UV-Vis-Induced Degradation of Phenol over Magnetic Photocatalysts Modified with Pt, Pd, Cu and Au Nanoparticles

The combination of TiO2 photocatalyst and magnetic oxide nanoparticles enhances the separation and recoverable properties of nanosized TiO2 photocatalyst. Metal-modified (Me = Pd, Au, Pt, Cu) TiO2/SiO2@Fe3O4 nanocomposites were prepared by an ultrasonic-assisted sol-gel method. All prepared samples were characterized by X-ray powder diffraction (XRD) analysis, Brunauer-Emmett-Teller (BET) method, X-ray photoelectron spectroscopy (XPS), scanning transmission electron microscopy (STEM), Mott-Schottky analysis and photoluminescence spectroscopy (PL). Phenol oxidation pathways of magnetic photocatalysts modified with Pt, Pd, Cu and Au nanoparticles proceeded by generation of reactive oxygen species, which oxidized phenol to benzoquinone, hydroquinone and catechol. Benzoquinone and maleic acid were products, which were determined in the hydroquinone oxidation pathway. The highest mineralization rate was observed for Pd-TiO2/SiO2@Fe3O4 and Cu-TiO2/SiO2@Fe3O4 photocatalysts, which produced the highest concentration of catechol during photocatalytic reaction. For Pt-TiO2/SiO2@Fe3O4 nanocomposite, a lack of catechol after 60 min of irradiation resulted in low mineralization rate (CO2 formation). It is proposed that the enhanced photocatalytic activity of palladium and copper-modified photocatalysts is related to an increase in the amount of adsorption sites and efficient charge carrier separation, whereas the keto-enol tautomeric equilibrium retards the rate of phenol photomineralization on Au-TiO2/SiO2@Fe3O4. The magnetization hysteresis loop indicated that the obtained hybrid photocatalyst showed magnetic properties and therefore could be easily separated after treatment process.


Introduction
Degussa (Evonik) TiO 2 P25 consisting of a mixture of anatase (∼78%), rutile (∼14%) phases and a minor amount of amorphous phase (∼8%) is a well-known commercial material frequently used to oxidize organic and inorganic compounds in air and water due to its strong oxidative ability and long-term photo-stability [1,2]. The photocatalytic activity of TiO 2 P25 is affected by light absorption, charge creation/recombination rate and surface activity. However, based on practicality, TiO 2 -based photocatalysis has some technical limitations that impede its commercialization. The first one is

122
The estimated values of phase content and crystallite sizes are presented in Table 1. The average 123 primary size of anatase crystals was about 20 ± 0.5 nm, whereas, for the rutile phase, it was 30 ± 1 nm.

124
Anatase phase content estimated according to the Rietveld method fluctuated from 57 ± 0.5 to 64 ± 3 125 wt %, whereas, for rutile, it was from 6.5 ± 0.2 to 8 ± 1.4 wt %. The diffraction peaks at 30°, 35 [42,43]. The average size of commercial magnetite crystallites estimated using the 128 Scherrer equation for the main peak of magnetite [311] was about 44 ± 1 nm. Moreover, the broad 129 diffraction peak at 2θ = 15-25° corresponded to an amorphous silica layer in the nanocomposite 130 structure [44,45]. For Au-TiO2/SiO2@Fe3O4 nanocomposite, the presence of gold nanoparticles (NPs) The estimated values of phase content and crystallite sizes are presented in Table 1. The average primary size of anatase crystals was about 20 ± 0.5 nm, whereas, for the rutile phase, it was 30 ± 1 nm. Anatase phase content estimated according to the Rietveld method fluctuated from 57 ± 0.5 to 64 ± 3 wt %, whereas, for rutile, it was from 6.5 ± 0.2 to 8 ± 1. 4 [42,43]. The average size of commercial magnetite crystallites estimated using the Scherrer equation for the main peak of magnetite [311] was about 44 ± 1 nm. Moreover, the broad diffraction peak at 2θ = 15-25 • corresponded to an amorphous silica layer in the nanocomposite structure [44,45]. For Au-TiO 2 /SiO 2 @Fe 3 O 4 nanocomposite, the presence of gold nanoparticles (NPs) with diameter of about 21 ± 1 nm was confirmed by diffraction peaks at 64.8 • and 77.8 • 2θ for Au [220] and [311]. However, other metal nanoparticles were hardly detected on the surface of TiO 2 indicating either a small amount of deposited metal or a small size of metallic NPs. The magnetic properties of Me-TiO 2 /SiO 2 @Fe 3 O 4 nanocomposites were measured at room temperature (293 K), and obtained results are presented in Figure 2 and Table 1. with diameter of about 21 ± 1 nm was confirmed by diffraction peaks at 64.8° and 77.8° 2θ for Au

132
[220] and [311]. However, other metal nanoparticles were hardly detected on the surface of TiO2

133
indicating either a small amount of deposited metal or a small size of metallic NPs. The magnetic properties of Me-TiO2/SiO2@Fe3O4 nanocomposites were measured at room 137 temperature (293 K), and obtained results are presented in Figure 2 and Table 1.

144
The surface composition of Me-TiO2/SiO2@Fe3O4 nanocomposites and oxidation states of C 1s,

145
Ti 2p and Fe 2p were determined by X-ray photoelectron spectroscopy (XPS) analysis, and the 146 obtained data are presented in Table 2.  For all obtained photocatalysts, the magnetic saturation reached ca. 10-12 emu·g −1 and did not depend on the amount of ferrite fraction in the nanocomposite and noble metal presence at the surface of TiO 2 .
The surface composition of Me-TiO 2 /SiO 2 @Fe 3 O 4 nanocomposites and oxidation states of C 1s, Ti 2p and Fe 2p were determined by X-ray photoelectron spectroscopy (XPS) analysis, and the obtained data are presented in Table 2. The main fraction of each nanocomposite was oxygen, as a component of TiO 2 , SiO 2 and Fe 3 O 4 . The content of oxygen varied from 53.5 to 59.6 at.%. In all Me-TiO 2 /SiO 2 @Fe 3 O 4 nanocomposites, the peak at binding energy (BE) 289-284 eV attributed to Si 2p was observed. The peak attributed to C 1s was observed at around 284-288 eV. The C 1s region could be deconvoluted for five peaks. It was found that carbon appeared as the COOH (BE~288.8 eV), C=O (BE~287.3 eV), C-OH (BE~285.3 eV) and C-C (aromatic and aliphatic) bonds (BE~283.8 eV and BE~284.3 eV, respectively). Carbon content varied from 9.5 to 13.4 at.% and was observed in the surface layer for all obtained photocatalysts, even for pure TiO 2 : see Table 2. Carbon content originated from the organic precursor of silica (TEOS) and the reaction environment. The correlation between carbon content and UV-visible irradiation-induced activity was not observed. Moreover, the band gaps for TiO 2 and noble metal modified magnetic photocatalysts were similar and equaled 3.1 eV. Based on energy-dispersive X-ray spectroscopy (EDS) analysis, the carbon content was about 2 wt % in all obtained photocatalysts.
The Ti 2p spectrum could be deconvoluted into two components at 458.6 and 458.1 eV binding energies and could be identified with TiO 2 and Ti 2 O 3 /Ti 3+ (in lattice, i.e., "self-doped titania"), respectively [46]. The presence of Ti 2 O 3 is not expected in well-crystallized titania P25, where even thermal treatment did not cause meaningful phase transition, e.g., an increase in rutile content from 14.4 to 18.8 was observed after 2 h calcination of P25 sample at 500 • C [47]. Intensities of Ti 4+ and Ti 3+ components showed that Ti 4+ was the dominant surface state (96-97%) for all obtained photocatalysts. The presence of all deposited metals was confirmed by XPS analysis, but their presence did not change significantly the surface composition of photocatalyst.
Moreover, the presence of platinum and silver was confirmed by (scanning electron microscopy/energy-dispersive X-ray spectroscopy (SEM/EDS) as is shown in Table 3. It must be pointed that, although metals were detected in all samples by SEM/EDS, their amounts were higher compared to XPS and slightly lower than used for deposition.

202
The frequency dispersion is observed due to the electrode porosity. An exemplary impedance 203 spectrum with appropriate fitting and used equivalent circuit (EQC) is presented in Figure 4b. The

211
The capacitance of the semiconductor/electrolyte interphase is fitted using constant phase 212 element (CPE). The diffusional impedance "W" is related to the ionic transport inside the pores of Sample Pt-TiO 2 /SiO 2 @Fe 3 O 4 had a Pt particle size in the range from 20 to 24 nm deposited on the titania surface. The size of Au nanoparticles of ca. 21 nm determined from XRD patterns was in good agreement with the size obtained from the STEM analysis (24 nm). It is well known that the efficiency of the photocatalytic process depends strongly on the particle size of metal deposits and titania physicochemical properties. It was reported that small and monodisperse silver nanoparticles below 10 nm exhibited the highest photocatalytic and antimicrobial activity [48,49]. However, for gold nanoparticles deposited on titania, an increase in gold NP size resulted in the enhancement of photocatalytic activity, probably due to the ability of the photoabsorption of more photons by larger and rod-like gold nanoparticles of broad plasmonic absorption band [50,51]. On the contrary, a considerable increase of photocatalytic activity was observed for small platinum and palladium NPs (2-3 nm) deposited on TiO 2 [52]. Therefore, it is assumed that the highest photocatalytic activity should reveal magnetic TiO 2 -based photocatalyst surfaces modified by fine palladium nanoparticles.
Mott-Schottky analysis was used to determine the location of flat band energy (E fb ) by measuring the space charge region capacitance (C sc ) at electrode/electrolyte interface. Exemplary Mott-Schottky plots of Pt/TiO 2 electrode are presented in Figure 4a. The capacitance of the space charge region was calculated using three different frequencies. As can be seen, different values of E fb can be interpolated. The frequency dispersion is observed due to the electrode porosity. An exemplary impedance spectrum with appropriate fitting and used equivalent circuit (EQC) is presented in Figure 4b. The proposed EQC consists of the lowest possible amount of elements.

211
The capacitance of the semiconductor/electrolyte interphase is fitted using constant phase 212 element (CPE). The diffusional impedance "W" is related to the ionic transport inside the pores of The capacitance of the semiconductor/electrolyte interphase is fitted using constant phase element (CPE). The diffusional impedance "W" is related to the ionic transport inside the pores of deposited TiO 2 to the blocking electrode (Pt). The comparison of Mott-Schottky analyses for un-modified and metal-modified TiO 2 nanocomposites is presented in Figure 4c. The Mott-Schottky plots showed that TiO 2 acts as an n-type (negative slope) semiconductor in all prepared photocatalysts. The flat band potential of TiO 2 nanocomposite was estimated at −0.69 V vs. Ag/AgCl (0.1 M KCl) and it is in good agreement with our previous results and literature data [53]. The flat band potentials were −0.925, −1.0, −1.405 and −1.5 V for Cu, Pd, Au and Pt modified nanocomposites, respectively. For all obtained photocatalysts, a significant cathodic shift of the flat band gap potential was observed. Tanabe and Ozaki [54] reported that photocatalytic properties of Me-TiO 2 photocatalysts depended on the work function of the used metal. Therefore, it can be concluded that the differences of work functions of Cu, Pd, Au, and Pt may also affect the location of flat band potential of modified TiO 2 . Thus, the presence of metal nanoparticles on the surface of titania should enhance the ability of TiO 2 to oxidize adsorbed species due to efficient electron trapping [55]. The Mott-Schottky plots also demonstrate the difference in the slopes of the curves of TiO 2 and metal-modified TiO 2 nanocomposites. There was a significant decrease in the slope of copper-modified TiO 2 compared to that of TiO 2 , indicating an enhanced charge carrier density and faster charge transfer for Cu-modified TiO 2 , which should contribute to higher photocatalytic activity.

Phenol Photocatalytic Degradation
The photocatalytic activity of the as-prepared nanocomposites was studied by examining the reaction of phenol degradation. No phenol was degraded in the absence of illumination, indicating that there was no dark reaction at the surface of Me-TiO 2 /SiO 2 @Fe 3 O 4 nanocomposites.
Efficiency of phenol photodegration and total organic carbon (TOC) reduction under ultraviolet-visible (UV-vis) light in the presence of titanium(IV) oxide magnetic nanocomposites modified with platinum, palladium, copper or gold nanoparticles are presented in Figure 5a,b, respectively.

240
The apparent first-order constant rate of phenol photodegradation increased from 8.
The stability of the magnetic particles (i.e., iron oxides) was studied by measuring the photodissolution of the iron oxides using atomic absorption spectroscopy (ASA). As presented in Table 4, the concentration of iron in the reaction medium was below the limit of quantification (1 mg·dm −3 ). The iron content for blank test, i.e., water without photocatalyst, equaled 0.49 ± 0.06 mg·dm −3 and in post-reaction medium was about 0.55 ± 0.10. It has been previously reported that combining magnetite and titanium dioxide may lead to the photodissolution of iron oxide phase and the leaching of iron ions to reaction medium [4,5]. However, the introduction of a silica layer results in the prevention of iron oxide photodissolutions [56]. Additionally, a nanocomposite analysis of copper species dissolution for Cu-TiO 2 /SiO 2 @Fe 3 O 4 was also performed. The content of Cu in post-reaction medium was defined as 0.41 ± 0.08 mg·dm −3 . The obtained concentration was also below the limit of quantification (0.5 mg·dm −3 ). Therefore, the obtained data clearly indicates that photocatalysts were stable during photocatalytic process and metal leaching was not observed.

Identification of Degradation Intermediates
Phenol is a non-volatile and common contaminant that is frequently present in industrial wastewaters. The US Environmental Protection Agency (EPA) and the European Union (EU) have classified phenolic compounds as priority pollutants since they are harmful to organisms even at low concentrations. Moreover, a higher content of phenol than 30 mg·dm −3 inhibits biological treatment or even eliminates sensitive microorganisms from activated sludge in biological wastewater plants and significantly reduces the biodegradation of other components. Therefore, it is important to study the pathway and intermediates formed during the photo-oxidation of phenol. It is well known that catechol, hydroquinone and 1,4-benzoquinone are the three most important aromatic intermediates in the reaction of phenol degradation [57]. The oxidation route of phenol occurs by the hydroxylation of its molecule to hydroquinone and catechol as a first step with a further oxidation of the dihydroxylbenzenes to benzoquinones [57]. Similarly, in our study, the highest yield was observed for hydroquinone, and then catechol generation during phenol photocatalytic decomposition. As shown in Figure 6, the concentration of hydroquinone and p-benzoquinone in aqueous phase during 60 min of irradiation was much higher for Au-TiO 2 /SiO 2 @Fe 3 O 4 than that for Pd-TiO 2 /SiO 2 @Fe 3 O 4 and Cu-TiO 2 /SiO 2 @Fe 3 O 4 photocatalysts, which revealed a significant enhancement of phenol photomineralization.
Generally, p-benzoquinone can be formed by (1) electrophilic attack of • OH radical, (2) direct photooxidation of phenol by holes h + , which may form phenoxyl radical and directly oxidize phenol to benzoquinone, or (3) direct oxidation of hydroquinone by oxygen dissolved in water [57]. Hydroquinone could be produced directly by • OH radical attack on a phenol molecule or by electron e cb − reduction of benzoquinone molecule. Interestingly, catechol was not formed during the first 10 min of irradiation only on Au-TiO 2 /SiO 2 @Fe 3 O 4 nanocomposite. It is proposed that keto-enol tautomeric equilibrium between hydroquinone and benzoquinone acts as a buffer and is responsible for the inhibition of phenol mineralization in the Au-TiO 2 /SiO 2 @Fe 3 O 4 photocatalytic system [37]. The highest formation of catechol, the lowest amount of formed benzoquinone and lack of hydroquinone generation during first 10 min of irradiation was observed for the Cu-TiO 2 /SiO 2 @Fe 3 O 4 photocatalyst, whereas, for Pd-TiO 2 /SiO 2 @Fe 3 O 4 photocatalyst, a high yield of catechol and benzoquinone generation was determined in the reaction mixture. The oxalic acid (aliphatic intermediate) in the reaction mixture appeared after 10 min of irradiation for Pd-TiO 2 /SiO 2 @Fe 3 O 4, and after 20 min for Cu-, Pt-and unmodified-TiO 2 /SiO 2 @Fe 3 O 4 , suggesting that the phenyl ring was destroyed, forming carboxylic acid intermediates through photocatalytic degradation. No oxalic acid generation was found for Au-TiO 2 /SiO 2 @Fe 3 O 4 nanocomposite during 60 min of irradiation.

300
For Pt-TiO2/SiO2@Fe3O4 photocatalyst only benzoquinone was detected in the first 10 min of 301 irradiation, whereas after 20 min of irradiation, hydroquinone and oxalic acid were also formed.

302
However, for platinum-modified photocatalyst, a significant change in the photoactivity with respect 303 to TiO2/SiO2@Fe3O4 nanocomposite was not noticed. Therefore, it is proposed that Pt particle sizes 304 were probably decisive for the photocatalytic activity under UV-vis irradiation. Previously, we have 305 reported that platinum deposited on anatase with a particle size below 3 nm exhibited the highest 306 photocatalytic activity. Larger particles (> 5 nm diameter) contained a decreased number of surface matrix; therefore, the size of titania determines much more the size of platinum than palladium or 309 copper.

310
To clarify the mechanism of phenol degradation, the analysis of hydroxyl radical formation on  Au-TiO2/SiO2@Fe3O4 Cu-TiO2/SiO2@Fe3O4 Pt-TiO2/SiO2@Fe3O4 Pd-TiO2/SiO2@Fe3O4 For Pt-TiO 2 /SiO 2 @Fe 3 O 4 photocatalyst only benzoquinone was detected in the first 10 min of irradiation, whereas after 20 min of irradiation, hydroquinone and oxalic acid were also formed. However, for platinum-modified photocatalyst, a significant change in the photoactivity with respect to TiO 2 /SiO 2 @Fe 3 O 4 nanocomposite was not noticed. Therefore, it is proposed that Pt particle sizes were probably decisive for the photocatalytic activity under UV-vis irradiation. Previously, we have reported that platinum deposited on anatase with a particle size below 3 nm exhibited the highest photocatalytic activity. Larger particles (>5 nm diameter) contained a decreased number of surface Pt atoms and therefore revealed decreased activity [48]. Platinum ions interact strongly with TiO 2 matrix; therefore, the size of titania determines much more the size of platinum than palladium or copper.
To clarify the mechanism of phenol degradation, the analysis of hydroxyl radical formation on the surface of Me-TiO 2 /SiO 2 @Fe 3 O 4 photocatalysts under UV-vis irradiation was performed by photoluminescence (PL) spectroscopy using terephtalic acid as a probe molecule. According to Figure 7, it was observed that obtained photocatalysts can produce • OH under UV-vis light irradiation. After 60 min of irradiation, the highest amount of • OH was observed for Cu-TiO 2 /SiO 2 @Fe 3 O 4 nanocomposite, but modification with Pd resulted in a significant decrease in • OH radical formation (much lower amount than that for unmodified nanocomposite).

Verification of the Degradation Mechanism Using Scavengers and Under N 2 Purging
The active species were investigated to understand the photocatalytic reaction mechanism. The holes (h + ), hydroxyl radicals ( • OH) and superoxide radical anion ( • O 2 − ) are the probable active species taking part in the photodegradation of organic pollutants. Results of the photocatalytic activity in reaction of phenol degradation in the presence of e − , h + , • O 2 − and • OH scavengers, i.e., silver nitrate, ammonium oxalate, benzoquinone and tert-butyl alcohol, respectively, and during N 2 purging are presented in Figure 8.   The degradation constant rates, determined without scavengers, serve as reference materials for particular photocatalysts. In the presence of no scavenger, the highest activity was exhibited by Pd-TiO 2 /SiO 2 @Fe 3 O 4 and Cu-TiO 2 /SiO 2 @Fe 3 O 4 photocatalysts. The lowest activity was found for SiO 2 @Fe 3 O 4 , with a degradation constant rate of 0.84 × 10 −2 min −1 . Compared to metal-modified TiO 2 /SiO 2 @Fe 3 O 4 nanoparticles, the activity of silica-magnetite nanocomposite was negligible, and therefore no further experiments were carried out for this sample. The low activity of SiO 2 @Fe 3 O 4 indicates that, despite the presence of ferric ions in the structure of nanocomposite, the photocatalytic activity results only from the presence of TiO 2 on the surface of magnetic nanocomposite, whereas SiO 2 @Fe 3 O 4 plays the role for an inert structure.

Effect of N 2 Purging
Molecular oxygen dissolved in water solution acts as an electron acceptor in photodegradation process, which limits the recombination rate between electron and hole through the formation of reactive oxygen species. In this study, the effect of dissolved oxygen was investigated by N 2 purging. As shown in Figure 8, the decrease concentration of oxygen in reaction solution resulted in inhibition of phenol degradation for all obtained magnetic photocatalysts. This observation demonstrates the predominant role of reactive oxygen species in mechanism of phenol degradation.

Effect of Benzoquinone
To identify the contribution of superoxide radical species on the photocatalytic degradation, prior to irradiation 500 mg·dm −3 of benzoquinone (BQ) was added into phenol solution. The photocatalytic performance was significantly suppressed after BQ was introduced, indicating that • O 2 − played a crucial role in the photodegradation process. Phenol degradation efficiency was reduced by 88% for Pt-TiO 2 /SiO 2 @Fe 3 O 4 and Pd-TiO 2 /SiO 2 @Fe 3 O 4 photocatalysts. For Au-TiO 2 /SiO 2 @Fe 3 O 4 , Cu-TiO 2 /SiO 2 @Fe 3 O 4 and TiO 2 /SiO 2 @Fe 3 O 4 nanocomposites, the addition of BQ inhibited the photocatalytic efficiency by 75-77% and slightly lower for TiO 2 P25 (60%).

Effect of Ammonium Oxalate
Ammonium oxalate was introduced as the scavenger of photogenerated holes (h + ). The photodegradation reaction was partly suppressed by 33% and 42% for Pd-TiO 2 /SiO 2 @Fe 3 O 4 and Pt-TiO 2 /SiO 2 @Fe 3 O 4 nanocomposites, suggesting that h + also played a role in the process of phenol oxidation. For Cu-TiO 2 /SiO 2 @Fe 3 O 4 , Au-TiO 2 /SiO 2 @Fe 3 O 4 , TiO 2 /SiO 2 @Fe 3 O 4 and TiO 2 P25 photocatalysts, ammonium oxalate capturing of h + have little or almost no effect on the efficiency of phenol degradation. Thus, h + is not the important active species for the reaction of phenol degradation in the presence of P25, TiO 2 /SiO 2 @Fe 3 O 4 and TiO 2 /SiO 2 @Fe 3 O 4 modified by copper or gold nanoparticles.

Effect of Silver Nitrate
Silver nitrate was utilized to trap the photogenerated electrons. The phenol degradation for TiO 2 P25, Cu-TiO 2 /SiO 2 @Fe 3 O 4 , Pt-TiO 2 /SiO 2 @Fe 3 O 4 and Pd-TiO 2 /SiO 2 @Fe 3 O 4 photocatalysts in the presence of AgNO 3 caused a negligible decrease in the photocatalytic efficiency. Moreover, for TiO 2 /SiO 2 @Fe 3 O 4 and Au-TiO 2 /SiO 2 @Fe 3 O 4 photocatalysts, the addition of silver ions has a rate-enhancing effect. Ag + ion could trap photogenerated electrons to avoid the recombination of electrons and holes. The positive effect of electron trapping may also result from the in-situ formation of bimetallic particles of Ag-Au deposited on the surface of TiO 2 /SiO 2 @Fe 3 O 4 nanocomposite. It was reported that such coupling of Ag and Au increases the photocatalytic performance of the photocatalyst [58,59].
These results confirm the crucial role of reactive oxygen species ( • O 2 − and • OH) in the photocatalytic degradation of phenol in the presence of Me-TiO 2 /SiO 2 @Fe 3 O 4 photocatalysts.

Discussion and Proposed Mechanism
The degradation of organic pollutants proceeds through the formation of intermediates, which often are very stable and toxic for the environment. In order to do the above, it is essential to investigate the mechanism of pollutant degradation. Of particular interest should be studies on correlation between activity, composition and selectivity of photocatalysts.
Oxidation or reduction pathways of TiO 2 -based photocatalytic reaction may depend on the crystalline structure, surface or bulk modification of semiconductor, pH and salinity of the surrounded medium [60][61][62].
The first step of photocatalytic reaction involves excitation of TiO 2 . Photogenerated hole and electron pairs may recombine or migrate on the semiconductor surface, taking part in reaction with water, oxygen and other species present on the TiO 2 surface or in the surrounded medium. However, the adsorbed molecules may be oxidized via direct or indirect methods [63]. A direct oxidation proceeded through the reaction of free holes/electrons with adsorbed organics, while an indirect oxidation proceeds through oxidation by reactive oxygen species (ROS). ROS are formed, depending on the pH of surrounding solution, by reaction of photogenerated holes with terminal oxygen ions or terminal hydroxyl groups. ROS may be also generated during a recombination of organic radicals. In our study, the pH of the aqueous phase was slightly basic, indicating (pH > pH IEP ) formation mainly of superoxide radicals. Miyazaki et al. [64] observed that, at room temperature, the generation of dihydroxybenzenes is favored during both direct and indirect oxidation. Kim and Choi [35] observed that photocatalytic activity of phenols may be also affected by the formation of surface complexes of adsorbed molecule and terminal titanium atoms. They proposed that the complex formation included covalent bonding as well as physical adsorption and depended on the surface area affected by TiO 2 structure. The complex formation could induce visible light activity, whereas, under UV light, hydroxyl radical generation played the main role. The formation of TiO 2 -organic molecule complex was observed for bare titanium(IV) oxide P25, while for Pt-TiO 2 and F-TiO 2 the complexation did not occur, due to blocking Ti-OH sites and the changing acidity of OH groups. Murcia et al. [65] have also investigated the formation of surface complex on Pt-TiO 2 surface. They observed that platinum nanoparticles favored the adsorption of phenol on photocatalyst surface by the formation of bidentate phenolates species. The adsorption depended on platinum NP size and the oxidation state of Pt species [65]. Small palladium nanoparticles preferentially are formed on lattice defects. Similarly, our previous studies for Ag/Au-, Pt/Pd-and Ag/Pt-modified TiO 2 showed the dependence of particle size of platinum and gold NPs on the crystallite size of titania particles, where small anatase with a larger number of surface defects than that in well crystallized larger particles of rutile and anatase stimulated formation of fine noble metal nanoparticles [48,59,66].
Combining the obtained results, a possible mechanism for the photocatalytic degradation of phenol with metal-modified (Me = Pd, Pt, Cu, Au) TiO 2 /SiO 2 @Fe 3 O 4 photocatalysts is proposed and illustrated in Figure 9. For Cu-TiO 2 /SiO 2 @Fe 3 O 4 , superoxide and hydroxyl radicals are mainly active species involved in phenol degradation, which attack the phenyl ring yielding catechol, and hydroquinone generation (benzoquinone was also detected but at significantly lower concentration). Then, the phenyl rings in these compounds disintegrate and short-chain organic acids are produced, mainly oxalic acid, which further mineralized to CO 2 and H 2 O (see Figure 9a). For Au-TiO 2 /SiO 2 @Fe 3 O 4 , the surface plasmon resonance induces electrons to be transferred to the conduction band of TiO 2 and are trapped by dissolved oxygen, resulting in generation of • O 2 − . The main intermediates determined in the first minutes of irradiation are benzoquinone and hydroquinone. However, the induction period was observed for generation of catechol by oxidation of phenol in the presence of h vb + . As shown in Figure 9b, the keto-enol tautomeric equilibrium between hydroquinone and benzoquinone retards the rate of phenol photomineralization.
The mechanism of phenol degradation in the presence of Pd-TiO2/SiO2@Fe3O4 photocatalyst 442 proceeds by generation of reactive oxygen species, e.g., ˙O2 − , which oxidize phenol to benzoquinone, 443 hydroquinone and catechol (see in Figure 9c). The highest concentration of benzoquinone and 444 hydroquinone was observed at the beginning of the reaction. However, hydroquinone concentration 445 decreased slower compared to the catechol amount due to different pathways of phenol photo-446 oxidation. Our observations are in good agreement with literature. Santos et al. [67] found that 447 catechol oxidation did not yield either benzoquinone or maleic acid formation, but oxalic acid, which 448 finally was mineralized to CO2. However, benzoquinone and maleic acid are products, which are 449 determined in the hydroquinone oxidation. The induction period for hydroquinone oxidation was 450 reported, meaning that fewer or more-quickly oxidizable intermediates are produced from catechol 451 than from hydroquinone [67]. In our study, a much higher mineralization rate was observed for Pd-

452
TiO2/SiO2@Fe3O4 and Cu-TiO2/SiO2@Fe3O4 photocatalysts, which produced the highest concentration 453 of catechol during photocatalytic reaction. For Pt-TiO2/SiO2@Fe3O4 nanocomposite, catechol was not 454 detected after 60 min of irradiation, resulting in a low oxidation to oxalic acid and mineralization to 455 CO2 (see in Figure 9d). Therefore, it is proposed that enhanced activity is related to a decrease in  The mechanism of phenol degradation in the presence of Pd-TiO 2 /SiO 2 @Fe 3 O 4 photocatalyst proceeds by generation of reactive oxygen species, e.g., • O 2 − , which oxidize phenol to benzoquinone, hydroquinone and catechol (see in Figure 9c). The highest concentration of benzoquinone and hydroquinone was observed at the beginning of the reaction. However, hydroquinone concentration decreased slower compared to the catechol amount due to different pathways of phenol photo-oxidation. Our observations are in good agreement with literature. Santos et al. [67] found that catechol oxidation did not yield either benzoquinone or maleic acid formation, but oxalic acid, which finally was mineralized to CO 2 . However, benzoquinone and maleic acid are products, which are determined in the hydroquinone oxidation. The induction period for hydroquinone oxidation was reported, meaning that fewer or more-quickly oxidizable intermediates are produced from catechol than from hydroquinone [67]. In our study, a much higher mineralization rate was observed for Pd-TiO 2 /SiO 2 @Fe 3 O 4 and Cu-TiO 2 /SiO 2 @Fe 3 O 4 photocatalysts, which produced the highest concentration of catechol during photocatalytic reaction. For Pt-TiO 2 /SiO 2 @Fe 3 O 4 nanocomposite, catechol was not detected after 60 min of irradiation, resulting in a low oxidation to oxalic acid and mineralization to CO 2 (see in Figure 9d). Therefore, it is proposed that enhanced activity is related to a decrease in palladium particle size, an increase in adsorption sites and efficient separation of charge carriers.
XRD analysis was performed using a Rigaku Intelligent X-ray diffraction system (SmartLab, Neu-Isenburg, Germany) equipped with a sealed tube X-ray generator (Neu-Isenburg, Germany). Measurements were performed on the 2θ range of 10-80 • with the scan speed of 1.00 • ·min −1 and scan step of 0.01 • . Crystallite size of photocatalysts in the direction vertical to the corresponding lattice plane was determined using the Scherrer equation, based on the corrected full width at half maximum (FWHM) of the XRD peak and angle of diffraction.
To characterise the light-absorption properties of modified photocatalysts, diffuse reflectance (DR) spectra were recorded, and data were converted to obtain absorption spectra. To characterise the light-absorption properties of modified photocatalysts, diffuse reflectance (DR) spectra were recorded, and data were converted to obtain absorption spectra. The band gap energy of photocatalysts was calculated from the corresponding Kubelka-Munk function, F(R) 0.5 E 0.5 ph against E ph , where E ph is photon energy. The measurements were carried out on ThemoScientific (Waltham, MA, USA) evolution 220 spectrophotometer equipped with PIN-757 integrating sphere (Waltham, MA, USA). Commercial TiO 2 P25 was used as a reference sample.
XPS analysis was carried out in multichamber ultrahigh vacuum (UHV) system (Omicron nanoTechnology, Taunusstein, Germany), at room temperature in a ultra-high vacuum conditions, below 1.1 × 10 −8 mBar. The photoelectrons were excited by an Mg-K α X-Ray source. The X-ray anode was operated at 15 keV and 300 W. Omicron Argus hemispherical electron analyser with round aperture of 4 mm was used for analysis of emitted photoelectrons. Measurements were carried out in a constant analyser energy (CAE) mode with pass energy equal 50 eV. The binding energies were corrected using the background C 1s line (285.0 eV) as a reference. XPS spectra were analysed with Casa-XPS software (Prevac, Rogów, Poland) using a Shirley background subtraction and Gaussian-Lorentzian curve as a fitting algorithm.
The particle size, dispersion uniformity and chemical composition of obtained nanocomposites was examined using transmission electron microscopy (HRTEM) using FEI Europe, Tecnai F20 X-Twin (Waltham, MA, USA).
Magnetic hysteresis loops were carried out using psychical properties measurements system (PPMS, Quantum Design, San Diego, CA, USA). Measurements were performed at temperature of 293 K, in the range of 0-30,000 Oe.
Electrochemical impedance spectroscopy (EIS) measurements were performed using the potentiostat-galvanostat AutoLab PGStat302N system (Utrecht, The Netherlands) under GPES/FRA software control. Electrochemical experiments were performed in a three-electrode cell in 0.2 M K 2 SO 4 in pH adjusted to 7. Ag/AgCl (0.1 M KCl) was used as a reference electrode and platinum mesh acted as a counter electrode. The impedance spectra were taken at the frequency range from 10 kHz to 100 Hz with 10 mV amplitude of the alternating current.
AC signal. Space charge region capacitances have been calculated from single frequency (1000 Hz) using Equation (1).
C-capacitance, Z -imaginary impedance, ω-angular frequency. Mott-Schottky analysis was performed in order to examine the influence of metal nanoparticles on the flat band potential of metal-modified TiO 2 . Flat band potential can be estimated from intercept of C sc −2 vs. E, according to Equation (2) assuming that the term k B T/e is small and can be neglected, where C sc -capacitance of space charge region, ε and ε 0 -dielectric constant of the material and permittivity of free space, e-electronic charge, N D -the number of donors, E-applied potential, E fb -flat band potential, k B -Boltzmann's constant, T-temperature.
Impedance spectra analysis and fitting were performed using EIS Spectrum Analyzer software. Atomic absorbance spectroscopy (AAS) measurements were performed using SENS AA DUAL spectrometer (GBC EQUIPMENT, Hampshire, IL, USA) equipped with hollow cathodic lamps Cu (GBC EQUIPMENT) and Fe (CPI International, Palo Alto, CA, USA).

Preparation of TiO 2 /SiO 2 @Fe 3 O 4 and Me-TiO 2 /SiO 2 @Fe 3 O 4 Photocatalysts
TiO 2 /SiO 2 @Fe 3 O 4 nanocomposites were prepared by an ultrasonic-assisted sol-gel method. Firstly, 1 g of commercial Fe 3 O 4 magnetic nanoparticles was dispersed in 50 cm 3 of ethanol and ultrasonicated for 15 min. Subsequently, 100 cm 3 of ethanol and 20 cm 3 of water were added to the suspension and sonicated for 30 min. Then, 150 cm 3 of ammonia ethanolic solution was dropwise added into magnetite dispersion and sonicated for another 15 min. In the next step, 7.71 cm 3 of tetraethyl orthosilicate (TEOS), preliminarily diluted in ethanol, was added to the magnetite particles suspension and ultrasonicated for the next 15 min. The weight ratio of magnetite to titanium(IV) oxide was equal to 1:2, whereas molar ratios of TEOS to Fe 3 O 4 and NH 4 OH to TEOS were 8:1 and 16:1, respectively. After aging of silica gel the suspension of commercial TiO 2 (P25) in 50 cm 3 of ethanol was added into SiO 2 @Fe 3 O 4 dispersion and stirred for 2 h. The obtained suspension of TiO 2 /SiO 2 @Fe 3 O 4 photocatalyst was separated, dried at 70 • C to dry mass and calcined at 400 • C for 2 h.
In order to obtain metal modified TiO 2 on SiO 2 @Fe 3 O 4 nanoparticles, 1.25 cm 3 of 0.1 M aqueous solution of particular metal precursor (Cu, Au, Pt, and Pd) was added dropwise into the suspension of TiO 2 /SiO 2 @Fe 3 O 4 photocatalyst and stirred for 30 min. The amount of deposited metal on TiO 2 was established to be 0.5 mol % regarding the amount of TiO 2 . Metal ions were reduced by adding of 0.1 M aqueous solution of sodium borohydride. The molar ratio of NaBH 4 to metal ions equaled to 1.5. The obtained suspension of Me-TiO 2 /SiO 2 @Fe 3 O 4 photocatalyst was centrifuged, dried at 70 • C to dry mass and calcined at 400 • C for 2 h.
Preparation of samples of Au-TiO 2 @SiO 2 /Fe 3 O 4 and Cu-TiO 2 @SiO 2 /Fe 3 O 4 were repeated three times, and preparation of Pt-TiO 2 @SiO 2 /Fe 3 O 4 was repeated two times without any changes in physicochemical properties (XRD, BET, colour) and photocatalytic activity.

Measurements of Photocatalytic Activity
Photocatalytic activities of the obtained samples were estimated by measurement of the rate of phenol decomposition in an aqueous solution under UV-vis irradiation. For each experiment, 50 mg of the photocatalyst was suspended in 25 cm 3 of aqueous solution of phenol (Co. = 20 mg·dm −3 ). The obtained suspension was mixed in darkness for 30 min to provide uniform adsorption of phenol on photocatalyst surface, and then irradiated under continuously stirring using 300-W xenon lamp of 50-mW·cm −2 power flux. Aliquots of 1.0 cm 3 of the aqueous suspension were collected after equal time invertals of irradiation, and filtered through syringe filters (ϕ = 0.2 m) to remove the photocatalyst particles. The temperature of the aqueous phase during irradiation was kept at 20 • C using a water bath. Phenol concentration was estimated by colorimetric method using a UV-vis spectrophotometer (Thermo Evolution 220). Phenol and phenol degradation intermediates were determined chromatographically using HPLC system (Kyoto, Japan) with UV-vis detector Shimadzu SPD-6A (detection wavelength: 254 nm) (Kyoto, Japan) and a WAKOSIL-II 5C18 AR column (dimensions 4.6 × 250 mm) (Wako Pure Chemical Industries, Tokyo, Japan) with a mobile phase contained water, acetonitrile and phosphoric acid with volume ratio of 70:29.5:0.5, respectively. Photocatalytic degradation analysis were repeated at least three times for each obtained magnetic photocatalyst.
The formation of hydroxyl radicals in suspension of Me-TiO 2 /SiO 2 @Fe 3 O 4 photocatalyst under UV-vis irradiation was evaluated by photoluminescence (PL) spectroscopy using terephthalic acid as a probe molecule under alkaline conditions. Hydroxyl radicals, produced during photocatalytic process, reactedwith terephtalic acid (TA) generating 2-hydroxyterephtalic acid, which emited fluorescence at around 426 nm [68]. Formation of hydroxyl radicals was estimated using the same experimental set-up as for measuring the decomposition rate of phenol under UV-vis light. After irradiation, the solution was filtered and analyzed on a Perkin Elmer LS55 (Waltham, MA, USA) fluorescence spectrophotometer with an excitation wavelength of 315 nm using NG3 and NG5 (Opole, Poland) cut-off filters. The spectra were recorded in the range of 360-550 nm.

Conclusions
The preparation procedure and characterization of new metal-modified (Me = Pd, Au, Pt, Cu) TiO 2 /SiO 2 @Fe 3 O 4 nanocomposites was reported. XPS analysis revealed that the deposition of different metals changed the surface composition of photocatalysts. The highest content of oxygen vacancies (Ti 3+ ) was observed for Pd-TiO 2 /SiO 2 @Fe 3 O 4 and Cu-TiO 2 /SiO 2 @Fe 3 O 4 nanocomposites. For all obtained photocatalysts, the magnetic saturation was about 10-12 emu·g −1 and did not depend on the amount and kind of metal deposited on the surface of TiO 2 . Mott-Schottky analysis showed a significant decrease in the slope of copper-modified TiO 2 compared to that of TiO 2 , indicating an enhanced charge carrier density and faster charge transfer for Cu-modified TiO 2 nanoparticles.
The highest photooxidatation rate of phenol and mineralization, measured as TOC reduction, was observed for Pd-TiO 2 /SiO 2 @Fe 3 O 4 and Cu-TiO 2 /SiO 2 @Fe 3 O 4 photocatalysts. Based on fluorescence spectra and analysis of scavenger formation, it has been found that superoxide and hydroxyl radicals are main active species involved in the degradation, which attacks the phenyl ring yielding catechol, hydroquinone and benzoquinone generation, followed by oxalic acid and CO 2 formation. It was found that the pathways for hydroquinone and catechol oxidation were different. Catechol was directly oxidized to oxalic acid and then mineralized to CO 2 , while the pathway of degradation for hydroquinone proceeded through the formation of a larger amount of intermediates, e.g., benzoquinone, maleic acid, which were further oxidized to aliphatic carboxylic acids and finally to CO 2 . The highest concentration of catechol and then oxalic acid during photocatalytic reaction was observed for the most active Pd-TiO 2 /SiO 2 @Fe 3 O 4 and Cu-TiO 2 /SiO 2 @Fe 3 O 4 photocatalysts. The enhanced activity was related to a decrease in noble metal and semi-noble metal particle size, an increase in the adsorption sites and efficient separation of charge carriers. For Au-TiO 2 /SiO 2 @Fe 3 O 4 the keto-enol tautomeric equilibrium retarded the rate of phenol photomineralization. Therefore, it is proposed that the hydroquinone pathway is a limiting step for phenol degradation.