Introduction

Pyridine and pyridinium derivatives are important in various biological processes. For example, pyridinium salts are present in the structure in NAD+ and NADP+ cofactors, which are essential for the functioning of all living cells. Since pyridines and pyridinium derivatives are photochemically reactive, [1, 2] in principle, excitation by light allows for the activation or deactivation of biochemical processes involving NAD+ and NADP+ cofactors. Furthermore, biologically important molecules from the family of vitamin B6 and its derivatives are methylpyridines (also known as picolines), which are also prone to photochemical reactions [3].

Interest in pyridine photochemistry was initiated by the discovery of photochemical rearrangements giving Dewar-type products [4]. Later, in a seminal paper by Kaplan et al. [5], photochemical rearrangement of pyridinium salts was reported. Under basic conditions, pyridinium salt 1 undergoes rearrangement and hydration to azabicyclo[3.1.0] derivative 3, via a cationic intermediate 2 (Eq. 1) [5, 6]. It is a relatively simple reaction that delivers complex heterocyclic structures with well-defined stereochemistry. Thus, Mariano et al. extended the scope of the pyridinium salts photorearrangement and reported on acid- [7] or base-induced ring opening of the photochemically formed aza-heterocycles, [8] whereas Simpson et al. [9] and Burger et al. [10] used the photorearrangement as the key step in the synthesis of natural compounds.

(1)

Pyridinium photochemistry has been used in the development of photocages. Falvey et al. developed photocages for alcohols and esters by applying single electron transfer (SET) photochemistry of pyridinium salts in the presence of pyrene and heterocyclic electron donors, [11] or ruthenium(II) sensitizers [12]. Boncella et al. studied N-methylpicolinium triflate photocages for aliphatic amines [13]. Furthermore, photocages were developed wherein the electron donor and pyridinium salt were in the same molecule [14]. Photodecaging of carboxylic acids was accomplished by use of golden nanoparticles and pyridinium salts, [15] whereas carboxylic acids, amino acids, and phosphates can be photodeprotected from pyridinium photocages in the presence of cumarine and BODIPY derivatives [16]. Recently, Falvey et al. reported a simple approach for photodecaging of carboxylic acids by use of visible light, based on charge transfer transition in pyridinium iodide salts [17]. Consequently, a deeper understanding of photochemical transformation of pyridine and pyridinium derivatives is of great relevance and could redirect future studies in the life sciences.

Herein, we report a general investigation of photochemical reactivity in a series of α-, β-, and γ-picolines 4–6, their methyl iodide, and perchlorate salts 7–9. The investigation is conducted to probe for anticipated competing photochemical reactions including excited state proton transfer (ESPT), charge transfer, photohydration, and photorearrangement. Irradiations of 7–9 were conducted under different conditions, and photoproduct formation studied by NMR and UV–vis spectroscopy. The current investigation indicates occurrence of two competing photochemical processes, charge transfer and photorearrangement, neither of which should be neglected in the applications of pyridines photochemistry. Moreover, we observe an interesting effect of the counterion and the meta effect of the methyl group to the photochemical reactivity.

figure a

Experimental

General

1H and 13C NMR spectra were recorded on a Bruker Spectrometer at 300, 400, or 600 MHz. Picoline isomers 4–6 were purchased from the usual commercial sources and were purified by distillation. Pyridinium iodide and perchlorate salts 7–9 were prepared according to the known procedures [8, 18]. Experimental procedures for the preparation of 7–9 and the corresponding NMR and UV–vis spectra are in the supporting information. Solvents used for the photochemical experiments were of HPLC grade. UV–vis spectra were recorded on a PG T80/T80 + spectrometer, and fluorescence on a Cary Eclipse Varian spectrometer. MS spectra were obtained on a HPLC–MS–MS Agilent 1200 series, 6410.

Irradiation experiments in NMR tubes

Irradiation experiments were conducted in a Luzchem photochemical reactor equipped with eight lamps (one lamp 8 W, output at 254 nm).

Picoline isomers 4–6 (1.0–2.4 mg) were dissolved in a mixture of CD3CN (1 mL) and D2O (100 μL) in quartz NMR tubes (c = 1.91–2.34 × 10−2 M). The solutions were irradiated in a Luzchem reactor at 254 nm over 5 min-2 h. After different irradiation times, NMR spectra were recorded.

Pyridinium salts 7–9 (1.5–2.6 mg) were dissolved in a mixture of CD3CN (500 μL) and D2O (50 μL) in quartz NMR tubes (c = 1.2–2.0 × 10−2 M). The solutions were irradiated in a Luzchem reactor at 254 nm, and, after certain irradiation intervals, NMR spectra were recorded.

In the preparative irradiation experiments, pyridinium salt 8a (376 mg, 0.04 M) was dissolved in aqueous KOH solution (40 mL, 0.05 M). A quartz tube was filled with the solution, purged with Ar for 30 min and sealed. After 8 h irradiation, the solution was extracted with diethyl ether, ethyl acetate, and CH2Cl2. The collected extracts were dried over Na2CO3, filtered, and the solvent was removed on a rotary evaporator (without heating) to afford the residue that was analyzed by 1H NMR.

UV–vis experiments

A solution of 7a, 8a, or 9a in CH3CN–H2O (10:1, 3 mL) was divided into three quartz cells for fluorescence spectroscopy. The first solution was purged with Ar, the second with O2, and the third was not purged. The absorbances of the solutions at 234 nm were adjusted to 0.75 (c ≈ 5 × 10−5 M). The solutions were irradiated in a Luzchem reactor at 254 nm in certain time intervals (2 min-2 h), and after each irradiation UV–vis spectra were recorded.

Alternatively, to the solutions of 7a–9a in CH3CN (c ≈ 5 × 10−5 M), different amounts of H2O were added (corresponding to the volumetric ratio of 1, 5, 10, 20, and 50 %). The solutions were irradiated at the same time in a Luzchem reactor at 254 nm. After certain irradiation intervals (2 min-2 h) UV–vis spectra were recorded.

Measurements of reaction quantum efficiencies

Quantum efficiency for the hydration of 7a–9a was measured by a valerophenone actinometer (Φ 254 = 0.65 ± 0.03) [19, 20]. For the actinometer, a solution of valerophenone in CD3CN–D2O (10:1) c = 1.4 × 10−2 M) was used. Solutions of 7a–9a in CD3CN–D2O (10:1) were prepared with the same concentration (c = 1.4 × 10−2 M, A 254 > 2). Quartz NMR tubes were filled with the solutions, the solutions were purged with N2 15 min and the tubes were sealed. The solutions were irradiated in a Luzchem reactor with four lamps at 254 nm 15 min (valerophenone) or 2 h (7a–9a). After the irradiation, NMR spectra were taken. From the conversion of reactant to photoproducts, quantum yields were calculated (for the equations to calculate Φ, see the supporting information). The measurements were performed in triplicate.

Quantum efficiency for the formation of I3 − from 7a–9a was measured by a KI/KIO3 actinometer KI/KIO3 (Φ 254 = 0.74), [19, 21]. For the actinometer, a fresh solution of KIO3 in borate buffer (0.01 M, pH 9.25) was prepared, and then, a solution of KI (0.6 M) in the borate solution of KIO3. Fresh solutions of 7a–9a in CH3CN–H2O (10:1) were prepared in cells for fluorescence spectroscopy in the concentration range corresponding to A 265 ≈ 0.5 (c ≈ 1 × 10−4 M). The volume in cells was 2.5 mL, and the room temperature varied from 17 to 22 °C. The solutions were purged with N2 15 min (or were not purged). Prior to the irradiations, the UV–vis spectra were recorded. The solutions were irradiated in a Luzchem reactor with one lamp at 254 nm 0.5, 1, and 2 min. During the irradiations the cells were kept at a constant distance from the lamp and were covered by a dark paper at three sides to ensure photon flux to the cell only from the front face. Concentration of I3 − was calculated from the absorbance increase at 352 nm, whereas decomposition of 7a–9a was obtained from the decrease of absorbance at 235 or 228 nm (for the equations to calculate Φ, see the supporting information). The measurements were performed in triplicate.

Results and discussion

Absorption spectra of picolines 4–6 have an absorption band at ≈263 nm (253 nm for 6) populating S1 state, which is not fluorescent. Attempts to record fluorescence spectra failed. Pyridinium iodide salts exhibit two absorption bands, at 265 nm (255 nm for 9a), corresponding to the pyridinium excitation, and at ≈235 nm corresponding to a charge transfer transition due to the presence of iodide [17] (see Figs. S1–S4 in the supporting information). Similar to 4–6, 7a–9a are not fluorescent.

All irradiation experiments were performed by use of lamps with the output at 254 nm. Electronically excited states are often characterized by different acid–base properties compared to the corresponding ground states [22]. Thus, pyridine derivatives become more basic upon excitation to S1 (pK a S0 = 5.5, pK a S1 = 12.7) [23]. Moreover, phototautomerization of 4 was investigated by computational methods indicating that excitation to S1 or T1 lowers the activation barrier for the formation of 4-T (Eq. 2) [24].

(2)

Upon electronic excitation, the pyridine nitrogen in picolines is anticipated to become more basic, whereas the methyl group may behave as a carbon photoacid leading to the formation of 4-T, as shown in Scheme 1. Subsequent thermal tautomerization of 4-T to 4 should lead to the regiospecific deuterium incorporation in the methyl group in compound 4-D. Similarly, it is anticipated that pyridinium salts 7–9 behave as carbon photoacids. For example, deprotonation of 7 is anticipated to give 7-T that should upon thermal protonation in D2O give 7-D with the regiospecificaly deuterated methyl group (Scheme 2). It is plausible that meta and para derivatives undergo similar photoreactions via the corresponding intermediates 5-T–9-T (Fig. 1).

Scheme 1
scheme 1

Plausible ESPT reactivity of 4

Full size image
Scheme 2
scheme 2

Plausible ESPT reactivity of 7

Full size image
Fig. 1
figure 1

Anticipated intermediates in the ESPT photochemistry of 4–9

Full size image

Anticipated ESPT reactivity and phototautomerization of 4–9 was probed by irradiations in CD3CN–D2O. After the irradiation, the position and the extent of deuteration was examined by NMR and MS, respectively. It should be noted that protonation (deuteration) of pyridine nitrogen cannot be probed in such a way. Interestingly, after the irradiation of 4–7, no deuterium incorporation in the methyl group was observed, but it took place in the aromatic ring. Results are compiled in Table 1. The plausible mechanism for deuteration of the aromatic ring may be electrophilic aromatic substitution in S1, or a radical-cation mechanism similar to the one proposed for the deuteration of pyrroles and similar aza-heterocycles [25]. In such a mechanism, picolines are first oxidized to the corresponding radical-cations 10 that undergo D-abstraction from CD3CN to give cations 11 and eventually yield D-incorporated molecules (Scheme 3). Very low efficiency of D-incorporation is in line with such a mechanism. Namely, photoionization of pyridine should take place with low efficiency due to its relatively high oxidation potential (in CH3CN, E ox = 1.82 V vs. Ag/AgCl) [26].

Table 1 Deuterium content in 4–6 after irradiation in CD3CN–D2Oa
Full size table
Scheme 3
scheme 3

Plausible mechanism for D-exchange in 4 upon irradiation in CD3CN–D2O

Full size image

In addition to deuteration of 4–6, no other photochemical processes were observed. In particular, phototransposition products were not formed, contrary to the previous report by Pavlik et al. [27] on phototransposition of picolines in gas phase. Obviously, photochemical reactivity in gas phase and protic solution differ significantly, presumably due to the protonation of pyridine nitrogen in S1 [23].

Contrary to picolines, irradiation of the pyridinium derivatives 7–9 did not give rise to deuterated compounds, but gave H2O-adducts and rearranged products. In addition, during the irradiations of 7a–9a, the color of the solution changed from transparent to yellow, indicating formation of I3 − [28]. Therefore, we examined two processes separately, photochemical formation of I3 − by UV–vis spectroscopy and formation of H2O-adducts by irradiation in quartz NMR tubes and recording NMR spectra.

Upon irradiation of 7a–9a in NMR tubes, a new set of signals between 3 and 6 ppm in 1H NMR spectra was observed. Conversion to these primary photoproducts after 2 h irradiation was ≈10 % for 7a and 9a, whereas for the meta isomer 8a it was ≈50 %. From 7a or 9a, in addition to primary products formed in yields lower than 10 %, competing phototransposition took place, giving mixtures of all three isomers 7a–9a (Scheme 4). Our observation is in agreement with a preceding theoretical report that photolysis of meta-alkylated pyridinium salts takes place without nitrogen migration [29]. Attempts to isolate the primary photoproducts after preparative irradiations in CH3CN–H2O failed. In the isolation process, the primary products rearranged in a transposition reaction giving mixtures of 7a–9a. Similarly, upon heating of a NMR tube containing only starting molecule 8a and primary products 14 and 15, a mixture of 7a–9a was obtained (Scheme 5).

Scheme 4
scheme 4

Competing photohydration and phototransposition of 7a and 9a

Full size image
Scheme 5
scheme 5

Photohydration of 8a

Full size image

In principle, primary photoproducts could be associated to the hydration products 12–16, or iodide adducts such as 17 (Scheme 4). Similarity of the signals in 1H NMR to H2O-adducts described in precedent literature [5] encouraged us to assign the primary products to 12–16. To verify this assignment, the irradiation of 8a has been conducted under basic conditions, the same as in the report by Kaplan et al. [5] where I-adducts are not expected due to the presence of a large excess of the better nucleophile, OH−. Indeed, the same products were obtained in the irradiation of neutral and basic solution, indicating that the primary products are correctly assigned to H2O-adducts. Additional evidence that the primary products are H2O-adducts came from the lack of their formation upon irradiation in neat CH3CN. Prolonged irradiation resulted only in the inefficient phototranspositon. Note that 12–16 have in principle different diastereomeric forms. Since we did not manage to separate the H2O-adducts, we were not able to assign their stereochemistry. However, it is known that nucleophile attacks the azabicyclic [3.1.0] cation stereospecifically, anti- to the aziridine ring, as shown, for example, in Eq. (1) [5, 7–10].

Efficiency of the photohydration was determined by use of a valerophenone actinometer [19, 20]. Results are compiled in Table 2. It is interesting to note that photohydration of meta- isomer 8a takes place about 2–3 times more efficiently than for the ortho or para isomers. This result represents one more example of a meta-effect in photochemistry [30, 31]. The observation is probably due to the positive inductive effect of the methyl substituent that stabilizes the cationic intermediate, more pronounced in 8-cat then in 7-cat or 9-cat. Our observation of meta-effect is also in agreement with precedent theoretical investigation [29].

Table 2 Quantum yield of photohydration (Φ H2O) of 7a–9a a
Full size table

Photochemical reactivity of iodide salts 7a–9a in charge transfer reactions was investigated by recording UV–vis spectra during the irradiation, where formation of I3 − species was clearly observed. For example, Fig. 2 shows the changes in spectra upon irradiation of 8a, where a decrease of absorbance at 235 and 265 nm indicate a disappearance of I− and pyridinium chromophores, whereas an increase of absorbance at 290 and 360 nm indicates formation of I3 −. In particular, the absorption band at 235 nm corresponds to a charge transfer transition [32]. It is clearly demonstrated by UV–vis spectra at different H2O content where the absorption band shifts hypsochromically with the increase of solvent polarity (See Fig. S4 in the supporting information). Thus, upon excitation, an electron is transferred from I− to the pyridinium forming a radical pair. Escape of the iodine radicals from the cage leads to the formation of I2 and I3 − (Scheme 6). However, by 1H NMR we did not detect the presence of any bispyridine or reduction product, indicating that the recombination of pyridyl radicals does not take place or leads to the formation of high molecular weight products.

Fig. 2
figure 2

Changes in the UV–vis spectra of 8a upon irradiation in CH3CN–H2O (9:1). The irradiation was performed at 254 nm and the spectra were recorded in the irradiation intervals of 0.5–5 min

Full size image
Scheme 6
scheme 6

Charge transfer transition in 8a leading to the formation of I3 −

Full size image

Whereas charge transfer transition and formation of I3 − are important in the application of pyridinium salts as photocages, [17] it is an energy wasting process in the synthetic applications. Therefore, we determined quantum efficiency for the chromophore disappearance and formation of I3 − from 7a to 9a by use of a KI/KIO3 actinometer (Table 3) [20, 21]. Moreover, we investigated the influence of solvent polarity and presence of O2 to the reaction efficiency. Irradiations were performed in N2-purged, air-saturated and O2-purged solutions where we examined the efficiency of the iodine formation, as well as the decay of the photochemically formed I3 −. Results in Table 3 and Fig. S7 in the supporting information indicate that I3 − is formed with similar efficiency when the solution contained O2, suggesting that no quenching or negligible quenching of the excited state took place.

Table 3 Quantum efficiency for the compound degradation (Φ d) and I3 − formation (Φ I) for 7a–9a in N2-purged and air-saturated solutionsa
Full size table

Monitoring of the absorbance of 8a after the irradiation at 254 nm indicates that some intermediates or primary products formed in the photochemical reaction are not stable. Their decay was monitored over 1 h wherein we observed slower kinetics in the presence of O2 than in the Ar-purged solution (see Fig. S8 in the supporting information). However, we were not able to isolate intermediates in the charge-transfer photochemistry or get more confident spectroscopic proofs for their structure.

Efficiency of the I3 − formation was also tested at different H2O contents and different 8a substrate concentrations. Whereas in the tested substrate concentration range no significant influence to the reaction efficiency was observed, H2O concentration affected the charge transfer transition and the efficiency for the I3 − formation. At higher H2O concentration the efficiency of the I3 − formation decreases (see Fig. S9 in the supporting information), presumably due to competing photohydration. Thus, at higher H2O concentration, disappearance of the substrate was more efficient.

Irradiation of pyridine iodide salts at 254 nm gives rise to the charge transfer transition and formation of iodine. On the contrary, perchlorate has a high oxidation potential rendering such a transfer energetically impossible. Therefore, we performed photochemical experiments with pyridinium iodide and perchlorate isomers to probe for the counter-ion effect. Interestingly, NMR spectra after the irradiation indicate formation of different products than from the iodide salts. Again, attempts to isolate the products failed. However, the observed difference can be associated to acid-catalyzed aziridine opening by photorelease of a strong acid, HClO4, as shown in Scheme 7. 1H NMR spectrum after irradiation of 8b showed th presence of two compounds in approximate ratio 1:1 with characteristic signals clearly corresponding to 18 and 19, in agreement with precedent literature [7].

Scheme 7
scheme 7

Photohydration of 8b followed by aziridine ring opening

Full size image

Conclusions

Photochemical reactivity of a series of picolines and their N-methylated salts has been investigated. Picolines are not reactive in phototransposition, probably due to protonation of pyridine nitrogen in the excited state. Deuterium exchange was observed upon irradiation in CD3CN–D2O, presumably due to photoionization and radical formation. On the other hand, N-methylated picoline salts are not protonated upon excitation. They undergo photohydration and phototransposition. A difference in reactivity was observed between perchlorate and iodide salts. Upon irradiation of iodides, azabicyclic[3.1.0] hydration products were obtained, whereas due to photoelimination of perchloric acid thermal aziridine ring opening takes place. Excitation of iodide pyridinium salts gives rise to change transfer transition forming iodide radicals that eventually give I3 −. Revealing the photochemical reactivity of these simple chromophores is of significant importance in the application of these molecules as photocages or in the synthesis of complex polycycles.