The effect of hydrogen-bonding on the ultrafast electronic deactivation dynamics of indigo carmine

Excited-state dynamics and mechanisms of the rapid deactivation process of indigo carmine (InC) were investigated by means of femtosecond transient absorption spectroscopy and steady-state Raman spectroscopy.

Solvent dependence of the excited-state lifetime revealed that intermolecular hydrogen-bonding with the solvent molecule is more effective than the intramolecular ones to accelerate the deactivation process.

Steady-state Raman spectra in the low-frequency region indicated a loss of molecular planarity in protic solvents.

It was concluded that the hydrogen-bonding in the excited state, which leads to the twisting around the central CC bond and/or to the out-of-plane deformation, was of crucial importance in the rapid deactivation.


Indigo is one of the natural dyestuffs utilized by human beings from the dawn of civilization.

The description “indigo” can refer either to the blue dyestuff obtained from several species of plants, or to the chemical compound accountable for the color.

In this report, it represents the latter.

The Nobel Prize in 1905 was awarded to a German research chemist, A. von Baeyer, for the synthesis and the structural formulation of indigo.1,2

Influenced by this success, search for artificial indigo derivative with superior chemical properties took place and the first artificial derivative, thioindigo, was synthesized in .19063

It is well known that thioindigo and other indigo-derivatives shown in Scheme 1 (a) undergo photochromism, i.e., reversible transcis isomerization around the central CC bond on photo-excitation.4–9

However, for indigo (X = NH) itself, photo-isomerization and/or the cis-form have never been observed.

On the basis of the peculiar molecular structure enabling intramolecular hydrogen(H)-bonds between the two adjacent pairs of carbonyl and NH groups,10–13 two possibilities were considered as the origin of the absence of photo-isomerization; (1) double intramolecular H-bonds “lock” the molecular structure in planer trans-form, or (2) intramolecular proton-transfer from the NH group to the carbonyl group occurs on the excited state prior to the isomerization, which leads back to the trans-configuration in the ground state.14,15

For highly fluorescent thioindigo, it is generally accepted that transcis isomerization takes place through triplet intermediate.15–21

It was concluded that the planar trans configuration is in thermal equilibrium with the twisted configuration in the T1 state.20

T1 state is also considered to be the intermediate in the case of N,N′-dimethylindigo.22

Intersystem crossing is usually a slow process which can be inhibited by a faster reaction channel.

Elsaesser and coworkers have carried out time-resolved picosecond fluorescence and IR spectroscopy of 4,4′,7,7′-tetramethylindigo, a indigo derivative that can form intramolecular H-bonds.23

They concluded that proton transfer does not occur in the first excited electronic state from the following experimental results, (a) their time-resolved IR spectra revealed that the NH stretching in the S1 state was shifted to lower frequency only by 40 cm−1 compared to the ground state, (b) no evidence of a transient OH band was found around 3200 cm−1 in the excited molecule, (c) the CO stretching mode at 1640 cm−1 remained virtually unchanged upon photo-excitation and (d) the steady state fluorescence spectrum was Stokes shifted only by 900 cm−1 which was rather too small for a fluorescence form a proton transferred tautamor which can be as large as ∼10 000 cm−1.

Although, there are many reports on excited state intramolecular proton transfer (ESIPT) taking place from phenolic –OH group to the H-bonded adjacent –N group, i.e., 2-(2′-hydroxy-5′-methylphenyl)benzotriazole,24,25 2-(2′-hydroxyphenyl)benzothiazole,26,27 and 10-hydroxybenzo[h]quinoline,28 we could not find any report on ESIPT from NH group to CO group.

Intramolecular H-bond forming 5- or 6-membered ring is considered to be strong, because of the good overlap of the molecular orbitals.29

If the isomerization of indigo is blocked by the intramolecular H-bond bridge, formation of intermolecular H-bonds with the solvent may break the bridge and lead to photo-isomerization.

However, addition of alcohols to the solution of photochromic indigo dyes is known to quench the fluorescence and diminish the isomerization.30–32

It was concluded that intermolecular H-bond opens a rapid nonradiative deactivation channel to the trans-formed ground state by transferring the energy to the solvent bath through the H-bond.30

In the case of indigo, which is only weakly fluorescent, intramolecular H-bonds are suggested to provide a channel for the rapid radiationless deactivation.32,33

For 4,4′,7,7′-tetramethylindigo in chloroform, the fluorescence lifetime was only 30 ps and the overtones of the NH stretching mode and the out-of-plane deformation mode were considered to be the decay channels for the rapid internal conversion.23

Another H-bonded system that exhibits rapid nonradiative decay without formation of any detectable intermediate is the intermolecular H-bonded 1-pyrenol-pyridine system.34

It was concluded that, immediately after the electron-transfer form 1-pyrenol to pyridine, a large scale and ultrafast proton shift takes place which induces an ultrafast nonradiative crossing to the ground state.

Hydrogen-bonding is one of the most important topics in chemistry and in molecular biology.

H-bonds in the excited state are known to mediate not only proton transfer but also electron transfer.35,36

However, the role of H-bonds in energy deactivation and dissipation is not well understood.

The ultrafast deactivation channel of indigo can be utilized as a model of energy dissipation and related structural reorganization in much complicated H-bonding systems.

It is needless to say that H-bonds determine the structure and dynamics of protic liquids like water as well as of biopolymers like protein and DNA.

Here we report a study on ultrafast deactivation process of indigo carmine (InC) to elucidate the relation between deactivation dynamics and intra-/intermolecular H-bonding.

InC is a trans-indigo derivative which can form intramolecular double H-bonds as shown in Scheme 1(b).

We have chosen this molecule because of its high solubility in protic solvents compared to indigo.

It is also reported that InC forms solute/solvent complexes with methanol and water because of its solvent dependence of the absorption and emission spectrum.37

Solvent dependence of the excited state lifetime was carefully examined by ultrafast time-resolved spectroscopy.

Steady-state resonance Raman measurement was also carried out to elucidate the solvent dependence of the molecular structure of InC.

Experimental section

Cr : forsterite laser system with 25 fs time-resolution

The details of the home-made cavity-dumped Kerr lens mode-locked Cr : forsterite laser and the femtosecond pump–probe measurement setup were reported elsewhere.38

The repetition rate of the cavity-dumping was 100 kHz and the output was focused into a LBO crystal to generate the second harmonic centered at 630–640 nm.

The second harmonic pulse energy was 4 nJ and the FWHM was calculated to be 27 fs.

The beam was divided into pump and probe pulses by a 50% beam splitter.

The energy of the pump pulse was 0.6–0.8 nJ and that of the probe pulse was attenuated to 60–80 pJ by a ND filter.

The pulses were focused into the sample by a 10 cm focusing lens.

The pump–probe signals were detected at a magic angle polarization with a photodiode coupled to a lock-in amplifier.

The sample path length was 0.5 mm and the absorbance of the sample was set to 1.0 at ∼610 nm, i.e., the peak of the absorption.

In this experiment, signal intensity was measured in terms of transmittance difference, ΔT.

The measured ΔT was regarded to be equivalent to absorbance difference, ΔAbs, because the signals were very weak, i.e., the ΔAbs was less than 10−4 for our experimental condition.

Ti : sapphire laser DOPA system with 120 fs time-resolution

Transient absorption measurement with 120 fs time-resolution was carried out utilizing a dual OPA femtosecond laser system based on a Ti : sapphire laser.

The details of the system were described elsewhere.39

Briefly, the output of a femtosecond Ti : sapphire laser (Tsunami, Spectra-Physics) pumped by SHG of cw Nd3+ : YVO4 laser (Millennia V, Spectra-Physics) was regeneratively amplified with 1 KHz repetition rate (Spitfire, Spectra-Physics).

The amplified pulse (1 mJ/pulse energy and 85 fs fwhm) was divided into two pulses with same energy.

These pulses were guided into two OPA systems (OPA-800, Spectra-Physics) or into a 1 cm fused silica cell filled with H2O/D2O (3 : 7) mixture for super-continuum generation.

The wavelengths of the OPA output pulses were converted by SHG, THG, FHG, or by sum-frequency generation with the fundamental 800 nm pulse.

The output pulses were obtained with energy of 1–10 mW, fwhm of ∼120 fs, and a wavelength range that covered 300–1200 nm.

When two OPA systems were utilized, one of the energy of the output pulse was attenuated to <1/5000 and utilized as a probe pulse.

The pulse duration at the sample position was estimated to be 120 fs from the cross correlation trace at the same position.

The intensity of the probe, reference, and the pump pulses were monitored simultaneously by photodiodes and sent to a microcomputer for further analysis.

When the super continuum was utilized as the probe pulse, both the pump and the probe pulses were passed through an optical chopper rotating with a frequency of 100 Hz to reduce the optical damage of the sample.

The signal and the reference pulse were detected with two pairs of a monochrometer and a MCPD.

The obtained transient spectra were calibrated for group velocity mismatch.

Sample cell with an optical length of 2 mm was utilized and the optical density of the sample was set to ∼1.0.

Picosecond YAG laser system with 15 ps time-resolution

Picosecond laser photolysis system with a mode-locked Nd3+:YAG laser was utilized for transient absorption measurement.40

The second harmonic at 532 nm with 15 ps fwhm and 0.5–1.0 mJ of power was utilized for excitation.

The excitation pulse was focused into a spot with a diameter of ca. 1.5 mm.

A picosecond white continuum generated by focusing the fundamental pulse into a 10 cm quartz cell containing H2O/D2O mixture (1 : 3) was employed as a probe pulse.

The white continuum was divided into two beams by a half-mirror and the reflected beam was monitored by a monochrometer and a MCPD.

The penetrated beam was utilized as a probe beam and also detected by a monochromator and a MCPD.

To cancel the effect of molecular rotational diffusion, the polarization of the white continuum was changed from linear to circular by a λ/4 plate.

Steady state absorption spectrum, fluorescence spectrum, and Raman scattering measurements

Steady state absorption and emission spectra were measured by Hitachi U-3500 spectrophotometer and Hitachi 850E fluorescence spectrophotometer, respectively.

The optical density at the peak absorption wavelength of the InC solution was 0.25–0.15 for the emission measurement.

Resonance Raman scattering was measured by the 514.5 nm line of an Ar ion laser (STABLELITE 2017, Spectra-Physics) and a CCD (1152F, Princeton Instruments).

The sample concentration was ∼1.0 × 10−3 M and a rotating cell with diameter of 11.7 mm was utilized.

The laser beam was injected into the sample cell from the bottom and the perpendicular scattering was measured.

Indigo carmine (InC, >97.5%) and dimethylsulfoxide (DMSO, >99.7%) were purchased from Kanto chemicals, N,N′-dimethylformamide (DMF, infinity pure grade), methanol (MeOH, infinity pure grade), D2O (>99.5%), and polyvinylalcohol (PVA, MW ≅ 40 000, completely hydrolyzed) were from Wako Pure Chemical Industries, N-methylformamide (MFA, >99%) and formamide (FA, >99%) were from Tokyo Chemical Industries.

H2O was filtered by ion exchange resin.

The purity of InC was checked by measuring the IR absorption spectrum of InC before and after recrystallization, which bared no difference.

The InC/PVA film was produced by mixing 2.0 mg of InC and 1.5 g of PVA in 20 ml of hot water and dropping 2 ml of the solution onto a glass plate.

The sample was dried over night under atmosphere and then kept inside a vacuum desiccator for several days.

The peak optical density of the film was about 0.7–1.0.

Results and discussions

Steady-state absorption and fluorescence spectra

The steady-state absorption and fluorescence spectra of InC in various solvents are shown in Fig. 1.

The electronic spectrum of InC shows a modest solvatochromic shift with virtually similar shape.

The absorption, λabs, and fluorescence peak wavelengths, λflu, and Stokes shift, νStokes, are listed in Table 1, respectively.

The value of νStokes increased in the order of DMF < DMSO < MFA < FA ≅ MeOH < H2O which is in good agreement with the previous report.37

It was reported that plot of νStokesvs. the Mataga–Lippert polarity function, f(D,n), exhibits some degree of linearity while that of νStokesvs. ET(30) exhibits more linearity.

ET(30) is an empirical solvent polarity parameter that considers both solvent polarity and H-bonding influences on solvatochromic shifts.41,42

These solvent parameters are also listed in Table 1.

The chromophore of InC is composed from two pairs of adjacent CO and NH group separated by the central CC bond that form a H-shaped structure (Scheme 1).

The molecular orbital calculations suggest that charge-transfer takes place upon photo-excitation from the NH group to the CO group.32,43,44

However, the dipole contribution to the Stokes shift should be minor because InC possesses a quadropole moment, if the C2h symmetry of the molecule is perfect.

The finite Stokes shift suggests the breakdown of the symmetry upon photo-excitation, which may be caused by the twisting around the central CC bond or the out-of-plane deformation.

For N,N′-diacetylindigo (X = N–COCH3 in Scheme 1 (a)), X-ray crystallography shows that the molecule is not completely flat even in the ground state.43

Moreover, poor linearity of νStokesvs. f(D,n) suggests that the Stokes shift and the solvatochromism are not the results of a simple dipole solvation but also involves structural reorientation arising from the H-bonding to the solvent molecules as the ET(30) correlation suggests.

The electrostatic repulsion between the CO groups is also utilized to explain the blue shift of the absorption spectrum upon transcis isomerization, i.e., instability of the cis-configuration compared to the trans-configuration is much larger in the S1 state than in the S0 state which leads to a larger vertical S1 ← S0 transition energy.44

Note that indigo derivatives that exhibit efficient transcis isomerization have atomic groups (see Scheme 1: X = S, O, Se, N–COCH3,) that posses weaker electron donating ability than that of indigo (X = NH).

N,N′-dimethylindigo (X = N–CH3) also exhibits transcis isomerization although its quantum yield is low, 0.08, and the lifetime of the cis-form is rather short, a few seconds.22

Moreover, isomerization of thioindigo is known to take place through T1 state,15–21 and avoids the electrostatic repulsion in the S1 state.

In the case of oxoindigo with less charge-transfer character, the quantum yield of the transcis isomerization is reported to be 0.63 and trans-oxoindigo is non-fluorescent, indicating a direct isomerization from the S1 state.8

Usually cis-form is more polar than the trans-form because of the lower symmetry.

This property is utilized to crystallize cis-form of N,N′-diacethylindigo by inducing transcis photoisomerization in non-polar solvent.45

For indigo, polar solvation may not be enough to reduce the electrostatic repulsion of the carbonyl groups to stabilize the cis-form.

Transient-state absorption spectra

On the contrary to the ground state absorption spectra, the band shapes of the transient absorption shown in Figs. 2–4 depended drastically on solvents.

Immediately after the photo-excitation, transient absorption with maximum at ∼465 nm and 635–645 nm appeared in DMF and MeOH solution, which were assigned to the Sn ← S1 absorption.

The absorption peak at ∼645 nm in DMF was slightly blue shifted to ∼635 nm in MeOH.

These transient absorptions are similar to the Sn ← S1 absorption of thioindigo which peaks at ∼470 nm and ∼600 nm.21,46,47

Because no bleaching can be seen around 640 nm where the S1 ← S0 and Sn ← S0 absorption overlap in DMF, extinction coefficient of the Sn ← S1 transition was concluded to be larger than that of the S1 ← S0 transition.

The shoulder appearing around 605 nm in the DMF spectrum can be caused by the bleaching of the S1 ← S0 absorption with its maximum at 618 nm.

The negative signal appearing in the region of >700 nm is assigned to the stimulated emission.

In aqueous solution, the lifetime of the transient absorption was extremely short and the measurement was carried out utilizing the femtosecond Ti : sapphire DOPA system.

In Fig. 4, the transient absorption revealed that the absorption at ∼640 nm is weak compared to those in organic solvents.

Negative signal can be seen at 610–640 nm which is safely ascribed to the bleaching of the S1 ← S0 absorption.

These results indicate that the Sn ← S1 absorption in H2O is broader and blue shifted compared to those in the organic solvents.

Note that the transient absorption around 500–550 nm is relatively stronger than that in the organic solvents.

The excited state deactivation dynamics

Fig. 5 shows the kinetics of the ultrafast part (<1.5 ps) of the transient absorption signal pumped and probed at 635 nm with time-resolution of 30 fs.

In addition to the solution phase, the temporal profile in PVA polymer film was also measured.

In the early stage after the excitation, the signals are strongly modulated by the coherent intramolecular vibrations caused by the resonant impulsive stimulated Raman scattering process.

The vibrational dephasing times were 0.5–0.8 ps, which did not strongly depend on solvent, and the vibrational frequencies and their assignments will be discussed later.

Note that the phase of the oscillation is similar in all solvents, including H2O where the signal possesses negative intensity.

The negative signal observed in H2O corresponds to the bleaching of the ground state absorption as can be seen in Fig. 4.

Fig. 6 shows the picosecond decay of the transient absorption pumped and probed at 635 nm in various solvents except H2O. The time constants listed in Table 2 were obtained by multi-exponential fitting of the decays shown in Figs. 5 and 6.

The intense coherent oscillation and overlap of the ground state bleach and transient absorption prevent the rational analysis of the dynamics around the time origin.

The fitting range was set to ≥53 fs, therefore, any ultrafast dynamics that occurred within the pulse duration were neglected.

In the organic solvents, three components including two with negative amplitudes were necessary to fit the signals satisfactorily.

It can be seen in Fig. 5 that the excited state absorption rapidly increased in the organic solvents.

The time constants for the rise, τ1 and τ2, were not very sensitive to the solvent property, although in PVA, the rise was significantly slow compared to other fluid solvents and can be observed in Fig. 6.

Because the probe wavelength of 635 nm is near the peak of the Sn ← S1 absorption spectrum, the rise component corresponds to the narrowing of the transient absorption caused by a conformational relaxation in the excited state.

It can be noticed in Fig. 6 that the decay profile in DMSO is identical to that in DMF and those in MeOH and FA are parallel.

The longest time constant, τ3, in organic solvents and PVA was assigned to the lifetime of the S1 state.

The fluorescence lifetime of InC in DMF was reported to be 110 ps48 which is in good agreement with our present value of 92 ± 0.4 ps.

The ultrafast time profile in aqueous solution was rather complicated and needed four components to fit the data satisfactorily.

To elucidate the accurate S1 lifetime in aqueous solution, the decay of the transient absorption and the recovery of the ground state bleach was probed at 470 nm and 610 nm, respectively, with excitation at 585 nm (Fig. 7).

The time constants of the decay and the rise measured at 470 nm and 610 nm were 3.9 ps and 4.0 ps, respectively; thus, the average of these two were regarded as the S1 lifetime of InC in aqueous solution.

The dynamics taking place faster than 4.0 ps in aqueous solution is also related to the relaxation in the excited state.

Interestingly, from Tables 1 and 2, it can be seen that the lifetime of InC decreases in solvents with larger Stokes shift.

These observations indicate that ultrafast structural reorganization and solvent reorientation have a strong influence on the deactivation process.

Thus, we have carried out experiments on deuterium isotope effect and viscosity effect.

Provided that the H-bonding is involved in the deactivation process, deuteration could cause a dynamical influence.

The time dependence of the ground state bleach signal pumped and probed at 635 nm in H2O was compared with that in D2O in Fig. 8.

The PP signals showed ultrafast decay into the negative amplitude regime and it reached the minimum value at ∼250 fs.

After the decay, the signals recovered multi-exponentially, and the PP signal overshot to positive value at ∼4 ps.

To fit the time-dependence of the signal, four components, including two with negative amplitude, were necessary (Table 2).

The signal exhibited an ultrafast decay of ∼200 fs which was not observed in other organic solvents.

The solvation dynamics of aqueous solution is known to be extremely fast, i.e., inertial response with time constant of ∼30 fs.49

It is clear from Fig. 8 that the negative bleach signal in H2O recovers to zero faster than that in D2O. This can be the evidence of the H-bond influencing the ultrafast structural reorganization and the deactivation process.

The amino-hydrogen of InC can be easily exchanged to deuterium by simply dissolving in D2O. Therefore, this could be either the effect of intra- or intermolecular H-bond.

It should be noted that the viscosities of H2O and D2O are slightly different, 1.00 cP and 1.25 cP at 20 °C and this difference also affects the deactivation process in some cases.50

The frequencies and dephasing times of the coherent oscillation observed in the signal were similar in both solvents.

The longest lifetime in highly viscous PVA (260 ± 10 ps) and slower dynamics in D2O compared to that in H2O indicate that the S1 lifetime may depend on the fluidity of the solvent.

To examine this presumption, the S1 state lifetime was plotted against viscosity in Fig. 9(a), which exhibited no correlation.

The lifetime in FA with viscosity of 3.3 cP (22 ± 0.2 ps) was the same as that in MeOH with viscosity of only 0.54 cP.

On the other hand, the lifetime in DMF with viscosity of 0.8 cP was as long as 92 ± 0.4 ps.

The lifetime is shorter in protic solvents than those in aprotic solvents.

Thus, the lifetime was plotted against ET(30) in Fig. 9(b) which showed a good correlation.

See Table 1 for the values of the solvent parameters.

ET(30) is an empirical measure of solvent polarity which also depends on H-bonding ability of the solvent.41,42

It was reported that the value of the Stokes shift also had a good linear relation with ET(30).37

When lifetime is compared with the Mataga–Lippert polarity function, f(D,n), the correlation was not as good as ET(30).

The f(D,n) of MeOH (0.324) is close to that of H2O (0.326), although the lifetimes were rather different, ∼22 ps and ∼4 ps, respectively.

The lifetime in MeOH was the same as that in FA (∼22 ps) with rather different value of f(D,n) which is 0.287.

Correlation can be also found for the solvent H-bond donor acidity parameter, α, which describes the ability of the solvent to donate a proton in a solvent-to-solute H-bond.51

The value of α is as large as 1.17 in water, whereas, those of DMF and DMSO are 0.00.

On the contrary, there was an opposite correlation with the average solvation time, 〈τ〉, obtained from the dynamic Stokes shift measurement of coumarin .15352

The values of 〈τ〉 for MeOH and FA were both 5.0 ps, while those in DMSO and DMF were shorter, i.e. 2.0 ps.

These results demonstrate that H-bond ability of the solvent plays a crucial role rather than the polarity.

Intramolecular vibrations of InC

Because the ultrafast part of the transient absorption signal of InC was modulated strongly by intramolecular vibrations, the Fourier transform was performed on the signals shown in Fig. 5 and the real parts of the Fourier transformed spectra are shown in Fig. 10.

The strongest mode was observed at ∼260 cm−1 in protic solvents and interestingly this mode became a doublet, ∼254 cm−1 and ∼265 cm−1, in aprotic solvents.

In our femtosecond measurement, coherent oscillation can either result from the electronic ground or excited state, because both absorptions are overlapped at the prove wavelength.

To confirm their origin, resonance Raman scattering measurement was carried out in various solvents with an excitation wavelength of 514.5 nm and the results are shown in Figs. 11 and 12.

In the low-frequency region (Fig. 11), a mode appeared at ∼260 cm−1 in protic solvents which became a doublet, 255 cm−1 and 270 cm−1, in aprotic solvents, in good agreement with the Fourier transformed spectra.

In the near-IR Fourier transform Raman spectrum of solid state indigo, three modes are observed in this frequency range, 275 cm−1 (δOC–CC), 266 cm−1 (δN–CC), and 252 cm−1 (γCC), which are assigned to in-plane and out-of-plane bending modes.53

If the intramolecular H-bonds are replaced by the intermolecular ones, slight twisting or out-of-plane deformation of the molecule can occur.

Therefore, the molecular structure will be thermally distributed along the twisting/deformation coordinate and result as a broadening of the Raman bands.

Other mode that showed strong solvent dependence was the one at 613–617 cm−1.

This mode was sharp and comparatively strong in aprotic solvents whereas it was broad and weak in protic solvents.

For indigo, out-of-plane N–H motion was located at 635 cm−1 (Raman) and 633 cm−1 (IR).53

If the mode at 613–617 cm−1 of InC can be assigned to the same mode, this also suggests that molecular planarity has declined in protic solvents.

However, in high-frequency region, 1200–1800 cm−1 (Fig. 12), we were not able to observe any notable solvent dependence.

The most important modes for indigo in this region are the CC and CO stretching modes, i.e., 1582 cm−1 and 1701 cm−1, respectively.53

For InC, strong mode can be observed at ∼1582 cm−1 and ∼1705 cm−1.

The frequency of these modes may be slightly higher in protic solvents although it is hardly noticeable.

It seems that exchange of intra- to intermolecular H-bond does not significantly change the conjugated double bonding structure of InC. Usually it is known that intramolecular H-bond that forms a five- or six-membered ring is rather strong, because of the good overlap of the molecular orbitals, and the frequency of the hydrogen accepting CO stretching mode is reduced by 10–100 cm−1.29

However, symmetric ν6 CO stretching mode of indigo observed in Raman spectrum is not significantly lower than other non-H-bonding indigo derivatives.

In solid state near-IR Fourier transformed Raman spectra, the frequency was 1701, 1674, and 1660 cm−1 for indigo, thioindigo, and selenoindigo, respectively (the frequency of indigo was the highest).

In solution, the frequency was reported to be 1705, 1688, and 1714 cm−1 for indigo, N,N′-dimethylindigo and N,N′-diacetylindigo, respectively.54

For the IR active out-of-phase ν61 CO stretching mode, the frequency of indigo was notably lower than that of other non-H-bonding indigo derivatives, i.e., 1627, 1638, 1690, 1642, 1656, 1692, and 1724 cm−1 for indigo, N,N′-dimethylindigo, N,N′-diacetylindigo, selenoindigo, thioindigo, oxoindigo, and dehydroindigo, respectively.11,29,54

Usually this is treated as the evidence of double intramolecular H-bonds, however, it should be noted that, in crystal, indigo is considered to form intermolecular H-bonds with each other.

The shifting and the broadening of the absorption spectrum in solid phase are regarded as the evidence of the self-association of indigo.13

The NH stretching mode appears as a rather broad band at 3268 and 3270 cm−1 in IR and in Raman spectrum, respectively, and the presence or absence of the H-bond is difficult to judge.53

To investigate the nature of the H-bonding, temperature dependent measurement of the Raman spectrum of InC in DMF was carried out (Fig. 13).

Usually the bonding energy of H-bond is in the order of 3–5 kcal mol−1 which can be partially broken by increasing temperature.

However, noticeable temperature dependence was not observed for the CC and CO stretching modes of InC in the range of 278–363 K (Fig. 13(a)).

The relative intensity of the CO stretching mode of InC at 1703 cm−1 was insensitive to temperature and no new bands appeared.

There are solvent bands at 1630 cm−1 and 1730 cm−1 which can disturb the observation of the temperature dependence.

Note that the solvent band at 1630 cm−1 is shifting to higher frequency with increasing temperature.

We also carried out similar temperature dependent experiment in aqueous solution and no intensity change or appearance of a new band was observed.

If H-bonding and non-H-bonding molecules are in chemical equilibrium, CO stretching is likely to appear as a doublet.

However, the observed CO stretching modes were all singlet indicating that chemical equilibrium is strongly one-sided in the studied temperature range.

There may be slight temperature dependence for the solvent sensitive low-frequency bands, i.e., the band at 255 cm−1 slightly down shifted to 252 cm−1 and the intensity of the band at 270 cm−1 seemed to be decreasing at high temperatures, although the changes were close to the limit of our experimental accuracy.

From the solvent dependent low-frequency Raman modes, it can be concluded that molecular planarity depends on solvent property, although from the constancy of the CC and CO stretching modes, the twisting or the out-of-plane deformation is not as significant as to disturb the conjugated CC structure of InC. The observation of only a single CO stretching mode in all the studied solvents and temperatures rises a question to the existence of intra- or intermolecular H-bonding in the ground state.

Even if there were an interaction between CO and NH groups or with the solvent, it should be a feeble one in the ground state.

Temperature dependence of the fluorescence intensity

Temperature dependence of the fluorescence intensity was measured and shown in Fig. 14.

The Arrhenius plot reveals the activation energy for the radiationless deactivation to be 8.8 kJ mol−1 and 3.5 kJ mol−1 in DMSO and H2O, respectively.

The deactivation is a barrier-crossing process and the activation energy in protic solvents was reduced to less than half the value of that in aprotic solvents.

These activation energies are similar to that expected for H-bond breaking.

We assume that slight twisting or out-of-plane deformation lead to the deactivation of the excited InC. In the case of aprotic solvents, intramolecular H-bonds exist in the excited state, which interrupt the twisting/deformation process, whereas such interruption is weakened in the protic solvents and the activation barrier is reduced.

Conclusions and the deactivation scheme proposed

From the steady state absorption and fluorescence spectroscopy, poor linearity of νStokesvs. f(D,n) suggests that the Stokes shift and the solvatochromism are not the results of a simple dipole solvation but also involves structural reorientation arising from the H-bonding to the solvent molecules as the ET(30) correlation suggests.

From the steady state resonance Raman spectrum, it was observed that the low-frequency bands which can be assigned to the in-plane and out-of-plane bending modes, ∼255 cm−1 and ∼270 cm−1, respectively, became broad and indistinguishable at ∼260 cm−1 in protic solvents.

This indicates the reduction of molecular planarity in protic solvents.

However, the CO stretching at ∼1705 cm−1 did not show any solvent or temperature dependence.

On the contrary to the ground state absorption spectrum, the transient Sn ← S1 absorption spectrum in the aqueous solution was quite different from those in the organic solvents.

The ultrafast dynamics in D2O was noticeably slower than that in H2O which suggests the involvement of H-bonding in the deactivation process.

Elsaesser et al. concluded from their time-resolved IR absorption experiment that the NH stretching in the S1 state was shifted to lower frequency by 40 cm−1 compared to the ground state.23

This also supports the formation of the H-bonds in the excited state.

The deactivation process was strongly solvent dependent, and it was faster in protic solvents, especially in H2O, and the lifetime had a linear correlation with the empirical solvent polarity parameter, ET(30).

The deactivation was a barrier crossing process and the activation energy was lower in protic solvents.

From these observations, we propose the following deactivation scheme.

The H-bond interaction is weak in the ground state whereas it is strong in the excited state.

The central CC bond will weaken upon photo-excitation by the increased anti-bonding nature of the molecular orbital in the excited state which leads to the loss of molecular planarity.

Loss of the planarity opens a channel to the ultrafast radiationless deactivation.

However, in aprotic solvents, intramolecular H-bonds retain the molecular planarity and disturb the deactivation process.

In protic solvents, intermolecular H-bonds take place of the intramolecular ones and the activation barrier for the radiationless deactivation is reduced.

The rise components observed in ultrafast kinetics of the transient absorption (Fig. 5) could be the formation of the intra-/intermolecular H-bonds and the related conformational reorganization.

Dissipation of vibrational energy through the solute-solvent H-bond may also play a role in the deactivation process.

It is reported for betaine 30, the probe molecule utilized to determine the ET(30) values of solvents, that the vibrational energy is transferred to solvent through the H-bond.55

The cis-form of indigo is energetically unstable than the trans-form and energy barrier does not exist between these two isomers.

It is known that thermal transcis isomerization of N,N′-dimethylindigo can be catalyzed by acid22 indicating a low energy barrier in protic solvents.

Therefore, the deactivation process only leads back to the trans-form.