Surface Modification of Magnetic TiO2 Core-Shell with Doped Cerium for Enhancement of Photocatalytic Performance

The core-shell structure of Ce-doped TiO2@SiO2@(Ni-Cu-Zn) ferrite noted of CTSF as composite nanoparticles (NPs) was synthesized using a modified sol-gel method. The physicochemical properties of as-prepared products were characterized completely by X-ray diffraction (XRD), Brunauer-Emmit-Teller (BET), X-ray photoelectron spectroscopy (XPS) and superconducting quantum interference device (SQUID), serially. Meanwhile, assessment of the photocatalystic activity of catalyst was performed by ultraviolet-visible spectrometry (UV-vis). The results of the study show that the anatase phase related to the TiO2 structure was constructed on the outer shell coating of composite NPs. However, the second phase associated with the Ce structure was not easy to be detected on the XRD pattern, confirming that the doping Ce had been incorporated into the TiO2 crystal structure. The mesoporous structure of Ce-doped TiO2 layers was demonstrated by the type IV isotherm and H3 type hysteresis loop. The homogenous pore size was generated with the specific surface area up to 111.916 m2/g and 0.241 cc/g of pore volume. The stoichiometry of the chemical composition formed with fewer defects on the surface of TiO2 layers was exhibited by the symmetry curve of Ti 2p3/2 and Ti 2p1/2 peaks of XPS spectra. Meanwhile, the redox couple corresponding to Ce3+/Ce4+ was incorporated inside the thin TiO2 coating. Furthermore, the catalyst magnetic NPs can be also separated by using an external magnetic field from the reaction system. The product performance associated with the degradation efficiency was achieved to be 50 % in the aqueous solution of methylene blue (MB).

tive semiconductor material that has been widely utilized in this system because of its superior photoelectric properties, non-toxicity, low cost and high stability [8][9][10][11]. In the working mechanism of the TiO 2 system, the electrons in the upper valence band will jump into the conduction band as they received the ultraviolet light, generating the electron and holes pair [12]. They create the formation of reactive oxygen species, which play directly for the degradation of organic pollutants via oxidation process. However, the rapid recombination rate for charge-carrying of electron pair holes and the narrow range of light response become a serious problem, which inhibits the photocatalytic process in practical applications [13,14]. In addition, the difficulty of separating the catalyst from the reaction system also leaves a new issue as they are applied in liquid waste treatment.
Several attempts have been made to overcome the issue by expanding the gap of the TiO 2 absorption band with reducing the possibility of recombination for electron-hole pairs. The strategies carried out include the design of sensitivity, doping impurities by the metal ions, and incorporation with other semiconductors [15][16][17]. In another part, the magnetic catalyst nanoparticles were also a good model to be developed for separating the catalyst product from the reaction system, which further will be described in more detail in the literature review. Thus, the collaboration between both the doping material and the magnetic catalyst model is an interesting strategy to be investigated in order to reduce the issue above.

Literature review and problem statement
Modification of TiO 2 catalyst by doping ions mechanism using the rare earth materials proven can enhance the photocatalytic activity [18]. Cerium (Ce) is one of the attractive rare earth materials that could prevent the electron-hole recombination rate in TiO 2 [19,20]. The study reported that doping of Ce up to 1 % (w/w) in TiO 2 can increase the photocatalytic efficiency in the 4-chlorophenol dye to be optimum up to 86.6 % for 3h at doping of 0.6 % Ce. However, the positive effect is maintenance with the crystalline structure as the synthesis of TiO 2 products was prepared by the solvothermolysis method. Another study developed with hydrothermal technique reported that the incorporation of Ce of 0.01 gram/Nitrogen for TiO 2 (0.01Ce/N-TiO 2 ) in the AO-7 dye increased the photocatalytic degradation up to 79.04 %, which was facilitated by the reduction of recombination rate of electron holes pair [21]. Likewise, the attending of Ce 3+ /Ce 4+ mixture in TiO 2 formed by the sol-gel process was an effective mode for inhibiting the electron-holes recombination, enhancing the photocatalytic activity [22]. The experiment of doping (0-2.0 %Ce)/La-TiO 2 demonstrated that product yields for 1.5 wt %Ce/La/TiO 2 were around 1.8, 1.9 and 2.1 times compared to La/TiO 2 for CH 3 OH, CH 4 and CO. Several recent investigations also reported the superiority of cerium in doping TiO 2 . A study of co-doped TiO 2 with 0.6 %Mn and 1 %Ce molar ratios exhibited the strongest photoactive catalyst with 94 % of DCF removal [23]. The cerium led to be an effective dopant for increasing the response of TiO 2 by an efficient bandgap narrowing and hindering the electron-hole recombination. Thus, it can improve the quantum yield in the photocatalytic process. In addition, an experimental result revealed that the doping model of 0.02 g Ce in TiO 2 -NPs-AC at pH 5.0 has shown the higher concentration of BR 46 leading to a decrease in removal percentage [24]. Unfortunately for these catalysts model, failure to recycle these nanostructures not only increases processing costs but can also lead to new types of pollution.
For the reason, incorporation of magnetic materials with nano-based catalyst TiO 2 particles is of particular interest because of their unique magnetic response, chemically modified surface, and low toxicity [25][26][27]. The studies reported that the magnetic catalyst of TiO 2 composite could be recovered from the reaction solution by using a permanent magnet. However, the direct contact between the TiO 2 layer and magnetic component results in the negative effect, which facilitates the increasing rate in electron-hole recombination. For reducing the adverse influence of magnetic materials, the interface layer between the magnetic core and TiO 2 catalyst should be modified. Some works reported that using SiO 2 materials as an intermediate layer of the core-shell structure has been introduced for creating the magnetic catalyst of TiO 2 /SiO 2 / ferrite [28][29][30][31]. So, the developing study of the magnetic catalyst by the intermediate layer is still promising in the TiO 2 core-shell structure.
In this study, surface modification of TiO 2 magnetic core-shell doped cerium as the catalyst product of Ce-doped TiO 2 @SiO 2 @(Ni-Cu-Zn) ferrite for enhancement of photocatalytic performance is reported. The doping composition of 2 %wt Ce in TiO 2 by the sol-gel method will be introduced. The characteristics of the structure, surface behavior and magnetic properties of the catalyst product are investigated and discussed in detail as well.

The aim and objectives of the study
The aim of the study is to improve the photocatalytic performance of the magnetic catalyst of the TiO 2 core-shell structure by adding cerium element.
To achieve this aim, the following objectives are accomplished: -Synthesis of the catalyst magnetic nanoparticles (NPs) by the sol gel method to create the TiO 2 core-shell structure.
-Surface modification at the outermost layer of coreshell structure with doping cerium in TiO 2 .
-Characterization of the structure, surface composition and magnetic properties of the product.
-Performance test of the product using the MB dye solution in a dark condition and under visible light.

Materials and methods for surface modification of TiO 2
magnetic core-shell structure doped cerium

1. Catalyst synthesis
The applied procedure for the NPs synthesis of Ce-TiO 2 @SiO 2 @(Ni-Cu-Zn) ferrite magnetic photocatalyst noted as CTSF is schematically designed in the core-shell model. The detailed synthesis process was reported in our previous study [32]. The final model of CTSF was created from tetrabutlytitanate (Ti(OC 4 H 9 ) 4 ) doped with 2wt %Ce, for which is coated on the SiO 2 @Ni-Cu-Zn fer-rite (SF) nanoparticles. Further, the model was calcined at 500 °C for 3 hours to generate the anatase phase of TiO 2 . All of the samples including (Ni-Cu-Zn) ferrite, SiO 2 @ (Ni-Cu-Zn) ferrite, TiO 2 @SiO 2 @(Ni-Cu-Zn) ferrite, and Ce-TiO 2 @SiO 2 @(Ni-Cu-Zn) ferrite were noted as F, SF, TSF, and 20CTSF, respectively. The samples are further characterized and tested to get the surface structure properties and photocatalystic performance.

2. Catalyst characterization
The crystalline structure was observed by using the X-ray diffraction (XRD) with Cu K α radiation (λ=0.154056 nm) (Rigaku D/Max-II, Tokyo, Japan) for a scan rate of 3°/minutes.The surface characteristics related to the surface area, pore volume, and pore size were measured by Brunauer-Emmit-Teller (BET) using nitrogen adsorption employing Autosorb-1C instrument (Quanta Chrome, Boynton Beach, FL). The nitrogen gas was used as an adsorbent, while pore volume and pore size were calculated from the BJH method cumulative adsorption. The surface chemical composition related to Ce element doped on the TiO 2 was measured by the X-ray photoelectron spectroscopy (XPS; VGS Thermo K-Alpha, Waltham, MA, USA) with an Al K α radiation as the exciting source. The magnetization properties were performed at ambient temperature using the superconducting quantum interference device (MPMS5; Quantum Design, San Diego, CA) set-up with a maximum magnetic field of 10 kOe.

3. Photocatalytic activity
The photo-degradation of MB concentration was carried out at 0.30 g powders of both TSF and 20CTSF magnetic photocatalysts in the 50 mL of 10 mg/L MB aqueous solution using UV-VIS spectrometer (Evolution 220; Thermo, Waltham, MA). Before the UV illumination process, the suspension was stirred with ultrasonic vibration in the dark room for 30 min to reach adsorption-desorption equilibrium. The performance of TSF and 20CTSF composite was measured under the 35-W Xe arc lamp acting as simulated sunlight for irradiation function. 5 mL of suspension was taken every 1 h during 6 h and centrifuged for separate powders with reacting liquid prior which was analyzed based on the absorbance intensity of MB at 664.6 nm.

Experiment results of characterization of structure
layer, surface area, surface composition, magnetic properties and photocatalytic performance Fig. 1 shows the XRD pattern of photocatalyst nanoparticles for F, SF, and TSF samples. The ferrite phase includes both of the Ni-Zn and Fe 2 O 3 associates of the core NPs shown in Fig. 1, a while the pattern of the SiO 2 coating is displayed in Fig. 1, c. Finally, Fig. 1, c, d demonstrates the pattern of anatase TiO 2 and Ce doped TiO 2 layer, serially.

1. XRD characterization
Based on the XRD results in Fig. 1, there are two main peaks pattern raised with both Fe 2 O 3 (◻) and Ni-Zn ferrite (○) phase in Fig. 1, a. Both peaks initiate from the core of magnetic ferrite NPs. While the same peaks in Fig. 1, b are obtained due to the amorphous phase of the SiO 2 layer on the magnetic ferrite core NPs. In Fig. 1, c, the third phase with the reduction in the intensity of both peaks before attend as the new peak of anatase (▽) from the TiO 2 structure. Meanwhile, there is no change of the pattern in Fig. 1, d with doped element Ce. Fig. 2, a displays the BET analysis results for isothermal adsorption-desorption of magnetic catalyst NPs of a -F; b -SF; c -TSF and d -20CTSF, regularly. While BJH pore size distribution of nanoparticles photocatalyst is shown in Fig. 2, b. Besides, the surface area, pore diameter and pore volume of the magnetic catalyst NPs are tabulated in detail in Table 1.

2. BET Analysis
On the BET curve results, the same trend with a narrow hysteresis loop with low pore volume both of F and SF samples were shown in Fig. 2, a. The broader hysteresis loop by higher pore volume was demonstrated by two kinds of the TSF and 20CTSF sample. For the BJH pore size distribution in Fig. 2, b the shaper peak of TSF compared to the 20CTSF sample was introduced. However, in the F and SF sample demonstrated the broader peak with low pore volume adsorbed.  Table 1 displays the surface characteristics of catalyst NPs associated with surface area, pore diameter, and pore volume, serially. The first column is the product sample. The second column is the surface area value that is the higher the S BET value, the greater the surface area formed. Interim, the third column is the value of pore diameter created on the surface product. In the end column, it is the value of pore volume obtained from the accumulation of the pore diameter with the depth hollow constructed on the surface.  Fig. 3 presents completely the XPS spectra of surface composition for Ce-TiO 2 /SiO 2 /Ni-Cu-Zn. The full range of observation for all of the elements is shown in Fig. 3, a. The characteristics of Ti 2p peak and the bonding element of Ti-O were shown in Fig. 3, b, c, respectively. Furthermore, the Ce 3d spectrum compared to the TSF sample was exhibited in Fig. 3, d.

3. XPS Spectra
According to the results of examination of XPS spectra, all of the elements were detected on the full range observa-tion up to 1,200 binding energy shown in Fig. 3, a. The Ti 2p peak including Ti 2p 3/2 and Ti 2p 3/2 correlated with titanium element present in the investigation in more detail at the range 456 to 468 eV in Fig. 3, b. While, the Ti-O bonding related to titanium oxide in Fig. 3, c was monitored in the range of 526 to 536 eV. Moreover, the attending of two photo-peaks compared to TSF peak representing the element cerium consist of Ce 3+ 3d (trivalent) and Ce 4+ 3d (tetravalent) was obtained for the range of 880 to 905 eV in Fig. 3, d. a b Fig. 2. N 2 adsorption-desorption isothermal of NPs photocatalyst: a -BET analysis using nitrogen gas for isothermal adsorption-desorption, b -BJH pore size distribution of NPs photocatalyst  Fig. 3. XPS spectra of magnetic NPs on the 20CTSF layer: a -full range survey, b -characteristics of Ti 2p, c -oxygen incorporated in Ti; d -Ce 3d spectrum Fig. 4 provides the magnetic characteristics of all of catalyst NPs samples, which are performed by the superconducting quantum interference device at room temperature. The saturation magnetization value with the minimum hysteresis loop was shown in Fig. 4. The curve hysteresis loop of catalyst magnetic is noted as F, SF, TSF, and 20CTSF, regularly.

4. Magnetization Test
The magnetization curve of all catalyst products was shown in Fig. 4. The saturation magnetization values of F, SF, TSF, and 20CTSF samples are 28. 12, 21.13, 4.73, and 9.23 emu/g, regularly. The greater value indicates the stronger performance of NPs for responding to the magnetic field. The inset of Fig. 4 demonstrates the photograph of 20CTSF NPs composite under a magnetic field after the completion of photo-degradation. The catalyst NPs can be collected from the deionized water solution after 30 seconds. Fig. 5 demonstrates the performance of magnetic catalyst NPs in MB dye solution, which is compared between both the TSF and 20CTSF samples. The catalytic activity was carried out in two steps, the first condition is in the dark room and the second phase is under visible light.

6. Photodegradation Performance
Two curves were exhibited in Fig. 5 where the blue line and black one are associated with the degradation rate of MB dye solution with the catalyst NPs of TSF and 20CTSF, respectively. In the first step, the MB dye concentration was reduced up to 27 and 29 % for the TSF and 20CTSF solution. Further, for the second process, both curves go down continually as the irradiation time under visible light was added to the photodegradation system. The final reduction at 6 hours, the degradation efficiency was achieved at 45 and 50 % for both catalyst samples.

Discussion of experimental results
The XRD pattern of photocatalyst nanoparticles for F, SF and TSF samples was performed in detail shown in Fig. 1. The spinel crystal structure of particle ferrite (Ni-Cu-Zn) is demonstrated by the pattern of sample F in Fig. 1, a. The pattern matched with the Joint Committee of Powder Diffraction Standard (JCPDS) with the number of 08-0234 for the crystal structure of ferrite Ni-Zn [33]. However, there is a second phase attending in the pattern, which was also matched with JCPDS of 72-0469 for Fe 2 O 3. Both kinds of those peak patterns seem broader, suggesting the particles in the nanometer size. After the deposited SiO 2 layer, there is no SiO 2 peak present on the pattern shown in Fig. 1, b. This could be due to the amorphous structure constructed on the SiO 2 layer. Since TiO 2 and Ce doped TiO 2 have coated on the SF particles, some of the new peaks attended on the XDR pattern in Fig. 1, c, d. Their presence was accompanied by a decrease in the intensity of some of the previous peaks. The new crystal peak located 25.27° at 2θ-XRD in (101) plane was originated from the anatase phase of TiO 2 . The pattern is following JCPDS number 21-1272 for the anatase crystal structure of TiO 2 [34]. Based on the result, it is explained that the TiO 2 layer was successfully formed on the outer shell of composite nanoparticles of TiO 2 @SiO 2 @(Ni-Cu-Zn) ferrite. However, no peaks of Ce were detected on the pattern in Fig. 1, d, which confirmed that all of the doping Ce had been incorporated into the TiO 2 crystal structure.
The adsorption analysis of nitrogen gas was done for the characterization both of specific surface area and pore-size distribution of photocatalyst magnetic particles. Fig. 2, a displays nitrogen gas isothermal adsorption-desorption for samples F, SF, TSF, and 20CTSF, serially. The type IV isotherm and H3 type hysteresis loop were investigated for all samples, indicating the product of TSF and 20 CTSF having the mesoporous structure [35]. The specific surface area (SSA) of both models, calculated with the standard multi-points Brunauer-Emmett-Teller (BET) method, was obtained at 126.831 m 2 g -1 and 111.916 m 2 g -1 , respectively. However, for F and SF models, the SSA value was obtained lower than both models tabulated in detail in Table 1. Besides in Fig. 2, b the peak pore size distribution of TSF based on Barret-Joyner Halenda (BJH) desorption isotherm shows shaper than 20CTSF, suggesting a homogenous pore size in the sample [36]. The greater SSA might facilitate the more surface contact area, which could be beneficial for accelerating the photocatalyst reaction for reducing dyes molecules.
The surface chemical composition and bonding element of 20CTSF catalyst NPs were observed in more detail by XPS displayed in Fig. 3. The full range survey up to 1,200 eV for XPS spectra is shown in Fig. 3, a, which composite main peaks are Si 2p, 2s, C 1s, Ti 2p, O 1s, Fe 2 2p, Ni 2p, Ce 3d, Cu 2p, and Zn 2p, serially. The attending of C 1s peak located at 285.1 related to carbon on the surface originates from the surface adventitious carbon. While the presence of Si 2p and Si 2s centered on 103 and 133.08 eV associate with the binding energy of Si 4+ and O 2in silica [37], identified due to the thin coating of TiO 2 on the SiO 2 layer. Meanwhile, the elements of Fe, Ni, Cu, and Zn corresponding to the magnetic core NPs are also detected at 709, 886, 975, and 1021 eV regularly, for which both layers of TiO 2 and SiO 2 are still in the very thin coating. Thus, the characteristic peaks of Ti 2p are demonstrated in detail in Fig. 3, b. There are two photo-peaks of Ti 2p 3/2 and Ti 2p 1/2 centered at 459.1 and 464.8 eV, serially [38]. The symmetry curve of both peaks without the shoulder on the lower energy sides indicates that the formation of TiO 2 was in the stoichiometries with fewer defects. Moreover, the O 1s peak binding energy with a prominent peak at 530 in Fig. 3, c was the oxygen incorporated in Ti, suggesting the Ti-O bonding constructed on the outer shell layers. Finally, the Ce 3d XPS spectrum peaks were identified in higher binding energy around 456 to 467 eV in Fig. 3, d [39], revealing that two photo-peaks of Ce 3+ 3d (trivalent) and Ce 4+ 3d (tetravalent) were attended for doping TiO 2 . The centered peaks of the trivalent are in two peaks positions of 886.18 and 904.23 eV, while the tetravalent peaks are also in two locations of 882.63 and 901.08 eV regularly, suggesting the presence of a mixed-valence state for Ce ions in the TiO 2 layer. Fig. 4 provides the magnetization curve of magnetic NPs tested at room temperature. The greater value indicates the stronger NPs influenced by a magnetic field. The magnetization curve of Ni-Cu-Zn ferrite as the catalyst core shows the minimal hysteresis loop ascribing a soft magnetic material of 28.12 emu/g compared to the bulk value of 80 emu/g. Furthermore, the decrease of saturation magnetization for SF TSF and 20CTSF corresponds to the nonmagnetic TiO 2 and SiO 2 layer coating over F NPs. The lower in these values is likely influenced by the thickness of the shell structure formed on the composite magnetic NPs. The inset of Fig. 4 demonstrates the photograph of 20CTSF NPs composite under a magnetic field after the completion of photo-degradation. It seems that the magnetic NPs of 20CTSF can be separated by using an external magnetic field for 30 seconds from the reaction system. The challenge of the model is to control the thickness of the layer as thin as possible during the synthesis process. The thinner layer coating can lead to the faster catalyst response to the external magnetic fields.
The photocatalystic activity of TSF and 20CTSF NPs was evaluated by the degradation of MB dye solution under visible light using a Xe arch lamp with 35 Watt in Fig. 5. Before photo-degradation under visible light, the adsorption-desorption was reacted in the dark room for 30 minutes whereas the decoloration of MB dye was achieved up to 27 and 29 % for TSF and 20CTSF NPs, respectively. It is well known that the surface area and many pores on the surface of catalyst NPs tabulated in Table 1 will facilitate the absorption of dyes contained in the MB solution. As the catalyst product was loaded under irradiation visible light, the photodegradation efficiencies were found to be increased gradually with increasing reaction time. The enhancement of the degradation efficiency was achieved to be 45 and 50 % for catalyst NPs of TSF and 20CTSF for 6 hours. Higher efficiency for the 20CTSF catalyst was suggested that the redox couple Ce 3+ /Ce 4+ lead to reduced bandgap energy. It can also facilitate the effective recombination delay of photo-induced charge carriers [40]. Moreover, they also have an effective dopant for improving the response of TiO 2 , which increases the quantum yield in the photocatalytic process [23]. However, the reduction of the specific surface area of 20CTSF based on the BET results can restrict the optimum performance of the product in degrading the MB dye. So, the right Ce doping composition is further needed to optimize the surface area for increasing the performance of the product. Also, the investigation related to the layer structure and surface morphology should be conducted in detail by the Transmission Electron Microscopy (TEM) and Scanning Electron Microscopy (SEM) to get a deep understanding of the character from the product. Thus, the cyclic loading should also be carried out to find out the performance of the product. Eventually, various dyes could be used to develop the product in the application.

Conclusions
1. The TiO 2 catalyst magnetic NPs as a synthesis product with the core-shell structure was successfully constructed by the sol gel method. The model of 2wt %Ce-TiO 2 @SiO 2 @ (Ni-Cu-Zn) ferrite noted as 20CTSF was introduced to the reduction of MB dye.
2. The attending of two photo-peaks of Ce 3+ 3d (trivalent) and Ce 4+ 3d (tetravalent) on the product layers was associated with the metal ions of Ce doping on the surface layer of TiO 2 .
3. The anatase phase of the TiO 2 structure has formed on the outer layer of the core-shell structure. The large surface area and great pore volume have been achieved by the 20CTSF product up to 111.916 m 2 /g and 0.241 cc/g, respectively. While the good enough magnetic field with 9.23 emu/g could separate the product by an external magnetic field from the reaction system. 4. The total performance of the 20CTSF product in the MB dye solution was achieved up to 50 % in the degradation efficiency during 6 hours. The performance of the catalyst was better up to 5 % of efficiency than without the doped Ce for the reduction of MB dye.