1972 229Primary P hot0 processes in lsoq u i no1 i ne N-OxidesBy Christian Lohse,’ Davy Faraday Research Laboratory of the Royal Institution, London W.1The photoisomerization of isoquinoline N-oxides (I) in a variety of solvents was investigated by use of conventionaland laser flash photolysis techniques, combined with fluorescence and quantum yield measurements. The form-ation of isoquinolones (11) in polar solvents and 1.3-benzoxazepines (111) in non-polar solvents proceeds via singletexcited states. A triplet excited state of isoquinoline N-oxides was observed and characterized, but was shownnot to contribute to the photochemical reactivity. A possible oxaziridine intermediate could not be observed.Experiments with a laser flash showed that the isoquinolone was formed within 20 ns.Since the singlet lifetimei s 1 ns. there is little time left for the formation of an intermediate. Parameters for the various modes of decay of thefirst excited singlet state (e.g. fluorescence quantum yields, radiative lifetimes, and quantum yield for productformation) were measured for reactions in various solvents. From these the rate constants of product formationwere calculated. The rate of formation of isoquinolone is nearly solvent-independent; the rate of formation ofoxazepine in polar solvents i s 20 times slower than that of isoquinolone and 30 times greater in non-polar solvents.THE photochemistry of heteroaromatic amine oxides hasbeen the subject of a number of recent publi~ations.l-~Quinoline N-oxide (IV) apparently undergoes only twomain types of rearrangement.In polar hydroxylicsolvents the photoproduct is the quinolone (V) 4-6 andin non-polar, non-hydroxylic solvents the photoproductis 3,l-benzoxazepine (VI) .5?6 Furthermore, a smallamount (2-5%) of the parent base, quinoline has beenR3&; 0 R’( I 1.)0 H~ : 2 0HOH H(Yl31isolated in many cases1* Isoquinoline N-oxides havenot been as fully investigated, but the availableresults 57798 indicate that similar processes take place.The main photoproducts in polar solvents are iso-quinolones, and in non-polar solvents, 1,3-benzoxa-zepines. This paper deals with some of the primary* Present address : The Chemical Institute, University of1 G. G. Spence, E. C. Taylor, and 0. Buchardt.Chem. Rev.,C. Kaneko, S. Yamada, and M. Ishikawa, Tetrahedron Letters,Odense, 6000 Odense, Denmark.1970, 70, 231.1970, 2329.3 J. Streith and P. Martz, Tetrahedron Letters, 1969, 4899.photoprocesses in the rearrangements of isoquinolineN-oxides.RESULTS AND DISCUSSIONSteady State Irradiations.-Irradiation of an iso-quinoline N-oxide with either a hydrogen atom or analkyl group in the 1-position in ethanol, gave thecorresponding isoquinolone in high yield (Table 1). TheTABLE 1Solvent(dielectric const.)Ethanol (24-30) (114Acetone (20.7) (IW(W(IIb)Ethanol (IW(IIC)(WAcetone (IId)(114EthanolAcetoneAcetoneEthanolEthanolWater (80.30) (IIf) (95)Ethanol (IIf) (75)Acetone (IIf) (26), (VIIf) (34),Ethylene chloride (IIf) (21)(VIIIf) (21)(30.65)Ethyl acetate (6.02) (IIf) (18)Benzene (2.28) (IIf) (6.2)Carbon tetrachloride (IIf) (4.3)(2.24)yield of isoquinolone was significantly less when theN-oxides were irradiated in acetone.Furthermore, newphotoproducts, derivatives of the phenol (VII) appeared.These are known products of the hydrolysis of stable1,3-benzoxazepines (the primary photoproduct froml-cyano- and 1-phenyl-substituted isoquinoline N -oxides). This provides good evidence for the formationof 1,3-benzoxazepines as primary photoproducts in thephotolysis of alkyl-substituted isoquinoline N-oxides innon-polar solvents.8 Although the chemical yields fromthe irradiations do not account for all the N-oxide4 0. Buchardt, Acta Chem, Scand., 1963, 17, 1467.6 M.Ishikawa, S. Yamada, H. Hotta, and C. Kaneko, Chem.6 0. Buchardt, P. L. Kumler, and C . Lohse, Ada Chem.7 0. Buchardt, C. Lohse, A. M. Duffield, and C. Djerassi,8 C. Lohse, Tetvahedron Letters, 1968, 5625.and Phavm. Bull. (Japan), 1966, 14, 1102.Scand., 1969, 23, 159.Tetrahedron Letters, 1967, 2741230 J.C.S. Perkin 11consumed, no isoquinolines were detected in theproduct mixture, in contrast to previous report^.^In order to investigate the influence of the solventfurther, 3-methylisoquinoline N-oxide was irradiated ina series of solvents with different dielectric constants.The results (Table 1) demonstrate a pronounced solventeffect. Comparison with the results for alkyl-substituted quinoline N-oxides indicates strongly thata similar mechanism is operating.The intervention of an oxaziridine intermediate hasbeen suggested to explain the results in the photolysisof quinoline N-0xides.l If a similar reaction pathway isassumed to be operating in the photolysis of isoquinolineN-oxides the excited N-oxide will rearrange to give theoxaziridine (IX) (Scheme 1).In polar solvents, theQNO RI m+ 0- R(c) flashing of an ethanolic solution containing l-methyl-isoquinoline N-oxide and anthracene through ananthracene filter permitted the observation of anthraceneI I I I \L / n m400 500 600FIGURE 1 Triplet-triplet absorptions of l-methylisoquinolincN-oxide, 4 p s delaytriplet-triplet absorption,1° but not of the abovetransient.The triplet from 1-methylisoquinoline N-oxide aswell as the triplet from other N-oxides were observed indifferent solvents (Table 2).The spectral region in whichTABLE 2Triplet-triplet absorption of isoquinoline N-oxidesCompound(14(XI 1 (XISCHEME 1next step involves the formation of the zwitterion (X),followed by a [1,2] shift of the alkyl group. In non-polar solvents, the pathway from the oxaziridine isformulated as a [1,5] sigmatropic shift to the epoxystructure (XI), followed by a new [1,5] shift to theoxazepine (111). This mechanism is attractive becausethe behaviour of alkyl-substituted N-oxides in differentsolvents can be explained in a convincing way.l But inspite of much work, oxaziridine intermediates have notso far been detected.We thought that conclusiveevidence might be obtained by flash-spectroscopicstudies.Flash Photo2ysis.-By use of flash-spectroscopic tech-niques it was possible to observe a short lived transientfrom a degassed solution of 1-methylisoquinoline N-oxidein ethanol (Figure 1). This transient is believed to bethe triplet state of l-methylisoquinoline N-oxide because(a) the decay is first order with a rate constant of1.3 x lo5 s-l; ( b ) it is quenched by oxygen; and9 G. Porter, ‘ Technique of Organic Chemistry,’ ed. A. Weiss-berger, Interscience, New York, 1963, vol. 8, part 11, p. 105(14(144-Bromo-l-cyano-isoquinolineN-oxideSolventWaterEthanolC yclohexaiieWaterEthanol90% EthanolC yclohexancEthanolWaterEthanolC yclohexane.\bsorptionregion(4360-650,389380-640,396400-620,418360-630, 388370-630,395,370-640,398400-620,422380-620, 392370-650,382410-620,420420-650,427418First-ordcrrateconstant4.91.32.81.31.31 0 - 5 k p1.11-73 410.42.55.9the absorption is observed and the absorption maximaand lifetimes are similar for all the observed transients.Since the photoproducts vary for these compounds, thepossibility that the transients could be triplets of a1.4,- an q 0 2L t I I IA,, / nmFIGURE 2 Transient from isoquinolone: a, 7 ps delay;b, 12 ps delay400 500 600 650photoproduct seems excluded.The conclusion wasconfirmed by flash photolysis of some isoquinolones.Transients were observed (Figure 2), but the absorption10 G.Porter and M. W. Windsor, DZ’SCZ~SS. Favaday SOC., 1954,7, 1781972 231spectra of these and their lifetimes were different fromthose of the corresponding N-oxides.Table 3 shows the effect of degassing on the lifetime ofthe triplet from 3-methylisoquinoline N-oxide. Al-TABLE 3'Triplet lifetimes for 3-methylisoquinoline N-oxideFirst-order rate constant 10-6k/s-1Solvent Degassed AeratedC yclohexane 3.1 50Ethyl acetate 2.5 62Ethanol 2.2 62Water 4.9 11TABLE 4Quantum yields (4) for3-tncthylisoquinoline N-oxideOxazeyine Yield ofN-Oxidc Isoquino- appearance isoquino-disappear- lone (differ- lone ("/o)Solvent ance appearance ence) Calc. Found('arbon tetra- 0.42 0.022 0-40 5.4 4.3fcthyl acetate 0.31 0.035 0.27 12 17Ethanol 0.19 0.15 0-035 76 75Water 0.16 0.16 0 100 95( ' yclohexane 0.42 0.014 0.4 1chloridethough the triplet is effectively quenched by oxygen,the quantum yield of product formation (Table 4) wasfound to be unaffected by degassing.Thus it is con-cluded that the observed photoproducts arise from anexcited singlet state.The quantum yield for the disappearance of 1 -methylisoquinoline N-oxide in cyclohexane and ethanolwas found to be independent of the light intensity fromthe flash-lamps. Therefore, no photochemically formedintermediate is subject to a further photochemicalreaction. The photoproducts are formed in two parallelreactions. If the number of photochemical steps ineach is the same, the ratio of the quantum yields will beequal to the ratio of the chemical yields.This allows acalculation of the chemical yield from the quantumyield. Fair agreement is obtained between the calcu-lated yields (Table 4) and those obtained by steady stateirradiations. Although two or more photochemicalsteps could be involved in each of the photochemicalrearrangements, the probability of only one photo-chemical step occurring, i.e. the excitation of the AT-oxide, is high. Thus, there is good evidence for a mono-photonic reaction.It was not possible to detect any common inter-mediates for the two photoproducts. On the contrary,there is evidence for the lack of such intermediates.Though no data could be obtained on the rate of form-ation of oxazepines, more informative results wereobtained for the formation of isoquinolones.When a5 x 10-5~-solution of isoquinoline N-oxide in water wasflash-photolysed, absorption bands identical with those* If a 10-6xvx-solution of isoquinoline N-oxide was flash-photolysed, a new transient with a lifetime of 00 ps appearedin the 310-640 nm region. This transient could not be observedfor more concentrated solutions of isoquinoline N-oxide, but atransient with a similar lifetime and absorption was formed froma solution of isoquinolone (Figure 2).of isoquinolone were observed after 4 ps, and theirintensity did not change thereafter. Since isoquinoloneis the only photoproduct from isoquinoline N-oxide inwater the reaction is complete in 4 ps.*A similar experiment using laser flash spectrophoto-metric measurements was unsuccessful because of a verylow signal-to-noise ratio due to light absorption by thetriplet and emission of fluorescence.However fluores-cence measurements of solutions of isoquinoline N-oxideand isoquinolone in ethanol gave supplementary inform-ation. Both solutions were 5 x 10-4~ with the sameoptical density at 347 nm, the frequency doubled wave-length of the ruby laser pulse. The relative fluorescenceintensities obtained by using a conventional fluorescencetechnique (i.e. low-intensity continuous excitation lightsource) were 0.299 for isoquinoline N-oxide and 1.00 forisoquinolone. However when identical solutions wereexcited with one single laser pulse (e.g. a high intensitylight pulse) the fluorescence intensities were 0.450 and1.00, respectively.The intensity of the laser pulse was10l4 times greater than that of the fluorimeter lightsource, and sufficiently intense to excite every singleN-oxide molecule in the solution. Isoquinolone ishowever photostable under these conditions and thechange in the ratio of the fluorescence intensities musttherefore be due to a photochemical conversion of iso-quinoline N-oxide during the light pulse.Because isoquinolone is the main photoproduct fromisoquinoline N-oxide in ethanol, and because thefluorescence of isoquinoline N-oxide changes significantlyto resemble that of isoquinolone under the influence of asingle laser pulse, it is inferred that the formation ofisoquinolone takes place within the duration of onelaser pulse, 20 ns.The singlet lifetime of isoquinoline N-oxide is 1 ns.tThe difference between this singlet lifetime and theapparent upper time limit for the formation of iso-quinolone is not sufficient for the formation and subse-quent thermal breakdown of any intermediate. It isthus concluded that there is no common intermediategiving rise to the two photoproducts.Since the photochemical reaction takes place from the* ' 1 1 $, 4 - so -=SCHEME 2 1, Fluorescence ; 2, isoquinolone formations;lowest excited singlet state, the parameters of this areimportant for a mechanistic discussion.Scheme 2t Measured with a pulse sampling fluorescence apparatus."I thank Dr. G. S. Beddard for this measurement.11 L.Hundley, T. Coburn, E. Garwin, and L. Stryer, Rev. Sci.I m f v . , 1967, 38, 488.3, oxazepine formations ; 4, triplet formation232 J.C.S. Perkin I1shows the processes responsible for the decay of the firstexcited singlet state of 3-methylisoquinoline N-oxide.The quantum yields for processes 2 and 3 in five solventsare given in Table 4, and Table 5 shows the fluorescencequantum yields ( c $ ~ ) , the reciprocal radiative lifetimes(k1),#' and the activation energies for the singlet decay(ES1-4). These values are sufficient to calculate the rateconstants for product formation, k, and k, (Table 5).TABLE 5Parameters of the first excited singlet state of 3-niethyliso-quinoline N-oxide in various solvents10-7k1/ 10-*k,/ 10-7k,/ EX1-&Solvent 1034, S-1 S-1 S-1 caltVater 3.8 1.07 4-51 <2*3 1410Ethanol 5.2 1.66 4.66 14.4 1070Ethyl acetate 1.1 2.17 6.92 536 2160Cyclohexane 0.60 3-08 7.18 2100 1930Carbon tetra- 0.43 3.14 17.0 3160 CG.0chloridea Subscripts refer t o processes shown in Scheme 2.The change in the photoproduct distribution whenchanging from a polar to a non-polar solvent haspreviously been explained by the interaction of thesolvent with a hypothetical oxaziridine intermediate.lHowever our experimental evidence indicates that theexcited singlet state is the last species common to thetwo photoproducts. The solvent effect is clearly dis-played in the magnitude of k , and k,. These show that,whatever the mechanism might be, the rate of iso-quinolone formation is largely independent of thesolvent, and the solvent effect should be explained interms of a change in the rate of oxazepine formation.Many parameters in this reaction are very similar tothose in the photochemical isomerization of N-benzylideneaniline N-oxide (XI I) to diphenyloxaziridinePhCH=NPh ,% PhCH-NPh t '0'0( X n l (XIII)(XIII).l3 Both are singlet reactions,14 the rates ofproduct formation are of equal magnitude,13 and thequantum yields change in a similar way from polar tonon-polar s01vents.l~ However, whereas it is wellestablished that the primary photoproducts fromnitrones are oxaziridines [ rl12 (2,3-diphenyloxaziridine)ca.2 h in ethanol 15], the present work shows that apossible oxaziridine intermediate from isoquinolineN-oxides must have rl/2 < 4 ps (from the flash-photolysisresults).This is a 109-fold decrease in the half-life, andthe laser experiments indicate a 4 x loll times shorterhalf-life. Such a destabilization of an oxaziridine inter-* Calculated by use of the formula given by Bowen andWokes. l2l2 E. J. Bowen and F. Wokes, ' Fluorescence of Solutions,'Longmans, London, 1953.l8 K. Shinzawa and I. Tanaka, J . Phys. Chenz., 1964, 68, 1205.l4 J. S. Splitter and &I. Calvin, Tetrahedron Letters, 1970, 3905.l6 J. S. Splitter and 11. Calvin, J . Org. Chem., 1965, 30, 3427.mediate in the N-oxide rearrangements is difficult toexplain by steric, or electronic considerations, and itseems likely that the mechanism in the photolysis ofN-oxides is significantly different from that of nitrones.EXPERIMENTAL1.r.spectra were taken with a Perltin-Elmer Infracorcland U.V. spectra with a Beckman DB instrument. Quan-tum yields were measured with a Perkin-Elmer 124 andfluorescence measurements were made with a Perkin-ElmerMPF-2A instrument. N.m.r. spectra were obtained with aVarian A60A instrument. G.1.c. was performed with aVarian Autoprep A 700 machine (SE 30 column). Thespectroscopic flash photolysis apparatus was of standarddesign9 and used with krypton-xenon filled lamps. Thedelay times were from 4 ps-5 s and were reproducible tof l O % or better. In flash experiments N-oxide concen-trations from 5 x 10-4-10-6~ were used. The kinetic laserflash photolysis apparatus has recently been described.laThe N-oxide concentration in these experiments was adjusteduntil the optical density of the solution was 0.3-0.5 a t347 nm in a 1 cm cell.1 - Methylisoquinoline N-Oxide .- 1 -Methylisoquinoline (2.0g ) was heated with acetic anhydride (2 ml) and 35%hydrogen peroxide (6 ml) on a steam bath for 2 h, afterwhich ice and sodium hydroxide were added.The homo-geneous solution was extracted with chloroform, and theorganic phase was dried (MgSO,) and evaporated. Thecrude N-oxide was redissolved in dry ether (100 ml) and thesolution was left in a water-saturated atmosphere at 4" for4 weeks, after which l-methylisoquinoline N-oxide di-hydrate (2.7 g) was isolated, m.p. 28-30" (Found: C, 61-95;H, 6-3; N, 7.3.Calc. for C&13hTO3: C, 61.55; €3, 6-65:N, 7.2%).Steady State Irradiations.-A 0.1 yo solution (w/v) of theN-oxide in ethanol or acetone was irradiated in argon untilthe N-oxide had been consumed. In general, photolysiswas complete in A 6 h for 250 mg samples. Evaporationof the solvent in vaczLo, p.1.c. on silica gel, and crystallizationyielded the products.Identi3cation of Photoproducts.-The isoquinolones (IIa) , l7(IIb),l* and (IIf) l7 are known. Compound (IIc) was pre-pared froin the oxide (Ic) by treatment with acetic an-hydride. Compounds (IId) and (He) were prepared fromthe quinolone (IIa) by N-alkylation with benzyl bromideand l-bromo- l-phenylethane, respectively. The identity ofphotochemical and synthetic isoquinolones was establishedby i.r.spectroscopy and mixed m.p.s.Identi$cation of N-(o-Hydroxy-a-methylstyryZ)formamide(VIIf) .-Hydrolysis of compound (VIIf) in M-hydrochloricacid yielded 2-methylbenzof~ran,~~ ammonia, and formicacid. Hydrolysis in ethanol-water yielded 2-hydroxy-phenylacetone.20 Compound (VIIf) showed vmK (KBr)3220 (NH), 2150, and 1650 (CHO) cm-l, 7 0-16br (lH, s,OH), 0-82br (lH, s, NH), 1.65br (lH, s, CHO), 2-7-3.3(4H, m, ArH), 4-25. (lH, m, HG), and 7.85 (3H, d, J 1 Hz,l6 G. Porter and h'l. R. Topp, Proc. Roy. Soc., 1970, A , 815,163.l7 M. M. Robison and B. L. Robison, J . Org. Chem., 1957, 21,1337.l* B. Elpern and C. S. Hamilton, J . Amer. Chem. Soc., 1946,68, 1436.l9 M. Bisagni, N. P. Buu-Hoi, and R. Royer, J . Chem.SOC.,1955, 3688.*O S. \\'. Tinsley, J . Oug. Chem., 1959, 24, 11971972 233Me).assigned similarly.The structures of compounds (VIIa, b, and d) were2.0*10 T -I/ K -l3.0 4.01-IGURE 3 Temperature dependence of the fluorescence quantumyields of 3-methylisoquinoline N-oxide : a, cyclohexane ;b, ethyl acetate: c, watcr; d, ethanolMeasurements.-The quantum yields were determined at313 nm with a high pressure xenon lamp as source. The21 C . G. Hatchard and C. A. Parker, Proc. Roy. Soc., 1956,-4, W, 518.escitation part of a Perkin-Elmer spectrofluorimeter wasused as monochromator. The half band width was 4 nm.Ferrioxalate was used as actinometer.21 The disappearanceof N-oxide was followed by U.V. spectroscopy. Theappearance of isoquinolone was followed by emissionspectroscopy, by comparing the fluorescence intensity of theirradiated solution with that of a standard solution ofN-oxide and isoquinolone. The reaction was normallyquenched at 10% conversion into photoproduct. Fluores-cence quantum yields were found relative to anthracene.The procedure recommended by Parker was used.22 Fromthe temperature dependence of the fluorescence quantumyield (Figure 3) the activation energy for the singlet decaywas calculated by use of the expression 23 (l/$f - 1) =kexp( -E/RT). Phosphorescence could not be detected ina matrix at 77 I<. Fluorescence lifetimes were measuredwith a pulse sampling fluorescence apparatus. l1I thank Professor G. Porter, F.R.S., for discussions andfor providing the opportunity to work in The Davy FaradayResearch Laboratory, and The Carlsberg Foundation for aWellcome Carlsberg Travelling Research Fellowship.[1/976 Received, June 14118, 1971122 C. A. Parker, ' Photoluminescence of Solutions,' Elsevier,23 E. J. Bowen and J. Sahu, J . Phys. C h e w , 1959, 63, 4.Amsterdam-London-New York, 1968