Solvent effects on the photooxidation of indolepyrazines

Photodestruction of 2-(pyrazin-2′-yl)-1H-indole and 2,5-di(1H-indol-2′-yl)pyrazine involves singlet oxygen generation and its rapid insertion into the indole ring with the formation of benzoxazinone derivatives: 2-(pyrazin-2-yl)-4H-3,1-benzoxazin-4-one and 2-[5-(1H-indol-2-yl)pyrazin-2-yl]-4H-3,1-benzoxazin-4-one. The quantum yield of this reaction strongly depends on the environment. It is definitely smaller in protic methanol than in aprotic acetonitrile or n-hexane. The observed effect of photostabilization is explained by formation of hydrogen bonded complexes between the chromophore and alcohol, which results in lower triplet formation efficiency and, in consequence, decrease of singlet oxygen formation quantum yield.

In a previous work we have shown that 2-substitued benzoxazinone can be easily formed upon photooxidation of 2-substitued indole derivative in acetonitrile in the presence of air. No additional photosensitizer is necessary in this reaction. The substrate itself acts as a photosensitizer: after UV excitation the energy is transferred to the triplet state of the chromophore and, next, to molecular oxygen with the formation of singlet oxygen and, finally, the product [5].
Photophysical and photochemical properties of bifunctional azaaromatic indole derivatives that possess both the proton donor and proton acceptor abilities can be changed by the environment. In protic solvents these compounds form hydrogen-bonded complexes, which can effectively diminish their photodestruction. To explain this effect, two mechanisms were proposed, each leading to an increased rate of depopulation of the excited singlet state: (i) fast excited state double proton transfer occurring in 1:1 cyclic complexes with water or alcohol and (ii) efficient S 0 ← S 1 internal conversion in 1:2 stoichiometry complexes. Both of these processes are faster than T 1 ← S 1 intersystem crossing and can therefore diminish the quantum yield of triplet state formation, as well as the singlet oxygen formation efficiency [14][15][16][17][18].
In this report we investigate photostability of two indolepyrazines: 2-(pyrazin-2′-yl)-1H-indole (1) and 2,5-di(1H-indol-2′-yl)pyrazine (2). These molecules possess the same structural motif as 3-5, consisting of pyridine and pyrrole nitrogen atoms separated by three bonds, but the presence of pyrazine ring suggests that the electronic properties and the ability to form intra-and intermolecular hydrogen bonds may be different. We find that the quantum yields of photodegradation of 1 and 2 are comparable and the molecules are about ten times more photostable in the protic solvents than in aprotic ones. The photostabilization ability due to hydrogen bonding interactions is, however, significantly lower than that previously observed for compounds 3-5(30 to 200 times). The mechanism of the photoreaction involves singlet oxygen mediated oxidation of the indole ring of the studied chromophore with formation of benzoxazinone derivatives: 2-(pyrazin-2-yl)-4H-3,1-benzoxazin-4-one (12) and 2-[5-(1H-indol-2-yl)pyrazin-2-yl]-4H-3,1benzoxazin-4-one (14) for molecules 1 and 2, respectively. The formation of the photoproducts is slower and less efficient in methanol because the quantum yield of singlet oxygen formation in this solvent is diminished by hydrogen bonding interactions with the chromophore.

Synthesis of 1 and 2
The pyrazinyl indoles 1 and 2 were prepared from the corresponding phenylhydrazone derivatives 8 and 11 by application of the Fisher indole synthesis (Schemes 3 and 4).
NMR spectra were obtained on a General Electric QE-300 spectrometer at 300 MHz for 1 H NMR and at 75 MHz for 13 C NMR. Chemical shifts were reported in parts per million relative to the residual proton or carbon of the corresponding solvent. Melting points were measured on Thomas Hoover capillary melting point apparatus and are not corrected. Elemental analyses were carried out by QTI, P.O. Box 470, Whitehouse, NJ 08,888-0470. All chemicals are commercially available, and all solvents were reagent grade.

Solvents
The following solvents (spectroscopic or LCMS grade) were used: n-hexane, acetonitrile, methanol, 1-butanol, and water (Aldrich or Merck). Prior to measurements the solvents were carefully examined and showed no traces of absorption or emission.

Absorption and fluorescence
Electronic absorption spectra were measured with a Shimadzu UV 2700 spectrophotometer. Stationary emission and fluorescence excitation spectra were obtained using Scheme 4. Preparation of 2,5-di(1H-indol-2′-yl)pyrazine (2) 1 3 an Edinburgh FS 5 spectrofluorometer. The spectra were corrected for the spectral sensitivity of the instrument. Fluorescence curves, recorded as a function of wavelength (λ), were subsequently multiplied by λ 2 in order to convert counts per wavelength into counts per wavenumber. The quantum yield of fluorescence, Φ f , was determined by a reference method, using 0.05 mol dm −3 quinine sulfate in sulfuric acid (Φ f = 0.51) as the standard [20]. Fluorescence decay lifetimes were determined by time-resolved single photon counting (SPC) on a home-built SPC setup consisting of an IBH Nanoled emitting diode (297 nm, and pulse width < 750 ps), Becker&Hickl PMC 100-4 photomultiplier, Spectral Products Digikröm CM 110 monochromator, and PicoQuant TimeHarp 100 PC card. PicoQuant Fluofit software (version 3.3) was used to analyze the decay curves.

Singlet oxygen generation
The quantum yield of singlet oxygen generation, Φ Δ , was determined with a home-built setup. An Opotek Radiant 355 laser was used for excitation at 340 or 370 nm. The emission was detected using a BENTHAM DTMc300 double monochromator, equipped with a Hamamatsu H10330C-75 thermoelectrically cooled photomultiplier. In order to determine Φ Δ amplitudes of singlet oxygen phosphorescence decay curves were measured at 1275 nm, the emission maximum, photosensitized by the standard (signal amplitude A 0 ) and by the compound under investigation (amplitude A x ). Both samples were compared using the same solvent (acetonitrile or methanol, depending on the experiment).
, where 1 − 10 −OD 0 and 1-10 −ODx correspond to fractions of light absorbed by the phenalenone and the measured compound, respectively, at a given excitation wavelength. At room temperature the quantum yield of singlet oxygen generation for phenalenone is equal to 0.98 both in n-hexane and methanol [21].

Photodegradation
The photodegradation of the compounds was studied in the regime of 365 nm irradiation. Solutions of 1 and 2 (~ 5 × 10 -5 mol dm −3 ) were prepared in n-hexane, acetonitrile, and methanol and then irradiated in a sealed quartz cuvette. High power (270 mW) 365 nm LED diode (Roithner Laser Technik, H2A1-H365-r4) was used for irradiation. The light intensity was determined by potassium ferrioxalate actinometry as 2.17 × 10 16 quanta s −1 [22]. The photodegradation process was monitored by UV-Vis absorption measurements. The detailed procedure of the determination of the quantum yield of photodegradation for the studied compounds is described in Supporting Information.

LC-MS analysis
The photoproducts were analysed using Shimadzu Nexara X2 LC/MS/MS 8050 system with positive or negative ESI ionization methods. The samples dissolved in acetonitrile (concentration of about 5 × 10 -5 mol dm −3 ) were irradiated by the 365 nm UV diode in a quartz cuvette and next transferred into UPLC vials and diluted 5 times with the solvent. The samples were subsequently injected into the LC/MS/MS system for the determination of the masses of the substrate and the photoproducts. Chromatographic separation was achieved with a 00D-4424-B0 Synergi™ 4 µm Fusion-RP 80 Å, LC column 100 × 2 mm (Phenomenex). The ESI-MS settings were as follows: capillary voltage 4000 V (positive mode) and 3000 V (negative mode), nebulizing gas flow 3.0 L/ min, and drying gas flow 5 L/min at 300 °C. The mobile phase was composed of acetonitrile or methanol (90%) and water (10%). The flow rate was 0.4 mL/min with an isocratic elution. The injection volume was 2 μL. The column temperature was set at 40 °C. The mass-to-charge ratio (m/z) scan range was from 80 to 1000 (positive and negative). The data were processed using LabSolutions software (Shimadzu).

Quantum chemical calculations
Gaussian 09 software package was used [23]. All the calculations were performed by applying the density functional theory (DFT) method using B3LYP functional [24,25] with 6−31+G(d,p) basis set. Vertical excitation energies and excited-state geometries were obtained by the time-dependent DFT (TDDFT) method. Vibrational frequency analysis was performed to predict zero-point vibrational energies (ZPVE) of different forms. The counterpoise technique [26,27] was used to correct for the basis-set superposition error (BSSE). energy than the anti conformer. For 2 the DFT calculations predict that the syn-anti and anti-anti forms have energies higher than the most stable syn-syn conformer by about 16 and 31 kJ mol −1 , respectively. Figure 2 shows the simulated electronic absorption spectra of the rotameric forms of 1 and 2.

Results and discussion
It is well known that solvent can influence a chromophore by electrostatic or hydrogen bonding interactions and change its spectral, photophysical, and photochemical characteristics. Therefore, the electronic absorption and fluorescence spectra of 1 and 2 have been measured in three solvents of different polarity and proticity: n-hexane (aprotic, nonpolar), acetonitrile (aprotic, polar), and methanol (protic, polar) (Fig. 3).
Spectral and photophysical characteristics of the experimental absorption and emission spectra in different solvents are presented in Table 1.
The spectroscopic and photophysical data of 1 and 2 show a distinct change with the nature of the solvent ( Table 1). The position of the lowest absorption band of 1 and 2 is almost the same in all solvents, but the maximum of the emission band is moving to the red with increasing polarity of the solvent. For molecule 1 the observed Stokes shifts  are larger in all solvents than for molecule 2. Most probably this relates to the differences in dipole moments of these compounds in S 0 and S 1 states (compare Table S3 in the Supporting Information). The fluorescence quantum yields, Φ f , are about one hundred (for 1) and ten times (for 2) higher in aprotic solvents (n-hexane and acetonitrile) than in methanol. In all solvents 2 exhibits higher values of Φ f than observed for compounds 1 and 3-5 (see Table 1 and Table S1 in the Supporting Information for comparison). For this type of bifunctional indole derivatives the fluorescence quantum yields in n-hexane were reported to span the range from 0.99 for 2-(3΄-pyridyl)indole to 0.04 for 3,3΄-tetramethylene-2-(2΄-pyridyl)indole [28].
In nonpolar n-hexane the fluorescence decays are monoexponential and the lifetime values are 0.92 ns for 1 and 1.5 ns for 2. The quantum yields (Φ f ) and lifetimes (τ) obtained in this medium lead to large values of the radiative rate constants (Eq. (1)), 2.6 × 10 8 s −1 (1) and 4.1 × 10 8 s −1 (2). Radiative and radiationless rate constants were calculated using the Eqs. (1) and (2), respectively. The data are shown in the Table 1.
The electronic transition dipole moments in fluorescence M fl , calculated from Eq. (3) [29,30] are 4.7 D and 6.5 D for 1 and 2, respectively.
In this equation k nr is the radiationless rate constant, h is the Planck constant, n is the refractive index of the solvent, and ̃fl is the wavenumber of fluorescence maximum.Also the M fl values (which are the best quantitative measure of the transition probability) are comparable to those of the electronic transition dipole moments M abs (Eq. 4 [30]) related to the lowest 1 (π,π*) ← S 0 absorption band In this equation (̃) is the molar absorption coefficient, ̃a bs is the wavenumber of absorption maximum, N A is the Avogadro number, and c is the speed of light. The similarity of M fl and M abs values indicates a similar nature of the Franck-Condon excited states initially reached in absorption and the solvent-equilibrated fluorescent states. Most probably, the absorption and fluorescence in n-hexane correspond to the syn forms of 1 and 2. In this solvent monoexponential fluorescence decay was observed.
Compounds 1 and 2 in acetonitrile show biexponential fluorescence decay which, similarly to the case of 2-(2΄-pyridyl)pyrrole [31,32] and 3 and 4 [17] in polar aprotic media is due most probably to the presence of both syn and anti rotameric forms. In acetonitrile both decay times are longer than the monoexponential decay time observed in n-hexane. The stabilization mechanism of the anti forms of 1 in in this polar solvent, is not obvious, because the syn and anti forms have predicted similar dipole moment values of equal 1.4 D and 0.9 D, respectively. In the case of 2 the calculations predict zero dipole moment values for centrosymmetric syn-syn and anti-anti forms and 2.2 D for the syn-anti structure. These data suggest that in a case of 2 the syn-anti conformer can be stabilized in polar acetonitrile; the fluorescence decay data show that in acetonitrile a few percent of fluorescence originate from the syn-anti form with longer decay time values (~ 8 ns).
In alcohols both compounds have very short decay times (< 0.2 ns), probably originating from the emission of hydrogen-bonded syn forms of the compounds. For 1:1 syn  Comp Solvent̃ abs

Complexation with methanol
The B3LYP/6-31+G(d,p) optimized structures of the most stable complexes of 1 and 2 with methanol of 1:1 are shown in Figs. 4 and 5, respectively. The stabilization energies of these complexes are listed in Table 2. These results indicate that the most stable structures of 1:1 stoichiometry are cyclic complexes of the syn rotamers, with proton donating and accepting units connected by a methanol link with two cooperative hydrogen-bonds. This type of complexes is predicted to be the most stable in the case of all molecules 1-5. The complexes of 1-3 have similar binding energies (ΔE C ) about −27 kJ mol −1 , whereas for 4 and 5 the stabilization energies are about 10 kJ mol −1 lower (Table S2 in the Supporting  Information).
To establish the stoichiometry of the ground state complexes of 1 with alcohols we performed the titration experiments with mixed solvents [14]. Upon adding small amounts of 1-butanol to n-hexane solutions of 1 the growth of the low-energy side intensities and isosbestic points in the absorption spectra are observed (Fig. 6). This shows the formation of equilibrium between two species in the ground state, which are assigned to the uncomplexed and complexed forms of 1.
For a reaction: which involves one molecule of chromophore X and n molecules of alcohol Y the equilibrium constant, K 1n , may be expressed as follows: where The absorption data yielded the 1:1 stoichiometry of the complexes of 1 for which the equilibrium constants is equal to 12 (± 5) dm 3 mol −1 . The obtained equilibrium constant is similar to these observed for 3 (14 (± 10) dm 3 mol −1 ),  [17,18]. This observation is in line with the results of DFT calculations, which predict a much higher binding energy for the cyclic doubly hydrogen-bonded syn complexes of 4 and 5 than for complexes of 1 and 3 (Table S2 in Supporting  Information). For molecule 2, an analogous experiment was also done, but because of poor solubility of the chromophore in n-hexane the obtained data are not quantitatively reliable and are not presented here.

Singlet oxygen quantum yield
The determined values of quantum yield of singlet oxygen generation are higher in acetonitrile than for methanol solution for all studied compounds 1-5 ( Table 3). The largest difference is observed for compound 1. In this case the Φ Δ in alcohol is about 35 times lower than in acetonitrile. For compound 2 this difference is about ten times smaller than for 1.

Photodegradation quantum yield
Photostability of 1 and 2 dissolved in n-hexane, acetonitrile, and methanol was studied by measuring the absorption intensity during continuous illumination at 365 nm by a high power LED. Figure 7 shows the spectral changes observed during the UV irradiation of 1 and 2 in acetonitrile. Upon irradiation the absorption bands of the indolepyrazines gradually disappear and new bands appear, belonging to the photoproduct. The appearance of clear isosbestic points in the absorption spectrum suggests that under irradiation the stoichiometry of the photoreaction is preserved. Most probably the direct substrate to product transformation occurs and only a one photoproduct is formed. A similar spectral behaviour was observed also for n-hexane and methanol solutions of these compounds.
The photodegradation quantum yields, Φ pho , determined for molecules 1 and 2 dissolved in n-hexane, acetonitrile, and methanol are listed in Table 4 and compared with the values obtained previously for chromophores 3-5. A detailed description of the procedures used for the determination of Φ pho values is given in the Supporting Information. The data show that alcohol acts for all compounds as a photostabilizer. The molecules 1 and 2 are, respectively, 20 and 10 times more photostable in methanol than in acetonitrile. These results are comparable with the data obtained for 3, but the stabilizing effect of hydrogen bonding interactions is smaller than observed for 4 and 5.

Photoproducts
To establish the nature of the photoproducts, solutions of 1 and 2 before and after 365 nm irradiation were analyzed using LC-MS. Figures S1 and S2  Based on our previous studies on 3 [5] and the existing literature data on singlet oxygen-mediated photooxidation of 2-phenylindole [4], we assign the ions with m/z = 226.1 to 2-(pyrazin-2-yl)-4H-3,1-benzoxazin-4-one 1 3 (12) and the ions with m/z = 243.3 to the product of hydrolysis of 12, namely 2-[(pyrazine-2-carbonyl)amino]benzoic acid (13). The products of hydrolysis of 4H-3,1-benzoxazin-4-one derivatives with analogous structure were identified during the synthesis of benzoxazinones from 2-phenylindoles [2]. The LC-MS data obtained for 2 in acetonitrile show that upon UV irradiation the amount of positive ions with m/z = 311.1 decreases and a new ion appears with the m/z = 341.2 (see Figure S3 in Supporting Information). The new ion can be assigned to 2-[5-(1H-indol-2-yl)pyrazin-2-yl]-4H-3,1-benzoxazin-4-one (14). For 2 no negative MS signals which could be assigned to the substrate or product were observed. The schematic photooxidation pathways of 1 and 2 are presented in Schemes 5 and 6, respectively.

Summary
The Therefore, all the studied molecules 1-5 are more photostable in methanol than in acetonitrile or n-hexane.
The photodegradation quantum yields of molecules 1-3 are, respectively, 20, 10, and 16 times lower in methanol than in acetonitrile. For chromophores 4 and 5 this difference is larger: 39-and 147-fold. The explanation of the origin of the observed differences in photodegradation quantum yields of molecules 1-5 requires additional investigations. The photodegradation quantum yield (Φ pho ) is proportional to the quantum yield of triplet state formation (Φ T ), the rate constant of photoreaction (k pho ) and the triplet state lifetime (τ T ). These three factors need not correlate for different molecules. Therefore, it is extremely difficult to provide a general model for predicting relative photostability in a series of heteroazaaromatic compounds. Assuming that the quantum yield of singlet oxygen generation is a low limit value of the quantum yield . The number of alcohol molecules in the complex, n, is given by the slope of the plot (Eqs. 5 and 6 in the text). Absorbance changes were monitored at 385 nm (26,000 cm −1 ), n = 1.04 ± 0.04, correlation coefficient r = 0.986 Table 3 Comparison of the quantum yield of singlet oxygen formation Φ Δ determined in acetonitrile (ACN) and methanol (MeOH) for compounds 1-5 Excitation wavelength 340 nm; estimated error 10% of the triplet state formation, the photodegradation quantum yield can be correlated with the quantum yield of singlet oxygen formation. The observed lack of such correlation in the case of compounds 1-5 suggests that the triplet state lifetime is an additional important parameter to be involved in the qualitative characterization of the photodegradation quantum yield. In our previous publication we showed that the triplet states lifetime of molecule 5 differs significantly between nonpolar and polar solvents (from tens of ns in hexane to a few ns in acetonitrile) [18]. In order to assess the role of three different factors in the photostability, we plan to measure triplet lifetimes of molecules 1-4 in different solvents. 365 nm irradiation of 1 and 2 in acetonitrile or methanol at ambient temperature in the presence of air, similarly as was observed for 3 [5], leads to formation of the photoproduct with benzoxazinone structure. Starting from 1 leads to 2-(pyrazin-2-yl)-4H-3,1-benzoxazin-4-one (12). Additionally, a product of hydrolysis of 12, namely 2-[(pyrazine-2-carbonyl)amino]benzoic acid (13) was also observed.
In the case of 2 as a substrate, 2-[5-(1H-indol-2-yl) pyrazin-2-yl]-4H-3,1-benzoxazin-4-one (14) was identified as the photoproduct. The mechanism of photoreaction most likely involves two steps: (i) the singlet oxygen formation and (ii) its insertion into the indole ring of the Fig. 7 Changes in absorption observed upon 365 nm irradiation of 1 and 2 in acetonitrile. The spectra recorded before and after irradiation are represented by thick black and red lines, respectively Table 4 The comparison of photodegradation quantum yield, Φ pho , of the indole derivatives 1-5 determined in n-hexane, acetonitrile, and methanol solutions under 365 nm irradiation Estimated error 30%, b this work, c reference [17], d reference [18]  studied compound. This photooxidation reaction does not require the use of any external photosensitizer. The studied indole derivatives act simultaneously as the substrate and the photosensitizer.