A novel photorearrangement of 1,3-diketonatoborinate: A photoinduced intramolecular alkylation on a conjugated carbonyl system

Article abstract of DOI:10.1139/v99-139

Irradiation of borinates derived from 9-borabicyclo[3,3,1]nonane and 1,3-diketones at their π-π* transition band caused a rearrangement from their triplet excited state to give boronate intermediates that could be hydrolyzed to the corresponding aldols. T

Full text of DOI:10.1139/v99-139

1374
A novel photorearrangement of
1,3-diketonatoborinate: a photoinduced
intramolecular alkylation on a conjugated
carbonyl system
Jun-ichi Ishiyama and Yuan L. Chow
Abstract: Irradiation of borinates derived from 9-borabicyclo[3,3,1]nonane and 1,3-diketones at their π–π* transition
band caused a rearrangement from their triplet excited state to give boronate intermediates that could be hydrolyzed to
the corresponding aldols. The primary photolysis involves the scission of one of the B—C bonds followed by a 1,3- or
1,5-migration and the formation of a more stable B—O bond by radical pathways. The triplet state reaction was
established by quenching and heavy atom effect experiments. The overall reaction pattern is an irreversible 1,2- or 1,4-
addition of an alkyl-boron group to an α,β-conjugated ketone system by radical processes. Evidence showed that the
non-isolated boronates also underwent secondary photoreactions followed by hydrolysis to afford enones corresponding
to dehydration products of aldols; mechanisms are suggested for these secondary steps. The photoreaction was more
facile in nonpolar solvents than in polar solvents; different solvents also gave different product compositions. In the
presence of dilute acetic acid, the photolysis gave a high yield of aldols, suggesting a rapid acetolysis of the B—C
bond in boronates to acetylborates. Under oxygen, the photolysis was complicated by radical oxidation to give totally
different products.
Key words: borinate photochemistry, charge transfer absorption, alkylborane addition, boronate intermediates, B—C
bond homolytic scission.
Résumé : L’irradiation de borinates dérivés du 9-borabicyclo[3.3.1]nonane et des 1,3-dicétones au niveau de leur bande
de transition π–π* provoque un réarrangement à partir de leur état excité triplet pour conduire à des boronates
intermédiaires que l’on peut hydrolyser en aldols correspondants. La photolyse primaire implique la scission de l’une
des liaisons B—C, suivie d’une migration 1,3 ou 1,5 et la formation d’une liaison B—O plus stable par le biais de
voies radicalaires. On a établi l’implication d’une réaction de l’état triplet à l’aide d’expériences de désactivation et
d’effets isotopiques. Le schéma de la réaction globale est une addition, 1,2 ou 1,4 irréversible, d’un groupe alkyl-bore
sur un système cétonique α,β-insaturé par des processus radicalaires. Les données montrent que les boronates qui n’ont
pas été isolés subissent aussi des photoréactions secondaires qui, suivies d’une hydrolyse, fournissent les énones qui
correspondent aux produits de déshydratation des aldols; on suggère des mécanismes pour ces étapes secondaires. La
photoréaction est plus facile dans les solvants non polaires; les divers solvants donnent aussi des compositions de
produits différentes. En présence d’acide acétique dilué, la photolyse fournit un rendement élevé d’aldols; ceci suggère
que, dans les boronates, il se produit une acétolyse rapide de la liaison B—C pour donner des acétylboronates. En
présence d’oxygène, la photolyse est compliquée par une oxydation radicalaire qui conduit à des produits totalement
différents.
Mots clés : photochimie du borinate, absorption du transfert de charge, addition d’alkylborane, intermédiaires borona-
tes, scission homolytique de la liaison B—C.
[Traduit par la Rédaction] Ishiyama and Chow 1383
Introduction
two short communications on the photoreaction of allied
borinates (where the bicyclononane ring in 5–7 is replaced
by di-n-butyl or dicyclohexyl groups) had already been pub-
lished in 1986 (1). These works were a study based on sug-
gestions derived from an earlier communication by another
group (2). While an additional related communication was
We observed some 10 years ago that dialkylborinates de-
rived from simple 1,3-diketones (such as 5–7 in Scheme 1,
see below) underwent photolysis to give rearranged prod-
ucts. The work has been left dormant since we noted that
Received February 17, 1999.
J. Ishiyama1 and Y.L. Chow.2 Department of Chemistry, Simon Fraser University, Burnaby, BC V5A 1S6, Canada.
1Visiting Professor from Fukushima National College of Technology, Taira, Fukushima, Japan; Overseas Research Fellow (1997)
supported by the Ministry of Education, Science, Sports and Culture, Government of Japan.
2Author to whom correspondence may be addressed. Telephone: (604) 291-3352. Fax: (604) 291-3765. e-mail: ychow@sfu.ca
Can. J. Chem. 77: 1374–1383 (1999)
© 1999 NRC Canada

Ishiyama and Chow
1375
Scheme 1.
published later (3), the detailed photochemical reaction of
allied borinates has not, to our knowledge, been described.
In the meantime, the photochemistry and photophysics of
boron compounds continue to attract investigations, and in-
teresting results have been published in relation to
photoinduced electron transfer (4), photo-cycloadditions (5),
exciplex formations (5), and unusual emission patterns
(6, 7). We were attracted by the abnormally deep-colored
appearance of these borinates, and were puzzled by what ap-
peared to be an unusually strong CT band in their UV–visi-
ble spectra. More recently, we undertook an in-depth
investigation of borinates 5–7, and discovered that our re-
sults were not in total agreement with the earlier communi-
cations (1). The chemical and physical behavior of these
borinates is interesting and promises to stimulate new stud-
ies of allied boron compounds, particularly in relation to
their excited states and radical reactions. We wish to de-
scribe the results from our group.
B-diketonato-9-borabicyclo[3,3,1]nonane, 5–7, in high
yields (8); these borinates were abbreviated as AABBN (for
B-acetylacetonato-, 6) (8b), BABBN (for B-benzoyl-
acetonato-, 7) and DBMBBN (for B-dibenzoylmethanato-,
5). They were prepared equally easily from 9-methoxy-BBN
(1, H = OCH3) and 9-(2-aminoethanolato)-BBN (1, H =
NH2CH2CH2O) in comparable yields. These colored (from
yellow to orange) crystalline dialkylborinates 5–7 promi-
nently showed strong UV absorption that is slightly blue-
shifted from that of the parent diketone (6); this UV band
was red-shifted in more polar solvents and must arise from a
π–π* transition. They also showed a broad and weaker but
well-defined band that was assigned as a CT transition
(Fig. 1). Since the intensity of these absorption bands
obeyed the Beer–Lambert law in cyclohexane and methylene
chloride, it was concluded that these borinates existed as
monomers up to 0.1 M. The borinates showed a strong IR
absorption band for the B—O bond at 1376–1378 cm–1 and
a
13C NMR signal for the C-B carbon at about 23 ppm,
which was slightly broadened owing to the boron
quadrupole couplings. It should be noted that whereas the
chemical shift of olefinic hydrogen moved downfield in the
order 6–7–5, the 13C chemical shift of the corresponding
Results
The reaction of 9-borabicyclo[3,3,1]nonane (BBN, 1) with
a 1,3-diketone (2–4) in THF readily gave the corresponding
© 1999 NRC Canada

1376
Can. J. Chem. Vol. 77, 1999
Fig. 1. The absorption spectra of (a) 6 at 1.05 × 10–4 M, (b) 7 at 3.69 × 10–5 M, and (c) 5 at 3.48 × 10–5 M in cyclohexane.
Table 1. Relative peak % of the photolysis of 6 from GC analysis.
RT(min)a
1.23
1.9
1.82
3.89
4.25
4.87
1.4
5.23
6.09
8.73
24.0
25.0
13.44
4.3
300 nm
254 nme
Products
13.0
11.0
15
1.8
0.6
14a
3.3
14.0
13
5.8
1.7
14
41.0
31.0
12
7.7
1.8
3.8
b
c
d
6
aThere are more peaks in the vicinity of solvent that were identified as acetylacetone, cyclooctane, and bicyclo[3,3,0]octane
by the coinjection of an authentic sample or by its mass spectral pattern.
bCyclooctanone by coinjection.
c1,1′-Bicyclooctyl by the MS pattern.
dUnknown.
eThere are more peaks of small areas over the whole range in this than in the photolysis with the 300 nm source.
olefinic carbon moved upfield in the same order. The re-
versed order of chemical shifts could be related to the ring
current generated by resonance involving the borinate ring.
formed from the retro-aldol reaction of 8 on injection into
GC. Similar photolysis of 5 in a solution containing 0.2% of
acetic acid (by volume) afforded aldol 8, in a 41% isolated
yield, but no enones 9 and 10; the yields of the latter two
were small as shown by GC analysis. However, in the pres-
ence of methanol (<1%) or acetic anhydride, analogous
photolysis caused marginal effects. On the other hand, simi-
lar photolysis of 5 in a mixture of methylene chloride and
acetonitrile (1:4) gave only enones 9 and 10 in 38 and 36%
yield, respectively. While enone 9 could be isolated by chro-
matography, enone 10 decomposed during the operation.
In the initial stage, AABBN 6 was photolyzed in a mix-
ture of cyclohexane and methylene chloride (9:1) under a
blanket of dry nitrogen with 300 nm light in order to obtain
a higher yield of aldol 12 at a reasonable rate. Irradiation
with 254 nm light was faster by a factor of two than that
with 300 nm light, but irradiation with 350 nm light (i.e., ex-
citation of the CT band), on the contrary, caused no signifi-
cant decomposition in 24 h. Photolysates were
chromatographed to give aldol 12 and enones 13 and 14 in
yields of 54, 7 and 1%, respectively, by 254 nm irradiation,
and 54, <1, and 14%, respectively, by 300 nm irradiation.
The GC spectrum of the crude product showed these and
other product peaks, including those of acetylcyclooctane
(15) and the geometric isomer 14a; both of them were too
small to be isolated (see Table 1). As is typical of a radical
reaction, there were also several small by-product peaks that
DBMBBN 5 gave a broad (380–500 nm) fluorescence
band with λmax 415 nm in methylene chloride on excitation
at the π–π* transition at 330 nm, but no emission on excita-
tion at the CT band. AABBN 6 in cyclohexane (and methy-
lene chloride) showed a narrow fluorescence peak at 280 nm
on excitation at 259 nm, and at 393 on excitation of the CT
band at 353 nm. In both spectra weak emission humps at
longer wavelengths (e.g., those at 320 and 370 nm in the
former and at 432 and 507 nm in the latter) were assumed to
arise from decomposed intermediates. While these fluores-
cence spectra were stable in cyclohexane, those in
acetonitrile showed complex shapes and changed with time
lapses. Their study remains to be continued later.
Irradiation of DBMBBN 5 in methylene chloride at
350 nm gave unconjugated enone 9 and a mixture of iso-
meric enones 10 in addition to a smaller amount of aldol 8;
the amount of all enones increased at the expense of 8 dur-
ing isolation. Injection of this sample into GC showed that 9
and 10 were present in 14 and 30%, respectively, but 8 was
thermally decomposed to give a new peak showing M+ =
217 by a CI mode. By the EI mode, this GC peak showed
the major m/e fragment at 216 (M+), 133 (C6H5COCH2CH2),
105 (C6H5CO, 100%), and 77 (C6H5); this MS pattern is
compatible with the structure of 11,which could be easily
© 1999 NRC Canada

Ishiyama and Chow
1377
a
Table 2. Relative product % from photolysis of 6.
of 11 and 15 in a 27%/15% ratio in the absence, and in a
56%/20% ratio in the presence, of acetic acid. Similar
photoreaction at 350 nm irradiation either in the presence or
absence of acetic acid recorded the ketone peaks in a ratio of
12:10 by GC analysis. As the ratio of 11:15 virtually repre-
sents the ratio of 16:17 in the crude product, the latter must
be coalescing to the 1:1 ratio.
RT (min)
3.66b
6.43
6.97
8.20
9.38
Product
Tube
15
2
14a
2
13
3
14
4
12
2
28
5
3
5
Tube 2c
Tube 3d
Tube 4d
Tube 7e
Tube 8e
Tube 9e
9
4
2
2
2
2
2
3
4
12
11
9
8
10
57
37
22
38
51
22
24
20
18
13
AABBN 6 was used as a model to study the excited state
reaction of these borinates. Photolysis of 6 with 300 nm
light in various solvents showed the decreasing efficiency
order of cyclohexane >> THF ~ benzene > acetonitrile >
methylene chloride in the ratio of 1.00:0.74:0.71:0.60:0.49.
When the photolysis was carried out in the presence of a
trace amount of water (i.e., using purging nitrogen directly
from a cylinder without drying), it was noteworthy that this
ratio was nearly leveled, although cyclohexane and benzene
were still the better solvents. GC examinations showed that
the major product was aldol 12 for the reaction in methylene
chloride and enone 14 for those in cyclohexane and benzene.
Sensitized photodecomposition of 6 with benzophenone by
350 nm irradiation was an ideal setup since the sensitizer ab-
sorbed strongly in the 350 nm region where 6 showed no
photoreactivity; this irradiation caused much more efficient
decomposition than that by direct irradiation at 300 nm, and
resulted in enormously accelerated disappearance of 6 as
well as the efficient formation of enones 13 and 14a in addi-
tion to a few unknown GC peaks. The reaction was compli-
cated by the sensitizer itself being consumed, but the
occurrence of triplet energy transfer was indicated by the ab-
sence of benzopinacol in the crude product; this implied that
triplet excited benzophenone had been quenched (9). While
the interpretation was not unambiguous, this observation,
coupled with the efficient product yields, supported that effi-
cient sensitization must have occurred. Under the oxygen
blanket, the photolysis of 6 was not totally quenched but in-
stead slightly accelerated, and gave oxidation products obvi-
ously due to the participation of O2 in the product formation
as is often observed in free radical reactions (10). While the
result was inconclusive, in view of the smallness in the rate
increase, oxygen quenching had probably occurred but was
masked by the enhanced radical reactions.
The photoreaction of 6 was quenched efficiently by 1,3-
pentadiene at the 3.5 mM level, and also by vinyl ethyl ether
and furan (at 5 mM). The former diene is a typical triplet
quencher, but the latter ethers do quenching by some other
mechanisms, most likely intercepting the singlet state by
electron transfer (9). No new compound was formed from
these quenching experiments. Further, the consumption of 6
was accelerated by more than an order of magnitude in the
presence of bromobenzene (4 mM) and 1,2-dibromoethane
(5 mM). In both cases, both enones 13 and 14, particularly
the former unconjugated one, were produced in significantly
increased amounts with increasing bromide concentrations,
while the yield of aldol 12 was affected little, as shown in
Table 2. The fluorescence peak of AABBN 6 in cyclohexane
at 280 nm by excitation at λex 259 nm was quenched by the
presence of increasing concentrations of 1,2-dibromoethane.
The Stern–Volmer plot was a straight line with the slope
(Ksv) of 20.0. This demonstrated that singlet excited 6 inter-
acted with the bromide and mutated rapidly to the triplet
state by heavy atom effects (11). Similar spectroscopic ex-
6
8
aSee text for details of the solutions; the irradiation time was 4 h for
tubes 1–4 and 2.5 h for tubes 7 and 8; the corresponding factor for
products to the internal standard (hexadecane) was arbitrarily set at 10:1.
bThis peak is derived from 12 upon injection of photolysates.
cTube 3 contains 0.167 M acetic acid.
dTubes 3 and 4 contain phenyl bromide, 0.0082 and 0.0041 M,
respectively.
eTubes 7–9 contain 1,2-dibromoethane, 0.02, 0.04, and 0.08 M,
respectively.
were more visible in the 254 nm irradiation. GC–MS and
(or) the coinjection of authentic samples identified these mi-
nor GC peaks as those of cyclooctane, cyclooctene, 1,1′-
bicyclooctyl, and cyclooctanone. In the presence of <1%
acetic acid, a similar 300 nm photolysis of AABBN 6 gave
predominantly aldol 12 and only trace amounts of enones
13–14, though the rate of AABBN conversion was scarcely
affected (Table 2); indeed 12 was prepared in good yield by
this procedure. Aldol 12 was partially decomposed on GC
injection to give the peaks of itself and 15 in a 6:4 ratio;
GC–MS by the CI mode showed two peaks to have M+ at
213 and 155. The latter GC peak exhibited the m/e at 154
(M+), 139, 111, 69 (100%), and 43, which supported the
structure of 15. Aldol 12 also slowly decomposed to 15 ei-
ther in solution or neat; decomposition was more rapid on
heating, during which enones 13 and 14 were also formed in
small amounts as shown by GC and NMR monitors. In di-
lute acidic (HCl) methanol, 12 decomposed slowly to 14 and
15 at room temperature, but rapidly and mainly to enone 13
on refluxing. The enone distribution suggested that
unconjugated enone 13 is thermally more stable than the
conjugated 14, probably because of steric crowding.
BABBN 7 in methylene chloride was irradiated with
300 nm light to give low yields of four enones that could be
derived from aldols 16 and 17; these aldols continued to de-
compose during separation. With difficulty, enones 20–21
were finally isolated only in semi-pure states by preparative
TLC. The same solution containing 1% of acetic acid was ir-
radiated under similar conditions to give aldol 16 (46%) and
a small amount of a mixture containing the alternative aldol
17. Aldol 16 when injected into the GC gave only one peak,
which showed the MS pattern corresponding to 11 both by
EI and CI modes. The fact that both aldols 8 and 16 decom-
posed to give the common ketone 11 provided a rationale for
the structural correlation among them. The GC injection of
the impure fraction of 17 under the same conditions gave the
major peak corresponding to 15 as shown by the RT and MS
pattern. A crude product obtained from terminating a 60%
photoreaction of 7 at 300 nm irradiation was subjected to
GC analysis conditions as above to record the ketone peaks
© 1999 NRC Canada

1378
Can. J. Chem. Vol. 77, 1999
Scheme 2.
periments with bromobenzene and 1,3-pentadiene could not
be carried out owing to the overlap of the absorption in the
260 nm region.
21 exhibited (i) an olefinic proton singlet at 6.05, 6.05, and
6.72 ppm, (ii) a set of pertinent 13C signals for an α,β-unsat-
urated carbonyl system, and (iii) five 13C signals for the
cyclooctyl group due to its rapid rotation. In addition, the
NMR spectra of these enones also showed the required sig-
nals with appropriate chemical shifts.
The structure of the products was primarily determined
from NMR spectral evidence, in some cases with partially
pure samples. The primary photoproducts were obviously
aldols 8, 12, 16, and 17, which served as the source for the
observed products, enones and acylcyclooctanes, derived
from the secondary reactions as shown in Scheme 1. The
isolated aldols showed, among other required signals, the
following key NMR signals to define the structure unambig-
uously. Firstly, they showed a pair of AB quartets due to the
diastereotopic proton of -CO-CH2-, with the chemical shifts
progressively moved downfield as a CH3 group was re-
Discussion
The overall reaction resembles the well-known classic
1,2- or 1,4-addition of an alkylborane onto α,β-unsaturated
carbonyl compounds (12), except that the present reaction is
promoted by light and occurs intramolecularly through free
radical mechanisms. Thus, electronic excitation provides an
alternative pathway to promote this alkylborane addition.
The failure for excitation in the CT band to promote the re-
action must imply that this excited state is rapidly deacti-
vated to the ground state, i.e., that is the only channel
available. These borinates 5–7 undergo a deep-seated rear-
rangement upon excitation of the π–π* transition band in a
wide range of solvents; the photolytic efficiency is better in
nonpolar than in polar media and the product patterns vary
widely depending on reaction conditions. The overall result
indicates that the products are formed in consecutive
multistep processes involving intermediates that are sensi-
tive to the photolysis conditions. The overall reaction is ap-
proximated as the primary photoinduced rearrangement to
give the precursors of aldols 8, 12, 16, and 17, followed by
secondary decomposition of the precursors under the
photolysis conditions.
placed by a C6H5 group. Secondly, they had the carbonyl 13
C
signal at 211 ppm for the acetyl and at 202 ppm for the
benzoyl group. Finally, the monosubstituted cyclooctyl
group showed eight 13C signals indicating that the eight car-
bons are nonequivalent, that is, in each pair the correspond-
ing -CH2- groups were diastereotopic to one another due to
the vicinal chiral centers. The unconjugated enones 9, 13,
and 20 showed (i) a two-proton singlet in the 3.0–4.0 ppm
region and a carbonyl 13C signal at 199–208 ppm for the
CH2-CO, and (ii) eight 13C signals (including one olefinic at
about 137 ppm) for the cyclooctylidene group. Among the
conjugated enones, only 14, 14a, and 21 were isolated, the
last two as mixtures. Both 14 and 21 were always formed as
the major isomer and assigned to the configuration shown
for steric reasons. Both enones 10 and 19 (and (or) their
isomers) could be detected with GC–MS but could not be
isolated because of decomposition during column chromato-
graphy. The isomeric enone pair 14 and 14a showed nearly
identical spectral patterns not only in MS but also in the 13C
The experimental probes described provide good qualita-
tive evidence that the primary photoreaction is derived from
the triplet excited state of these borinates 5–7. Two of these
probes are unambiguous; that is, efficient triplet quenching
(9) by 1,3-pentadiene, and a heavy atom effect to enhance
1
and H NMR spectra, which gave little evidence to assign
the cis–trans isomerism. These two conjugated enones and
© 1999 NRC Canada

Ishiyama and Chow
1379
Scheme 3.
the population of the triplet state induced by bromine-
containing compounds (11). The other two, the benzophe-
none sensitization and oxygen quenching, are complicated
by the participation of the reagents in the reactions; the for-
mer, which cannot be analyzed quantitatively owing to the
change in the product pattern, also supports triplet state en-
ergy transfer on the basis of the detailed observations.
In the previous communications (1), the authors stated
that the photoreaction of some similar dialkylborinates could
not be quenched by either oxygen or 1,3-pentadiene, which
led them to suggest a singlet state reaction. Though the pres-
ent dialkylborinates 5–7 are not exactly the same as theirs,
the corresponding results for these two quenchers are just
the opposite to, and more complicated than, those stated in
their reports. It is interesting to compare details of the two
experimental descriptions.
The reaction mechanism is proposed in Scheme 2, in
which the carbonyl triplet excited state 22, as represented by
the diradical model (9), is the species initiating the radical
pathway. The step 22 to 23 is analogous to the homolytic
substitution of a borane by an alkoxy radical, the driving
force of which is derived from a bond energy gain of about
30–40 kcal/mol in the conversion of a B—C bond to a B—O
bond (13). Triplet excited state biacetyl has been shown to
interact with borane either directly or through the acetyl rad-
ical (14). The intramolecular attack in 22–23 should be more
facile than any intermolecular attacks. The subsequent re-
combination of the radical pair 23 can occur at the carbon of
either O—C bond; the direction is controlled by the nature
of the R groups to give boronate 24. The migration can be
highly regiospecific if the R and R′ groups possess a large
electronic gradient (1). Thus, the primary photolysis is as-
sumed to terminate on the formation of 24. Its simple hydro-
lysis should give aldols 25, the process of which appears to
be relatively sluggish under the photolysis conditions; thus
GC analysis at times shows poor reproducibility depending
upon sampling times. As will be discussed below, the con-
version of 24–25 can also involve an excitation step. How-
ever, the work-up does not require highly nucleophilic
hydrogen peroxide anion (1) to hydrolyze borate esters.
While the efficiency of the borinate photodecomposition
does not correlate strictly with solvent polarity, its rate is
generally faster in hydrocarbons than in polar solvents. It is
noteworthy that water, even in a trace amount, can retard the
efficiency; the retardation is more significant in cyclohex-
ane. The mechanism may be related to the coordination pro-
cess at the boron center and remains to be elucidated. Efforts
to isolate boronate 24 for an intermediate identification un-
der anhydrous conditions have not been successful in pro-
viding a workable sample.
The mechanistic studies have been encumbered by unpre-
dictable variations of the product pattern on changing the
photolysis conditions; their effects have been traced to the
secondary reactions and the reactivity of accompanying in-
termediates. The probable concurrence of thermal as well as
photolytic pathways for intermediates adds further complex-
ity that poses a formidable problem in portraying the mecha-
nism(s). Nevertheless, certain observations can be explained
with obvious and straightforward rationales. First of all,
about 1% acetic acid does not affect the rate of photolysis,
as shown in Table 2, but may rapidly attack boronate 24 by a
pre-rate-determining nucleophilic coordination (15, 16) to
form acetyl-borate 26, which should be more easily hydro-
lyzed than 24 to afford 25. Unless absorbing excitation light,
both 25 and 26 are not photolabile and can afford good
yields of aldols such as is the case for 8 and 12, which are
obtained clean even after prolonged irradiation under the
given conditions. In contrast, the photolysis of borinate 7
generates borate 26 with R = C6H5 and R′ = CH3 (corre-
sponding to aldol 17), which possesses the α-boratoxystyryl
chromophore and must be labile to 300 nm irradiation, but
not to 350 nm irradiation. Indeed, two observations support
this argument; they are the formation of much less 17 than
16 (as measured by the GC peak of 15 and 11) under the for-
mer (300 nm) irradiation and that of as much 17 as 16 under
the latter (350 nm) irradiation. The high yield of aldol 8 in
the photolysis of 5 under 350 nm irradiation must have a
similar cause. Likewise, if aldols 25 with R = C6H5 are
formed by adventitious water, they will be sensitive to
300 nm light and may cause a further complication to prod-
uct patterns. To avoid these uncertainties, photolysis should
be carried out under dry conditions. Further, there is no
doubt that ketones 11 and 15 are formed by thermal and (or)
acid-catalyzed retro-reactions from the aldols, slowly during
isolation and more rapidly on injection in the GC port; when
the port is cleaned the decomposition of 12 is drastically re-
duced.
Explanations of other observations are less straightfor-
ward and must rely on indirect rationales. In view of the
sluggish acid-catalyzed dehydration of 12 at a room temper-
ature (16), enone formation through simple thermal reactions
cannot make a substantial contribution to high yields of
enones such as those shown in Table 1. On the basis of the
following evidence, we propose, as the cause of the enone
formation, that boronate 24 undergoes photo-rearrangement
followed by hydrolysis (Scheme 3). Firstly, the
photoreaction of borinates 5–7 affords lower yields of aldols
in the absence of acetic acid than in the presence, suggesting
that exposure of 24 to light causes photoreactions to give
enones (e.g., see Table 2). The borinate photoreactions car-
© 1999 NRC Canada

1380
Can. J. Chem. Vol. 77, 1999
ried to a high conversion always furnish more enones. Sec-
ondly, the photolysis of 7 with 300 nm irradiation always
gives much less 17 than 16, particularly in the absence of
acetic acid. This is a good indication that the intermediate 24
(R = C6H5, R′ = CH3), corresponding to 17, having an α-
boronatostyryl chromophore, can absorb more light energy
from 300 nm irradiation and undergo photolysis more exten-
sively. The corroborating evidence is the formation of nearly
the same amount of 16 and 17 in 350 nm irradiation, in
which both corresponding intermediates 24 are scarcely irra-
diated. Thirdly, the dominance of either 13 or 14, depending
on the wavelength of excitation (see Table 1), supports the
participation of an excited state of 24, which is the interme-
diate in the absence of acetic acid. Finally, in the presence
of benzophenone as a sensitizer or a bromide as a triplet
enhancer, enones 13 and 14 are disproportionately increased
in comparison to aldol. We speculate that the excited state of
either promoter interacts with intermediates, through
exciplexes or otherwise, which may increase either conjugated
14 (e.g., in the presence of bromobenzene) or unconjugated
enone 13 (e.g., in the presence of 1,2-dibromoethane) 12, as
shown in Table 2.
× 0.2 mm BB-1 or OV-1 capillary column and HP 3390A
integrator.
Irradiation was carried out in a Rayonet Photochemical
Reactor in quartz (for 254 and 300 nm irradiation) or Pyrex
(for 350 nm irradiation) test tubes (12 × 100 mm) placed in
a “merry-go-round.” A bottom fan circulated the air to main-
tain a constant temperature of 30–34°C. For analytical and
sensitized photolysis, solutions were added with a known
concentration of hexadecane or eicosane as the internal stan-
dard before irradiation. The corresponding factor of AABBN
to hexadecane was determined for each column to calculate
percent conversions.
Preparation of borinates 5–7
To a solution of a 1,3-diketone in THF, a calculated
amount of 9-BBN was added by syringe under a nitrogen
blanket at room temperature while the solution was stirred
magnetically. The precipitate in nearly a quantitative yield
was filtered and recrystallized under nitrogen purging. 9-
Methoxy-9-borabicyclo[3,3,1]nonane (1.0 M solution in
hexane) or 9-(2-aminoethyleneoxy)-9-BBN was also used to
give nearly quantitative yields.
While 24 is not available for study at this stage, the bond
reorganization as shown in Scheme 3 is suggested to account
for the photochemical transformations to give the enone pre-
cursors 27 and 28. The former, 27, can be arrived at by a
retro-Diels–Alder reaction; the latter, 28, by a 1,3-hydrogen
shift from 24. The alkyl B=O bond may be regarded as an
analogue of an acyl carbonium ion, and is expected to react
with water vigorously to give the observed enones. The pro-
posed boronate intermediates in Scheme 3 involve much un-
explored chemistry of their ground and excited states, which
promises to open up a new area of investigation. In this re-
gard, the origin of the broad CT absorption band must be re-
lated to the electronic structure of borinates 5–7 and will be
studied.
DBMBBN (5)
The precipitate was recrystallized in CH2Cl2 and hexane
(2:1) or THF and hexane (2:1) and cooled to –20ºC to give
deep red crystals, mp 266–267ºC; IR (Nujol): B–O stretch-
ing at 1376.5 cm–1; H NMR (CDCl3) δ: 1.00 (2H, br s),
1
1.55–1.65 (2H, m), 1.70–1.80 (4H, m), 1.90–2.10 (6H, m),
6.90 (1H, s), 7.50 (4H, t), 7.60 (2H, t), 8.10 (4H, d); 13C
NMR (CDCl3) δ: 23.0 (br), 24.91, 31.41, 92.67, 128.24,
128.86, 133.73, 134.14, 183.11; MS (EI) m/e (relative inten-
sity): 344 (16, M+), 287 (18), 261 (28), 248 (43), 235 (17),
105 (100). Anal. calcd. for C23H25BO2: C 80.23, H 7.27;
found: C 80.32, H 7.35. This compound is soluble in THF
and methylene chloride but much less soluble in benzene
and acetonitrile.
Experimental section
BABBN (7)
The precipitate was recrystallized in THF and diethyl
ether (1:1) by cooling to –20°C to give brilliant red crystals,
mp 196–197ºC; IR (Nujol): B–O stretching at 1377.7 cm–1;
1H NMR (CDCl3) δ: 0.90 (2H, br s), 1.50–1.60 (2H, m),
1.60–1.70 (4H, m), 1.80–2.00 (6H, m), 2.25 (3H, s), 6.30
(1H, s), 7.45 (2H, t), 7.60 (1H, t), 7.95 (2H, d); 13C NMR
(CDCl3) δ: 23.0 (br), 24.79, 24.83, 24.98, 31.14, 31.36,
96.42, 128.15, 128.78, 133.53, 133.63, 181.79, 191.84; MS
(EI) m/e (relative intensity): 282 (22, M+), 239 (13), 225
(35), 199 (55), 186 (100), 173 (40). Anal. calcd. for
C18H23BO2: C 76.61, H 8.21; found: C 76.75, H 8.45.
Materials and instrumentation
Commercial 1,3-diketones (Aldrich) were recrystallized or
distilled, but 9-borabicyclo[3.3.1]nonane (9-BBN, 0.5 M so-
lution in THF) was used as received. For preparative
photolysis, reagent grade solvents were distilled with proper
dehydrating agents. Solvents of spectroscopic grade (BDH)
were used for optical spectra. Benzophenone, 1,3-
pentadiene, bromobenzene, and 1,2-dibromoethane were pu-
rified as described in earlier publications (5, 10, 11). The
sample of 1,3-pentadiene was a mixture of trans/cis 55:45
by GC.
AABBN (6)
Melting points were determined on a Fisher–Johns block
and not corrected. Infrared spectra were recorded with a
Perkin–Elmer model FT1605 spectrometer, ultraviolet and
visible spectra with a Varian–Cary 210 spectrophotometer,
fluorescence spectra with either a PTI LS-1 or Perkin–Elmer
MPF-44B, and GC–MS with a Hewlett–Packard 5985B.
NMR spectra were taken with Bruker AMX-400 or AMX-
600 using TMS as the internal standard. Gas chromatogra-
phy was carried out at the injection port temperature of 150–
250º C with a Hewlett–Packard 5790 equipped with a 15 m
The precipitate was recrystallized in degassed hexane to
give bright yellow crystals, mp 121–122º C; IR(Nujol): B–O
stretching at 1376.6 cm–1; H NMR (CDCl3) δ: 0.80 (2H,
1
br s), 1.40–1.50 (2H, m), 1.58–1.66 (4H, m), 1.70–1.90
(6H, m), 2.10 (6H, s), 5.55 (1H, s); 13C NMR (CDCl3) δ:
23.0 (br), 24.29, 24.73, 31.07, 100.63, 190.44; MS (EI) m/e
(relative intensity): 220 (25, M+), 177 (12), 163 (50), 137
(80), 124 (100), 111 (31). Anal. calcd. for C13H21BO2: C
70.91, H 9.55; found: C 70.68, H 9.45. 6 is soluble in cyclo-
© 1999 NRC Canada

Ishiyama and Chow
1381
hexane and benzene but less soluble in methylene chloride
and acetonitrile.
tained 21 (RT 9.7) as the major component; 1H NMR
(CDCl3) δ: 1.50–1.80 (14H, m), 2.18 (3H, s), 2.35 (1H, tt),
6.71 (1H, br s), 7.45 (2H, m), 7.51 (1H, m), 7.90 (2H, m);
13C NMR (CDCl3) δ: 18.09, 26.08, 26.43, 26.70, 31.78,
41.19, 118.73, 123.38, 128.14, 139.09, 139.76, 166.44,
192.22; MS (EI) m/e: 256 (5, M+), 241 (6), 171 (12), 145
(71), 105 (100), 77 (41); MS (CI) m/e: 257 (M++1). Fraction
Preparative photolysis of borinates
DBMBBN
A solution of 5 (0.10 g, 0.29 mmol) in anhydrous THF
(50 mL) was distributed in 12 Pyrex tubes, sealed with
septa, and purged with dry nitrogen for 5 min. The tubes
were rotated in a merry-go-round under irradiation with
RPR 3500 A lamps (24 W × 16) for 24 h. The photolysate
gave two major GC peaks, at RT 3.60 and 4.20. The crude
product was chromatographed on a short column of silica
gel 60 (20 cm3) to afford an impure fraction of the former
peak that was rechromatographed to give 9; 1H NMR
(CDCl3) δ: 1.50–1.80 (10H, m), 2.11 (2H, br t), 2.30 (2H, br
t), 4.10 (2H, s), 7.20–7.60 (8H, m), 7.81 (2H, m); MS (EI)
m/e: 318 (5, M+), 213 (11), 105 (100), 91 (15), 77 (25); MS
(CI) m/e: 319 (100, M++1). Anal. calcd. for C23H26O: C
86.75, H 8.23; found: C 86.95, H 8.64.
1
three contained 18 (RT 7.6); H NMR (CDCl3) δ: 1.50–1.70
(10H, m), 2.05 (3H, s), 2.10 (2H, br t), 2.35 (2H, br t), 3.45
(2H, s), 7.05 (2H, m), 7.20 (3H, m); MS (EI) m/e: 256 (60,
M+), 241 (100), 238 (26), 213 (45), 198 (25), 143 (51), 131
(80), 117 (75), 91 (90); MS (CI) m/e: 257 (100, M++1), 239
(25), 199 (81). Fraction four contained 19 (RT 8.5); 1H NMR
(CDCl3) δ: 1.40–1.60 (10H, m), 1.61 (2H, m), 1.70 (3H, s),
1.81 (2H, m), 2.60 (1H, m), 6.05 (1H, s), 7.10 (2H, m), 7.30
(3H, m); MS (EI) m/e: 256 (15, M+), 241 (3), 105 (100), 95
(22), 77 (32); MS (CI) m/e: 257 (M++1).
In a similar experiment, 7 (0.142 g, 0.51 mmol) in a solu-
tion of methylene chloride (50 mL) and acetic acid (0.5 mL)
was irradiated for 23 h, and the photolysate was neutralized
with K2CO3. The crude product was chromatographed twice
on a silica gel column to give a small fraction containing 17
When the photolysis was carried out in a mixture of meth-
ylene chloride and acetonitrile (1:1), the GC spectra showed
both peaks in 38 and 36% yields. The fraction of RT 4.20
(10) decreased rapidly on chromatography; one of its impure
1
and a fraction of 16 (0.026 g). H NMR (CDCl3) δ: 1.20–
1.90 (15H, m), 1.98 (3H, s), 2.87 and 3.22 (AB quartet,
diastereotopic -CH2-, J = 16.3 Hz), 4.45 (1H, br), 7.19
(1H, m), 7.28 (2H, m), 7.36 (2H, m); 13C NMR (CDCl3) δ
26.20, 26.35, 26.72, 26.85, 27.04, 27.20, 27.46, 27.79,
32.22, 46.99, 49.88, 79.06, 125.61, 126.43, 127.65, 145.94,
211.72. The injection of 16 to GC gave a single peak at RT
1.77 min, and the fragmentation pattern of 11. The former
fraction containing 17 was contaminated by benzoylacetone,
and showed, on GC–MS determination, a peak at RT 1.45
and the fragmentation pattern of 15 (vide infra).
A solution of 7 (0.023 g, 0.081 mmol) and eicosane
(0.02 g, 0.071 mmol) in methylene chloride (10 mL) was di-
vided in two quartz tubes; acetic acid (10) was added to the
second one. These tubes, sealed with septa and purged with
dry nitrogen, were irradiated with RPR 3000A for 19 h to
give 70% conversions. The GC analysis of the photolysates
(injection temperature 250°C) gave two peaks at 1.40 min
(for 15) and 1.59 min (for 11) with a yield of 27%/15% for
the first and 56%/20% for the second tube (with acetic acid).
Separately, two tubes of methylene chloride solution con-
taining 7 (0.0078 M) and eicosane (0.0071 M), one with the
presence (10 µL) and the other with the absence of acetic
acid, were irradiated under similar conditions except for us-
ing RPR 3500A lamps for 24 h to cause a 40% conversion.
The GC analysis gave a yield for 15/11 of 12%/10%, within
experimental error, for both photolysates.
1
fractions showed H NMR (CDCl3) δ: 0.90 (2H, m), 1.50–
2.00 (12H, m), 2.72 (1H, m), 6.60 (1H, s), 7.11 (2H, m),
7.22 (3H, m), 7.30 (2H, m), 7.70 (2H, m), 8.01 (H, m); MS
(EI) m/e (relative intensity): 318 (60, M+), 207 (5), 192 (5),
105 (100), 91 (11), 77 (31); MS (CI) m/e: 319 (100, M++1).
In a separate photolysis, DBMBBN (0.17 g) in methylene
chloride (50 mL) containing acetic acid (0.5 mL) was irradi-
ated in a similar way to afford a 65% conversion. Repeated
chromatography on a long silica gel column (40 cm3) gave 8
as colorless crystals; 1H NMR (CDCl3) δ: 1.20–1.80
(14H, m), 1.96 (1H, q), 3.25 and 3.95 (AB quartet,
diastereotopic -CH2-, J = 16.3 Hz), 4.88 (1H, br s), 7.12
(1H, t), 7.23 (2H, t), 7.40 (2H, d), 7.43 (2H, t), 7.55 (1H, t),
7.84 (2H, d); 13C NMR (CDCl3) δ: 26.31, 26.40, 26.70,
26.93, 27.08, 27.76, 28.03, 44.13, 47.49, 79.43, 125.75,
126.34, 127.79, 127.98, 128.59, 133.49, 137.35, 146.19,
202.56. Anal. calcd. for C23H28O2: C 82.10, H 8.39; found:
C 82.45, H8.12. This compound on injection to GC–MS
gave a single peak for 11, at an early RT of 1.77 min, which
showed M++1 = 217 by the CI mode, and m/e: 216 (M+,
10%), 133 (12), 120 (12), 105 (100), and 77 (35) by the EI
mode.
BABBN
A solution of 7 in methylene chloride was irradiated for
24 h with RPR 3000A lamps, and the crude product was
chromatographed to give four fractions, each containing a
major component, that showed the RT of the major compo-
nent at 7.6, 8.5, 8.8, and 9.7 min and M++1 = 257 of the ma-
jor peak by the CI mode. Fraction one contained mostly 20
AABBN
A solution of AABBN 6 (0.048 g, 0.22 mmol) in a mix-
ture of cyclohexane and methylene chloride (9:1, 20 mL)
was distributed in four quartz tubes and sealed under nitro-
gen. The solution was irradiated with RPR 2540A lamps for
8 h to cause about 80% conversion. A separate solution pre-
pared from 0.04 g of 6 was irradiated with RPR 3000A
lamps under the same conditions for 18 h to cause a 75%
conversion. Both solutions were examined by GC–MS anal-
ysis to afford the major peaks shown in Table 1. A similar
experiment was carried out with RPR 3500A lamps for 24 h
1
(RT 8.8); H NMR (CDCl3) δ: 1.50–1.68 (10H, m), 1.69
(3H, s), 2.20 (2H, br t), 2.31 (2H, br t), 3.75 (2H, s), 7.40
(2H, m), 7.51 (1H, t), 7.95 (2H, m); 13C NMR (CDCl3) δ:
19.30, 16.19, 26.22, 26.67, 26.99, 27.56, 31.41, 32.11,
44.38, 121.92, 128.09, 128.52, 132.70, 132.76, 137.63,
198.96; MS (EI) m/e: 256 (15, M+), 241 (3), 105 (100), 95
(21), 77 (31); MS (CI) m/e: 257 (M++1). Fraction two con-
© 1999 NRC Canada

1382
Can. J. Chem. Vol. 77, 1999
to give a 15% conversion of 6. The crude product from
254 nm irradiation was chromatographed on silica gel col-
umns with n-hexane elution to give unreacted 6. With in-
creasing amounts of methylene chloride in n-hexane, 13
(7.8 mg), 14 (trace amount), and 12 (18 mg, 54%) were iso-
lated. The physical data of 12: IR (Nujol): 3488 (O–H
stretching), 1699.7 (C=O stretching), 1171.8 (tert-C–O
17.42, 26.34, 26.54, 25.97, 31.60, 31.82, 48.80, 121.56,
165.09, 199.31. This sample showed a GC peak at RT 3.7 in
contrast to that of 5.4 for 14; the GC–MS of the former peak
showed MS (EI) m/e: 194 (40, M+), 179 (42), 151 (10), 136
(12), 123 (25), 109 (71), 95 (100), 81 (50), 69 (40), 43 (98).
Quantitative analysis of the photoreaction of AABBN 6
A solution of 6 (7.2 mg, 0.0364 mmol) in cyclohexane
(25 mL) was divided between five test tubes (5 mL each);
tubes 3 and 4 also contained acetic acid (2 µL). Tubes 1–4
were sealed under dry nitrogen and kept in a dark place.
Tube 5 was sealed under dry oxygen. After 5 days, GC anal-
ysis showed a peak at RT 1.23 min, in trace amounts, for
tubes 1 and 5, and 2% for tube 3; after 8 days the intensity
was 2–3% for all tubes.
Solutions of 6 (0.054–0.056 M) in cyclohexane, benzene,
THF, methylene chloride, and acetonitrile were prepared to
contain fixed concentrations of hexadecane (0.0131–0.0121
M). These solutions, placed in quartz tubes and sealed under
nitrogen dried over solid CaCl2 and NaOH pellets, were irra-
diated with RPR 3000A lamps for 21 h. The photolysates
were analyzed by GC, and the % conversion of 6 was deter-
mined to give >96, 71, 74, 52, and 61% with respect to the
internal standard. A set of similar experiments purged with
nitrogen directly from a tank without drying gave the % con-
version of 60, 64, no data, 41, and 56%.
Four tubes of a solution of 6 (0.007 M) in cyclohexane
containing hexadecane as the internal standard and 1,3-
pentadiene (45:55 mixture of cis/trans) at 0.014, 0.007,
0.0035, and 0.0 M were prepared and sealed under nitrogen.
A solution containing the same substrate concentrations in a
mixture of cyclohexane and methylene chloride was also
prepared and distributed in four tubes to make the second
set. They were irradiated for 5 h, and analyzed by GC for
the % conversion of 6 to give 21, 22, 23, and 60% for the
first set in cyclohexane, and 16, 19, 20, and 64% for the sec-
ond set in the mixture. No new reaction product with the
dienes was detected.
1
stretching) cm–1. H NMR (CDCl3) δ: 1.12 (3H, s), 1.15–
1.25 (2H, m), 1.35–1.83 (13H, m), 2.19 (3H, s), 2.55 and
2.66 (diastereotopic -CH2-, J = 16.3 Hz); 13C NMR (CDCl3)
δ 23.36, 26.40, 26.53, 26.98, 27.14, 27.75, 29.01, 32.10,
46.60, 50.04, 75.09, 211.52. Anal. calcd. for C13H24O2: C
73.54, H 11.39; found: C 73.82, H 11.48. The injection of
12 into GC at 250°C gave two peaks at RT 1.85 and
6.04 min in the ratio of 6:4. The former peak arising from 15
showed M++1 at 155 by theCI mode, and m/e at 154 (1), 139
(1), 111 (10), 97 (14), 69 (100%), 43 (60) by the EI mode.
The latter peak was due to 12 itself and showed m/e by the
CI mode at 213 (M++1, 2), 195 (31), 155 (100) and by the
EI mode at 197 (1), 111 (10), 101 (100), 69 (21), 43 (61).
1
The physical data of 13: H NMR (CDCl3) δ: 1.40–1.55
(6H, m), 1.15–1.72 (4H, m), 1.68 (3H, s), 2.10 (3H, s),
2.14–2.20 (2H, m), 2.20–2.26 (2H, m), 3.13 (2H, s); 13C
NMR (CDCl3) δ: 19.18, 26.07, 26.22, 26.70, 26.90, 27.59,
29.15, 31.29, 32.02, 49.80, 121.58, 138.15, 207.56; MS (EI)
m/e: 194 (15, M+), 179 (16), 151 (10), 136 (12), 109 (26), 95
(100), 81 (35), 69 (42). Anal. calcd. for C13H22O: C 80.35,
H 11.41; found: C 80.11, H 11.58.
The crude product from 300 nm photolysis was worked
up similarly to give 13 (trace amount), 14 (4.3 mg), and 12
(14.3 mg, 54%). 14 showed the following physical con-
1
stants: H NMR (CDCl3) δ: 1.40–1.78 (14H, m), 2.09 (3H,
d, J = 1.5 Hz), 2.16 (1H, br s), 2.18 (3H, s), 6.05 (1H, q, J =
1.5 Hz); 13C NMR (CDCl3) δ: 17.47, 26.02, 26.39, 26.59,
31.65, 48.84, 121.60, 166.11, 199.32; MS (EI) m/e: 194 (10,
M+), 179 (41), 151 (10), 136 (15), 111 (41), 109 (55), 98
(30), 95 (81), 81 (45), 69 (75), 43 (100). Anal. calcd. for
C13H22O: C 80.35, H 11.41; found: C 80.28, H 11.21.
Five tubes of solution in a mixture of cyclohexane and
methylene chloride (9:1) containing 6 (0.0082 M) and
hexadecane (0.0046 M) were prepared. Tube 1 served as the
control run. Acetic acid (0.167 M) was added to tube 2, and
phenyl bromide (0.0082 and 0.0041 M) to tubes 3 and 4, re-
spectively. These tubes were sealed under dry nitrogen ex-
cept for tube 5, which was bubbled with a slow stream of
dry oxygen while all of them were irradiated with RPR 3000
lamps. GC analysis of the photolysates showed the remain-
ing % of 6 in each tube to be 100, 100, 85, 88, and 90%, re-
spectively, at 1 h irradiation, and 88, 84, 44, 46, and 52%,
respectively, at 4 h irradiation. The details of GC analysis of
photolysates 1–4 after 4 h irradiation are given in Table 2.
The photolysate of 6 under oxygen contained completely
different products of highly oxidized materials with RT 2–
5 min; column chromatography failed to give pure products.
Four tubes of similar solutions containing 6 (0.0088 M)
and hexadecane (0.0034 M) were prepared as tubes 6–9, to
which were added 1,2-dibromoethane at 0.0, 0.02, 0.04, and
0.081 M, respectively. These tubes were irradiated as above.
GC analysis of the photolysates gave % of remaining 6 to be
100, 67, 63, and 52%, respectively, after 1 h irradiation, and
98, 47, 46, 29%, respectively, after 2.5 h irradiation. The de-
Sensitized photolysis of AABBN 6
AABBN (0.101 g, 0.46 mmol), hexadecane (0.05 mmol),
and benzophenone (0.042 g, 0.23 mmol) in a mixture of cyc-
lohexane and methylene chloride (9:1, 50 mL) were irradi-
ated with RPR 3000A as above for 6 h to cause a 76%
conversion of 6. GC analysis of the photolysate showed the
peak % of 15, 14a, 13, 14, and 12 to be 4.3, 11, 16, 3.1, and
4.2%. Separately, a cyclohexane solution of 6 (0.0052 M),
hexadecane (0.00033 M), and benzophenone (0.0032 M)
was irradiated with RPR 3500A for 2.6 h to give a 60% con-
version; the control experiment without benzophenone gave
a <2% conversion. The GC analysis gave the peak % for 15,
14a, 13, 14, and 12 of 0.5, 3.5, 6.1, 3.2, and 0.5%. In both
GC analyses the benzophenone peak decreased and several
minor peaks were observed in RT >10 min. Examination of
the crude products with HPLC gave no benzopinacol peak
by coinjection.
The former solution was worked up in the usual way to
give 13 (5.4 mg, 8%), 14 (10.0 mg, 15%), 12 (30.3 mg,
42%), and a mixture of 14a and benzophenone. This mixture
1
showed H NMR (CDCl3) δ: 1.40–1.80 (14H, m), 2.08 (3H,
d, J = 1.2 Hz), 2.15 (3H, s), 6.05 (1H, br s); 13C NMR δ:
© 1999 NRC Canada

Ishiyama and Chow
1383
6. (a) Y.L. Chow, Y.-H. Zhang, M.X. Zhang, and A. Rassat.
Chem. Phys. Lett. 272, 471 (1997); (b) H.-D. Ilge, E. Birckner,
D. Fassler, D. Kozmenko, M.D. Kuz’min, and H. Hartmann. J.
Photochem. 32, 177 (1986).
7. J. Yoon and A.W. Czarnik. J. Am. Chem. Soc. 114, 3874
(1992).
tailed product analyses of the 2.5 h photolysate by GC for
tubes 7–9 are given in Table 2, but those for the control run
(tube 6) were too small and are assumed to be similar to that
shown for tube 1.
8. (a) M.F. Hawthorne and M. Reintjes. J. Org. Chem. 30, 3851
(1965); (b) R. Koster and G.W. Rotermund. Liebigs Ann.
Chem. 689, 40 (1965); (c) L.H. Toporcer, R.E.Dessy, and
S.I.E. Green. Inorg. Chem. 4, 1649 (1965); (d) R. Boese, R.
Koster, and M. Yalpani. Chem. Ber. 118, 670 (1985).
9. (a) N.J. Turro. Modern molecular photochemistry. Benjamin–
Cummings, Menlo Park, Calif. 1978; (b) J. Gersdorf, J.
Mattay, and H. Gorner. J. Am. Chem. Soc. 109, 1203 (1987);
(c) Y.L. Chow, M. Vanossi, and S.-S. Wang. J. Photochem.
Photobiol. A: 88, 125 (1995).
Acknowledgments
The authors wish to thank the Natural Sciences and Engi-
neering Research Council of Canada for generous support of
this project. J.I. gratefully acknowledges the award of an
Overseas Research Fellowship from the Japanese Govern-
ment through the Ministry of Education, Science, Sports,
and Culture.
10. K.L. Pillay, K. Hanaya, and Y.L. Chow. Can. J. Chem. 53,
3022 (1975).
References
1. (a) K. Okada, Y. Hosoda, and M. Oda. J. Am. Chem. Soc. 108,
321 (1986); (b) Tetrahedron Lett. 27, 6213 (1986).
2. K. Utimoto, T. Tanaka, and H. Nozaki. Tetrahedron Lett. 1167
(1972).
3. K. Okada, H. Inokawa, and M. Oda. Tetrahedron Lett. 30,
4395 (1989).
4. (a) S. Chatterjee, P.D. Davis, P. Gottschalk, M.E. Kurz, B.
Sauerwein, X. Yang, and G.B. Schuster. J. Am. Chem. Soc.
112, 6329 (1990); (b) J.D. Wilkey and G.B. Schuster. J. Am.
Chem. Soc. 113, 2149 (1991).
5. (a) Y.L. Chow and X.-E. Cheng. Can. J. Chem. 69, 1331
(1991); 69, 1575 (1991); (b) Y.L. Chow, S.S. Wang, C.I.
Johansson, and Z.-L. Liu. J. Am. Chem. Soc. 118, 11 725
(1996).
11. Y.L. Chow and X. Liu. Can. J. Chem. 69, 1261 (1991).
12. (a) K. Negishi and H.C. Brown. J. Am. Chem. Soc. 95, 6757
(1973); (b) P. Jacob III. J. Organomet. Chem. 156, 101 (1978).
13. K.U. Ingold and B.P. Roberts. Free-radical substitution reac-
tions. John Wiley & Sons, New York. 1971. pp. 46–62.
14. (a) J. Grotewold and E.A. Lissi. Chem. Commun. 1367
(1968); (b) J. Chem. Soc. (B), 264 (1968).
15. L.H. Toporcer, D.E. Dessy, and S.I.E. Green. J. Am. Chem.
Soc. 87, 1236 (1965).
16. (a) W. Gerrard, M.F. Lappert, and R. Shafferman. J. Chem.
Soc. 3648 (1958); (b) E.W. Abel, W. Gerrard, and M.F.
Lappert. J. Chem. Soc. 5051 (1957); 1451 (1958).
© 1999 NRC Canada