Excited state enol-keto tautomerization in salicylic acid - Physics, IITM

Excited state enol-keto tautomerization in salicylic acid: A supersonic free jet study Prem B. Bisht,a) Hrvoje Petek,b) and Keitaro Yoshihara Institute for Molecular Science and The Graduate University for Advanced Studies, Myodaiji, Okazaki 444, Japan

Umpei Nagashima Ochanomizu Women’s University, Outsuka, Bunkyoku, Tokyo 112, Japan

~Received 23 March 1995; accepted 12 June 1995! Excited state enol-keto isomerization in salicylic acid ~SA! monomer and dimer has been studied in a supersonic free jet expansion. Two carboxylic group rotamers of SA with significantly different photophysical properties are found in the expansion. Rotamer I, the major form of SA in the expansion, has an intramolecular hydrogen bond and can undergo excited state tautomerization reaction. Its S 1 origin is at 335.34 nm. Single vibronic level emission spectra are dominated by progressions in OH stretching ~3230 cm21!, and in-plane bending of the carboxylic group ~240 cm21!. The spectra appear to consist of two components, normal ~UV! and tautomer ~BLUE! emissions, even at the origin. The intensity of the BLUE relative to the UV emission depends on the vibronic state, rather than the excess vibrational energy between the origin and 1100 cm21. The fluorescence decay time profiles for both the emission components of rotamer I are identical within ;1 ns experimental time resolution. A nonradiative decay process with an activation energy of ;1100 cm21 is deduced from an abrupt decrease in fluorescence lifetimes above this energy. The rotamer II cannot undergo excited state tautomerization. Its electronic origin is at 311.52 nm and emits only UV fluorescence. Upon increasing the concentration of the SA sample, a new spectrum is observed. Due to a nonlinear concentration dependence of the intensity and the propensity of SA to form dimers in solution, it is assigned to the SA dimer. This spectrum shows possible evidence of double proton transfer in the S 1 state. © 1995 American Institute of Physics.

I. INTRODUCTION

The excited state enol-keto tautomerization reaction is of great scientific and technological interest in organic synthesis, photochemical energy conversion, photoprotection, oscillation of laser dyes, and design of functional materials.1–7 To gain experimental and theoretical understanding of the mechanism for excited state intramolecular proton transfer ~ESIPT! in complex biological systems and practical applications, it is desirable to study simple systems, such as salicylic acid, under supersonic expansion conditions. In a large class of molecule, the intramolecular proton transfer process can be described by a transformation between two degenerate, or nearly degenerate, limiting enol and keto resonance structures,

Such a transformation is often simply called an intramolecular proton transfer. However, in this work, the process will be described as a ‘‘enol-keto tautomerization’’ to emphasize that along with the proton transfer, a significant redistribution of a!

Address for correspondence: Photophysics Laboratory, Kumaun University, Naini Tal, 263001, India. b! Present address: Advanced Research Laboratory, Hitachi Ltd., Hatoyama, Saitama 350-03, Japan. 5290

J. Chem. Phys. 103 (13), 1 October 1995

electron density results in changes in the bond orders and lengths of heavy atoms such as C and O. The correct description of electronic structure of such a molecule involves contributions from both of the limiting enol and keto forms. Nagaoka and Nagashima have introduced a ‘‘nodal plane’’ model to predict from simple ab initio calculations which of the two forms will be more stable in electronically excited states.8 The calculated position of a nodal plane on the benzene ring, on a molecule such as SA, is a useful qualitative guide for predicting the more stable form in a particular electronic state. The nodal plane in electronically excited states will coincide with single bonds of the more stable form. This theoretical approach can explain, for instance, why the S 1 state of 2-hydroxybenzoic acid undergoes tautomerization, while the S 2 state does not.8 This theory demonstrates that the proton transfer occurs in the response to the changes of electron distributions on the heavy atoms of the benzene ring. Tautomerization reaction can occur between equivalent or inequivalent structures. In a molecule where the enol and keto structures are identical, the emission spectra are indistinguishable, and the proton-transfer can be deduced only from a splitting of vibronic bands due to tunneling between the two equivalent minima. The tunneling splittings depend on the vibronic state and are very sensitive to the vibrational motions of heavy atoms, which may promote or hinder the transformation.9,10 When the two structures are inequivalent, the relative stability of the enol and keto forms depends on the electronic state, and there may be a barrier to intercon-

0021-9606/95/103(13)/5290/18/$6.00

© 1995 American Institute of Physics

Bisht et al.: Tautomerization in salicyclic acid

version. The experimental signature of tautomerization is a large Stokes-shift of the emission spectrum, which can be observed both in condensed and gas phases.2– 6,8 –20 Such redshifted emission was first observed by Marsh in solutions of methyl salicylate ~MS!.1~a! Later, Weller explained this anomalous emission from salicylic acid ~SA! derivatives in terms of ESIPT.1~b!,1~c! Subsequent studies of the fluorescence excitation spectra of MS in the solution and gas phases revealed that there are actually two carboxylic group rotamers present in this system; the interpretation of the emission spectra are complicated by the fact that only one can undergo ESIPT.11–17 The presence or absence of a barrier can have a strong effect on the rates of transformation. When the enol and keto forms are equivalent, interconversion between tautomers can occur by tunneling on picosecond to subpicosecond time scales through barriers of several kcal/mol. The barrier heights and tunneling rates are strongly dependent on the electronic state. Due to the coupling of heavy atom motions the tunneling rate is strongly dependent even on the vibronic states, which do not involve large hydrogen atom motion. Asymmetric molecules may have single or double minima. Methyl salicylate and related molecules appear to have a single minimum in the enol form for the ground, and keto form for S 1 states.2,11–17 From the studies of relative fluorescence quantum yields in organic glasses below 100 K of bis-2,5-~2-benzolyl!-hydroquinone, an S 1 state barrier of 0.35 kcal/mol has been reported.4~a! For 3-hydroxyflavone in solution, the tautomer emission rise time of 100–300 fs has been measured.4~b! In case of 2-~28-hydroxyphenyl! benzoxazole ~HBO!, the ESIPT rate of 60630 fs has been deduced by pump–probe transient absorption measurements in solution.5 Homogeneous broadening of vibronic bands of supersonic jet spectra of 7-azaindole and 1-azacarbazole dimers,9 3-hydroxyflavone,4~b!,18 and HBO ~Ref. 5! have been attributed to ultrafast tautomerization. In a recent study on MS in vapor phase, the measured tautomer rise time of 60 fs is reported for the S 1 state.15~b! Since these rise times correspond to many periods of the O–H stretching vibration, it appears that the tautomerization time scale is determined by the motions of the heavy atoms rather than of the H atom. This raises the problem of whether these processes can be described by few degrees of freedom involving mainly hydrogen motion, or whether a multidimensional surface involving adiabatic motions of heavy atoms is more appropriate. Since resonance Raman spectroscopy gives information on the initial dynamics on the excited states, it provides important information on how to describe the reaction coordinate. Significantly, resonance Raman spectra of 2-hydroxy acetophenone show displacements of a large number of the skeletal vibrational coordinates, but not of the OH stretch.13 It appears that the motion of the hydrogen occurs in response to the adiabatic motions of heavy nuclei, which in turn are caused by a redistribution of electron density in the excited states. Therefore, models which treat tautomerization as onedimensional H atom transfer may not have sufficient complexity to describe spectroscopic and kinetic observations on this class of molecule. Even though SA is one of the simplest systems in which

5291

tautomerization can occur, there are as yet no reported studies in the gas phase and only a few in the condensed phase.19,20 Because SA readily dimerizes in the condensed phase and the electronic spectra of the monomer and dimer overlap, it has not been possible to study electronic spectra of the monomer in solution. The possibility of the double proton transfer in the ground and S 1 state has been suggested for the dimer.19 Molecular orbital calculations of SA indicate a small or no barrier to proton transfer.21 Several studies have been devoted to the derivatives of SA, particularly methyl salicylate ~MS! in solution, matrices, gas phase, and free jet expansions.1,11–17 Goodman and Brus have shown that in 4.2 K neon matrix, MS does not appear to have a barrier to tautomerization.11~a! A number of studies on MS fluorescence excitation and emission spectra in supersonic jets have shown the following: ~i! two MS rotamers exist in the jet; ~ii! the more abundant rotamer undergoes tautomerization; ~iii! this rotamer decays by a nonradiative process with an activation energy of 1300 cm21; and ~iv! its emission spectrum shows large displacements in the O–H, C–O, and low frequency ~176 cm21! modes, with an emission maximum at two quanta of O–H stretch.14 –17 The nonradiative process has been attributed variously to the torsion of the –COOMe group, crossing of the p* – p and p* –n surfaces, breaking of the hydrogen bond, and analogous decay to the channel three process in benzene.14 –16 Of particular interest is a recent time-resolved fluorescence depletion measurement of S 1 state of MS, where a 60 fs transient was attributed to the hydrogen atom transfer reaction.15~b! Though there have been several studies on MS and several derivatives in supersonic expansions, a number of spectroscopic and dynamical details are still controversial. Though SA readily dimerizes in condensed phase, we have found it possible to generate supersonic jets of mostly SA monomer or dimer by controlling SA partial pressure before the expansion. In this work, we report fluorescence excitation, dispersed fluorescence spectra, and decay rates of SA and SA dimer ~SA2!. The goal is to understand tautomerization in SA and SA2 ~double proton transfer! under isolated conditions. Both rotamers of SA monomer ~Fig. 1! exist in the expansion. Rotamer ~I! has dual ~small and large Stokes shift! emission even at the origin, while rotamer ~II! has only one emission component with a small Stokes shift. The assignments of the fluorescence excitation spectra have been made on the basis of the available data on phenol, MS, benzoic acid dimer, benzaldehyde, as well as theoretical calculations. The S 1 state potential energy surface appears to have a single minimum, which has more keto character than the ground state. Furthermore, SA has an analogous nonradiative channel to MS with an activation energy of 1100 cm21. The fluorescence excitation spectrum of SA2 shows progressions in several low frequency intermolecular modes. The emission spectra show evidence for large displacements in hydrogen bonded H atoms in the excited state. The relative simplicity of SA compared to MS, makes it a better model for studying excited-state tautomerization reactions in simple aromatic systems.

J. Chem. Phys., Vol. 103, No. 13, 1 October 1995


5292

FIG. 1. Structures of rotamer I, II, and dimer of SA.

II. EXPERIMENT

Experimental arrangement for fluorescence excitation spectral measurements in free jet expansions has been described elsewhere.22 Supersonic expansion was formed by flowing He seeding gas at stagnation pressure of 1–3 atm over a solid sample of SA, and by expanding the gas mixture through 0.50 mm orifice during 300– 400 ms opening of a pulsed valve. The partial vapor pressure of SA could be varied between ;0.04 –0.8 Torr by adjusting the SA reservoir temperature between 35 and 100 °C. Before introducing the sample into vacuum system, the nozzle was heated for 48 h up to 100 °C under vacuum to remove water, which can form clusters with SA. The vacuum chamber was at ,1025 Torr during the opening of the pulsed valve. The fluorescence excitation spectra were measured with an excimer-pumped dye laser ~Lambda Physik-LPX 105/LPD 3002!. The laser and molecular beams intersected approximately 12 mm below the nozzle orifice. Fluorescence from the sample was detected through optical filters, which selected UV ~Schott WG 335, UG 11; 340–370 nm! or BLUE ~Schott GG 385 or UY42, BG 3; 380– 480 nm! emissions. Transmitted light was detected by a photomultiplier tube ~Hamamatsu H3177!, gated by a Stanford Research Systems boxcar integrator ~SR250!, and averaged and stored by a microcomputer. Ten laser shots were averaged for each 0.0025 nm laser scan interval

with 10 Hz laser repetition rate. The laser wavelength was calibrated by measuring the optogalvanic spectrum of a Ne hollow cathode lamp. The laser resolution was ;0.3 cm21. Fluorescence spectra were recorded by dispersing the total fluorescence in a monochromator, and detecting emission with an optical multichannel analyzer ~PAR 1420! having a 700 detector array. The exposure times were 7 min with excitation rates of 8 –10 Hz. Emission spectra were calibrated by measuring atomic emission spectrum of mercury. The fluorescence decay time profiles were measured in the following manner. The excitation source was a Coherent 700 synchronously pumped dye laser, which operated with Rh 6G and DCM Special dyes. The picosecond oscillator pulses were amplified at 10 Hz rate with a three stage amplifier pumped by a regenerative YAG amplifier ~Quantel RGA60-10!. The excitation wavelength was measured with a calibrated monochromator. Excitation bandwidth of ;20 cm21 in UV could be achieved with a three-plate birefringent filter in the oscillator. The SA sample was excited by the doubled output of the amplifier, which had a ;10 ps pulse width and ;25 mJ energy. The resulting fluorescence was detected by a Hamamatsu Photonics H3284 photomultiplier, which had a specified rise time of ;300 ps. The fluorescence decay traces of monomer were detected and averaged 128 – 256 times by a Tektronix 602A digitizing oscilloscope with a 11A72 vertical amplifier ~1 GHz bandwidth!. For the dimer we used a LeCroy 7200 digital oscilloscope with 500 MHz bandwidth. The decay data were analyzed using a nonlinear least square fitting program. The validity of the fitting to a single or biexponential functions was checked using x2 and weighted residuals. SA ~.99.9%! from Wako was used without purification. Mixture of the SA-h and monodeuterated SA ~SA-d! was prepared by keeping SA in D2O atmosphere at 60 °C, in the sample holder of the pulsed valve, for 14 h. The valve was operated at 1 Hz to allow D2O to flow over the sample. To confirm that the new bands are due to SA-d, different sets of experiments were carried out by varying the exposure of SA to D2O. A broad background due to clusters is observed when the spectra are recorded immediately after introducing D2O in the sample holder. After flowing D2O over the sample for about 1 h, the sample was visually inspected. The top layer of the sample had recrystallized due to H–D exchange. In longer scans, the intensity of bands due to SA-d, decreased and after about 10 h of the experiments at room temperature vapor pressure of the sample, all monodeuterated sample disappeared and only spectrum due to the SA-h was observed. The spectrum was recorded in several parts and combined into one. The intensities of the bands may have 50% error due to depletion of the SA-d with time. III. RESULTS AND DISCUSSION A. Spectroscopy of SA rotamers

The two rotamers of the SA are shown in Fig. 1. The rotamer I forms an intramolecular hydrogen bond and it can tautomerize in the S 1 state. The rotamer II does not have an intramolecular hydrogen bond and enthalpy difference is 2.5 kcal/mol in case of MS.14~b! Rotamer I can convert to rotamer



5293

FIG. 2. The fluorescence excitation spectra of the rotamer I on monitoring ~a! BLUE fluorescence, ~b! UV fluorescence. Asterisk ~*! indicates the bands due to the SA dimer. The spectra are normalized at the origin.

II by internal rotation of the carboxylic group, which requires an activation energy of ;10 kcal/mol.12~c! The identification of the observed spectra with rotamer I and II is based on similar assignments for MS.14 1. Fluorescence excitation spectra of SA (rotamer I)

The emission from gas phase SA has some unusual features, which need to be described before presenting the fluorescence excitation spectra. The emission spectrum is similar to MS in that the intensity maximum has a large Stokes shift with a maximum at ;430 nm, with a tail extending to .600 nm. Such a broad spectrum with a large Stokes shift implies that either there is a large change in the excited state geometry, or that the absorbing and emitting states are different. As a result of tautomerization in the S 1 state of SA, it is possible that the S 0 and S 1 states have a single minimum, where S 0 is described by enol structure and S 1 keto, or that the S 1 surface has two minima due to enol and keto forms, which can interconvert on time scale of the emission. To distinguish between these two cases, we measured fluorescence excitation spectra monitoring separately UV ~340–370 nm! and BLUE ~380– 480 nm! emissions. The fluorescence excitation spectrum of SA measured while monitoring the BLUE emission is shown in Fig. 2~a!. This spectrum resembles literature spectra of MS under similar conditions. By analogy to MS, the spectrum is assigned to the S 1 ←S 0 ~p*←p! transition of the rotamer I.14,15 Although

SA and MS have very similar chromophores, there are some significant differences in the spectra. For instance, the origin of SA ~335.35 nm; 29 820 cm21, air! is considerably weaker, relative to higher frequency bands, than the origin of MS ~The origin of MS is at 30 052 cm21.!14 Also, due to the methoxy group, MS has additional bands and the density of vibrational states is significantly higher. The stronger vibrational bands in the spectra of SA are listed in Table I along with frequencies and assignments of the corresponding bands of MS. Fluorescence excitation spectrum of SA measured while monitoring the UV emission is shown in Fig. 2~b!. Except for several very weak bands in the UV spectrum, which we will show are due to the dimer, most bands appear in both spectra. Significantly, when the spectra are normalized at the origin, some vibronic bands have considerably higher intensities in the UV spectrum. For example, the 849 cm21 band is about 2.2 times more intense in UV than in BLUE fluorescence excitation spectrum. It will be shown that these differences in intensities are consistent with a single, displaced minimum in the S 1 state of SA. The fluorescence excitation spectra can be divided into three regions; 0–1000 cm21, 1000–1400 cm21, and above 1400 cm21. The spectrum is sparse up to 500 cm21 above the origin. Sharp, isolated vibronic bands appear in the 0–1000 cm21 region with an intensity maximum at 849 cm21. The widths of the vibrational bands of ;1 cm21 are probably determined by the laser resolution and rotational band con-



5294

TABLE I. Observed vibrational frequenciesa and assignments for SA-d 0b and SA-d 1 ~rotamer I!. Frequencya/ cm21

UVc intensity

Blued intensity

Assignments ~SA-d 0!

Frequency ~SA-d 1!

0 88 364 370 431 511 704 727 734 795 800 811 835 849f 859 900 906 910 914 930 944 956 958 989 994 998 1030 1054 1072 1079 1092 1098 1104 1125 1132 1138 1140 1150 1156 1162 1175 1185 1205 1244 1251 1261 1281 1379 1449

36 12 85 30 17 81 106 52 117 81 40 262 215 623 25 65 68 301 114 5 90 36 36 3 20 43 39 171 43 142 77 106 36 178 86 76 199 107 59 4.5 30 63 34 81 50 45 48 37 40

36 9 45 27 20 66 91 47 80 61 35 134 133 283 29 39 58 167 95 16 60 31 20 26 131 44 30 173 30 108 59 85 34 136 73 57 130 91 44 33 24 69 28 60 30 34 28 31 27

0-0 ~29 820 cm21! n1 n2 n3 n4 n5 n6 2n2 n21n3 n51n1 , n21n4 n31n4 2n21n1 , n7 2n31n1 , n8 n9 2n4 n71n1 n10 n11 n12 n81n1 n91n1 , n12 n13 n14 n15 n101n1 n111n1 n121n1 n16 n21n6 n61n3 n17 3n2 n18 n19 n20 n141n1 n21 n412n2 n22 n714n1 2n23 3n21n1 n24 n71n4 n715n1 n81n4 n91n4 n121n31n1 n25

0-0 ~29 840 cm21! 84 363 432 489 674 or 688 720 724 787 795 804 825 844

MS assignmentse and frequencies 0-0 ~30 052.3 cm21! n1 ~176! n2 ~347! n216 n4 ~424! n11n2 ~521! n5 ~569! 2n2 ~694! n7 ~739! n61n1 , n21n4 ~770! n21n416 ~776! n8 ~810! 2n4 ~847!

899 882

n9 ~904! n916 ~908! n21n5 ~913!

932

n11 ~956! n11n8 ~987!

1023 1028 1053

n12 ~999! n112n4 ~1022! n11n9 ~1080! n13 ~1094!

1116

2n5 , n21n31n4 ~1140! 1145

n51n6 ~1164! 1185 1198

n31n9 ~1280! 4n112n2 ~1396!

Frequency shift from the 0-0, estimated accuracy 63 cm21. Intensities are normalized at the origin. c On monitoring UV fluorescence ~340–370 nm!. d On monitoring BLUE fluorescence ~380– 480 nm!. e Assignments of MS are taken from Heimbrook et al. ~Ref. 11!. The numbers in parentheses indicate the frequency shifts from 0-0 of MS. f Most intense transition.

a

b

tours. Excess broadening in the 1000–1400 cm21 region is probably due to intramolecular vibrational coupling. In MS, a larger density of vibrational states due to the methyl torsion results in excess broadening appearing above ;900 cm21. Above ;1400 cm21 the SA and MS spectra are structureless and fluorescence intensity decreases with energy.

Vibrational analysis. The vibrational analysis is based on comparisons of SA spectrum with available data on related molecules such as benzene, phenol,23 MS,14 benzoic acid dimer,24,25 benzaldehyde,26 salicylaldehyde,27 as well as theoretical calculations.28 The lowest frequency mode ~88 cm21; n1 of Table I! is not found in phenol,23 therefore it is



likely to be the lowest frequency mode associated with the carboxylic group, i.e., the torsion of the –COOH ~calculated to be 99 cm21 in S 0 ,28 and observed at 92 cm21 in the S 1 state of naphthoic acid!.29 This band also is found in combination with other major vibrational bands, however, it has no overtones. Considering just the reduced mass for the torsion, the corresponding band in MS would be expected at lower frequency. However, the lowest frequency band in MS is at 136 cm21.16 This mode may be either methyl torsion, or –COOMe torsion, but with a substantially larger force constant than the corresponding mode in SA. The 364 and 511 cm21 modes ~n2 and n5 of Table I! correspond to the doubly degenerate n6 (e 1u ) in-plane breathing modes of benzene, and are commonly found in benzene containing molecules such as MS, phenol, and benzoic acid dimer.14,23–25 These modes have been reported at 347 and 557 cm21 in case of MS. The 370 and 431 cm21 modes of SA probably correspond to the 353 and 424 cm21 modes of MS. Relative to other bands in the spectrum, the 557 cm21 mode of MS is considerably more intense that the 511 cm21 mode of SA, while the relative intensities are reversed for the 431 and 424 cm21 modes.14 These two modes probably involve the OH in-plane bending mode. This is supported by the calculated ground state values for SA ~382 and 485 cm21!,28 and a significant deuterium isotope shift ~29 cm21! for the 424 cm21 mode of MS.14 The 704 cm21 mode of SA does not have an obvious counterpart in MS. Therefore, it probably involves the motion of the –COOH group. The calculated frequencies of ground state modes with large components of –COOH in-plane bend are 243 and 701 cm21. These modes are expected to have large Franck– Condon factors due to a large change in the angle of the bond between the –COOH and benzene ring ~ab initio values for benzaldehyde are 119.3° in S 0 and 122.7° in S 1!.30 Due to much larger inertial mass, the corresponding –COOMe bend may be assigned to either 569 or 595 cm21 bands of MS. Transitions in the 800–1100 cm21 region are the most intense in the SA fluorescence excitation spectrum. These bands probably involve considerable distortion of the benzene ring ~ring breathing!, which is enhanced by the excitedstate tautomerization. Significant displacements along these modes result in large Franck–Condon factors. The strongest band in the spectrum of SA is at 849 cm21. By contrast, the most intense band in MS is at 1094 cm21 ~tentatively assigned to carbonyl stretching!.14 Such differences in relative intensities for a number of vibronic bands of SA and MS may be due to a large vibronic state dependence of fluorescence quantum yields, or due to different Franck–Condon factors. However, lifetimes of neither molecule show large state-dependent variations, that would be indicative of a similar trend in quantum yields ~vide infra!. Therefore, the differences in relative intensities can be attributed more probably to different molecular structures and extent of tautomerization in the S 0 and S 1 states of the two molecules. Bands with large Franck–Condon factors help in determining the change in the molecular structure upon excitation to the S 1 state. The available spectroscopic information on the S 1 state indicates changes in the ring bond lengths, and –COOH bending angle. Heimbrook et al. predicted the

5295

CvO and OH stretches at 1094 and 2582 cm21.14 There are a number of bands in 1000–1200 cm21 region, which may be candidates for CvO stretching. However, by contrast to MS, these are not the strongest bands in the spectrum. Due to the loss of intensity and structure in the fluorescence excitation spectrum, it may not be possible to identify the CvO and OH stretches in the S 1 state of SA by fluorescence techniques. However, evidence for displacement along these modes comes from emission spectra, which show strong activity of OH and double bond ~CvC and CvO! regions consistent with expected changes in structure due to the excited state tautomerization. 2. Fluorescence excitation spectrum of rotamer II

Figure 3 shows the fluorescence excitation spectrum of the rotamer II up to 1300 cm21 above the origin, which was recorded by monitoring fluorescence between 320–360 nm. The spectrum is considerably weaker than for the rotamer I, which is consistent with the lower concentration and assignment to p*←n transition as suggested for MS.14 Another factor contributing to lower signal to noise ratio of this spectrum, as compared with the rotamer I, is that the spectrum is superimposed on a structureless fluorescence excitation spectrum due to other species present in the beam. On monitoring the fluorescence longer than 360 nm, only a structureless spectrum, which decreases in intensity at high energy, is observed. Since rotamer II cannot tautomerize, it does not have substantial long wavelength emission such as the BLUE emission of rotamer I. Hence, the continuous spectrum observed when long wavelength emission is monitored, probably is due to high energy tails of the rotamer I and the dimer ~see Sec. III B!. It is difficult to estimate of the relative populations of the two rotamers in the expansion, because different experimental conditions are necessary to measure the spectra, and the transition moments and fluorescence quantum yields are not known. Since the enthalpies of formation for the two rotamers of SA probably are similar to those of MS, the rotamer II concentration is probably ;30%, as for MS in near room temperature supersonic expansions.14 The most intense band at 311.52 nm ~32 101 cm21! is assumed to be the origin ~;2280 cm21 above the origin of rotamer I!. Another peak, which is 240 cm21 lower in energy, only appears when monitoring UV emission and hence cannot be assigned to the rotamer I. As its intensity strongly depends on the expansion conditions, it probably is due to a hot-band or a cluster of rotamer II. For comparison, the origin of rotamer II spectrum of MS is at 32 303 cm21.14 Since there is no internal hydrogen bond in rotamer II, it also is possible that rotamers of the phenolic OH can be observed in the expansion. Table II lists the vibrational assignments for rotamer II. Since the structures of rotamers I and II are related, a number of vibrational modes appear with similar frequencies. For instance, the bands of rotamer I at 364, 431, and 511 cm21 appear at 363, 403, and 521 cm21 for rotamer II. The low frequency bands at 44 and 69 cm21 may be due to the torsion and out-of-plane bending of the carboxylic group. The most active mode in the spectrum is the ;240 cm21 band ~n4 , Table II!, which has several overtones and combination


5296


FIG. 3. Fluorescence excitation spectrum of rotamer II on monitoring UV fluorescence.

bands with other modes. We assign this mode to the in-plane bend of –COOH group, as will be explained in Sec. IV. 3. Fluorescence excitation spectra of SA-d

Figure 4 shows the first 800 cm21 of the S 1 ←S 0 fluorescence excitation spectrum of partially deuterated SA ~SA-d n ; n, number of substituted H atoms!. The spectrum was recorded by detecting the total emission longer than 350 nm. Interpretation of the spectrum is complicated by the presence of two acidic protons—one on the carboxylic and the other on the phenolic groups—which are only partially exchanged. Therefore, we expect to see features in the spectrum due to at least four distinct species corresponding to unsubstituted SA, monosubstituted carboxylic and phenolic SA, and disubstituted SA. By contrast, MS has only one acidic hydrogen on the phenolic group. The most pronounced origin of deuterium substituted SA is shifted by 121 cm21 from the SA-d 0 origin. For comparison, the deuterium isotope shift for MS in free jet is 199 cm21 for rotamer I, and 121 cm21 for rotamer II.14 The large difference in isotope shifts between MS and the most prominent SA isotopomer suggests that the substitution is on different functional groups. The difference in isotope shifts for rotamers of MS suggests that substitution at an intramolecular hydrogen bond results in a larger isotope shift. As result of excited state tautomerization, the largest change in frequencies between S 1 and S 0 states is expected for the modes, which involve intramolecular hydrogen bond. This reasoning sug-

gests that the relatively small shift for the most intense SA isotopomer origin is due to the substitution is on a nonhydrogen bonded OH, i.e., the carboxylic acid. This assignment is consistent with relative acidities of the carboxylic and phenolic OH groups. The origin of the phenolic OD species is probably in a cluster of four lines between 88 –104 cm21 above the origin of SA-d 0 . We tentatively assign the lines at 88 and 104 cm21 to the mode 88 cm21 of SA-d 0 and SA-d 1 ~carboxylic acid!, respectively, while the 93 and 99 cm21 probably are due to the origins of SA-d 1 ~phenolic! and SA-d 2 . That the band at 193 cm21 is an origin can be confirmed by observation of vibronic development of higher frequency modes, which is similar to SA-d 0 . The assignment of the 199 cm21 band to the SA-d 2 origin is more tentative, since the isotope shifts do not appear to be additive, and the intensities are more sensitive to the degree of deuteration. We are unable to identify higher vibronic bands belonging to this species. The frequencies of the vibronic bands involving substantial H motion are expected to have the largest shifts. The presence of three isotopomers makes it difficult to uniquely identify these bands. For example, the 364 cm21 band, which is due to the vibration of the ring, does not have a significant isotope shift. However, bands at 511 and 704 cm21 are found at 489 and 674 or 688 cm21 for SA-d 1 . These two bands probably involve significant motion of the –COOH group. For rotamer II of SA, the only isotopomer is observed at 134 cm21. Since this shift is significantly different than for


Bisht et al.: Tautomerization in salicyclic acid TABLE II. Frequency shift, intensity, and assignments of rotamer II of SA.

a

Shifta

Intensity

2226 0 44 69 203 240 358 363 403 481 521 546 561 584 598 603 643 683 720 801 806 809 832 842 861 882 920

13.5 36 14 14.2 12.8 24.8 11.5 14.6 18.2 14.7 7.8 9.2 11.2 7.6 7.8 12.9 13.8 13.1 13.6 18 16 14 21 15.5 16.6 18.2 27.3

Assignments ? Origin ~32 098 cm21! n1 n2 n3 n4 n5, n11n21n4 n6 n51n1 2n4 n612n11n2, n7 n8 n71n1 n71n2 n81n1 n41n6 n41n51n1 n41n512n1 3n4 , n71n3 , 2n5 2n512n1 2n41n5 2n41n6 2n41n51n1 n61n71n1

Relative to the origin ~cm21!.

MS, we assign it again to the –COOD species based on the above arguments. For comparison monodeuteration of benzoic acid dimer results in 23 cm21 blue shift of the origin of the first p – p* state.24~a!,24~b! Due to poor signal to noise ratio, it was not possible to identify spectra of other isotopomers of rotamer II or the SA dimer. This in part is due to low concentrations, a number of partially substituted species, and overlap of the spectra with a background due to SA: water clusters and SA rotamer I continuum. 4. Emission spectra

Figure 5 shows emission spectra from the S 1 origin and 849 cm21 bands of the rotamer I. The emission spectra from SA and MS origins are similar. The main Franck–Condon active modes in the emission spectrum from the SA origin are 240 ~in-plane bending of the carboxylic acid! and 3230 cm21 ~phenolic O–H stretch! bands. In the 1500–1700 cm21 region there are a number of bands probably due to CvC and CvO stretches and their combination bands with the 240 cm21 mode. These modes clearly have significant Franck–Condon factors, but due to spectral congestion it is difficult to make specific assignments as was possible in case of MS.15~a! The spectrum has a maximum at ;430 nm, which approximately corresponds to the excitation of two quanta of OH stretch. Emission from vibronic bands with up to 1000 cm21 excess energy has resolved vibrational structure ~Fig. 6!. However, the emission spectrum is structureless for 11057 cm21, the highest energy band for which the emission spectrum was measured. This probably is due to the

5297

relaxation of the prepared state by intramolecular vibrational redistribution ~IVR!, which becomes a significant factor for this and higher energy bands. The emission spectrum from 849 cm21 band is of particular interest. It shows the characteristic 240 cm21 progression at low frequencies and becomes congested for Stokes shifts of .1000 cm21. Under low resolution there are two broad maxima at ;380 and 430 nm, which give the appearance that the emission consists of two overlapping spectra. Such spectra are sometimes observed when optical excitation produces significant population in two emissive electronic states. Similar two-component emission spectra are observed for other bands in the 800–1100 cm21 wavelength region. The ratio of UV-to-BLUE emissions is clearly state dependent, as can be seen in the comparison of UV and BLUE fluorescence excitation spectra in Fig. 2. Although it is tempting to assign UV emission to enol, and BLUE emission to the keto isomers, we will argue in the discussion section that these unusual state dependent variation can be explained by differences in the extent of tautomerization in the ground and excited states. We will now focus here on assignments of individual modes. Vibrationally resolved emission spectrum from the S 1 origin of MS in supersonic jet reported by Felker et al.15~a! shows three prominent modes; 180 ~distortion of the ring containing the hydrogen bond!, 1690 ~carbonyl stretch!, and 3220 cm21 ~OH stretch!.15~a! In relatively low resolution emission spectra ~;50 cm21!, the 180 cm21 mode appears to form progressions of five or more quanta from the origin and prominent bands such as CvO and O–H stretches. While the assignment of the CvO and O–H stretches appears straightforward and easy to justify considering the excited state tautomerization, the assignment of the 180 cm21 is more controversial. Although the 180 and 176 cm21 bands in S 0 and S 1 states have very similar frequencies, contrary to previous assignments, we believe that they should be assigned to different vibrational motions. Zewail and co-workers suggested that the 180 and 176 cm21 bands in the S 0 and S 1 states of MS, respectively, are due to the out-of-plane bending motion of the ‘‘ring’’ that includes hydrogen bond.15~a! Heimbrook et al. suggested that the 176 cm21 mode is a vibration of the ‘‘ring’’ containing the hydrogen bond, and they assigned a number of bands in the S 1 state to its overtones and combination bands based on the observed frequencies. However, for the monodeuterated molecule the overtones of the 176 cm21 mode are not observed, which provides an argument against some of the assignments for MS-h and suggests that this mode is not displaced with respect to the ground state. Furthermore, Heimbrook et al. report an isotope shift of 110 cm21 for this mode.14 However, deuterium isotope substitution should decrease the vibrational frequency of any mode which involves the hydrogen bond. Work of Nishiya et al.17 also casts doubt on assignment of the 176 cm21 mode to a distortion involving the hydrogen bond. They measured the fluorescence excitation and emission spectra ~in durene mixed crystals at 4.2 K and in supersonic jets! of molecules in which the methoxy group is replaced by other substituents ~H, CH3 , and NH2!. These molecules have the same hydro-


5298


FIG. 4. Fluorescence excitation spectra of mixture of SA-d 0 on monitoring emission longer than 420 nm ~a!, and SA-d n on monitoring the total emission longer than 350 nm ~b!.

gen bonding ring and should show analogous bands to the 176 and 180 cm21 modes, if the above assignments to outof-plane distortions are correct. Since these spectra do not have the corresponding bands, they concluded that the 176 cm21 mode is due to hindered methyl torsion.17 This conclusion may be consistent with the deuterium isotope shift for the 176 cm21 mode. By contrast, the 180 cm21 mode in the ground state of MS is strongly displaced between S 0 and S 1 states, and is observed in MS analogs.17 Therefore, this mode probably does derive intensity from the excited state tautomerization. Though the vibrational frequencies are deceptively similar, the 176 and 180 cm21 modes clearly are due to unrelated vibrations. Corresponding to the 180 cm21 mode of MS, we find a 240 cm21 progression in the ground state of SA. Although there is a significant increase in frequency from MS to SA, the fact that both modes form long progressions suggests that the Franck–Condon activity is due to similar changes in the chromophore. Different frequencies probably can be attributed to different reduced masses of –COOH and –COOMe groups. Therefore, this vibration is localized mostly on the ring containing the intramolecular hydrogen bond. The assignment to the out-of-plane distortion is unlikely because an antisymmetric mode, such as out-of-plane bend, would form a progression only if the molecular structure is planar in one state and nonplanar in the other. However, there is no reason to believe that tautomerization would affect the planarity of the molecule. We prefer to assign the 240 cm21 mode of SA and 180 21 cm mode of MS to the in-plane-bending of the carboxylic

group. This assignment is corroborated by the calculated vibrational frequency for this mode ~243 cm21!,28 and is consistent with the large frequency difference between SA and MS. Analogous mode with 264 cm21 frequency and a large displacement also is seen in the emission and resonance Raman spectra of 2-hydroxyacetophenone.13 The calculated increase in angle between the benzene and carbonyl groups in salicylaldehyde is 13.4° upon excitation to the S 1 state.30 This change occurs in response to excited state tautomerization and it results in a large Franck–Condon factors for these related modes. In the S 1 state of SA, a corresponding band to the 176 cm21 mode of MS is not observed. This is to be expected if the 176 cm21 band is due to methyl torsion. In the S 1 state of SA there are no obvious bands which can be identified with the 240 cm21 band in the S 0 state. Tautomerization in the S 1 state is expected to increase the bond order ~force constant! of the C–C bond between the carboxylic group and the benzene ring. Hence, the in-plane-bend is expected at higher frequency than in the S 0 state. The 240 cm21 normal mode of the S 0 state, probably has a higher frequency and undergoes Dushinsky rotation in the S 1 state.31 A possible candidate is the 704 cm21 mode, as suggested in Sec. III A 1. Further experimental and theoretical work is necessary to identify the in-plane bending of the carboxylic group in S 1 state of SA and related molecules. B. Spectroscopy of SA dimer

In condensed phase, SA exists predominantly as a dimer consisting of two units of rotamer II ~Fig. 1!. Absorption and



5299

FIG. 5. Low resolution emission spectra of rotamer I upon pumping the origin ~solid line! and 1849 cm21 bands ~dotted line!. Inset shows low frequency bands observed in higher resolution emission from the origin.

emission spectra of the two rotamers and the dimer of SA overlap. As a result, there have been no reliable studies of excited electronic states of SA in the condensed phase. In a supersonic jet, control over the vapor pressure of the sample, and the simplification of the spectra by vibrational and rotational cooling allowed us to separate the dimer and monomer spectra. The dimer can consist of a combination of two units of rotamer I and/or II ~Fig. 1!.19~c! Since dimerization of rotamer I requires weakening of two intramolecular hydrogen bonds to form two intermolecular bonds, structure ~IV! is less stable with respect to dissociation to monomer units than structure ~III!. In crystalline SA, IR and Raman spectroscopy has been used to deduce an energy difference of 154 cm21 between the two structures.19 Structure III can be formed by dimerization of two units of rotamer I followed by double proton transfer, or by 180° torsion of the two phenyl rings, or by dimerization of rotamer II. At present, the relative stabilities and activation energies for interconversion of structures I–IV under isolated conditions are not known. The concentrations of the monomer and dimer in the supersonic jet probably are determined both by thermodynamics and kinetics of interconversion between the rotamers, and cluster formation and dissociation. In condensed phase, the relative stabilities and kinetics of interconversion probably are influenced strongly by the presence of a solvent. It appears that nonpolar solvents strongly favor dimerization, while at room temperature in gas phase rotamer I is the dominant

FIG. 6. High resolution emission spectra of rotamer I from several S 1 state bands indicated by their frequencies.


5300


species. As a result of prevalence of the dimer in solution phase, only vibrational assignment of its ground state are available.24~c! 1. Fluorescence excitation spectra

The fluorescence excitation spectra of rotamer I contain some weak features, which are more pronounced relative to the monomer when UV emission is monitored ~see Fig. 2!. Increasing partial pressure of the SA sample increased the intensity of these weak features relative to the monomer, as expected for the dimer. At ;0.8 Torr SA partial pressure in the sample reservoir ~90 °C temperature!, the dimer bands dominate the fluorescence excitation spectra @Fig. 7~a!#. The relative intensities of the dimer origin bands do not change with the He stagnation pressure ~1–3 atm!. We assign the origin of SA2 to the intense band at 30 049 cm21, which is 229 cm21 above the origin of rotamer I. The spectrum has sharp, well separated lines in the low energy region, with a maximum at 378 cm21. There is a continuous background, which reaches a maximum at ;1000 cm21 above the origin. Above this energy, the spectrum is dominated by a continuous background, which decreases in intensity. The only sharp structure at high energies is due to the rotamer II. Another very weak progression is observed between 27–34 cm21 below the dimer origin. These bands appear only when UV emission is monitored, and have maximum intensity at high He stagnation pressures ~3 atm!. These bands may be due to other isomers of the dimer or due to clusters of the dimer with He or H2O. In case of 7-azaindole dimers, the less stable isomers also are observed.9 The 0–500 cm21 region of the SA2 spectrum is dominated by progressions of four low frequency modes ~57, 95, 101, and 151 cm21!. The assignments are based on those for the benzoic acid dimer reported by Tomioka et al., as given in Table III.24 The low frequency bands are due to hydrogen bond ~HB! twist, out-of-plane bend, and stretch, and C–COOH torsion, in-plane and out-of-plane bends.24~c! At higher frequencies the in-plane vibrations of benzene rings, such as 354, 437, 501 cm21 appear at similar frequencies as the monomer @Fig. 7~b!#. Essentially the whole spectrum can be assigned to combination and overtone bands of the above low frequency modes associated with the intramolecular hydrogen bond, –COOH bending and torsion, and benzene ring distortions. Another interesting feature of the dimer fluorescence excitation spectrum is 1.5– 4 cm21 splitting of the bands, which are localized on the benzene rings such as 354 ~n7!, 409 ~n71n1!, 501 ~n14!, and 505 cm21 ~n71n4!. This splitting probably arises from weak coupling of benzene ring modes, which are localized at either end of the dimer and couple through the hydrogen bond. The spectrum contains a broad background, which is noticeable above ;200 cm21 and has a maximum at ;1000 cm21 excitation energy. Several factors probably contribute to this background. Due to several low-frequency, Franck– Condon active intermolecular modes, at high energies there is a large density of bright states, which contribute to spectral congestion and IVR. Also, the dimer can undergo intermolecular double proton transfer, inter converting forms III and

IV ~Fig. 1!. The barrier for this process is not known and there is no strong spectroscopic evidence that this process occurs, such as been reported for 7-azaindole dimers.9 However, such a process could induce homogenous broadening of the vibronic bands and greatly increase the density of vibrational states.

2. Emission spectra

The emission spectrum from the dimer origin is shown in Fig. 8. It has a few resolved bands near the origin, but at higher energies individual bands cannot be resolved due to the congestion. The dimer emission spectrum also appears to consist of two broad, barely resolved components with maxima at ;360 and ;400 nm. The intensities of each component appear to be nearly equal. The two components may reflect the Franck–Condon intensity distribution for S 1 state emission of SA2 ; or it may be evidence that the emission occurs from two different species, for instance the normal and tautomer forms. It is conceivable that the structure of the dimer in the S 1 state is intermediate between structures III and IV ~partial excited state proton transfer!, so that Franck– Condon factors are nearly the same for emission to III and IV structures. Or it may be that even at the dimer S 1 origin there is a rapid equilibration between structures III and IV, so that the emission occurs to both ground state species. In condensed phase ~concentrated solutions or crystalline solid! the dimer emission also shows two overlapping emission bands at ;370 and ;440 nm.19~a!,19~b! The single exponential decay with the same rate constant has been attributed to the excited state equilibrium between the two structures.19~b! Fuke et al.9 have reported strong evidence for ESDPT in dimers of 7-azaindole and derivatives in supersonic expansions. They observed fluorescence from both the normal and tautomer forms of the dimers. On increasing the excitation energy, the emission from the normal form decreased and that of the tautomer form increased. Due to the presence of a barrier to ESDPT, the rate of formation of tautomer increased with excitation energy. However, for SA2 the shapes of the low resolution emission spectra are independent of the excitation wavelength. Due to overlap of absorption spectra of several species, this could be confirmed only for the strongest SA2 bands. If dual emission can be assigned to the normal and tautomer forms, it may suggest that the ESDPT is much faster than emission, and the equilibrium between the normal and tautomer forms is independent of the excitation energy. Although our data are consistent with a single minimum, we cannot exclude rapid equilibration between two nearly isoenergetic, strongly fluorescent species. Higher resolution emission spectra from several low frequency S 1 bands of SA2 are shown in Fig. 9. A ;100 cm21 progression is prominent for emission from the origin, 157, and 195 cm21 bands; and a 300 cm21 progression appears for emission from 195 and 151 cm21. These bands probably correspond to dimer torsional and stretching bands observed in condensed phase Raman spectra at 106 and 301 cm21.24~c!



5301

FIG. 7. ~a! Fluorescence excitation spectrum of dimer on monitoring fluorescence between 340–380 nm; and ~b! the 300–500 cm21 region is presented in higher resolution to show the splitting of several bands discussed in the text. J. Chem. Phys., Vol. 103, No. 13, 1 October 1995


5302

TABLE III. Salicylic acid dimer spectrum. Experimental conditions temperature of the sample holder 90 °C ~;0.8 Torr!, He5800 Torr, emission monitored at 340–380 nm. Shift from SA dimer origina 2229 2141 0 57 95 101 111 112 151 158 169 189 195 206 246 252 263 274 283 289 298 300 329 340 354 355 366 378 383 393 395 404 409 411 422 437 446 450 460 466 477 492 501 503 505 507 516 540 556 572 574 596 620 649 660 671 679 691 707 716 726 742 762 771 799

Benzoic acid dimerb S 0 , ~S 1!c

57, ~57! ~70! 117, ~110! 168, ~209!

223, ~221! 258, ~248!

388

422

616

Assignments of dimer

Intensity 5.6 1.8 190 197 143 36 92 51 195 32 52 108 37 83 109 39 48 66 45 19 180 66 132 41 223 223 46 176 62 59 43 70 203 217 55 105 180 171 52 104 109 50 140 119 95 105 35 115 87 59 54 198 130 187 96 90 87 72 109 106 75 93 94 84 83

Origin SA ~I! 29 820 cm21 188 SA ~I! Origin ~dimer! 30 049 cm21 n1 ~HB twist! n2 ~HB op bend! n3 ~C–COOH tor! 2n1 2n1 n11n2 , n4 ~HB str! n11n3 3n1 2n2 n21n3 n212n1 n21n4 , n112n2 n11n21n3 n213n1 , n412n1 n5 ~C–COOH op bend! 3n2 n312n2 n6 ~COOH ip bend! 2n4 , 3n3 n51n1 n412n2 n61n1 , n7 ~skeletal defor! n711 n51n2 n51n3 4n2 n61n2 n212n4 n11n21n31n4 , 4n3 n71n1 n71n112 n51n11n2 n8 ~skeletal defor! n61n4 n61n11n2 , n61n414 4n31n1 n712n1 n512n3 , 5n2 2n212n4 , n81n1 n9 ~skeletal defor! n912 n71n11n2 , n41n7 , 5n3 n71n412, n61n112n2 4n312n1 n61n41n2 , n10 ~skeletal defor! n91n1 2n6 , n91n2 , n612n4 ~n8 of monomer! n11 2n512n1 n12 n61n51n3 2n61n2 2n7 , n111n1 2n513n1 n121n1 2n71n1 , n1112n1 2n514n1 n1212n1



5303

TABLE III. ~Continued.! Shift from SA dimer origina

Benzoic acid dimerb S 0 , ~S 1!c

813 825 889 911 943 956 999

Intensity

Assignments of dimer

49 112 91 75 91 79 61

2n513n11n2 n12 of monomer n13 ~n121n1 of monomer! n131n1 ~3n21n1 of monomer! n131n3

Estimated accuracy 63 cm21; HB5hydrogen bond, tor5torsion, ip5in-plane, op5out-of-plane, defor 5deformation. b Reference 24. c The values in parenthesis are for the S 1 state.

a

C. Time-resolved emission

1. Rotamer (I) of SA

The lifetimes for rotamer ~I! were measured as a function of the excitation energy. The fluorescence decays of rotamer II were too fast to measure with our experimental time resolution ~,1 ns!. Figure 10 shows the dependence of the SA rotamer I decay rates on the excitation energy. The decay times for the BLUE and UV components are same within our experimental error ~65%!. The lifetime at the origin is 9.660.5 ns ~12.0 ns for MS!. Between the origin and ;1100 cm21, the lifetime decreases monotonically to ;4.3 ns. Above ;1100 cm21 there is an abrupt acceleration in the decay rate, which coincides with the fall off in the spectral intensity. At higher energies ~.1000 cm21! the decay curves are biexponential. The longer component has a nonlinear dependence on SA concentration, and we therefore, attribute it to the dimer. Observation of the same kinetics for the UV and BLUE emission implies that the emission comes from the same species or that the equilibrium between the two species is established on a subnanosecond time scale. We attribute the sudden increase in the observed decay rates above 1100 cm21 to an efficient intramolecular nonradiative decay process. The decrease in the fluorescence quan-

FIG. 8. Emission spectrum from the origin of the dimer.

tum yield in solution and gas phase has been observed in several related molecules.8,13–15 The fluorescence decay rates in MS show similar behavior, except that the activation energy for nonradiative decay is ;1300 cm21.14,15 Several mechanisms have been proposed to explain the nonradiative decay of MS. SA, MS, and related molecules are known to have two or more close lying electronic states.14,15 The observed spectrum has been assigned to a ~p,p*! transition. One or more singlet ~n,p*! states are predicted to have nearly the same energy. It is conceivable that there is a surface crossing between the two electronic states at ;1100 cm21. However,

FIG. 9. Emission spectra of dimer on pumping the indicated low frequency bands in the S 1 state.


5304


FIG. 10. Fluorescence decay rates as a function of energy of the rotamer I on monitoring BLUE ~l! and UV ~s! fluorescence.

since the ~n,p*! states do not correlate to the ground state, strong coupling between ~n,p*! and ~p,p*! states does not in itself lead to fast nonradiative decay. The fluorescence quenching could be accomplished by intersystem crossing from the ~n,p*! states. An additional problem with this mechanism is the difference in the activation energy between SA and MS. The fact that both the origin and threshold for nonradiative decay of SA are ;200 cm21 lower than for MS, implies that the ~n,p*! state of SA must be ;400 cm21 lower than for MS ~assuming similar shapes and displacements of potential energy surfaces!. Since methyl substitution usually decreases rather than increases the 0-0 transition energies of ~n,p*! states, an increase in excitation energy upon methyl substitution by 400 cm21 is unlikely. Two mechanisms are proposed for the nonradiative decay of MS.15 The first one involves the breaking of the intramolecular hydrogen bond by out-of-plane torsion of the –COOMe group. We prefer the second explanation for the following reasons. The ground state barrier for the analogous isomerization of salicyldehyde is estimated to be 7– 8 kcal.32 In the ~p,p*! state the bond connecting the carboxylate to the phenyl has more p character than the ground state. Thus, for salicyldehyde, the S 1 state barrier is 6200 cm21.32 Therefore, the 1300 cm21 activation energy probably is too low to represent breaking of the hydrogen bond by torsional motion of the –COOMe. Also, if motion of the hydrogen atom were involved in the nonradiative decay, it would probably show a significant deuterium atom effect, contrary to the observations in Ref. 15~b!. The second mechanism proposed in Ref. 15~b! is the analog of the channel 3 decay process in benzene. The nonradiative decay of benzene above ;3000 cm21 energy in the S 1 state has been attributed to internal conversion to the ground state via either Coriolis coupling33 or isomerization to benzvalene.34,35 Since the S 1 states of MS, SA, and benzene are all of ~p,p*! nature, a channel 3 like process also may lead to nonradiative decay in SA and its analogs. If the nonradiative decay occurs via a crossing between S 0 and S 1 states, then the ;200 cm21 difference in origins of MS and

FIG. 11. Fluorescence decay rates as a function of energy of the dimer on monitoring emission longer than 350 nm.

SA is consistent with the similar decrease in the activation energy for internal conversion to S 0 . Further experimental and theoretical work will be necessary to determine the mechanism for nonradiative decay of SA and related molecules. The much faster decay of rotamer II is not well understood. However, since the observed spectrum is thought to be due to the p*←n transition,14~b! it is not surprising to observe different photophysics than for rotamer I. 2. SA dimer

Lifetimes of the dimer were measured by detecting the total fluorescence above 350 nm. The decay rates are shown in Fig. 11. At the origin the lifetime is approximately equal to that of the monomer, i.e., ;960.5 ns. Up to 1100 cm21 above the monomer origin it is difficult to measure dimer fluorescence decays without a contribution from the monomer. Since the lifetimes are similar, there is significant correlation in parameters determined in fitting the data to biexponential decay profiles. Due to much more rapid decay of the monomer than the dimer for excitation 1100 cm21 above the monomer origin, the high energy lifetime values are more reliable. The lifetimes of the dimer show a monotonic decrease up to 2500 cm21 above the origin. Due to much larger density of states in the dimer, a sharp threshold for the nonradiative decay is not observed. The nonradiative process could be the same as for the monomer, or it may involve processes such as dissociation or double proton transfer. D. Potential energy surface for tautomerization

Figure 12 shows a schematic potential energy surface for the excited state tautomerization of SA. Such a potential surface was originally proposed by Goodman and Brus to explain matrix isolation spectra of MS,11~a! and it was later adopted by Zewail and co-workers to explain their femtosecond time-resolved measurements.15~b! The same schematic potential also explains the fluorescence excitation and emission spectra of SA. Although we have used the limiting enol and keto resonance structures to label the two minima, we



FIG. 12. The potential energy surface for the excited state tautomerization of SA. Wave functions for the vibrational ground states, and modes involved in tautomerization are schematically indicated.

should emphasize that the extent of hydrogen transfer is rather small. This is corroborated by the ab initio calculations on salicyldehyde and related systems.8,30 The observed spectra support a single minimum, rather than a double minimum originally proposed by Weller,1 for the following reasons. Spectra of 3-hydroxyflvone ~Ref. 18! and 1-azacarbazoles ~Ref. 9! are characteristic of excited state tautomerization by tunneling through an excited state barrier between two excited state minima. The origin of the spectrum is due to the vertical excitation. Tunneling through the barrier to tautomerization results in Stokes-shifted emission. The intensity of the Stokes-shifted emission increases and that of vertically excited state decreases with excitation energy, since the tunneling rates increase and equilibrium favors the more stable tautomer form. If the tunneling rate is rapid, it is also possible to observe lifetime broadening in the fluorescence excitation spectrum. However, in case of SA, the UV emission intensity increases as a function of excitation energy, rather than decreasing as in the double minimum case. The origin appears to be due to a nonvertical excitation into the excited state minimum, which is displaced in the tautomer coordinate. The intensity of individual bands appear to reflect Franck–Condon factors rather than energy dependent tautomerization between two excited state minima. The Franck–Condon factors for transition between the ground ‘‘enol’’ to the excited state ‘‘keto’’ minima ~0-0 transition! is small but observable. With nanosecond laser excitation we do not expect to see any dynamics associated with tautomerization. The 0-0 excitation optically prepares the enol wave function by a ‘‘nonvertical’’ transition. At higher excitation energies we expect to find ‘‘vertical’’ transitions associated with the ‘‘reaction coordinate.’’ Such vibronic

5305

states will have significant probability density at both classical turning points of the tautomerization potential, as shown in Fig. 12, and will have strong BLUE and UV emissions. The normal modes, which appear stronger in UV than in BLUE fluorescence excitation spectra, probably are related to the tautomerization coordinate. Significantly, these modes do not appear to involve significant hydrogen motion. Modes like the prominent 849 cm21 band probably involve the skeletal distortions of the benzene ring. Since the excited state partially loses aromaticity, the bond orders of benzene ring CC bonds either increase or decrease, and in response, the bond lengths change. Calculations for salicyldehyde show that the bond lengths change from nearly uniform 1.40 Å in the ground state, to 1.34 Å for double bonds, and 1.47 Å for the single bonds in the S 1 state.30 For comparison the bond length of O–H is predicted to increase by only 0.01 Å. This is consistent with the analysis of the resonance Raman spectrum of 2-hydroxyacetophenone, which shows displacement of a number of skeletal modes and no significant displacement of the O–H stretch.13 In case of MS and SA the emission spectra show evidence of hydrogen motion. In both cases the most prominent emission band is at two quanta of the O–H stretch. Thus there is evidence that tautomerization involves the motion of both the phenolic H atom and of the heavy atoms. However, the emission spectrum gives information on long time scale dynamics. Instantaneous excitation of SA results in motion along several coordinates on different time scales. The change in electronic structure mostly affects the p-bonding framework. The initial motion is along the high frequency modes with steepest gradients. These probably are CvO and CvC stretching coordinates. The O–H stretching potential probably changes with time in response to the adiabatic motion of the heavy nuclei. The final motion of H probably involves in-plane bending of the –COOH, which is the lowest frequency strongly displaced mode. The time resolution in our experiment is not sufficient to resolve the initial dynamics. However, Zewail and coworkers monitored the nuclear motion in S 1 state of MS using a time resolved fluorescence depletion technique with 80 fs duration pump and probe pulses.15~b! Fluorescence depletion technique is a useful method for studying excited state dynamics from the time evolution of the emission spectrum.36 In case of MS, UV, and BLUE emissions were monitored as a function of delay between the excitation and gate ~fluorescence quenching pulses!. Under the bulb conditions of these experiments, most of the UV emission came from rotamer II of MS. Both emissions showed a 60 fs rise, followed by the characteristic decay for rotamer II and I, respectively. Substitution of the phenolic OH by deuterium resulted in the same rise time as the parent molecule. The rise time of the BLUE emission was attributed to the excited state tautomerization, while similar rise time for the UV emission was explained by the evolution of the excited state wave packet due to the response of the nuclei to the nearly instantaneous change in the potential surface. We suggest that in both cases, the 60 fs rise time represents the same dynamics, namely the response of the nuclear motion to formation of a node in the electronic wave function bisecting


5306


the benzene ring, as described in Ref. 8. Therefore, propose that such rise times should be common to ~p,p*! and ~n,p*! states of benzene ring containing molecules. It is our opinion that the spectroscopy and dynamics of SA, MS, and related molecules can be mainly described by the motion of heavy nuclei, while the changes in the O–H•••O structure is relatively small and strongly influenced by the heavy atom motion.

IV. CONCLUSIONS

In conclusion, we have studied the SA monomer and the dimer in supersonic free jet. Two rotamers of SA are found to exist in the expansion with different populations. The rotamer I emits from normal and tautomer forms even at the electronic origin. The ratio of UV-to-BLUE emissions depends on the vibronic state, and is large with respect to the origin for selected bands in the 800–1100 cm21 region. Apparently these vibronic states have better Franck–Condon factors than the origin, and probably have a significant projection on the reaction coordinate for the tautomerization. These bands provide evidence for significant motion of the heavy atoms in response to changes in electron densities following the S 1 ←S 0 excitation. The emission spectrum is dominated by progressions in OH stretching ~3230 cm21! and in-plane distortion of carboxylic group ~240 cm21!, and show significant activity for the CvC and CvO stretches. The large Franck–Condon factors for these modes are a consequence of excited state tautomerization. The decay rates for both the emission components of rotamer I are the same indicating that on the ;1 ns time scale, the species leading to the two emission components is the same. A sudden increase in the nonradiative rates 1100 cm21 energy above origin has been observed. This is most likely due to an analogous internal conversion process to the channel 3 in benzene. The rotamer II does not undergo tautomerization and as a consequence shows only UV fluorescence. Upon increasing the concentration of the sample in the expansion a new spectrum due to the SA dimer is observed. The emission spectra of the dimer show possible evidence of double proton transfer. The fluorescence decay rates indicate a nonradiative process, which may be attributed to the proton transfer, dissociation, or another nonradiative process. Salicylic acid is one of the simplest molecules which undergoes excited state tautomerization, and it is convenient to study experimentally and theoretically. Therefore it is a promising system for further studies of excited state tautomerization.

ACKNOWLEDGMENTS

We thank Instrument Center of IMS for the loan of the excimer-pumped dye laser system, and Professor S. Nagaoka for participating in helpful discussions. P.B.B. acknowledges Ministry of Education, Science and Culture of Japan for a fellowship. This work was supported by a Grant-in-Aid for Scientific Research on New Program ~05NP0301! by the Ministry of Education, Science, and Culture of Japan.

~a! J. K. Marsh, J. Chem. Soc. 125, 418 ~1924!; ~b! A. Weller, Z. Elektrochem. 60, 1144 ~1956!; ~c! A. Weller, Prog. React. Kinet. 1, 187 ~1961!; ~d! V. E. Bondybey, R. C. Haddon, and J. H. English, J. Chem. Phys. 80, 5432 ~1984!; ~e! D. McMorrow and M. Kasha, J. Phys. Chem. 88, 2235 ~1984!; J. Am. Chem. Soc. 105, 5133 ~1983!. 2 ~a! P. F. Barbara, P. K. Walsh, and L. E. Brus, J. Phys. Chem. 93, 29 ~1989!; ~b! P. F. Barbara, L. E. Brus, and P. M. Rentzepis, J. Am. Chem. Soc. 102, 6531 ~1980!. 3 ~a! P. T. Chou, D. McMorrow, T. J. Aartsma, and M. Kasha, J. Phys. Chem. 88, 4596 ~1984!; ~b! N. P. Ernsting and B. Nikolaus, Appl. Phys. B 39, 155 ~1986!; ~c! P. T. Chou, S. L. Studer, and M. L. Martinez, Appl. Spectrosc. 145, 513 ~1991!; ~d! M. L. Martinez, W. C. Cooper, and P. T. Chou, Chem. Phys. Lett. 193, 151 ~1992!. 4 ~a! U. Brackmann, N. P. Ernsting, D. Ouw, and K. Schmitt, Chem. Phys. Lett. 110, 319 ~1984!; ~b! N. P. Ernsting and B. Dick, Chem. Phys. 136, 181 ~1989!. 5 T. A.-Engeland, T. Bultmann, N. P. Ernsting, M. A. Rodriguez, and W. Thiel, Chem. Phys. 163, 43 ~1992!. 6 P. F. Barbara and H. D. Trommsdorff, Special Issue, Chem. Phys. 136 ~1989!. 7 T. Miyasaka, K. Koyama, and I. Itoh, Science 255, 342 ~1992!. 8 ~a! S. Nagaoka, U. Nagashima, N Ohta, M. Fujita, and T. Takemura, J. Phys. Chem. 92, 166 ~1988!; ~b! S. Nagaoka and U. Nagashima, ibid. 94, 1425 ~1990!; ~c! 95, 4006 ~1991!; ~d! Chem. Phys. 136, 153 ~1989!. 9 ~a! K. Fuke and K. Kaya, J. Phys. Chem. 93, 614 ~1989!; ~b! K. Fuke, K. Tsukamoto, F. Misaizu, and K. Kaya, J. Chem. Phys. 95, 4074 ~1991!. 10 ~a! H. Sekiya, Y. Nagashima, and Y. Nishimura, J. Chem. Phys. 92, 5761 ~1990!; ~b! T. Tsuji, H. Sekia, Y. Nishimura, R. Mori, A. Mori, and H. Takeshita, ibid. 97, 6032 ~1992!, and references therein. 11 ~a! J. Goodman and L. E. Brus, J. Am. Chem. Soc. 100, 7472 ~1978!; ~b! A. U. Acunã, F. Toribio, F. Amat-Guerri, and J. Catalan˜, J. Photochem. 30, 339 ~1985!; ~c! J. Catalan˜, F. Toribio, and A. U. Acunã, J. Phys. Chem. 86, 303 ~1982!; ~d! R. Lopez-Delgado and S. Lazare, ibid. 85, 763 ~1981!. 12 ~a! W. Klo¨pffer and G. Naundorf, J. Lumin. 8, 457 ~1974!; ~b! 20, 283 ~1979!; ~c! W. Klo¨pffer, Adv. Photochem. 10, 311 ~1977!; ~d! K. Sandros, Acta Chem. Scand. A 30, 761 ~1976!. 13 L. A. Peteanu and R. A. Mathies, J. Phys. Chem. 96, 6910 ~1992!. 14 ~a! L. A. Heimbrook, J. E. Kenny, B. E. Kohler, and G. W. Scott, J. Chem. Phys. 75, 5201 ~1981!; ~b! J. Phys. Chem. 87, 280 ~1983!. 15 ~a! P. M. Felker, W. R. Lambert, and A. H. Zewail, J. Chem. Phys. 77, 1603 ~1982!; ~b! J. L. Herek, S. Pedersen, L. Banãres, and A. H. Zewail, ibid. 97, 9046 ~1992!. 16 J. W. Kuper and D. S. Perry, J. Chem. Phys. 80, 4640 ~1984!. 17 T. Nishiya, S. Yamauchi, N. Hirota, M. Baba, and I. Hanazaki, J. Phys. Chem. 90, 5730 ~1986!. 18 A. Ito, Y. Fujiwara, and M. Itoh, J. Chem. Phys. 96, 7474 ~1992!. 19 ~a! H. C. Joshi, H. B. Tripathi, T. C. Pant, and D. D. Pant, Chem. Phys. Lett. 173, 83 ~1990!; ~b! D. D. Pant, H. C. Joshi, P. B. Bisht, and H. B. Tripathi, Chem. Phys. 185, 137 ~1994!; ~c! P. Upadhyay, D. D. Pant, and H. D. Bist, in Proceedings of the XII International Conference Raman Spectroscopy, edited by J. R. Durig and J. P. Sullivan ~Wiley, Singapore, 1990!; ~d! P. Fischer, P. Jolliker, B. H. Meier, R. R Ernst, A. W. Hewat, J. D. Jorgensen, and F. J. Rotella, J. Solid State Commun. 61, 103 ~1986!. 20 ~a! S. E. Morsi and J. O. Williams, J. Chem. Soc. Perkin Trans. II, 1281 ~1978!; ~b! P. J. Kovi, C. L. Miller, and S. G. Schulman, Anal. Chim. Acta 61, 7 ~1972!. 21 J. Catalan˜ and J. I. F.-Alonso, J. Mol. Struct. 27, 59 ~1975!. 22 H. Petek, A. J. Bell, and K. Yoshihara, J. Chem. Phys. 95, 4739 ~1991!. 23 H. Abe, N. Mikami, M. Ito, and Y. Udagawa, Chem. Phys. Lett. 93, 217 ~1982!. 24 ~a! D. E. Poelt and J. K. Mcvey, J. Chem. Phys. 78, 4349 ~1983!; ~b! Y. Tomioka, H. Abe, N. Mikami, and M. Ito, J. Phys. Chem. 88, 2263 ~1984!; ~c! V. Volovsek, L. Colombo, and K. Furic, J. Raman Spectrosc. 14, 347 ~1983!. 25 J. C. Baum and D. S. McClure, J. Am. Chem. Soc. 102, 720 ~1980!. 26 A. J. Duben, L. Goodman, and M. Koyanagi, in Excited States, edited by E. C. Lim ~Academic, New York, 1974!, Vol. 1. 27 C. J. Seliskar, J. Mol. Spectrosc. 53, 140 ~1974!. 28 U. Nagashima ~unpublished results!. 29 S. Jagannathan and D. W. Pratt, J. Chem. Phys. 100, 1874 ~1994!. 30 S. Nagaoka, Kagaku to Kogyo 44, 182 ~1991!. 31 F. Duschinsky, Acta Physiochim. URSS 7, 551 ~1937!. 1


Bisht et al.: Tautomerization in salicyclic acid 32

E. Orton, M. A. Morgan, and G. C. Pimental, J. Phys. Chem. 94, 7336 ~1990!. 33 U. Schubert, E. Riedle, H. J. Neusser, and E. W. Schlag, J. Chem. Phys. 84, 6182 ~1986!.

34

5307

M. Sumitani, D. V. O’Connor, Y. Takagi, K. Kamogawa, N. Nakashima, Y. Udagawa, and K. Yoshihara, Chem. Phys. 93, 359 ~1985!. 35 S. Kato, J. Chem. Phys. 88, 3045 ~1988!. 36 H. Petek and K. Yoshihara, J. Chem. Phys. 87, 1458 ~1987!.
