Efficiency of La-doped TiO₂ calcined at different temperatures in photocatalytic degradation of β-blockers

2.1 Synthesis

The nanopowders have been synthesized by sol–gel process with titanium tetrachloride (TiCl₄, Merck Chemicals) as precursor. The Ti(OH)₄ hydrogel has been obtained by hydrolysis of TiCl₄ at 0 °C, with controlled addition of 2.5 wt.% alkalic solution NaOH (MP Hemija) into the aqueous solution of TiCl₄ (0.3 M) and careful control of the solution pH value (9.3) (Golubović et al., 2009; Ikehata et al., 2006). After aging in the mother liquor for 5 h, the as-prepared hydrogel was filtered and washed out with distilled water to remove chloride ions. Obtained Ti(OH)₄ hydrogel was converted to its alcogel by repeated exchange with amyl alcohol (Lacheme). The alcogel was dried at 280 °C and then calcined for 7 h at various temperatures (450, 550, 650, 700, or 750 °C). To obtain titania–based nanoparticles doped with 1 mol% of lanthanum, LaCl₃⋅7H₂O (Merck) has been dissolved in water prior to the hydrolysis of TiCl₄. All chemicals in this experiment were analytical grades, and used as received. Synthesized pure TiO₂ nanopowders calcined at 450 and 650 °C were labeled as T(CAN)₄₅₀ and T(CAN)₆₅₀, respectively, whereas La-doped titania−based nanopowders were labeled as TL, with calcination temperature indicated in parentheses.

2.2 Characterization methods

The phase composition of synthesized nanopowders was determined by X-ray powder diffraction (XRPD) analysis on Ital Structures APD2000 diffractometer, using Cu–Kα radiation (λ = 1.5406 Å). The PowderCell software (Kraus and Nolze, 2000) has been used for preliminary phase analysis, whereas the unit cell parameters and volumes have been calculated by the UnitCell software (Holland and Redfern, 1997). The mean crystallite sizes have been estimated by Williamson–Hall method (Williamson and Hall, 1953).

Morphology and composition/quality of the synthesized catalysts have been analyzed on SEM (JEOL JSM–6460LV, with the operating voltage of 20 keV) equipped with an EDS INCAx-sight detector and “INAx-stream” pulse processor (Oxford Instruments).

The porous structure has been evaluated from adsorption/desorption isotherms of N₂, at −196 °C, using the gravimetric McBain method. The main parameters of the porosity have been estimated by Brunauer–Emmett–Teller (BET) method and α_s–plot (Kaneko et al., 1998). The pore size distributions have been calculated by Barrett−Joyner−Halenda (BJH) and corrugated pore structure (CPSM) methods from experimental nitrogen sorption data (Androutsopoulos and Salmas, 2000; Barrett et al., 1951; Golubović et al., 2013).

The Raman scattering measurements have been performed in the backscattering geometry at room temperature in the air using Jobin–Yvon T64000 triple spectrometer (with 1800 g mm⁻¹ gratings) and TriVista TR557 triple spectrometer with 300/300/500 g mm⁻¹ grating combination. Both spectrometers were equipped with confocal microscope and nitrogen–cooled charge coupled device detector. The spectra measured at Jobin–Yvon T64000 system have been excited by 514.5 nm line of Ar⁺/Kr⁺ ion laser with output power less than 5 mW. The high frequency Raman spectra have been collected at TriVista system by using VerdiTM G–Series Optically Pumped Semiconductor Laser operating at 532 ± 2 nm, with power at the sample less than 1 mW.

2.3 Measurements of photocatalytic activity

The photocatalytic activity of the titania–based nanopowders doped with 1 mol% of La was evaluated by the degradation of the solution of MET (⩾99%, Sigma–Aldrich) and PRO (⩾99%, Sigma–Aldrich). The photocatalytic degradation was carried out as described in our previous research (Grujić-Brojčin et al., 2014; Šćepanović et al., 2012). Experiments were performed using 20 ml of 0.05 mM of MET or PRO and the TiO₂ loading was 1.0 mg ml⁻¹, except in the study of direct photolysis and in experiments where optimal concentration of catalysts was investigated. All experiments were performed at the natural pH which changed during the photodegradation, from pH 9 to pH 6 with sol–gel synthesized nanopowders in the case of MET and from pH 9 to pH 7 in the case of PRO. Results have been compared with photocatalytic degradation of investigated compounds in the solution of TiO₂ Degussa P25, with the starting pH value set to 9. In order to investigate the effect of hydroxyl radicals (^•OH), superoxide radicals $\left({\text{O}}_2^{\text{•}-}\right)$ , singlet oxygen (¹O₂), and photogenerated hole (h⁺) scavengers on kinetics of photocatalytic degradation, isopropanol (IPA, Sigma–Aldrich), benzoquinone (BQ, ⩾98%, Sigma–Aldrich), sodium azide (SA, ⩾99%, Sigma–Aldrich), and ammonium oxalate monohydrate (AO, Sigma–Aldrich), respectively, have been added to suspensions of MET and PRO. Kinetics of MET and PRO photodegradation was monitored with Ultra Fast Liquid Chromatography with Diode Array Detection (UFLC–DAD, Shimadzu) (Golubović et al., 2013; Grujić-Brojčin et al., 2014) at 225 nm for MET and 215 nm for PRO. The use of the gradient mode to follow the degradation kinetics was necessary in order to separate the peaks originating from investigated compounds and their intermediates and to shorten the time of the LC–DAD analysis. Reproducibility of repeated runs was around 3–10%.

2.4 Computational details

All DFT simulations were performed using Jaguar 9.0 (Bochevarov et al., 2013) program, as implemented in Schrödinger Materials Science Suite 2015-1. Geometries of MET and PRO were obtained employing the hybrid, non-local exchange and correlation functional of Becke-Lee, Parr i Yang (B3LYP) exchange and correlation functional with 6-31G+(d, p) basis set (Becke, 1993). To make the simulations more realistic, solvent effects of water were taken into account using Poisson-Boltzmann solver (Marten et al., 1996; Tannor et al., 1994). Optoelectronic properties, e⁻ and h⁺ reorganization energies (ERE and HRE respectively), were obtained using the screening calculation method which is intended to produce high quality results using small basis set. Detailed information on calculation procedure of HRE and ERE is provided in previous paper (Armaković et al., 2015a). Some key structural properties such as total surface area (TSA), polar surface area (PSA), and molecule volume are calculated with VegaZZ (Pedretti et al., 2002; Pedretti et al., 2003; Pedretti et al., 2004) using optimized geometries of MET and PRO. Additionally, hydrophobic areas of MET and PRO were calculated with Maestro (SchrödingerInc, 2015).

3.1 Structure and composition of La-doped titania-based catalysts

The XRPD patterns of synthesized nanopowders doped with 1 wt% La, calcined at different temperatures, are shown in Fig. 1. In the diffraction patterns of the samples calcined at 450, 550, and 650 °C, the most intensive diffraction peaks have been ascribed to the anatase crystal structure (JCPDS card 78–2486). The average size of anatase nanocrystallites was estimated by Williamson–Hall analysis as 10.8, 12.3, and 13.2 in the samples TL(450), TL(550), and TL(650), respectively, revealing slight crystallite growing with the increase in calcination temperature. Retaining the position (within experimental error) of main diffraction peaks of anatase with La-doping stands in favor of the assumption that La³⁺ ions, with much greater ionic radius (1.22 Å), did not substitute Ti⁴⁺ with smaller ionic radius (0.69 Å) in TiO₂ lattice. Such mismatch of the ionic radii of Ti⁴⁺ and La³⁺ has made the formation of Ti—O—La bonds or La₂O₃ oxide at the surface of anatase nanoparticles much more likely than real incorporation of La³⁺ into the TiO₂ crystal lattice (Liqiang et al., 2004; Grujić-Brojčin et al., 2014). The absence of the XRPD peak characteristic for lanthanum oxide in all XRPD data of La-doped samples may indicate that La oxide is highly dispersed and/or mainly amorphized (Shi et al., 2012). In the diffractogram of the sample calcined at 650 °C, low intensity diffraction peaks, which may be ascribed to sodium hexatitanate phase, Na₂Ti₆O₁₃ (JCPDS 73–1398) have been noticed. According to diffractogram of TL(750) sodium hexatitanate phase has became the dominant in that sample calcined at 750 °C. In the sample calcined at 750 °C the crystallite size of sodium hexatitanate has been estimated as 61.2 nm.

Close

Figure 1 XRPD patterns of pure and some La–doped TiO₂ nanopowders, with characteristic reflections of anatase phase given in parentheses. The asterisk denotes the peak at ∼31.8° Corresponding to NaCl.

Export to PPT

The EDS results, collected from corresponding framed areas of SEM micrographs, shown in the Supplement (Fig. S1), are presented in Table 1. The Na content has been in all samples, whereas Na/Cl ratio is increasing with the increase in calcination temperature above 550 °C, which is in accordance with XRPD results, showing the presence of sodium hexatitanate phase. The La-content has been detected in the sample calcined at 450 and 650 °C (2.01 and 1.53 wt.% of La, respectively). Note that in the samples with detected La-content, the oxygen weight percent is estimated as ∼45 wt.% in measured areas, which is greater than in stoichiometric TiO₂ (40 wt.%).

The structure and composition of synthesized nanopowders has been also analyzed by Raman spectroscopy. The most intensive Raman features in the spectra of TL(450), TL(550) and TL(650) shown in Fig. 2a have been assigned to anatase nanocrystalline phase (Šćepanović et al., 2012): E_g (∼143 cm⁻¹), E_g (∼199 cm⁻¹), B_1g (∼399 cm⁻¹), A_1g + B_1g (∼518 cm⁻¹), and E_g (∼639 cm⁻¹). The shift and broadening of the most intensive Raman E_g mode in these nanopowders, relative to its position and linewidth in the bulk anatase, are mainly ascribed to the nanosize effect, without significant influence of lattice strain or nonstoichiometry. By applying the phonon confinement model (PCM), described in detail in Ref. (Šćepanović et al., 2007), to this E_g mode the size of nanocrystallites has been estimated between 10 and 12 nm, slightly growing with the increase in calcination temperature. The PCM fitting curves, together with experimental data, are shown in the Fig. S2 in the Supplementary material. Beside the dominant Raman modes ascribed to anatase, low intensity modes detected in the range from 150 to 400 cm⁻¹ of the Raman spectra of the samples calcined at 450–650 °C can be ascribed to disordered and/or partially amorphized brookite phase (Golubović et al., 2009; Golubović et al., 2013): A_1g (∼153 cm⁻¹), A_1g (∼247 cm⁻¹), B_3g (∼288 cm⁻¹), B_1g (∼322 cm⁻¹), and B_2g (∼366 cm⁻¹).

Close

Figure 2 (a) Raman spectra of La-doped nanopowders calcined at 450, 550 and 650 °C with specified anatase Raman modes (A). The inset: Magnified spectra of the same samples with designated brookite (B) and sodium-titanate (ST) Raman modes. (b) Raman spectra of La-doped nanopowders calcined at 700 and 650 °C with dominant sodium-titanate structure. Some spectra were upshifted for clarity.

Export to PPT

The most prominent Raman modes in the spectra of the samples T(700) and T(750) (Fig. 2b) can be attributed to sodium hexatitanate, Na₂Ti₆O₁₃ (Su et al., 2000; Viana et al., 2011; Viana et al., 2009; Zarate et al., 2008). In the spectrum of the sample calcined at 700 °C a low intensity anatase modes have appeared, together with titanate modes, whereas in the spectra of sample calcined at the higher temperature, 750 °C, anatase modes have not been observed. The modes corresponding to the brookite phase have not been registered in Raman spectra of neither TL(700) nor TL(750). It should be noticed that some modes of Na₂Ti₆O₁₃ may also be recognized in the spectra of the sample calcined at 650 °C, coexisting with anatase and brookite modes. The phase composition of synthesized nanopowders, derived from a detailed analysis of their Raman spectra (presented in Fig. S3 and Table S1 in the Supplementary material), is summarized in Table 2.

The results obtained from XRPD and Raman measurements have shown that sol–gel synthesized titania–based nanopowders, doped with 1% La, retain their anatase structure as dominant phase, with some small amount of brookite, when calcined at temperatures up to 650 °C. The size of anatase crystallites in those samples has been slightly increasing with the increase in calcinations temperature, but still remained significantly lower (10–13 nm) than the crystallite size in the undoped samples (17.5–21 nm) synthesized under similar conditions and calcined at the same temperatures (Šćepanović et al., 2012). Such relative retaining of crystallite nanosize at different calcination temperatures originates from the suppressing of the anatase crystallite growth due to the La-doping (Parida and Sahu, 2008). Presented results have also revealed a gradual transformation from anatase to sodium hexatitanate with increasing calcination temperature from 650 to 750 °C.

It should be noted, that the analysis of XRPD results has supported the assumption that La³⁺ ions have rather been included in formation of Ti—O—La bonds on the surfaces of the TiO₂ nanoparticles and/or formed La oxide, than substituted titanium in anatase lattice. According to the results of scanning tunneling microscopy and spectroscopy (STM/STS), as well as spectroscopic ellipsometry (SE), relying on similar synthesis procedures and level of La-doping (Grujić-Brojčin et al., 2014), it is reasonable to believe that La oxide is mainly formed on the surface of anatase nanoparticles, significantly modifying their surface properties. Furthermore, the absence of the diffraction peak characteristic of lanthanum oxide in XRPD data, as well as lack of Raman peaks in the spectra of La-doped samples, could indicate that La oxide was very small and highly dispersed.

In order to identify the specific surface OH groups and estimate their amount at the surface of titania nanoparticles, the high frequency Raman spectra of anatase−based nanopowders are shown in Fig. 3. These spectra have dominated by the broad feature between 2800 and 3700 cm⁻¹ which can be assigned to O—H stretching vibration of water molecules adsorbed on the surface of the particles and condensed in the pores of TiO₂ nanopowders (Šćepanović et al., 2012). Thereby, the lower frequency part of this stretching OH band, located around 3200 cm⁻¹, can be related to the vibrations of OH groups with strong hydrogen bonds, the region of the stretching OH band around 3420 cm⁻¹ is due to the vibrations of OH oscillators with weak hydrogen bonds, whereas the higher frequency region of the stretching band (above 3620 cm⁻¹) can be associated with the vibrations of water molecules with broken hydrogen bonds (Burikov et al., 2010). The peak at about 2,933 cm⁻¹ and its shoulder at lower frequency side, which are superimposed on the broad stretching OH band, can be assigned to the stretching modes characteristic for CH - species (Šćepanović et al., 2012). However, low intensity sharp peak located at ∼3700 cm⁻¹ in the spectra of all anatase-based samples deserves spatial attention. Namely, this feature, assigned to the O—H stretching mode of isolated (free) hydroxyls, can be related to the ability of titania−based samples to form the surface O—H groups with the important influence on their catalytic activity (Šćepanović et al., 2012). This feature is more pronounced in the pure TiO₂ samples (T(CAN)₄₅₀ and T(CAN)₆₅₀) and La-doped sample (TL(650)) calcined at 650 °C, and Degussa P25, than in the La-doped samples calcined at lower temperatures (TL(450) and TL(550)). However, the intensity ratio of the feature at ∼3700 cm⁻¹ to the broad feature between 2800 and 3700 cm⁻¹ is the highest in the pure and doped samples calcined at 650 °C. As this intensity ratio could be taken as a measure of the amount of free hydroxyls compared to the amount of bonded hydroxyls in titania nanopowders, it can be concluded that the relative content of free hydroxyls is the highest in the samples T(CAN)₆₅₀ and TL(650).

Close

Figure 3 High frequency Raman spectra of synthesized anatase − based nanopowders and Degussa P25. The spectra were normalized on the intensity of the highest frequency anatase E_g Raman mode. After suitable baseline correction some spectra were upshifted for clarity.

Export to PPT

3.2 Morphology and pore structure of La-doped titania−based catalysts

The SEM images, presented in Fig. 4, have shown the evolution of the nanopowders morphology with increasing calcination temperature. As it can be seen in the micrographs, most of the particles in the samples with dominant anatase phase (TL(450), TL(550) and TL(650)) have spherical shape (Fig. 4(a–c)). The change in powder morphology, as a consequence of agglomeration, may be clearly observed. Namely, the diameters of grains agglomerated in those samples are at least twice larger than the anatase crystallite size estimated from XRPD and Raman data. The diameters are roughly estimated from micrographs (Fig. 4(a–c)), and they increase with the calcination temperature in the following manner: ∼25–27 nm in TL(450), ∼32–38 nm in TL(550), and up to ∼50 nm in TL(650). Much greater particles/agglomerates may be observed in the micrographs of the samples TL(700) and TL(750), as shown in Fig. 4(d–e). According to the analysis of XRPD and Raman results, these agglomerates may be ascribed to sodium hexatitanate.

Close

Figure 4 SEM micrographs of La-doped anatase–based nanopowder samples calcined at (a) 450, (b) 550, (c) 650, and sodium hexatitanate based samples calcined at (d) 700, and (e) 750 °C.

Export to PPT

In order to study the effects of calcination temperature on the pore structure and adsorption abilities of titania–based nanopowders, BET nitrogen sorption measurements have been carried out. The parameters of porosity, determined from the α_s–plots (Fig. S4 in the Supplementary material) obtained from the standard nitrogen adsorption isotherms (Fig. S5 in the Supplementary material) have shown that the samples with dominant anatase phase, TL(450), TL(550) and TL(650), are fully mesoporous. The pore parameters of the samples have been strongly influenced by the calcination temperature. Namely, in anatase–based samples the values of specific surface area (S_BET) and pore volume (V_p), obtained by BET method (Table 3), have been decreasing with calcination temperature increasing. The specific surface area decreases sharply from ∼83 m² g⁻¹ in TL(450) to the values of 49, and 45 m² g⁻¹, registered in TL(550), and TL(650), respectively. Decrease in the specific surface area with increasing calcination temperature continues down to S_BET = 4 m² g⁻¹ in the sample calcined at 750 °C.

The pore size distribution (PSD), estimated from desorption branch of the hysteresis isotherm loops by BJH method (Barrett et al., 1951), together with PSD obtained as a result of CPSM simulations (Androutsopoulos and Salmas, 2000; Golubović et al., 2013) for the samples with dominant anatase phase is shown in Fig. 5. These results have indicated that PSD was similar for all of these samples, whereas the mean pore diameters, calculated by BET, BJH and CPSM methods ( ${\bar{D}}_{\mathit{BET}}\text{,}{\bar{D}}_{\mathit{BJH}}\text{,}{\bar{D}}_{\mathit{CPSM}}$ , respectively, Table 3) slightly varied from sample to sample. The number of pore cylindrical segments with equal length and randomly distributed diameters (N_S) and the pore tortuosity factor (τ), as the parameters which measured the pore structure complexity according to CPSM (Androutsopoulos and Salmas, 2000; Golubović et al., 2013), had the highest and lowest values in the TL(450) and TL(550) samples, respectively (Table 3).

Close

Figure 5 Pore size distribution for La-doped nanopowders calcined at 450, 550, and 650 °C, calculated by BJH (dashed lines) and CPSM (thick lines) methods.

Export to PPT

3.3 Photocatalysis

The influence of calcination temperature of La-doped titania−based nanopowders on photocatalytic activity was studied through the photocatalytic degradation of MET and PRO under UVA light irradiation. According to kinetic curves (Fig. S6 in the Supplementary material), a linear dependence of ln(c/c₀) on the illumination time was obtained in the first 10 min of heterogeneous photocatalysis for all synthesized samples suggesting that the degradation reactions of the MET and PRO are of pseudo-first order kinetics (the linear correlation coefficients varied in the range 0.991–0.999). The dependence of calculated reaction rate constant (k′) on calcination temperature is presented in Fig. 6.

Close

Figure 6 Influence of calcination temperature of La-doped titania (1.0 mg ml⁻¹) on reaction rate constant of photocatalytic degradation of MET and PRO (0.05 mM). k′ determined for the first 10 min of irradiation.

Export to PPT

The highest photocatalytic efficiency has been found for the anatase samples (TL(450), TL(550) and TL(650)), as all of them eliminated about 97% of MET within 60 min and about 96% of PRO within 45 min. TL(450) exhibited higher efficiency in degradation of PRO, while TL(650) was the best in degradation of MET, Fig. 6. These differences are certainly the consequence of different characteristics of investigated molecules which will be analyzed in detail in chapter 3.9 and due to the inherent properties of catalysts. Most certainly the unexpected result is related to TL(550) comparing with TL(650) and TL(450). Such result could be a consequence of lower pore structure complexity (described by τ parameter, Table 3) of the TL(550) in comparison with the samples TL(450) and TL(650), in spite of slightly higher mean pore diameter in this sample. The lower photocatalytic efficiency of TL(550) may also be related to lower content of both isolated (free) hydroxyls (registered by Raman spectroscopy) and excess oxygen (registered by EDS) in this nanopowder.

It is already known that thermal treatment of titania samples synthesized in the presence of NaOH may lead to transformation into sodium hexatitanate at temperatures of ∼500 °C (Nam et al., 2013; Zarate et al., 2008). Due to La-doping this effect is in case of our samples apparently shifted to higher temperatures (650–700 °C), until the sodium hexatitanate as sole phase occurs at calcination temperature of 750 °C. Both samples with sodium hexatitanate as dominant phase, TL(700) and TL(750), have eliminated about 83% of MET within 60 min and 90% of PRO within 45 min of photocatalytic degradation process. So, photocatalytic efficiency obtained in the presence of these photocatalysts has been lower than the efficiency of anatase–based samples, but still good enough in the first 60 min or less.

Kinetics of photocatalytic degradation of the most efficient photocatalysts, TL(650) for degradation of MET and TL(450) for degradation of PRO, was compared to the activity of TiO₂ Degussa P25 and undoped TiO₂ synthesized under the same conditions (Fig. 7). Since starting pH value of suspension with doped and undoped TiO₂ nanopowders was around 9, pH value of suspension with TiO₂ Degussa P25 was set to mentioned value. As can be seen, the usage of La(III) for doping of TiO₂ has lead to the increased photocatalytic efficiency during the degradation of MET and PRO comparing with undoped TiO₂. However, comparing with TiO₂ Degussa P25, nanopowder TL(650) has higher photocatalytic efficiency during degradation of MET, while TL(450) has shown somewhat lower efficiency during degradation of PRO (Fig. 7).

Close

Figure 7 Kinetics of (a) MET and (b) PRO degradation under UVA irradiation using of different types of TiO₂ catalyst at pH 9 and direct photolysis. When present, catalyst loading was 1.0 mg ml⁻¹.

Export to PPT

The usage of La(III) for doping of TiO₂ led to the improved photocatalytic efficiency during degradation of MET and PRO probably due to anatase nanostructure and surface modification caused by doping, as well as better separation of e⁻–h⁺ (Li et al., 2004). Also, high concentration of organic pollutants on the surface of La-doped photocatalysts can be achieved because La(III) has the ability to form complexes with Lewis bases (amines, aldehydes, alcohols, thiols, etc.) (Wu et al., 2010).

Reasons for different efficiencies of investigated nanopowders in photocatalytic degradation of MET and PRO should be searched in subtle interplay between catalysts and substrates properties. For example, the catalyst TL(450) with the smallest anatase nanocrystallite size (10 nm) and the largest specific surface area (83 m² g⁻¹) was the most efficient in degradation of PRO among all synthesized catalysts, but slightly less efficient at pH 9 in comparison with commercial Degussa P25 with greater crystallite size (∼25 nm) and smaller surface area (∼53 m² g⁻¹) (Tomić et al., 2015). On the other side, the synthesized catalyst TL(650), with smaller specific surface area (45 m² g⁻¹) in comparison with Degussa P25, was still more efficient in degradation of MET than commercial catalyst in the first 20 min of irradiation. Such behavior could be explained by more pronounced interaction between specified substrate and nanoparticle’s surface, which was not always strictly correlated with the smaller particles and/or greater surface area of the catalysts.

Since TL(650) and TL(450) nanopowders exhibited the highest photocatalytic efficiency for MET and PRO photodegradation, respectively, the effect of their loading was examined in the range from 0.5 to 5.0 mg ml⁻¹ (Fig. 8).

Close

Figure 8 Effect of (a) TL(650) and (b) TL(450) nanopowder loading on the kinetics of (a) MET and (b) PRO degradation.

Export to PPT

As it can be seen in Figs. 8 and 9, the degradation rate increased with the increase in the titania loading up to optimal concentration of TiO₂, while it slightly decreased with the further increase in loading. A similar effect has also been observed in the cases of T(CAN)₆₅₀ and T(CAN)₄₅₀ samples (Fig. 9), wherein the optimal TiO₂ loading of T(CAN)₆₅₀ and T(CAN)₄₅₀ was 2 mg ml⁻¹ in both causes. Theoretically, the increase in the catalyst loading above the optimum value has no effect on the photodegradation rate, since all available light is already utilized. However, higher loading of TiO₂ has led to the aggregation of its particles and thus to a decrease in the contact surface between the reactant and photocatalyst particles. Also, when catalyst was overdosed, the intensity of the incident UVA light was attenuated due to the decreased light penetration and increased scattering. This effect has attenuated the positive effect of the dosage increment, leading to overall decrease in performance (Abramović et al., 2011a; Šćepanović et al., 2012).

Close

Figure 9 Influence of catalyst loading on reaction rate constant of photocatalytic degradation of (a) MET and (b) PRO. k′ determined for the first 10 min of irradiation.

Export to PPT

It should be pointed out that TL(650) and TL(450) have shown much higher photocatalytic activity in the whole range of the catalyst loading of MET and PRO degradation, respectively, compared with T(CAN)_650/450. Also, it can be seen that PRO is significantly faster removed from investigated systems comparing with MET, which is most certainly the consequence of differences in structural and electronic properties of investigated compounds, which will be analyzed in the next chapter.

Information about the reaction type and mechanism is very important in view of practical application of photocatalysts. The experiments have shown that decomposition by photolysis is negligible (Figs. 7 and 9), whereas reactive radicals and h⁺ could be generated during the photocatalytic degradation reaction in UV/semiconductor systems, significantly increasing the efficiency of degradation (Lin et al., 2016a). A decrease in organic compounds degradation efficiency in the presence of scavenger of reactive radicals and h⁺, suggests that the holes and generated radicals are together involved in the oxidation of this compound (Lee et al., 2015; Chou et al., 2016a; Yang et al., 2016). Photodegradation mechanisms of MET and PRO have been examined by adding 2.0 mM of IPA, BQ, SA, and AO (Fig. 10).

Close

Figure 10 Effect of different scavengers on the kinetics degradation of (a) MET using TL(650) and (b) PRO using TL(450) under UVA irradiation.

Export to PPT

As shown in Fig. 10a, decrease in MET degradation efficiency as a consequence of IPA quenching is more pronounced than in the case of AO and SA quenching, whereas the MET degradation efficiency is even more decreased due to BQ quenching. After 20 min of degradation without aforementioned scavengers, around 9% of MET was present in solution. However, 45%, 22%, 37%, and 68% of MET retained in the solution after 20 min of degradation in the presence of IPA, AO, SA, and BQ scavengers, respectively. Literature data confirm significant contribution of ${\text{O}}_2^{\text{•}-}$ in degradation of organic compounds (Chou et al., 2016b; Lin et al., 2016a; Lin et al., 2016b). Besides, the presence of IPA confirms outstanding contribution of ^•OH in the process of MET degradation, in comparison with contribution of h⁺, which is also in accordance with literature data (Abramović et al., 2011a; Yang et al., 2010). Singlet molecular oxygen, as an important agent in a number of chemical processes (Thomas et al., 2003), is also one of the main activated species responsible for degradation of MET.

It should be noted that the mechanism of photocatalytic degradation of PRO is significantly different in comparison with MET. It can be seen in Fig. 10b that the degradation of PRO is less influenced by the presence of IPA and SA in comparison with the presence of AO. Namely, after 20 min of degradation without scavengers only 6% of PRO left in the water suspension, while in the presence of IPA and SA 19% and 12% of PRO left, respectively. The ${\text{O}}_2^{\text{•}-}$ has somewhat more important role in the degradation of PRO in comparison with ^•OH and ¹O₂, because 28% of PRO has left after 20 min of degradation in the presence of BQ. However, the decrease in photocatalytic efficiency in degradation of PRO is the most pronounced in the presence of AO, which confirms that h⁺ has the most significant contribution in degradation of PRO. Yang et al. (2010) have also investigated photocatalytic degradation of selected β-blockers and they have emphasized larger influence of h⁺ in degradation of PRO, in comparison with the other investigated compounds.

3.4 Theoretical analysis of the MET and PRO

3.4.1

3.4.1 Structural consideration, frontier molecular orbitals and quantum molecular descriptors

In order to understand fundamental properties of MET and PRO, useful for the understanding of different influence of photocatalysts, we will analyze the properties of mentioned molecules obtained with computational DFT approach. Geometries of MET and PRO and structural molecular descriptors are provided in Fig. 11.

Close

Figure 11 Equilibrium geometries of (a) MET and (b) PRO with polar and total surfaces (Ų) and corresponding volumes (ų).

Export to PPT

From the aspect of interaction of organic molecule with photocatalyst, the lower the TSA and PSA the higher will be its possibility to incorporate within photocatalytic medium and thus to achieve more pronounced interaction with surface charge of photocatalyst. According to the results presented in Fig. 11, PRO molecule could more easily penetrate the medium as it has significantly lower TSA, PSA and volume than MET molecule. This further indicates that PRO molecule can be easier located at the surface of catalyst and thus easier interact with surface charge.

Since TL(450) has the largest surface area and pore volume, with higher pore complexity (greater τ parameter, Table 3), its higher efficiency in degradation of PRO, as a smaller molecule, could be expected. Lower efficiency of PRO degradation with TL(650) in comparison with MET could be the consequence of significantly smaller surface area and pore volume of the catalyst calcined at 650 °C. Nevertheless, MET is more efficiently degraded in the presence of catalysts calcined at 650 °C, TL(650) and T(CAN)₆₅₀, comparing with catalysts calcined at 450 °C, T(CAN)₄₅₀ and TL(450), (Fig. 6). Possible explanation could be the higher relative content of free —OH groups compared to the content of bonded OH groups originating from water on the surface of catalysts calcined at 650 °C (confirmed by higher relative intensity of the peak at around 3700 cm⁻¹ in the Raman spectra of these samples, Fig. 3). According to the results collected in this work it can be seen that ^•OH on the catalyst surface are more significant then h⁺ for the degradation of MET. These results are in agreement with experimental results where the ^•OH is included in the mechanisms of degradation of MET (Abramović et al., 2011a; Armaković et al., 2015b; Yang et al., 2010).

Highest Occupied Molecular Orbital (HOMO) and Lowest Unoccupied Molecular Orbital (LUMO) are the most important orbitals for the reactivity and are considered as frontier molecular orbitals. Frontier orbitals principally determine the way in which the investigated molecule interacts with other molecules (Devi and Ramkumaar, 2015; Srivastava et al., 2015). Distribution of HOMO and LUMO orbitals emphasizes the role of aromatic rings of MET and PRO and yields very similar conclusions for both molecules, since both HOMO and LUMO orbitals are mainly delocalized over aromatic rings and oxygen atoms (O17 for MET and O16 for PRO), Fig. 12.

Close

Figure 12 HOMO and LUMO orbitals of (a) MET and (b) PRO.

Export to PPT

The HOMO–LUMO gap of MET is 5.68 eV while HOMO–LUMO gap of PRO is 4.54 eV, indicating that MET possesses higher energy barriers to transfer e⁻ between the occupied and unoccupied orbitals. Consequently, according to HOMO–LUMO gap the MET molecule possesses significantly higher stability over PRO molecule.

When it comes to the interaction with other molecules, quantum molecular descriptors (QMD) such as chemical hardness (η), chemical potential (μ) and electrophilicity index (ω) can be very useful for determination of global stability and reactivity of investigated molecules (Srivastava et al., 2015; Xavier et al., 2015). Values of QMD for MET and PRO molecules, together with dipole moments are listed in Table 4.

Table 4 QMD of MET and PRO.

Structure	Dipole moment (D)	HOMO (eV)	LUMO (eV)	H (eV)	μ (eV)	ω (eV)
MET	3.42	−6.30	−0.62	2.84	−3.46	2.11
PRO	3.04	−5.92	−1.37	2.27	−3.64	2.92

Dipole moment is known QMD which indicates the extent of charge polarization of molecule. Significantly higher charge separation in case of MET, 3.42 D, comparing with 3.04 D in case of PRO, indicates that consequently MET has significantly more pronounced interaction with polar solvents such as water. However, both molecules have relatively high dipole moments indicating the possibility to interact with other molecules, such as with newly formed radicals in the observed system (Armaković et al., 2015b). Stability of molecules is often assessed employing η. This quantity is interpreted as the resistance toward change in number of e⁻. According to η, PRO is significantly less stable than MET, with the value of 2.27 eV, comparing with the η of MET which is 2.84 eV.

Chemical potential measures the power of an atom or a group of atoms to attract e⁻ (Parr et al., 1978). This quantity can be used for calculation of ω, which measures the propensity of a molecule to accept e⁻. Both μ and ω describe the affinity toward e⁻; thus, the higher the μ and ω are, the higher electrophilic character of the molecule is. According to results presented in Table 4 PRO molecule is stronger electrophile since both μ and ω are higher than for MET molecule. This leads to the conclusion that according to frontier orbitals, PRO molecule is significantly more reactive than MET molecule.

3.4.3

3.4.3 Average local ionization energy

Bearing in mind that reactivity is of local character, obtained results naturally impose the task to locate the reactivity sites and confirm the above statements saying that PRO molecules are more prone to interaction with positive charge. For these purposes average local ionization energy (ALIE) is particularly useful. ALIE, I(r), is defined as the energy which is necessary to remove an electron from point $\overrightarrow{r}$ in the space of a system (Politzer et al., 2010; Sjoberg et al., 1990). The lowest values of ALIE indicate the areas where electrons are the least tightly-held, which further tells a lot about the molecule sites most vulnerable to electrophilic or to free radical attacks (Politzer et al., 2010). ALIE can be calculated as the sum of orbital energies weighted by the orbital densities (Politzer et al., 1998; Sjoberg et al., 1990). The positions at which ALIE has the lowest values are representing areas of least tightly bound electrons and sites to be most reactive toward electrophiles (Murray et al., 1990; Sjoberg et al., 1990).

In this work, we mapped the ALIE to the electron density surface and the obtained representative ALIE surfaces are presented in Fig. 13. Closer inspection of ALIE surfaces indicates that PRO molecule has more significantly pronounced sites prone to interaction with positive charge. These sites are given by the red color, representing sites where electrons are least tightly-held, which is much less present for MET molecule. It is also very useful to know the particular numerical values ALIE of certain atoms within investigated molecules. Therefore, in Fig. 13 we are providing maximal and minimal ALIE values of MET and PRO as projected on electron density surface.

Close

Figure 13 Representative ALIE surfaces of (a) MET and (b) PRO.

Export to PPT

Taking into account that PRO contains two aromatic rings and MET only one, it is easily concluded that PRO molecule is more prone to interaction with positive charge comparing with MET, concretely on those parts of molecule. Obtained results are in agreement with the reported degradation mechanism of PRO (Fig. 10b), which implies that the vast majority of intermediates were formed through opening of one aromatic ring in comparison with ^•OH binding to the ring (Santiago-Morales et al., 2013).