Magnetically Recoverable Core–Shell Nanocomposites with Enhanced Photocatalytic Activity



Core–shell structured Fe3O4/SiO2/TiO2 nanocomposites with enhanced photocatalytic activity that are capable of fast magnetic separation have been successfully synthesized by combining two steps of a sol–gel process with calcination. The as-obtained core–shell structure is composed of a central magnetite core with a strong response to external fields, an interlayer of SiO2, and an outer layer of TiO2 nanocrystals with a tunable average size. The convenient control over the size and crystallinity of the TiO2 nanocatalysts makes it possible to achieve higher photocatalytic efficiency than that of commercial photocatalyst Degussa P25. The photocatalytic activity increases as the thickness of the TiO2 nanocrystal shell decreases. The presence of SiO2 interlayer helps to enhance the photocatalytic efficiency of the TiO2 nanocrystal shell as well as the chemical and thermal stability of Fe3O4 core. In addition, the TiO2 nanocrystals strongly adhere to the magnetic supports through covalent bonds. We demonstrate that this photocatalyst can be easily recycled by applying an external magnetic field while maintaining their photocatalytic activity during at least eighteen cycles of use.


Heterogeneous photocatalysis using semiconducting oxide catalysts is an efficient method for the purification of wastewater.1, 2 Nanosized TiO2 is one of the most widely used photocatalysts among the semiconductors being studied due to its favorable characteristics such as its low cost, good chemical stability, high photocatalytic activity, and nontoxic nature.36 Although it has been proven that most organic contaminants in water can be effectively mineralized under UV irradiation, efficient recovery of this finely powdered TiO2 from treated water is still a challenge, which prevents its widespread application. Nanocrystalline TiO2 immobilized on supporting materials such as glass, sand, or zeolite can improve the separation efficiency.7, 8 In an ideal case, the nanocrystals would deposit as a monolayer covering the support surface to maximize the catalytically active surface area. Unfortunately, deposition is generally nonuniform, creating difficulty in reaching the ideal value of overall photocatalytic activity due to a decreased surface-to-volume ratio.9 In addition, the immobilization is typically realized through noncovalent bonding such that catalyst nanoparticles may easily detach from the support, making their complete recovery from the treated solution difficult.

Magnetic separation provides a very convenient approach for removing and recycling magnetic particles (such as magnetite, ferrite, and barium ferrite) by applying external magnetic fields. The incorporation of magnetic components into TiO2 nanoparticle-based catalysts may, therefore, enhance the separation and recovery of nanosized TiO2.1017 However, compared with titania, the magnetic oxide particles are much more sensitive and unstable, especially under acidic conditions. It has been reported that the direct coating of the surface of magnetic particles with a layer of titania can protect the magnetic particles from chemical dissolution. Unfortunately, a photodissolution phenomenon may occur, which not only changes the properties of magnetic oxides but also deteriorates the photocatalytic activity of titania.1820 Furthermore, because the crystallization of sol–gel TiO2 usually involves high-temperature annealing, magnetic materials such as γ-Fe2O3 or Fe3O4, if treated concurrently, may rapidly transform to the antiferromagnetic α-Fe2O3 phase,21 therefore, diminishing the magnetic response. Inserting a passivation layer between the magnetic particle and the titania coating might help to overcome these problems. It has been found that the addition of a silica layer between an iron oxide core and a titania shell promotes the photocatalytic activity of the catalyst by decreasing the adverse influence of the magnetic oxide core.22, 23 However, in prior studies the photocatalytic activity of the hybrid spheres did not show much improvement when compared with single-phase anatase particles, probably due to the limited control over the crystallinity and size of the supported TiO2 nanocatalysts.

Herein, we report a new and efficient method for preparing Fe3O4/SiO2/TiO2 core–shell composite nanostructures featuring enhanced photocatalytic activity and fast magnetic separation. Briefly, this method involves sequential coating of highly field-responsive superparamagnetic colloidal nanocrystal clusters (CNCs) of magnetite (Fe3O4) with a layer of SiO2 and a relatively thin layer of TiO2, followed by calcination at high temperature, which converts the amorphous TiO2 layer into a monolayer of anatase nanocrystals. The average size of the anatase nanocrystals can be conveniently tuned by controlling the thickness of the TiO2 coating. The anatase-coated samples exhibit high photocatalytic efficiency, which, for nanocrystals with small average sizes, is better than that of well-known commercial photocatalyst Degussa P25. Unlike conventional immobilization methods, in this case the TiO2 catalyst particles connect to the SiO2 surface through covalent Ti[BOND]O[BOND]Si bonds so that catalyst nanoparticles strongly attach to the support. The thickness of the SiO2 interlayer can be controlled to stabilize the magnetite effectively and prevent the photodissolution of Fe3O4 during photocatalysis, making the catalyst reusable after multiple reaction cycles. As a result, the catalysts can be efficiently recovered from the reaction solution by using external magnetic fields for many cycles without significant loss of either materials or photocatalytic activity. This composite structure of silica and titania can improve the total photocatalytic activity, which was evaluated by photocatalytic degradation of Rhodamine (RhB). The influence of the thickness of the titania layer on the photodegradation of RhB has also been investigated. Finally, the recoverable photocatalytic activity of the Fe3O4/SiO2/TiO2 composites calcined at 500 °C was demonstrated by monitoring their photocatalytic activity during eighteen cycles of use.

Results and Discussion

Highly field-responsive superparamagnetic cores of Fe3O4 with a mean diameter of approximately 222 nm were prepared by using a high-temperature hydrolysis reaction that we reported previously, as shown in the transmission electron microscopy (TEM) image in Figure 1 a.24 Our prior studies suggest that each Fe3O4 core is composed of small primary nanocrystals of approximately 10 nm in diameter, thus retaining their superparamagnetism at room temperature while showing a much higher saturation magnetization than individual nanodots. As a result, they strongly interact with external magnetic fields and can be easily separated from solution in a low magnetic field gradient. By using the classical Stöber method, the magnetite particles can be easily coated with a silica layer with a controllable thickness (Figure 1 b). Comparing Figure 1 b with Figure 1 c, one can clearly see that the originally smooth surface of the Fe3O4/SiO2 particles becomes rough after coating with an additional thin layer of titania by using tetrabutyl titanate (TBOT) as the precursor and hydroxypropyl cellulose (HPC) as the surfactant. The surface roughness increases further after calcination of the dry powders of the sample at 500 °C for 2 h, suggesting the crystallization of the TiO2 layer. Importantly, the TiO2 nanocrystals still remain as a uniform coating on the surface of the silica and no apparent aggregation of the composite particles has been observed (Figure 1 d).

Figure 1.

TEM images of a) Fe3O4, b) Fe3O4/SiO2, c) Fe3O4/SiO2/TiO2, and d) calcined Fe3O4/SiO2/TiO2 particles at the same magnification. The insets in c) and d) are magnified images of portions of the composite particles showing the morphological change in the TiO2 shell due to calcination.

The uniform and nonaggregated nature of the composite particles can be more easily appreciated in scanning electron microscopy (SEM) images (Figure 2). The thickness of the TiO2 layer was estimated by calculating the mean diameters of the Fe3O4/SiO2, Fe3O4/SiO2/TiO2, and calcined Fe3O4/SiO2/TiO2 composites, by measuring at least 200 particles from the SEM images. In typical samples as shown in Figure 2, the average diameter is 311.0 nm for Fe3O4/SiO2, 340.0 nm for Fe3O4/SiO2/TiO2, and 338.7 nm for calcined Fe3O4/SiO2/TiO2 composites, suggesting the average thickness of the TiO2 layer is 14.5 and 13.8 nm before and after calcination, respectively. The slight decrease of the thickness during the calcination process may be attributed to the evaporation of physically and chemically adsorbed water and the densification of the TiO2 network. As expected, the thickness of the TiO2 layer can be controlled by adjusting the total amount of TBOT precursor in the initial coating process. For example, 200 μL of TBOT were used to produce TiO2 shells of approximately 13.8 nm, whereas for another two sets of samples prepared by adding 150 and 250 μL of TBOT under otherwise identical conditions, the thicknesses of the TiO2 layers were estimated to be approximately 12.6 and 14.7 nm, respectively (Figure 3).

Figure 2.

SEM images of a) Fe3O4/SiO2 (average diameter=311.0 nm), b) Fe3O4/SiO2/TiO2 (average diameter=340.0 nm), and c) calcined Fe3O4/SiO2/TiO2 (average diameter=338.7 nm) particles at the same magnification.

Figure 3.

Representative TEM images of Fe3O4/SiO2/TiO2 core/shell/shell structures after calcination at 500 °C with different amounts of TBOT precursor a) 0 μL (average diameter=252.9 nm), b) 150 μL (average diameter=278.2 nm), and c) 250 μL (average diameter=282.4 nm).

The successful coating and subsequent crystallization of the titania layer have been further confirmed by using X-ray powder diffraction (XRD, Figure 4). Before calcination, all diffraction peaks can be perfectly indexed to the magnetite phase of Fe3O4 (JCPDS 19-629). No characteristic peaks of other materials were detected, indicating that the TiO2 and SiO2 were both amorphous. After calcination at 500 °C for 2 h, the amorphous TiO2 transformed to the anatase phase, which is confirmed by the new peaks in the XRD pattern. As shown in Figure 4 b, the appearance of relatively strong and sharp diffraction peaks shows that the product became crystalline, and all of them can be indexed to the anatase phase of TiO2 (JCPDS 21-1272). The average crystal sizes of the TiO2 were estimated by using Scherrer’s formula [Eq. (1)], in which B is the width of the XRD peak at the half-peak height in radians, λ is the X-ray wavelength in nanometers, and θ is the angle between the incident and diffracted beams in degrees.((1))

equation image((1))
Figure 4.

XRD patterns of Fe3O4/SiO2/TiO2 core–shell structures a) before and b) after calcinations (A=anatase and M=magnetite).

The crystal size of the titania was estimated to be around 14 nm, suggesting that the thin outermost layer is composed of a ring of TiO2 nanocrystals. This is consistent with the TEM observation of the grainy surface. Owing to the protection of the SiO2/TiO2 double shells, the magnetite phase of Fe3O4 remains essentially unchanged during the high-temperature treatment, except for a slight increase in the crystallinity judging by the peak width at the half-height of the main diffraction peak.

The surface area and porosity of the Fe3O4/SiO2/TiO2 composite structures before and after calcination were investigated by using nitrogen adsorption–desorption isotherms, as shown in Figure 5 a. Before calcination, the isotherm is a combination of the typical type I and IV patterns with distinct H2 and H3 hysteretic loops in the range of 0.2–0.9 P/P0 and 0.9–1.0 P/P0, respectively, indicating the existence of ink-bottle- and tubular-shaped pores according to the IUPAC classification.25 It is believed that this bimodal distribution of pore sizes results from having pores with two different origins: the smaller mesopores are related to interstices between primary particles (the surface TiO2 nanocrystals), whereas the larger ones are associated with secondary interaggregation of the composite spheres.26, 27 Such bimodal pores can promote the rapid diffusion of various reactants and products during the photocatalytic reaction and enhance the rate of the photocatalytic reaction. After calcination, both of the hysteresis loops shifted to higher relative pressure (P/P0) ranges and the areas of the hysteresis loops gradually became smaller, suggesting that the average pore size was increased and the pore volume was decreased during calcination. The corresponding pore-size distribution of the Fe3O4/SiO2/TiO2 composites with and without calcination was determined by using the Barrett–Joyner–Halenda (BJH) method from the desorption branch of the isotherm (see Figure 5 b). The initial Fe3O4/SiO2/TiO2 composites clearly show features of micro-, meso- and macropores, whereas the micropores disappear after calcination mainly due to the densification of silica networks. The calculated average pore diameter and BET surface area of the Fe3O4/SiO2/TiO2 composites before and after calcination were found to be 4.2 nm and 288.9 m2 g−1, and 8.5 nm and 35.5 m2 g−1, respectively.

Figure 5.

a) Nitrogen adsorption–desorption isotherm (▪=adsorption □=desorption) and b) BJH pore-size distribution curve of the Fe3O4/SiO2/TiO2 core/shell/shell structures before and after calcination (▪=before and ▴=after).

The Fourier transform infrared spectroscopy (FTIR) spectrum of the calcined sample was measured to confirm the composition and structure of the nanocomposites. As shown in Figure 6, the calcined Fe3O4/SiO2/TiO2 composites show many more signals than Degussa P25 TiO2. It has been reported that the signal at the wave number around 800 cm−1 corresponds to the symmetric vibration of Si-O-Si, 1080–1100 cm−1 for asymmetric stretching vibration of Si-O-Si, 940–960 cm−1 for Si-O-Ti vibration, and 500–900 cm−1 originates from Ti-O-Ti.2830 By combining the results of FTIR and XRD, one can see that silica exists as a segregated amorphous phase in the anatase titania. The presence of water is evidenced by the appearance of the bending mode at 1640 cm−1 and the stretching mode at 3400 cm−1. This surface hydroxylation is advantageous for the photocatalytic activity of TiO2 because it provides higher capacity for oxygen adsorption.31, 32

Figure 6.

FTIR spectra of the calcined Fe3O4/SiO2/TiO2 core–shell structures and Degussa P25.

To demonstrate the potential applications of the magnetic photocatalysts for the removal of contaminants from wastewater, we investigated their photocatalytic activity by employing the degradation of RhB as a model reaction. For comparison, we have also prepared pure anatase TiO2 particles (Figure 7 a) and SiO2/TiO2 core–shell structures (Figure 7 b) by sol–gel processes and then calcined them at 500 °C for 2 h. The XRD measurements again confirm the anatase phase for both cases, although the core–shell particles contain smaller crystalline domains than pure TiO2 particles (15.3 nm and 18.8 nm, respectively, as estimated from the XRD patterns in Figure 7 c).

Figure 7.

TEM images of a) pure TiO2 spheres and b) SiO2/TiO2 core–shell particles. c) XRD patterns of TiO2 spheres and SiO2/TiO2 core–shell particles. Both samples have been calcined at 500 °C.

Figure 8 a shows the absorption spectra of an aqueous solution of RhB exposed to UV light for various time periods. The typical absorption peak at 553 nm gradually diminishes as the UV exposure time increases, and completely disappears after 40 min, suggesting the complete photodegradation of RhB by the calcined Fe3O4/SiO2/TiO2 composites. The changes in RhB concentration (C) over the course of the photocatalytic degradation reaction are shown in Figure 8 b. While keeping the total amount of TiO2 the same, we found that the single-phase anatase TiO2 particles and Fe3O4/SiO2 core–shell structures show low photocatalytic efficiency under identical UV-light exposure (especially for the Fe3O4/SiO2 samples, no apparent photocatalytic activity was detected). However, the anatase nanocrystal-coated samples (the calcined SiO2/TiO2 and Fe3O4/SiO2/TiO2 core–shell structures) exhibit high photocatalytic efficiency, which is comparable to that of well-known commercial photocatalyst Degussa P25. The good photocatalytic activity of the Fe3O4/SiO2/TiO2 composites may be caused by two factors. One is the small size of the anatase nanocrystals formed during the calcination of the shell. It has been pointed out in a number of prior works that the optimal size of an anatase particle for photocatalysis is around 10 nm.3335 Another reason is the SiO2/TiO2 composite structure, which has been reported in many papers to show enhancement in the overall photocatalytic efficiency.32, 36 In addition, the Fe3O4 core has essentially no contribution to the photocatalytic activity. The SiO2 layer may prevent photogenerated electrons transferring into the lower lying conduction band of the iron oxide core, thus eliminating the possible photodissolution of iron oxide in the aqueous reaction medium.19

Figure 8.

a) Absorption spectra of a solution of RhB in the presence of calcined Fe3O4/SiO2/TiO2 nanocomposite under exposure to UV light. b) Photocatalytic performance of various samples of catalysts (▪=UV, •=TiO2 spheres, ▴=Fe3O4/SiO2, ▾=SiO2/TiO2, ◂=Fe3O4/SiO2/TiO2, ▸=P25).

The effect of the thickness of the titania layer on the photodegradation of RhB was also investigated, with the results shown in Figure 9. The photocatalytic degradation of RhB follows pseudo first-order kinetics, and the photocatalytic reaction can be described simply by ln (C0/C)=kt, in which C and C0 are the actual and initial concentration of RhB, and k is the apparent degradation rate constant. It can be seen that thickness of the titania layer has a great influence on the photocatalytic activity of the nanocomposites. The sample with a 12.6 nm thick titania shell exhibits the highest photocatalytic efficiency. Its apparent rate constant of RhB degradation is 0.1729 min−1, which is much higher than that of Degussa P25 (k=0.1579 min−1). The increase of the thickness of the titania layer leads to a decrease in the photocatalytic activity, consistent with the relatively smaller surface area of thicker titania layers when the total amount of titania is kept constant. The large specific surface area is beneficial for photocatalytic activity by not only adsorbing more RhB molecules but also offering more reaction sites.

Figure 9.

Effect of thickness of titania layer on a) decomposition reactions and b) apparent rate constants of the calcined Fe3O4/SiO2/TiO2 composites (▪=12.6 nm, •=13.08 nm, ▴=14.7 nm, ⧫=P25).

Magnetically responsive photocatalysts also have the advantage of convenient separation and recycling of the catalyst in liquid-phase reactions by applying external magnetic fields. Recyclability is very important for a catalyst because it allows for multiple uses and, therefore, reduces costs. The complete removal of catalysts from treated water is also critically required to prevent additional contamination. As shown in Figure 10 a, the dispersion of Fe3O4/SiO2/TiO2 nanocomposites in water leads to a turbid suspension. Upon application of an external magnetic field (a 1 inch cubic NdFeB magnet), the solution quickly becomes transparent within 1 min due to the rapid harvest of the majority of the magnetic photocatalyst, and only 10 min is necessary to completely remove all of the composite particles. We have also monitored the stability of these magnetic photocatalysts by monitoring the photocatalytic activity during eighteen cycles of use. To ensure the RhB was completely mineralized and had no influence on next cycle, we extended the irradiation time to 1 h for each cycle. As shown in Figure 10 b, the catalyst did not exhibit any loss of photocatalytic activity even after eighteen cycles of reaction. In addition, we also checked the chemical stability of iron oxide in such a reaction solution. Photocatalysis was performed in an aqueous solution for many cycles by repeatedly adding RhB and illuminating with UV light. After 18 cycles, the concentration of Fe3+ in the solution was measured at 40.6 μg L−1, which was about only 0.0083 % of total Fe3O4 in the system (out of a total of ca. 24.6 mg in 50 mL solution; Figure 10 c). Under acidic conditions (pH 1), the dissolved Fe3+ in the solution after one cycle of reaction was found to be 273 μg L−1, a value higher than that in water. However, as this value corresponds to only 0.057 % of the total Fe3O4, one can still conclude an overall high stability of the samples even in acidic conditions in contrast to the uncoated Fe3O4 particles that can be completely dissolved at such a pH. These measurements demonstrate that the silica/titania shell can prevent the Fe3O4 core from chemical dissolution during photocatalysis.

Figure 10.

a) Separation process of Fe3O4/SiO2/TiO2 by using a cubic NdFeB magnet. b) 18 cycles of the photocatalytic degradation of RhB in the presence of Fe3O4/SiO2/TiO2 core–shell structures. c) Accumulated amount of Fe3+ dissolved in the solution after different cycles of photocatalysis. For comparison, the amount of Fe3+ dissolved in an acidic solution (pH 1) after one cycle of photocatalysis is also displayed.


In summary, we have prepared Fe3O4/SiO2/TiO2 core–shell photocatalysts with enhanced photocatalytic activity and fast magnetic separability. The structures were produced by converting a sol–gel deposited layer of TiO2 through calcination into a layer of anatase nanocrystals that were strongly bound to the surface of a Fe3O4/SiO2 magnetic core. The as-prepared nanocomposite shows a number of important features as a recyclable photocatalyst for wastewater treatment: it contains a highly field-responsive superparamagnetic Fe3O4 core for efficient magnetic separation, a silica interlayer for protection from chemical- and/or photodissolution of Fe3O4, an outer layer of size-controllable anatase nanocrystals, and a well-defined SiO2/TiO2 interface for great enhancement of the photocatalytic efficiency. The enhanced photocatalytic activity, excellent chemical stability, and fast magnetic separation make these multifunctional nanostructures potentially useful in practical settings of photocatalysis.

Experimental Section

General methods and materials: Diethylene glycol (DEG, reagent grade) and sodium hydroxide (NaOH, 98.8 %) were purchased from Fisher Scientific. Anhydrous iron(III) chloride (FeCl3, 98 %) was obtained from Riedel-de Haën. Tetraethyl orthosilicate (TEOS, 98 %), poly(acrylic acid) (PAA, MW=1800) and hydroxypropyl cellulose (HPC, Mw=80 000) were obtained from Sigma–Aldrich. Absolute ethanol (200 proof) was purchased from Gold Shield Chemical, and tetrabutyl titanate (TBOT, 99 %) from Fluka. All chemicals were directly used as received without further treatment. RhB was purchased from Acros Organics. Degussa P25, which contains about 80 % anatase and 20 % rutile with an average particle size of 30 nm and BET surface area of (50±5) m2 g−1, was obtained from Degussa AG (Germany).

Synthesis of Fe3O4spheres: Fe3O4 superparamagnetic cores were synthesized by using a high-temperature hydrolysis reaction that we developed previously.24 Typically, a NaOH/DEG stock solution was prepared by heating a mixture of NaOH (50 mmol) and DEG (20 mL) at 120 °C for 1 h under nitrogen. The solution was then cooled and kept at 70 °C. A mixture containing PAA (4 mmol), FeCl3 (0.4 mmol), and DEG (17 mL) was heated to 220 °C in a nitrogen atmosphere for 30 min under vigorous stirring to form a transparent, light-yellow solution. NaOH/DEG stock solution (1.8 mL) was injected rapidly into the above hot mixture, and the resulting mixture was further heated for 1 h to yield magnetite nanocrystal clusters with an average diameter of approximately 200 nm. The final products were washed with a mixture of distilled water and ethanol several times and then dispersed in 3 mL distilled water.

Synthesis of Fe3O4/SiO2core–shell structures: The interlayers of SiO2 were prepared through a modified Stöber method.37 In a typical process, the above Fe3O4 aqueous solution (3 mL) was mixed with ethanol (20 mL) and ammonium hydroxide (1 mL) under vigorous magnetic stirring. TEOS (0.1 mL) was injected into the solution every 20 min until its total amount reached 0.3 mL. After washing with ethanol three times, the products were redispersed in absolute ethanol (3 mL).

Synthesis of Fe3O4/SiO2/TiO2core/shell/shell structures: The outer layer of TiO2 was prepared by using a method developed by Kim et al.38 Typically, the above Fe3O4/SiO2 aqueous solution (1.5 mL) was mixed with distilled water (0.12 mL), HPC (50 mg), and absolute ethanol (25 mL) under vigorous magnetic stirring. An appropriate amount of TBOT dissolved in ethanol (5 mL) was introduced to the system drop by drop, followed by heating the solution at reflux at 85 °C for 90 min. The final products were washed with ethanol several times, dried at 60 °C for 4 h, and finally calcined in air at 500 °C for 2 h.

Synthesis of single-phase TiO2spheres: In a mixture of distilled water (0.12 mL), HPC (100 mg), and absolute ethanol (25 mL) under vigorous magnetic stirring, an ethanol solution (5 mL) of TBOT (1 mL) was introduced drop by drop, followed by heating the system at reflux at 85 °C for 90 min. The product was washed with ethanol several times, then dried at 60 °C for 4 h, and finally calcined in air at 500 °C for 2 h.

Characterization: The crystalline structures of the samples were evaluated by X-ray diffraction analysis on a Bruker D8 Advance Diffractometer with Cu radiation (λ=1.5418 Å). The size and morphology of the products were analyzed by using a Philips ESEM XL30 scanning electron microscope equipped with a field-emission gun operated at 10 kV and a Philips Tecnai 12 transmission electron microscope at 120 kV. The porosity of the products was measured by the nitrogen adsorption–desorption isotherm and BJH methods on a Micromeritics ASAP 2020M accelerated surface area and porosimetry system. A probe-type Ocean Optics HR2000CG-UV-NIR spectrometer was used to measure the UV/Vis absorption spectrum of the solutions to monitor the concentration of RhB at different time intervals. The concentration of the Fe3+ dissolved in the solution was determined by using a Perkin–Elmer inductively coupled plasma optical-emission spectrometer (ICP–OES) Optima 2000 DV. IR spectra were measured by using KBr pellets on a Spectrum One Perkin–Elmer FTIR spectrometer.

Photocatalytic Degradation of Rhodamine B: Photocatalytic reactions for degradation of RhB were carried out in a 100 mL beaker that contained the reaction slurry (50 mL). Agitation was provided by a magnetic stirrer. The aqueous slurry, prepared with different catalysts and RhB (1.0×10−5M) was stirred in the dark for 30 min to ensure that the RhB was adsorbed to saturation on the catalysts. A 15 W UV lamp (254 nm, Spectroline XX-15G, USA) that was placed 6 cm above the reaction slurry was used as the UV radiation source. The average light intensity striking the surface of the reaction solution was estimated to be 1.55 mW cm−2. In the recycling experiments, after the Fe3O4/SiO2/TiO2 composites were separated from solution by an external magnetic field, we removed the clear upper solution, and then redispersed the Fe3O4/SiO2/TiO2 in RhB solution (50 mL, 1.0×10−5M) for another cycle. To ensure the RhB was completely mineralized and had no influence on the next cycle, we increased the irradiation time to 1 h for each cycle. The photocatalytic efficiency of the Fe3O4/SiO2/TiO2 composites was compared to TiO2 solid particles and SiO2/TiO2 core–shell structures, which were prepared by using procedures similar to that for the TiO2 coating as described above but without adding seeds or by replacing the Fe3O4/SiO2 seeds with SiO2 spheres. The mass of the TiO2 layer was determined by measuring the weight difference before and after TiO2 coating, and the concentration of TiO2 in the reaction solution was about 500 mg L −1for all the runs.


Y.Y. thanks the University of California, Riverside for startup funds. Acknowledgement is made to the donors of the Petroleum Research Fund, administered by the American Chemical Society for support of this research. Financial support of this work was also provided by Basic Energy Sciences-U.S. DOE, SISGR-Catalysis for Energy grant No. DE-SC0002247. Y.Y. is a Cottrell Scholar of Research Corporation for Science Advancement. M.Y. thanks Dr. Xiaowei Liu from Harbin Institute of Technology for help with BET measurements and Ms. Jingjing Yao from UCR for help with ICP-OES measurements.