Recyclable magnetic CoFe₂O₄/BiOX (X = Cl, Br and I) microflowers for photocatalytic treatment of water contaminated with methyl orange, rhodamine B, methylene blue, and a mixed dye

Abstract

The recycling of photocatalysts and improving their activities by hybridizing two materials are important. Herein, nanosize ferromagnetic (M_s = 62.3 emu g⁻¹) CoFe₂O₄ nanoparticles (NPs) were embedded into nanosize-assembled BiOX (X = Cl, Br and I) microflowers and examined by scanning electron microscopy, transmission electron microscopy, X-ray diffraction, UV-visible absorption spectroscopy, Fourier-transform infrared spectroscopy, and photoluminescence spectroscopy. The adsorption and photocatalytic performance of CoFe₂O₄/BiOX for methyl orange (MO), rhodamine B (RhB), methylene blue (MB), and a mixed dye (MO + RhB + MB) were examined under UV and visible light irradiation. The adsorption capacity of CoFe₂O₄/BiOI for RhB was 160 mg g_cat⁻¹, which is significantly larger than the <5 mg g_cat⁻¹ obtained for CoFe₂O₄/BiOCl and CoFe₂O₄/BiOBr. The photocatalytic activity was observed in the order of CoFe₂O₄/BiOBr < CoFe₂O₄/BiOCl < CoFe₂O₄/BiOI for RhB. Their adsorption and photocatalytic performances were also investigated with pure MO, pure MB and a mixed dye. MO in the mixed dye was the most easily removed by the catalysts under light exposure. Based on the scavenger tests, h⁺ and ˙O₂⁻ play major and minor roles in the photodegradation of the dyes, respectively. Although the ˙OH radical was formed for CoFe₂O₄/BiOBr and CoFe₂O₄/BiOCl, it has a much smaller role than the other active species.

---

1. Introduction

The development of efficient photocatalysts and adsorbents has been a major research focus for energy and the environment. As an environmental issue, removing pollutants present in water has been a persistent goal. In recent years, bismuth oxyhalides, BiOX (X = Cl, Br and I), have exhibited good adsorption/photocatalytic performance and been studied extensively.^1–10 A major advantage of these materials is the ease of band gap tuning between 1.7 eV and 3.4 eV by simply changing the (relative amount of) halide ions.^11,12 More recently, hybrid materials with BiOX have been developed to further increase the photocatalytic activity. The enormous hybrid materials include phthalocyanine (CuPc)/BiOCl,¹³ Bi₂S₃/BiOCl,¹⁴ Bi/BiOBr_xI_1−x,¹⁵ BiOBr-g-C₃N₄,¹⁶ (Rh, Pd, Pt)/BiOX(Cl, Br, I),¹⁷ Ag–BiOBr,¹⁸ Ag/AgBr/BiOBr,¹⁹ Ag–BiOBr_xI_1−x,²⁰ Bi₂₄O₃₁Cl₁₀,²¹ Bi₂WO₆,²² BiOCl/Bi₂O₂CO₃,²³ BiOI/BiPO₄,²⁴ Ag₃PO₄/BiOI,²⁵ Bi/BiOI,²⁶ BiOI–MWCNT,²⁷ Ti–BiOBr,^28,29 and graphene–BiOBr.³⁰ The development of magnetic photocatalysts are of interest with the benefit of recyclability, even though in some cases, the catalytic activity is degraded after introducing magnetic materials into the host photocatalyst.^31–40 Magnetic SrFe₁₂O₁₉ was found to inhibit the growth of BiOCl along the [001] direction to expose the {001} facet. Furthermore, it induced visible light absorption (change in band gap from 3.3 eV to 2.8 eV after loading the magnetic material) and charge separation to increase photocatalytic MB degradation.³² Yao et al. prepared dandelion-like magnetic Fe₃O₄@C@BiOCl and Fe₃O₄@BiOCl composites. They reported that the surface area, adsorption capacity and photocatalytic activity were improved by introducing an interfacial carbon layer between Fe₃O₄ and BiOCl.³³ The total destruction of organic pollutants is important for eliminating pollution by secondary products. Zhang et al. solved this problem by synthesizing BiOBr@SiO₂@Fe₃O₄ microspheres through a multi-step synthesis process. They found only one main intermediate during the photocatalytic degradation of 2,2-bis(4-hydroxyphenyl)propane (BPA), where active h⁺ and ˙OH were mainly and partly involved in the direction oxidation, respectively.³⁴ For a photocatalytic reaction, the role of active species is very important and has been studied extensively using a number of methods, such as LC-MS,^34,41,42 scavenger tests³⁴ and DMPO (5,5-dimethyl-1-pyrroline N-oxide) electron paramagnetic resonance (EPR) spin trapping techniques.^21,31,43

A literature summary (ESI, Table S1†)^32–40 clearly showed the novelty of the present study. The previously reported BiOX-based magnetic photocatalysts include BiOCl–SrFe₁₂O₁₉,³² Fe₃O₄@C@BiOCl,³³ BiOBr@SiO₂@Fe₃O₄,³⁴ Fe₃O₄/BiOCl,^35,36 Fe₃O₄@SiO₂@BiOBr,³⁷ BiOBr/Fe₂O₃,³⁸ BiOBr–ZnFe₂O₄,³⁹ and Fe₃O₄/BiOI.⁴⁰ However, all the catalysts^{32,33,35–40} were only tested with pure MO, RhB and MB dyes. There are no reports on a more complicated dye system such as mixed dyes. Because a real polluted area is complicated it is necessary to test a more complicated simulated dye system as presented in this study. The catalysts in the literatures were prepared with only a single BiOX matrix (for example, BiOCl, BiOBr or BiOI) and mainly tested under visible light. However, we prepared magnetic catalysts with three BiOX (for example, BiOCl, BiOBr and BiOI) matrices and tested both under UV and visible light conditions for systematic comparison and analysis.

In the present study, p-type BiOX was first hybridized with n-type CoFe₂O₄ to prepare magnetic CoFe₂O₄/BiOX (X = Cl, Br and I) composites for the purpose of recycling a photocatalyst and a simulated real water treatment. The catalysts were fully tested with MO, RhB and MB and a mixed dye (MO + RhB + MB) under UV and visible light. Hybridization of two different materials is an important strategy to increase a performance.^{32–40,44–48} A deeper understanding of the different roles of the photocatalysts to a more complicated dye system will be very important for developing a photocatalyst applicable to a more complicated system.

2. Experimental section

2.1. Catalysts preparation

For the synthesis of CoFe₂O₄ nanoparticles (NPs), stoichiometric amounts (1 : 2 mole ratio) of Co nitrate and Fe nitrate were weighed and dissolved completely in ethylene glycol (EG) solvent. After heating the solution at 160 °C for 1 h, a 0.2 M Na₂CO₃ solution was added and the solution was stirred for 1 h. Subsequently, the solution was cooled naturally and the products were filtered. For the synthesis of CoFe₂O₄/BiOX (X = Cl, Br and I) composites, CoFe₂O₄ NPs was first dispersed in EG solvent and a stoichiometric amount of Bi(NO₃)₃·5H₂O was added. The mole ratio of CoFe₂O₄/Bi was fixed to 0.25. A stoichiometric amount of KX (X = Cl, Br and I) was then added. The Bi/X molar ratio was fixed to 1.0. The solution in a tightly capped Teflon bottle was placed at 120 °C in an oven for 5 h. The final products were washed several times with deionized water and ethanol, and fully dried in an oven (80 °C).

2.2. Catalysts characterization

The surface morphology of the hybrid samples was characterized by field-emission scanning electron microscopy (FE-SEM, Hitachi SE-4800). The crystal phase was analyzed by powder X-ray diffraction (XRD, PANalytical X'Pert Pro MPD) using Cu Kα radiation (λ = 0.154056 nm) operated at 40 kV and 30 mA. Transmission electron microscopy (TEM) was used to examine the microstructure of a selected sample placed on a carbon-coated Cu grid. The diffuse UV-Vis absorption spectra were used to measure the band gaps of the powder samples using a SCINCO NeoSys-2000 double beam UV-Vis spectrophotometer. The Fourier-transform infrared (FT-IR, Thermo Scientific Nicolet iS10) spectra were obtained in attenuated total reflection (ATR) mode. The photoluminescence (PL) spectra of the powder samples were obtained using a SCINCO FluoroMate FS-2. Magnetic properties of the powder samples were examined using a MPMS SQUID magnetometer (Quantum Design, USA) with an applied magnetic field ranging from −70 to 70 kOe at room temperature. The Brunauer–Emmett–Teller (BET) surface areas for the powder samples were measured using a ChemBET TPR/TPD analyzer (Quantachrome Instruments, USA) equipped with a thermal conductivity detector.

2.3. Adsorption, photocatalytic activity and active species

For the BiOCl and BiOBr catalysts, a 20 mg L⁻¹ RhB solution was used, while for BiOI, a 40 mg L⁻¹ dye solution was used. For a mixed dye to test the photocatalytic performance of the BiOCl and BiOBr catalysts, 10 mg L⁻¹ concentrations of MO, RhB and MB solutions were mixed. For a mixed dye to test the BiOI catalyst, 20 mg L⁻¹ MO, RhB and MB solutions were mixed. For adsorption, 25 mg of a catalyst was dispersed into a 100 mL dye (RhB or the mixed dye) solution. The solution was fully stirred in the dark for 1 hour. The photocatalytic experiments were performed under visible and UV light with different light exposure times. The concentration of the dye solution was monitored using a Jasco V-530 UV-Vis spectrophotometer. A scavenger test was used to examine the roles of active ˙OH and ˙O₂⁻ species. Isopropyl alcohol (IPA) and benzoquinone (BQ) were added to the test solution as ˙OH and ˙O₂⁻ scavengers, respectively.^49,50 To examine ˙OH radical formation over the catalysts, photoluminescence spectroscopy was performed using a terephthalic acid solution.³⁴ A catalyst was dispersed in the terephthalic acid solution and irradiated with UV or visible light for a specified time. The PL spectrum of the solution was taken at an excitation wavelength of 315 nm using a SCINCO FluoroMate FS-2 fluorometer.

3. Results and discussion

3.1. Morphologies

Fig. 1 presents SEM images of the CoFe₂O₄/BiOX nano-assembled microflowers prepared in EG. The SEM images commonly showed distorted spheres with sizes of 1–3 μm. The assembled average sizes were in the order of CoFe₂O₄/BiOCl < CoFe₂O₄/BiOI < CoFe₂O₄/BiOBr. The magnified SEM images showed that the spheres differed slightly according to halide ions used and consisted of nanosize structures. The reported surface areas of bare BiOCl, BiOBr and BiOI microflowers were 17 m² g⁻¹, 19 m² g⁻¹ and 61 m² g⁻¹, respectively (Table 1).^11,12 The similar surface areas for BiOCl and BiOBr may indicate that they had similar morphological structures as shown in Fig. 1. Upon loading CoFe₂O₄ NPs, we observed the BET surface areas of 32.7 m² g⁻¹, 30.9 m² g⁻¹ and 62.0 m² g⁻¹ for CoFe₂O₄/BiOCl, CoFe₂O₄/BiOBr and CoFe₂O₄/BiOI, respectively (Table 1). The surface areas of BiOCl and BiOBr were substantially increased while that of BiOI showed no significant change after loading CoFe₂O₄ NPs. The BiOCl and BiOBr with smaller surface areas could be increased a lot further by loading with smaller size CoFe₂O₄ NPs.

Fig. 1 SEM images of CoFe₂O₄/BiOX (X = Cl, Br and I) nano-assembled microflowers.

Table 1 BET surface areas (m² g⁻¹) of BiOX before and after loading CoFe₂O₄ NPs

Catalysts	BiOCl	BiOBr	BiOI	Ref.
Before loading	17	19	61	11 and 12
After loading NPs	32.7	30.9	62.0	This work

---

Fig. 2 shows TEM images of the selected CoFe₂O₄/BiOI microflowers and CoFe₂O₄ nanoparticles (NPs). The energy-dispersive X-ray (EDX) analysis confirmed the chemical species of the CoFe₂O₄ NPs and the hybrid materials (ESI, Fig. S1†). The magnetic CoFe₂O₄ NPs were very small with sizes of <5 nm. The 3D-microflowers were formed by the assembly of nanosize structures. The BiOX seeds were formed via the reaction, Bi³⁺ + X⁻ + H₂O → BiOX + 2H⁺. The nano-seeds assemble and grow into a larger sphere through Ostwald ripening.⁸ During this growth, the initially dispersed much smaller NPs (unreactive to BiOX) became embedded into the μm-size BiOI microflowers to form hybrid CoFe₂O₄/BiOI microflowers. Upon completion of the growth, BiOX microflowers generally showed {011} and {020} crystalline lattices, indicating higher percentage of {001} facets.^11,12 In the HRTEM image of Fig. 2, a clear lattice fringe with a spacing of 0.199 nm corresponds to the (020) plane of tetragonal BiOI (Scheme 1).¹²

Fig. 2 Typical TEM images of the CoFe₂O₄/BiOI microflowers and CoFe₂O₄ nanoparticles. Left top inset shows the HRTEM image of the edge of BiOI.

Scheme 1 Growth mechanism of the hybrid materials.

3.2. Crystal structures

Fig. 3 shows the powder XRD patterns of CoFe₂O₄ NPs and CoFe₂O₄/BiOX (X= Cl, Br and I) microflowers. The XRD patterns of pure CoFe₂O₄ NPs were relatively weaker than those of the microflowers due to the lower crystallinity arising from the extremely small NP size (<5 nm). The patterns matched those of cubic (Fd3̄m) CoFe₂O₄ (JCPDS 3-0864). The peaks at 2θ = 34.8° and 60.5° were assigned to the (311) and (440) planes of the crystal structure. The other peaks included the (220), (222), (400), (422), and (511) planes. For the CoFe₂O₄/BiOX (X = Cl, Br and I) microflowers, the XRD peaks of CoFe₂O₄ NPs could not be discriminated clearly because of the smaller hybrid amount in the BiOX matrices and the weak XRD peaks of pure CoFe₂O₄ NPs. The XRD patterns of CoFe₂O₄/BiOCl were similar to those of tetragonal (P4/nmm) BiOCl (JCPDS 1-085-0861). Three major peaks were observed at 2θ = 25.9°, 32.6° and 36.4°, which were assigned to the (101), (110) and (102) planes of the tetragonal phase, respectively. The XRD patterns of CoFe₂O₄/BiOBr matched those of tetragonal (P4/nmm) BiOBr (JCPDS 01-085-0862). Two major peaks at 2θ = 31.7° and 32.3° were assigned to the (102) and (110) planes, respectively. The XRD peak of the (101) plane was relatively weaker. For the CoFe₂O₄/BiOI microflowers, the XRD peaks matched those of tetragonal (P4/nmm) BiOI (JCPDS 1-073-2062). Two major peaks were observed at 2θ = 29.5° and 31.8° corresponding to the (012) and (110) planes, respectively.

Fig. 3 Power X-ray diffraction patterns of CoFe₂O₄ NPs, and CoFe₂O₄/BiOX (X = Cl, Br and I) microflowers.

3.3. Optical properties

Fig. 4 shows the UV-Vis diffuse reflectance (DR) absorption spectra of CoFe₂O₄/BiOX (X = Cl, Br and I) microflowers. The Y-axis (absorbance) was obtained from the DR data using the Kubelka–Munk method. The sharp absorption edges (not shown) for bare BiOCl, BiOBr and BiOI microflowers were observed at 3.2, 2.8 and 1.8 eV, respectively. Upon loading the CoFe₂O₄ NPs, the absorption in the visible region were increased substantially and the colors of the samples became darker. We could expect a change in color and an increase in visible light absorption upon introducing black color CoFe₂O₄ NPs with a band gap of 1.2 eV. An increase in visible light absorption facilitates photocatalytic activity under visible light.⁵¹ Although the visible light absorption of bare BiOCl and BiOBr microflowers was much lower than that of BiOI, the absorption became higher than BiOI after loading the CoFe₂O₄ NPs. Upon embedding CoFe₂O₄ NPs, the CoFe₂O₄/BiOX microflowers showed magnetic properties (as shown in the inset photos in Fig. 4).

Fig. 4 UV-visible diffuse reflectance absorption spectra of CoFe₂O₄/BiOX (X = Cl, Br and I) microflowers. The insets show photographs of the corresponding samples.

Fig. 5 shows the photoluminescence (PL) spectra of CoFe₂O₄/BiOX (X = Cl, Br and I) microflowers taken at an excitation wavelength of 250 nm. The origin of the PL is the recombination of an electron and a hole. A higher the PL intensity indicates more recombination. Higher recombination indicates poor charge separation, generally leading to lower catalytic activity.⁵² The PL spectra of the three different samples were quite different with an intensity in the order of CoFe₂O₄/BiOI < CoFe₂O₄/BiOBr < CoFe₂O₄/BiOCl. As the band gap is larger (or the excitation energy is closer to the CB of a material), the PL becomes stronger. In other words, the band gap edge shifts to a longer wavelength as the halide ion is changed from Cl, Br to I. As a consequence, the PL peak maximum is shifted to a longer wavelength. For the CoFe₂O₄/BiOCl microflowers, broad regions were observed between 320 nm and 600 nm. For CoFe₂O₄/BiOBr, a broad peak at 460 nm was dominant. This was attributed to defects. For CoFe₂O₄/BiOI, the intensity decreased dramatically. A broad maximum was observed at 600 nm.

Fig. 5 Photoluminescence spectra of CoFe₂O₄/BiOX (X = Cl, Br and I) microflowers at an excitation wavelength of 250 nm.

FT-IR spectra (ESI, Fig. S2†) were taken for CoFe₂O₄/BiOX (X = Cl, Br and I) and BiOX microflowers. Several peaks were observed and the profiles were similar before and after loading the CoFe₂O₄ NPs. A broad peak at 3400 cm⁻¹ and a sharper peak at 1600 cm⁻¹ were attributed to the stretching and deformation vibrations of the adsorbed OH group.³²

3.4. Magnetic properties

The magnetic properties of CoFe₂O₄ NPs and CoFe₂O₄/BiOX (X = Cl, Br and I) microflowers were examined by SQUID. Fig. 6 shows the magnetization (M–H) curves measured with magnetic fields from −70 to 70 kOe at room temperature. The M–H curves commonly showed saturation magnetization (M_s). A saturation magnetization of 0.0623 emu mg⁻¹ (or 62.3 emu g⁻¹) was measured for CoFe₂O₄ NPs. This value was in good agreement with the literature.⁴⁴ A magnetic hysteresis loop was observed (ESI, Fig. S3†), suggesting ferromagnetic behavior of the NPs. The hysteresis loop became significantly narrower for the hybrid materials. For CoFe₂O₄/BiOCl, CoFe₂O₄/BiOBr and CoFe₂O₄/BiOI microflowers, the magnetization values were decreased and measured to be 0.00747, 0.0067 and 0.00621 emu mg⁻¹ (or 7.47, 6.7 and 6.21 emu g⁻¹), respectively. The decrease in saturation magnetization is due to hybridization with non-magnetic BiOX (X = Cl, Br and I) materials.^44,47 Although the magnetization of the hybrid sample was decreased^44,47 the powder sample dispersed in a dye solution was easily (in a few seconds) attracted (or recycled) by a magnet (ESI, Fig. S4†).

Fig. 6 Magnetization (M–H) curves for CoFe₂O₄ NPs and CoFe₂O₄/BiOX (X = Cl, Br and I) microflowers with applied magnetic fields from −70 to 70 kOe at room temperature.

3.5. Adsorption and photocatalytic activity of CoFe₂O₄/BiOCl and CoFe₂O₄/BiOBr for pure RhB

The adsorption and photocatalytic activities of CoFe₂O₄/BiOCl microflowers for RhB were tested. Fig. 7 shows the UV-Vis absorption spectra of RhB tested with the catalysts under visible (Fig. 7A–C) and UV (Fig. 7A1, B1 and C1) light irradiation. The roles of active ˙O₂⁻ and ˙OH radical species were also examined using a scavenger test method. Benzoquinone (BQ) and isopropyl alcohol (IPA) were added as ˙O₂⁻ and ˙OH scavengers, respectively. RhB showed a broad UV-Vis absorption band at approximately 550 nm. The peak decreased after adsorption over the catalyst for 1 hour under dark conditions. Under visible light without a scavenger (Fig. 7A), with increasing light exposure time, the UV-Vis absorption peak decreased substantially and the peak position shifted gradually to a shorter wavelength of 490 nm. The pink color of the original solution changed to a green color. This indicates that a secondary product had formed during the photocatalytic reaction of RhB over the catalyst. After visible light irradiation for 6 h, >95% of the RhB was degraded. The blue-shift in the peak for RhB is generally attributed to de-ethylation products by ˙O₂⁻ radical species.^12,20,38 To confirm this, benzoquinone (BQ) was added as a ˙O₂⁻ radical scavenger and the photocatalytic experiments were performed (Fig. 7B). Although the peak intensity decreased with increasing light exposure time, the peak position showed no change. Based on the results, it was concluded that the ˙O₂⁻ radical was involved in the photocatalytic reaction, but it did not affect the dye degradation greatly. To test the role of ˙OH radical species, IPA was added to the catalyst solution during photoirradiation (Fig. 7C). The UV-Vis absorption peak decreased with increasing irradiation time, and changed its peak position. After adding IPA, the degradation rate was increased slightly due to the greater dispersion of the catalysts. The scavenger did not negate the photodegradation rate. This suggests that the ˙OH radical was not involved in the photocatalytic reaction. Li et al. also obtained a similar result for RhB over Fe₃O₄/BiOI and reported that h⁺ and ˙O₂⁻ were the major active species while ˙OH was a minor species for dye degradation.⁴⁰ Under UV irradiation, the degradation rate decreased and no shift in the peak position was observed (Fig. 7A1). This suggests that the ˙O₂⁻ radical was not involved significantly in the photocatalytic reaction (Fig. 7B1). After adding the ˙OH radical scavenger (Fig. 7C1), IPA, a shift in the peak position under UV light irradiation with time was observed.

Fig. 7 Adsorption (in dark) and photodegradation (under UV and visible lights) tests of RhB (20 mg L⁻¹, 100 mL) over 25 mg CoFe₂O₄/BiOCl microflowers without (A and A1) and with scavengers of BQ (B and B1) and IPA (C and C1). The insets show the corresponding photographs displaying a change in the dye color with photodegradation time.

The adsorption and photocatalytic activity of the CoFe₂O₄/BiOBr microflowers for RhB were examined by UV-Vis absorption spectroscopy (ESI, Fig. S5†). As discussed above, the catalysts were tested with and without active species scavengers under UV and visible light irradiation. The behaviors of the UV-Vis absorption spectra were similar to those of the CoFe₂O₄/BiOCl microflowers. The adsorption performance of the CoFe₂O₄/BiOBr microflowers was somewhat lower than that of CoFe₂O₄/BiOCl microflowers. In addition, the photocatalytic activity was also lower than that of the CoFe₂O₄/BiOCl microflowers. After adding BQ and IPA, similar results to those for the CoFe₂O₄/BiOCl microflowers were obtained.

3.6. Adsorption and photocatalytic activity of CoFe₂O₄/BiOI for pure RhB

Fig. 8 displays the UV-Vis absorption spectra of RhB solution (40 mg L⁻¹) over CoFe₂O₄/BiOI microflowers upon adsorption and with UV (Fig. 8A) and visible (Fig. 8B) according to the light exposure time. The adsorption capacity of the catalyst for RhB was extremely high (160 mg g_cat⁻¹). The high adsorption performance was attributed to a high surface area and electrostatic interactions between the catalyst and RhB.^11,12,53 The negatively charged CoFe₂O₄/BiOI surface may interact strongly with the positively charged RhB.^11,12 Because of the high adsorption capacity, visible light is absorbed mainly by the uppermost adsorbed RhB. The UV absorption by the catalyst is also protected by the adsorbed RhB. Because of the UV protection, the adsorbed dye was photodegraded slowly (Fig. 8A). As the visible light exposure time was increased, the extremely weak peak was shifted to a shorter wavelength (Fig. 8B). This suggests that the RhB was de-ethylated by active ˙O₂⁻ species. Under UV light (Fig. 8A), the shift in the peak position was much slower, indicating dull photodegradation. The catalyst sample after the UV light experiments was much darker than that after the visible light experiments.

Fig. 8 Adsorption (in dark) and photocatalytic dye degradation ((A) under UV light, (B) under visible light) of Rh B (40 mg L⁻¹, 100 mL) over 25 mg CoFe₂O₄/BiOI microflowers. The insets show photographs of the corresponding solutions showing a change in dye color with photoirradiation time. The weak peaks were multiplied by 5×.

3.7. Adsorption and photocatalytic activity of CoFe₂O₄/BiOX for pure MO and MB

We tested the adsorption and photocatalytic performances with pure MO and MB for comparison (ESI, Fig. S6†). Fig. 9 displays adsorption (C/C_o) photocatalytic dye degradation (C_6h/C_Ad) performances for pure MO and MB over the CoFe₂O₄/BiOX (X = Cl, Br, I) microflowers under visible light irradiation. For pure MO, the adsorption and photocatalytic degradation performances showed an order of CoFe₂O₄/BiOCl < CoFe₂O₄/BiOBr < CoFe₂O₄/BiOI. For pure MB, the adsorption and photocatalytic degradation performances became an order of CoFe₂O₄/BiOBr < CoFe₂O₄/BiOCl < CoFe₂O₄/BiOI. Because the surface area of CoFe₂O₄/BiOI was larger than those of CoFe₂O₄/BiOCl and CoFe₂O₄/BiOBr (Table 1) it was expected that the adsorption and photocatalytic degradation performances were higher for the CoFe₂O₄/BiOI microflowers. For MO, the adsorption and photocatalytic degradation performances of CoFe₂O₄/BiOCl was poor than those of CoFe₂O₄/BiOBr although the surface area of CoFe₂O₄/BiOCl was somewhat larger than that of CoFe₂O₄/BiOBr. This reflects that other factors such as surface interactions play also important role in the adsorption and photocatalytic dye degradation.^11,12,53 The adsorption capacities of CoFe₂O₄/BiOI for MO and MB were measured to be 24 and 21 mg g_cat⁻¹, respectively. This value was significantly smaller than that (160 mg g_cat⁻¹) for RhB, as mentioned above.

Fig. 9 Adsorption (C/C_o) and photocatalytic dye degradation (C_6h/C_Ad) performances for pure MO and MB over the CoFe₂O₄/BiOX (X = Cl, Br, I) microflowers. C_Ad and C_6h are the UV-Visible absorption intensities after adsorption and 6 h visible light photoirradiation.

3.8. Adsorption and photocatalytic activity of CoFe₂O₄/BiOX for a mixed dye

The adsorption and photocatalytic performances of the CoFe₂O₄/BiOCl (Fig. 10A and A1), CoFe₂O₄/BiOBr (Fig. 10B and B1), CoFe₂O₄/BiOI (Fig. 10C and C1) microflowers for a mixed dye (MO + RhB + MB) were tested under UV (Fig. 10A–C) and visible (Fig. 10A1, B1 and C1) light and the results are displayed in Fig. 10. The mixed dye showed the absorptions of MO, RhB and MB as expected. The three peak positions of 450, 550 and 650 nm correspond to the MO, RhB and MB dyes, respectively. For the analysis, the three peak positions were selected and the heights of the peaks were examined. Upon adsorption of the mixed under the dark conditions, the UV-Vis absorption peak decreased. The adsorption performance of the mixed dye showed an order of CoFe₂O₄/BiOBr < CoFe₂O₄/BiOCl ≪ CoFe₂O₄/BiOI. On the other hand, the three peaks over the catalyst were not evenly decreased. Fig. 11 shows the adsorption performance of the CoFe₂O₄/BiOX (X = Cl, Br and I) microflowers. In the mixed dye, MO showed the best adsorption over the CoFe₂O₄/BiOCl (Fig. 11A) and CoFe₂O₄/BiOBr (Fig. 11B) microflowers, whereas MB showed the best adsorption over the CoFe₂O₄/BiOI microflowers (Fig. 11C). MB was almost completely removed by the CoFe₂O₄/BiOI microflowers. In contrast, RhB was the most poorly adsorbed on the CoFe₂O₄/BiOX microflowers. Because of the relative difference in the adsorption performances over the catalysts, the mixed solutions tested with CoFe₂O₄/BiOBr and CoFe₂O₄/BiOCl became violet while that with CoFe₂O₄/BiOI was pink after the adsorption in the dark condition.

Fig. 10 Adsorption (in dark) and photodegradation (under UV and visible lights) tests of the mixed dye (MO + RhB + MB) over 25 mg CoFe₂O₄/BiOX (X = Cl, Br and I) microflowers. For CoFe₂O₄/BiOCl (A and A1) and CoFe₂O₄/BiOBr (B and B1), a dye mixed with the three dyes with a concentration of 10 mg L⁻¹ (=ppm) was used. For CoFe₂O₄/BiOI (C and C1), a dye mixed with the three dyes with a concentration of 20 mg L⁻¹ was used. The insets show photographs of the corresponding dye solutions as a function of the photodegradation time. The weak peaks were multiplied by 15.

Fig. 11 Adsorption performance of the mixed dye (MO + RhB + MB) over the CoFe₂O₄/BiOCl (A), CoFe₂O₄/BiOBr (B) and CoFe₂O₄/BiOI (C) microflowers.

Fig. 12 displays the photodegradation rates of a mixed dye over CoFe₂O₄/BiOX (Cl, Br, I) microflowers based on Fig. 10. As the UV (Fig. 12A–C) and visible (Fig. 12A1, B1 and C1) light exposure times were increased, the UV-Vis absorption peaks were decreased. The catalytic activity for MO degradation under UV light (Fig. 12A) was observed in the order of CoFe₂O₄/BiOI < CoFe₂O₄/BiOCl < CoFe₂O₄/BiOBr. For RhB (Fig. 12B) and MB (Fig. 12C), the catalytic activities were similar. For MO in the mixed dye under visible light (Fig. 12A1), the catalytic activity showed a same order of CoFe₂O₄/BiOI < CoFe₂O₄/BiOCl < CoFe₂O₄/BiOBr. However, for RhB (Fig. 12B1) and MB (Fig. 12C1) under visible light irradiation, the order became CoFe₂O₄/BiOCl ≤ CoFe₂O₄/BiOBr < CoFe₂O₄/BiOI. Based on the degradation rates of MO, RhB and MB over the photocatalysts in the mixed dye, MO was degraded most rapidly under visible light irradiation. The order of catalytic activity can be explained as follows. The catalytic activity showed no linear relationship with the BET surface area (Table 1). This indicates that other factors (e.g. exposed crystal facets, surface interactions, charge transfer, and energy level alignment) play more important role in the dye degradation than the surface area. It was reported that exposed crystal facets showed different catalytic reactivity for UV and visible light because the wavelength of incident light determines charge transfer direction. A charge transfer rate is very important for photocatalytic dye degradation.^11,12 Different molecules are differently interacted with the different crystal facets and the energy levels (lowest unoccupied/highest occupied molecular orbitals and conduction/valence bands) are differently realigned. The charge transfer rate is promoted by a well-aligned energy level (or a good interfacial wavefunction mixing).

Fig. 12 Photodegradation rates of MO (A and A1), RhB (B and B1) and MB (C and C1) dyes in the mixed dye over CoFe₂O₄/BiOX (Cl, Br, I) microflowers under UV (top: A, B and C) and visible (bottom: A1, B1 and C1) light.

3.9. ˙OH radical formation probed by PL

The formation of ˙OH radicals over the catalysts during photoirradiation was probed by photoluminescence (PL) spectroscopy.³⁴ A catalyst is dispersed in a terephthalic acid (TA) solution and irradiated with UV (or visible) light while stirring. When a ˙OH radical is formed, TA reacts with the radical to form 2-hydroxyterephthalic acid, which emits a blue emission signal at approximately 425 nm. The intensity of the emission peak reflects the amounts of ˙OH radical produced during photoirradiation. Fig. 13 shows the photoluminescence spectra taken after UV and visible light irradiation for 6 h. For CoFe₂O₄/BiOCl and CoFe₂O₄/BiOBr, the photoluminescence signal was increased significantly under UV irradiation, while the intensity under visible light was observed to be weaker. Zhang et al. also reported ˙OH radical formation under visible light for BiOBr@SiO₂@Fe₃O₄ microflowers.³⁴ For CoFe₂O₄/BiOI, no photoluminescence signal was observed under UV and visible light irradiation, indicating no ˙OH radical formation. Under UV light, the PL intensity for CoFe₂O₄/BiOBr was stronger than that for CoFe₂O₄/BiOCl. Under visible light irradiation, the intensities became similar.

Fig. 13 Photoluminescence spectra of a terephthalic acid solution over the CoFe₂O₄/BiOX (X = Cl, Br and I) microflowers after UV (left) and visible (right) light irradiation for 6 h.

3.10. Photocatalytic mechanism and reliability/stability

The following photocatalytic degradation mechanism (as shown in Fig. 14) can be proposed based on the present results and the literature.^{41,49–51,54–63} Three active species of hydroxyl radicals (˙OH), superoxide radicals (˙O₂⁻) and holes (h⁺) act as the major species for photocatalytic dye degradation.

Fig. 14 Photocatalytic dye degradation mechanism of a dye over the CoFe₂O₄/BiOX (X = Cl, Br and I) microflowers. The red and blue vertical arrows indicate direct UV and visible light absorptions, respectively.

Under UV or visible light irradiation:

Mixed and RhB dyes + visible light → Dye (e_CB⁻ + h_VB⁺)

BiOBr (partly) and BiOI + visible light → BiOI and BiOBr (e_CB⁻ + h_VB⁺)

CoFe₂O₄ + UV and visible lights → CoFe₂O₄ (e_CB⁻ + h_VB⁺)

CoFe₂O₄/BiOX + UV light → CoFe₂O₄/BiOX (e_CB⁻ + h_VB⁺)

Dye (e_CB⁻ + h_VB⁺) + CoFe₂O₄/BiOX → CoFe₂O₄/BiOX (e_CB⁻) + dye (h_VB⁺)

CoFe₂O₄ (e_CB⁻ + h_VB⁺) + BiOX → BiOX (e_CB⁻) + CoFe₂O₄ (h_VB⁺)

CoFe₂O₄/BiOX (e_CB⁻) + O₂ → ˙O₂⁻ + CoFe₂O₄/BiOX

CoFe₂O₄/BiOX (e_CB⁻) + ˙O₂⁻ + 2H⁺ → 2˙OH

CoFe₂O₄/BiOX (h_VB⁺) + H₂O → H⁺ + ˙OH + CoFe₂O₄/BiOX

Dye + ˙O₂⁻, h⁺ or ˙OH + dye → degradation

Under visible light irradiation, the dye and the catalyst (with a lower band gap) absorb light to create an electron and a hole in the conduction and valence bands (CB and VB) of the materials, respectively. BiOI (E_g = 1.8 eV) and BiOBr (E_g = 2.8 eV) strongly and partly absorb visible light, respectively. BiOCl (3.2 eV) does not absorb visible light. The electron in the CB of the dye transfers to that of the BiOX material. This is often called a dye-sensitized mechanism. CoFe₂O₄ (E_g = 1.2 eV) also absorbs visible light well,⁶⁴ and electron in the CB of the CoFe₂O₄ transfers to that of BiOX. The electron in the CB is captured by adsorbed oxygen (an electron acceptor) to produce active ˙O₂⁻ species. Active ˙OH radicals can be produced by the reactions of ˙O₂⁻ species and electron (e_CB⁻) or water and hole (h_VB⁺). The active ˙O₂⁻, h⁺ and ˙OH species will consequently degrade the adsorbed dye. The CoFe₂O₄/BiOI microflowers showed no creation of ˙OH radical species while CoFe₂O₄/BiOCl and CoFe₂O₄/BiOBr formed some ˙OH radical species. Under visible light, the amount formed was very small compared to that under visible light. Based on the results discussed above, the role of the ˙OH radical was much weaker than that of the other active species. The role of ˙O₂⁻ species depends on the amount of oxygen dissolved in the dye solution.⁶⁵ In the present study, ˙O₂⁻ species were weakly involved in the photocatalytic reaction. Consequently, h⁺ plays a major role in dye degradation, as reported for the Fe₃O₄/BiOI catalyst.⁴⁰ Under UV irradiation, the catalysts mainly absorb incident UV light and directly create an electron and hole in the CB and VB of the CoFe₂O₄/BiOX catalysts, respectively. The electrons in the CBs of the catalysts commonly initiate the creation of active species. Therefore, the catalytic activities were not much different than those found under visible light, as shown in Fig. 11.

Finally, we tested reliability and stability of the catalyst samples after the photocatalytic dye degradation experiments. We found that the crystal structure of the catalyst samples showed no change after photocatalytic dye degradation experiments. This indicates that the catalysts were very stable based on the XRD patterns (ESI, Fig. S7†). We tested the reusability of the catalysts samples, and found that about 80% of the initial performance was obtained in the second cycle. It took more time to completely photo-dissociate dye in a solution (ESI, Fig. S8†). This is due to some aggregation of the magnetic catalyst particles. However, in the third cycle the catalytic activity showed no significant decrease compared with the second cycle.

4. Conclusion

Highly efficient recyclable magnetic CoFe₂O₄/BiOX (X = Cl, Br and I) microflower photocatalysts were prepared by a solvothermal method and tested with a mixed dye (MO + RhB + MB) and pure MO, RhB and MB. The adsorption performance and photocatalytic activity of a photocatalyst for a complicated polluted system were examined. Several new important results are listed below.

(1) A saturation magnetization of 62.3 emu g⁻¹ was measured for CoFe₂O₄ NPs. The magnetization values were decreased by 8–10× upon hybridization.

(2) For the adsorption of RhB, the adsorption performance showed an order of CoFe₂O₄/BiOBr < CoFe₂O₄/BiOCl ≪ CoFe₂O₄/BiOI. The adsorption capacity of CoFe₂O₄/BiOI was 160 mg g_cat⁻¹, which is much larger than <5 mg g_cat⁻¹ for CoFe₂O₄/BiOCl and CoFe₂O₄/BiOBr.

(3) For the photocatalytic activity of pure RhB over CoFe₂O₄/BiOX microflowers, the photocatalytic activity showed an order of CoFe₂O₄/BiOBr < CoFe₂O₄/BiOCl < CoFe₂O₄/BiOI.

(4) The adsorption capacities of CoFe₂O₄/BiOI for MO and MB were 24 and 21 mg g_cat⁻¹, respectively. For pure MO and MB, the adsorption and photocatalytic degradation performances showed orders of CoFe₂O₄/BiOCl < CoFe₂O₄/BiOBr < CoFe₂O₄/BiOI and CoFe₂O₄/BiOBr < CoFe₂O₄/BiOCl < CoFe₂O₄/BiOI, respectively.

(5) The adsorptions in the mixed dye were different. In the mixed dye, MB showed the best adsorption over the CoFe₂O₄/BiOI microflowers while MO showed the best adsorption over the CoFe₂O₄/BiOCl and CoFe₂O₄/BiOBr microflowers. On the other hand, RhB was the most poorly adsorbed on the CoFe₂O₄/BiOX microflowers.

(6) The degradation rate of each dye in the mixed system was different. The MO dye was removed most easily by the catalysts under UV and visible light exposure. CoFe₂O₄/BiOBr showed the highest photocatalytic activity for MO degradation while CoFe₂O₄/BiOI showed the highest activity for RhB and MB degradation.

(7) For CoFe₂O₄/BiOX microflowers, h⁺ and ˙O₂⁻ play major and minor roles in photodegradation of the dyes, respectively. Although the ˙OH radical was formed for CoFe₂O₄/BiOBr and CoFe₂O₄/BiOCl under visible light, the role was found to be negligible. No ˙OH radical was formed for CoFe₂O₄/BiOI both under UV and visible light irradiations.

The newly developed magnetic photocatalysts provide new insights into the design of recyclable adsorbents and photocatalysts for a more complicated polluted system.

Acknowledgements

This study was supported financially by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MEST) (NRF-2014R1A1A2055923).

References

1. L. Ye, Y. Su, X. Jin, H. Xie and C. Zhang, Environ. Sci.: Nano, 2014, 1, 90–112.
2. H. Cheng, B. Huang and Y. Dai, Nanoscale, 2014, 6, 2009–2026.
3. H. Zhao, F. Tian, R. Wang and R. Chen, Reviews in Advanced Sciences and Engineering, 2014, 3, 3–27.
4. J. Li, Y. Yu and L. Zhang, Nanoscale, 2014, 6, 8473–8488.
5. L. Zhang, W. Wang, S. Sun, Y. Sun, E. Gao and J. Xu, Appl. Catal., B, 2013, 132–133, 315–320.
6. K. Li, Y. Tang, Y. Xu, Y. Wang, Y. Huo, H. Li and J. Jia, Appl. Catal., B, 2013, 140–141, 179–188.
7. J. Xiong, G. Cheng, F. Qin, R. Wang, H. Sun and R. Chen, Chem. Eng. J., 2013, 220, 228–236.
8. J. Hu, S. Weng, Z. Zheng, Z. Pei, M. Huang and P. Liu, J. Hazard. Mater., 2014, 264, 293–302.
9. S. Wang, L. Wang, W. Ma, D. Johnson, Y. Fang, M. Jia and Y. Huang, Chem. Eng. J., 2015, 259, 410–416.
10. L. Ai, Y. Zeng and J. Jiang, Chem. Eng. J., 2014, 235, 331–339.
11. W. Kim, D. Pradhan, B. Min and Y. Sohn, Appl. Catal., B, 2014, 147, 711–725.
12. Y. Park, Y. Na, D. Pradhan, B. Min and Y. Sohn, CrystEngComm, 2014, 16, 3155–3167.
13. X. Zhang, X. Liu, C. Fan, Y. Wang, Y. Wang and Z. Liang, Appl. Catal., B, 2013, 132–133, 332–341.
14. H. Cheng, B. Huang, X. Qin, X. Zhang and Y. Dai, Chem. Commun., 2012, 48, 97–99.
15. C. Zheng, C. Cao and Z. Ali, Phys. Chem. Chem. Phys., 2015, 17, 13347–13354.
16. L. Ye, J. Liu, Z. Jiang, T. Peng and L. Zan, Appl. Catal., B, 2013, 142–143, 1–7.
17. C. Yu, F. Cao, G. Li, R. Wei, J. Yu, R. Jin, Q. Fan and C. Wang, Sep. Sci. Technol., 2013, 120, 110–122.
18. L. Lu, L. Kong, Z. Jiang, H. Lai, T. Xiao and P. Edwards, Catal. Lett., 2012, 142, 771–778.
19. T. Li, S. Luo and L. Yang, J. Solid State Chem., 2013, 206, 308–316.
20. Y. Lei, G. Wang, P. Guo and H. Song, Appl. Surf. Sci., 2013, 279, 374–379.
21. F. Li, Q. Wang, X. Wang, B. Li, Y. Hao, R. Liu and D. Zhao, Appl. Catal., B, 2014, 150–151, 574–584.
22. H. Li, Y. Cui and W. Hong, Appl. Surf. Sci., 2013, 264, 581–588.
23. X. Zhang, T. Guo, X. Wang, Y. Wang, C. Fan and H. Zhang, Appl. Catal., B, 2014, 150–151, 486–495.
24. J. Cao, B. Xu, H. Lin and S. Chen, Chem. Eng. J., 2013, 228, 482–488.
25. Z. Cui, M. Si, Z. Zheng, L. Mi, W. Fa and H. Jia, Catal. Commun., 2013, 42, 121–124.
26. C. Chang, L. Zhu, Y. Fu and X. Chu, Chem. Eng. J., 2013, 233, 305–314.
27. M. Su, C. He, L. Zhu, Z. Sun, C. Shan, Q. Zhang, D. Shu, R. Qiu and Y. Xiong, J. Hazard. Mater., 2012, 229–230, 72–82.
28. X. Wei, H. Cui, S. Guo, L. Zhao and W. Li, J. Hazard. Mater., 2013, 263, 650–658.
29. R. Wang, G. Jiang, X. Wang, R. Hu, X. Xi, S. Bao, Y. Zhou, T. Tong, S. Wang, T. Wang and W. Chen, Powder Technol., 2012, 228, 258–263.
30. X. Zhang, X. Chang, M. Gondal, B. Zhang, Y. Liu and G. Ji, Appl. Surf. Sci., 2012, 258, 7826–7832.
31. Z. Wu, W. Li, P. A. Webley and D. Zhao, Adv. Mater., 2012, 24, 485–491.
32. T. Xie, L. Xu, C. Liu, J. Yang and M. Wang, Dalton Trans., 2014, 43, 2211–2220.
33. Y. Yao, W. Huang, H. Zhou, Y. Zheng and X. Song, J. Nanopart. Res., 2014, 16, 2451–2459.
34. L. Zhang, W. Wang, S. Sun, Y. Sun, E. Gao and Z. Zhang, Appl. Catal., B, 2014, 148–149, 164–169.
35. L. Zhang, W. Wang, L. Zhou, M. Shang and S. Sun, Appl. Catal., B, 2009, 90, 458–462.
36. C. Tan, G. Zhu, M. Hojamberdiev, C. Xu, J. Liang, P. Luo and Y. Liu, J. Cluster Sci., 2013, 24, 1115–1126.
37. Y. Yao, W. Huang, H. Zhou, H. Yin, Y. Zheng and X. Song, Mater. Chem. Phys., 2014, 148, 896–902.
38. C. Guo, Y. He, P. Du, X. Zhao, J. Lv, W. Meng, Y. Zhang and J. Xu, Appl. Surf. Sci., 2014, 320, 383–390.
39. L. Kong, Z. Jiang, T. Xiao, L. Lu, M. Jones and P. Edwards, Chem. Commun., 2011, 47, 5512–5514.
40. X. Li, C. Niu, D. Huang, X. Wang, X. Zhang, G. Zeng and Q. Niu, Appl. Surf. Sci., 2013, 286, 40–46.
41. J. Xu, W. Meng, Y. Zhang, L. Li and C. Guo, Appl. Catal., B, 2011, 107, 355–362.
42. J. Xu, L. Li, C. Guo, Y. Zhang and W. Meng, Appl. Catal., B, 2013, 130–131, 285–292.
43. X. Wang, W. Yang, F. Li, J. Zhao, R. Liu, S. Liu and B. Li, J. Hazard. Mater., 2015, 292, 126–136.
44. M. Kooti and E. Nasiri, J. Mol. Catal. A: Chem., 2015, 406, 168–177.
45. Z. Cai, Y. J. Liu, X. Lu and J. Teng, J. Phys. Chem. C, 2013, 117, 9440–9445.
46. Z. Cai, J. Teng, Z. Xiong, Y. Li, Q. Li, X. Lu and X. S. Zhao, Langmuir, 2011, 27, 5157–5164.
47. B. Tian, T. Wang, R. Dong, S. Bao, F. Yang and J. Zhang, Appl. Catal., B, 2014, 147, 22–28.
48. Z. Cai, Z. Xiong, X. Lu and J. Teng, J. Mater. Chem. A, 2014, 2, 545–553.
49. J. Cao, B. Xu, H. Lin, B. Luo and S. Chen, Chem. Eng. J., 2012, 185–186, 91–99.
50. M. Yin, Z. Li, J. Kou and Z. Zou, Environ. Sci. Technol., 2009, 43, 8361–8366.
51. X. Qin, H. Cheng, W. Wang, B. Huang, X. Zhang and Y. Dai, Mater. Lett., 2013, 100, 285–288.
52. J. Jiang, K. Zhao, X. Xiao and L. Zhang, J. Am. Chem. Soc., 2012, 134, 4473–4476.
53. D. Zhang, J. Li, Q. Wang and Q. Wu, J. Mater. Chem. A, 2013, 1, 8622–8629.
54. Y. Li, J. Wang, H. Yao, L. Dang and Z. Li, J. Mol. Catal. A: Chem., 2011, 334, 116–122.
55. K. Vignesha, A. Suganthi, B. Min and M. Kang, Appl. Surf. Sci., 2015, 324, 652–661.
56. C. Yu, F. Cao, G. Li, R. Wei, J. Yu, R. Jin, Q. Fan and C. Wang, Sep. Purif. Technol., 2013, 120, 110–122.
57. Y. I. Choi, H. J. Jung, W. G. Shin and Y. Sohn, Appl. Surf. Sci., 2015, 356, 615–625.
58. M. J. Kim, Y. I. Choi, S. W. Joo, M. Kang and Y. Sohn, Ceram. Int., 2014, 40, 16051–16059.
59. Y. Na, Y. I. Kim, D. W. Cho, D. Pradhan and Y. Sohn, Mater. Sci. Semicond. Process., 2014, 27, 181–190.
60. N. Roy, Y. Park, Y. Sohn, K. T. Leung and D. Pradhan, ACS Appl. Mater. Interfaces, 2014, 6, 16498–16507.
61. Y. Na, S. W. Lee, N. Roy, D. Pradhan and Y. Sohn, CrystEngComm, 2014, 16, 8546–8554.
62. A. K. Nayak, S. Lee, Y. Sohn and D. Pradhan, CrystEngComm, 2014, 16, 8064–8072.
63. M. M. Khan, S. A. Ansari, M. O. Ansari, B. K. Min, J. Lee and M. H. Cho, J. Phys. Chem. C, 2014, 118, 9477–9484.
64. B. Holinsworth, D. Mazumdar, H. Sims, Q. Sun, M. Yurtisigi, S. Sarker, A. Gupta, W. Butler and J. Musfeldt, Appl. Phys. Lett., 2013, 103, 082406–082409.
65. L. Ye, L. Tian, T. Peng and L. Zan, J. Mater. Chem., 2011, 21, 12479–12484.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra17616f

---