J. Phys. Chem. B 1997, 101, 10645-10652
10645
Femtosecond Fluorescence Study of Proton-Transfer Process in Thermochromic Crystalline Salicylideneanilines Taro Sekikawa and Takayoshi Kobayashi* Department of Physics, UniVersity of Tokyo, 7-3-1, Hongo, Bunkyo-ku, Tokyo 113, Japan
Tamotsu Inabe Department of Chemistry, Hokkaido UniVersity, 10-8, Sapporo 060, Japan ReceiVed: May 14, 1997; In Final Form: September 3, 1997X
In order to clarify a proton-transfer mechanism, the dependence of proton-transfer rate on the hydrogen-bond length was investigated. The proton-transfer rate in four thermochromic N-salicylideneanilines (SAs) in a crystalline phase was estimated by a femtosecond time-resolved fluorescence spectroscopy. The transfer rate is given by k0 exp(-Rδr) (R ) 9.0 Å-1) as a function of the tunneling distance δr, where k0 is the O-H stretching frequency. The proton-transfer rate is insensitive to a sample temperature and shows a deuteration effect. From these results, it is concluded that the photoinduced proton transfer in the excited state of SAs takes place by the quantum mechanical tunneling.
1. Introduction Intramolecular “proton transfer”, including neutral hydrogenatom motion and zwitterion formation, in the electronically excited state is one of the simplest photochemical reaction and is related to some biological processes. In addition, the proton transfer might be used for optical memory or switch utilizing large spectral changes induced by a π-electron configurational change. Therefore, the process of proton transfer has been attracting many researchers and has been studied in a large number of compounds using the time-resolved spectroscopy.1-23 The development of femtosecond lasers has enabled to trace the dynamical process of proton transfer.8,10,12-23 According to the previous studies, there is no barrier in the energy surface of proton transfer.15-17,19,22 These studies, however, have been performed in a liquid or gas phase. The hydrogen-bond length in the liquid phase might be dependent on solvent even when a solvent-solute interaction is relatively weak as in a nonpolar solvent. In addition, the viscosity of solvent influences the proton-transfer dynamics as shown in many previous studies.7,14,24-27 In solution, partial moieties in a hydrogenbonded system may not be coplanar to each other in the ground and/or excited states, but there is no such information available. In crystalline phase, the structures of some of the hydrogenbonded system have been determined and offer useful information to the discussion of the proton transfer mechanism. Therefore, the investigation of proton transfer in a crystalline phase, where the positions of the molecules are fixed within the thermal fluctuation amplitude by the intermolecular interaction, is expected to reveal the intrinsic dynamics of proton in the hydrogen bond. In order to discuss the mechanism in relation with the tunneling, it is important to study the dependence of the transfer rate on the transfer distance in the crystalline phase. The transfer or tunneling distance would be proportional to the hydrogenbond length. If the proton tunnels through the potential barrier, the transfer rate would depend on the hydrogen-bond length sensitively. Hence, we report the dependence of the femtosecond dynamics of proton transfer in the crystalline phase on X
Abstract published in AdVance ACS Abstracts, November 15, 1997.
S1089-5647(97)01588-5 CCC: $14.00
Figure 1. (a) Energy diagram of a thermochromic SA and (b) the constitutional formulas of the SA derivatives studied: (1) Cl4SPY, (2) DNP, (3) ClSA, and (4) BSP.
the hydrogen-bond length using the time-resolved fluorescence spectroscopy.28 For such a purpose, four thermochromic N-salicylideneanilines (SAs) shown in Figure 1 in a crystalline phase are picked among the many systems with an intramolecular hydrogen bond,29-40 because the intramolecular hydrogen-bond lengths in the SAs are known from an X-ray structural analysis.38-40 The energy diagram of the thermochromic SA suggested from the X-ray analysis,33,34 stationary absorption and fluorescence spectroscopies,29-32 and NMR study35,36 is schematically shown in part a of Figure 1. In the ground state, the enol form is more stable than the keto form. The energy separation between these tautomers is small enough for the keto form to be populated thermally. Thus, the absorption spectrum changes drastically with temperature. On the other hand, the proton moves spontaneously to the opposite side in the hydrogen © 1997 American Chemical Society
10646 J. Phys. Chem. B, Vol. 101, No. 50, 1997 TABLE 1: Hydrogen-Bond Lengths
a
compound
hydrogen-bond length (Å)
Cl4SPYa DNPb ClSAc BSPd
2.534(5) 2.537(3) 2.584 2.607(2)
Reference 39. b Reference 40. c Reference 33. d Reference 38.
bond very rapidly in the excited state, which is the origin of the large Stokes shift of the fluorescence. Therefore, the thermochromic SA is considered to perform a closed reaction cycle with enol-keto-type proton transfer by photoexcitation. The lifetime of the excited enol form is expected to depend on the proton-transfer time, which may be estimated by the fluorescence lifetime. In the present study, the proton-transfer rates in the following four thermochromic SAs were estimated by the femtosecond fluorescence spectroscopy: (l) N-tetrachlorosalicylidene-1pyrenylamine (Cl4SPY),39 (2) N,N′-bis(2-hydroxy-1-naphthylmethylene)-p-phenylenediamine (DNP),40 (3) N-5′-chlorosalicylideneaniline (ClSA),23,33 and (4) N,N′-bis(salicylidene)p-phenylenediamine (BSP).38 The constitutional formulas are shown in Figure 1b. The X-ray structural analysis shows that the planar SA molecules stack in the crystal, which is consistent with the thermochromic property.33 The hydrogen-bond lengths of the samples are listed in Table 1. They are longer than the critical length, 2.5 Å, below which the proton potential is considered to be a single well.41 In fact, the calculation of the electron-density map from the X-ray analysis suggests that the above four SAs have both enol and keto forms at room temperature.31,33,38-40 The present paper is organized as follows: Firstly, the temperature dependence of the fluorescence emission and excitation spectra are measured to clarify the electronic structures of the thermochromic SAs. Then, the time dependence of the fluorescence intensity is shown. The results of the deuterated analogues are also presented. The estimated protontransfer rate in the excited state of a thermochromic SA in the crystalline phase is plotted against the hydrogen-bond length, which is the distance between the hydroxy oxygen and the imine nitrogen. The process of proton transfer is discussed in terms of the proton tunneling. 2. Experimental Section The single crystals of Cl4SPY, DNP, ClSA, and BSP were prepared by the methods described in refs 39, 40, 23, and 37, respectively. The deuterated analogue, ClSA-d, was prepared by refluxing 40 mg of ClSA in 10 ml of methanol-d4 for 24 h. Slow evaporation of the solvent over a week yielded orange planar crystals typically of 1 × 1 × 0.1 mm3 size. The deuterated analogue BSP-d2 was prepared by refluxing 40 mg of the BSP in 20 ml of mixed solvent of methanol-d4/benzened6 (v/v ) 1/1) for 24 h. Slow evaporation of the solvent over a week yielded orange planar crystals typically of 0.5 × 0.5 × 0.02 mm3 size. The stationary spectra of fluorescence emission and excitation were measured by a spectrofluorometer (RF-5000, Shimadzu). The light source was a Xe lamp. The fluorescence and excitation spectra at various temperatures were measured using a continuous liquid-He flow cryostat. The excitation spectra were corrected for the light source spectrum using a thermopile. The fluorescence decay in the femtosecond region was measured using a fluorescence up-conversion method. The fundamental of Ti:sapphire laser at 820 nm (1.51 eV, 12 200 cm-1) with an average power of 600 mW and a repetition rate
Sekikawa et al. of 100 MHz was used to produce a second harmonic (SH) at 410 nm (3.02 eV, 24 400 cm-1) with an average power of 10 mW in a 3 mm LBO crystal (type I). The SH was used to excite the samples after passing through a dichroic mirror to be separated from the fundamental. The excitation power was less than 1 mW. The remaining fundamental was used as a probe beam after passing through the optical delay to up convert the fluorescence from the sample in a 0.5 mm BBO crystal (type I) by angle tuning. The time dependence of the fluorescence intensity was measured by changing the optical delay. The cross correlation trace between the SH and the fundamental had a full width at half-maximum of 260 fs. The measured cross correlation was used as a response function of the system. The up-converted signal was detected by a photomultiplier with a photon-counting system attached to a monochromator. 3. Results and Discussion Stationary Fluorescence and Excitation Spectra. Figure 2 shows the stationary fluorescence emission and excitation spectra of samples at various temperatures. All the samples were excited at 3.1 eV (25 000 cm-1), which agreed with the laser photon energy and excites both the enol and keto forms simultaneously at higher temperature. Although the sample was excited with the lower excitation energy down to the absorption edge, no substantial changes in the spectrum were observed. In addition, no clear features of an excitonic transition were also observed. The spectral features are discussed using a molecular energy diagram shown in Figure 1. Since the profile of the excitation spectrum of a crystalline sample does not completely agree with the absorption spectrum because of reabsorption effect, only the peak position is paid attention to and the profile is not discussed here. (1) ClSA. Figure 2a shows the temperature dependence of the fluorescence spectra and the excitation spectra probed at 2.14 eV (17 300 cm-1). No clear change in the excitation spectra could be observed, even when higher photon energies were probed. These spectra can be explained using the energy diagram shown in Figure 1 as follows: As the temperature rises, a new band grows around 2.4 eV (19 400 cm-1) in the excitation spectrum. This indicates that the absorption band around 2.4 eV is due to the keto form. The mirror symmetry between the thermochromic absorption and the fluorescence spectra indicates that the fluorescence is originated from the excited state of the keto form. Then, the excitation spectrum at 14 K is attributed to the enol form. Below 100 K, a large Stokes shift of the fluorescence was observed. The fluorescence intensity at 2.45 eV (19 800 cm-1) was less than 0.1% of that of the band peak. It suggests that the proton transfer takes place within a few picoseconds after the photoexcitation, since the radiative lifetime of the allowed transition is about a few nanoseconds in typical organic crystals.42 (2) Cl4SPY. Figure 2b shows the temperature dependence of the fluorescence emission and excitation spectra probed at 1.91 eV (15 400 cm-1), which is in the lower part of the fluorescence spectrum, of Cl4SPY. Cl4SPY has the shortest hydrogen-bond length among the samples studied. No clear changes in the excitation spectra were observed, although the excitation spectrum was measured at higher photon energies. The increase in the fluorescence intensity above about 2.2 eV is due to the excitation-light scattering because of smaller amount of the sample than the others. In the excitation spectra, a small hump indicated by the arrow around 2.0 eV (16 100 cm-1) observed at 17, 30, and 70 K grows with temperature, indicating the thermochromism of
Thermochromic Crystalline Salicylideneanilines
J. Phys. Chem. B, Vol. 101, No. 50, 1997 10647
(a) (b)
(c)
(d)
Figure 2. (a) Stationary fluorescence (-) and excitation (‚‚‚) spectra of ClSA at various temperatures. The excitation energy of the fluorescence spectra was 3.10 eV (25 000 cm-1). The probed photon energy in the excitation spectra is 2.14 eV (17 300 cm-1). (b) Stationary fluorescence (-) and excitation (‚‚‚) spectra of Cl4SPY at various temperatures. The excitation energy of the fluorescence spectra is 3.10 eV (25 000 cm-1). The probe photon energy of the excitation spectra is 1.91 eV (15 400 cm-1). The arrows indicate the “small hump”, which is a thermochromic band. (c) Stationary fluorescence (-) and excitation spectra of DNP at various temperatures. The excitation energy of the fluorescence spectra is 3.10 eV (25 000 cm-1). The probe photon energies in the excitation spectra are 2.00 eV (16 100 cm-1, - ‚‚‚ -), 2.21 eV (17 800 cm-1, -‚-), and 2.25 eV (18 100 cm-1, ‚‚‚). (d) Stationary fluorescence (-) and excitation (‚‚‚) spectra of BSP at various temperatures. The excitation energy of the fluorescence spectra is 3.10 eV (25 000 cm-1). The probed photon energy of the excitation spectra is 2.07 eV (16 700 cm-1).
Cl4SPY of which electronic structure is shown in Figure 1. The evolution of the small hump corresponds to the thermally activated population of the keto form. The peak photon energy of the fluorescence at 17 K is almost the same as that of the hump observed in the excitation spectrum. Therefore, the fluorescence is attributed to the keto form. The excitation band with the peak at 2.08 eV at 17 K corresponds to the absorption band of the enol form. The small Stokes shift at 17 K suggests that the energy separation between the enol and the keto forms in Cl4SPY is smaller than in ClSA. In fact, the excitation spectrum shows that the keto form of Cl4SPY is populated enough even at 17 K, which is not the case in ClSA. The energy separation between the tautomers would be less than 2 meV (16 cm-1), which corresponds to the temperature of 17 K. The fluorescence emission and excitation spectra overlap sufficiently at higher temperatures, while they do not at low temperatures because of the large Stokes shift. Thus, it is
considered that the temperature dependence of the emission spectrum is mainly caused by reabsorption. (3) DNP. Figure 2c shows the temperature dependence of both fluorescence and excitation spectra probed at 2.25 eV (18 100 cm-1), 2.21 eV (17 800 cm-1), and 2.00 eV (16 100 cm-1) of DNP. Two fluorescence bands at 2.11 eV (17 000 cm-1) and at 2.26 eV (18 200 cm-1), corresponding to A and B, respectively, are observed at low temperatures. There are three bands denoted as C, D, and E in the excitation spectra at 15 K, indicating that the origin of A band is different from that of B band. With increasing temperature, the lower parts of D and E bands grow and the peaks of these bands disappear. Thus, C and D bands correspond to the thermochromic bands. Since DNP has two intramolecular hydrogen bonds, three forms are expected in the ground state: enol-enol, enol-keto, and keto-keto forms shown in Figure 3. The energy of the enol form is usually lower than that of the keto form in the
10648 J. Phys. Chem. B, Vol. 101, No. 50, 1997
Figure 3. Energy diagram of DNP and the constitutional formulas of DNP tautomers. The asterisk (*) indicates the excited state.
ground state. Hence, it is expected that the enol-enol form is the most stable and the keto-keto form has the highest energy among the three forms in the ground state. Hence, the observed thermochromic bands C and D are assigned as the enol-keto or keto-keto form. From the hydrogen-atom-transfer (proton-transfer) mechanism proposed by Nagaoka and Nagashima,43-45 the stabilization of the keto form due to the nodal pattern of the wave function of the excited naphthalene induces an intramolecular proton transfer, suggesting that the proton transfer takes place almost independently in two intramolecular hydrogen bonds of DNP. Thus, four types of the excited species are considered: enolenol*, enol-keto*, enol*-keto, and keto-keto* forms, where an asterisk (*) indicates the excited part of the molecule. The enol-enol* and enol*-keto states relax to the enol-keto* and keto*-keto states with proton transfer, respectively, while the enol-keto* and the keto-keto* states relax to the enol-keto and keto-keto states radiatively or nonradiatively, respectively. Using the energy diagram shown in Figure 3, the fluorescence emission and excitation spectra can be explained as follows: Since C band grows with temperature and its peak energy is the lowest among C, D, and E bands, C band is assigned to the absorption band of the keto-keto form. Hence, the corresponding fluorescence A band can be attributed to the fluorescence from the keto-keto* state. If C band is due to the enol-keto form, the fluorescence energy would be much lower than A band. The other fluorescence band, B, is due to the enol-keto* state. The fluorescence of the enol*-keto form is hardly observed, because the excited enol would relax to the excited keto very rapidly. The peak energy of B band is close to that of D band. Thus, D band is assigned to the absorption from the enol-keto state to the enol-keto* state. From the peak energy and the temperature dependence of the intensity, E band is assigned to the absorption to the enol-enol* state. The absorption band to the enol*-keto state (F) would be between D and E bands, although the distinct band was not observed. The fluorescence band at room temperature is considered to have two different origins: One is the enol-keto* state at the higher energy side and the other is the keto-keto* state at the lower energy side. The lower fluorescence intensity of B band than that of A band is probably due to the reabsorption effect,32 because of the substantial overlap between the fluorescence and absorption bands. (4) BSP. Figure 2d shows the temperature dependence of the fluorescence spectra and the excitation spectra probed at 2.07 eV (16 700 cm-1), which is located in the lower photon energy region of BSP fluorescence spectrum. No clear change
Sekikawa et al. in the excitation spectra is observed at the higher photon energies in contrast with DNP. A thermochromic absorption band around 2.35 eV (19 000 cm-1) grows with temperature. Since BSP has also two intramolecular hydrogen bonds like DNP, the energy diagrams of these two SAs are considered to be similar to each other. However, only one thermochromic band was found in the excitation spectrum unlike DNP. Since the hydrogen-bond length in BSP is longer than that in DNP, the energy separation between the enol and keto forms in BSP is expected to be larger. Thus, the thermochromic band in the excitation spectra is considered to be due to the enol-keto form. The keto-keto form could not be observed at room temperature probably because of the large energy separation between the enol-keto and keto-keto forms. From these results, the fluorescence spectrum at room temperature is suggested to be composed of two bands: One is from the enol-keto* state at the higher energy side and the other is from the keto-keto* state at the lower energy side. The temperature dependence of the fluorescence spectrum is also considered to be caused by reabsorption. Time-Resolved Fluorescence. Figure 4 shows the timeresolved fluorescence decay in ClSA, Cl4SPY, DNP, and BSP probed at corresponding transition energies at room temperature. It was found that the kinetics of the fluorescence strongly depends on the probed photon energy: In the higher energy region of the fluorescence spectrum, the shorter fluorescence decay time of several hundred femtoseconds was observed, while the relatively slow decay time of several hundred picoseconds was observed in the lower energy region. In the energy range between them, biexponential decay with the short and long lifetimes was observed. These indicate the existence of two fluorescent species. The solid lines represent the leastsquare fitting of the decay convoluted with the response function. The fast decay time increases with the hydrogenbond length as shown in Table 2. Since the time-integrated intensity of the slow component was about 500 times higher than that of the fast component in every compound and the slow component was observed in the same region of the stationary fluorescence spectrum, the slow component contributes dominantly to the stationary fluorescence spectrum. Consequently, the slow component is due to the relaxation of the excited keto state in the cases of Cl4SPY and ClSA. In the cases of BSP and DNP, the slow component has two origins: One is the enol-keto* state and the other is the keto-keto* state. Even if there are these two components with different lifetimes, they could not be determined within the limited time sweep span of 20 ps. The fast decay component of the thermochromic SA was observed just below or inside of the absorption band of the enol form, suggesting that it is due to the enol* fluorescence. In addition, no fast decay components were observed in the previous study when the thermochromic band was selectively excited.23 Thus, the decay time of the fast component corresponds to the lifetime of the enol* state. In the cases of DNP and BSP, the fluorescence from both enol*-keto and enol*enol states contributes to the decay dynamics at room temperature. However, the fast decay component can be fitted well with the single exponential function. Thus, the decay rate of the enol*-keto state is considered to be close to that of the enol*-enol state. The rise time of the slow component at 77 K of ClSA agrees with the decay time of the fast component within the experimental error,23 suggesting that the excited enol state is a precursor to the excited keto state. There are following three decay channels of the excited enol state: (1) proton
Thermochromic Crystalline Salicylideneanilines
J. Phys. Chem. B, Vol. 101, No. 50, 1997 10649
(a)
(b)
(c)
(d)
Figure 4. (a) Fluorescence decays at 2.58 eV (20 800 cm-1), 2.48 eV (20 000 cm-1), and 2.18 eV (17 600 cm-1) of ClSA. (b) Fluorescence decays at 2.18 eV (17 600 cm-1) and 1.97 eV (15 900 cm-1) of Cl4SPY. (c) Fluorescence decays at 2.43 eV (19 600 cm-1), 2.34 eV (18 900 cm-1), and 2.10 eV (16 900 cm-1) of DNP. (d) Fluorescence decays at 2.53 eV (20 400 cm-1), 2.41 eV (19 400 cm-1), and 2.18 eV (17 600 cm-1) of BSP.
transfer, (2) nonradiative decay by the intersystem crossing to triplet states, and (3) radiative decay to the ground enol state. Since the radiative decay time of the allowed transition is usually of the order of nanoseconds in organic crystals and since the time for the singlet-triplet intersystem crossing in typical aromatic compounds is usually longer than several hundred picoseconds,42 the above channels (2) and (3) contribute little to the relaxation dynamics of the excited enol state. Therefore,
the decay rate of the excited enol form is considered to be mainly determined by the proton-transfer process. This is supported by the deuteration effect on the fluorescence decay as will be presented later. Figure 5 shows the relation between the protontransfer rate and the hydrogen-bond length, which will be discussed later in more detail. In the case of enol*-keto forms of DNP and BSP, the Fo¨rster-type energy transfer to the enol-keto* form may take
10650 J. Phys. Chem. B, Vol. 101, No. 50, 1997
Sekikawa et al.
TABLE 2: Time Constants (τf) of the Fast-Decay Component compound
τf (ps)
Cl4SPY DNP ClSA BSP
0.44 ( 0.08 0.71 ( 0.09 0.78 ( 0.09 1.00 ( 0.07
place as a decay channel between two moieties composing DNP and BSP, i.e., enol and keto parts in the enol-keto form of these molecules. It is, however, considered that the Fo¨rstertype mechanism makes minor contribution in the relaxation process in DNP and BSP from the following three reasons: (1) Since the Fo¨rster-type energy transfer is not expected to take place in the enol*-enol state, the relaxation of the fast component is expected to be biexponential decay due to the proton transfer and the Fo¨rster-type energy transfer. However, the fast decay component is well reproduced by the single exponential. (2) If the Fo¨rster-type mechanism contributes as much as the proton transfer to the relaxation process, F band should be observed in the excitation spectrum of B band. However, F band was not observed, indicating that the Fo¨rstertype mechanism is not the main relaxation pass of enol*-keto state. F band would be blurred by C band. (3) The Fo¨rstertype energy transfer is considered to be inefficient because of the small transition dipole associated with the nπ* transition in the keto moiety. Potential Surface along the Proton-Transfer Path in the Excited State. In the enol ground state, the O-H group forms a strong intramolecular hydrogen bond to the nitrogen atom. Photoexcitation of the enol form creates vibronic Sn states via vertical transition, i.e., the nuclear coordinates remain unchanged in the excitation process. The rapid charge redistribution after the excitation results in an electronic potential surface with a slope and an energy minimum shifted towards the proton position in the keto form. The structure of the potential surface in the excited state of a proton-transfer system has been continuing interest. In TIN (2-(2′-hydroxy-5′-methylphenyl)benzotriazole)13,14 and HBT (2(2′-hydroxyphenyl)benzotriazole)12,15 in solution, the protontransfer times were in the range of 100 to 160 fs and no deuteration effect was observed in the proton kinetics. From these two results, it was concluded that the potential energy surface for proton transfer is barrier free.12-15 The protontransfer time was considered to be determined by vibrations of the low-frequency, large-amplitude modes of the hydrogenbonded groups, of which frequencies are in the range of 100 to 200 cm-1.13,15 Recently it is proposed that the highly anharmonic in-plane deformation mode at 470 cm-1 controls proton transfer by opening a barrierless transfer channel on the excited state.22,46 In the case of thermochromic SAs studied in the present paper, the proton-transfer times are several times longer than those of TIN and HBT. The relatively longer transfer times suggest the existence of barrier in the thermochromic SAs. If there is a potential barrier, the transfer dynamics depends on the mass of the transferring particle. In order to investigate the hydrogen/deuterium isotope effect on the dynamics of the proton, the time-resolved fluorescence of ClSA-d and BSP-d2 was measured. Figure 6 shows the semilogarithmic plots of the fluorescence intensity against time for ClSA and ClSA-d at 2.58 eV (20 800 cm-1) and for BSP and BSP-d2 at 2.53 eV (20 400 cm-1). While the kinetic behaviors of the deuterated analogues were similar to those of ClSA and BSP, the lifetimes of the fast-decay components become longer by deuteration. The deuterium-transfer times in ClSA-d and BSP-d2 were 1.1 ( 0.1 and 1.3 ( 0.1 ps, respectively, while the proton-transfer
Figure 5. Dependence of the proton-transfer rate on the hydrogenbond length. The solid line represents the fitting result. See text for detail.
Figure 6. (a) Fluorescence decay of ClSA and its deuterated analogue, ClSA-d at 2.58 eV (20 800 cm-1). The solid lines represent the fitting results. The broken lines indicate the decay of the fast component. (b) Fluorescence decay of BSP and its deuterated analogue, BSP-d2 at 2.53 eV (20 400 cm-1). The solid lines represent the fitting results.
times in ClSA and BSP were 0.78 ( 0.09 and 1.0 ( 0.1 ps, respectively. The data of BSP-d2 scatter in the time range longer than 2 ps as shown in Figure 6. Even when the scattered data longer than 2 ps are excluded for the fit, the obtained lifetime is 1.2 ( 0.1 ps, which is still longer than the lifetime of BSP. Thus, it is concluded that the isotope effect is observed, suggesting the potential surface with barrier in the excited state. In the previous study,23 the proton-transfer time in ClSA at 77 K and at room temperature were close to each other. The proton transfer by a thermally activated process must be slower at the lower temperature. Therefore, it is concluded that the proton tunnels through the potential barrier quantum mechanically in SA. The differences in the potential curves of the excited state among TIN, HBT, and SA are considered to be due to the energy separation between the excited enol and keto forms in the excited state. In the case of HBT and TIN, the Stokes shifts are approximately 1.36 eV (11 000 cm-1),15,21 while Figure 2 shows that those in SAs are less than 0.2 eV (1600 cm-1). The much smaller Stokes shifts in SAs suggest that the energy levels of the excited enol and keto forms are closer to each other than the cases of HBT and TIN. In order to confirm this, however, it is necessary to estimate the energy difference between the enol and keto forms in the ground state. While that of ClSA was estimated as 31 meV (250 cm-1) experimentally in the present authors’ previous study,23 no estimation has been made for TIN and HBT to our knowledge. Thus, the calculated value of TIN 270 meV (2180 cm-1) is adopted for comparison.46 These values show that the energy difference in the excited state
Thermochromic Crystalline Salicylideneanilines
J. Phys. Chem. B, Vol. 101, No. 50, 1997 10651
of TIN is expected to be much larger than that of ClSA. Since the potential barrier of the double well decreases with increasing the energy separation between the tautomers, the potential barriers of HBT and TIN would not exist or affect the protontransfer dynamics. The proton-tunneling rate is considered to be related with the O-H/O-D stretching vibration frequency. Thus, the deuterium-transfer time in the case of the harmonic vibration is expected to be x2 times longer, which is close to the present experimental result. The small difference may be due to the anharmonicity of the vibration, which is large in the hydrogen bond as expected from the IR absorption spectrum of the O-H stretching mode.47 In fact, in the ground state of the enolenol form of BSP, the frequencies of the O-H stretching mode and that of the O-D stretching mode were 2600 and 2000 cm-1, respectively.37 The frequency ratio is 1.3, which deviates from x2 and is closer to the above-observed ratio of the protontransfer rate 1.3. The Dependence of the Proton-Transfer Time on the Hydrogen-Bond Length. Figure 5 shows the relation between the proton-transfer rate and the hydrogen-bond length. The logarithm of the proton-transfer rate decreases linearly with increase in the hydrogen-bond length. This relation can be explained by the proton-tunneling mechanism as follows: In the WKB approximation,48 the proton-tunneling rate k is
(
k ) k0 exp -
∫x2µ(U(x) - E) dx)
4π h
(1)
where k0 is the O-H stretching frequency, x is the position of the proton in the hydrogen bond, U(x) and E are potential and total energies of the proton, respectively, µ is the reduced mass of the proton, and h is the Planck constant. The integral is expanded into a power series of the tunneling distance δr ) r - r0, defined by the hydrogen-bond length r and the critical length r0 below which the potential surface has no barriers for the proton transfer. Then, k is approximated by the following equation,
k ) k0 exp(-Rδr)
(2)
where R is a constant. Hence, the proton-transfer rate has an exponential relation with the hydrogen-bond length. With the assumption of k0 ) 7.5 × 1013 s-1, which corresponds to the wave number of 2500 cm-1 for the hydrogen-bonded O-H stretching mode.37-40 R and r0 are obtained as 9.0 and 2.1 Å, respectively, by fitting eq 2 to the data as shown in Figure 5. Since the energy separation between the enol and keto forms in the excited state also depends on the hydrogen-bond length, this approximation is considered to include the effect of the change in the energy separation. In addition, it is not expected that the energy levels change as much in SA derivatives as in HBT and TIN, because the electronic circumstances around the hydrogen bonds are similar in SA derivatives. In fact, the Stokes shift in Cl4SPY is less than 0.1 eV (800 cm-1), indicating that the energy separation between the enol and keto forms in the excited state is much smaller than that in TIN. Therefore, the hydrogen-bond length is the major factor determining the protontransfer mechanism and its rate in the SAs. Using r0, the tunneling distances in Cl4SPY, DNP, and BSP are estimated to be 0.41, 0.42, and 0.49 Å, respectively. Taking account of the amplitude of the zero-point vibration of the proton 0.12 Å, the distances between the two minima in the hydrogen bond are 0.65, 0.66, and 0.73 Å, respectively. These obtained distances appear to be consistent with the distances between the two minima in the hydrogen bond suggested from the
electron-density maps, 0.43, 0.61, and 0.90 Å in Cl4SPY, DNP, and BSP, respectively.38-40 The change in the distances between the excited and the ground states reflects the drastic electronic rearrangement caused by photoexcitation. 4. Summary The present paper focuses on the relation between the proton dynamics and the hydrogen-bond length in crystalline SAs. Especially the proton-transfer rate in the excited state is estimated by measuring the lifetime of the excited enol form using the femtosecond fluorescence spectroscopy, taking account of the stationary fluorescence emission and excitation spectra. In the excited state, the proton-transfer rate was found to increase exponentially with decreasing hydrogen-bond length. This relation is explained by the quantum-mechanical tunneling of the proton in the excited state, which is supported by the observed deuteration effect on the proton-transfer dynamics. The hydrogen-bond length is an important factor to determine the proton-transfer dynamics. References and Notes (1) Nakagaki, R.; Kobayashi, T.; Nakamura, J.; Nagakura, S. Bull. Chem. Soc. Jpn. 1977, 50, 1909. (2) Nakagaki, R.; Kobayashi, T.; Nagakura, S. Bull. Chem. Soc. Jpn. 1978, 51, 1671. (3) Kobayashi, T. J. Phys. Chem. 1978, 82, 2277. (4) Kobayashi, T.; Rentzepis, P. M. J. Chem. Phys. 1978, 70, 886. (5) Barbara, P. F.; Walsh, P. K.; Brus, L. E. J. Phys. Chem. 1989, 93, 29. (6) Photoinduced proton transfer, Michael Kasha Festschrift; J. Phys. Chem. 1991, 95, 10215. (7) Kaiser, W., Ed. Ultrashort Laser Pulses, 2nd ed.; Topics in Applied Physics; Springer: Berlin, 1993; Vol. 60, Chapter 8, Addendum G and references therein. (8) Douhal, A.; Lahmani, F.; Zewail, A. H. Chem. Phys. 1996, 207, 477. (9) Elsaesser T.; Kaiser, W. Chem. Phys. Lett. 1986, 128, 231. (10) Elsaesser, T.; Schmetzer, B. Chem. Phys. Lett. 1987, 140, 293. (11) Elsaesser, T.; Schmetzer, B.; Lipp, M.; Bauerle, R. J. Chem. Phys. Lett. 1988, 148, 112. (12) Laermer, F.; Elsaesser, T.; Kaiser, W. Chem. Phys. Lett. 1988, 148, 119. (13) Wiechmann, M.; Port, H.; Laermer, F.; Frey, W.; Elsaesser, T. Chem. Phys. Lett. 1990, 165, 28. (14) Wiechmann, M.; Port, H.; Frey, W.; Laermer, F.; Elsaesser, T. J. Phys. Chem. 1991, 95, 1918. (15) Frey, W.; Laermer, F.; Elsaesser, T. J. Phys. Chem. 1991, 95, 10391. (16) Frey, W.; Elsaesser, T. Chem. Phys. Lett. 1992, 189, 565. (17) Schwartz, B. J.; Peteanu, L. A.; Harris, C. B. J. Phys. Chem. 1992, 96, 3591. (18) Arthen-Engeland, Th.; Bultmann, T.; Ernsting, N. P.; Rodriguez, M. A.; Thiel, W. Chem. Phys. 1992, 163, 43. (19) Herek, J. L.; Pendersen, S.; Banares, L.; Zewail, A. H. J. Chem. Phys. 1992, 97, 9046. (20) Lenz, K.; Pfeiffer, M.; Lau, A.; Elsaesser, T. Chem. Phys. Lett. 1994, 229, 340. (21) Chudoba, C.; Lutgen, S.; Jentzsch, T.; Riedle, E.; Woerner, M.; Elsaesser, T. Chem. Phys. Lett. 1995, 240, 35. (22) Chudoba, C.; Riedle, E.; Pfeiffer, M.; Elsaesser, T. Chem. Phys. Lett. 1996, 263, 622. (23) Sekikawa, T.; Kobayashi, T.; Inabe, T. J. Phys. Chem. A 1997, 101, 644. (24) Kobayashi, T.; Dengenkolb, E. O.; Rentzepis, P. M. J. Phys. Chem. 1979, 83, 2431. (25) Flom, S.; Barbara, P. F. Chem. Phys. Lett. 1983, 94, 488. (26) Strandjord, A. J. G.; Barbara, P. F. J. Phys. Chem. 1985, 89, 2355. (27) Dick, B.; Ernsting, N. P. J. Phys. Chem. 1987, 91, 4261. (28) Shah, J. IEEE J. Quantum Electron. 1988, 24, 276. (29) Cohen, M. D.; Schmidt, G. M. J. J. Phys. Chem. 1962, 66, 2442. (30) Cohen, M. D.; Schmidt, G. M. J.; Flavian, S. J. Chem. Soc. 1964, 2041. (31) Cohen, M. D.; Flavian, S.; Leiserowitz, L. J. Chem. Soc. B 1967, 329. (32) Cohen, M. D.; Flavian, S. J. Chem. Soc. B 1967, 334. (33) Bregman, J.; Leiserowitz, L.; Schmidt, M. J. J. Chem. Soc. 1964, 2068. (34) Bregman, J.; Leiserowitz, L.; Osaki, K. J. Chem. Soc. 1964, 2086.
10652 J. Phys. Chem. B, Vol. 101, No. 50, 1997 (35) Hadjous, E.; Milia, F.; Seliger, J.; Blinc, R.; Zagar, V. Chem. Phys. 1980, 47, 105. (36) Hadjous, E. J. Photochem. 1981, 17, 355. (37) Hoshino, N.; Inabe, T.; Mitani, T.; Maruyama, Y. Bull. Chem. Soc. Jpn. 1988, 61, 4207. (38) Inabe, T.; Hoshino, N.; Mitani, T.; Maruyama, Y. Bull. Chem. Soc. Jpn. 1989, 62, 2245. (39) Inabe, T.; Gautier-Luneau, I.; Hoshino, N.; Okaniwa, K.; Okamoto, H.; Mitani, T.; Nagashima, U.; Maruyama, Y. Bull. Chem. Soc. Jpn. 1991, 64, 801. (40) Inabe, T.; Luneau, I.; Mitani, T.; Maruyama, Y.; Takeda, S. Bull. Chem. Soc. Jpn. 1994, 67, 612. (41) Hamilton, C.; Ibers, J. A. Hydrogen-Bonding in Solids; Benjamin: New York, 1968; p 52.
Sekikawa et al. (42) Wirkinson, F. In Organic Molecular Photophysics; Birk, J. B., Ed.; John Wiley & Sons: New York, 1975; Vol. 1, Chapter 3. (43) Nagaoka, S.; Nagashima, U.; Ohta, N.; Fujita, M.; Takemura, T. J. Phys. Chem. 1988, 92, 166. (44) Nagaoka, S.; Nagashima, U. Chem. Phys. 1989, 136, 153. (45) Nagaoka, S.; Nagashima, U. J. Phys. Chem. 1990, 94, 1425. (46) Pfeiffer, M.; Lau, A.; Lenz, K.; Elsaesser, T. Chem. Phys. Lett. 1997, 268, 258. (47) Sekikawa, T.; Miyakubo, K.; Takeda, S.; Kobayashi, T. J. Phys. Chem. 1996, 100, 5844. (48) Landau, L. D.; Lifshitz, E. M. Quantum Mechanics; Pergamon: Oxford, 1978; p 178.