2 + 2 Can Make Nearly a Thousand! Comparison of Di- and Tetra-Meso-Alkyl-Substituted Porphycenes

Two porphycenes, substituted at the meso positions with two and four methyl groups, respectively, reveal similar absorption spectra, but their photophysical properties are completely different. 9,20-dimethylporphycene emits fluorescence with about 20% quantum yield, independent of the solvent. In contrast, fluorescence of 9,10,19,20-tetramethylporphycene is extremely weak in nonviscous solvents, but it can be recovered by placing the chromophore in a rigid environment. We propose a model that explains these differences, based on calculations and structural analogies with other extremely weakly emitting derivatives, dibenzo[cde,mno]porphycenes. The efficient S1 deactivation involves delocalization of two inner cavity protons coupled with proton translocation toward a high-energy cis tautomer. The latter process leads to distortion from planarity. The probability of deactivation increases with the strength of the intramolecular NH···N hydrogen bonds. The model also explains the observation of biexponential fluorescence decay in weakly emitting porphycenes. It can be extended to other derivatives, in particular, the asymmetrically substituted ones. We also point to the possibility of using specific porphycenes as viscosity sensors, in particular, when working in single molecule regime.


■ INTRODUCTION
Arguably, the majority of optical-sensing techniques are based on luminescence, fluorescence, in particular. 1,2 These techniques are widely used in various areas, ranging from biology, medicine, and pharmacology to materials and environmental sciences. Fluorescence sensors exploit the sensitivity of the emission to the environment characteristics, such as temperature, viscosity, and polarity, 3−6 or to the presence of specific analytes, for example, metal ions. 7 Obviously, a rational design of a sensor must be preceded by thorough characterization and understanding of its photophysical properties. In this work, we describe a series of compounds based on porphycene which seem to be a very good candidate for applications as viscosity sensors.
Porphycenes, structural isomers of porphyrins, have been the objects of intense investigations since the synthesis of the parent compound reported in 1986. 8 Among these studies, particularly important in terms of possible applications are those devoted to spectral and photophysical properties. 9,10 Porphycenes are very promising as sensitizers in photodynamic therapy, 11 the main, but not the only theme of numerous patents. Other areas of possible use of porphycenes based on interaction with light include photoelectrochemical devices, 12 near-IR absorbing and emitting fluorophores, 13−15 preparation of new electrochromic, 16 liquid crystalline, 17 and elastomeric 18 materials. However, comparative research on the photophysical properties of different porphycenes has been rather limited. Most of porphycenes studied until now reveal quite large fluorescence quantum yields and nanosecond decay times. A noteworthy exception is that of meso-tetraalkyl-substituted porphycenes, 9,10,19,20-tetramethyl-, and tetrapropyl-derivatives. 19 Roomtemperature fluorescence of these molecules is extremely sensitive to solvent properties, in particular, viscosity. In lowviscosity solvents, fluorescence is extremely weak (quantum yield of 10 −4 to 10 −3 ), whereas it dramatically increases in paraffin and even more (by about 3 orders of magnitude) for molecules embedded in polymer films. It was demonstrated that the ultrafast deactivation of the excited state is due to S 0 ← S 1 internal conversion, and the triplet state is not populated. 19 Moreover, specific interactions with the solvent could also be eliminated as a possible origin of this effect, because the deactivation is equally efficient in nonpolar, polar protic, and polar aprotic solvents, as long as the viscosity remains similar.
The data accumulated so far strongly suggest a link between the photophysical and tautomeric properties; the latter have been extensively investigated for porphycenes in various environments, ranging from liquid and solid condensed phases, 20−28 supersonic jets, 29−32 and helium nanodroplets, 33 to the regime of single molecules probed by fluorescence, 34−37 Raman, 38 and scanning probe microscopies. 39−45 We have proposed the explanation of the unusual photophysical behavior of meso-alkylated porphycenes using a model which assumes that the species responsible for very efficient nonradiative S 1 depopulation is a rare cis2 tautomeric form (Scheme 1). 46 In contrast to the other tautomeric species of porphycene, such a structure is nonplanar. The observed viscosity dependence can be understood by realizing that a large amplitude molecular motion resulting in distortion from planarity along the trans−cis2 tautomerization path is required to initiate rapid excited-state deactivation. Assuming that the deactivation rate is proportional to the trans(S 1 )−cis2 energy difference, the calculations predicted that the efficient deactivation should increase in the order: parent porphycene, 9,20-dimethylporphycene, 2, 3,6,7,12,13,16,17-octamethylporphycene, 9,10,19,20-tetramethylporphycene. This prediction was experimentally confirmed for the parent, tetra-and octasubstituted porphycenes, but the doubly substituted derivative was not known at that time. This motivated us to obtain and study 9,20-dimethylporphycene. In this work, we compare the spectral and photophysical properties of this compound (2) and a related, 2,7-tetra-t-butyl-10,19-dimethylporphycene (3), with the data obtained, on the one hand, for the parent porphycene (1), its mono-(6), di- (7), and tetra-β-substituted (8) derivatives, and, on the other hand, with the characteristics of meso-tetramethyl (4) and meso-tetrapropylporphycene (5) (Scheme 2). A spectacular finding is that removing two out of four meso alkyl substituents leads to a complete change in photophysics and tautomerism. In contrast to 4 and 5, both 2 and 3 strongly fluoresce, irrespective of solvent viscosity. Moreover, only one tautomeric (trans) form is detected, in contrast to the case of meso-tetraalkyl-substitued porphycenes, where both trans and cis species are present. This behavior can be rationalized on the basis of calculations, which reveal very different geometry changes along the trans−cis2 pathway in 2 as compared to 4. For the latter, a rather unusual pattern of out-ofplane bending which accompanies the trans to cis2 tautomerization is obtained. A refined model is proposed that postulates that a small distance between the H-bonded inner nitrogen atoms leads to delocalization of inner hydrogen atoms, which is responsible for efficient nonradiative S 1 depopulation. This model seems to be general, as it can also explain the biexponential fluorescence decay profiles observed in 4, 5, and several asymmetric porphycenes. It also accounts for the lack of fluorescence and ultrafast, temperature-independent S 1 −S 0 internal conversion observed in both, planar and nonplanar dibenzo[cde,mno]porphycenes.

Scheme 2. Investigated Porphycenes
The Journal of Physical Chemistry A pubs.acs.org/JPCA Article Electronic absorption spectra were measured on Shimadzu UV 3100 and 2700 spectrophotometers. Fluorescence spectra were recorded using FS 900 CDT or FS5 spectrofluorometers (Edinburgh Instruments). The FS5 instrument was used for determination of fluorescence quantum yields of 1 in various solvents, applying a recently developed SAFE method. 51−53 The quantum yields were calculated with respect to rhodamine 6G water solution (QY = 0.92). 54,55 The solvents used for the fluorescence measurements were of the highest available purity. Acetonitrile, n-hexane, dimethyl sulfoxide, methanol, ethanol, 1-butanol, toluene, liquid paraffin, and dichloromethane were acquired from Merck (Uvasol grade solvents). Dodecane, tetradecane (both with purity >99%), and methanol-d 1 (monodeuteromethanol, 99 atom % D) were purchased from Aldrich.
Fluorescence lifetimes were obtained on homebuilt setups using time-resolved single photon counting. Picoquant LDH-D-C-375 laser (379 nm, 60 ps pulse) or Fianium FemtoPower 1060 supercontinuum laser ( Alternatively, fluorescence lifetimes were also obtained using a Becker & Hickl Simple-Tau 150 module. The emitted photons were collected by means of a multiwavelength detection system: a polychromator and a 16 channel TCSPC detector (PML-16, Becker & Hickl), which registers 16 individual fluorescence decays spanning a 200 nm spectral range (12.5 nm per channel). PicoQuant 375 nm laser diode (LDH-D-C-375) with a PDL 800-D controller was used for excitation. The impulse response function signal measured for this setup was 130 ps fwhm. The data were collected using SPCM software. The analysis of fluorescence decays was performed using SPCImage 5.0 package.

■ CALCULATIONS
Quantum chemical modeling of the ground and the lowest excited states was performed at the density functional theory (DFT) and time-dependent DFT (TD-DFT) levels, respectively. The hybrid B3LYP functional was used. For the excited states, we used also its long-range-corrected CAM-B3LYP version, 56 because it provides a better description of excited states with a charge-transfer character. The 6-31+G(d,p) basis set was employed.
Tautomerization reaction paths were obtained using generalized internal coordinates (GICs). The difference between distances of a moving hydrogen atom from donor and acceptor nitrogen atoms (r NH −r HN ) was specified as the only one fixed and scanned coordinate. Transition states were fully optimized with TS or saddle = 2 keywords. All calculations were carried out using Gaussian 09 57 or Gaussian 16 58 suites of programs.

■ RESULTS AND DISCUSSION
Absorption and emission spectra of unsubstituted, parent porphycene 1, meso-dimethyl derivative 2, and meso-tetramethyl   The Journal of Physical Chemistry A pubs.acs.org/JPCA Article viscosity. For each compound, solvent dependence of the band shape and position is very weak, both in absorption and fluorescence. As could have been expected, the spectra are somewhat better structured in nonpolar solvents. Alkyl substitution at the meso position results in a moderate red shift that scales in the approximately linear fashion with the number of substituents. Thus, the S 0 −S 1 transition energy in 2 lies about 400 cm −1 lower than in 1, whereas the difference between 4 and 1 amounts to ca. 800 cm −1 . These trends are correctly predicted by TD-DFT calculations.
The additivity breaks down completely with respect to the photophysical parameters (Table 1). Double meso substitution leads to approximately 50% reduction in fluorescence quantum yield and lifetime. In contrast, upon adding two more methyl groups fluorescence intensity decreases dramatically. For instance, the quantum yields of 1 and 4 in n-hexane or acetonitrile differ by 3 orders of magnitude. We have previously reported very low fluorescence quantum yield of 4 and 5 and detected that the emission of meso-tetraalkyl-substituted porphycenes is crucially dependent on solvent viscosity. 19 In a rigid environment, obtained by freezing the solution or, alternatively, by embedding the chromophore in a polymer matrix at room temperature, fluorescence intensity and lifetime largely increase, reaching the values comparable with those obtained for 2. On the other hand, Table 1 shows that the emission of the doubly substituted porphycene 2 does not exhibit any viscosity dependence, revealing the same quantum yield in nonviscous n-hexane and in viscous paraffin.
We have also synthesized another derivative, 2,7-tetra-t-butyl-10,19-dimethylporphycene (3), bearing four alkyl substituents in a different arrangement than in 4 and 5: two at the meso and two at the β positions. The photophysical properties of 2 and 3 are practically the same (Table 1): both molecules show quite strong fluorescence, not sensitive to solvent viscosity, polarity, and proticity.
Similar photophysical behavior has been observed for porphycenes substituted with one, two, and four tert-butyl groups at the β positions (molecules 6, 7, and 8, respectively). All these porphycenes fluoresce nearly as efficiently as The Journal of Physical Chemistry A pubs.acs.org/JPCA Article porphycene 1 ( Table 1). As in the cases of 2 and 3, no significant solvent dependence could be detected. These results imply that the huge decrease of fluorescence intensity observed for 4 and 5 is an effect specific for meso substitution. What needs to be explained, however, is the complete lack of additivity: the drastic reduction of fluorescence intensity is only observed for porphycenes with four alkyl groups, but not when two such substituents are present at the meso positions.
As mentioned in the Introduction section, calculations based on the model of deactivation involving the role of the nonplanar cis2 tautomer predicted that the fluorescence should decrease in the order 1 > 2 > 4. 46 However, because only the stationary structures were considered, the reaction path leading from the initially excited form to the species that is rapidly deactivated remained unknown. We have, therefore, attempted to carry out calculations aiming to elucidate the mechanism of photoinduced tautomerization, in order to understand the completely different photophysics of doubly and quadruply mesosubstituted porphycenes. The goal was to compare trans−cis1 and trans−cis2 paths in dimethyl-substituted porphycene 2 and tetramethyl derivative 4. The results of calculations are presented in Table 2 and Figure 4.
The obtained trans−cis1 tautomerization profiles are similar in the two molecules, both in S 0 and S 1 . In contrast, a dramatic difference was revealed for the trans−cis2 conversion. Figure 5 compares geometry changes calculated for the ground state of 2 and 4 along the trans−cis2 tautomerization path. A completely different picture is obtained for the two molecules. The dimethyl derivative 2 remains planar until the moving hydrogen is The Journal of Physical Chemistry A pubs.acs.org/JPCA Article transferred to the other nitrogen atom. Then, as expected, two pyrrole rings assume an out-of-plane conformation, with the two NH bonds pointing upward and downward, respectively (cf. also Scheme 1). In contrast, while the optimized cis2 form of tetramethylporphycene 4 has essentially the same structure as in 2, the path leading to it from the trans species is completely different. Already at the initial stage of hydrogen transfer, the molecule becomes strongly nonplanar, assuming a saddle-like shape ( Figure 5).
The trans−cis2 energy profile calculated for 2 in S 1 is similar to that obtained for S 0 (Figure 4). On the other hand, the full S 1 tautomerization profile could not be obtained for 4 because the energy was not converging, even before reaching the maximum along the reaction path. The molecule was becoming strongly bent, and its ground-state energy was going strongly upward, leading to a very small S 0 −S 1 energy gap. This behavior, suggesting the presence of S 1 −S 0 conical intersection, which TD-DFT calculations are not able to handle properly, could have been expected. It should also be noted that the S 1 −S 2 energy gap in meso-tetraalkylporphycenes is very small (<700 cm −1 compared to ca. 1000 cm −1 in parent porphycene 59 ). The situation can be even more complicated: the orbital nature of the optimized form of cis2 in the lowest excited state of 4 was found to correspond to an nπ* transition, of which the energy significantly decreases as the hydrogen atom migrates from the trans to the cis2 tautomer. This close proximity of several electronic transitions, combined with the size of the molecule, makes the computational task of finding the structure responsible for efficient S 1 −S 0 deactivation extremely challenging, if not impossible. Our future plans include search for conical intersections using such methods as, for example, computationally relatively inexpensive spin-flip TD-DFT. In terms of explaining the nature of the ultrafast depopulation of excited meso-tetraalkylporphycenes, the most significant result is the nonplanar structure of 4 along the trans−cis2 tautomerization path. Close inspection of the nonplanar geometry reveals structural similarity to 3,6-13,16-dibenzo-[cde,mno]porphycene (Figure 6), of which two derivatives, 2,7,12,17-tetramethyl-and 2,7,12,17-tetra-tert-butyl-, have been experimentally investigated in our laboratory. 60,61 Both were found to be nonemissive, not only at room, but also at cryogenic temperatures. Picosecond transient absorption studies revealed that the recovery of the ground-state population via S 1 −S 0 internal conversion occurs in 10−20 ps in a rigid argon matrix at 27 K. Interestingly, dibenzoporphycene substituted with four methyl groups is not planar, but the tetra-tert-butyl analogue is. This indicates that the nonplanarity is not a necessary requirement for efficient excited-state deactivation. What matters is the very small distance between the nitrogen atoms, which results in a very strong hydrogen bond and, finally, proton delocalization. Indeed, while the calculations performed for dibenzoporphycenes predict a double minimum potential with extremely small barriers for trans−trans, cis1−cis1, and cis1− trans conversions (Table S1), inclusion of zero-point energies results in delocalization of both inner hydrogen atoms.
Based on the above experimental and theoretical findings, we propose that delocalization of inner hydrogens is responsible for the efficient S 1 deactivation in porphycenes. Once the structure   The Journal of Physical Chemistry A pubs.acs.org/JPCA Article with delocalized inner hydrogens has been achieved, the molecule undergoes fast S 0 ← S 1 radiationless transition ( Figure  7). The probability of obtaining the appropriate geometry is a strong function of the NH···N distance and is the largest in dibenzoporphycenes, for which this distance is the shortest (2.51 Å). In this case, the ground-state geometry, planar or not, ensures rapid S 1 depopulation. Therefore, lowering of temperature and/or placing the molecule in a rigid environment has no effect on the emission, which is weak under any conditions. In contrast, in meso-tetraalkylporphycenes, NH···N separation is slightly larger (2.53 Å) and, in order to reach the "delocalized" geometry, distortion from planarity is required (cf. Figure 5). Because this distortion represents a large amplitude motion, one can expect viscosity dependence, exactly as observed experimentally.
Our model can explain another, rather intriguing finding: observation of biexponential emission decay in 4, with the lifetimes increasing from picoseconds to nanoseconds upon passing from solutions to a rigid environment. We can now propose that these lifetimes reflect the times required to reach the point of efficient S 1 depopulation from the initially excited trans and cis1 species. Naturally, as the solvent is getting more viscous, the probability of reaching the nonplanar configuration decreases.
The proposed scheme of S 1 deactivation is similar to the model put forward by Nonell and co-workers in order to explain the photophysics of asymmetric porphycenes. 24,62 Analyzing the properties of 9-substituted n-propyl and methoxyethylporphycenes, they came to the conclusion that "tautomerization is coupled with an efficient deactivation pathway". 24 The cis2 tautomer was postulated as the species responsible for efficient S 1 deactivation. 62 Our present results indicate that deactivation may occur along the trans−cis2 tautomerization coordinate before reaching the relaxed cis2 structure. Such a process is resemblant of fluorescence quenching by "unsuccessful" or "aborted" chemical reaction, a theoretical model proposed by Olivucci and co-workers. 63 One should note that hydrogen delocalization, which we postulate to be crucial for S 1 deactivation, occurs between trans and cis1 forms, which can be illustrated by the vertical movement of one of the protons in Scheme 1, but the efficient radiationless transition is triggered by the proton movement toward cis2, that is, in the horizontal directions. Such situation is not unusual. It has been demonstrated by calculations that the proton density in hydrogen bonds may have a disc-like shape, with the plane of the disc perpendicular to the H-bond axis. 64

■ SUMMARY AND CONCLUSIONS
Experimental and theoretical results clearly indicate the link between the unusual photophysical behavior and tautomeric properties of porphycenes, as illustrated by the scheme of excited scheme tautomerism in meso-tetraalkylated porphycenes, as shown in Figure 7. We demonstrated earlier that in the electronic ground state of 4 and 5, trans and cis1 tautomers coexist in about 4:1 ratio. 19 In the same work, it was found that (i) the excited state decay is slower than trans−trans conversion (picoseconds vs 100 fs or less, respectively) and that (ii) no equilibrium between trans and cis1 is established in S 1 , as demonstrated by bi-exponential fluorescence decay. These observations were explained by the multidimensional character of the tautomerization coordinate, involving the rotation of the alkyl moiety. We now refine this model, postulating that the probability of rapid S 1 depopulation should, in general, increase with the strength of intramolecular H-bonds and therefore need not be substituent-specific. In fact, efficient deactivation can be expected for any substituent that modifies the inner cavity dimensions in a way that leads to a stronger H-bond. This hypothesis is experimentally confirmed by our ongoing research on mixed meso aryl/alkyl as well as halogenated porphycenes.
In the present model, the distance between the H-bonded N atoms seems to be a crucial parameter that determines the probability of fast excited state deactivation. We note, however, that low fluorescence quantum yields (albeit about an order of magnitude higher than in 4 and 5) have been found also for porphycenes substituted with eight alkyl groups on the pyrrole rings. In these molecules, the NH···N distances are significantly larger than those in "normally" emitting porphycenes, 2.80 versus ca. 2.60 Å 10 (see also Table S1). However, the octaalkyl derivatives were found to be nonplanar in S 0 . Moreover, the "horizontal" N···N distance is now smaller than that in other porphycenes. These factors can partially compensate the effect of weak H-bonds.
The two fluorescence decay times observed in 4 and 5, assigned to the decay of trans and cis1 forms, indicate that these tautomeric forms are not in equilibrium. This strongly suggests that the excited-state trans−cis1 tautomerization path cannot bypass the molecular geometry responsible for the efficient internal conversion to S 0 . If this were to be the case, the trans− cis1 equilibrium would be established for the overwhelming majority of excited molecules in the time much shorter than the experimentally measured fluorescence decays, in particular, those obtained for viscous solvents.
Unfortunately, it was not possible to safely assign each of the decay components in 4 and 5 to a specific form by selectively exciting and probing at specific wavelengths because both absorption and emission of trans and cis1 lie at practically the same energies (calculations predict a difference of only 100 cm −1 for S 0 −S 1 transitions in the two tautomers).
The scheme presented in Figure 7 is quite general. We believe that it can be applied for both symmetrically and asymmetrically substituted porphycenes. For the latter, it is natural to assume that the trans−trans tautomerization occurs in a stepwise fashion with the cis1 intermediate, because the two intramolecular NH···N bonds have different strengths. Also, in symmetric derivatives, the stepwise mechanism should be favored at high temperatures, as demonstrated recently by nuclear quantum dynamics calculations. 65 In order to further test the present model, it would be instructive to find other arguments for proton delocalization. As of now, we note a significant difference in the absorption spectra of 4 recorded in the Q region for two isotopomers: the nondeuterated molecule and the species with two inner protons replaced by deuterons. 19 The spectra are much broader and lessstructured in the former. For parent porphycene, such an effect is not observed. Another relevant feature is the activation energy of the process responsible for temperature dependence of fluorescence determined for 4 and 5 in the MeOH/EtOH and MeOD/EtOD mixtures. For both molecules, a larger value was obtained for deuterated solvents. 19 Finally, we consider a possibility of using custom-designed porphycenes as viscosity sensors. It may at first appear unrealistic because the fluorescence quantum yields of viscosity-sensitive porphycenes are quite low. However, when single molecule fluorescence sensors are considered, the crucial factor is not the emission quantum yield, but the yield of triplet formation and the triplet lifetime. A molecule trapped in the dark The Journal of Physical Chemistry A pubs.acs.org/JPCA Article "OFF" state does not emit fluorescence. Because the triplet lifetimes are of the order of several hundred microseconds under typical conditions of single molecule experiments (thin polymer film at room temperature), the number of photons emitted per time unit is drastically reduced for chromophores with considerable yields of triplet formation. It is instructive, in this context, to compare strongly emitting 1 with weakly emitting 4 and 5. Parent porphycene 1 has the fluorescence quantum yield of ca. 50% and a similar yield of triplet formation, whereas 4 or 5 emit fluorescence with quantum yields of ca. 10 −3 , but they exhibit negligible triplet formation yields. It can be readily simulated (assuming the triplet lifetime of 200 μs, a value we know from the experiment) that the number of photons emitted per time unit in the single molecule regime will be orders of magnitude larger for seemingly "weaker" luminophores. An additional advantage of these molecules may be higher photostability, because photodecomposition usually involves the triplet state.
Other obvious advantages of porphycenes are: (i) large scale of fluorescence intensity changes with solvent viscosity; (ii) insensitivity to temperature (similar intensities and lifetimes for low temperature glasses and room temperature polymer matrices); (iii) insensitivity to solvent polarity (although this may not be true for certain porphycenes); (iv) strong absorption in the convenient spectral region; (v) emission located in the red part of the visible range. Initial studies indicate that porphycene 4 fulfills all the above criteria. A particularly attractive variant may involve combining two porphycenes, such as, for example, 2 and 4, that differ dramatically in the dependence of fluorescence quantum yield on viscosity, otherwise being very similar. Such a sensor could work in the ratiometric mode, enhancing sensitivity and accuracy.
Calculated structural and spectral parameters for several porphycenes (PDF)