J. CHEM.SOC. FARADAY TRANS., 1994, Wl), 59-67 Role of Twisted Intramolecular Charge-transfer States in the Decay of 2-(2'-Hydroxyphenyl)benzothiazole following Excited-state Intramolecular Proton Transfer Charles A. S. Potter and Robert G. Brown* Chemistry Department, University of Central Lancashire, Preston, Lancashire, UK PR 1 2HE Friedrich Vollmer and Wolfgang Rettigt lwan-NStranski Institut, Technical University of Berlin, Strasse des 17 Juni 112, D-10623 Berlin 12, Germany The photophysics of 2-(2'-hydroxyphenyl)benzothiazole in non-polar and alcoholic solution is reported for tem- peratures in the range 96-298 K. In all solvents a rise in both fluorescence quantum yield and lifetime is observed as the temperature is decreased. It is proposed that a viscosity-dependent non-radiative process leading to a non-emissive, twisted excited state accounts for these observations.Results of quantum chemical calculations are consistent with this interpretation. 242'-Hydroxyphenyl)benzothiazole(HBT) is an example of a molecular system which undergoes excited-state intramolecu- lar proton transfer (ESIPT) to yield an excited keto form of the original enol. The system has been studied by a number of different workers'-8 and it is clear that depending on the medium in which the HBT is situated, there are three ground-state species which may be important in terms of the absorption properties of the molecule; the neutral enol, anion and cation. All three can also contribute to the excited-state properties of HBT, together with the excited keto form resulting from ESIPT.Scheme 1 shows the various species which can potentially contribute to the photophysics of HBT together with some of the related spectral and pK data. In the solid state at room temperature, only fluorescence from the excited keto species is ~bserved.~.~ This is also the case at low temperature in argon at 12 K6and in methyl- cyclohexane at 77 K.7 Phosphorescence from the keto excited triplet state peaking at 648 nm was also reported in the latter t Present address: W. Nernst Institut fur Phys. und Theor. Chem., Humboldt University of Berlin, Bunsenstr. 1, D-10117 Berlin, Germany. study. In solution at room temperature, the keto fluorescence band is usually observed independent of the medium.Fluo- rescence from the excited anion, enol or cation requires a strongly hydrogen-bonding solvent to disrupt the intramole- cular hydrogen bond so that other processes such as inter- molecular proton transfer and the normal radiative and non-radiative decay processes are able to compete with ESIPT.'*5,8 Under conditions where ESIPT is not disrupted, its kinetics have been found to be extremely rapid with only a lower estimate for the rate constant usually being calculable. In tetrachloroethylene, Laermer et al. have determined a rate constant of cu. 6 x 10l2 for the ESIPT process from the 170 fs risetime of the keto fluores~ence.~ It is a reasonable assumption that similar rate constants for ESIPT will be the case in other non-hydrogen-bonding solvents.It has been suggested that the observed fluorescence from the keto tauto- mer consists of contributions from both the cis and trans con-formations.2,'0 Further evidence to support this proposal are reports of two ground-state transient species with micro- second lifetimes. '**' ' We have been studying the HBT system for a number of c lh ; 510nm I 7 -a;*-0r-p-0;q -0 0 cis -keto trans -keto -9 Scheme 1 J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 perature dependence of HBT fluorescence in a variety of solvent media which we have undertaken in order to explore the possible contribution of twisted intramolecular charge transfer (TICT) states12-14 in the photophysics of HBT.The literature reports of two excited- and ground-state keto tau- tomers suggest that twisting about the 2-1’ carbon-carbon bond may play a significant role. We also report the results of some molecular orbital calculations which model the excited- state energetics of the various species which may contribute to the photophysics in this system. Experimental HBT was prepared and purified as described previously.2v5 All the solvents used here were of spectroscopic grade (mostly Merck uvasol) and were used as supplied. Absorption spectra were measured on Perkin-Elmer Lambda 3 or Cary 14 spec-trophotomers and corrected fluorescence spectra on Perkin- Elmer LS5 or 650/60 spectrofluorimeters. Fluorescence quantum yields were determined relative to quinine sulfate in 0.05 mol dmP3 sulfuric acid (& = 0.5515).The error in the quantum yield values is estimated at f10%. Fluorescence lifetime measurements were undertaken at the BESSY synchrotron radiation source using low-temperature equipment described previously.l6 The time- correlated, single-photon counting technique’ ’ is used to accumulate the data using the single bunch mode of oper- ation of the synchrotron. Excitation wavelengths in the 340- 360 nm region were employed. The decay profiles were analysed by iterative reconvolution and the ‘goodness of fit’ judged on the basis of x2 values and the distribution of residuals. With an apparatus response function of 0.5-0.7 ns, decays could be analysed down to 0.05-0.1 (f0.05) ns.The quantum chemical calculations for the ground state were conducted using the GAUSSIAN88 package” for ab initio calculations on the STO-3G level and for AM1. Energy differences between the ground state and the various excited states were derived by the CNDO/S-CI method according to Del Bene and Jaffe. 9t Results and Discussion The photophysical properties of HBT vary considerably with the medium in which the HBT is situated. In this study we have measured the fluorescence properties of HBT as a func- tion of temperature between 96 and 298 K in a non-polar solvent mixture (methylcyclohexane-2-methylbutane:MCH-2-MB) and in polar, hydrogen-bonding solvents (ethanol, propan-1-01 and mixtures of the two). Although the absorp- tion spectra of HBT in these solvents are very similar (see Fig.1A) and are in agreement with literature parameters:.’ the fluorescence spectra differ considerably between the two classes of solvents. As may be seen from Fig. lB, HBT in the alcohols exhibits two emission bands which can be attributed to the enol (E, short-wavelength band) and the keto species (K, long-wavelength band). Only the K band is observed in the aprotic non-polar solvent MCH-2-MB. We have pre- viously reported three emission bands for HBT in aqueous solution depending on the PH.~ The additional fluorescence band in water originates from the HBT anion and, in wave- length terms, lies between the two bands observed in the alcohols. In other polar, non-hydrogen-bonding solvents such ~ t Program QCPE no. 333 from Quantum Chemistry Programme Exchange, Bloomington, Indiana, was used with the original param- eters, 50 singly excited configurations were used for configuration interaction.0.8 4 wavelengt h/nm 100 B 300 350 400 450 500 550 600 E wavelength/nm Fig. 1 A, Absorption and B, fluorescence spectra of HBT in MCH- 2-MB (a)and ethanol (b) (concentration 2.0 x mol dm-’) as acetonitrile, a single, long-wavelength fluorescence band attributable to K is observed as in non-polar Solvents. The origins of the three fluorescence bands observable in the various solvents are summarised in Scheme 1. The intensity of HBT fluorescence at room temperature in the solvent used here is weak.As Tables 1 and 2 show, fluo- rescence quantum yields in the range 0.01-0.05 are found. As the temperature is decreased, the fluorescence quantum yields uniformly increase such that at least an order of magnitude increase over room temperature is observed at 96 K. In the alcohol solvents it is noticeable that, as the temperature is lowered, there is a greater increase in the amount of keto tautomer fluorescence relative to the enol such that the ratio ~enoJ4ke,ofalls from a value in the region of 10 at room tem- perature to a value closer to unity at 96 K. These quantum yield changes are depicted in Fig. 2 for the pure ethanol solvent which typifies the quantum yield behaviour in alco- hols as a function of temperature. At room temperature, the fluorescence from both the enol and keto species is weak and causes considerable scatter in the values of the ratio 4enoJ4keto(Table 2).We are therefore not convinced that there is a real solvent effect on this ratio at room tem-perature. At reduced temperature, where the quantum yield values are higher, the variation of 4enoJ4kctowith the com- position of the alcohol solvent may be significant. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 1 Fluorescence quantum yield and decay data for HBT in MCH-2-MB as a function of temperature i/nm x2 4f 1.Ooo 0.11 --1.15 0.1 1 0.095 8.9 298 {500 0.01 1 540 1.Ooo 0.08 --1.26 0.08 0.141 13.2 223 -------0.062 --0.210 0.98 0.790 2.51 1.02 2.19 0.089 0.37 173 {500 0.200 520 1.Ooo 2.60 --1.29 2.60 0.075 0.3 1 0.278 1.16 0.722 3.74 0.92 3.03 0.095 0.24 0.290 1.Ooo 5.01 --1.29 5.01 0.057 0.14 1.Ooo 5.30 --0.95 5.30 0.060 0.13 0.320 96 1.Ooo 5.16 --1.14 5.16 0.06 1 0.13 We have also measured fluorescence decay profiles for calculated by eqn.(1): HBT in these solvents over the same temperature range as the fluorescence spectra and quantum yields. The data (7) = xui 7Jxai (1) obtained are presented in Tables 1, 3 and 4 and some typical Note that all of the fluorescence decay profiles exhibit a decay profiles are shown in Fig. 3. Corresponding emission prompt rise. There is no evidence of a risetime slower than cu. spectra are shown in Fig. 4. For the most part the decays are 50 ps in any of the profiles in the different solvents or at the not monoexponential except for some of the keto fluores- different temperatures.These observations are clearly in cence profiles at the lowest temperature used. Here the decay accord with the rapid ESIPT step obtained by Laermer et time is very similar to that found for solid HBT at room ~1 as far as the keto fluorescence is concerned. They also .~ temperature3 although the keto fluorescence yield is still only suggest that the enol fluorescence observed in the alcohols is a fraction of the fluorescence quantum yield for the solid. either excited directly from the ground state (i.e.excitation of The majority of the fluorescence decays require a sum of an HBT molecule intermolecularly hydrogen bonded to two exponential components to fit the data adequately.In a solvent molecules) or is produced from the intramolecularly few cases, three exponential components are required to give hydrogen-bonded enol tautomer of HBT which is the precur- a x2 value < 1.2 and a random distribution of residuals. The sor of the keto tautomer. As the excitation spectra for enol use of three exponential components in the analysis of the and keto fluorescence are identical, there is no obvious spec- data does not tend to alter the mean lifetime (T) which is tral evidence for the presence of both intra- and inter-Table 2 Fluorescence quantum yields for the enol(4,J and keto (4K)forms of HBT in alcohol solvents at various temperatures solvent temperature/K 4cotal 4E 4K 4J~K 100% EtOH 298 0.021 0.019 0.002 8.8 223 0.048 0.042 0.006 7.6 173 0.102 0.090 0.012 7.7 123 0.148 0.125 0.023 5.3 96 0.161 0.128 0.033 3.9 75% EtOH-25% PrOH 298 0.02 5 0.022 0.003 8.3 223 0.067 0.06 1 0.006 10.2 173 0.150 0.136 0.014 10.0 123 0.21 3 0.180 0.033 5.5 96 0.230 0.177 0.053 3.4 50% EtOH-50% PrOH 298 0.044 0.034 0.010 3.4 223 0.078 0.065 0.013 5.1 173 0.136 0.109 0.027 4.0 123 0.227 0.172 0.055 3.1 -96 25% EtOH-75% PrOH 298 0.012 0.010 0.002 6.1 223 0.068 0.055 0.013 4.3 173 0.154 0.120 0.034 3.6 123 0.257 0.180 0.077 2.4 96 0.280 0.173 0.107 1.6 100% PrOH 298 0.010 0.008 0.002 4.7 223 0.029 0.024 0.005 4.5 173 0.064 0.050 0.014 3.4 123 0.088 0.059 0.029 2.0 96 0.093 0.056 0.037 1.5 50 100 150 200 250 300 temperature/K Fig.2 Variation of fluorescence quantum yields and lifetimes for the enol and keto forms of HBT in 100% ethanol as a function of temperature:(a)enol lifetime, (0)keto lifetime; (H)enol yield, (A)keto yield 10 1 10000 g10 000 3 2 1000 2 1000 v) Q)c. 100 .E 100 Q) c a 10 5: 10 E -2 ’c Fig. 3 Fluorescence decay profiles for the enol (a) and keto (b)forms of HBT in ethanol as a function of temperature J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 molecular hydrogen-bonded ground-state forms of HBT in alcohols. However, the rapidity of ESIPT does raise the ques- tion as to whether breaking of the intramolecular hydrogen bond to yield the intermolecularly hydrogen-bonded species (and hence enol fluorescence) can compete efficiently with ESIPT.The data are more compatible with a scheme involv- ing ground-state solvates (Scheme 2) which differ only mar- ginally in their absorption properties. Eointer solvation EOinra -KO Scheme 2 Although both the enol and keto fluorescence decay pro- files are multiexponential under most of the conditions studied, there is no evidence from the analysis of the decays for a kinetic relationship between the two excited species once they are formed. There appears to be no relationship between the decay parameters for the enol and keto fluores- cence from HBT in a given alcohol solvent at a given tem- perature. Reverse ESIPT in the excited keto tautomer is therefore not competitive with its radiative and non-radiative decay routes.Neither is reformation of the intramolecular hydrogen bond in the HBT enol with intermolecular hydro- gen bonds to solvent molecules competitive with fluorescence and non-radiative decay in this species. The two fluorescent species appear to be kinetically isolated from one another. The origin of the multi-exponential decay profiles is there-fore not immediately apparent. In the absence of any obvious risetime components and with the two fluorescent species kinetically isolated, the multi-exponential decays must be an inherent property of the two species. In the alcohol solvents where a range of solvated species could be anticipated to exist, the observation of multi-exponential decays which reflect a range of excited species with a distribution of life- times is not unreasonable and echoes conclusions some of us have drawn elsewhere.*’ A possible kinetic mechanism accounting for the data observed in alcohols is given in Scheme 2.Here, the indices inter and intra refer to differently solvated HBT species, with Eintrabeing the intramolecularly hydrogen-bonded enol tautomer prepared for the ESIPT process, and Einterthe HBT species intermolecularly hydro- gen bonded to other alcohol molecules. This can happen in a variety of fashions, hence a distribution of rate constants k: and k:, (see below) is expected if excited-state equilibration is not fast enough. Given that the fluorescence quantum yield will present an ‘averaged’ view of the emission efficiencies of this range of excited species and is effectively the integrated intensity under the decay curve, we feel that it is reasonable to use the (z) values from eqn. (1) in our later analysis.The biexponential decays, even though many are only slightly biexponential, observed in MCH-ZMB are not as easy to explain in these non-interacting solvents. Once again (7) values will be used in our further analysis of the data. Combination of the measured yields with the (z) values gives radiative (k,) and non-radiative (k,,,) rate constants for the excited enol (k:, k:,) and the keto (kr, kf,) tautomers. These values are shown in Tables 1 and 5. In MCH-2-MB, kr is independent of temperature whereas kfr decreases sharply with decreasing temperature.A temperature-dependent non-radiative process is indicated which appears to be almost completely ‘frozen out’ at 96 K. We can there- fore calculate the temperature-dependent part of the non- J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 3 Fluorescence decay parameters for the enol and keto forms of HBT in alcohol solvents at various temperatures solvent Alnm temperature/K a1 r,/ns a2 t,/ns X2 (t>/ns 100% EtOH 380 298 0.998 0.39 0.002 4.50 1.22 0.40 223 0.771 0.47 0.229 0.77 1.65 0.54 173 0.908 0.64 0.092 1.21 1.30 0.69 123 0.877 0.64 0.123 1.32 1.26 0.72 96 0.860 0.73 0.140 1.59 1.08 0.85 510 298 0.988 0.14 0.012 3.00 1.71 0.18 223 0.982 0.24 0.018 1.86 1.63 0.27 173 0.67 1 0.73 0.329 1.65 1.14 1.03 123 1.Ooo 5.13 - - 1.14 5.13 96 1.Ooo 5.71 - - 1.01 5.71 75% EtOH-25Oh PrOH 380 298 0.999 0.40 0.00 1 3.65 1.43 0.40 223 0.968 0.55 0.032 1.38 1.67 0.58 173 0.935 0.64 0.065 1.35 1.05 0.68 123 0.889 0.64 0.111 1.36 1.35 0.72 96 0.882 0.67 0.118 1.39 1.17 0.76 510 298 0.990 0.16 0.010 2.97 2.25 0.19 223 0.98 1 0.29 0.019 2.04 1.32 0.32 173 0.600 0.96 0.400 2.04 1.15 1.40 123 0.143 2.97 0.857 5.66 1.01 5.27 96 1.Ooo 5.70 - - 1.17 5.70 50% EtOH-50% PrOH 380 29 8 0.999 0.40 0.001 2.68 1.29 0.40 223 0.890 0.50 0.1 10 0.92 1.34 0.55 173 0.819 0.56 0.181 1.03 1.20 0.66 123 0.868 0.60 0.132 1.23 1.17 0.68 96 0.854 0.63 0.146 1.27 1.10 0.72 510 29 8 0.994 0.13 0.006 3.1 1 1.32 0.15 223 0.986 0.34 0.014 1.98 1.24 0.36 173 0.602 1.25 0.398 2.40 0.9 1 1.71 123 1.Ooo 5.39 - - 1.22 5.39 96 1.Ooo 5.77 - - 1.14 5.77 25% EtOH-75% PrOH 380 298 0.999 0.39 0.00 1 3.98 1.36 0.40 223 0.969 0.53 0.03 1 1.32 1.48 0.56 173 0.986 0.57 0.014 1.14 1.23 0.64 123 0.884 0.6 1 0.116 1.29 1.15 0.69 96 0.886 0.65 0.1 14 1.37 1.04 0.74 510 298 0.990 0.14 0.010 3.49 1.32 0.17 223 0.984 0.42 0.016 2.04 1.42 0.44 173 0.49 1 1.56 0.509 2.75 0.93 2.17 96 1.Ooo 5.76 - - 1.18 5.76 100% PrOH 380 298 1.OOo 0.38 - - 0.63 0.38 173 0.735 0.47 0.265 0.94 1.20 0.60 96 0.796 0.57 0.204 1.23 1.29 0.70 510 298 0.990 0.11 0.010 2.78 1.25 0.13 173 0.75 1 2.29 0.249 4.04 1.30 2.72 123 0.182 0.53 0.818 5.50 1.17 4.60 96 0.070 1.89 0.930 5.83 2.47 5.55 radiative rate [k;,(T)]from eqn.(2): the range (9.2-11.7) f2.0 kJ mol-'. However, given that the lifetime values at room temperature can be considered as 1 1 only an upper limit owing to the instrumental responsekf,(T)= ---profile, the true activation energy values will be somewhat r(T) ~(96K) higher than these and will be similar to the activation energy An Arrhenius plot of log kf,(T)0s. 1/T yields an activation in MCH-2-MB. The corresponding rate constants for the energy of 12.2 f2.0 kJ mol-'.enol fluorescence in alcohols exhibit much smaller changes In the alcohol solvents the same observations as in MCH-with temperature. 2-MB may be made with respect to k: and k;,. The values of Most interestingly, there is a reasonably clear trend for k:, kr are more scattered than those calculated for the non-polar which increases by a factor of ca. 5 between 298 and 96 K, solvent but there is no obvious trend to the values. We attrib- whereas kfi, decreases by a factor of only ca. 3 between these ute the scatter to the low fluorescence quantum yield of the two temperatures (Table 5). The temperature dependence of keto species at all temperatures in the alcohols coupled with k: as well as the non-exponential nature of the decay, sup- short fluorescence lifetimes at the higher temperatures.These ports the above view of a number of emitting states, Eintcr properties, together with the presence of the two ground-state (incapable of ESIPT owing to their intermolecular solvation), enols, introduce sufficient uncertainty into the kr values that with different individual k: and k:, values, which may depend we have to conclude that kr is probably invariant with tem- differently on temperature, The temperature dependence of perature in all the solvents studied. The slopes of the Arrhe- the /cEr values transforms into changes of the relative popu- nius plots are all quite similar and give activation energies in lations of the different Einte, species and therefore into a 64 J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 4 Fluorescence decay parameters for the HBT keto form in alcohol solvents at various temperatures and emission wavelengths solvent temperature/K I/nm a1 '51/ns a2 Tz/ns xz ('5>/ns 100% EtOH 298 510 0.988 0.14 0.012 3.00 1.71 0.18 520 0.993 0.13 0.007 3.00 1.60 0.14 530 0.995 0.1 1 0.005 3.01 1.55 0.12 173 510 0.67 1 0.73 0.329 1.65 1.14 1.03 520 0.652 0.73 0.348 1.60 1.22 1.03 530 0.727 0.79 0.273 1.69 1.23 1.04 96 510 1.Ooo 5.7 1 1.01 5.71 520 1.Ooo 5.72 1.22 5.72 530 1.ooo 5.73 0.94 5.73 75% EtOH-25% PrOH 298 510 0.990 0.16 0.010 2.97 2.25 0.19 520 0.993 0.14 0.007 2.86 1.62 0.16 530 0.99 1 0.19 0.009 2.95 1.37 0.21 50% EtOH-50% PrOH 298 510 0.994 0.13 0.006 3.1 1 1.32 0.15 520 0.997 0.1 1 0.003 3.14 1.39 0.12 530 0.998 0.010 0.002 3.27 1.34 0.1 1 100% PrOH 96 510 0.070 1.89 0.930 5.83 2.47 5.55 520 1.Ooo 5.58 3.00 5.58 530 1.Ooo 5.59 3.16 5.59 change of the effective kfi,, which is an amalgam of the state and in the non-polar solvent at 96 K.This assumption weighted contributions of the individual Einterspecies. is borne out by the fluorescence lifetimes of 5.0-6.0 ns mea- Although the data are insufficient to draw any detailed con- sured at 96 K in the various alcohol solvents. The reduced & clusions on these distributions the above observations clearly values (Table 2) in these solvents may be attributed to the rule out the possibility that there is a single emitting Einter absorbed light being distributed between Einlcrand Eintra species in alcohols, whereas in hydrocarbon solvents the clas- species such that only a portion actually leads to the excited sical two-state scheme (Scheme 2 without the contribution of keto species.Since the calculation of &assumes that all of Einler)seems to be appropriate. the absorbed light contributes to the formation of the keto Note that, with the exception of the keto fluorescence yield species, these quantum yield values will automatically be in alcohols at 96 K, the photophysical properties of the HBT reduced from their true values which would be expected to lie keto form in frozen solution at 96 K and as a solid at room in the region of 0.3. temperature are virtually identical. It is therefore reasonable This allows us to suggest that ca.10% of the HBT mol- to assume that in the alcohol solvents the excited keto species ecules are intramolecularly hydrogen bonded in 100% will have similar properties to those observed in the solid ethanol [&(observed) :&(expected) =0.033 :0.31 and 100% Table 5 Radiative and non-radiative rate constants for the enol and keto forms of HBT as a function of temperature in ethanol-propanol mixtures derived from fluorescence quantum yields and mean lifetimes enol tautomer keto tautomer solvent T/K & rJns k,E/10-9 s-l k310-9 s-1 $J~ 7Jns kr/10-9 s-' kfr/10-9 s-l 100% EtOH 298 0.019 0.40 0.049 2.48 0.002 0.18 0.013 5.70 223 0.042 0.54 0.079 1.77 0.006 0.27 0.02 1 3.70 173 0.090 0.69 0.130 1.31 0.012 1.03 0.01 1 0.96 123 0.125 0.72 0.173 1.22 0.023 5.13 0.005 0.19 96 0.128 0.85 0.151 1.03 0.033 5.71 0.006 0.17 75% EtOH-250/, PrOH 298 223 0.022 0.06 1 0.40 0.58 0.056 0.106 2.43 1.62 0.003 0.006 0.19 0.32 0.014 0.019 5.3 1 3.11 173 0.136 0.68 0.199 1.26 0.014 1.40 0.010 0.71 123 0.180 0.72 0.252 1.15 0.033 5.27 0.006 0.18 96 0.177 0.76 0.235 1.09 0.053 5.70 0.009 0.17 50% EtOH-50% PrOH 298 0.034 0.40 0.086 2.42 0.010 0.15 0.069 6.69 223 0.065 0.55 0.1 19 1.70 0.013 0.36 0.036 2.72 173 0.109 0.65 0.169 1.38 0.027 1.71 0.016 0.57 123 0.172 0.68 0.252 1.22 0.055 5.39 0.010 0.18 96 0.72 5.77 25% EtOH-75% PrOH 298 0.010 0.40 0.026 2.49 0.002 0.17 0.010 5.87 223 0.055 0.56 0.100 1.70 0.013 0.44 0.029 2.24 173 0.120 0.64 0.188 1.37 0.034 2.17 0.01 5 0.45 123 0.180 0.69 0.262 1.19 0.077 - - - 96 0.173 0.74 0.235 1.12 0.107 5.76 0.019 0.16 10Oo/;, PrOH 298 0.008 0.38 0.022 2.61 0.002 0.13 0.014 7.62 223 0.024 - - - 0.005 - - - 173 0.050 0.60 0.083 1.59 0.014 2.72 0.005 0.36 123 0.059 - - - 0.029 4.60 0.006 0.2 1 96 0.056 0.57 0.099 1.66 0.037 5.55 0.007 0.17 J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Fig. 4 Fluorescence spectra for the HBT keto form as a function of temperature in (a) MCH-ZMB and (b)ethanol. The spectra are all normalised to the same value to allow the spectral shapes to be com- pared propanol [&(observed) : &(expected) = 0.037 : 0.31, whereas the proportion is apparently greater in the mixed ethanol- propanol solvents reaching a maximum of some 35% in 25% ethanol-75% propanol (quantum yield ratio 0.107 : 0.3).For the keto fluorescence in alcohols, the radiative rate, k:, depends much less on temperature (Table 5), roughly consis- tent with emission from a single intramolecularly hydrogen- bonded species. However, in view of the observed slight non-exponentialities of the decays, different Kintrasolvates may be present here too, involving some intermolecular hydrogen bonding in addition to the intramolecular one of main importance. On the other hand, the non-radiative process (kfr) competing with the keto fluorescence depends much more strongly on temperature than kr or the rate con- stants of the enol species.The temperature dependence of ktr could be due to a viscosity-dependent non-radiative process available to the excited keto form. This may involve twisting of the 2-1’ bond connecting the aromatic systems as pro- posed for 3-hydroxy-2,2’-bipyridyI2’to lead to a non-emissive, lower-energy TICT state. The proposal of a viscosity-dependent non-radiative decay process for K is further substantiated by the results presented in Table 5, where different mixtures of ethanol and propanol are com- pared at the same temperature. Fig. 5 depicts the data for 173 K where the lifetime changes are most clear cut and signifi- cant.For the keto tautomer, the lifetime lengthens by a factor of more than two when changing from the less viscous ethanol to the more viscous propanol, and the derived kfr decreases. For the enol tautomer, these effects are marginal and in the opposite sense. The activation energy associated with the temperature dependence of k:r in ethanol is some- what lower than that for the solvent viscosity, consistent with the Kramers theory that the non-radiative process is intrinsi- cally barrierless as typified by other TICT molecules.22 Quantum chemical calculations can help to elucidate the question regarding the nature of the viscosity-dependent non- radiative decay channel of K. As mentioned above, one possi- bility is the involvement of a TICT-like specie^'^-'^ which is non-emissive in nature.As a formal double bond is involved in the twisting of the K species (cis-keto in Scheme l), this process can be described within the general theory of biradicaloid which, for example, can interpret the simultaneous twisting of the double and single bonds in sub- stituted ~tilbenes.~’,~~Such a twisting process towards a biradicaloid state A* with twisted geometry is possible in the excited state if the twisted product is less energetic than the planar precursor. The same argument holds for the ESIPT process in that the observation of K fluorescence tells us immediately that the energy of the K* state has to be less than that of the E* precursor. To decide whether a possible TICT state A* could be lower in energy than the other states E* and K* with planar geometry, a combination of quantum chemical methods for ground (ab initio STO-3G and AM1) and excited states (CNDO/S-CI) was used. To start with, the ESIPT reaction was modelled.In a first step, the ground-state geometry of E and K was optimized to two levels, starting with the X-ray structure of E determined by Sten~on.~’ At the first levels (a), the six most relevant bonds and bond angles which change in the ESIPT process were optimized, and at the second level (b),seven more geo- metric parameters were optimized, as in the formula in Table 6. In order to obtain a stable energetic minimum for K, the N-H bond length, r3, had to be fixed to an assumed value of 100 pm.If r3 was left free, the molecule relaxed to the E structure in the ab initio calculation. This behaviour indicates 3 1.6 ‘K1.4 -1.2 0 25 50 75 100 propan-2-01(Oh) Fig. 5 Variation of mean lifetime and derived non-radiative rate constant of the enol and keto forms of HBT as a function of viscosity in ethanol-propanol mixtures at 173 K; (0)enol lifetime, (0)keto lifetime; (B)kfi, ,(A)kfr J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 6 Structural parameters for the optimized geometries of the enol and keto forms of HBTin the ground state' (a) Enol form HI3 0 -+r2 ~~ X-ray structureb 148.13 130.48 -128.07 140.42 137.20 175.73 123.08 124.09 -110.74 115.39 109.21 {::I1;;{ 149.28 136.71 99.68 ----121.52 124.95 104.85 ---149.13 136.65 99.71 130.55 143.18 140.51 174.60 121.20 125.08 105.00 110.08 114.19 110.18 146.89 135.92 97.23 ----127.24 129.13 110.81 --146.07 135.87 97.27 132.02 141.38 142.35 168.44 127.55 129.10 110.86 110.66 114.03 1&59 (b) Keto form STO-3G (a) (b) 142.56 140.98 128.97 128.03 100.00 (fixed) 100.00 -135.26 -141.24 -140.62 -176.33 115.55 116.71 119.88 121.10 138.21 136.40 -113.99 -111.72 -111.20 (fixed) AM1 (a) (b){ 142.57 140.58 126.69 126.38 (fixed) 100.00 100.00 -137.12 -140.71 -142.01 -169.47 127.17 128.06 123.90 124.05 123.97 123.97 -113.38 -112.52 -110.13 (fixed) Bond lengths in pm, angles in degrees.Ref. 28. that on the ground-state hypersurface, there is no energetic barrier between the E and K structures.The geometric and energetic results are collected in Table 6 and Fig. 6 and 7. The results for the optimized E structure closely corre-spond to the experimental X-ray determination, with some difference only for the C-0 bond length, r2. In the opti- mized K structure, the main geometric changes are the short- ening of the inter-ring C-C bond, the shortening of the C-0 bond, and a shrinkage of the angle wl, which leads to an approach of the N and 0 atoms. The optimized structures for E and K are shown in Fig. 6. Owing to the approach of N and 0 in the structure for the keto species, the hydrogen atom is not only close to N but also close to the oxygen atom, thus facilitating ESIPT.The ESIPT process can be viewed as initiated and accompanied by an in-plane molecu- lar vibration modifying the hydrogen-tunnelling distance. A similar conclusion was recently reached regarding the related molecule 2-(2'-hydroxyphenyl)benzoxazole (HB0).28 In a second step, the energy differences between ground and excited states were calculated for the various optimized ab initio (STO-3G) ENOL _______ KETO structures by CNDO/S-CI (Fig. 7). With the geometries for ab initio STO-3G, the first excited state is of nn* nature, and the first allowed nn* state is situated some 0.6 eV above. This is clearly in contradiction to the experimental results with the rather high fluorescence quantum yields at low temperature indicating that S, is an allowed state.Furthermore, the energy of K* is calculated as being greater than that of E*, which is also contrary to the experimental observation of K fluorescence, indicating an exothermic E* +K* reaction. This is mainly due to a rather high energy for reaching K in the ground state which diminishes considerably within the AM1 framework so that the excited-state reaction E* +K* xx*nn* T nx* . . . . - - - - . . . . . . . . to.2 Fig. 7 Ground-and excited-state energies (eV) for the ESIPT Fig. 6 Structural representation of the optimized geometries of the process in HBT as derived from ab initio (left) and AM1 (right) calcu- enol (-) and keto (---) tautomers in the ground state according lations at optimization levels (a) and (b) (see text for details), com- to the ab initio (STO-3G)calculations [optimization level (a)] bined with CNDO/S-CI calculations J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 67 E* K* A' support and BESSY for access to beamtime via the EC 10.1 Large-scale Installations Programme (GE 1-00 18-DCB). I I4.1 3.2 K A Fig. 8 Ground- and excited-state energies (eV) of precursor and products on the excited-state hypersurface of HBT. The results are derived from a combination of the methods AM1 and CNDO/S and involve partial geometrical optimization. Whereas the products are highly endothermic on the ground-state surface, they are exothermic on the excited-state surface, and the photoreaction E* -,K* -+ A* can take place becomes slightly exothermic.Moreover, with the AM1 opti- mized geometries, the nz* state for K* is not situated below but approximately isoenergetic with the lowest m* state. In summary, the results in Fig. 7 indicate that the E* +K* reaction can be modelled qualitatively to be consistent with experiment by a combination of AM1 and CNDO/S. Ab initio ground-state calculations at the optimization levels and with the basis set STO-3G used here seem to overestimate strongly the energy difference between E and K. In a third step, the TICT reaction was modelled by twist- ing around the inter-ring bond using the K* structure from AM 1, optimization level (a). The ground state increases further in energy, but the energy difference between ground state and S, (of TICT nature, with electron transfer from the benzothiazole to the ketonic moiety) becomes rather small, less than half that in E*, so that the reaction from K* to the TICT state A* becomes exothermic (Fig.8). These results support the involvement of a TICT state responsible for the viscosity-dependent non-radiative decay channel of HBT. The fluorescence quenching process of HBT can thus be viewed as the result of a two-step photochemical reaction E* -+K*+A* with the first step being in the femtosecond range and the second step ranging from nanoseconds at low temperature to some tens of picoseconds at room tem-perature and being strongly viscosity dependent owing to the involvement of a large-amplitude molecular relaxation (intramolecular twisting process).We thank DAAD, the British Council, the Ciba-Geigy Trust, NATO and the BMFT (project 05 414 FAB1) for financial References 1 M. D. Cohen and S. Flavian J. Chem. SOC. B, 1967, 317; 321; 329; 334. 2 D. L. Williams and A. Heller, J. Phys. Chem., 1970,74, 4473. 3 K. Anthony, R. G. Brown, J. D. Hepworth, K. W. Hodgson, B. May and M. West, J. Chem. SOC., Perkin Trans. 2, 1984,2111. 4 R. S. Becker, C. Lenoble and A. Zein, J. Phys. Chem., 1987, 91, 3509. 5 C. A. S. Potter and R. G. Brown, Chem. Phys. Lett., 1988,153, 7. 6 K. Ding, S. J. Courtney, A. J. Strandjord, S. Flom, D. Friedrich and P. F. Barbara, J. Phys. Chem., 1983,87, 1184. 7 P-T. Chou, S. L. Studer and M. L. Martinez, Chem. Phys. Lett., 1991,178, 393.8 T. Elsaesser and B. Schmetzer, Chem. Phys. Lett., 1987, 140, 293. 9 F. Laermer, T. Elsaesser and W. Kaiser, Chem. Phys. Lett., 1988, 148, 119. 10 M. Itoh and Y. Fujiwara, J. Am. Chem. SOC., 1985, 107, 1561. 11 W. E. Brewer, M. L. Martinez and P-T. Chou, J. Phys. Chem., 1990,94, 1915. 12 Z. R. Grabowski, K. Rotkiewicz, A. Siemiarczuk, D. J. Cowley and W. Baumann, Nouv. J. Chim., 1975,3,443. 13 W. Rettig, Angew, Chem., Int. Ed. Engl., 1986, 25,971. 14 M. Van der Auweraer, Z. R. Grabowski and W. Rettig, J. Phys. Chem., 1991,95,2083. 15 S. R. Meech and D. Phillips, J. Photochem., 1983,23, 193. 16 M. Vogel and W. Rettig, Ber Bunsenges. Phys. Chem., 1987, 91, 1241. 17 D. V. O'Connor and D. Phillips, Time-correlated Single Photon Counting, Academic Press, London, 1984.18 M. J. Frisch, M. Head-Gordon, H. B. Schlegel, R. Raghavachari, J. S. Binkley, C. Gonzales, D. J. Defrees, D. J. Fox, R. A. White- side, R. Seeger, C. F. Melius, J. Baker, R. Martin, L. R. Kahn, J. J. P. Stewart, E. M. Fluder, S. Topiol and J. A. Pole, GAUSSIAN88, Gaussian Inc. Pittsburgh, PA, 1988. 19 J. Del Bene and H. H. Jaffe, J. Chem. Phys., 1968, 48, 1807; 4050; 49, 1221 ;1969,50, 1126. 20 J. A. T. Revill and R. G. Brown, Chem. Phys. Lett., 1992, 188, 433. 21 F. Vollmer and W. Rettig, to be published. 22 E. Lippert, W. Rettig, V. BonaEic-Koutecky, F. Heisel and J. A. Miehe, Ado. Chem. Phys., 1987,68, 1. 23 J. Michl and V. BonaEic-Koutecky, Electronic Aspects of Organic Photochemistry, Wiley, New York, 1990. 24 V. BonaEic-Koutecky, J. Koutecky and J. Michl, Angew. Chem., 1987,99,216, Angew. Chem., Int. Ed. Engl., 1987,26, 170. 25 W. Rettig, W. Majenz, R. Lapouyade and G. Haucke, J. Photo-chem. Photobiol A: Chem. 1992,62,415. 26 R. Lapouyade, K. Czeschka, W. Majenz, W. Rettig, E. Gilabert and C. Rulliere, J. Phys. Chem., 1992, 96,9643. 27 P. Stenson, Acta Chem. Scand., 1970,24,3729. 28 N. P. Ernsting, T. Arthen-Engeland, M. A. Rodriguez and W. Thiel, J. Chem. Phys., 1992, 97, 3914; T. Arthen-Engeland, T. Bultmann, N. P. Ernsting, M. A. Rodriguez and W. Thiel, Chem. Phys., 1992, 163, 43. Paper 31027626; Received 17th May, 1993