Proton‐Transfer Dynamics of Photoacidic Merocyanines in Aqueous Solution

Abstract

Photoacids attract increasing scientific attention, as they are valuable tools to spatiotemporally control proton‐release reactions and pH values of solutions. We present the first time‐resolved spectroscopic study of the excited state and proton‐release dynamics of prominent merocyanine representatives. Femtosecond transient absorption measurements of a pyridine merocyanine with two distinct protonation sites revealed dissimilar proton‐release mechanisms: one site acts as a photoacid generator as its pK_a value is modulated in the ground state after photoisomerization, while the other functions as an excited state photoacid which releases its proton within 1.1 ps. With a pK_a drop of 8.7 units to −5.5 upon excitation, the latter phenolic site is regarded a super‐photoacid. The 6‐nitro derivative exhibits only a phenolic site with similar, yet slightly less photoacidic characteristics and both compounds transfer their proton to methanol and ethanol. In contrast, for the related 6,8‐dinitro compound an intramolecular proton transfer to the ortho‐nitro group is suggested that is involved in a rapid relaxation into the ground state.

Transient absorption studies of protonated pyridine‐ and nitro‐merocyanines reveal a pH‐gated photoacidic behaviour. The O‐protic site shows an ultrafast excited state super‐photoacid reactivity at low pH values that is even capable of proton transfer to alcoholic solvents. Remarkably, the pyridine compound exhibits an additional N‐protic site which functions as photoacid generator at pH 5.5 due to its lowered ground state pK_a value after photoisomerization to the ring‐closed spiropyran structure.

Introduction

Photoacids (PAs) turn into strong acids upon irradiation and thus enable light‐stimulated proton dissociation. They steadily gain scientific attention, as they provide a convenient way to convert an optical input into a desired chemical response. If chosen properly, photoacids or photobases can in principle exert control over any proton‐ or pH‐driven system with light as an outstanding external stimulus with unrivalled spatiotemporal precision. As a versatile trigger for proton‐mediated processes, they can be exploited to regulate protein activity,[¹, ²] to design photoelectric circuits[³, ⁴] and to advance functional materials.[⁵, ⁶, ⁷]

The generic term photoacid originally refers to compounds whose acidity is significantly enhanced when promoted to their electronically excited state (Scheme 1A). This entails an ultrafast proton dissociation or rather an excited state proton transfer (ESPT) to the solvent. First described as Förster cycle^[8] this mechanism implies that the released proton is short‐lived and reassociates thermally when the excited species relaxes back to its less acidic ground state.^[9] An essential feature of a PA is its excited state dissociation constant pK_a*, which is correlated with the proton transfer rate. The pK_a* can be estimated by the Förster cycle model, while the proton transfer rates are typically determined by time‐resolved spectroscopic techniques. In this context, various kinds of aryl‐OH compounds are among the most extensively investigated photoacids. Examples like phenol or 1‐naphthol (1N) exhibit lowered but still positive pK_a* values of 3^[10] and 0.5,^[11] respectively. In this acidity range 8‐hydroxypyrene‐1,3,6‐trisulfonic acid (HPTS) also known as pyranine[¹², ¹³] is a prominent representative with a pK_a*=1.3. The proton transfer process can take up to nanoseconds in aqueous solution for such photoacids with positive pK_a* values. Though, photoacids that adopt negative pK_a* values (<−1) were termed super‐photoacids^[14] and are even able to transfer their proton to alcohols within a few picoseconds. In several pioneering studies Tolbert and co‐workers reported a series of 1N and 2‐naphthol (2N) derivatives, where 5,8‐dicyano‐2‐naphthol (DC2N) was found to be the strongest PA among them with a pK_a* =−4.5.[¹¹, ¹⁴, ¹⁵, ¹⁶] Further related and even more photoacidic examples are hydroxyquinolinium compounds, which adopt pK_a* values of ∼−7. Topp,^[17] Ernsting^[18] and Solntsev^[19] reported the ESPT dynamics of N‐methyl‐6‐hydroxyquinolinium (NM6HQ) and showed that proton dissociation occurs in a solvent‐controlled manner within an excited state lifetime of 2–3 ps in aqueous solutions. More recently, Huppert et  al. investigated several quinone cyanine dyes[²⁰, ²¹] (QCy), which are typically used as fluorophores for super‐resolution microscopy. The highly conjugated dye QCy9 for example, exhibits a pK_a* value as low as −8.5 and is thus the strongest photoacid studied so far.^[22]

Scheme 1.

Generalized proton‐release reaction schemes A) of excited state photoacids (PA) following the Förster cycle, and B) of photoacid generators (PAG). C) Typical proton‐release reaction of merocyanine‐spiropyran photoswitches.

Following a mechanism different from ESPT, a proton‐release can also occur as a result of a photoreaction cascade which makes the respective compound a photoacid generator (PAG, Scheme 1B).^[23] After activation, PAGs allow for a high increase in proton concentration, but often irreversibly. They have been successfully utilized to initiate acid‐catalyzed reactions^[24] or cationic polymerization,^[25] to create pH jumps^[26] or photoacidic polymers.^[27] They have also been applied as photoresists in microlithography.^[28]

To this end, using photoswitches as photoacidic compounds can be very advantageous since they allow for a reversible operation of their protonation state, regardless of the nature of the photoacid itself. The reassociation of the proton is dictated by the isomerization of the photochromic compound and can be driven either thermally, photochemically or both.^[29] Moreover, the (de‐)protonation itself can essentially affect the properties of the respective photoswitch which can be applied to design multistimuli‐responsive molecular logic systems.[³⁰, ³¹] Therefore, merocyanine‐based photoacids, in particular benzo‐indolino‐pyrano‐spiran (BIPS)[³², ³³] derivatives have been studied extensively for miscellaneous applications.[³, ³⁴, ³⁵, ³⁶, ³⁷, ³⁸]

Under acidic conditions, the ring‐opened merocyanine form is protonated at its phenolate moiety (Scheme 1C). This proton is released upon photoisomerization towards the spiropyran structure and the compound thermally reverts to the protonated ring‐opened form in the dark.[³⁹, ⁴⁰, ⁴¹, ⁴², ⁴³, ⁴⁴] The pK_a value of this protonation site can be tuned by the choice of substituents on the respective phenyl ring (R₂ in Scheme 1C). For an unsubstituted ring a pK_a value of 7.8^[39] is reported, while, for example, attachment of electron‐withdrawing functionalities like an NO₂‐ or a CN‐group in para‐position drastically lowers the pK_a to 3.7 and 4.4, respectively.^[35] Different proton‐release pathways were postulated like an initial cis/trans photoisomerization of the protonated merocyanine followed by proton dissociation and ring‐closure. Also the formation of a highly acidic spiropyran protonated on the indoline nitrogen has been discussed.[³¹, ⁴²] At very low pH values, an intramolecular transfer of the phenolic proton to the indoline nitrogen upon ring closure was suggested, where the latter protic site adopts a pK_a value below 1.6.^[35] Yet, it has not been clarified whether the proton‐release occurs from the excited merocyanine prior to the ring‐closing sequence (excited state PA) or as consequence of the photoreaction due to the formation of an acidic photoproduct (PAG).

We previously investigated the photochromic properties of several pyridine derivatives of the parent BIPS compound, which we referred to as Py‐BIPS.[⁴⁵, ⁴⁶, ⁴⁷] Originally, the pyridine nitrogen of Py‐BIPS‐type compounds was methylated and therefore positively charged. More recently, we reported on an unmethylated Py‐BIPS derivative, which shows refined photo‐ and acidochromic behaviour in aqueous solution.^[48] The merocyanine isomer of this compound 1 and its protonated states are shown in Scheme 2A.

Scheme 2.

Nomenclature of protonated species of the investigated A) Py‐BIPS derivative 1 and B) Nitro‐BIPS‐type merocyanines 2 and 3. C) Illustration of merocyanine cis (C)/trans (T) ‐isomers, referring to the configurations of the central double bond bridge. Rotation around the angles α, β and γ may occur thermally or photochemically. Among the eight possible isomers, those four with a T configuration at the middle bond are stable and the shown TTC and TTT (to a minor extent) are typically present in solution.

Due to the absence of the methyl group, merocyanine 1 exhibits a slightly acidic protonation site on its pyridine nitrogen (pK_a ^I=6.8), in addition to the phenolic site. Interestingly, the incorporation of the pyridine moiety lowers the pK_a ^II value of the phenolic site to 3.2, which is even lower than the values of Nitro‐BIPS derivatives. The pK_a of Nitro‐BIPS 2 is 3.7^[35] and that of Dinitro‐BIPS 3 is 3.9^[49] (Scheme 2B). Upon exposure to light, each of the shown protonated merocyanines is capable of photoinduced ring‐closure which is accompanied by proton dissociation. In addition to protonated states, the existence of several cis/trans‐isomers (the configurations of the three central double bonds are typically denoted by T for trans and C for cis) should be considered when investigating merocyanine compounds spectroscopically (Scheme 2C).^[32] Eight isomers are possible in principle, but only those with a central T bond are stable ones. It is known that the two most stable isomers – mainly TTC and a minor fraction of TTT – exist in an equilibrated solution. The absorption band of TTT is shifted bathochromically with respect to the dominant TTC isomer. For Nitro‐^[50] as well as Dinitro‐BIPS^[51] derivatives distinguishable photoisomerization pathways were assigned for the two main isomers in organic solvents, which might also be true for protonated merocyanines in general.

Here we present the first time‐resolved investigation of the proton‐release dynamics of photoacidic merocyanines by means of UV/vis‐transient absorption (TA) spectroscopy in the fs‐ns time window. We address the proton‐release reactions of both protonation sites of compound 1 and of the phenolic sites of compounds 2 and 3 in order to shed light on the underlying mechanisms. The observed dynamics allow for a clear discrimination between photoacid generation and excited state photoacidity and provide clear evidence that the phenolic protonation site of the photoacidic merocyanines functions as a super‐photoacid. Comparison with additional experiments in protic organic solvents give further insights into the ESPT dynamics of excited state PAs and the photoprotolytic cyle in general.

Results and Discussion

Photodynamics of MC and HMC of compound 1

The compounds MC and HMC (Scheme 2A) were investigated at pH 7.4 and 5.5, respectively, to ensure the maximal amount of the particular isomers. The photoinduced ring‐closure reaction causes the pK_a ^I value of the pyridine nitrogen to shift from 6.8 down to 4.8 for the spiropyran structure. Due to this outstanding side effect of photoswitching, we could realize a light‐controlled steady‐state pH value regulation of aqueous solution in a pH range roughly from 4.5 to 7. A pH drop of about 1.5 units was monitored upon exposure of an equilibrated HMC sample to visible light (520 nm) and the initial pH value was recovered within 5 minutes in the dark.^[48] A comprehensive summary of the photophysical acidochromic properties of compound 1 is provided in the Supporting Information (see Figure S1). However, the fact that a significant persistent pH drop can be observed, implies that a thermal reassociation of the released proton is hindered after photoisomerization to the ring‐closed structure. Thus, the observable pH drop apparently relies on the change of acidity of the peculiar N‐protic site and the reprotonation is only feasible via ring opening. It remains unclear though, whether the proton is released as consequence of ring‐closing or due to an increased excited state acidity of HMC* (excited states are indicated by asterisks).

The time‐dependent spectral evolution during the first 1.5 ns of the ring‐closing sequence of MC and MCH upon excitation with 520 nm at pH 7.4 and 5.5, respectively, is shown in Figure 1. The contour plots display the absorption difference relative to the ground state absorption. Hence, positive signals can be assigned to excited state absorption (ESA) or to the absorption of emerging intermediate species. Negative signals appear because of the ground state bleach (GSB) upon electronic excitation and due to stimulated emission (SE) induced by the probe pulse. In the transient map of MC (Figure 1A, left panel), a positive signal at 425 nm (ESA₁) and a more pronounced one around 505 nm (ESA₂) is observed. Both appear directly upon excitation and can thus be ascribed to the excited S1 state MC*. Moreover, the negative SE signal around 620 nm resembles the steady‐state emission of MC and the GSB is visible around 540 nm, where the ground state absorption band is centred (Figure S1). All of the excited state signals (both ESAs and SE) decay on a similar timescale and are almost vanished after 100 ps.

Figure 1.

A) TA contour plot of MC of compound 1 at pH 7.4 (left panel) and corresponding DAS (right panel). B) TA contour plot of HMC of compound 1 at pH 5.5 (left panel) and corresponding DAS (right panel). C) Selected transients showing the time‐dependent progression of the signals ESA₁ (blue) and ESA₂ (cyan) for MC (circles) and HMC (plus signs). Data points are represented by symbols and the fit from GLA by solid lines.

The kinetics observed in the TA spectra were analyzed by fitting the data sets globally with a sum of a given number of exponential lifetime components, which is referred to as global lifetime analysis (GLA).^[52] From this procedure, decay‐associated spectra (DAS) are obtained, which show the wavelength‐dependent amplitudes of the determined lifetimes (Figure 1A and B, right panels). The reading of the DAS is as follows: a positive amplitude captures the decay of a positive absorption difference signal or a build‐up of a negative signal. Inversely, a negative amplitude indicates a decay of a negative or an increase of a positive signal. Analysis of the TA spectrum of MC via GLA revealed that two main time constants are sufficient to describe the observed dynamics adequately. An additional infinite time constant τ_∞ reflects the residual signal at the end of the measurement after 1.5 ns. A schematic depiction of the assumed mechanism is given in Scheme 3A. The determined 1.2 ps lifetime (τ₁) models the solvent reorganization and vibrational relaxation within the S1 state MC* as the sigmoidal shape of its DAS captures a slight blue‐shift of the ESA signals and a dynamic Stokes shift of the SE. However, the slower constant τ₂=22 ps accounts for the decay of the signals associated with the electronically excited state and is therefore ascribed to its lifetime. The large negative contribution of τ₂ around 560 nm indicates that the decay of MC* mostly recovers the initial MC ground state through SE but also non‐radiatively. Last, the infinite time constant τ_∞ reveals that residual GSB is present at the end of the measured time frame due to the formation of the ring‐closed photoisomer SP. Interestingly, there is no indication for the formation of a protonated state upon excitation of MC. Thus, the pyridine group plausibly does not acquire a proton from the solvent when excited as one could expect since related compounds like quinoline or acridine are known to be photobases.[⁵³, ⁵⁴]

Scheme 3.

Photoreaction pathway schemes of Py‐BIPS 1. A) Ring‐closure dynamics of MC at pH 7.4. B) Ring‐closure dynamics of HMC and subsequent proton‐release from the HSP ground state at pH 5.5.

Just as the transient map of MC, the spectrum of HMC (Figure 1B, left panel) exhibits two separated ESA signals, although both are shifted to longer wavelengths by about 30 nm. Thus, the GSB appears in between the positive ESA signals. Similarly, the ESAs shift hypsochromically with time. Apart from the GSB, there is a pronounced negative SE signal around 610 nm. Data analysis indicates two main time constants reflecting an analogous spectral evolution as for MC (Figure 1B, right panel). First, the fast time constant τ₁=1.1 ps models a rapid relaxation within the S1 state HMC*, due to an initial motion away from the Franck‐Condon region and to solvent reorganization. Then, the decay of the excited state is monitored by τ₂, which is significantly accelerated compared to MC* (13 ps vs. 22 ps). At the end of the measured time, again GSB is the only observed residual signal. The significant acceleration of the photodynamics of HMC compared to MC is also evident in view of the transient signal evolution depicted in Figure 1C. Similar dynamics for each of the signals are observed, where it takes approximately 80 ps for the MC* excited state signals to decay thoroughly and only about 40 ps for MCH*. In both cases, the more pronounced ESA₂ seems to increase during the first ps, but this effect can be ascribed to the superposition of the respective ESA and the GSB.

The similarities in the MC and HMC photodynamics hint at analogous ring‐closure pathways for both compounds. Furthermore, the observed dynamics is in agreement with the singlet pathway of the previously reported Py‐BIPS^[45] compound, where the decay of the excited state is further accelerated to 3.2 ps. These observations account for a distinct correlation between the substitution on the pyridine nitrogen and the excited state lifetime of the ring‐open merocyanine. Starting from a mere lone pair on the nitrogen (MC) to a proton (HMC) to a methyl‐group (Py‐BIPS), the electron density on the N‐atom decreases, which enhances the excited state decay. Although the formation of the photoproduct HSP could not be directly monitored here due to limitations of the probe light (broken arrows, Scheme 3), we suggest that ring‐closure proceeds similarly to the parent Py‐BIPS derivative. Hence, upon excitation of the thermodynamically favoured trans‐merocyanine, ring‐closure occurs via a series of cis‐trans isomerizations from the vibrationally hot ground state of the ring‐open isomer.[⁴⁵, ⁵⁵] An indirect indication for these isomerizations is that the remaining GSB signal is slightly shifted with respect to the steady‐state absorption band. The planar trans‐merocyanine converts into a cisoid intermediate which already adopts a perpendicular structure and finally establishes the C−O bond.

In our previous study,^[48] a fluorescence quantum yield of ∼9 % was determined for MC by integrating sphere measurements. Hence, non‐radiative decay such as internal conversion or vibrational relaxation occurs with a ∼91 % probability. Starting from a vibrationally hot MC ground state, ring‐closure is assumed to occur with a yield of ∼5 %, according to steady‐state switching experiments. Compared to MC, the excited state decay of HMC is significantly faster and a lower fluorescence quantum yield of ∼3.5 % was found, which entails a probability of ∼96.5 % for non‐radiative relaxation. Moreover, the quantum yield for photoproduct formation (SPH) is slightly higher at pH 5.5 (∼7 %) than at pH 7.4 (SP). This points to a favoured establishment of the ring‐closed form when the pyridine nitrogen is protonated.

Concerning the proton‐release of HMC it appears that the decay of the excited state is not accompanied by the emergence of an intermediate state. Hence, a dissociation of the proton in the excited state can be ruled out, as this would imply the formation of the excited deprotonated MC*. HMC therefore does not release H⁺ within the excited state but directly converts into the ring‐closed structure HSP (Scheme 3B). This entails that HMC has to be considered as a reversible PAG rather than an excited state PA. The proton is released subsequently after formation of HSP and the observable pH jump therefore merely results from the altered pK_a value of the respective protic site.

Photo‐ and ESPT‐dynamics of HMCH of compound 1

The TA spectrum of the doubly protonated HMCH in aqueous solution at pH 1 is shown in Figure 2A. Several signals are present directly upon excitation – the GSB at 420 nm and two ESA signals around 480 nm (ESA₁) and above 625 nm (not denoted). Moreover, an SE signal is detectable already at early times in between the ESA bands. Besides the GSB, the listed signals correspond to the excited HMCH* species. The positive amplitude of the time constant τ₁=1.1 ps between 450 nm and 500 nm, where ESA₁ is located (Figure 2A, DAS), is indicative of an ultrafast decay of the initial excited species. The pronounced negative amplitude of τ₁ around 530 nm reveals that the decay of ESA₁ is accompanied by the emergence of a second positive signal (ESA₂) at 530 nm. The positive amplitude of τ₁ above 570 nm additionally captures the formation of the main SE₂ signal around 610 nm. Therefore, this time constant corresponds to a transition from HMCH* to a distinct new transient species. Comparison of the transients between 0.2 and 4 ps (Figure 2B) illustrates that the decay of ESA₁ and the increase of ESA₂ and SE₂ indeed occur with one rate. The spectral signature of the state, evolving with the constant τ₁ (ESA₂ and SE₂), coincides with that of the excited single protonated HMC* (ESA₂ and SE, see Figures 1B and S2) when the GSB is considered. The decay of ESA₂ and SE₂ is described by a similar time constant τ₂=13 ps with a DAS that resembles the one found for HMC* which further confirms the assignment of the HMC* state. Hence, the release of the phenolic proton of HMCH* can be directly observed here and evidently occurs in the excited state. This process interestingly takes place on the same timescale as relaxation within the excited state in the cases of MC and HMC. Consequently, the proton is transferred to the solvent rather efficiently even before the HMCH* state is thoroughly relaxed on the S₁ potential energy surface. Competing deactivation channels for HMCH* are non‐radiative decay and fluorescent emission. A fluorescence maximum at 470 nm was observed in the steady‐state spectra and the red edge of this emission band that extends to roughly 600 nm is detected in the shape of SE₁.

Figure 2.

A) TA contour plot of HMCH of compound 1 at pH 1 (left panel) and corresponding DAS (right panel). B) Selected transients showing the time‐dependent progression of the signals ESA₁ (blue), ESA₂ (cyan) and SE (orange). Data points are represented by symbols (crosses) and the fit from GLA by solid lines.

After the proton transfer and the transition to HMC*, the decay towards the ring‐closed structure progresses analogously to the dynamics observed upon direct excitation of HMC. A simplified photoreaction and proton‐release mechanism is provided in Scheme 4. The weaker ESA band of HMC* at 440 nm is not clearly visible in the contour plot (Figure 2A) due to the superposition with the GSB. Another small, positive signal below 370 nm can be noticed between 1 ps and 10 ps that also corresponds to HMC*. Furthermore, the amplitudes of τ₁ and τ₂ indicate a rise and decay behaviour of the respective signal. Interestingly, the 13 ps lifetime exhibits no negative amplitude in the range of the GSB, which means that a recombination process of the proton from HMC* straight to the HMCH ground state is not indicated here.

Scheme 4.

Photoreaction pathway scheme of the excited state proton‐release and ring‐closure of HMCH of Py‐BIPS 1 at pH 1.

After roughly 10–30 ps the sharp ESA₂ signal of HMC* superimposes with an emerging broad positive signal between 450 and 580 nm, which decays on a longer timescale but is almost vanished after 1.5 ns. The slow depletion of this broad signal can be observed in the spectral range where ESA₁ was located (Figure 2B). This broad absorption signal matches the time‐integrated absorption spectrum of HMC and can thus be attributed to its ground state absorption (GSA). The positive amplitude of the time constant τ₃=424 ps around 500 nm reflects the decay of the respective signal. Since the negative amplitude of τ₃ in the wavelength range of the GSB captures a repopulation of the HMCH ground state, we assign the respective time constant to the thermal reassociation of the proton. This reprotonation process completes the photoprotolytic Förster cycle roughly within the first nanosecond after excitation. At the end of the monitored time window the remaining bleach signal around 420 nm, described by the infinite time constant, reflects the conversion of some of the molecules into the spiro form.

Similar to HMC, a fluorescence quantum yield of ∼3.5 % was determined for HMCH.^[48] Non‐radiative relaxation of HMCH* certainly represents a considerable decay channel as the amplitude of τ₃ exhibits a ground state recovery contribution. However, the proton dissociation channel is suggested to be the predominant decay pathway, which is further supported by the fact that ESA₁ (HMCH*) seems to thoroughly convert into ESA₂ (HMC*) with mostly one lifetime. Assuming that radiative relaxation of HMC* again occurs with ∼3.5 % probability, non‐radiative decay is again the major channel. Though, an overall quantum yield of SPH formation of ∼10 % was obtained by steady‐state experiments, which is even higher than upon excitation of HMC. If this is due to relative excess energy provided by the high efficiency of the photoprotolytic decay channel towards HMC* remains speculative.

The pK_a* of the excited state can be estimated by applying the Förster cycle, which provides a simplified thermodynamic model for photoprotolytic processes. Equation (1) gives the correlation between the free energies ΔG of the proton dissociation processes in the ground and excited state and the transition energies of the acid AH and conjugate base A⁻.[⁵⁶, ⁵⁷]

$${\displaystyle E_{AH}-E_{A^{-}}=\Delta G-{\Delta G}^{{}^{*}}=-RT(ln{K}_a-ln{K_a}^{{}^{*}})}$$ (1)

To determine these energies, the transition energy differences between the vibrational ground states of S0 and S1 have to be estimated. Therefore, the most reasonable approach is to average the frequencies of the absorption and emission maxima of the acid (${\nu }_{\mathrm{R}\mathrm{O}\mathrm{H}}$ ) and the conjugate base (${\nu }_{{\mathrm{R}\mathrm{O}}^{-}}$ ) as observed in steady‐state experiments. Then the difference of the pK_a values of the ground and excited state is obtained by Equation 2.

$${\displaystyle {{pK}_a-{pK}_a}^{{}^{*}}=N_{A}h({\nu }_{ROH}-{\nu }_{{RO}^{-}})/2.3RT}$$ (2)

Here, N_A is Avogadro‘s number, h Planck‘s constant and R the universal gas constant. With a ground state pK_a=3.2 of HMCH this estimation results in a substantial pK_a drop of 8.7. Therefore, an excited state pK_a* of −5.5 is determined, which makes this merocyanine isomer one of the strongest super‐photoacids reported, significantly more acidic than 1N or HPTS. Regarding its pK_a* value, HMCH is most comparable to the naphthol derivative DC2N^[11] (pK_a* =−4.5) or to the QCy derivatives like, for example, QCy7^[20] (pK_a* =−5.7). The ultrafast proton transfer within less than 1.1 ps and the striking pK_a drop indicate a pronounced enhancement of the acidity of HMCH* upon excitation. The photoacidity basically depends on the charge distribution in the excited state and the stabilization of the negative charge of the deprotonated form. In case of merocyanines, the negative charge is in general highly delocalized due to the expanded conjugation of the molecule and can be displaced from the phenolate oxygen to the indolinium nitrogen. Because of that, the neutral quinoid mesomeric resonance structure of the core chromophore strongly contributes to the stabilization. Moreover, a pronounced redistribution of π‐electrons is suggested to take place upon excitation of aromatic photoacids, which facilitates the proton‐release process.^[58]

Following theory, the ESPT mechanistically occurs in two consecutive steps.[⁵⁸, ⁵⁹, ⁶⁰, ⁶¹, ⁶²] Upon electronic excitation, the proton is transferred from an aromatic photoacid ROH* to an adjacent water or solvent molecule with an intrinsic rate. This creates a solvent stabilized contact ion pair, with a distinct contact radius [Eq. 3]:

$${\displaystyle {ROH}^{{}^{*}}\overrightarrow{\leftarrow }\left[{RO}^{-{}^{*}}{\cdots H}^{+}\right]}$$ (3)

With further increase of the distance between the ions the proton becomes thoroughly solvated or rather is transferred to larger water clusters. This step is apparently dictated by the diffusion of the proton away from the anion into the bulk solvent and is therefore strongly affected by solvent properties and electrostatic interactions. Therefore, the timescale of this diffusive step is typically much longer than that of the initial proton abstraction on contact [Eq. 4]:

$${\displaystyle \left[{RO}^{-{}^{*}}\cdots H^{+}\right]\overrightarrow{\leftarrow }{RO}^{-{}^{*}}+H^{+}}$$ (4)

A mathematical treatment engages the Debye‐Smoluchowski equation (DSE) that takes the random thermal motion into account as well as the Coulomb potential between the ions.^[63] The back reaction is referred to as diffusion‐assisted geminate recombination, which is substantially influenced by the electric field of the conjugate base anion.

In the case of HMCH*, the initial step of the photoprotolytic reaction occurs within 1.1 ps in water. This translates into an overall proton transfer rate k_pt=9.1×10¹¹ s⁻¹. The formation of ESA₂ (Figure 2A) is ascribed to the contact ion pair, as this is assumed to adopt identical spectral properties as the deprotonated RO⁻*. There are no clear indications for a major contribution of the back reaction since ESA₁ seems to convert entirely into ESA₂. Though, the existence of the equilibrium can be verified since a pH dependent fluorescence emission of HMCH* is observed in the steady state. Interestingly, the initial proton transfer occurs on a similar timescale as solvent relaxation dynamics. The Debye relaxation time of water around a solute molecule is found to be approximately 1 ps^[17] and the proton hopping time from one water molecule to another is roughly 1.5 ps.^[62] Hence, the proton transfer from HMCH* to a water molecule proceeds even faster than the proton transfer in between water molecules via the Grotthuss mechanism. The subsequent diffusion is supposed to occur similarly fast in aqueous solution, although this is influenced by the electric field of the conjugate base anion. Apart from the negative phenolate charge, the deprotonated merocyanine HMC* exhibits two positively charged nitrogen atoms. In addition, the compound bears a flexible negatively charged sulfonic acid residue that could exert enhancing effects on geminate recombination.

To gain more insights into the ESPT dynamics and to verify whether HMCH is capable of transferring its proton to poorer proton acceptors like organic solvents, we conducted additional TA experiments in the protic solvents methanol (MeOH) and ethanol (EtOH) and the aprotic solvent acetonitrile (MeCN) at pH 1. Although, both protic solvents are capable of H‐bonding, they are significantly less polar. While water exhibits a dielectric constant ϵ of 78, those of MeOH and EtOH are 33 and 24, respectively. The dielectric constant of MeCN is 37 and is therefore comparable to MeOH. The obtained transient maps in MeOH and EtOH show almost the same spectral signature as observed in water, although the signals are slightly shifted hypsochromically (see Figures S3 and S4).

The decay of the initially formed ESA₁ and SE₁ of HMCH* is accompanied by the emergence of ESA₂ and SE₂ within the first ps. The latter two signals are again assigned to the excited state of the conjugate base HMC*. An ultrafast ESPT of the phenolic proton therefore evidently occurs also in protic organic solvents.

Apart from that, the decay of ESA₁ is slowed down with decreasing proton acceptor capabilities of the solvents. As the formation of ESA₂ and SE₂ is also delayed it can be noted that the ESPT is decelerated in the order from water to MeOH to EtOH. Furthermore, the relative intensities of the ESA₂ signals in the protic solvents compared to ESA₁ are smaller than in water, which hints at a less efficient process. Accordingly, the SE₂ signal is significantly less pronounced, too. In the kinetic analysis of the data sets, a similar number of lifetime components as in water was determined (Figure 3A). The respective lifetimes are of similar orders of magnitude and their amplitudes mostly resemble analogous contributions. Yet, the measurement in aprotic MeCN revealed no emergence of the excited conjugate base signals (see Figure 3B). Only the signals originating from HMCH* are visible, which decay substantially slower than in water and the protic solvents. Therefore, the lifetime of the excited state HMCH* is elongated. However, a proton transfer to MeCN can be ruled out, although it is the organic solvent with the highest polarity among the investigated ones.

Figure 3.

A) Comparison of the determined DAS of the TA measurements of HMCH of compound 1 in H₂O (solid lines), MeOH (dashed lines) and EtOH (dotted lines). B) TA contour plot of HMCH of compound 1 at pH 1 in MeCN (left panel) and corresponding DAS (right panel).

In the GLA of the MeOH and EtOH measurements, time constants of 1.7 ps and 1.9 ps, respectively, were determined to describe the proton transfer processes (Figure 3A). Despite the before mentioned hypsochromic shift of the signals, the amplitudes of these time constants are in excellent agreement with the 1.1 ps constant found in aqueous solution. Hence, the ESPT is significantly slower than in water but roughly similarly fast in MeOH and EtOH. The time constants found for the subsequent decay of HMC* are also slower in MeOH (16 ps) and EtOH (24 ps) than the 13 ps constant in water. The respective time constants exhibit positive contributions in the range of ESA₁ around 490 nm, which are absent in the DAS of the lifetime determined in water. This might indicate that a pronounced fraction of ESA₁ decays on a 10–30 ps timescale and the actual lifetime of the HMCH* state is prolonged in both protic solvents. The biphasic decay of ESA₁ furthermore shows that a slowed down decay channel competes with proton dissociation. A similar timescale of the ESA decay was also observed in the MeCN measurement. A 10 ps lifetime was determined for the main decay channel of HCMH* and an additional 33 ps lifetime was found. The latter captures a decay of a positive band around 500 nm as well as the formation of a positive absorption signal above 550 nm and a repopulation of the ground state. The respective lifetime might thus account for the decay of a distinguishable merocyanine isomer, which is excited simultaneously or which is formed in the excited state through isomerization. Although the photodynamics are suggested to be dominated by the TTC isomer (Scheme 2C), the observed signals may be superimposed with those of the TTT isomer. The lifetime components τ₂ found in MeOH and EtOH therefore supposedly represent mixed lifetimes that comprise additional decay dynamics of HMCH*. In any case, the biphasic decay of ESA₁ modelled by τ₁ and τ₂ implicates competing deactivation pathways that lower the efficiency of proton dissociation.

The lifetimes τ₂ furthermore exhibit negative amplitudes below 475 nm and thus model a recovery of the HMCH ground state, in contrast to the corresponding lifetime found in water. Hence, in organic solvents a proton recombination quenching of HMC* seems to occur, which could be assisted by the negatively charged sulfonate residue. In aqueous solution this process might occur too, but significantly less pronounced. As in water, the formation of the broad GSA of HMC (450–550 nm) can be observed in both TA maps of the protic solvents (Figures S3 and S4) after the ESA₂ decay. The relative intensity of this long‐term feature decreases significantly in the order from water to MeOH to EtOH, which provides another evidence that the efficiency of the photoprotolytic decay of HMCH* also decreases in that order. In contrast to the measurement in water, the depletion of this signal is not completed at the end of the accessible time window after 1.5 ns. The HMC ground state is therefore still present in both protic solvents.

However, to fit the data sets adequately by GLA, a third lifetime component had to be engaged, that captures a partial decay of the HMC ground state towards HMCH. This lifetime τ₃ exhibits a much smaller relative amplitude than the corresponding 424 ps lifetime determined for reprotonation in water and it is accelerated to 320 ps in MeOH and 254 ps in EtOH. The amplitudes of τ₃ in the protic solvents exhibit a minor contribution to the decay of ESA₂ and a slower decay component of the SE₂ signal. This hints at a non‐exponential long‐term decay of the excited conjugate base. An accelerated reprotonation process seems plausible, if it is due to geminate recombination, as the solvation of the free proton is less favoured in both organic solvents than in water. Here, again the sulfonate residue supposedly plays an important role in arranging the nearby solvation shell of the photoacid. An exclusively diffusion‐driven reprotonation process, following the DSE model, should be much slower in the protic solvents, since the diffusion coefficients of the proton are significantly smaller than in water. Thus, the electrostatic potential of the negatively charged and flexible residue supposedly affects the diffusion of the proton and assists to the recombination. Moreover, the amplitudes of τ₃ reveal a decay contribution in the red flank of the overall GSA signal (575 nm) in the protic solvents. The blue region of the signal around 550 nm is still present after 1.5 ns, which is captured by the infinite lifetimes. This might again be ascribed to the presence of two merocyanine conformers, where one undergoes the reprotonation process faster than the other. Upon excitation, the main TTC‐merocyanine converts into a ground state cisoid structure which either undergoes ring‐closure or reverts back to the initial trans‐configuration and adopts a red‐shifted absorption band.^[64] Hence, the decay of the red flank of the GSA signal might account for the cisoid‐form which favours the reprotonation process due to its particular geometry.

As additional model‐free data evaluation method, lifetime distribution analyses (LDA) were performed on each data set. The obtained lifetime density maps (LDM) for the three solvents are depicted and further compared in the Supporting Information (see Figure S6). The LDMs represent the time‐dependent distribution of lifetime components, obtained from fitting the data sets with a high number of lifetimes (>100). In contrast to the GLA, this analysis is thus not biased by a kinetic scheme and a given set of contributing lifetimes. It therefore provides a better representation of the actual timescales of the assigned processes. The reading of the LDM contour plots, however, is similar as for the DAS. A positive amplitude indicates the decay of a positive signal or the increase of a negative one and vice versa.

Proton‐transfer dynamics of Nitro‐BIPS derivative 2

TA experiments were conducted with the protonated form of the extensively applied 6‐nitro‐BIPS derivative 2 (Scheme 2B). In acidified aqueous solution, the spiropyran‐merocyanine equilibrium is shifted to the energetically favoured protonated merocyanine MCH and thermal conversion to the ring‐closed isomer is not observed anymore. The photophysical properties in neutral aqueous solution as well as the ultrafast dynamics are reported elsewhere.^[47] A brief description of the steady‐state behaviour in the neutral and acidic pH range may be found in the Supporting Information, together with the absorption and emission spectra of the MC and MCH states (Figure S7). The ground state pK_a value of compounds 2 is 3.7.^[35] Thus, the TA measurements were carried out at pH 1 to accumulate the pure protonated state.

The transient map of Nitro‐BIPS 2, shown in Figure 4A, reveals initial signals that can clearly be assigned to the excited S1 state MCH*. Around 560 nm the negative signal SE₁ appears directly upon excitation, as well as a prominent positive signal (ESA₁) and a minor ESA band below 425 nm (not denoted). The latter overlaps with the GSB around 410 nm, which also emerges instantaneously upon excitation. The minor ESA below 425 nm vanishes after around 1 ps just as the signal SE₁, which is accompanied by the formation of the even more pronounced SE₂ around 610 nm. This spectral evolution already hints at the formation of the excited conjugate base, as SE₂ fits the steady‐state emission thereof.^[47] The prominent ESA₁ prevails until roughly 100 ps but evolves into a broad positive signal between 450 nm and almost 600 nm. Here, the interpretation is not unequivocal at first, due to the superposition of multiple signals. Yet, in comparison with the reported TA data of the deprotonated MC ^[47] the assignment becomes clear. The deprotonated form exhibits an SE signal between 575 nm and 650 nm and one pronounced ESA signal around 450 nm, both lasting until approximately 100 ps. The similarity of this spectral signature and the observed signal of MCH (ESA₁) indicate that both the acid and the conjugate base exhibit ESA signals around 450 nm, impeding a precise discrimination in the spectrum in Figure 4A.

Figure 4.

A) TA contour plot of MCH of compound 2 at pH 1 (left panel) and corresponding DAS (right panel). B) Selected transients showing the time‐dependent progression of the signals GSB (blue), ESA₁ and ESA₂ (cyan), SE₁ and GSA^MC (green) and SE₂ (orange). Data points are represented by symbols (crosses) and the fit from GLA by solid lines.

The ESA₁ transient at 485 nm (Figure 4B) reveals an initial ultrafast decrease within the first 700 fs, followed by a slower depletion of the signal. According to the DAS (Figure 4A, right panel), a sub‐ps lifetime was found (τ₁=0.4 ps), which is ascribed to an initial cooling process within the excited S1 state MCH*, as it models a slight shift of the corresponding signals. The lifetime τ₂=2.1 ps captures the decay of the minor ESA below 425 nm as well as the decay of SE₁ and the subsequent formation of SE₂. In the wavelength range around ESA₁ the amplitude of this lifetime is almost zero. However, we assign τ₂ to the proton‐release process, as illustrated in Scheme 5, primarily because it accounts for the transition from SE₁ (MCH*) to SE₂ (MC*). Yet, as the signal SE₂ already emerges within the first 0.5 ps, the proton transfer is supposed to start during the vibrational relaxation process. If the transition from SE₁ to SE₂ is just fitted with one lifetime component, a lifetime of 1.6 ps is determined, which therefore reflects the mean value of the proton transfer lifetime component. This estimation results in a proton transfer rate k_pt=6.3×10¹¹ s⁻¹. However, compared to compound 1 the quantum yield of the ESPT process is suggested to be similarly high, since the SE₁ signal is entirely converted into SE₂. The minor ESA below 400 nm shows similar decay dynamics as SE₁ which also speaks against a large contribution of additional competing dynamics.

Scheme 5.

Photoreaction pathway scheme of the excited state proton‐release and ring‐closure of MCH of Nitro‐BIPS 2 at pH 1.

The subsequent decay of ESA₂ and SE₂, that correspond to MC*, is modelled by time component τ₃=71 ps (positive amplitude around 450 nm and negative amplitude above 575 nm). Therefore, it can be attributed to the lifetime of the respective state and a lifetime in the same order of magnitude and with a similar DAS was found upon direct excitation of MC.^[47] The negative amplitude of τ₃ between 490 nm and 575 nm accounts for the emergence of another positive signal (ESA₂) whose decay is modelled by the broad positive amplitude of τ₄=413 ps in this range. This absorption band is related to the MC ground state. As τ₄ features a negative amplitude below 450 nm it also captures a repopulation of the MCH ground state. Hence, τ₄ describes the reprotonation process, which completes the photoprotolytic cycle on the same timescale as observed for compound 1. Interestingly, the reprotonation in the ground state proceeds with almost similar lifetimes for both merocyanine derivatives 1 and 2, as the pK_a values of the acid ground states are rather similar. This is indicative of a diffusion‐driven recombination with the solvated proton. Apparently, the alkyl residue attached to the indoline nitrogen does not affect the reprotonation in aqueous solution, as the two compounds exhibit differently charged functional groups.

Based on the steady‐state absorbance and fluorescence data (Figure S7), again the excited state pK_a* of MCH* could be obtained. With a pK_a=3.7, the Förster cycle estimation yields a pK_a drop of 7.9 units and a pK_a* =−4.2. This is only slightly less acidic than the Py‐BIPS PA 1 discussed before and Nitro‐BIPS 2 has to be considered a super‐photoacid as well. The attachment of the NO₂‐group exerts a smaller electron‐withdrawing effect on the phenolic oxygen than the incorporated nitrogen of Py‐BIPS and the charge displacement upon optical excitation is less pronounced.

The photoprotolytic reaction of MCH of compound 2 was also measured in the solvents MeOH, EtOH and MeCN. The TA spectra of MCH in MeOH (Figure 5A) and EtOH (Figure S8) show very similar features and monitor clear indications for a proton transfer to the solvent. In the protic solvents, the signal SE₁ of MCH* is longer lived and SE₂ of MC* is significantly less pronounced which is indicative of the proton transfer. A time constant of τ₁=2.4 ps was found for the proton transfer in both protic solvents as it models the transition from SE₁ to SE₂ (Figure 5B). It also captures a partial decay of the superimposed ESA₁/ESA₂ signal and the decay of the minor ESA below 400 nm, that also corresponds to MCH*. The following decay of the ESA₁/ESA₂ feature seems significantly accelerated compared to water and it evolves into the broad absorption signal of the MC ground state, which is also less pronounced. The decay of the ESA signal is captured by two more lifetime components. While the main decay is modelled by a 17 ps (18 ps) in MeOH (EtOH), a considerable contribution decays with a significantly longer lifetime of 290 ps (298 ps). The corresponding amplitudes are in good agreement for both protic solvents. The amplitude of the 71 ps lifetime found in water for the particular ESA decay is rather resembled by the slower lifetime components than the faster ones, regarding the minimum in the near‐UV and the region above 500 nm (Figure 5B). However, both the fast and the slow components show contributions to the formation of the broad GSA signal and the decay of SE₂ but at different wavelength ranges. Therefore, the respective two lifetimes are supposedly attributed to different cis‐trans merocyanine isomers that undergo the discriminable relaxation pathways.

Figure 5.

A) TA contour plot of MCH of compound 2 in MeOH (left panel) and corresponding DAS (right panel). B) Comparison of the DAS determined through GLA for the three solvents H₂O (solid lines), MeOH (dashed lines) and EtOH (dotted lines). The lifetime components are sorted by highest agreement.

Especially concerning Nitro‐BIPS derivatives, different isomerization sequences have been reported for particular conformers in various solvents.^[64] Ruetzel et al. found that in MeCN the TTC isomer converts into the TTT in the excited state unidirectionally within 200 fs and both forms decay on different timescales afterwards.^[50] An SE signal around 625 nm was assigned to the TTC isomer while that of the TTT form is shifted bathochromically by 30 nm. Therefore, we ascribe the lifetime τ₂, which models the decay of the blue region of SE₂ to the TTC conformer. The longer lifetime τ₃ with a smaller relative amplitude then presumably corresponds to TTT which represents the isomer, typically existing to a smaller extent. After 1.5 ns, there is still a pronounced positive signal around 550 nm, which is in good accordance with the reported ground state absorption of the TTC isomer. This, as well as a prominent residual fraction of the MCH bleach signal, is reflected by the infinite time constants determined in both protic solvents (Figure 5B). Their amplitudes are in good agreement with the 412 ps lifetime component determined in water. Hence, the ground state reprotonation seems to be essentially slower than in water and also than in the case of compound 1. Here again, this might be influenced by the alkyl residue attached to the indoline nitrogen. Compound 2 bears a positively charged functionality, which is supposed to exert a repulsive effect on the released proton and therefore hinders a diffusion‐assisted recombination.

The essentially reduced efficiency of the photoprotolytic decay of MCH* in MeOH and EtOH hints at the existence of other major reaction channels than in water. Besides fluorescence relaxation also non‐radiative decay into the ground state may occur. Internal conversion is reported to be a considerable pathway for Nitro‐BIPS derivatives, also because of the enhancing effect of the nitro group.^[65] Moreover, the contribution of triplet states to the photodynamics of Nitro‐BIPS is well‐known,^[64] although no distinct indications were found here. Non‐radiative decay is suggested to yield a vibrationally hot ground state which undergoes further cooling. This cooling process causes the observed bathochromic shift of the GSB signal after 100 ps (Figure 5A). However, the remaining bleach after 1.5 ns resembles the ground state absorption spectrum quite well.

In the measurement in MeCN (Figure S11) similar MCH‐associated signals were detected as in the protic solvents, like an ESA around 480 nm and SE around 570 nm and both decay on a similar timescale within 10–20 ps. As no emergence of signals that could be assigned to the excited conjugate base was observed, a proton transfer can be ruled out. Though, a positive absorption signal from 500 nm to 625 nm appearing after 20 ps is detected, which accounts for a deprotonated merocyanine ground state absorption. This could either arise from small amounts of water within the measured sample that the proton is transferred to or to residual amounts of non‐protonated merocyanine that undergoes the isomerization observed by Ruetzel et  al. The GLA revealed a vibrational relaxation of MCH* with a lifetime of 1.1 ps. Interestingly, a lifetime of 7.5 ps was determined here for the MCH* decay of the main isomer (TTC), which is essentially faster than the lifetimes reported for the non‐protonated Nitro‐BIPS analogue.

Proton‐transfer dynamics of Dinitro‐BIPS derivative 3

For the Dinitro‐BIPS derivatives 3 (Scheme 2B), the ring‐opened MC isomer is even more stabilized than for the nitro compound 2. In aqueous solution, compound 3 is therefore more prone to hydrolysis. Though, the photoswitching properties are quite comparable to compound 2 just as the excited state dynamics of the unprotonated MC isomer.^[47] With a pK_a value of 3.9,^[49] the protonated MCH state is formed upon acidification, which is thermally stable at pH 1 and shows fluorescence emission around 495 nm. A brief summary of the behaviour in aqueous solution may be found in the Supporting Information, with corresponding absorption and emission spectra of the MC and MCH states (Figure S7). Based on these spectra, again the pK_a* value can be estimated for compound 3 by use of the Förster cycle approximation. The resulting pK_a drop of 8.3 units is even more pronounced than that of the nitro compound 2. Ultimately, a pK_a* value of −4.2 is determined for MCH of compound 3 is, which makes it a super‐photoacid.

The TA measurement of MCH of the Dinitro‐BIPS derivative 3 at pH 1 revealed an essentially different behaviour than the before presented PAs 1 and 2. The TA spectrum (Figure 6A) displays two ESA signals centred around 470 nm and 600 nm, where the latter is more intense by a factor of roughly 2.5. Those positive signals as well as the GSB at 400 nm decay almost completely within 1 ps. This implies, that the compound mainly relaxes back into its ground state non‐radiatively upon optical excitation. After roughly 30 ps, all observed signals virtually decay to zero, as the transients in Figure 6B illustrate. However, a small portion of the GSB signal remains until the end of the measurement. The ESA features decay with a 0.8 ps time constant and leave two small blue‐shifted positive signals around 460 nm and 550 nm behind. After approximately 1.5 ps, a small negative SE signal is noticeable above 600 nm which decays with the same time constant (9.4 ps) as the respective minor positive signals. These signals and the corresponding lifetime are assigned to the conjugate base MC*, as they are in good agreement with the reported spectral signature, although the weak ESA around 550 nm could not be observed upon direct excitation of MC at pH 7.4 due to the superimposed prominent bleach signal.^[47] The reported lifetime of the excited MC*state is 8.7 ps, which further confirms this assignment. The amplitude of the lifetime determined here exhibits a small contribution to the ground state recovery, which indicates a proton‐recombination induced quenching process. The 0.8 ps thus models the transition from MCH* to MC* and is associated with the proton transfer rate k_pt=1.3×10¹² s⁻¹. After the depletion of the MC* signals, there is almost no residual signal observable. Only the infinite time constant reveals a weak positive residual signal around 520 nm that corresponds to the conjugate base ground state. An additional lifetime component for the reprotonation process could not be determined.

Figure 6.

A) TA contour plot of MCH of compound 3 in MeOH at pH 1 (left panel) and corresponding DAS (right panel). B) Selected transients showing the time‐dependent progression of the signals GSB^MCH (blue), ESA₁ (cyan), ESA₂ (green) and ESA₂ and SE₂ (orange).

The peculiar photodynamics of compound 3, compared to compounds 1 and 2, apparently arise from the additional nitro group in ortho‐position to the acidic hydroxyl group of MCH. With a partially negative charge on both oxygen atoms of the nitro group, it may stabilize the hydroxyl proton already in the ground state and react as proton acceptor upon excitation. The photodynamics of o‐nitro benzyl (oNB) compounds is known to involve an initial excited state intramolecular proton transfer (ESIPT) but typically from an alkyl α‐H‐atom to the nitro group.[⁶⁶, ⁶⁷, ⁶⁸] This results in the formation of the tautomeric aci‐nitro form (=NO₂H) through parallel singlet and triplet pathways, which is frequently exploited for the subsequent removal of photolabile protecting groups.[⁶⁹, ⁷⁰] Regarding compound 3, an ultrafast ESIPT to the 8‐nitro group might therefore be an considerable reaction channel, although no distinguishable spectroscopic characteristics of this species are observed here. However, since MCH* of compound 3 mostly decays into its ground state, the oNB group might facilitate the proton‐induced recombination quenching of the MC* state via transient aci‐nitro formation.

Similar to Nitro‐BIPS derivatives, the presence of the TTC and TTT isomer has been reported for the non‐protonated dinitro‐analogues.^[71] The TTC form also represents the most stable one with a main absorption band around 560 nm, while the TTT absorbance is shifted to longer wavelengths. An interconversion between the two conformers is not supposed to be a significant reaction pathway. In this study, we found no indications for discriminable isomers and the photodynamics are suggested to be dominated by the TTC isomer.

Correlation between free‐energy and ESPT rates

The determined photoacidic characteristics of the three investigated compounds are summarized in Table 1. It is noticeable that the ground state pK_a values are in a similar range. This entails, that the electron distribution at the phenolate moiety is influenced in a comparable way. The attachment of the electron withdrawing nitro substituents and the incorporation of the pyridine ring evidently lower the pK_a value to a similar extent and the estimated pK_a changes upon optical excitation are quite similar, too. Compound 1 yet exhibits the lowest pK_a* value and thus the most pronounced charge displacement in the excited state.

Table 1.

Photoacidic properties of the excited state PAs of the Py‐BIPS derivative 1, the nitro‐BIPS compound 2 and the dinitro‐BIPS compound 3 in water.

	1 (HMCH)	2 (MCH)	3 (MCH)
pKa	3.2	3.7	3.9
ΔpKa	−8.7	−7.9	−8.3
pKa*[a]	−5.5	−4.2	−4.4
τ (ROH*) [ps]	1.1	2.1	0.8
kpt [b] [s−1]	∼9.1×1011	∼6.3×1011	∼1.3×1012

[a] Excited state pK_a* values, determined by the Förster cycle approach, [b] ESPT rate coefficients obtained from GLA.

The relationship of the extracted proton transfer rate constants and the estimated pK_a* values can be further rationalized by a semi‐empirical correlation, which is based on the Marcus theory for electron transfer processes^[72] yet modified for proton transfer. This gives a major role to the alignment of solvent molecules in the vicinity of the dissociating proton.[⁷³, ⁷⁴] Accordingly, the proton transfer process is described by a solvent coordinate along an pre‐existing hydrogen bond. The rate k_pt is then expressed as a function of the intrinsic solvent‐dependent activation free‐energy change ΔG^# of the reaction [Eq. 5]:

$${\displaystyle k_{pt}=k{}^{*}·exp\left(-\frac{\Delta G^{\#}}{RT}\right)}$$ (5)

Here, (k^*)⁻¹ represents the frequency factor of the transfer reaction, R is the gas constant, and T the absolute temperature. The free energy change ΔG^# is estimated by the Marcus bond‐energy‐bond‐order (BEBO) model [Eq. 6]:^[75]

$${\displaystyle {\Delta G}_0=\frac{{\Delta G}_0}{2}+{\Delta G}^{\#}+\frac{{{\Delta G}_0}^{\#}}{ln2}ln\left(cosh\left[\frac{{\Delta G}_{0}ln2}{2{{\Delta G}_0}^{\#}}\right]\right)}$$ (6)

The free‐energy of the charge‐exchange with the solvent is given by ΔG₀ ^# if the total free‐energy in the PT process ΔG₀ is zero [Eq. 7]:

$${\displaystyle {\Delta G}_0=RTln10{\Delta pK}_a}$$ (7)

The correlation of the determined rate constants k_pt of the compounds 1–3 and the pK_a* values, estimated by the Förster cycle, is depicted in Figure 7. The solid line represents the Marcus BEBO model fit including the reported PT rates of a variety of comparable photoacids with pK_a* values around zero or below. Several phenol,^[10] 1N,[⁷⁵, ⁷⁶] 2N^[16] and hydroxy quinoline (HQ)[⁷³, ⁷⁷] derivatives are shown, as well as the beforementioned QCy[²⁰, ⁷⁸] compounds.

Figure 7.

Free‐energy correlation of determined rates k_pt of compounds 1,2 and 3 (filled diamonds) compared with comparable reported photoacids (empty circles). The solid line represents the Marcus bond‐energy‐bond‐order (BEBO) model according to Equations 5–7.

For PT reactions to bulk water the pK_a* is corrected for the purely electrostatic contribution pK_a,el=R_D/2.3a.^[79] A charge of −1 for the conjugate base is assumed for the Debye radius R_D and the contact radius a was set to be 6.5 Å.^[80] In the fitting procedure, k* was treated as a free parameter and a value of roughly 5×10¹² s⁻¹ was obtained which represents the proton transfer rate for an activationless process. Similarly, an intrinsic free‐energy barrier of ΔG₀ ^#=3.5 kcal mol⁻¹ was found, which is close to reported values.[⁸¹, ⁸²] Consequently, the determined rates k_pt of the herein investigated merocyanine PAs are in well agreement with the applied literature‐based structure‐reactivity correlation, including super‐photoacids with pK_a* values lower than −4. The compounds 1–3 are significantly more acidic than most naphthol derivatives. They are situated in the highly exothermic regime of the correlation curve together with the structurally related QCy7, S‐QCy7 and TS‐QCy compounds. The photoacidity of the studied merocyanine PAs as well as the QCy dyes is additionally enhanced due to the high degree of charge delocalization through the molecular structure and the formation of the quinoid form. The strongest reported photoacid QCy9 even exceeds the boundaries of the correlation model with its remarkable proton transfer rate of ∼10¹³.[²², ⁷⁸] This is close to the stretching mode frequency of an OH‐bond which is on the order of almost ∼10¹⁴ s⁻¹.

The findings further corroborate the estimated pK_a* values, that were derived from equilibrium conditions. These values can therefore hardly describe dynamic characteristics of ESPT processes. For instance, the compounds 1–3 exhibit several charged functional groups, such as the flexible ionic alkyl residues at the indoline nitrogen, that may contribute to a pre‐organization of the solvent shell which facilitates the ESPT processes. The establishment of an aligned solvent network stabilized by polar or hydrogen bonding moieties affects proton dissociation differently than bulk water arrangements and also extends the lifetime such water wires.^[83] Especially the anionic sulfonate group of 1 might exert a significant effect in this fashion. For S‐QCy7 and TS‐QCy, bearing similar functional groups, a sulfonate‐assisted proton dissociation is suggested, too.^[20] Yet, because of the different PT rates found in the protic solvents MeOH and EtOH a direct proton transfer to the sulfonate group can be excluded here.

Regarding the measurements in alcoholic solvents, it has to be noted that minor amounts of water were present in the sample solutions, because of acidification with aqueous HCl. Again, the observed varying proton transfer rates compared to water indicate that it‘s not a pure water cluster that is arranged around the acidic proton and that it is transferred to, although the solvent distribution might not be homogeneous.^[79] Moreover, the measurements in MeOH and EtOH revealed an increased contribution of alternative decay channels of the excited PA that lower the efficiency of proton dissociation. For the Py‐BIPS derivative 1, ESPT rates of 5.9×10¹¹ s⁻¹ and 5.3×10¹¹ s⁻¹ were estimated in MeOH and EtOH, respectively. For Nitro‐BIPS 2, the contribution of competing pathways such as internal conversion is even more pronounced and similar ESPT rates of roughly 4.2×10¹¹ s⁻¹ were found in both protic solvents.

Conclusion

Our results show that the phenolic protonation sites of the Py‐ as well as the Nitro‐BIPS merocyanine photoacids investigated herein represent outstanding excited state PAs. They are capable of transferring their phenolic proton to water within a few ps upon optical excitation. Except for the dinitro derivative 3, ESPT is highly efficient and the preferred deactivation pathway in aqueous solution. Compounds 1 and 2 are even capable of ESPT to protic organic solvents, although this is less efficient and slightly slowed down compared to water. Regarding compound 3, an ESIPT to the ortho‐located nitro group seems plausible that results in the formation of a transient aci‐nitro form. Internal conversion and proton‐induced quenching of the excited conjugate base are suggested be the major deactivation channels.

The proton‐release mechanism itself has not yet been monitored directly on an ultrafast time scale but distinguishable pathways have been suggested, based on observations made in different steady‐state experiments. Our findings imply that the release of the phenolic proton is the first step in the photodynamics upon excitation of merocyanine PAs. Hence, the involvement of the aforementioned protonated spiropyran species in the ring‐closure reaction can be ruled out, as the proton is released prior to ring closing. This might furthermore not only be true for the presented photoacidic Py‐ and Nitro‐BIPS derivatives but also for protonated merocyanine derivatives in general. The TA measurements of the ultrafast ESPT reactions of compound 1 and 2 enabled the assignment of the entire photoprotolytic cycle, including the reprotonation in the ground state. As similar reprotonation lifetime components were determined for both compounds in water, the respective process is supposed to be solvent‐ and thus diffusion‐controlled. The differently charged residues attached to the indoline nitrogen do not cause differences in the reprotonation dynamics. In contrast, this step seems to be influenced in the protic solvents. The negative charge of the sulfonate group of compound 1 is supposed to accelerate the reprotonation process partially with decreasing solvent polarity. On the contrary, the positive charge of the trimethylammonium group of compound 2 slows down the reprotonation in the protic solvents.

In addition, we could show that the N‐protic site of compound 1 exhibits a fundamentally different reactivity. The proton is not released due to an enhanced excited state acidity, but subsequently upon isomerization to the spiropyran structure because of the ground state pK_a change. Hence, the respective protic site has to be considered a PAG rather than an excited state PA, although it also functions as a reversible photoacid. Yet, knowledge of the timescale and the dynamics of the proton‐release is essential in view of potential applications of PAs. Especially when applied as a trigger for time‐resolved studies of proton‐mediated processes the proton transfer dynamics must be taken into account as well as limitations caused by diffusion.

The integration of the pyridine ring into the merocyanine structure of Py‐BIPS derivatives has fascinating effects on the photochemistry and provides a valuable motif for future molecular design. Not only that this position allows for straightforward attachment of diverse substituents, but the photodynamics is understood quite well. Following a singlet pathway just as the original BIPS compound, the decay of the excited state is accelerated drastically starting from the unsubstituted pyridine nitrogen (MC) through the protonated (HMC) to the methylated Py‐BIPS form.^[45] Besides the retained ultrafast singlet pathway, the photophysical steady‐state properties are affected severely by the respective substituents and can thus be tuned. Essential differences are for example found when comparing the behaviour of compound 1 at pH 7.4 with the methylated Py‐BIPS compound. The lack of the methyl group of compound 1 results in a pronounced stabilization of the ring‐opened merocyanine, which allows for an operation of the photoswitch as a negatively photochromic one using visible light only.

Experimental Section

Sample preparation. The synthesis of compound 1 was performed following the published procedure via aldol condensation of the alkylated indoline and the corresponding pyridine salicyle aldehyde.^[48] The Nitro‐BIPS compounds 2 and 3 were synthesized using the respective nitro‐salicyl aldehydes.^[47] The spectroscopic investigation was carried out in phosphate buffered saline (PBS buffer) except for the measurements in organic solvents. The pH values were adjusted with concentrated aqueous HCl. The absorption spectra of the particular photoisomers were recorded with a Specord S600 spectrophotometer (Analytik Jena AG, Jena, Germany) in 10 mm×10 mm UV‐grade quartz glass cuvettes (Starna GmbH, Pfungstadt, Germany). The emission spectra were measured with a JASCO FP 8500 spectrofluorometer (JASCO Germany GmbH, Groß‐Umstadt, Germany) using 4 mm×10 mm UV‐grade quartz glass cuvettes (Starna GmbH). All spectra were offset corrected and the fluorescence spectra were additionally corrected for reabsorption effects as well as for the detector sensitivity of the spectrometer.

Femtosecond TA spectroscopy. The TA data were acquired with self‐assembled UV/vis‐pump/vis‐probe setups, either supplied by a 1 kHz Ti:Sapphire amplifier (CPA, Clark‐MXR, Michigan, USA) with a pulse duration of approximately 150 fs and a central wavelength of 775 nm or a Ti:Sapphire amplifier (Spitfire Ace, Newport Spectra‐Physics GmbH, Darmstadt, Germany) with a pulse duration of 120 fs and a fundamental wavelength of 800 nm. The excitation pulses in the vis‐range (520 nm) were generated with a non‐collinear optical amplifier (NOPA) by guiding a supercontinuum pulse into a beta‐barium borate crystal (β‐BaB₂O₄, BBO) together with a second harmonic pulse at 388 nm. UV‐pulses for excitation (420 nm) were generated by subsequent sum frequency generation (SFG) of the NOPA output and the fundamental in another BBO crystal. The white light continuum for probing (∼350 nm–650 nm) was created by guiding the fundamental of the laser system through a 2 mm CaF₂ window. The samples for the time‐resolved experiments were prepared in 1 mm UV‐grade quartz glass cuvettes (Starna GmbH) and the concentrations were adjusted to an optical density of approximately 0.6. During the pump/probe experiments, the samples were illuminated with light emitting diodes (LED, ThorLabs Inc., Newton, New Jersey, USA) to accumulate the photoisomer of interest. In order to prevent a reexcitation of already excited molecules, the sample cuvettes were moved in an y‐z‐plane perpendicular to the excitation pulses. To eliminate anisotropy effects, the relative polarizations of the pump and probe pulses were adjusted to the magic angle (54.7°).

The obtained transient maps were processed and corrected for the group‐velocity dispersion and the coherent artefact. The latter is fitted by a Gaussian function or its first and second derivative and subtracted from the data set. The TA spectra were then subjected to multiexponential fitting via global lifetime analysis (GLA) by using the kinetic fitting software OPTIMUS.^[52] From this, lifetime components and their corresponding decay associated spectra (DAS) could be extracted. Additionally, a lifetime distribution analysis (LDA) was performed as a complementary and model‐free approach (see Supporting Information).

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

C. Kaiser, T. Halbritter, A. Heckel, J. Wachtveitl, Chem. Eur. J. 2021, 27, 9160.