: determined in aqueous solutions.

Figure 2b shows the set of EPR spectra monitored during the exposure of TiO2/DMPO/air in mixed solvent water/DMSO (5:1 v:v). The EPR spectra obtained are more complex compared to those found in water suspensions (Figure 2a), and represent a superposition of the dominant six-line signal attributed to • DMPO-CH3 and the low-intensity signal of • DMPO-OH with slightly modified hyperfine coupling constants (hfcc) caused by the DMSO presence in the system [56]. The experimental and simulated EPR spectra found upon 400 s exposure are depicted in Figure 4a and the corresponding spin Hamiltonian parameters elucidated from the simulated spectra are summarized in Table 2.

Further experiments using 3,5-dibromo-4-nitrosobenzene sulfonate (DBNBS) spin trapping agent in aerated aqueous TiO2 suspensions containing DMSO or DMSO-*d*6 (water/DMSO, 5:1 v:v) unambiguously confirmed the generation of methyl radicals via the reaction of hydroxyl radicals with the solvent, as the EPR spectra monitored upon UVA irradiation are well-matched to • DBNBS-CH3 or • DBNBS-CD3 spin-adducts (Figure 4b,c), respectively, with the spin Hamiltonian parameters well correlated with literature data (Table 2).

**Figure 4.** Experimental (**black**) and simulated (**red**) EPR spectra obtained upon irradiation (Ȝmax = 365 nm; irradiance 15 mW·cm<sup>í</sup><sup>2</sup> ; exposure 400 s) of the aerated TiO2 P25 suspensions in mixed solvent water/DMSO (5:1 v:v) in the presence of various spin trapping agents (TiO2 concentration 0.167 mg·mL<sup>í</sup><sup>1</sup> ): (**a**) DMPO (*SW* = 8 mT; *c*0,DMPO = 0.035 M); (**b**) DBNBS (*SW* = 10 mT; *c*0,DBNBS = 0.008 M); (**c**) DBNBS using DMSO-*d*6. The spin Hamiltonian parameters of corresponding spin-adducts are listed in Table 2. (a) • DMPO-CH3 (relative concentration in %, 94), • DMPO-OH (6); (b) • DBNBS–CH3 (100); (c) • DBNBS-CD3 (100).

The application of 5-(ethoxycarbonyl)-5-methyl-1-pyrroline *N*-oxide (EMPO), 5- (diisopropoxyphosphoryl)-5-methyl-1-pyrroline *N*-oxide (DIPPMPO) and Į-(4-pyridyl-1-oxide)-*Ntert*-butylnitrone (POBN) spin trapping agents suitable for the detection of oxygen-centered radicals in the aerated aqueous TiO2 suspensions upon exposure confirmed the dominant generation of hydroxyl radical spin-adducts (Figure 5). The chiral centre in EMPO and DIPPMPO molecules may result in the production of *trans* and *cis* spin-adduct diastereoisomers. Indeed, the simulation analysis of the corresponding experimental EPR spectra summarized in Table 2, revealed the superposition of two individual EPR signals belonging to hydroxyl radical spin-adduct diastereoisomers. Additionally, low-intensity EPR signals of radical intermediates originating from the spin traps decomposition were detected as the carbon-centered spinadducts (• EMPOdegr, • DIPPMPOdegr) or the four-line signal of hydroxy *tert*-butylnitroxide (• POBNdegr) [61].

**Table 2.** Spin Hamiltonian parameters (hyperfine coupling constants and *g*-values) of the spin-adducts elucidated from simulations of the experimental EPR spectra obtained upon UVA irradiation (Ȝmax = 365 nm) of aerated TiO2 P25 suspensions in water and water/dimethylsulfoxide mixed solvent (5:1 v:v) in the presence of the corresponding spin trapping agents.


Symbols ȕ and Ȗ denote the position of the interacting hydrogen nuclei.

Previously, the photoinduced generation of O2 •í on the irradiated TiO2 nanoparticles was confirmed by low temperature EPR measurements below 160 K [25,31,32,34,52]. Even though the spin trapping agents EMPO and DIPPMPO were specially designed for the detection of superoxide radical anion in aqueous media and biological systems [60,63–66], the EPR signals reflecting the presence of • EMPO-O2 í /O2H and • DIPPMPO-O2 í /O2H spin-adducts were not found in the irradiated aqueous TiO2 suspensions (Figure 5). Most probably, under the given experimental conditions, the superoxide radical anions are preferably transformed to hydrogen peroxide by a disproportionation with protons [24,26,67]. Moreover, a significantly lower rate constants for the addition of O2 •í / • O2H to the nitrone spin traps may cause the limited production of spin-adducts [46,48]. The rapid transformation of superoxide radical anions, as well as their very slow reaction with nitrone spin traps caused that at room temperature in aerated aqueous TiO2 suspensions superoxide detection using conventional spectroscopic techniques failed, and only chemiluminescence with luminol or luciferin analog was applied as a suitable experimental method [68–70].

**Figure 5.** Experimental (**black**) and simulated (**red**) EPR spectra obtained upon irradiation (Ȝmax = 365 nm; irradiance 15 mW·cm<sup>í</sup><sup>2</sup> ; exposure 400 s) of the aerated aqueous TiO2 P25 suspensions in the presence of various spin trapping agents (TiO2 concentration 0.167 mg·mL<sup>í</sup><sup>1</sup> ): (**a**) EMPO (*SW* = 8 mT; *c*0,EMPO = 0.05 M); (**b**) DIPPMPO (*SW* = 16 mT; *c*0,DIPPMPO = 0.035 M); (**c**) POBN (*SW* = 8 mT; *c*0,POBN = 0.05 M). Simulations represent linear combinations of corresponding spin-adducts (hfcc parameters listed in Table 2): (a) *trans*- • EMPO–OH (relative concentration in %, 68), *cis*- • EMPO-OH (28), • EMPOdegr (4); (b) *trans*- • DIPPMPO-OH (77), *cis*- • DIPPMPO-OH (10), • DIPPMPOdegr (13); (c) • POBN-OH (92), • POBNdegr (8).

#### *2.2. Spin Trapping in Non-Aqueous TiO2 Suspensions*

EPR spin trapping investigations of reactive radicals produced in the irradiated TiO2 dispersions so far were focused mainly on aqueous systems and on the detection of hydroxyl radicals [20,31,35–37,49]. Analogous experiments performed in organic solvents may provide interesting information concerning the radical intermediates generated [13,71–74], consequently we carried out spin trapping experiments with TiO2 nanoparticles dispersed in DMSO, acetonitrile (ACN), methanol and ethanol. The generation of electron-hole pairs upon TiO2 irradiation and their consecutive reactions resulting in the free radicals formation are substantially influenced by the solvent properties [54,75,76]. The increased solubility of molecular oxygen plays an important role in these processes (Table 3), together with the stabilization effect of the aprotic solvents on the superoxide radical anions and the reactivity of holes and hydroxyl radicals with the solvents (Table 1).


**Table 3.** Solubility of molecular oxygen in various solvents at 25 °C.

#### 2.2.1. Dimethylsulfoxide

The EPR spectra monitored upon UVA photoexcitation of TiO2 suspensions in aerated DMSO (Figure 6a) differ from those found when water was used as a solvent (Figure 3a), and the dominating signals represent spin-adducts • DMPO-O2 í and • DMPO-OCH3 with the spin Hamiltonian parameters summarized in Table 4. The superoxide radical anion stabilization in the aprotic solvent explains its favourable generation and consequently also trapping [24,25]. The production of • DMPO-OCH3 adduct is initiated by the oxidation of hydroxide anions or water molecules adsorbed on the titanium dioxide surface producing reactive hydroxyl radicals, which immediately attack the DMSO solvent forming methyl radicals (Equation (7)), as shown above in the mixed water/DMSO solvent (Figure 4). The rapid reaction of methyl radicals with molecular oxygen results in the generation of peroxomethyl radicals serving as a source of • DMPO–OCH3 spin-adducts (Equations (9)–(13)) [73]. Further low-intensity oxygen-centered spin-adduct assigned to • DMPO– OR originates from the solvent and most probably represents • DMPO–OCH2S(O)CH3:

$$\text{?CH3} + \text{Oz} \rightarrow \text{CHzOO}"\tag{9}$$

$$\text{LMPO} + \text{CH}\_3\text{OO}^\* \rightarrow \text{'DMPO} - \text{OOCH}\_3 \tag{10}$$

2 • DMPO–OOCH3 ĺ O2 + 2 • DMPO–OCH3 (11)

$$2\text{ CH}\_3\text{OO}^\* \rightarrow 2\text{ CH}\_3\text{O}^\* + \text{O}\_2 \tag{12}$$

$$\text{DMPO} + \text{CHxO}^{\bullet} \rightarrow \text{'DMPO} + \text{OCHx} \tag{13}$$

The photoinduced generation of methyl radicals upon the irradiation of titania-DMSO dispersions was confirmed by the nitroso spin trapping agent DBNBS, suitable for the identification of carbon-centered radicals, as typical signal of • DBNBS-CH3 and • DBNBS-CD3 (using DMSO-*d*6) were detected (Figure 6b,c). Depending on the experimental conditions the spin trap decomposition may occur during the photocatalytic processes demonstrated by the generation of • DBNBS-SO3 <sup>í</sup> (Table 4).

**Figure 6.** Experimental (**black**) and simulated (**red**) EPR spectra (*SW* = 8 mT) obtained upon irradiation (Ȝmax = 365 nm; irradiance 15 mW·cm<sup>í</sup><sup>2</sup> ; exposure 400 s) of the aerated TiO2 P25 DMSO suspensions in the presence of various spin trapping agents (TiO2 concentration 0.167 mg·mL<sup>í</sup><sup>1</sup> ): (**a**) DMPO (*c*0,DMPO = 0.035 M); (**b**) DBNBS (*c*0,DBNBS = 0.008 M); (**c**) DBNBS using DMSO-*d*<sup>6</sup> (83% vol.). The spin Hamiltonian parameters of corresponding spin-adducts are listed in Table 4. (**a**) • DMPO–O2 í (relative concentration in %, 46), • DMPO–OCH3 (46), • DMPO–OR (6), • DMPO–CH3 (2); (**b**) • DBNBS–CH3 (90), • DBNBS–SO3 í (10); (**c**) • DBNBS–CD3 (86), • DBNBS–CH3 (14).

$$
\bigwedge
\bigwedge
\bigwedge
\bigwedge
\bigwedge
\bigwedge
\bigwedge
\bigwedge
$$

(**a**)

#### 2.2.2. Acetonitrile

Due to the increased solubility of molecular oxygen in ACN (Table 3) the EPR spectra monitored in the irradiated systems TiO2/DMPO/ACN/air represent a four-line EPR signal with significantly broadened spectral lines (Figure 2c, Figure 7 blue line). Consequently, to obtain spectra suitable for the identification of spin-adduct parameters, the saturation of the exposed sample with argon is necessary. The experimental EPR spectrum recorded immediately after a post-radiation Ar-saturation and EPR spectrometer re-tuning, along with its simulation is shown in Figure 7 (black and red lines). Acetonitrile as an aprotic solvent stabilizes superoxide radical anions, consequently the spin-adduct • DMPO-O2 í dominates the EPR spectrum. The use of dried ACN solvent indicates that the hydroxyl radicals are generated by the oxidation of OH<sup>í</sup> /H2O adsorbed on the TiO2 surface via the photogenerated holes [80]. A lower reactivity of the photogenerated hydroxyl radicals towards acetonitrile (Table 1) allows the hydroxyl radicals to be trapped by DMPO and the • DMPO-OH was found in the spectra (Figure 7). The formation of • DMPO-OCH3 most probably relates to the interaction of hydroxyl radicals with the solvent [81] producing CH3OO• radicals trapped as the • DMPO-OCH3 spin-adducts [73]. The spin Hamiltonian parameters of the individual spin-adducts elucidated by the simulation of experimental EPR spectra obtained in TiO2/DMPO/ACN/air are summarized in Table 4.

**Figure 7.** Experimental (**blue**) EPR spectra (*SW* = 8 mT) obtained upon irradiation (Ȝmax = 365 nm; irradiance 15 mW·cm<sup>í</sup><sup>2</sup> ; exposure 400 s) of the aerated TiO2 P25 (0.167 mg·mL<sup>í</sup><sup>1</sup> ) suspensions in ACN containing DMPO spin trap (*c*0,DMPO = 0.035 M), along with EPR spectra measured after post-radiation saturation with argon (**black**) and their simulations (**red**). The spin Hamiltonian parameters of corresponding spinadducts are listed in Table 4. • DMPO-O2 í (rel. conc. in %, 60), • DMPO–OH (9), • DMPO-OCH3 (23), • DMPOdegr (8).

Additional experiments with nitrosodurene (ND) spin trap were performed in order to identify the structure of the carbon-centered radicals produced during the exposure of TiO2/ACN/air. The EPR spectra measured upon continuous irradiation (spectra not shown) revealed the presence of a nine-line signal of • ND–CH2CN, produced via the interaction of • OH with ACN [54], and a broad, three-line signal of ND•+ generated by the spin trap oxidation (Table 4).

#### 2.2.3. Methanol and Ethanol

The protic solvents methanol and ethanol are characterized with increased concentration of dissolved molecular oxygen (Table 3), and these solvents are well-known as efficient scavengers of photogenerated holes [82]. By their application radical intermediates created via interaction with both photogenerated charge carriers may be observed, *i.e*., electrons are scavenged by molecular oxygen forming O2 •<sup>í</sup> and holes react with alcohols producing the primary alkoxy radical species, • OCH3 and • OCH2CH3 (Equations (14) and (15)) [12,13,62].

$$\text{CHrOH} + \text{h}^+ \rightarrow \text{CHrO}^\* + \text{H}^+ \tag{14}$$

$$\text{CHrCHzOH} + \text{h}^+ \rightarrow \text{CHrCHzO}^\* + \text{H}^+ \tag{15}$$

**Table 4.** Spin Hamiltonian parameters (hyperfine coupling constants and *g*-values) of spin-adducts elucidated from the simulations of experimental EPR spectra obtained upon UVA irradiation (Ȝmax = 365 nm) of aerated TiO2 P25 suspensions in organic solvents in the presence of spin traps.

