Primary photochemical processes for Pt( IV ) diazido complexes prospective in photodynamic therapy of tumors †

Diazide diamino complexes of Pt( IV ) are considered as prospective prodrugs in oxygen-free photodynamic therapy (PDT). Primary photophysical and photochemical processes for cis , trans , cis -[Pt(N 3 ) 2 (OH) 2 (NH 3 ) 2 ] and trans , trans , trans -[Pt(N 3 ) 2 (OH) 2 (NH 3 ) 2 ] complexes were studied by means of stationary photolysis, nanosecond laser ﬂ ash photolysis and ultrafast kinetic spectroscopy. The process of photolysis is multistage. The ﬁ rst stage is the photosubstitution of an azide ligand to a water molecule. This process was shown to be a chain reaction involving redox stages. Pt( IV ) and Pt( III ) intermediates responsible for the chain propagation were recorded using ultrafast kinetic spectroscopy and nanosecond laser ﬂ ash photolysis. The mechanism of photosubstitution is proposed.

Both 1 10 and 2 11 are the subjects of reductive cross-linking with nucleotides under the action of light. The reaction can be induced both by UV and visible irradiation. The light-induced cytotoxicity of both complexes is comparable with the cytotoxicity of cisplatin in dark. 11 The trans complex 2 is considered as more potent anticancer agent than the cis one, 16 mainly because of the red shift of the absorption bands. Note that for complex 2 there is no limitation of trans complex anticancer activity which is characteristic for transplatin. The acting species linking to DNA in the cases of both 1 13 and 2 18 photolysis is the diamino diaqua complex of Pt(II), as in the case of cisplatin. In spite of that, the mechanism of cancer cell damage by photoirradiated complex 1 and 2 is different from the case of cisplatin. 12,24 This is probably due to the effect of active transients, such as N 3 • radicals and (or) Pt(IV) nitrenes. 13,20 Photolysis of 1 and 2 is the multistage process. In this work we are focused on the first stage. It was reported that the mechanism of photolysis is solvent-dependent. Two types of aqueous solutions were examined, namely, slightly acidic aqueous solutions ( pH ∼5) and phosphate buffered saline (PBS), pH 7.4. Conclusions concerning the primary photochemical processes were made based on the changes of NMR spectra in the course of irradiation. 13,14,18 The results are schematically presented in Scheme 2. In the case of 1 in PBS Phillips et al. 14 found that the primary photochemical reaction caused by UVA irradiation is the exchange of the N 3 − anion to a water molecule or a hydroxide anion ( product 1-1 in Scheme 2A). In this case the irradiation time was probably small enough to examine the primary stage of photolysis. UVA irradiation of 1 in slightly acidic solution was found to result in photoreduction of Pt(IV) to Pt(II) and N 2 release. 13 Absence of N 3 − in reaction products and results of trapping experiments allowed authors 13 to conclude that photolysis occurs via the formation of nitrenes (which is represented by product 1-2 in Scheme 2B). However, it should be noted that irradiation time in experiments 13 was rather long (2 h). As a result of prolonged irradiation, azide anions could decay in secondary photochemical processes. In fact, there is no evidence that nitrene 1-2 is the primary product of photolysis. Photolysis of 2 was reported to demonstrate the similar solvent effect as in the case of 1. 18 Primary processes caused by UVA irradiation of 2 in PBS (Scheme 2C) were identified as the exchange of azide anions to water molecules (or hydroxide anions). The results of photolysis in acidic solutions were interpreted in ref. 18 as the parallel reactions of photoaquation and N 2 release with the nitrene formation (Scheme 2D). Again, the irradiation time in experiments performed in ref. 18 was rather long, which did not allow one to be sure that the secondary photochemical processes were not involved.
In this paper we describe the case study of the first stage in photochemistry of diazide diamino complexes 1 and 2 by means of stationary photolysis, nanosecond laser flash photolysis and ultrafast kinetic spectroscopy. We present evidence that photolysis of both complexes is started with the chain photoaquation.
UV absorption spectra were recorded using Agilent 8453 spectrophotometer (Agilent Technologies).
Stationary photolysis was performed using the irradiation of a high-pressure mercury lamp with a set of glass filters for separating necessary wavelengths. In several experiments, excimer XeBr lamp (excilamp) was used as a quasicontinuous source of UV-irradiation at 282 nm (half width of light pulse, 5 nm; pulse duration, 1 µs; frequency, 200 kHz; incident light flux, 2.7 × 10 16 photons per cm 2 s). 25 To measure the radiation intensity for the quantum yield calculations we used the IrCl 6 2− complex in methanol solutions as a chemical actinometer (quantum yield of photoreduction in air-saturated solutions is 0.1 when initial concentration is less than 10 −3 M (ref. 26)). Nanosecond laser flash photolysis experiments were performed using excitation by radiation of 3 rd (355 nm) and 4 th (266 nm) harmonics of a YAG laser (Lotis TII, Belarus, 5 ns pulse duration, 10 mJ per pulse energy). Light power meter SOLO 2 (Gentec, Canada) was used to measure the laser pulse energy. The setup was described in detail in ref. 27.
We used pump-probe spectroscopy to study transient absorption of the samples in femto-and picosecond time domains. The experimental setup was described in detail elsewhere. 28 The samples were excited by ∼60 fs pulses at Scheme 2 Primary stages of 1 and 2 photolysis according to literature. A -Complex 1 in PBS; 14 Bcomplex 1 in acidic solution; 13 Ccomplex 2 in PBS; 18 Dcomplex 2 in acidic solution. 18 ∼400 nm (second harmonic of a Ti:sapphire generator-amplifier system (Tsunami -Spitfire Pro, Spectra Physics)) and by ∼100 fs pulses at ∼320 nm (fourth harmonic of the signal wave of the TOPAS parametric amplifier). The excitation energy in both cases was about 1 μJ per pulse, the pulse repetition rate was 1 kHz. 200 pulses were used to record a single timeresolved spectrum. In order to decrease the magnitude of the so-called "coherent artifact" signal 29 we set the angle between electric field vectors of the pump and probe beams at 54.7°( "magic angle"). To do that we rotated the polarization of the pump beam using Berek's variable wave plate. Each kinetic curve contained 110 points (60 points with a 100 fs step, 20 points with a 500 fs step, and 30 points with a 3 ps step). The investigated solutions (total volume of 20 ml) were pumped through a 1 mm thick quartz cell at room temperature to provide uniform irradiation and to avoid possible degradation due to photochemical reactions.
ExciPro program (CDP System Corporation) was used for corrections of the group velocity dispersion. The corrected experimental data were further globally fitted (typically using a three-exponential model).

Results and discussion
3.1. UV spectra and photochemistry of complexes 1 and 2 UV spectra of complexes 1 and 2 are shown in Fig. 1a and 2a (curve 1) correspondingly. Absorption in the near UV region observed in the spectra of azide complexes of Pt(IV) was assigned as the N 3 → Pt CT band. 30,31 . Comparing the spectra of 1 and 2, one can see that the N 3 → Pt band of trans-complex is red-shifted (λ max = 257 and 286 nm) and more intense (ε max = 13 200 and 19 500 M −1 cm −1 correspondingly).
The spectral parameters for 1 and 2 are coincident with the values reported in literature. 2,11 Note that the increase in temperature results in the blue shift of the N 3 − → Pt IV LMCT band maximum, which is evident from comparison of spectra of 2 obtained in this work at 295 K and in work of ref. 19 at 310 K (authors 11 did not specify the temperature; according to the origin of the spectrum it was about 295 K).
Although the major absorption of 1 and 2 is in the UV region, both complexes exhibit tails which extend out to the visible region. E.g., molar absorption coefficients at 458 nm are 50 and 280 M −1 cm −1 for 1 2 and 2 (this work) correspondingly. Tails are evidently caused by the LF transitions. LF bands are photoactive; visible light-induced reductive photochemistry was reported both for 1 10 and 2. 16 As a result, the use of these compounds in visible light-induced PDT is possible. It is essential that to excite the intense UV bands is experimentally more convenient than to excite weak LF bands. That is why in the mechanistic studies performed in this work we used the irradiation into the N 3 − → Pt IV LMCT band.
Spectral changes caused by the UV irradiation of 1 are shown in Fig. 1. On the first stage of photolysis ( Fig. 1a) three isosbestic points at 209, 237 and 287 nm are conserved. According to ref. 14 the primary photoreaction is the exchange  of an azide anion to a water molecule or OH − . It was found that no changes in pH occur at this stage. Therefore, N 3 − is exchanged to a water molecule (forming product 1-1 in eqn (1)) but not to a hydroxide anion. ð1Þ Changes in the UV spectrum caused by prolonged irradiation of 1 are shown in Fig. 1b. No conservation of isosbestic points is observed. According to literature, 13,14 several Pt(IV) and Pt(II) products are formed. Further irradiation results in precipitation; the precipitates may arise from multinuclear oxygen-bridged platinum complexes. 11 First stage of photolysis of complex 2 ( Fig. 2a) is similar to the case of 1. Three isosbestic points at 222, 262 and 320 nm are conserved. According to ref. 18 the primary photoreaction is the exchange of an azide anion to a water molecule or OH − . No sufficient changes in pH were detected. Therefore, the first stage of photolysis is photoexchange with the formation of product 1-2 described by eqn (2).

ð2Þ
Deep photolysis (Fig. 2b) is characterized by the absence of isosbestic points. Complete disappearance of absorption in the near UV region (curve 8 in Fig. 2a) corresponds to complete photoreduction of Pt(IV) to Pt(II).
To estimate the quantum yields of the first stage of photolysis one needs to know molar absorption coefficients of products 1-1 and 2-1. In this work we can present brief estimations of these values. From the origin of the spectral changes in the course of photolysis ( Fig. 1 and 2) it could be estimated that the molar absorption coefficients of the primary products 1-1 and 2-1 at the wavelengths of the near UV bands maxima of the initial complexes (measured in M −1 cm −1 ) fall within the ranges 4500 < ε 1-1 257 nm < 9500 and 6000 < ε 2-1 286 nm < 11 000 correspondingly. Taking values in the middle of these ranges we determined that the quantum yields of reactions (1) and (2)  The reason to specify the initial concentration and photon flux density when presenting quantum yields comes from the experiments with varied concentrations of complexes. The concentration dependencies of quantum yields for complexes 1 and 2 are presented in Fig. 3. The linear dependence of the quantum yield on the concentration of the initial reagent indicates the chain character of the process. In the case of quadratic chain termination the dependence of the quantum yield for excitation at the wavelength λ (φ λ ) vs. concentration of the initial reagents (c) could be described by the equation: 32 where φ 0 λ is the quantum yield at c → 0, I 0 is the photon flux density (Einstein L −1 s −1 ), k p and k d are the apparent rate constants for chain propagation and termination. The slopes and intercepts of the straight lines in Fig. 3 are 0.009 ± 0.002 L mol −1 and 0.10 ± 0.01 for complex 1 and 0.013 ± 0.003 L mol −1 and 0.09 ± 0.02 for complex 2. In fact, intercept of the straight line (eqn (3)) is the quantum yield of the reactive intermediate responsible for the chain character of the reaction.
Chain character of photoaquation reactions (1) and (2) testifies that photochemical reactions involve redox stages with the participation of Pt(III) intermediates. A kinetic analogue could be found in the photochemistry of the PtCl 6 2− complex in water [33][34][35] and acetonitrile. 36 For mechanistic study of chain reactions nanosecond laser flash photolysis was applied. In several microseconds the initial intermediate absorption band is transformed into another band with the maximum at 400 nm (curve 2 in Fig. 4b) attributed to another intermediate further marked as Int 1-2. In spite of coincidence of the bands maxima, Int 1-1 and Int 1-2 are different species. This is evident from the conservation of the isosbestic point in the region of 340 nm (Fig. 4b). Kinetic curves in the region of 320 and 535 nm (Fig. 4a) demonstrate opposite kinetic behavior in the time range of several microseconds. The corresponding kinetic curves were fitted by two-exponential functions eqn (4) (red lines in Fig. 4b). The characteristic time of Int 1-1 → Int 1-2 transition was determined as τ 1 = (3.9 ± 0.7) μs. The decay of Int 1-2 is evident from the kinetic curves at the wavelengths close to the band maximum (see curve corresponding to 400 nm in Fig. 4a). The biexponential fit of kinetic curves in this region resulted in characteristic time of ca. 4 and 32 μs. The first time again corresponds to Int 1-1 → Int 1-2 transition, and the second time τ 2 = (32 ± 7) μs corresponds to decay of Int 1-2.
Photochemical properties of complex 2 and 1 are similar. The two intermediates marked as Int 2-1 and Int 2-2 were recorded in the course of laser flash photolysis of 2 (Fig. 5). Again, the amplitude of the intermediate absorption was linear vs. laser pulse energy, and no effect of dissolved oxygen on the intermediate absorption behavior was observed.
Opposite to the case of complex 1, no conservation of isosbestic points was observed in the intermediate absorption spectrum (Fig. 5b). In spite of that, conclusion on the existence of the two successive Pt(III) intermediates with the similar absorption spectra was made from the analysis of the kinetic curves   (Fig. 5a). Initial intermediate Int 2-1 (curve 1 in Fig. 5b) is transformed to the second intermediate Int 2-2 with the characteristic time τ 1 = (2.3 ± 0.3) μs. The decay Int 2-2 is evident from the kinetic curves at the wavelengths close to the band maximum (see curve corresponding to 400 nm in Fig. 5a). The decay time of Int 2-2 (extracted from the biexponential fit of the kinetic curves (4)) is τ 2 = (13 ± 2) μs.
Therefore, we can conclude that both for 1 and 2 two successive Pt(III) intermediates are observed. The first intermediates (Int 1-1 and Int 2-1) are formed with the characteristic time less that the resolution of the experimental setup (50 ns). The second intermediates (Int 1-2 and Int 2-2) are formed with the characteristic time of several microseconds and decay in time interval of several tens of microseconds. The kinetic law of the second intermediates decay is rather complicated; this is out of the scope of the current study. We believe that the second intermediates are the chain carriers in the mechanisms of chain photoaquation of complexes 1 and 2.
Photochemistry of Pt(IV) azide complexes was studied by Vogler et al. 30,37 Based on these works, one could expect that a Pt(III) intermediate is formed as a result of a light-induced electron transfer from an azide anion to the Pt(IV) cation. In this case an azide radical should release. Sadler and co-workers 38 have detected by spin-labeling and trapping experiments the presence of azidyl radicals formed upon photolysis of a related diazido-Pt(IV) complex trans,trans,trans-[Pt(N 3 ) 2 (OH) 2 ( py) 2 ]. The • N 3 radical exhibits moderate absorption only in the UV range with a sharp maximum at 274 nm, maximal molar absorption coefficient 2025 M −1 cm −1 , and the width of spectral band is about 20 nm at half-maximum height. 39,40 Therefore, it is a big problem to record its spectrum, especially in a case like ours, when absorption of the radical is masked by absorption of the initial compound.
In this work the UV spectrum of • N 3 was not recorded because of the experimental difficulties. In spite of that, based on works of Vogler et al. 30,37 and Sadler et al. 38 we may consider its formation as highly probable. Because of strong oxidative properties of • N 3 40 it should be included into the scheme of possible reactions caused by laser irradiation of complexes 1 and 2. Laser flash photolysis experiments were performed both in neutral ( Fig. 4 and 5) and slightly acidic ( pH 4.5-5) solutions ( Fig. S1 and S2 of ESI †). No difference was observed in the framework of the experimental accuracy. It supports our suggestion that the formation of molecular nitrogen (Scheme 2) is presumably the result of prolonged irradiation but not of the primary photochemical processes.

Possible mechanism of chain photoaquation
Based on the experiments described in items 3.1 and 3.2, we believe that the photoaquation of complexes 1 and 2 is the chain process occurring via the redox reactions. Tentative reaction mechanism is presented by eqn (5)-(11) (because the reactions proposed for complexes 1 and 2 are similar, we do not specify whether the azide anions are in cis-or transpositions): Chain initiation Chain propagation Chain termination According to the proposed mechanism, the primary photochemical process is the inner-sphere electron transfer in the excited complex followed by an azide radical release (5). Two successive Pt(III) intermediates participate in chain initiation (reactions (5) and (6)). The [Pt III (NH 3 ) 2 (OH) 2 ] + complex is proposed to be the chain carrier (reaction (7)). Chain termination seems to be complicated because of oxidative properties of the azide radical. In addition to the direct chain termination reactions ( (8) and (9)) one can propose the indirect decay of chain carriers started by oxidation of the initial complex by the azide radical (10) and followed by disproportionation of Pt(V) and Pt(III) intermediates (11). The possibility of Pt(V) complexes formation as a result of Pt(IV) oxidation by free radicals was demonstrated in ref. 33 and 41. The chain mechanism of photoaquation (5)-(11) is consistent with the concentration dependence of quantum yields obtained in the stationary experiments and with the observation of two successive Pt(III) intermediates (Int 1-1, Int 1-2 for the case of 1 and Int 2-1, Int 2-2 for the case of 2) in the laser flash photolysis experiments. Identification of successive intermediates as [Pt III (NH 3 ) 2 (N 3 )(OH) 2 ] and [Pt III (NH 3 )(OH) 2 ] + complexes requires explanation of similarity in their UV spectra (curves 1 and 2 in Fig. 4b and 5b). It could be possible if the absorption at 400 nm is defined by LMCT transition from the NH 3 ligand to Pt(III) cation. If the N 3 − ligand in the [Pt III (NH 3 ) 2 (N 3 )(OH) 2 ] complex is weakly bound to the central cation, its elimination would not much affect the absorption spectrum. As an example, the square-pyramid structure with N 3 − as the axial ligand could be proposed.
Details of the proposed mechanism (including direct observation of azide radicals) are the subjects of further study.

Ultrafast kinetic spectroscopy of 1 and 2
Experiments on the ultrafast kinetic spectroscopy for complex 2 were performed with excitation at 320 and 400 nm. The 320 nm excitation falls into the long-wavelength tail of the N 3 − → Pt IV LMCT band (Fig. 2), The 400 nm excitation corresponds to weak (ε = 440 M −1 cm −1 ) absorption, which is probably caused by the d-d transitions of Pt IV . Complex 1 was excited at 320 nm to the N 3 − → Pt IV LMCT band. In this case the 400 nm irradiation was not used because the absorption of 1 at this wavelength is too weak. The time-dependent intermediate absorption spectra were recorded in the range 440-780 nm. Time profiles of the intermediate absorption are shown in Fig. S3-S5 of ESI. † The kinetic curves for different complexes and different excitation wavelengths are shown in Fig. 6a, 7a and 8a. As it was mentioned in the Experimental section, kinetic curves were affected with the coherent solvent artifact. 29 For the case of 320 nm excitation the coherent artifact was efficiently decreased (see Experimental); the kinetic curves were corrected for its residuals by subtracting the second derivative of Gaussian function. For the case of the 400 nm excitation the coherent artifact, in spite of the use of magic angle geometry, remained rather large (see Fig. S5 of ESI †). Therefore, for this case the points corresponding to the time delays falling into the range (−500 fs) < τ > (550 fs) were omitted (Fig. 8a).
The kinetic curves obtained in the experiments on the ultrafast pump-probe spectroscopy with the 320 nm excitation were globally fitted using a two-exponential function with the residual absorption (12).
Fitting curves are shown in Fig. 6a, 7a and 8a as red lines. The extracted characteristic lifetimes are collected in Table 1. The increase in number of exponential functions used for fit from two to three did not result in better fitting quality (see Fig. S6-S8 of ESI †).
When the kinetic curves are fitted using function (12), the sequential decay of the transient absorption A → B → C is proposed. Here C can include both the ground state of the initial complexes and photolysis products. The species associated difference spectra (SADS) of the individual components were calculated by means of the formulae (13)- (15). 42 If the characteristic lifetimes are sufficiently different, the SADS could be presented as sums of corresponding amplitudes A i (λ).
Immediately after excitation of complex 1 at 320 nm the transient absorption spectrum appears as a wide band with the flat maximum in the region 450-600 nm (SADS S A in Fig. 6b). The similar absorption occurs in the case of complex 2 (SADS S A in Fig. 7b). The initial absorption bands are transformed into other bands with the maxima in the region of 420 nm (SADS S B in Fig. 6b and 7b) with the characteristic lifetimes 0.5-1.4 ps. The second intermediates decay with the characteristic lifetimes 7-17 ps. The disappearance is not complete, but the spectra of the residual absorption (SADS S C in Fig. 6b and 7b) do not differ much of the SADS S B .
In the case of excitation of complex 2 at 400 nm the origin of intermediate absorption is similar to the case of the 320 nm excitation. Nevertheless, the large value of the coherent artifact (see Fig. S8 of ESI †) did not allow us to determine the SADS of the initial intermediate S A correctly. In fact, the SADS S A in Fig. 8b is the superposition of the spectra of first and second intermediates. Correspondingly, the value of its lifetime (0.7 ps, see Table 1) should be considered as estimation.
The final SADS S C obtained in experiments on ultrafast kinetic spectroscopy could be compared with the spectra of initial intermediates Int 1-1 and Int 2-1 recorded in the experiments on nanosecond laser flash photolysis. This comparison is presented in Fig. 9. SADS S C refer to the 320 nm excitation, while spectra of Int 1-1 and Int 2-1 refer to the 266 nm excitation. For both complexes excitation wavelengths 266 and   320 nm fall within the same absorption bands ( Fig. 1 and 2), therefore one can expect no difference in corresponding photochemical processes. The spectra of SADS S C are clearly different of the spectra of Int 1-1 and Int 2-1 (Fig. 9). It means that the species C in the sequence A → B → C used for analysis of ultrafast kinetic data do not coincide with the initial intermediates Int 1-1 and Int 2-1 observed in the nanosecond time domain. Therefore, additional reactions C → Int 1-1 and C → Int 2-1 are necessary to fit the results of ultrafast and nanosecond experiments. The characteristic times of these reactions fall within the time interval 0.1-10 ns. These lifetimes were not extracted from analysis of the data on ultrafast kinetic spectroscopy because signal-to-noise ratio was not enough to make a choice between biexponential and three-exponential models (see Fig. S6-S8 of ESI †). In spite of that, we add this reaction to the mechanism of primary processes (Table 1). Our interpretation of the ultrafast spectroscopy data is as follows (Table 1 and Fig. 10). The initial SADS S A corresponds to the spectrum of the Franc-Condon (FC) state. The first observed process with the characteristic time of 0.5-1.5 ps corresponds to the transition of FC to the lowest electronic excited state (LE). This process is probably accompanied by vibrational cooling and solvent relaxation having characteristic lifetimes comparable with the characteristic time of electronic transition. The different processes (e.g., intersystem crossing and vibrational relaxation) could have similar characteristic times and manifest itself as a single, convoluted process. 43 The multiplicities of the ground state (GS) and the LE state are most likely different (the electronic configuration of Pt 4+ is 5d 6 . If the geometry of the GS is not very different of octahedral, it should be singlet, while the LE should be triplet). In this case, the lifetime of LE could be 10-20 ps, which is observed in the experiment.
As it was discussed earlier, the LE state could not transit directly to the reaction product. We assume that the result of its transformation is the radical complex (further -RC) [Pt III (NH 3 ) 2 (N 3 )(OH) 2 ⋯N 3 • ], which is in fact the ion-radical pair of Pt(III) complex and N 3 • radical situated in the solvent cage.
The weakly-bound RC were experimentally observed as intermediates in photochemistry of Cu(II), 44 Pt(IV), 45 Ir(IV), 46 Fe(III). 47 Its lifetime in common solvents at room temperature can reach 1 ms. 47 Escape of an azide radical from RC into the solution bulk results in formation of intermediates Int 1-1 and Int 2-1 (Table 1 and Fig. 9). The scheme of processes corresponding to the described tentative model is represented by eqn (16)

Conclusions
The results obtained in this work have demonstrated that the photolysis of mixed-ligand diazide complexes of Pt(IV), even at the first stage, is rather complicated. The first stage which is a release of an azide anion from the first coordination sphere of Pt(IV), is shown to be the chain process. Several intermediates were recorded both in picosecond an in microsecond time domains. Tentative scheme of processes is proposed.
Analyzing the results of time-resolved experiments, we have concluded that the pathway of photoaquation includes at least five different Pt(IV) and Pt(III) intermediates, which are listed in Table 2.  Table 2. For further study of mechanism, it seems attractive to find experimental evidence of the N 3 • radical formation directly in the laser flash photolysis experiments. The approach to solve this problem is to find a proper stable molecule (trap) reacting with N 3 • with the formation of easily detected product. The rate constant of the reaction between an azidyl radical and a trap should be known or measured in a model system. Performing this experiment will allow one to quantify the chain photoaquation mechanism (5)- (11). The second possible direction of the mechanistic study is to perform experiments in frozen (to 77 K) aqueous salt matrices with the simultaneous registration of intermediates by UV and ESR spectroscopy. This approach can be reasonable if the quantum yield does not drop to zero with decrease in temperature. Probably, in this case the ESR detection of the stabilized radical complex (RC) could be possible.