group.7 We
also reveal herein significantly different photoreaction patterns
when these ortho-substituted arylglyoxylates are irradiated
in benzene. This results from the difference in the rates of
hydrogen abstraction by the excited carbonyl group from the
ortho-substituents.
the aryl group in 1 was systematically altered. Since it was
known that 13C NMR chemical shifts of carbonyl groups in aryl
carbonyl compounds are sensitive to the degree of coplanarity
of the carbonyl group with the aryl section of the molecule,8
the ground state geometry of these molecules can be inferred
from their NMR spectra. Twisting the carbonyl group out of
the plane of the aryl ring increases the chemical shift (δ) of
the carbonyl carbon signal. The signals of the ester carbonyl
carbons do not vary significantly among these compounds
whereas those of the keto carbons change with different aryl
groups. The δ values of the keto carbonyl group in 1a and 1b are
significantly lower than that of the unsubstituted phenyl com-
pound 1c, indicating that the keto groups in 1a and 1b are more
coplanar with the adjacent aryl rings than their counterpart
in 1c. It has been shown that most of the excited state energy in
the lowest triplet excited states of the carbonyl chromophore
in 1a and 1b is delocalized over the neighboring aryl rings, and
that the lowest triplet states of 1a and 1b are π, π* in character
as compared to the lowest n, π* triplet excited state of 1c.9 The
keto carbonyl group of 1d is forced out of planarity with the
Results and discussion
Photocycloaddition
Arylglyoxylates 1 (Scheme 1) were synthesized and character-
ized as described in the Experimental section. The bulkiness of
† 1H and 13C NMR (APT) spectra for all new compounds are available
as supplementary data available from BLDSC (SUPPL. NO. 57569,
pp. 19) or the RSC Library. See Instructions for Authors available via
the RSC web page (http://www.rsc.org/authors).
‡ Present address: PPG Industries, Inc., 4325 Rosanna Drive, Allison
Park, PA 15101, USA.
J. Chem. Soc., Perkin Trans. 2, 1999, 1771–1778
1771

View Article Online
Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
attached phenyl ring by the o-methyl groups as indicated by its
high δ value.
Irradiating a mixture of 1 and an equimolar quantity of
furan in benzene (0.06 M) results in cycloaddition products 2,
Scheme 2. Reaction of 1a with cyclohexa-1,3-diene under the
GC and GC/MS analyses of the reaction mixtures. We could
not isolate 2b from this mixture. Oxetane 2c was isolated in 65%
yield from the reaction of 1c. The exo-phenyl product was also
observed and the ratio of 2c/exo-phenyl product is determined
to be 90:10 determined from GC measurements. Compound 2d
was isolated in 33% yield in addition to 5 from reactions of 1d
with furan. No other isomers of 2d can be detected by GC or
GC/MS analyses of the reaction mixtures. Oxetane 2e is the
only product from the reaction of 1e in quantitative yield.
Compound 1f is photostable under identical reaction con-
ditions and the starting materials were recovered after extended
irradiation.
It seems that the yield of 2 and the diastereoselectivity of
the photoproduct depend on the steric demand of the aryl rings
in 1. Bulky aryl groups favor the endo-aryl geometry in the
product and increase the yield of 2. The endo-aryl selectivity
follows the order 2e ≈ 2d > 2a > 2c > 2b. This observation is
rationalized as being analogous to that of other excited carbonyl
compounds.7
Scheme 2
same conditions results in 3a. The structures of 2 and 3a are
established by complete spectroscopic characterizations. The
endo-aryl stereochemistry is derived from the NOE experi-
mental data such as those for compounds 2a and 3a shown in
Scheme 3, and is further confirmed by comparing the 1H NMR
A general mechanistic picture for the cycloaddition reaction
between an excited carbonyl compound and an alkene is dis-
played in eqn. (1). The initial formation of a C–O bond between
the excited carbonyl compound (T) and the ground state alkene
leads to the triplet biradical, 3B. The rate restriction due to spin-
forbidden intersystem crossing (ISC) has to be overcome before
the transition to the singlet potential energy surface can occur
Scheme 3
3
spectra (especially the chemical shifts of the protons on the
former furan ring) of 2 with that of the known endo-phenyl
compound 4,7 Table 1. The 1H NMR spectra of 2, in particular
chemical shifts of H3 and H4, are in good agreement with those
of endo-4.
from B. This is followed immediately by the formation of a
1
C–C bond in B leading to the oxetane product. Spin-orbit
coupling (SOC) has been recognized to control the rate of ISC.
Factors influencing the magnitude of SOC have also been
included in the Salem–Rowland rules.10 Recent work by Michl11
as well as Adam et al.12 reveals that the spatial orientation of
two singly occupied orbitals in 3B is highly important while the
through-space distance between the two radical centers plays a
subordinate role in determining the rate of ISC. Intersystem
Table 1 1H NMR chemical shifts of selected protons on compounds 2
and 4
3
1
crossing from B to B is expected to be concerted with the
formation of a new C–C bond leading to a product whose
stereochemical outcome is associated with the conformation
adopted by the biradical at this instant. ISC may also occur
with concurrent cleavage of the initially formed C–O bond to
restore the starting carbonyl compound and alkene.
According to the Salem–Rowland rules,10–12 strong SOC
occurs when the p orbitals at the spin-bearing atoms are
orthogonal to each other. The possible conformers (C) of
the triplet biradical (3B) are represented by the Newman pro-
jections C1–C3 and C1Ј–C3Ј in Scheme 4. Bulky aryl groups pre-
fer to stay away from the former furan ring so that conformers
C1–C3 are favored over C1Ј–C3Ј. Among C1–C3, C1 and C2 are
expected to be similarly populated whereas C3 is higher in
energy. ISC from C1 leads to immediate C–C bond formation
that rotates the aryl group over the former furan ring plane
resulting in the endo-Ar product. ISC from C2 leads to cleavage
of the initially formed C–O bond restoring the starting
materials. Similarly, ISC from C3 furnishes the exo-Ar product.
It was demonstrated that the relative stabilities of C1 and C3
depend on the size of the α-substituent of the aryl carbonyl
compound.7 When the α-substituent is small (such as H)
compared to the aryl ring, C3 is the conformer responsible for
δ/ppm
H1
H2
H3
H4
exo-4
endo-4
2a
6.54
6.36
6.36
6.40
6.38
6.39
6.71
6.52
6.53
6.37
6.39
6.37
5.47
4.68
4.86
4.80
4.88
4.77
3.65
4.21
4.38
4.47
4.61
4.48
2c
2d
2e
Compound 2a was isolated in 92% yield from the reaction of
1a and furan. The starting materials were consumed completely
and the only other observed product as revealed by GC and
GC/MS analyses of the photosylate is an isomer of 2a, pre-
sumably the exo-aryl compound. Reaction between 1b and
furan results in four adducts in almost equal yields based on
1772
J. Chem. Soc., Perkin Trans. 2, 1999, 1771–1778

View Article Online
Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Scheme 4
Scheme 5
product formation and the exo-Ar product is selectively pro-
duced.7 When the α-substituent is large as in the case of the
COOCH3 group in compound 1, C1 is the dominating con-
former and furnishes primarily the endo-Ar compound as
the product.7,13 If only conformers C1–C3 are responsible for
product formation, the endo-/exo-Ar selectivity in producing 2
should not alter significantly with the change of the aryl group
of 1 because the same α-substituent (COOCH3) is involved in
all these compounds, and therefore the energy differences
between C1 and C3 are expected to be similar with different
Ar groups. However, we observed significant changes in the
product stereochemistry with alteration of the aryl groups,
indicating that the other set of conformers, C1Ј–C3Ј, also plays
a role in the product formation process. Analogously to the
situation in C1–C3, C1Ј leads to the exo-Ar product, C3Ј leads to
the endo-Ar products and C2Ј restores the starting materials
after ISC. Conformer C3Ј is highly congested and probably does
not contribute significantly to the formation of the endo-Ar
product. However, C1Ј may become populated to some degree
when the steric demand of the aryl group is low. Therefore,
when the aryl group is small enough to populate conformer C1Ј
as in the case of 1a, exo-Ar product results in competition with
the endo-Ar 2. Bulky aryl groups in 1 favor only conformer C1
and this leads to exclusive formation of endo-Ar oxetanes.
D1, usually undergoes an efficient 1,5-sigmatropic hydrogen
shift to regenerate 1d, accounting for, among other things, the
inefficient reaction of 1d. Compound 5 presumably results from
cyclization of diene D2. However, attempts to trap this reaction
intermediate (D2) with maleic anhydride are unsuccessful in this
study. Irradiating 1d in the presence of an equimolar amount of
maleic anhydride in benzene solution led to rapid consumption
of 1d. The expected Diels–Alder adduct of D2 to the anhydride
could not be observed.
Normal Norrish type II γ-hydrogen abstraction from the
ester methyl group leading to 1,4-biradical BR2 was suppressed
when 1d was irradiated in benzene. The products (benzaldehyde,
21
CO, and formaldehyde) expected from the cleavage of BR2
were not observed. The rate constant for the normal Norrish
type II hydrogen abstraction in a typical alkyl phenylglyoxylate,
k2, is estimated to be ≈106 sϪ1 21
Absence of Norrish type II
.
products implies that the rate constant for hydrogen abstraction
from the o-methyl group, k1, is higher than 106 sϪ1. This con-
clusion is further supported by the transient spectroscopy
of 1d.
Nanosecond laser flash photolysis of a 0.006 M benzene
solution of 1d results in transient absorptions being displayed
(insert Fig. 1). The absorption spectrum is significantly dif-
ferent from that of the triplet absorption of a typical alkyl
phenylglyoxylate.21 It is suspected that the triplet excited state
of 1d is deactivated rapidly by o-hydrogen abstraction and
escapes detection on the nanosecond time scale, i.e. k1 is close to
Photolysis of ortho-substituted methyl phenylglyoxylates
We further studied the photolysis of ortho-substituted methyl
phenylglyoxylates (1d, 1e, 1f). Irradiation of a benzene solution
of 1d (0.02 M) resulted in the formation of benzocyclobutenol
5 as the only product with low efficiency (ca. 50% conversion
achieved after 7 days of irradiation). Compound 5 is proposed
to derive from a triplet γ-hydrogen abstraction from the o-
methyl groups, Scheme 5. Triplet biradical BR1 is produced
after hydrogen abstraction and decays to ground state isomers
of the o-xylylenol (o-xylylene = o-quinodimethane) with both
Z (D1) and E (D2) configurations at the enol carbon.14,15 Based
on earlier reports on similar compounds,16 the E isomer, D2, is
expected to react with various dienophiles,17,18 cyclize to benzo-
cyclobutenol,19 or revert to starting material.20 The Z isomer,
108–109 sϪ1. Therefore the Norrish type II process (k2 ≈ 106 sϪ1
)
cannot compete. The transient decay monitored at 440 nm
(Fig. 1, upper) is best fitted by a biexponential decay with life-
times of 25 µs and 96 µs. The quality of this fitting is indicated
by the residual plot (Fig. 1, lower) representing the difference
between the experimental data and the fitted curve. The bi-
radical intermediate BR1 is also expected to be short-lived,16,22
and may also absorb in the same region (420–440 nm) as the
transients detected in Fig. 1. This further prevents its direct
observation. Based on the similarities between the absorption
spectra as well as the lifetimes of the transients detected herein
J. Chem. Soc., Perkin Trans. 2, 1999, 1771–1778
1773

View Article Online
Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
together with CO and formaldehyde (not studied herein) are
the only products from the photolysis of 1e and are proposed
to result from the decomposition of BR4 following a Norrish
type II γ-hydrogen abstraction in the triplet excited state.21
Abstraction of a δ-hydrogen from the o-tert-butyl group by the
excited carbonyl chromophore producing BR3 which sub-
sequently cyclizes to an indanol (Scheme 6) is expected from
1e considering that the analogous reaction occurred at a rate
constant of ≥1 × 109 sϪ1 in the triplet state of o-tert-
butylbenzophenone 7, (Scheme 7).23 The total absence of
Fig. 1 Transient decays monitored at 440 nm when a 0.006 M benzene
solution of 1d was flashed (upper). The insert is the transient absorp-
tion spectra at different delay times after the laser pulse. The residual
plot (lower) reveals the quality of the biexponential fitting of the decay
trace.
(Fig. 1) and those of other o-xylylenols reported in the litera-
ture,16,22 we assign the longer lived transient (96 µs) to D2
and the short lived species (τ = 25 µs) to D1 due to the allowed
1,5-hydrogen shift causing reformation of the ketone.
When a benzene solution (0.02 M) of 1e was irradiated,
the starting material was rapidly consumed and 2,4,6-tri(tert-
butyl)benzaldehyde (6) was isolated, Scheme 6. Aldehyde 6,
Scheme 7
δ-hydrogen abstraction products from 1e is initially
surprising. Laser flash photolysis (vide infra) further indicates
that the lack of indanol product from 1e is due to the absence
of a significant δ-hydrogen abstraction process (rate constant
smaller than 106
s
Ϫ1). Such a low rate constant for the
δ-hydrogen abstraction was also observed in 2,2Ј,4,4Ј,6,6Ј-
hexa(tert-butyl)benzil (8) though it was not satisfactorily
accounted for.24 It is also noted that the δ-hydrogen abstraction
process in 2,4,6-tri(tert-butyl)acetophenone (11) resulted in a
cyclization product from the resulting 1,5-biradical. The rate of
this hydrogen abstraction process was supposedly low due to
the observed inefficient reaction.25
On the other hand, δ-hydrogen abstraction from ortho-
substituents is not as atypical for α-keto esters as it is for α-
diketones. The slower δ-hydrogen abstraction in the case of 1e
cannot be, even in part, attributed to the influence of the α-keto
group on the excited carbonyl chromophore as it was in the case
of 8.24 Pappas et al. have discovered that δ-hydrogen abstraction
dominates the photoreaction of benzylglyoxylate 12 furnishing
13 as the only product, Scheme 8,26 indicating that the Norrish
type II γ-hydrogen abstraction reaction in 12 could not compete
with the δ-hydrogen abstraction. It is also noted that δ-
hydrogen abstraction in 12 proceeded at a rate comparable with
that of the phenyl ketone 10,19 probably indicating that
replacing an ester functionality (COOCH3) with a phenyl group
does not significantly alter the δ-hydrogen abstraction rate from
the ortho-groups. Therefore, the difference in the rate constants
of the δ-hydrogen abstraction between glyoxylate 1e (and to a
lesser degree, diketone 8) and ketone 7 is probably due to the
steric hinderance introduced by three tert-butyl groups on the
phenyl ring of the former.27
To exclude the possibility that the absence of an indanol
product from 1e is due to reformation of the ketone from BR3
following a δ-hydrogen abstraction, 1e was subjected to laser
flash photolysis. Transient absorptions attributable to the
Scheme 6
1774
J. Chem. Soc., Perkin Trans. 2, 1999, 1771–1778