Control of the Photo-Isomerization Mechanism in 3H-Naphthopyrans to Prevent Formation of Unwanted Long-Lived Photoproducts

1. Introduction

Discovery of photochromic chromenes by Becker and Michl in 1966 initiated numerous studies related to their fundamental characterization and applications, especially as photochromic lenses [1]. One of the most frequently studied compounds belonging to this family is 3,3-diphenyl-3H-naphtho[2,1-b]pyran (see Scheme 1, CF-H, also known as 2,2-diphenyl-5,6-benzo(2H)chromene) [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21].

The coloration process in this molecule—which is based on the photoinduced opening of the pyran ring in its closed form (CF)—is ultrafast as it takes place on a few ps time-scale leading to the transoid-cis TC form (see Scheme 1) [3,11,14,22,23,24,25]. In the next step transoid-trans TT form is produced by TC photoexcitation or is formed as a direct product of the pyran ring-opening [12,17]. A serious drawback of 3H-naphthopyran family is the long fading time of the transoid-trans TT isomer. While the TC form relaxes over tens of seconds, the persistence of the long-lived TT form over hours is a limiting factor for classical applications such as photochromic lenses, due to inconvenient slow fading response to UV light switch off. Thus, recent research has been focused on minimization of the TT formation yield in the photoreaction by designing new 3H-naphthopyran derivatives with appropriate substituents [21,22,26,27]. Aryl substituents in position 2 of 3H-naphthopyrans can exert both steric and electrostatic repulsion effects which have been suggested to suppress TT formation [22]. Exceptionally short TC lifetime, on the order of tens of microseconds in solution, make TC→TT photoisomerization inefficient upon stationary UV irradiation. However in this case, the photostationary formation yield of TC is also low. Keeping the TC lifetime of seconds and the suppression of the TT formation have been recently reported for 3H-naphthopyran with methoxy group substitution at position 10 (CF-OCH₃, see Scheme 1) [21]. It has been proposed that the alkoxy group effectively reduces the formation of the transoid-trans form due to C−H···O intramolecular hydrogen bonding in the TC form [21]. The mechanism of TT suppression requires detailed studies of TC excited-state energy landscape.

Recently we have reinvestigated TT formation mechanism for unsubstituted 3H-naphthopyran (CF-H). We have shown that TC→TT photoisomerization competes with the other deactivation channels such as: S₁→S₀ internal conversion (major), S₁→T₁ intersystem crossing, S₁ state fluorescence and allenyl-naphthol AP formation (hydrogen shift reaction) [25]. With this knowledge, we undertook comprehensive experimental and theoretical studies to explain why the presence of a methoxy group at position 10 of 3H-naphthopyran can so dramatically decrease the yield of TC→TT photoreaction. We apply multi-scale (from hundreds of femtoseconds to seconds) UV-vis time resolved absorption spectroscopy and quantum chemical calculations for 3H-naphthopyrans: CF-H and CF-OCH₃. We have shown here for the first time two isomerization mechanisms TC→TT: (i) sole double bond isomerization with a single-twist competing with (ii) a bicycle-pedal isomerization (Scheme 1). The latter one is the dominant pathway for TC-OCH₃ while the single-twist is the main one for TC-H. In the case of bicycle-pedal mechanism, isomerization is aborted in halfway to TT. Altogether, the bicycle-pedal has been shown a photostabilizing channel for TC form, thus this pathway plays a key role to effectively reduce TT concentration in the photoreaction.

2. Results and Discussion

2.1. The Effect of Methoxy Group Presence on the Ground-State Energy Landscape and the UV Absorption Spectrum

As far as the UV absorption spectrum is concerned, the calculations for CF-OCH₃ molecule predict a very slight red-shift of the absorption maximum, in comparison to its position for the model chromene (CF-H), which in energy scale is not higher than 0.1 eV. This is in excellent agreement with experiment which gives the maximum of absorption at 361 nm in cyclohexane (3.44 eV) for CF-H [28] and 370 nm (3.35 eV) for CF-OCH₃ (Figure S1). The observed red-shift is the effect of H substitution with a π-electron donating group, O−CH₃, that decreases the excitation energy value of all the S₀→S_n(ππ*) electron transitions in CF. Such behavior has been observed earlier for methoxy substituted 3H-naphthopyrans (6-, 7- and 8-methoxy derivatives) [5,29].

3H-Naphthopyrans typically produce both TC and TT isomers (see Scheme 1) under conditions of continuous UV irradiation tuned to the CF absorption band. These TC and TT isomers are structurally related through the rotation about the C₁₃−C₁₄=C₁−C₂ dihedral angle. The UV absorption spectra of TC and TT are then quite similar (see Tables S1 and S2) since the π-electron conjugation in those two forms is alike. Even though, in both forms the strongest S₀→S₂ transition is slightly red shifted for TC relative to its position for TT (Figure 1b). The methoxy group substitution does not change much the chromene’s ground-state energy landscape (calculated at the MP2/cc-pVDZ theory level) leaving the S₀-state energy barriers almost unchanged [28]. Thus, similarly as in the H counterpart, TC-OCH₃ and TT-OCH₃ forms do not differ much energetically. The difference between the relative energies of the respective H and OCH₃ derivatives is marginal, less than 0.02 eV, and both forms lie much above the closed-pyran form CF, app. by + 0.6 eV.

2.2. TT Formation, Single-vs. Two-Photon Mechanism

To unmix TT and TC contribution at the photo-stationary state under UV irradiation, the usual method is to follow thermal recovery to CF. Indeed after the decay of TC, only TT form remains. Figure 1a shows then time evolution of transient absorption spectra recorded for CF-OCH₃ in cyclohexane after switching off UV irradiation (experimental set-up in Scheme S1). The initial absorption band corresponds to the mixture of TC (major) and TT (minor) species.

Figure 1b shows that the absorption maximum of TC-OCH₃ occurs at 437 nm (2.84 eV), while that of TT-OCH₃ isomer at 412 nm (3.01 eV), in cyclohexane (compare expt. vs theory in Table S1), corresponding to the spectral locations for H-derivative (427 nm, for TC-H, and 412 nm, for TT-H, Table S2). The TC lifetime is slightly longer for the methoxy form (11.7 s vs. 9.3 s in cyclohexane, see

Table 1. Photophysical properties of TC in solution: S₀ and S₁ absorption band maxima, lifetime τ_S0 and excited state lifetime τ_S1.

Compound	Solvent	S0(TC) λabsmax	τS0, s	S1(TC) λabsmax	τS1, ps
TC-OCH3	cyclohexane	437 nm	11.7 1	525 nm 3	0.47 2
	acetonitrile	447 nm	17.0 1	545 nm	0.27 2
TC-H	cyclohexane	427 nm	9.3 1	520 nm 3	0.87 2
	acetonitrile	430 nm	8.6 1	513 nm	0.31 2

Temperature: ¹ 21 °C; Accuracy: ¹ ± 5%; ² ± 10%; τ_S1 retrieved from band integral kinetics; ³ in agreement with calculations (Tables S3 and S4, respectively).

). A polar solvent as acetonitrile stabilizes the TC-OCH₃ species (lifetime 17.0 s), indeed calculations show a substantial dipole moment of 4.2 D (Table S1). Figure 1c shows that TT isomer in cyclohexane is formed in smaller amount from CF-OCH₃ than from CF-H precursor, in agreement with the data reported by Inagaki et al. [21]. On the basis of DAS amplitudes selected at the TT and TC maxima we estimate that CF-OCH₃ in relation to CF-H described recently [25], produces four times less TT form under the same experimental conditions. Two parallel reaction mechanisms leading to TT isomer have been considered in literature [12,17]. One is a two-photon absorption process:

$$\mathrm{CF} \stackrel{365 \mathrm{nm}}{\to }\mathrm{TC} \stackrel{365 \mathrm{nm}}{\to }\mathrm{TT}$$

while the second one is a single photon excitation path:

$$\mathrm{CF} \stackrel{365 \mathrm{nm}}{\to }\mathrm{TT}$$

Experiments performed for CF-H to characterize the influence of the UV irradiation power on the TT absorbance signal (Figure 2a) show the slope of this relation to be 1.5 in agreement with reported data [28]. The slope value close to 2 is expected for a purely biphotonic process. The discrepancy can be explained by a small contribution of the single-photon reaction, additionally to the main consecutive two-photon absorption reaction path. We performed numerical simulations (see SI for details) for CF-H in cyclohexane using photophysical properties determined recently [28] that quantify for the first time the yields of both processes. A quantum yield of 0.003 for the single-photon channel explains the experimental data (Figure 2b, the slope of 1.5). The single photon process leading to TT has intriguing nature, since it requires the initial ring-opening followed by significant changes in geometry, which contradicts the intuition. However, one can expect that upon ring-opening process TC form is initially generated in the vibrationally excited state. Such excess of vibrational energy can lead the hot TC molecule to overcome relatively high (~1.2 eV) S₀-state energy barrier to form TT isomer.

The experimental slope determined for CF-OCH₃ is closer to unity (1.2, Figure 2a), which indicates a greater contribution of the single photon excitation path in the TT-OCH₃ formation (CF→TT) that correlates to a lower TT-OCH₃ signal level. The reason for the decrease in the share of biphotonic channel (CF→TC→TT) is linked to the photo-dynamics of TC-OCH₃ and needs thus to be determined precisely.

2.3. Photophysical Properties of TC-OCH₃ in the Singlet Excited State

Ultrafast transient UV-vis absorption experiments were performed for TC-OCH₃ in cyclohexane with photoexcitation at 475 nm (2.61 eV). In these experiments, a solution of CF-OCH₃ was under a continuous LED UV irradiation at 365 nm (3.40 eV) to ensure a constant TC-OCH₃ concentration. Although TT is also produced, its concentration is over 20 times lower than that of TC. Moreover the selected pump excitation wavelength at 475 nm (2.61 eV) favors TC excitation over TT on the basis of the respective absorption band locations (Figure 1b). Thus, the measured transient absorption spectra can be securely assigned to the sole excitation of TC-OCH₃. Figure 3a shows the evolution of the transient absorption bands, which resemble the data reported for TC-H [25]. The initial positive transient absorption band peaking at 525 nm (2.36 eV) corresponds to TC-OCH₃ in the singlet excited state (S₁→S_n), which is in agreement with theoretical calculations (Table S3). The band undergoes a substantial decay (88%) in the time window 0.3–2 ps which is concomitant with the recovery of the negative band peaking at 435 nm. The negative band corresponds to the depopulation band, i.e., depletion of the TC-OCH₃ in the S₀ state caused by laser pulse excitation at 475 nm (2.61 eV). The global analysis indicates two characteristic time-constants (see Figure 3b): 0.45 ps - related to the lifetime τ_S1 of TC-OCH₃ in the S₁ state, and 5.1 ps corresponding to the vibrationally hot S₀ species produced by S₁→S₀ internal conversion. The lifetime τ_S1 can be also obtained from analysis of the band integral kinetics (0.47 ± 0.05 ps, Figure S2). The offset shows a weak positive band at 535 nm (2.32 eV) (see the offset × 15 in Figure 3b), which is assigned to the triplet excited state T₁ produced by intersystem crossing S₁→T₁ as in the parent TC-H compound [25].

A significant shortening of singlet excited state lifetime for the methoxy TC derivative in comparison to the H-parent compound (0.47 vs. 0.87 ps) is observed in cyclohexane (Table 1).

Moreover, data comparison at 50 ps delay (Figure 3a) shows less pronounced S₀ depopulation. Both observations can be explained by a more effective S₁→S₀ internal conversion channel in TC-OCH₃ compound that rationalizes the decrease in TT formation in two photon process and then lower TT concentration in the photostationary state in comparison to those for the unsubstituted parent compound. The next step is to understand the reason for this increase in internal conversion rate and this requires the help of quantum chemical calculations.

2.4. Theoretical Modelling of TC→TT Photoisomerization Mechanism

In order to study the mechanism of the TC→TT photoisomerization reaction, it should be noted first that the molecular mechanism of rotation of the whole rotor unit vs. naphthalenone skeleton may be realized along the two different pathways (variations). They can be classified as single-twist or bicycle-pedal motion [30,31,32,33,34].

To visualize these two variations of the TC→TT photoisomerization process, the excited state (S₁) and ground-electronic state (S₀) two-dimensional minimum potential energy surfaces (S₁-PES, and S₀-PES, respectively) were constructed to compare the photoisomerization mechanism between the OCH₃ and H derivatives (see Figure 4). The molecule first evolves along the S₁-PES toward the region where the S₁→S₀ internal conversion process takes place (Figure 4a,b, for OCH₃ and H derivatives). Further evolution of the molecule follows the S₀-state gradient toward the stable minimum (Figure 4c,d). The PES minima of the ground state, S₀, and of the lowest excited state, S₁, were calculated using the MP2/cc-pVDZ and ADC(2)/cc-pVDZ methods, respectively. Each energy point at PES was obtained by optimization of the geometry of a given molecule imposing two constraints for driving coordinates: θ₁(C₁₄ = C₁) and θ₂(C₂ = C₃), separately, in a given electronic state. These two driving coordinates were frozen while all the remaining 3N-8 coordinates were optimized for each point in given electronic state. Thus, each PES is spread over the two driving coordinates: θ₁(C₁₄ = C₁) and θ₂(C₂ = C₃) (see Scheme 1, for definition) defined as the dihedral angles describing rotation about the respective double bond. Note also that the two phenyl rings in the rotor unit were distinguished and marked with letters: “a” and “b” in the molecular structure (see Scheme 1 and Figure 4) to discriminate between the single-twist and bicycle-pedal variant motions. In the excited state, the photoisomerization can be realized as the single-twist described as the rotation of the rotor unit vs. naphthalenone skeleton about the sole C₁₄ = C₁ bond and could be observed along the direction parallel to the θ₁(C₁₄ = C₁)-axes in Figure 4a,b, for OCH₃ and H derivatives, respectively. The isomerization process, however, can be realized alternatively – as a bicycle-pedal motion – in which the concerted rotation about the two double bonds: θ₁(C₁₄ = C₁) and θ₂(C₂ = C₃) takes place simultaneously. This movement can be seen as the two benzene rings: a and b moving in direction parallel to the naphthalenone moiety plane. The bicycle-pedal motion can be observed along the direction close to the diagonal of Figure 4a,b. It links upper-left and bottom-right corners of the respective S₁-PES so that the condition Δθ₁ = −Δθ₂ is approximately fulfilled.

Photoexcitation of TC isomer in its Franck-Condon region populates the ππ* excited state. Next, the excited-state relaxation proceeds in a barrierless fashion from the TC Franck-Condon region toward the relaxed excited-state minimum, S₁(TC) form. As shown for the unsubstituted molecule [25], this relaxation process is accompanied by elongation of the C₁₃ = O₄ carbonyl double bond which eventually becomes single and the excited state gains the nπ* character. This mechanism resembles that discovered for DNA bases in which the nπ* state is responsible for driving the system in the region of the conical intersection with the ground electronic state CI(nπ*/S₀) [35,36]. Thus, the populated excited-state minimum S₁(TC) is observed in the ultrafast experiment with transient absorption detection (Figure 3). This molecule evolves further along a few deactivation channels [25], including the one that drives the molecule toward the TT photoproduct. A detail analysis of the mechanism of this process was possible thanks to the use of the calculated PES. The relevant species obtained as a result of the excited-state geometry evolution are shown in Figure 5 and

Table 2. Comparison of geometric parameters (bond lengths, in Å, dihedral angles), adiabatic energies (E_a, in eV) for the excited state (S₁) geometries optimized at the ADC(2)/cc-pVDZ level of theory. σ and τ denote C₁₄−C₁−C₂−C₃ and C₅−C₁₃−C₁₄−C₁ dihedral angles, respectively.

S0- or S1-State Form	R1 C3−O4 [Å]	θ1 C13−C14=C1−C2 [°]	θ2 C1−C2=C3−C15 [°]	σ [°]	τ [°]	R2 C13−O4 [Å]	RD C1−C14 [Å]	RS C1−C2 [Å]	RE C2−C3 [Å]	Ea [eV]
H-derivative
S0 TC	4.165	2.5	−9.5	173.5	160.3	1.239	1.386	1.437	1.384	0.59
S1 TC	3.921	22.5	−18.7	−174.8	174.2	1.399	1.478	1.371	1.440	2.14
S1 TW	4.433	84.7	−14.9	−170.0	180.0	1.266	1.474	1.366	1.445	1.95
S1 BP	4.243	59.2	−68.5	−176.5	176.6	1.266	1.482	1.344	1.492	2.00
OCH3-derivative
S0 TC	4.086	1.8	−11.7	173.6	153.3	1.239	1.389	1.439	1.383	0.58
S1 TC	3.818	25.5	−27.0	−174.9	172.8	1.390	1.483	1.364	1.450	2.15
S1 BP	4.170	58.3	−69.1	−177.8	178.7	1.263	1.483	1.342	1.492	1.93
S1 TW	4.308	82.6	−13.7	−177.2	177.3	1.260	1.476	1.366	1.439	1.94

.

We start our analysis of the process with the ground-state S₀^TC geometry whose position is marked with the green dot in the upper-left corner of Figure 4a,b, respectively, for methoxy and H derivatives. The single-photon excitation of S₀^TC form of each molecule in a barrierless manner populates the excited-state S₁^TC minimum. The first difference between the molecules occurs in the position of the S₁^TC minimum on the corresponding S₁-PES (blue dot in the upper left corner of Figure 4a,b).

During the initial S₁-state relaxation process of TC, both double bonds (θ₁, θ₂) become twisted from their initial Franck-Condon region values (1.8°, −11.7°), for S₀^TC(OCH₃), down to (25.5°, −27°), for S₁^TC(OCH₃) minimum; and from the initial values (2.5°, −9.5°), for S₀^TC(H), down to (22.5°, −18.7°), for S₁^TC(H) minimum (see Table 2). In case of the H derivative, the S₁^TC geometry points toward the single-twist motion path along the direction parallel to the θ₁(C₁₄ = C₁)-axis (θ₁ = 22.5°, θ₂ = −18.7°).

However, the S₁^TC(OCH₃) geometry (θ₁ = 25.5°, θ₂ = −27.0°) lies much closer to the diagonal of the S₁-PES, where the θ₁ = −θ₂ bicycle-pedal condition is fulfilled. The S₁^TC form becomes an intermediate, whose geometry may determine further evolution of the molecule in the electronic excited state toward TT along the single-twist or bicycle-pedal pathway. These two mechanism variations populate a given type of the excited-state minimum: S₁^BP, for bicycle-pedal, or S₁^TW, for single-twist motion, respectively, shown in Figure 5.

For OCH₃ derivative (Figure 4a), the violet region forms a double minimum valley linking the two almost isoenergetic excited-state minimum geometries (blue dots): S₁^BP and S₁^TW separated by a very low energy barrier. S₁^TW [θ₁ = 82.6°, θ₂ = −13.7°] is located in the midpoint between the S₀^TC and S₀^TT geometries (green and red circles) illustrating twisting motion of the molecule. The second minimum, S₁^BP [θ₁ = 58.3°, θ₂ = −69.1°], is located in the region close to the diagonal of Figure 4a suggesting that the bicycle-pedal mechanism should prevail over the twisting motion in the OCH₃ molecule, for the energy reasons.

The S₁^BP and S₁^TW minima were also determined for the unsubstituted parent compound. In this case, the S₁^TW excited-state global minimum [θ₁ = 84.7°, θ₂ = −14.9°] is more energetically favored. It is by 0.05 eV more stable than that of the corresponding S₁^BP form [θ₁ = 59.2°, θ₂ = −68.5°]. As a consequence, the position of the deeper twist-type S₁^TW global minimum on S₁-PES suggests that the twisting motion prevails over the bicycle-pedal, for the model of unsubstituted compound. Furthermore in agreement with recently reported experiments [25] we can also rationalize that the use of polar solvent is another way to control the equilibrium between bicycle-pedal and single-twist motion leading to the suppression of the TT formation in the H derivative. Indeed, in a polar solvent, the bicycle-pedal pathway will be favored over single-twist motion since the former path leads to a more polar intermediate S₁^BP (Figure 5), which likely produces back TC form after S₁→S₀ internal conversion. Moreover, in a polar solvent such as acetonitrile, the lifetime τ_S1 of TC is shorter in comparison to that in a non-polar cyclohexane (0.27 ps vs. 0.47 ps for OCH₃ derivative and 0.31 ps vs. 0.87 ps for H derivative, see Table 1) [25].

Even though after photoexcitation of TC-OCH₃ the excited-state gradient pulls the molecule toward the S₁^BP region, the S₁-state PES still indicates the existence of a low-lying flat energy valley reaching toward the region located around θ₂ = −90° (blue region on the lower-left part of Figure 4c,d). This is the region illustrating sole single-twist motion around the C₂ = C₃ double bond of the H and OCH₃ derivatives. During this alternative twist motion, the proton attached to the C₂ carbon atom in allene chain is found in vicinity of the carbonyl oxygen atom, O₄, that pulls the proton to form a stable AP minimum. The ground-state energy surface S₀-PES of both molecules indicates the existence of a high-energy ground-state equilibrium naphthalenol AP form (see the structure in Table S2). Depopulation of the excited state in this region may generate the AP form in the electronic ground state. The formation of the AP form has been detected in NMR studies for the TC-H molecule [8].

2.5. Role of the Methoxy Group

The π-electron-donating character of the methoxy group has been studied recently in both, ground- [37] and excited state [38] of monosubstituted benzenes. The electron donating effect of the methoxy group is even greater in the excited S₁ state than in the S₀ state. The presence of a methoxy group in the species shown in Figure 5 results in a greater dipole moment of either the ground- or excited state due to a shift of electron-density from the methoxy group toward the carbonyl oxygen atom, O₄. The O₄ oxygen atom attracts phenyl b ring more in OCH₃ than in H derivative. It results in energetic stabilization of the corresponding excited-state species, S₁^BP and S₁^TW. In OCH₃ derivative, shortening of the C=O₄···H distance stabilizes the bicycle-pedal type excited-state species S₁^BP vs. the single-twist one. Thus, we claim here that the role of the methoxy group is mostly to shift electron density toward the carbonyl oxygen atom which attracts the whole rotor unit more strongly than in unsubstituted derivative. This strengthen attraction results in a slight tilt of the rotor toward the carbonyl group in all the structures in the singlet excited state (C₁₀-C₁ and C₁₀-C₂ distances increase by 0.1 Å). The electronic effect is affecting the photophysics of the process and is dominating over the steric effect caused by the incorporation of the relatively small methoxy group into the molecule. This thesis is supported by experimental finding that 10-bromo substituted compound photoproduces a similar TT yield as the H derivative [21]. However the steric effect with more bulky substituents needs further investigations.

The molecular system in the excited S₁ state can deactivate to the electronic ground state, S₀, through S₁→S₀ internal conversion. Thus, the ground state S₀-PES was constructed for both molecules (see Figure 4a,b, for H and OCH₃, respectively). The height of the S₀-state energy barrier precludes the TC→TT photoisomerization process to be thermally activated.

As one can see, the blue dot representing the geometry of twisting type excited-state S₁^TW minimum is located close to the top of the green region representing the S₀-state energy barrier for both molecules. The system after S₁-state deactivation follows the S₀-state energy gradient and eventually gains toward the TT photoproduct (S₀^TT), or goes back to TC form (S₀^TC).

In contrast to the twisting motion, the blue dot representing the bicycle-pedal type S₁^BP minimum is positioned on the left side of the S₀-state energy barrier. The deactivation of S₁^BP favors return of the system toward the ground-state TC form, since once being in the ground state, the molecule meets a high barrier for the TC→TT isomerization and the ground-state energy gradient pulls it back to TC isomer. Thus, the channel path populating the S₁^BP minimum may result in stopping the photoisomerization process. The bicycle-pedal motion in the S₁-state prevails over the single-twist for OCH₃ derivative, thus a lower TC→TT isomerization quantum yield for this molecule can be expected, which is in agreement with experimental finding.

3. Materials and Methods

3.1. Materials

CF-OCH₃ (10-methoxy-3,3-diphenyl-3H-benzo[f]chromene) was synthesized and purified accordingly to previously described procedures [21]. CF-H (3,3-diphenyl-3H-benzo[f]chromene) was purchased from TCI (Springfield, Virginia, USA).

3.2. Transient UV-Vis Absorption Spectra Over Seconds

UV LED (λ_exc = 365 nm, M365LP1, Thorlabs, Newton, New Jersey, USA) was used to perform photochromic reaction in a 1 cm × 1 cm fused silica cuvette placed in a temperature-controlled cuvette holder (Flash 300, Quantum Northwest, Liberty Lake, Washington, USA) with stirring (see Scheme S1). The volume of the solution was about 1.5 mL. Changes in UV-vis absorption spectra over seconds were recorded by a FLAME-T-VIS-NIR-ES USB spectrometer (6 s⁻¹ sampling rate, Ocean Optics, Largo, Florida, USA). White-light of 150 W xenon lamp (Applied Photophysics, Leatherhead, Surrey, United Kingdom) with intensity reduced to a small level was used as a probing beam. 150 ms were used for accumulation of each white-light continuum spectrum. An average of 60 initial spectra before sample UV-irradiation was used to calculate ${\mathrm{I}}_0\left(\mathsf{\lambda }\right)$ spectrum, the next spectra $\mathrm{I}\left(\mathsf{\lambda }\right)$ at subsequent times were measured upon or after UV-irradiation to determine the transient absorption spectra accordingly to formula: $\Delta \mathrm{A}\left(\mathsf{\lambda }\right)=\log \frac{{\mathrm{I}}_0\left(\mathsf{\lambda }\right)}{\mathrm{I}\left(\mathsf{\lambda }\right)}$.

In order to study influence of UV irradiation power on TT formation a V-550 spectrophotometer (Jasco, Hachioji-shi, Tokyo) equipped with the mentioned above temperature-controlled cuvette holder, stirring and UV LED source was used.

3.3. Transient UV-Vis Absorption Over Picoseconds

Femtosecond UV-vis transient absorption spectra were obtained using a commercially available system (Ultrafast Systems, Helios, Sarasota, Florida, USA) [39]. Photostationary state of TC was obtained by continuous UV irradiation (5 mW/cm²) at 365 nm of CF solution in a quartz cell 2 mm thick with stirring.

The instrument response function (IRF) was about 200 fs (FWHM) estimated from a stimulated Raman scattering observed in pure solvent. The energy of laser pulses at 475 nm was 1 μJ. The data set was composed of a series of 280 spectra acquired for pump-probe delay ranging from -1 to 50 ps. Each transient absorption $\Delta \mathrm{A}$ spectrum was calculated from 500 pairs of probe (pumpON) and reference (pumpOFF) spectra, it corresponded to 1 s of acquisition time (see Scheme S2). The obtained transient absorption spectra were analyzed using the global fitting procedure (ASUFIT program), satisfactory fits were obtained with a double-exponential function. Convolution with the instrument response function was included into the fitting procedure.

The accuracies of the obtained time-constants derived from analysis of transient absorption results were following: ± 5% (UV-vis data in time window over tens of seconds) and ± 10% (ultrafast UV-vis data).

3.4. Computational Details

The equilibrium geometries of the OCH₃ and H derivatives of 3H-naphthopyran and its isomers in their closed-shell ground state (S₀) were obtained with the MP2 method [40] without imposing any symmetry constrains. The energy of the most stable form CF is the reference energy for higher energy structures. The excited-state (S₁) equilibrium geometries were determined with the second-order algebraic diagrammatic construction ADC(2) method [41,42,43]. The correlation-consistent valence double zeta basis set with polarization functions on all atoms (cc-pVDZ) was used in these calculations as well as in potential energy surfaces [44]. The vertical excitation energies and response properties of the lowest excited singlet states were calculated using the ADC(2) and CC2 methods [45,46]. The aug-cc-pVDZ basis set augmented with the diffuse functions was also used to compute vertical excitation energies of the OCH₃ and H molecular system. All calculations were performed using TURBOMOLE program package [47].

4. Conclusions

The experiments in cyclohexane show TT photoreaction yield four times lower for methoxy derivative than for the parent unsubstituted compound. Ultrafast experiments demonstrated that the excited state lifetime τ_S1 of TC-OCH₃ derivative is shorter (0.47 ± 0.05 ps) than that of TC-H (0.87 ± 0.09 ps) due to more effective radiationless S₁→S₀ internal conversion channel, populating back the TC form. Theoretical studies show two possible variants of photoisomerization reaction: (i) single-twist and (ii) bicycle-pedal motions. As demonstrated by the calculations, in photoexcited TC-OCH₃ the bicycle-pedal motion is energetically more favored than the single-twist. The bicycle-pedal motion leads to a minimum in S₁^BP which is located in a close proximity of S₀-PES facilitating S₁→S₀ internal conversion, which through further structural evolution leads toward the relaxed TC minimum. The bicycle-pedal channel is less probable in H counterpart, which favors population of the single-twisted form S₁^TW. The excited-state decay of S₁^TW leads to the ground state at the geometry close-lying to the S₀-state isomerization barrier, thus the energy gradient can drive the molecule toward TT or back to the TC minimum. Our studies have shown that the choice of appropriate substituent or polar solvent can favor the bicycle-pedal TC→TT isomerization path, which can be aborted leading to the decrease in TT formation yield. Altogether, the bicycle-pedal path is seen as photostabilizing channel for TC form and opens the way to design new efficient naphthopyrans free from TT formation.