Controlled Synthesis of Hollow α-Fe2O3 Microspheres Assembled With Ionic Liquid for Enhanced Visible-Light Photocatalytic Activity

Porous self-assembled α-Fe2O3 hollow microspheres were fabricated via an ionic liquid-assisted solvothermal reaction and sequential calcinations. The concentration of the ionic liquid (1-butyl-3-methylimidazolium tetrafluoroborate [C4Mim]BF4) was found to play a crucial role in the control of these α-Fe2O3 hollow structures. Trace amounts ionic liquid was used as the soft template to synthesize α-Fe2O3 hollow spheres with a large specific surface (up to 220 m2/g). Based on time-dependent experiments, the proposed formation mechanisms were presented. Under UV light irradiation, the as-synthesized α-Fe2O3 hollow spheres exhibited excellent photocatalysis in Rhodamine B (RhB) photodegradation and the rate constant was 2–3 times higher than α-Fe2O3 particles. The magnetic properties of α-Fe2O3 hollow structures were found to be closely associated with the shape anisotropy.


INTRODUCTION
Nanostructured oxides with a variety of useful functionalities are widely applied in photocatalysis, energy storage, etc. (Guo et al., 2017;Li et al., 2017;Zhao et al., 2017Zhao et al., , 2018. In particular, hematite (α-Fe 2 O 3 ), an environmentally-friendly magnetic photocatalyst (E g = 2.2 eV), has been identified as an important material because of its potential for a wide range of practical applications (Brezesinski et al., 2011;Yang et al., 2014;Zhou et al., 2014;Ma et al., 2018). Recently, selfassembled hematite with highly specific hollow nano/micro-structures and unique properties has emerged as being of great interest to material scientists. To date, various self-assembled hollow α-Fe 2 O 3 nano/micro-structures have been prepared by different synthetic techniques, including the two-step reaction process , hydrothermal (Xu et al., 2012), and solvothermal approaches (Song et al., 2012;Zhu et al., 2012), thermal oxidation at high temperature (Xie et al., 2010), etc. For example, Yu and co-workers successfully fabricated cage-like Fe 2 O 3 hollow spheres and carbonaceous polysaccharide spheres were used as porous crystalline shell templates, followed by calcination at 500 • C for 4 h. Song and co-workers proposed a hydrolysis route to synthesis Fe 2 O 3 hierarchical hollow spheres, via a sodium dodecyl benzenesulfonate (SDBS)-assisted hydrolysis process. However, these template methods often suffer from a problem: it is difficult to remove the template and surface-active agent completely. Furthermore, template removal always leads to additional problems in sample morphology. Therefore, it is still necessary to develop a simple route to synthesizing assembled hollow α-Fe 2 O 3 structures.
Ionic liquids (ILs), or "designer liquids, " are known for their superior properties such as recyclability and stable chemical properties. They are used extensively in catalysis, separations, electrochemistry, and especially nanochemistry (Welton, 1999;Wasserscheid and Keim, 2000;Seddon, 2003). The most important advantage of ILs is that they can form extended hydrogen bond systems in the liquid, thereby enabling highlystructured nanostructures (Mele et al., 2003). This special performance can be used as the "entropic driver" for synthesis of self-assembled nanostructures. It has been proved that ILs can be used not only as solvents, but also as templates for preparing nanomaterials with improved properties (Endres et al., 2003;Cooper et al., 2004;Liu et al., 2006). For instance, a variety of nano/micro-structures have been synthesized via ILs, especially assembled hollow structures such as Bi 4 O 5 Br 2 and BiOBr spheres (Xia et al., 2011a;Mao et al., 2017) and α-Fe 2 O 3 hollow polyhedral (Xu et al., 2013). These hollow structures may also exhibit unique properties for photocatalysis.
In this work, we report a simple and feasible ionic liquidassisted solvothermal method to synthesize self-assembled α-Fe 2 O 3 hollow microspheres. The ionic liquid [C 4 Mim]BF 4 acts as a soft template can be easily removed by annealing in air. Compared with α-Fe 2 O 3 particles, the as-synthesized α-Fe 2 O 3 hollow spheres exhibit enhanced photocatalytic activity due to their porous self-assembly structure and higher surface area. As far as we known, α-Fe 2 O 3 porous hollow microspheres with high surface area (above 200 m 2 /g) prepared via this ionic liquidassisted solvothermal synthesis route is reported for the first time. Additionally, the possible formation mechanisms have also been proved. The as-prepared α-Fe 2 O 3 samples exhibit ferromagnetic properties and can be recycled easily.

Materials
Ferric chloride (FeCl 3 ·6H 2 O, 99%) and ethylenediamine (EDA) were purchased from Tianjin Damao Chemical Co. The ionic liquid ([C 4 mim]BF 4 ) was obtained from Beijing Solarbio Technology Co. Ethylene glycol (EG) was purchased from Tianjin Chemical Reagent Co. and distilled water was used throughout the experiment.

Synthesis
In a typical experimental process, FeCl 3 ·6H 2 O (0.81 g) was dissolved into 60.0 mL of ethylene glycol (EG). Then, 0, 0.1, 0.2, 0.3, 0.4, and 0.5 mL of [C 4 Mim]BF 4 were added to the above solution. After 2 h, 3 mL of ethylenediamine (EDA) was added into the mixture solution to form a well-distributed emulsion. After 1 h, the emulsion was transferred into a Teflonsealed autoclave with a capacity of 100 mL, sealed and heated at 200 • C for 20 h. After cooling to room temperature, the resultant precipitate was washed with deionized water and ethanol several times until the solution was neutral. After drying in a vacuum oven at 60 • C for 6 h, the dried products were calcined in air at 250 • C for 6 h. The reaction time was adjusted to study the mechanism.

Materials Characterization
X-ray diffraction (XRD) patterns were obtained with a Rigaku RINT 2500 diffractometer using a Cu-Kα radiation source (λ = 1.5406 Å). The morphology and microstructure of the as-prepared samples were examined with scanning electron microscopy (SEM, JEOL 6500F). TEM and HRTEM images were recorded with a Tecnai accelerating voltage of 120 and 200 kV, respectively. The nitrogen adsorption/desorption isotherms at 77.35 K were collected on an AUTOSORB iQ-MP instrument after heating the samples at 150 • C for 2 h. The surface areas were estimated using the Brunauer-Emmett-Teller (BET) method in the relative pressure range of 0.05-0.3 Pore size distributions were analyzed using nitrogen adsorption data in a Barrett-Joyner-Halenda (BJH) model. Magnetic properties of the as-synthesized samples were measured with a physical property measurement system (PPMS-9T, ever cool II, USA).

Photocatalytic Reaction
20.0 mg of α-Fe 2 O 3 photocatalyst catalyst and 50.0 mL of RhB dye aqueous solution (10.0 mg/L) were mixed in a Py"prex reactor and stirred in the dark for 30 min to reach a complete absorption/desorption equilibrium. Afterwards, the suspension was exposed to a 300 W Xenon lamp and stirred simultaneously. Subsequently, 3.0 mL of suspension was centrifuged at 5,000 rpm for 4 min to remove the photocatalyst. Finally, the concentration of RHB was monitored with a TU-1901 UV-vis spectrophotometer by monitoring the absorbance at 553 nm.

Structure and Morphology Characterization
The purity and crystallinity of the as-prepared powder samples are measured by XRD (Figure 1). The XRD shows that all diffraction peaks can be indexed in α-Fe 2 O 3 (JCPDS 33-0664) with structural parameters of a = b = 5.038 Å, c = 13.749 Å, α = β = 90 • , and γ = 120 • . No other peaks are observed, demonstrating that the α-Fe 2 O 3 products have high-purity and a single phase. Therefore, we can infer that anions and cations of the [C 4 Mim]BF 4 are not doped into the α-Fe 2 O 3 lattice. As the [C 4 Mim]BF 4 addition increases, the peak intensities of α-Fe 2 O 3 also gradually increase, indicating that sample (D) has the highest degree of crystallization. This result proves that the addition of ionic liquid can enhance the crystallization of assynthesized materials. A similar phenomenon is also observed in the previous report (Xu et al., 2013).
According to the full width at half-maximum (fwhm) of the diffraction peaks, the average crystallite size of the α- where D hkl is the particle size perpendicular to the normal line of (hkl) plane, K is a constant (it is 0.9), β hkl is the full width at halfmaximum of the (hkl) diffraction peak, θ hkl is the Bragg angle of (hkl) peak, and λ is the wavelength of X-ray.
The morphology and structure of the as-prepared samples are studied by SEM and TEM Figure 2 shows the typical SEM micrographs of the α-Fe 2 O 3 samples prepared with varying [C 4 Mim]BF 4 additions at 200 • C for 20 h. In the absence of [C 4 Mim]BF 4 (Figures 2A-C), only α-Fe 2 O 3 pseudo cubic particles are formed. The morphology and size of α-Fe 2 O 3 particles are non-uniform and the alignment is disordered. As the amount of [C 4 Mim]BF 4 increased from 0.1 to 0.3 mL, the as-prepared α-Fe 2 O 3 changed from irregular particles to the uniform microspheres. Figures 2D-F show the SEM image of [C 4 Mim]BF 4 = 0.1 mL, which is composed of nanoparticles. When the addition of [C 4 Mim]BF 4 is up to 0.2 mL, perfectly spherical α-Fe 2 O 3 is formed. As shown in Figure 2G, some broken α-Fe 2 O 3 microspheres reveal that the as-obtained α-Fe 2 O 3 microspheres are of a hollow structure. Therefore, we can infer that [C 4 Mim]BF 4 is used as a template in the process of α-Fe 2 O 3 synthesis. The three-tiered organization of crystallites for these hollow dandelions is shown in the magnified image ( Figure 2H). Numerous nanosheets are present on the surface of the dandelion spheres, with a puffy appearance and the average diameter being about 6.5 µm. From Figure 2I, we can see that individual nanosheets have an average size of about 100 nm, which is in agreement with the result calculated from Scherrer's formula. When the [C 4 Mim]BF 4 amount increases to 0.3 mL (Figures 2J,H,L), the similar nanosheets stack up in threedimensional ordered microspheres (5.0-7.0 µm) with a larger thickness (about 200 nm). Figure 3A shows the TEM image of an α-Fe 2 O 3 microsphere. However, the structure in [C 4 Mim]BF 4 = 0.3 mL demonstrates no internal void space. A high-resolution TEM (HRTEM) image ( Figure 3B) exhibited lattice spacing of 0.252 nm, corresponding to the (110) lattice plane. Additionally, some examples of synthesizing nanostructured α-Fe 2 O 3 using different kinds of ILs are given in Table 1.
Additionally, the self-assembly micron spheres are made up of a bunch of nanosheets, indicating that the spheres are porous in structure. It is striking that when the addition of [C 4 Mim]BF 4 reaches to 0.4 mL, precipitation is very sparse. Until 0.5 mL, no precipitation synthesized after reaction. This phenomenon can be explained as follows: EG and ionic liquids are not mutually soluble, so the two-phase interface is formed. With an increase in the addition of ionic liquids, the disorder of atoms at the twophase interface increases, resulting in a growing of interfacial energy, which lifts the nucleation resistance. Owing to large resistance, crystal cannot nucleate. Figure 4 shows the nitrogen adsorption-desorption isotherms and pore size distributions of the as-prepared α-Fe 2 O 3 . The isotherms of α-Fe 2 O 3 samples shown in Figure 4A are of type IV with a hysteresis loop from 0.5 to 1.0 (P/P 0 ). The BET-specific surface areas of α-Fe 2 O 3 hollow microspheres are calculated to be 183, 208, and 221 m 2 g −1 , all larger than that of α-Fe 2 O 3 synthesized without [C 4 Mim]BF 4 (only 44 m 2 g −1 ). Correspondingly, the BJH pore size of the as-synthesized α-Fe 2 O 3 ranges from 5.0 to 10.0 nm ( Figure 4B). The smaller pore structures may arise from the crystal growth, and the larger pores ascribe to the stacking of the α-Fe 2 O 3 nano-structures and the hollow structure. Additionally, the peaks of the pore sizes of the α-Fe 2 O 3 move to the right after adding [C 4 Mim]BF 4 , showing that average pore size increases. From this result, we can infer that the crystallinity of the as-prepared α-Fe 2 O 3 improved (Zhu et al., 2012), which is in agreement with the XRD analysis. Table 2 summarizes the mean pore size and S BET of the as-obtained α-Fe 2 O 3 . It shows that with increasing additions of [C 4 Mim]BF 4 , S BET increases dramatically. This result may be attributed to two factors: first, compared with irregular α-Fe 2 O 3 nanoparticles or nanosheets, hollow assembly structures have more opportunities to participate in nitrogen adsorption. Without the [C 4 Mim]BF 4 , the specific surface area of α-Fe 2 O 3 is only 44.0 m 2 g −1 (Xia et al., 2011b). Second, the presence of a large number of mesoporous structures (6-7 nm) leads to a large specific surface area.

Possible Formation Mechanism of α-Fe 2 O 3 Hollow Spheres
To better understand the formation process of the α-Fe 2 O 3 porous hollow spheres, time-dependent experiments were carried out. As shown in Figures 5A-D, representative SEM images at different time intervals are displayed. First, a hollow core-shell spherical structure was generated at 10 h. The mean diameter of the core is about 2.0 µm, which was further confirmed by the TEM image shown in Figure S1. Second, some  microparticles were deposited on its surface at 14 h. Then, as the reaction time reached 16 h, some nanosheets began to cover the surface of the nuclei, indicating that deposition was still in progress. When the duration reached 20 h, a hollow sphere covered with a great number of nanosheets was obtained.
Based on the above experimental results, the nucleation and growth mechanism for the microsphere in the two-phase system was proposed, as shown in Figure 6. First, Fe 3+ reacted with EDA in EG- [C 4 Mim]BF 4 solution to form a relatively stable complex ion of [Fe (EDA) 3 ] 3+ (Zhang et al., 2006). High temperature caused complex ions [Fe (EDA) 3 ] 3+ to decompose into FeOOH. Thus, FeOOH was first formed. Then, the FeOOH nanoparticles dissolved and reacted with the obtained Fe 2+ in a [C 4 Mim]BF 4 assisted system to form α-Fe 2 O 3 nanoparticles. From Figure 5E, the peaks of FeOOH and α-Fe 2 O 3 can be observed at 10 h. The formed α-Fe 2 O 3 nanoparticles were unstable and had a tendency to form larger congeries, which may have been driven by the minimization of interfacial energy. That is why after 14 h, the sample only showed the crystal image of α-Fe 2 O 3 . However, EG solution has greater viscosity and fewer surface hydroxyls than the aqueous solution, resulting in kinetically slower nucleation and aggregation of nanocrystals, which lead to the formation of perfectly oriented assemblies by the adequate rotation to find the low-energy configuration interface (Zhu et al., 2013). A longer reaction time leads to directional alignment to generate spherical nuclei at 10 h. During the subsequent   process, nanoparticles gradually assemble into nanosheets and stack on the surface of the template. The reason for the formation of nanosheets is that small particles gradually adhere to large particles to form thin sheets, which is commonly referred as Ostwald ripening. Finally, a specified morphology forms. Such a process is also observed in several other reports (Lou et al., 2006;Zhu et al., 2008;. In this study, the adjacent primary ferric alkoxide nanocrystals have high activity due to their high surface energy. They further grow into nanosheets by directional aggregation, which greatly reduces the interface energy of small primary nanocrystals. Then, by directional attachment and self-assembly, the nanosheets gradually evolve into 3D flower-like superstructures. At the same time, it is combined with the mature process of Ostwald to form a favorable hollow structure. In this formation process, the reaction time is one of the most important controlling factors. Apart from the reaction time, EDA is also a critical factor. EDA is used as a precipitant during this process. When a part of an EG molecule loses protons and coordinates with FeCl 3 to form ferric alkoxide, H + is produced in the reaction of EG with metal chloride. If H + cannot be removed, the accumulation of H + will inhibit the formation of further iron oxides. In this paper, EDA takes the lead to react with Fe 3+ to form a stable complex and inhibit the formation of H + in the reaction system.

Enhancement of Photocatalytic Activity
The photocatalytic degradation of RhB is monitored by measuring the absorption behavior of the solution at 553 nm. The evolutions of the spectrums are shown in Figure 7A for α-Fe 2 O 3 synthesized without [C 4 Mim]BF 4 and in Figure 7B for that synthesized with 0.3 mL [C 4 Mim]BF 4 . When the addition amount reaches 0.3 mL, 90% of RhB is decomposed after 90 min irradiation, while it is no more than 50% for the sample synthesized without [C 4 Mim]BF 4 . Thus, the use of [C 4 Mim]BF 4 in the synthesis process effectively improves the photocatalytic performance. To clarify the effect of [C 4 mim]BF 4 , we figure out the rate constants of the degradation processes for the α-Fe 2 O 3 synthesized with various concentrations of [C 4 Mim]BF 4 . To evaluate the reactivity, the apparent reaction rate constant (k) is calculated. Figure 7C shows ln (C/C 0 )-t plots for the α-   respectively, shown in Figure 7D. Thus, the photocatalytic performance is found to improve with increasing [C 4 Mim]BF 4 addition amounts. As shown in Figure 7E, we carried out a coarse comparison of degradation efficiencies between this study and other studies based on the photodegradation of RhB by α-Fe 2 O 3 .
The photocatalytic activities of the α-Fe 2 O 3 are closely related to its energy band. Figure 8A shows UV-vis diffuse reflectance spectra of α-Fe 2 O 3 with additions of [C 4 Mim]BF 4 = 0 and 0.3 mL in the wavelengths of 200-800 nm. Although the α-Fe 2 O 3 particles and spheres show similar optical properties, the microsphere with a [C 4 Mim]BF 4 addition of 0.3 mL shows stronger absorption, from 600 to 800 nm, resulting in a larger surface area that can absorb more light . Therefore, the as-prepared α-Fe 2 O 3 microspheres exhibit better photocatalytic performance. Figure 8B is the plot of (αhv) 2 vs. the energy of the absorbed light for α-Fe 2 O 3 . From Figure 8B, the band gap is determined to be 2.05 eV for the α-Fe 2 O 3 , indicating that the oxygen vacancies do not change the band gaps of α-Fe 2 O 3 microstructures significantly.
Furthermore, in the process of photocatalytic reaction, ·OH is considered to be another main reaction species leading to the oxidative decomposition of organic pollutants. However, it was revealed that ·OH could not be generated during irradiation for Fe 2 O 3 (Xiang et al., 2011). Because the photo-generated holes and electrons on Fe 2 O 3 could not react with OH − /H 2 O and O 2 to form ·OH and O 2− , respectively, no ·OH can be generated in Fe 2 O 3 (Xiang et al., 2011). This indicates that reactive species other than ·OH are present. However,  several research results suggest that photo-generated holes and electrons can directly give rise to photocatalytic oxidation. As shown in this study, the photocatalytic performances of these α-Fe 2 O 3 samples strongly depend on the [C 4 Mim]BF 4 concentrations in the synthesis process. Therefore, it is likely that the high photocatalytic activity results from the novel structure, large surface area and strong absorption of visible light. Generally, large specific surface area provides more unsaturated coordination, which helps to improve the efficiency of electron hole separation (Tang et al., 2004). Furthermore, the increase of unsaturated coordination sites may improve the surface electron transfer rate. The increase of the surface electron transfer rate leads to the reduction of the probability of recombination and, therefore, the photo-generated charge carriers could more easily transfer to the surface to degrade the adsorbed RhB.

Magnetic Properties
It is well-known that α-Fe 2 O 3 exhibits ferromagnetism (Sun et al., 2010). Figure 9 and Figure S3 shows the room temperature magnetic hysteresis loops and the magnetic field sweeping from −10.0 to 10.0 k Oe. Table 3 collects the values of remnant magnetization (Mr) and coercivity (Hc) of the as-synthesized α-Fe 2 O 3 samples. It can be determined from the shape of the hysteresis loop that the synthesized α-Fe 2 O 3 sample shows ferromagnetism. Additionally, the hysteresis loop did not reach saturation up to the maximum applied magnetic field, because of the presence of large-shape anisotropy (Bharathi et al., 2010). It is also shown in Table 3 that the residual magnetization and coercivity of assembled microspheres of α-Fe 2 O 3 are large than α-Fe 2 O 3 nanoparticles. It is widely understood that the morphology and structure of the as-synthesized samples will greatly affect the magnetization of ferromagnetic materials (Sorescu et al., 1999). Therefore, the assembly of the nano-sized and oriented particles and sheets into non-random structure results in the change of the single domain to the multidomain, leading to higher remnant magnetization and coercivity (Park et al., 2000). In summary, the difference can be attributed to the high crystallization, single domain size, surface (Tong et al., 2015), structure, and shape.

CONCLUSION
In summary, porous self-assembled α-Fe 2 O 3 hollow microspheres were successfully prepared via a facile ionic liquid assistant synthesis approach. A nucleationaggregation evacuation mechanism was the main formation of the porous hollow structures. EDA and [C 4 Mim]BF 4 play significant roles on the formation of porous α-Fe 2 O 3 hollow spheres. The increase in [C 4 Mim]BF 4 promotes photocatalytic activities and α-Fe 2 O 3 microspheres show higher photocatalytic activities. We believe that high specific surface area and porous hollow structure play an important role in improving the photocatalytic performance of as-prepared α-Fe 2 O 3 . Additionally, when the addition amount reaches 0.3 mL, the synthesized samples show ferromagnetism and can be easily recycled. The α-Fe 2 O 3 hollow porous spheres prepared by our method have high photocatalytic activity, ideal ferromagnetic properties, and high specific surface area. They are expected to exhibit use as applications in sensors, catalysis, separation technology, environmental engineering, controlled drug delivery, and more.

DATA AVAILABILITY
All datasets generated for this study are included in the manuscript and/or the Supplementary Files.

AUTHOR CONTRIBUTIONS
HY and YulZ partly designed the experiments and wrote the manuscript. QH, JZ, YuaZ, XX, and YL assisted in the analysis and interpretation of the data. JT and FW proposed the project and revised the manuscript.