Laser flash photolysis study of the photochemistry ofo-methylbenzils

V. Wintgens,a J. C. Netto-Ferreira b and J. C. Scaiano c

a Laboratoire de Recherche sur les Polymères, C.N.R.S., Thiais, 94320-Franceb Departamento de Química, Universidade Federal Rural do Rio de Janeiro Seropédica,

23851-970, Rio de Janeiro, Brazilc Department of Chemistry, University of Ottawa, 10 Marie Curie, Ottawa, K1N 6N5, Canada

Received 7th September 2001, Accepted 7th January 2002First published as an Advance Article on the web 15th February 2002

The photochemistry of 2-methylbenzil (1) was investigated by steady state and laser flash photolysis techniques.Laser excitation of 1 in benzene leads to a transient, which shows absorption at 370 and 470 nm and is quenched byknown triplet quenchers, whereas in methanol or acetonitrile a different transient absorption spectrum was observed.For the latter solvent the absorption bands are located at 350 and 410 nm when the spectrum is recorded 30 µs afterthe laser pulse. The kinetics associated with these two bands shows biexponential decay, from which lifetimes of 200and >200 µs were determined. These transients were attributed to a mixture of two possible photoenols. Similarbehavior was observed for 2,2�-dimethylbenzil (3), whereas 2,4,6-trimethylbenzil (4) did not show any detectable enol.Lamp irradiation of 1 in benzene leads to 2-hydroxy-2-phenylindan-1-one (2) as the major product, with a quantumyield of 0.09. Compound 3 also gives the corresponding hydroxyindanone as the main product (Φ = 0.25), whereasfor 4 the cyclization product is formed in a very low quantum yield, i.e., <0.004. From these data two possiblemechanisms were suggested to explain the formation of hydroxyindanones from methylbenzils, depending onthe nature of the solvent.

IntroductionPhotochemical excitation of aromatic ketones having o-alkylsubstituents with abstractable hydrogen leads to the formationof the corresponding enols, as shown in eqn. (1).1–4

This rection is known to involve two distinct excited tripletstates and to proceed through a biradical, formed by intra-molecular hydrogen transfer.1–4 This biradical gives two enolswith similar spectroscopic properties but with different kineticbehavior (Z-enols have shorter lifetimes than E-enols due torapid reketonization).

There are only few examples in which these enols undergointramolecular reaction to form new products: for example,2-substituted o-ethylbenzophenones give o-vinylbenzophenoneby elimination of the 2-substituent from the ethyl group;5 2,4,6-triisopropylbenzophenone leads to the corresponding cyclobut-enol by an intramolecular [2 � 2] cycloaddition from the enol;6

α-chloroketones undergo chloride elimination with concom-itant formation of indanones 7–10 and o-benzylbenzophenoneforms anthracene derivatives 11 and/or benzocyclobutenols.12

Photoenolizable diketones, such as 1-(o-tolyl)propane-1,2-dione, give upon irradiation the substituted 2-hydroxyindan-1-ones in good yields. The mechanism of photocyclization hasbeen investigated by several authors: Bishop and Hamer 13–15

suggested that product formation involves the intervention ofa benzocyclobutenol, whereas Ogata and Takagi 16,17 invoke theintermediacy of an enol.

Wagner and collaborators have shown that hexaalkylbenzilsphotocyclize to 2-hydroxyindan-1-ones. In all cases the 1,5-

(1)

biradical formed through δ-hydrogen abstraction is proposed asthe intermediate leading to product formation, even though itwas not possible to rule out the participation of a 1,4-biradicalgenerated by γ-hydrogen abstraction from the excited carbonylcompound.18

We report here the results of a study of the photochemistryof a related compound, 2-methylbenzil (1), which photocyclizesto 2-hydroxy-2-phenylindan-1-one (2) [eqn. (2)]. Laser flash

photolysis leads to the observation of different transientsdepending on the solvent; the long-lived photoenols have beendetected in methanol.

In order to elucidate the mechanism of photocyclization(through a 1,4-biradical leading to the enol and then to thecyclization product by an internal aldol condensation, orthrough a 1,5-biradical), two other benzil derivatives, 3 and 4,have been studied. We find that while 3 efficiently photocyclizes,4 is surprisingly less photoreactive.

(2)

184 Photochem. Photobiol. Sci., 2002, 1, 184–189 DOI: 10.1039/b108116k

This journal is © The Royal Society of Chemistry and Owner Societies 2002

Publ

ishe

d on

15

Febr

uary

200

2. D

ownl

oade

d by

Uni

vers

ity o

f V

icto

ria

on 2

7/10

/201

4 04

:03:

04.

View Article Online / Journal Homepage / Table of Contents for this issue

Experimental

Materials

The solvents (Aldrich) were of the highest quality available andused as received. Dienes (Aldrich) were distilled prior to use.

2-Methylbenzil (1) was synthesized in a two step reaction:addition of benzylmagnesium chloride to o-cyanotoluene(Aldrich) in anhydrous ethyl ether, followed by hydrolysis ofthe addition product, leads to the monoketone, o-tolyl benzylketone. This monoketone is oxidized with selenium dioxide inacetic acid to give 1, mp 56–57 �C (mplit 19 = 58 �C). This pro-cedure is identical to the one described for 2-ethylbenzil.20 Thespectroscopic properties of 1 agree with the proposed structure[1H NMR, δ (ppm) (CDCl3) 2.70 (3H, s), 7.20–8.00 (9H, m);IR ν/cm�1 1670 (νC��O)].

2-Hydroxy-2-phenylindan-1-one (2) was obtained by irradi-ation (λ > 300 nm) of 1 in deaerated benzene solution. Afterrecrystallization from benzene–hexane, it showed mp = 126–127 �C (mplit 21 = 127–128 �C).

2,2�-Dimethylbenzil (3) was synthesized in a two stepreaction; 2,2�-dimethylbenzoin was obtained by the usualbenzoin condensation, i.e., heating of o-formyltoluene (Aldrich)in ethanol with sodium cyanide for three hours. The oil obtainedwas then oxidized with an aqueous solution of nitric acid. Com-pound 3 was purified by thin layer chromatography (silica gel,20% ethyl acetate–hexane). Pale yellow crystals were obtained,mp = 85–86 �C. The spectroscopic properties of 3 agree with theproposed structure [1H NMR, δ (ppm) (CDCl3) 2.70 (6H, s),7.10–7.60 (8H, m); IR ν/cm�1 1670 (νC��O) ].

2,4,6-Trimethylbenzil (4) was obtained by a three stepreaction: addition of 2,4,6-trimethylbenzaldehyde (Aldrich) tobenzylmagnesium chloride in anhydrous ethyl ether, followedby the usual work-up, gives 2,4,6-trimethylphenyl(benzyl)-carbinol.† This alcohol was then treated by Jones’ reagent,which leads to the corresponding monoketone, 2,4,6-trimethyl-phenyl benzyl ketone. The yellow oil obtained was oxidizedwith selenium oxide in acetic acid to give 4, yellow crystals,mp = 136–137 �C (mplit 22 = 135 �C). The spectroscopic proper-ties of 4 agree with the proposed structure [1H NMR, δ (ppm)(CDCl3) 2.26 (6H, s), 2.29 (3H, s), 6.90 (2H, s), 7.50–8.20(5H, m); IR ν/cm�1 1665 (νC��O)].

General techniques1H-NMR spectra were recorded on a Bruker AC 200spectrometer (1H: 200 MHz) in CDCl3 using tetramethylsilaneas the internal standard.

Infrared spectra were obtained on a 1420 Perkin-Elmerspectrophotometer in CCl4 liquid film. UV–visible spectrawere recorded on a Hewlett-Packard 8451A diode arrayspectrometer.

Phosphorescence spectra were recorded in an EPA (ether–isopentene–alcohol) glass at 77 K on a Perkin-Elmer LS-5spectrofluorimeter equipped with a PE-3600 Data Station.

GC analyses were carried out on a Perkin-Elmer model 8320capillary gas chromatograph employing a 12 m J&W bondedphase vitreous silica BP1 silicone column. GC-MS analyseswere performed on a Hewlett-Packard model 5995 system.

Melting points were determined in a Mel-Temp apparatusand were not corrected.

Lamp irradiations were carried out in Pyrex tubes and in areactor fitted with 9 RPR-3500 lamps.

Laser flash photolysis

Laser flash photolysis experiments were carried out with thesystem described previously.23,24 The samples of 1 were excitedwith the pulses from a Lumonics TE-860/2 excimer laser

† The IUPAC term for a carbinol is a substituted methanol.

(308 nm, ∼5 ns, ∼20 mJ per pulse). In order to ensure that eachshot irradiates fresh solution, a flow system was used.

Lamp irradiation

Typical samples were 1 mL containing 0.03 M diketone inbenzene or in methanol and were deaerated by bubblingoxygen-free nitrogen. The products were analyzed by GCusing n-dodecane as an internal standard. For quantum yielddetermination, the photofragmentation of valerophenone inbenzene was used as an actinometer (Φ = 0.3 for acetophenoneformation 25).

Results

Product studies

Lamp irradiation of 1 in benzene leads to 2, as the majorphotoproduct. At high conversion, few other products wereobserved and accounted for less than 5% of the total productmixture. The quantum yield of formation of 2 was ∼0.09,which is low in comparison with the value of 0.54 reportedby Ogata and Takagi 16 for photoproduct formation from1-(o-tolyl)propane-1,2-dione. The reaction of 1 is efficientlyquenched by bubbling oxygen or by adding cyclohexa-1,3-diene(≥10�2 M) to the diketone solution. This suggests that thecyclization involves the triplet excited state of 1; this is notunexpected in view of the usual high intersystem crossingquantum yield of α-diketones (0.92 for benzil;26 0.97 forbutane-2,3-dione 27).

When the irradiation is carried out in methanol, 1 photo-reacts efficiently (Φ ≥ 0.20 for disappearance of startingmaterial); 2 is still the main product (Φ ≈ 0.03), but otherproducts were also observed. While these were not identified,they are likely to result from the photoreaction of 2-phenyl-2-hydroxyindan-1-one (2) due to its high photochemicalreactivity, leading probably to the corresponding ketoaldehydeby α-cleavage.21

Lamp irradiation of 3 in benzene also leads to the corre-sponding hydroxyindanone with a quantum yield of Φ ≈0.25, i.e., significantly higher than that for 1 (Burkoth andUllman reported that 3 is less reactive than 1-(o-tolyl)propane-1,2-dione 28). Lamp irradiation in methanol shows few otherphotoproducts beside the photocyclization product (Φ ≈ 0.16).

Irradiation of 4 in benzene gives the hydroxyindanone with avery low quantum yield (Φ ≤ 0.004). A value of 0.02 for thequantum yield of disappearance of 4 in tetrahydrofuran hasalready been reported.22

Laser flash photolysis studies

Laser flash photolysis of 1 (308 nm) in benzene gives a transientabsorption spectrum with maxima at 370 and 470 nm (seeFig. 1). This species, attributed to the triplet state of 1, decays at

Fig. 1 Triplet absorption spectrum obtained on 308 nm excitation ofo-methylbenzil in benzene. Inset: decay trace monitored at 480 nm dueto the triplet generated as above.

Photochem. Photobiol. Sci., 2002, 1, 184–189 185

Publ

ishe

d on

15

Febr

uary

200

2. D

ownl

oade

d by

Uni

vers

ity o

f V

icto

ria

on 2

7/10

/201

4 04

:03:

04.

View Article Online

470 nm by mixed first and second order kinetics (see insert inFig. 1), with an extrapolated lifetime of ∼5.7 µs at zero laserintensity. The second order component is attributed to triplet–triplet annihilation which is quite common in laser studies ofthis type. At shorter wavelengths (360 and 410 nm), the decaysshow the same lifetime but present some residual absorption(∼5% of the signal), which is probably due to product formationor to enol absorption. Under the same conditions, benzil givesa similar triplet absorption spectrum with maxima at 360 and480 nm, and with an extrapolated lifetime of ∼19.8 µs at zerolaser intensity. Both benzil and 1 lead to strong signals underlaser excitation (those from 1 being slightly weaker), reflectingtheir high intersystem crossing efficiencies.

The triplet state of 1 is readily quenched by dienes leading toquenching rate constants (kq) of 3.4 × 108 and 1.5 × 108 M�1 s�1

for cyclohexa-1,3-diene and 2,5-dimethylhexa-2,4-diene, respec-tively. These rate constants are significantly below the diffusioncontrol limit, a fact that can be attributed to the low tripletenergy of 1. We evaluated a triplet energy of 55 kcal mol�1 for 1from its phosphorescence spectrum in an EPA glass at 77 K.This compares with the value of 53.4 kcal mol�1 for benzil.29

Oxygen also quenches the triplet state of 1; the quenching rateconstant for oxygen is also rather low, kq = 5 × 108 M�1 s�1, withthe same behavior being observed for benzil, kq = 5.3 × 108 M�1

s�1. Relatively low rate constants for energy transfer are notunprecedented in the case of benzil; for example, the rate con-stant for triplet energy transfer to di-tert-butyl peroxide is twoorders of magnitude lower for benzil than for benzophenone.30

The triplet state of 1 is also quenched by hydrogen donors, suchas cyclohexa-1,4-diene with a rate constant of 3.8 ×105 M�1 s�1.

Laser irradiation of 1 in methanol solution leads to a differ-ent transient absorption spectrum with maxima at 350 and410 nm (see Fig. 2). The ratio between the two bands is slightly

different if the spectrum is recorded 70 or 700 µs after the laserpulse, which indicates the presence of transients showing verysimilar, but not identical, absorption spectra. The transientdecay monitored either at 350 or at 410 nm is multiexponential(lifetimes of ∼200 µs and >200 µs can be estimated). Weattribute these transients to a mixture of two possible photo-enols (see Scheme 1). Neither oxygen nor dienes has any effecton the enol lifetimes, as expected, but they both decrease theintensity of the signal, indicating quenching of a triplet precur-sor. On a shorter time scale, we observed a transient with amaximum at 480 nm, and decaying by first order kinetics witha lifetime of 1.8 µs. This species was assigned to the triplet stateof the diketone 1, as it was readily quenched by oxygen (kq =6.3 × 108 M�1 s�1) and by dienes (kq = 1.1 × 109 M�1 s�1 forcyclohexa-1,3-diene).

The big difference in the photobehavior of 1 betweenbenzene and methanol led us to carry out laser flash photolysis

Fig. 2 Enol absorption spectra obtained on 308 nm excitation ofo-methylbenzil in methanol recorded 70 µs (A) and 700 µs (B) after thelaser pulse.

experiments in another polar but not hydrogen bondingsolvent, such as acetonitrile. Fig. 3 shows the spectra recordedin acetonitrile. Immediately after the laser pulse, the spectrumcorresponds to the triplet excited state of 1 with maxima at 365and 465 nm and a lifetime of ∼3.1 µs. When the spectrum isrecorded 30 µs after the laser pulse, the bands present showmaxima at shorter wavelengths, and the spectrum shows simi-larities with that for the enols recorded in methanol. The signalsdue to the enols, while stronger than in benzene, were still tooweak to measure their lifetimes accurately.

Scheme 1

186 Photochem. Photobiol. Sci., 2002, 1, 184–189

Publ

ishe

d on

15

Febr

uary

200

2. D

ownl

oade

d by

Uni

vers

ity o

f V

icto

ria

on 2

7/10

/201

4 04

:03:

04.

View Article Online

Laser flash photolysis of 3 in benzene leads to a transientabsorption spectrum with maxima at 365 and 460 nm, when thespectrum is recorded 2 µs after the laser pulse. A spectrumrecorded 20 µs after the laser pulse shows only one bandcentered around 450 nm. Due to the extensive spectral overlap,the kinetics is clearly multiexponential and rather difficult toanalyze. By addition of a triplet quencher (oxygen, dienes) thedecay traces can be separated into a short component (due tothe triplet) and a long one (due to the enols). Typically, in anaerated solution, a lifetime of ∼1.0 µs can be attributed to thetriplet (more easily monitored at 365 nm); even with this shorttriplet lifetime, the decay at 450 nm—due to the enols—remainsmultiexponential (lifetimes of ∼10 µs and ≥40 µs can beestimated).

The behavior of 3 in methanol under laser excitation is quitesimilar to that of 1: a triplet with maxima around 360 and460 nm and a lifetime ∼0.9 µs in an aerated solution is observed.Again a spectrum recorded 20 µs after the laser pulse showsonly one band at 430 nm. The lifetime of the enols is longerthan in benzene (∼180 µs and >200 µs).

Laser flash photolysis of 4 in benzene leads to a transientabsorption spectrum with maxima at 380, 430 and 460 nm.This species decays with first order kinetics, with a lifetimeof 1.6 µs. The transient is quenched by oxygen and dienes(kq = 8.1 × 107 M�1 s�1 for 2,5-dimethylhexa-2,4-diene), and isthus assigned to the triplet state of 4. No transient, long orshort-lived, is observed in methanol.

DiscussionThe transient absorption spectra shown in Fig. 2 (in methanol)have been assigned to the mixture of enols derived from 1mainly by comparison of these absorptions with those for othero-xylylenes and photoenols: for example, the xylylene 5 hasmaxima at 340 and 460 nm,31 and the photoenol 6 has amaximum at 390 nm.1,3

We were unable to quench these photoenols within the timescale of these experiments by any common trapping agent (suchas dimethyl acetylenedicarboxylate or dimethyl malonate).Ogata and Takagi reported that irradiation of 1-(o-tolyl)-

Fig. 3 Transient absorption spectra obtained on 308 nm excitation ofo-methylbenzil in acetonitrile recorded 2 µs (A) and 30 µs (B) after thelaser pulse.

propane-1,2-dione in the presence of dimethyl acetylene-dicarboxylate leads to the [2 � 4] photoadduct (thermallyunstable).16,17 Thermal decomposition of this photoadductleads to hydroxyindanone, strongly indicating that photo-enols, produced via thermal reaction, do rearrange to thehydroxyindanones.

On the other hand, the yield of enol formation depends onoxygen concentration, and to a lesser extent on diene concen-tration, indicating a mechanism of enol formation from thetriplet excited state of 1. No growth for enol formation could beobserved due to the large spectral overlap of the triplet stateand the enols, as well as to the strong triplet signal when com-pared to that from the enols. However, some contribution to theformation of the Z-enol from the singlet excited state cannot beruled out; a singlet mechanism is operative in the related case ofo-methylphenacyl chloride.10

The apparent monoexponential decay of the triplet state of1 in methanol, with a relatively long lifetime, as well as thequenching of enol formation by triplet quenchers, indicate thatthe triplet state observed by laser flash photolysis must be theone which gives the 1,4-biradical by hydrogen abstraction andultimately the enols. We have no evidence for two distinct(syn and anti) triplet states; this means that the anti-to-syninterconversion of the triplet is fast compared with the intra-molecular hydrogen abstraction to form the correspondingbiradical (Scheme 1).

We were unable to observe the 1,4-biradical, an expectedprecursor of the enols formed in methanol. This is the resultof the long triplet lifetime that prevents the spectroscopicdetection of shorter lived intermediates derived from it.

As mentioned previously, the Z-enol is expected to revertvery rapidly to the diketone 1 by an intramolecular 1,5-hydrogen shift, and cyclization of the E-enol could occur by afast concerted process. The enols formed from 1 have partic-ularly long lifetimes, which does not agree with fast processes.These long lifetimes can also reflect stabilization by intra-molecular hydrogen bonding between the hydroxy-group andthe α-carbonyl (see Scheme 1). Such an intramolecular bondcould be formed for both E- and Z-enols. In the case of theZ-enol, this intramolecular hydrogen bond should stabilizethis species and lower the rate of reketonization leading to anunusually long lifetime for this enol; in the case of the E-enol,it should provide the appropriate geometrical arrangementfavoring cyclization to the hydroxyindanone 2 by an internalaldol-type condensation.21 Several authors have suggestedintramolecular hydrogen bonding 17,28 to explain the fact that nodeuterium incorporation is observed when the irradiation isperformed in deuterated methanol.14 The decays recorded forthe enols showed multiexponential behavior and the half-liveswere too long for our experimental set-up; it is thus ratherdifficult to attribute precise lifetimes to specific enols.

The behavior of 1 in benzene is quite different; we failed toobserve the enols, even though 2 is formed with a significantquantum yield. A first possibility is that both the retroketoniz-ation and the cyclization processes are faster in this solvent thanin methanol. In benzene, such a shortening of an enol’s lifetimehas already been observed for the case of α-chloro substitutedketones,7 and it was also noticed for the enols formed fromthe diketone 3. Another possibility, shown in Scheme 2, wouldbe a hydrogen abstraction by the other carbonyl group to givethe 1,5-biradical, followed by its cyclization to give directly thehydroxyindanone 2. Ogata and Takagi reported that such a 1,6-hydrogen shift does not occur in the case of α-(o-tolyl)acetone,17

but Wagner and co-workers showed that α-(o-tolyl)acetophen-one undergoes photocyclization to 2-phenylindan-2-ol througha short-lived 1,5-biradical with a quantum yield of unity.32,33

The possibility of the formation of 2 by 1,6-hydrogen shiftshould not be ruled out. This would be in good agreementwith the lack of enol formation in benzene, even if 2 is formedin this solvent with high efficiency. In fact, Wagner et al. clearly

Photochem. Photobiol. Sci., 2002, 1, 184–189 187

Publ

ishe

d on

15

Febr

uary

200

2. D

ownl

oade

d by

Uni

vers

ity o

f V

icto

ria

on 2

7/10

/201

4 04

:03:

04.

View Article Online

showed that at least for hexaalkylbenzils the 1,5-biradical mustbe the transient involved in the hydroxyindanone formation.18

As expected, we failed to observe the 1,5-biradical, which canbe anticipated to be short lived (≤50 ns) and comparable inlifetime to values reported for different 1,5-biradicals.32 Con-sidering the quantum yield for formation of 2 and the tripletlifetime of 1 in benzene, this gives a rate constant of 2 × 104 M�1

s�1 for intramolecular hydrogen abstraction, which is low, butcompatible with the modest triplet energy for 1 of only 55 kcalmol�1. For the sake of comparison, the monoketone α-(o-tolyl)-acetophenone shows a rate constant for the intramolecularhydrogen abstraction of 1.6 × 108 M�1 s�1, but has a muchhigher triplet energy.33 The low reactivity of 1 is also reflectedin the measured rate constant for intermolecular hydrogenabstraction from cyclohexa-1,4-diene of 3.8 × 105 M�1 s�1 (com-pared with the value for benzophenone 34 of 2.9 × 108 M�1 s�1).

The carbonyl bonded to the phenyl group is less stericallyhindered, and should be the most solvated in an alcoholicmedium; such a solvation usually destabilizes the excited state.Therefore the lower triplet state should be largely localizedon the carbonyl bonded to the o-tolyl group, thus leading tothe 1,4-biradical by intramolecular γ-hydrogen abstraction(Scheme 1). In agreement with this statement the change ofsolvent from benzene to methanol leads to the observationof both Z- and E-enols. The lowering of the quantum yieldof formation of hydroxyindanone in methanol compared tobenzene is consistent with a considerable fraction—or all—of2 being formed by the mechanism described in Scheme 2, i.e., bya 1,5-biradical mechanism.

In order to get a better insight into the mechanism for thisphotocyclization process, we also examined the diketones 3 and4. Due to the steric hindrance induced by the methyl groups in4, we expect that the excitation will predominantly be localizedon the carbonyl bonded to the unsubstituted benzene ring. The

Scheme 2

diketone 3 is symmetrical, and whatever the solvent is, theexcited carbonyl group will be in a position α to an o-tolylgroup.

Laser excitation of 3 leads to the observation of the enols inboth solvents, benzene and methanol. The efficiency of enolformation in benzene is quite high, and the contribution of enolabsorption to the signal intensity is at least six times higher thanin the case of 1 (comparison done at the enol maximum absorp-tion and on a short time scale). The quantum yield of photo-product formation is 2.7 times higher than that from 1. Thus, itseems that even if more of the enols are formed from 3 thanfrom 1, there is no comparable increase in the photoproductformation.

The diketone 4 does not give any detectable enols under laserirradiation, which was expected since the two o-methyl groupsinduce steric hindrance and the carbonyl is out of the plane ofthe phenyl (2,4,6-trimethylbenzophenone also does not formany enol under laser irradiation 35). On the other hand, the verylow photoreactivity of 4 could indicate that the δ-hydrogenabstraction is also unfavorable. We note that laser irradiation of4 leads to a weak triplet signal compared to the other benzils;the low photoreactivity may also reflect a low intersystem cross-ing efficiency, and not that 1,5-biradical formation does notoccur. Even though no particular explanation has been foundfor this apparent increase in the efficiency of the non-radiativeprocesses, this fact was mentioned previously.22

In conclusion, our studies demonstrate that steric effects,in combination with hydrogen bonding, can control excitedstate localization in α-diketones. In the case of 1 this effectleads to remarkable regioselectivity in the behavior of its tripletstate. Thus, in hydrogen bonding solvents 1,5-hydrogen transferis favored, leading to long-lived and readily detectable enols.In contrast, in benzene, where intermolecular hydrogen bond-ing is not involved, 1,6-hydrogen transfer dominates and leadsto the key product, 2, without any requirement for enolintermediacy.

Acknowledgements

We are grateful to the Natural Sciences and EngineeringResearch Council of Canada for generous support of thisresearch.

References1 P. K. Das, M. V. Encinas, R. D. Small, Jr. and J. C. Scaiano, J. Am.

Chem. Soc., 1979, 101, 6965.2 J. C. Scaiano, Acc. Chem. Res., 1982, 15, 252.3 L. J. Johnston and J. C. Scaiano, Chem. Rev., 1989, 89, 521.4 R. Haag, J. Wirz and P. J. Wagner, Helv. Chim. Acta, 1977, 60, 2595.5 S. S. Teng and E. F. Ullman, J. Am. Chem. Soc., 1976, 98, 541.6 Y. Ito, Y. Umehara, T. Hijiya, Y. Yamada and T. Matsuura, J. Am.

Chem. Soc., 1980, 102, 5917.7 J. C. Netto-Ferreira and J. C. Scaiano, J. Am. Chem. Soc., 1991, 113,

5800.8 W. R. Bergmark, J. Chem. Soc., Chem. Commun., 1978, 61.9 W. R. Bergmark, C. Barnes, J. Clark, S. Paparian and S. Marynowski,

J. Org. Chem., 1985, 50, 5612.10 A. P. Pelliccioli, P. Klán, M. Zabadal and J. Wirz, J. Am. Chem. Soc.,

2001, 123, 7931.11 J. C. Netto-Ferreira, V. Wintgens and J. C. Scaiano, Can. J. Chem.,

1994, 72, 1565.12 M. Sobczak and P. J. Wagner, Tetrahedron Lett., 1998, 39, 2523.13 R. Bishop and N. K. Hamer, J. Chem. Soc. (C), 1970, 1193.14 N. K. Hamer and C. J. Samuel, J. Chem. Soc., Perkin Trans. 2, 1973,

1316.15 N. K. Hamer, J. Chem. Soc., Perkin Trans. 1, 1979, 508.16 Y. Ogata and K. Takagi, Bull. Chem. Soc. Jpn., 1974, 47, 2255.17 Y. Ogata and K. Takagi, J. Org. Chem., 1974, 39, 1385.18 P. J. Wagner, B. S. Park, M. Sobczak, J. Frey and Z. Rappoport,

J. Am. Chem. Soc., 1995, 117, 7619.19 J. T. Craig and K. P. McNatty, Aust. J. Chem., 1967, 20, 1921.20 N. K. Hamer, J. Chem. Soc., Perkin Trans. 2, 1979, 1285.

188 Photochem. Photobiol. Sci., 2002, 1, 184–189

Publ

ishe

d on

15

Febr

uary

200

2. D

ownl

oade

d by

Uni

vers

ity o

f V

icto

ria

on 2

7/10

/201

4 04

:03:

04.

View Article Online

21 A. Padwa and A. Au, J. Am. Chem. Soc., 1976, 98, 5581.22 K. Maruyama, K. Ono and J. Osugi, Bull. Chem. Soc. Jpn., 1972, 45,

847.23 J. C. Scaiano, J. Am. Chem. Soc., 1980, 102, 7747.24 J. C. Scaiano, M. Tanner and D. Weir, J. Am. Chem. Soc., 1985, 107,

4396.25 P. J. Wagner, P. A. Kelso, A. E. Kempainen, J. M. McGrath, H. N.

Scott and R. G. Zepp, J. Am. Chem. Soc., 1972, 94, 7506.26 A. A. Lamola and G. S. Hammond, J. Chem. Phys., 1965, 43, 2129.27 H. H. Richtol and F. H. Klappmeier, J. Chem. Phys., 1966, 44, 1519.28 T. L. Burkoth and E. F. Ullman, Tetrahedron Lett., 1970, 1193.

29 S. L. Murov, Handbook of Photochemistry, M. Dekker, New York,1973 .

30 J. C. Scaiano and G. G. Wubbels, J. Am. Chem. Soc., 1981, 103,640.

31 J. C. Netto-Ferreira, V. Wintgens and J. C. Scaiano, TetrahedronLett., 1989, 30, 6851.

32 M. A. Meador and P. J. Wagner, J. Am. Chem. Soc., 1983, 105, 4484.33 P. J. Wagner, D. Subrahmanyam and B. S. Park, J. Am. Chem. Soc.,

1991, 113, 709.34 M. V. Encinas and J. C. Scaiano, J. Am. Chem. Soc., 1981, 103, 6393.35 B. Guerin and L. J. Johnston, Can. J. Chem, 1989, 67, 473.

Photochem. Photobiol. Sci., 2002, 1, 184–189 189

Publ

ishe

d on

15

Febr

uary

200

2. D

ownl

oade

d by

Uni

vers

ity o

f V

icto

ria

on 2

7/10

/201

4 04

:03:

04.

View Article Online