FeCl₃·6H₂O, FeCl₂·4H₂O, FeSO₄·7H₂O, and KOH were purchased from Tianjin Kermel Chemical Reagent Co., Ltd. (Tianjin, China); KNO₃, L(+)-glutamic acid (Gla, C₅H₉NO₄), tetrabutyl titanate (Ti(Bu)₄, Bu = OC₄H₉, CP) and methylene blue were purchased from Sinopharm Chemical Reagent CO., Ltd. (Shanghai, China); cetyltrimethylammmonium bromide (CTAB, C₁₉H₄₂BrN, ultrapure), MB and hexamethylenetetramine (C₆H₁₂N₄) were purchased from Aladdin Chemical Reagent CO., Ltd. (Shanghai, China); 3-aminopropyltriethyloxysilane (APTES) were purchased from Sigma (St. Louis, MO, USA), and all the reagents are analytical pure and used as received.

A solution of APTES was added into the above seed suspensions, stirred under Ar atmosphere at 25°C for 4 h. The prepared APTES-modified seeds were collected with a magnet, and washed with 50 mL of ethanol, followed by double distilled water for three times 40 .

In a typical synthesis, 0.2 g amino-functionalized seeds, 0.2 g CTAB, and 0.056 g HMTA were dissolved in 25 ml ethanol solution under ultrasonic condition at room temperature. The mixture solution was then transferred into a Teflon-lined tube reactor. Then, 1 ml Ti(Bu)₄ dropwise added in the tube, and was kept at 150°C for 8 h.

The prepared samples were weighed and added into 80 mL of methylene blue solutions (12 mg/L). The mixed solutions were illuminated under mercury lamp (OSRAM, 250 W with characteristic wavelength at 365 nm), and the MB solutions were illuminated under UV light in the photochemical reactor. The solutions were fetched at 10-min intervals by pipette for each solution and centrifuged. Then, the time-dependent absorbance changes of the transparent solution after centrifugation were measured at the wavelength between 500 and 750 nm.

TEM images were performed with a JEOL JEM-2010 (HT) (JEOL, Tokyo, Japan) transmission electron microscope operating at 200 kV, and the samples were dissolved in ethanol and dropped on super-thin cabon coated copper grids. SEM studies were carried out using a FEI Sirion FEG operating at 25 keV, samples were sprinkled onto the conductive substrate, respectively. Powder X-ray diffraction (XRD) patterns of the samples were recorded on a D8 Advance X-ray diffractometer (Germany) using Cu Kα radiation (λ = 0.1542 nm) operating at 40 kV and 40 mA and with a scan rate of 0.05° 2θ s^-1. X-ray photoelectron spectroscopy (XPS) measurements were made using a VG Multilab2000X. This system uses a focused Al exciting source for excitation and a spherical section analyzer. The percentages of individual elements detection were determined from the relative composition analysis of the peak areas of the bands. Magnetic measurements were performed using a Quantum Design MPMS XL-7 SQUID magnetometer. The powder sample was filled in a diamagnetic plastic capsule, and then the packed sample was put in a diamagnetic plastic straw and impacted into a minimal volume for magnetic measurements. Background magnetic measurements were checked for the packing material. The diffuse reflectance, absorbance and transmittance spectra, and photodegradation examination of the microspheres was carried out in a PGeneral TU-1901 spectrophotometer.

For the synthesis of the functional hybrid nanomaterials, we synthesized the colloidal solutions of iron oxides NPs with different shapes in ethanol at the first. These iron oxide NPs exhibit long sedimentation time, and are stable against agglomeration for several days. Then, iron oxides NPs were modified with amino group by APTES because silane can render highly stability and water-dispersibility, and it also forms a protective layer against mild acid and alkaline environment. As shown in Figure 2, hydroxyl groups (-OH) on the magnetite surface reacted with the -OH of the APTES molecules leading to the formation of Si-O bonds and leaving the terminal -NH₂ groups available for immobilization of TiO₂ 41 . The immobilization of TiO₂ can be explained by HSAB (hard and soft acids and bases) formula 42 . As a typical hard acid, Ti ions can be combined to the terminal -NH₂ groups (hard bases) easily, owing to there is small amount water in ethanol (95%), and then TiO₂ will be coated on the surface of amino-functionalized iron oxide NPs by hydrolysis and poly-condensation as follows:

We prepared the monodisperse spindle-like iron oxide NPs by ferric hydroxide precipitate method for evaluating and verifying our experimental mechanism and functional strategies. The electron micrograph of the starting weak-magnetic spindle-like hematite NPs are shown in Figure 3a, which have longitudinal diameter in the range from 120 to 150 nm and transverse diameter (short axis) around 40 nm. After TiO₂ coating (FT-1), the transverse diameter increased to around 50 nm, and the representative image is shown in Figure 3b. Moreover, the obvious contrast differences between the pale edges and dark centers further clearly confirms the composite structure. Therefore, the results reveal that this functional strategy for fabricating the TiO₂-functionalized iron oxide NPs is a feasible approach. Then, two strong magnetic iron oxide NPs with different shape and diameter as seeds were employed to fabricate the magnetic TiO₂ hybrid materials. As shown in Figure 3c, Fe₃O₄ NPs with an obviously hollow structure have diameters around 100 nm, and the insert field-emission SEM image illustrates the hollow NPs present sphere-like shape. In our previous report, we have confirmed that the hollow Fe₃O₄ NPs were formed by oriented aggregation of small Fe₃O₄ NPs 38 . Figure 3d shows bright field TEM image of the corresponding iron oxide NPs after the same TiO₂ coating process (FT-2). However, the hybrid NPs present a shagginess sphere-like shape and cannot observe the hollow structure. Additionally, the diameters of hybrid NPs increased about 5 to 10 nm. The results reveal that the hollow Fe₃O₄ NPs have been covered by TiO₂. Owing to the loose struture of Fe₃O₄ seeds, TiO₂ will fill to its internal and surface, and finally cause the hybrid products present a solid nature. The diameter of above two different iron oxide NPs including spindle-like and hollow is relatively large, subsequently, we employ the ultrafine Fe₃O₄ NPs as seeds to fabricate the hybrid NPs. Figure 3e presents the TEM images of ultrafine Fe₃O₄ NPs without any size selection, the size is about 5 to 8 nm. By introduce the TiO₂, the as-obtained products (FT-3) exhibit an aggregated nature and the ultrafine Fe₃O₄ NPs dispersing in the TiO₂ matrix, as shown in Figure 3f.

XRD and XPS surface analysis was used to further confirm the structure and composition of iron oxides/TiO₂ hybrid NPs. Figure 4a shows the XRD patterns of the as-synthesized α-Fe₂O₃ seeds and α-Fe₂O₃/TiO₂ (FT-1). From the XRD patterns of α-Fe₂O₃ seeds, it can be seen that the diffraction peaks conformity with that of rhombohedral α-Fe₂O₃ (JCPDS no. 33-0664, show in the bottom). After coating, compared with that data of JCPDS no. 33-0664 and JCPDS no. 21-1272 (pure anatase TiO₂ phase), the (101) and (200) peaks of anatase TiO₂ can be found in FT-1, suggesting that α-Fe₂O₃/TiO₂ composite NPs are successfully fabricated by this method. Figure 4b shows the XRD patterns of the as-synthesized Fe₃O₄ seeds and Fe₃O₄/TiO₂ (FT-2 and FT-3). All peaks in the XRD patterns of both seeds can be perfectly indexed to the cubic Fe₃O₄ structure (JCPDS no. 19-0629, show in the bottom). After coating, the (101) peak of anatase TiO₂ can be clearly found in FT-2 and FT-3, suggesting that Fe₃O₄/TiO₂ hybrid NPs are successfully synthesized.

Figure 5 is the typical XPS spectra of the naked, amino-functionalized, and titania coating ultrafine Fe₃O₄ NPs, where part (a) is the survey spectrum and parts (b) to (d) are the high-resolution binding energy spectrum for Fe, Si, O, and Ti species, respectively. According to the survey spectrum, the elements of Fe, O, and C are found in the naked ultrafine Fe₃O₄ NPs, of which the element of C is found on the surface as the internal reference, and the elements of Fe and O arise from the components of Fe₃O₄. The new signals of N 1s, Si 2s, and Si 2p are observed in APTES-coated Fe₃O₄ NPs, and the new signal of Ti 2p signals is observed in FT-3 hybrid NPs. These results indicate that the FT-3 are composed of two components, silane functionalized Fe₃O₄ and TiO₂. It is noteworthy that many studies demonstrated that if particles possessed a real core and shell structure, the core would be screened by the shell and the compositions in the shell layer became gradually more dominant, the intensity ratio of the shell/core spectra would gradually increase 43 44 45 46 47 . The gradually subdued XPS signals of Fe after TiO₂ coating are discerned in Figure 5b. APTES coating increases the intensity of carbon and oxygen, and decreases the concentration of Fe; further TiO₂ coating decreases the intensity of silicon and Fe (as shown in Figure 5b, c). Therefore, after TiO₂ coating, corresponding XPS signals of Fe, and Si rule also are decreased, C and O do not match with this rule due to the formation of TiO₂ and surfactant impurities (as shown in Figure 5d, e). Additionally, interactions should exist among APTES-coated Fe₃O₄ NPs and titania which cause the shift of binding energy of Fe. Usually, XPS measures the elemental composition of the substance surface up to 1 to 10 nm depth. Therefore, XPS could be regarded as a bulk technique due to the ultrafine particles size of the FT-3 (less than 10 nm). The XPS result indicates that the amino-functionalized Fe₃O₄ seeds have been coated by a TiO₂ layer, thus greatly reducing the intensity signals of the element inside. Table 1 lists the binding energy values of Fe, Si, O, N, and Ti resolved from XPS spectra of the above three different NPs. In three cases, the value of binding energy of Fe 2p and other elements are very close to the standard binding energy values. Relative to the standard values 48 , the binding energy values in FT-3 have decreased and this result is in agreement with the previous discussions.

Furthermore, XPS surface analysis is also used to quantify the amount of titanium and iron present in the near surface region of the three different hybrid NPs. Figure 6 is the typical XPS spectra of the FT-1, FT-2, and FT-3, where part (a) is the survey spectrum and parts (b)-(d) are the high-resolution binding energy spectrum for Fe, Si, O, C, N, and Ti species, respectively. According to the survey spectrum, all hybrid NPs exhibited typical binding energies at the characteristic peaks of Ti 2p, Fe 2p, Si 2p, N 1s and O1s in the region of 458, 710, 103, 400, and 530 eV, respectively. Details of the XPS surface elemental composition results of as-obtained products are shown in Table 2. The XPS data of the titanium-to-iron ratio of hybrid NPs is calculated in which the elemental composition ratio of FT-1, FT-2, and FT-3 (titanium/iron) are about 2:1, 3.5:1, and 5.5:1. The results reveal that the quantity of Ti element is higher than that of Fe element on the surface of samples. That is, it may deduce that iron oxide NPs have been coated by TiO₂. In all hybrid NPs, the amount of oxygen to titanium or iron calculated from XPS data is about 5:1, this results is in agreement with the other reports 49 . Nevertheless, the combined results from TEM and XPS suggest that the synthesized hybrid NPs are composed of amino-functionalized iron oxide NPs and TiO₂.

Magnetic measurements of the hybrid NPs were performed on a SQUID magnetometer. As shown in Figure 7, hysteresis loops demonstrate that FT-2 and FT-3 have no hysteresis, the forward and backward magnetization curves overlap completely and are almost negligible. Moreover, the NPs have zero magnetization at zero applied field, indicating that they are superparamagnetic at room temperature, no remnant magnetism was observed when the magnetic field was removed 50 . Superparamagnetism occurs when the size of the crystals is smaller than the ferromagnetic domain (the size of iron oxide NPs should less than 30 nm), the size of the ultrafine Fe₃O₄ component in our product is less than 10 nm, and the hollow Fe₃O₄ is consist of small magnetite NPs, there are reasonable to suppose that the hybrid NPs showed superparamagnetic behavior. The results reveal that the products have been inherit the superparamagnetic property from the Fe₃O₄ NPs, and the saturation magnetization value (M _s) of naked hollow Fe₃O₄ and ultrafine Fe₃O₄ is 89.2 and 72.1 emu/g, respectively. After TiO₂ coating, the corresponding value of M _s decreases to 16.2 and 5.0 emu/g, respectively. The M _s decreased significantly after coating with TiO₂ due to the surface effect arising from the non-collinearity of magnetic moments, which may be due to the coated TiO₂ is impregnated at the interface of iron oxide matrix and pinning of the surface spins 51 . Moreover, this decrease in magnetic behavior is very close to other reports 52 53 . As the most stable iron oxide NPs in the ambient conditions, the magnetic properties of hematite are not well understood 54 55 56 . We checked the magnetic properties of FT-1 hybrid NPs, the M _s is about 2 × 10^-4 emu/g, and the composite NPs exhibit a typical ferromagnetism. Thereby, as a weak magnetic hybrid NPs, FT-1 cannot be separate by common magnet.

We checked the magnetic responsibility of FT-2 and FT-3 hybrid NPs under the external applied magnetic field by a common magnet. As shown in Figure 8, both hybrid NPs gather quickly without residues left in the solid and solution state when the magnet presence. The gathered hybrid NPs can be redispersed in the solution easily by a slight shake. The results illustrate that the hybrid NPs display a good magnetic response, and this is also important for the industrial application in water cleaning as MRCs for preventing loss of materials and save cost.

The three different hybrid NPs were further characterized by UV-vis absorption spectra to compare their optical adsorption properties and the results are shown in Figure 9a. The spectra highlight a strong adsorption in the UV region, the results are in agreement with the other reports 57 58 . It is noteworthy that the hybrid NPs with different morphology (at same concentration) will cause the difference of adsorption intensity and peak location. Due to the small dimensions of semiconductor NPs, a discretization of the bandgap occurs with decreasing particle size, leading to smaller excitation frequencies. A blue shift of FT-3 is observed in the extinction behavior, and the absorption edge is positioned at smaller wavelengths 59 . The result confirms that the diameter of FT-1 hybrid NPs is large than the other two different types hybrid NPs. Additionally, a concomitant tail can be clearly observed in the visible region of the absorption curve owing to scattering losses induced by the large number of inorganic NPs in the composite nanostructure 60 .

In order to calculate the bandgap of hybrid NPs, the relationship between the absorption coefficient (α) and the photon energy (hν) have been given by equation as follows: αhv = A(hv-E _E ) ^m , where A is a constant, E _g is the bandgap energy, hν is the incident photon energy and the exponent m depends on the nature of optical transition. The value of m is 1/2 for direct allowed, 2 for indirect allowed, 3/2 for direct forbidden, and 3 for indirect forbidden transitions 61 . The main mechanism of light absorption in pure semiconductors is direct interband electron transitions. The absorption coefficient α has been calculated from the Lamberts formula 62 , α = 1 t ln 1 T , where T and t are the transmittance (can be directly measured by UV-vis spectra) and path length of the colloids solution (same concentration), respectively. A typical plot of (αhν)² versus photon energy (hν) for the samples are shown in Figure 9b. The value of FT-1, FT-2, and FT-3 is 2.85, 2.89, and 2.73 eV, respectively.TiO₂ is important for its application in energy transport, storage, and for the environmental cleanup due to its well known photocatalytic effect with a bandgap of 3.2 eV 63 . Comparing with the pure TiO₂ NPs, the bandgap of hybrid NPs is obviously decreased, and the absorption edge generates obvious red shift. This red shift is attributed to the charge-transfer transition between the electrons of the iron oxide NPs and the conduction band (or valence band) of TiO₂ 64 . Iron oxide NPs can increase energy spacing of the conduction band in TiO₂ and finally lead to the quantization of energy levels and causes the absorption in the visible region. The other is that amino groups can act as a substitutional dopant for the place of titanium and change metal coordination of TiO₂ and the electronic environment around them 65 . Similar phenomenon of red shift in the bandgap for iron oxide/TiO₂ hybrid NPs were also found by other reports 53 65 66 67 .

The photocatalytic activity was examined by a colorant decomposition test using MB, which is very stable chemical dye under normal conditions. In general, absorption spectra can be used to measure the concentration changes of MB in extremely dilute aqueous solution. The MB displays an absorption peak at the wavelength of about 664 nm. Time-dependent photodegradation of MB is shown in Figure 10. It is illustrated that MB decomposes in the presence of magnetic TiO₂ hybrid materials. Generally, the pure TiO₂ NPs can decompose 40% MB in 90 min 68 69 70 . In our previous report, the pure TiO₂ NPs with a average diameter of 5 nm can be decomposed 53% MB in 90 min 71 . However, in our system, 49.0%, 56.5%, and 49.6% MB decomposed by FT-1, FT-2, and FT-3 in 90 min, respectively. The result reveals that the introduction of iron oxide NPs not only improve the photocatalytic activity but also employ the corresponding magnetic properties from itself. Thus, the as-synthesized magnetic hybrid NPs with high photocatalytic efficiency are very potentially useful for cleaning polluted water with the help of magnetic separation. The photocatalytic degradation generally follows a Langmuir-Hinshelwood mechanism, which could be simplified as a pseudo-first order reaction as follows 72 73 : r = - d C t d t = k C t , where r is the degradation rate of reactant, C is the concentration of reactant, k is the apparent reaction rate constant. The k for FT-1, FT-2, and FT-3 was 1.066% min^-1, 1.331% min^-1, 1.054% min^-1, respectively. It was surprising that the FT-2 exhibited such higher activity. This may be explained by light absorption capability of the FT-2 due to their rough shell contributes to the good photocatalytic activity. Compared to smooth surface, the rough surface layers can absorb more light because the UV-vis light can have multiple-reflections among the shagginess surface structure 74 .