One-step solvothermal synthesis of Al-promoted Fe₃O₄ magnetic catalysts for the selective oxidation of benzyl alcohol to benzaldehyde with H₂O₂ in water

Abstract

Aluminium (Al)-promoted Fe₃O₄ magnetic catalysts were prepared by in situ solvothermal synthesis in ethylene glycol and characterized by XRD, SEM, XPS, BET and VSM techniques. The Al promoter significantly enhanced the catalytic activity of Fe₃O₄ microspheres in the selective oxidation of benzyl alcohol to benzaldehyde with H₂O₂ in water (yield: 5.6% vs. 43.7% before and after the addition of 1.42 wt% of Al, respectively). The effects of Al source and content, catalyst loading, H₂O₂ dosage, reaction temperature and time on the catalytic performance of Al-promoted Fe₃O₄ were also investigated. The Al-promoted Fe₃O₄ catalysts were magnetically recoverable and reusable at least five times without loss of catalytic activity and selectivity after being rejuvenated with NaBH₄ reduction. Theoretical and experimental investigations confirmed that the Al promoter introduced new Al active sites besides the primary iron (Fe) sites on the surface of Fe₃O₄ microspheres, facilitated the formation of Al–peroxo (Al–OOH) intermediates and reduced the apparent activation energy of the decomposition of H₂O₂ from 52.3 kJ mol⁻¹ to 41.1 kJ mol⁻¹. The synergistic effect between Fe and Al sites contributes to a much better catalytic activity of Al-promoted Fe₃O₄ as compared to pure Fe₃O₄, γ-Al₂O₃ or their physical mixtures as catalysts.

---

1. Introduction

Chlorine-free benzaldehyde (BzH) is in high demand in the perfumery and pharmaceutical industries.^1–15 However, the conventional industrial process via hydrolysis of benzal chloride derived from toluene chlorination only produces BzH that contains chlorinated impurities. Therefore, selective oxidation of benzyl alcohol (BzOH) to BzH has attracted considerable attention as an alternative to produce chlorine-free BzH. Up to now, great efforts have been made to explore environmentally benign methods for this conversion.¹⁶ Among these, the iron (Fe)-catalyzed selective oxidation of BzOH to BzH with H₂O₂ in water over magnetic iron oxides such as magnetite (Fe₃O₄) and maghemite (γ-Fe₂O₃) (Scheme 1) is more appealing because of its many attractive advantages.^13–16 Firstly, magnetic iron oxides are abundantly available, inexpensive, environmentally benign, easy to handle and magnetically separable. Secondly, H₂O₂ is a green oxidant, which has H₂O as its only by-product. Thirdly, water is a green solvent. In the catalytic system of BzOH–H₂O₂–Fe₃O₄, H₂O₂ preferentially reacts with Fe²⁺ on the surface of Fe₃O₄ to generate hydroxyl radicals (˙OH) according to a Fenton-like reaction mechanism,¹⁷ which then selectively oxidizes BzOH to BzH. However, it remains a great challenge to improve the catalytic activity of conventional Fe₃O₄ catalysts to activate H₂O₂ for practical applications.

Scheme 1 Selective oxidation of BzOH with H₂O₂ in water over Fe₃O₄.

The substitution of Fe with other transition metals was generally used to improve catalytic properties of Fe₃O₄ mainly due to the conjugation of redox pairs of iron species and imported active ions in the ˙OH production cycle.¹⁸ In contrast, modification of Fe₃O₄ by main group metals (e.g., aluminium (Al)) has been much less explored although Al-catalyzed selective oxidation of BzOH to BzH with H₂O₂ has also been reported in some papers.^9,10,19 Since Al and Fe are the first and second richest metal elements in the Earth's crust, respectively, they usually coexist in natural oxide mineral deposits such as magnetite and bauxite. For example, up to 1.8 wt% of Al has been detected in natural magnetite.^20,21 It is also well known that up to 4 wt% of Al₂O₃ (or 2.1 wt% of Al) has been added as promoter to stabilize industrial iron catalysts for ammonia synthesis.²² Therefore, it would be worthwhile to investigate the promotional effect of Al on catalytic properties of Fe₃O₄ in order to facilitate the practical catalytic application of natural magnetite or some commercial Fe₃O₄-based catalysts in the title reaction.

In the present work, Al-promoted Fe₃O₄ catalysts were prepared by in situ solvothermal synthesis in ethylene glycol (MEG). The Al promoter was found to significantly enhance the catalytic activity of Fe₃O₄ in the title reaction. Theoretical and experimental investigations confirmed that the Al promoter introduced new Al active sites, reduced the activation energy of H₂O₂ decomposition and facilitated the formation of ˙OH. The synergistic effect between Fe and Al active sites contributes to much better catalytic activity of Al-promoted Fe₃O₄ catalysts.

2. Experimental

2.1. Direct synthesis of Al-promoted Fe₃O₄ in MEG

In this work, different Al-promoted Fe₃O₄ particles were designated as Fe₃O₄-s-m, where s represents the Al source used (AIP for the saturated solution of aluminium isopropoxide in isopropanol, AC for AlCl₃·6H₂O, AN for Al(NO₃)₃·9H₂O and AS for Al₂(SO₄)₃·18H₂O), and m indicates the Al content in weight percentage. For example, Fe₃O₄-AIP-1.42% represents Fe₃O₄ particles that were modified by using AIP as the aluminium source and had the Al content of 1.42 wt%.

A typical procedure to prepare Fe₃O₄-AIP-1.42% is as follows: FeCl₃·6H₂O (3.703 g) and NaOAc (3.601 g) were stirred in 9 mL of MEG at 50 °C for 0.5 h. Then 2 mL of AIP solution was added and stirred for another 0.5 h. The whole mixture was heated in a 60 mL Teflon-lined stainless steel autoclave at 198 °C for 24 h, and then quenched to room temperature. Black particles were recovered by magnetic separation, washed with ethanol and water, and air dried. All chemicals are analytical reagents from Sinopharm Chemical Reagent Co., Ltd.

A series of Fe₃O₄-AIP-m (m: 0.08–5.53 wt%) samples were prepared according to the same procedure except different volumes of AIP solution added (see detailed experimental conditions in Table S1 in ESI†). For comparison, pure Fe₃O₄ microspheres (Fe₃O₄-blank) were also prepared without the addition of an aluminium source.

2.2. Catalytic oxidation of BzOH with H₂O₂

In a typical procedure, BzOH (40 mmol, 4.1 mL), deionized water (8.0 mL), and catalyst (0.200 g) were added to a 50 mL two-neck flask and refluxed at 100 °C. The reaction was initiated by adding 4.0 mL (40 mmol) of H₂O₂ (30 wt% in water). After 0.5 h, another 4.0 mL (40 mmol) of H₂O₂ (30 wt% in water) was added (note that unless otherwise mentioned, the total volume of H₂O₂ was divided into two equal portions and the second portion was added half an hour later after the addition of the first). The reaction continued for another 1 h. Then the catalyst was separated from the mixture with a magnet. The product was extracted by ethyl acetate (10 mL), washed with saturated sodium thiosulfate solution (2 × 10 mL), and then analyzed by gas chromatography equipped with an AE·FFAP capillary column and a FID. The conversion of BzOH and yield of BzH were calculated as follows:

where n_0,BzOH and Δn_BzOH are the initial and converted moles of BzOH, respectively, and n_BzH is the produced moles of BzH.

2.3. Reactivation of the recovered Fe₃O₄-AIP-1.42% catalysts

After the reaction, the Fe₃O₄-AIP-1.42% catalyst was recovered with the assistance of a magnet and washed with ethanol (3 × 10 mL) and deionized water (3 × 10 mL). The recovered Fe₃O₄-AIP-1.42% was reactivated by reduction with 10 mL of NaBH₄ (0.3 M) at 40 °C for 0.5–1 h and washed with deionized water to remove the remaining NaBH₄. The catalytic performance of the recovered Fe₃O₄-AIP-1.42% in the oxidation of BzOH with H₂O₂ was then tested immediately after the reactivation treatment with NaBH₄.

2.4. Characterization and analysis

XRD patterns were collected on Bruker D8 Advance Diffractometer with Cu Kα (0.15406 nm) radiation at 30 kV and 20 mA. SEM images were taken on a JEOL JSM-5510LV Scanning Electron Microscope. Typically the sizes of 100 particles were measured from the SEM images to make particle size distribution histograms. EDX patterns were performed on Horiba EMAX energy-dispersive X-ray spectrometer. XPS analyses were conducted on a Thermo Scientific K-Alpha XPS spectrometer in an ultrahigh vacuum, using a monochromatized Al Kα radiation (hν = 1486.6 eV) as the excitation source. All the binding energy (BE) values were calibrated by the C 1s peak of adventitious carbon at 284.6 eV. The Thermo Avantage v4.51 software was used to calculate the surface atomic ratio. Peak fitting was performed using XPS-PEAK software. Magnetic properties were investigated on a VSM JDAW-2000D vibrating sample magnetometer by sweeping the external field between ±6 kOe at room temperature. The concentration of H₂O₂ was analyzed by KMnO₄ titration that has a detection limit of 1 × 10⁻⁴ M.²³ The Fe³⁺/Fe²⁺ ratio in Fe₃O₄ particles and total dissolved iron in the aqueous solution were determined by the 1,10-phenanthroline spectrophotometric method.^23–26 The total acid content of the catalyst was measured by standard acid–base titration as follows: 0.4 g catalyst was stirred in 20 mL NaOH standard solution (0.1 M) at 60 °C for 7 h. Then 1 mL supernatant was back-titrated with HCl (0.01 M) after being diluted by 5 mL deionized water.²⁷

3. Results and discussion

3.1. Characterization of Al-promoted Fe₃O₄

During the investigation of catalytic performances of different Al-promoted Fe₃O₄ catalysts (see Section 3.2.1.), AIP was found to be the best aluminium source to make Al-promoted Fe₃O₄ with the highest catalytic activity. Therefore, only the characterization results of a series of Fe₃O₄-AIP samples were discussed in detail in the following parts. For comparison, Fe₃O₄-AC, Fe₃O₄-AN and Fe₃O₄-AS samples were also characterized by SEM and XRD. The SEM images and corresponding particle size distributions (Fig. S1 in the ESI†) show that spherical Fe₃O₄ particles with different sizes were synthesized with AlCl₃·6H₂O or Al(NO₃)₃·9H₂O as the Al precursor (e.g., 0.32 ± 0.06 μm for Fe₃O₄-AC microspheres and 0.27 ± 0.06 μm for Fe₃O₄-AN microspheres). However, Fe₃O₄-AS microspheres synthesized with Al₂(SO₄)₃·18H₂O as the Al precursor had bimodal particle size distribution at 0.3 μm and 0.5 μm, respectively. In addition, particles in Fe₃O₄-AS are more aggregated than those in other Fe₃O₄ samples. Fe₃O₄-AC, Fe₃O₄-AN and Fe₃O₄-AS samples show similar XRD patterns (Fig. S2†). Besides the diffraction peaks of Fe₃O₄ (PDF no. 65-3107), five unidentified extra peaks have also been observed in these XRD patterns (as marked by asterisks in Fig. S2†) and indicate the presence of other crystalline phases as impurities in these Fe₃O₄ samples.

The XRD patterns of both Fe₃O₄-AIP-1.42% and Fe₃O₄-AIP-5.53% are similar to that of blank Fe₃O₄ particles (Fig. 1) and consistent with the standard data of magnetite (PDF no. 65-3107). It confirms that no phase transformation occurs during in situ Al-modification of Fe₃O₄. No peak related to the crystalline phase of aluminium oxides or (oxo)hydroxides (denoted as AlO_x) is found in Fig. 1. It indicates the amorphous structure of AlO_x in Fe₃O₄-AIP-m with the Al content of 1.42–5.53%, which also explains that the signal-to-background ratio of the XRD pattern of Fe₃O₄-AIP-m (m: 1.42% or 5.53%) gradually decreases as the Al content increases.

Fig. 1 XRD patterns of blank and Al-modified Fe₃O₄. Shown at the bottom is the standard pattern of magnetite (PDF no. 65-3107).

The Al content and distribution in Fe₃O₄-AIP-m were measured by EDX analysis at different areas of the particles. As shown in Fig. 2 and Table 1, the Al element is uniformly distributed in Fe₃O₄-AIP-1.42%, and similar Al contents (1.25, 1.57, 1.36 and 1.50 wt%, respectively) were obtained by EDX analysis at four different areas. The uniform distribution of the Al element was also confirmed in other Fe₃O₄-AIP-m samples (see Fig. S3 and Table S2† for EDX analysis of Fe₃O₄-AIP-5.53%). Besides Fe, O and Al, EDX also detected 5.6 ± 2.6 wt% of C in some areas of Fe₃O₄-AIP-1.42%. The carbon source may be mainly from ethylene glycol or its derivatives (such as 2,3-diacetyl shown in eqn (1)) since considerable carbon content was even detected in Fe₃O₄-blank (6.17 ± 1.11%).²⁸ C 1s XPS spectra revealed that carbon existed in the same chemical state in both samples, as manifested by a single peak at ca. 284.6 eV.

Fig. 2 EDX analysis of Fe₃O₄-AIP-1.42%. (a) Detection areas and (b) corresponding EDX spectra.

Table 1 EDX analysis of Fe₃O₄-AIP-1.42%

Element	Content (wt%)				Average content (wt%)
	Area 1	Area 2	Area 3	Area 4
Fe	47.96	50.28	46.47	56.41	50.28
O	42.87	43.11	44.11	40.15	42.56
C	7.92	4.81	8.07	1.64	5.61
Al	1.25	1.57	1.36	1.5	1.42
Cl	—	0.23	—	0.29	0.26

---

Some differences in the morphology and size between blank and Al-promoted Fe₃O₄ particles are observed in SEM images (Fig. 3). Fe₃O₄-blank particles mainly consist of discrete and uniform microspheres of 0.36 ± 0.08 μm in size (Fig. 3a and b). However, spherical particles in Fe₃O₄-AIP-m become a bit more aggregated than those in Fe₃O₄-blank (Fig. 3a, c and e). The particle sizes gradually reduce with the increase of the Al content (0.27 ± 0.06 μm for Fe₃O₄-AIP-1.42% and 0.17 ± 0.02 μm for Fe₃O₄-AIP-1.42%). The changes in the size and morphology after Al-modification are related to the presence of the surface AlO_x layer on Al-promoted Fe₃O₄ (see XPS measurements below), which inhibits the particle growth on one hand but induces the aggregation of some particles like alumina adhesives on the other hand. As a result, there is no obvious change in the BET surface areas of Fe₃O₄ microspheres before and after modification by AlO_x (31.0 vs. 31.3 m² g⁻¹).

Fig. 3 SEM images and corresponding particle size distribution histograms. (a and b) For Fe₃O₄-blank, (c and d) for Fe₃O₄-AIP-1.42%, (e and f) for Fe₃O₄-AIP-5.53%.

XPS measurements were carried out to characterize the elemental composition and chemical states of the surface layer. XPS survey spectra confirm the presence of Fe, O and C in all three Fe₃O₄ particles (Fig. 4a). High resolution XPS spectra in the region of Fe 2p core level (Fig. 4b) indicate that the Fe 2p_3/2 levels of three Fe₃O₄ particles have similar BE values of magnetite at ca. 710.8 eV.^29,30 The Fe 2p_3/2 BE of maghemite (γ-Fe₂O₃) or hematite (α-Fe₂O₃) is only 0.4 eV higher than that of magnetite, which may interfere with the assignment of the surface phase as magnetite. However, we do not observe a characteristic shake-up satellite of maghemite or hematite on these spectra, which should otherwise appear at ca. 8.0 eV above the Fe 2p_3/2 peak (indicated by the arrow in Fig. 4b). The intensity of the shake-up satellite is originated predominantly from the coupling of the iron core-hole state with the valence band and has turned out to be a highly reliable indicator to distinguish the surface phase on iron oxides.^29,30 Thus, the surface phase of the three Fe₃O₄ catalysts is still similar to their bulk magnetite phase. However, the major peak at ca. 56 eV in the Fe 3p spectra in Fig. 4c indicates that the chemical state of the surface Fe species is predominantly Fe³⁺ ions.^31,32 The combination of Fe 2p and 3p spectra supports that the surface phase of the three Fe₃O₄ particles is non-stoichiometric magnetite with the large excess of Fe³⁺ ions.

Fig. 4 XPS spectra of Fe₃O₄ before and after Al-modification. (a) Survey scan, (b) Fe 2p, (c) Fe 3p, (d) Al 2p and (e) O 1s.

As expected, no signal is observed for Fe₃O₄-blank in the energy range of Al 2p core-level (Fig. 4d). Due to very low Al content on the surface of Fe₃O₄-AIP-1.42%, its Al 2p signal is also too weak to be distinguished from the background. In contrast, a main peak appears in the Al 2p spectrum of Fe₃O₄-AIP-5.53% (Fig. 4d), which can be fitted into two peaks at 73.8 eV and 74.4 eV, respectively, and confirms the presence of the Al–O and Al–OH bonds in the AlO_x surface layer.³³ The surface Al/Fe atomic ratio (0.97) of Fe₃O₄-AIP-5.53% is appreciably higher than its bulk value (0.22 in Table S2†), which indicates that the Al species is significantly enriched on the surface of Fe₃O₄-AIP-5.53%. In other words, the Fe₃O₄-AIP-5.53% particles might be encapsulated with the Al species.

The O 1s spectrum (Fig. 4e) of blank Fe₃O₄ particles is deconvoluted to two components at 530.1 eV and 531.9 eV, respectively. The first peak at 530.1 eV is associated with the lattice oxygen (O_L, O²⁻) of the surface magnetite phase. The second peak at 531.9 eV can be assigned to chemisorbed oxygen (O_chem), such as the surface hydroxyl group (Fe–OH).^34–37 Besides the O_L and O_chem species, the O 1s spectra of Fe₃O₄-AIP-1.42% and -5.53% contain a third weak peak at 533.1 eV, which may be associated with oxygen in chemisorbed water^29,38 or AIP-derived C–O residues.³⁵ The O_chem/O_L ratios are respectively 1.28, 0.72 and 0.53 for Fe₃O₄-blank, Fe₃O₄-AIP-1.42% and Fe₃O₄-AIP-5.53%. It indicates that the amount of chemisorbed oxygen gradually decreases with the increase of the Al content. This may be due to the formation of new bridging Fe–O–Al groups by dehydration of surface hydroxyl groups, which also facilitates the immobilization of AlO_x on the surface of Fe₃O₄ microspheres.

3.2. Catalytic properties of Al-promoted Fe₃O₄

3.2.1. Effects of the Al source and content. Four different Al sources, which are AIP, AlCl₃·6H₂O, Al(NO₃)₃·9H₂O and Al₂(SO₄)₃·18H₂O, respectively, were employed to synthesize Al-promoted Fe₃O₄ according to the procedure described in Section 2.1. As presented in Fig. 5, all four Al-promoted Fe₃O₄ show significantly improved catalytic activities than pure Fe₃O₄ for the selective oxidation of BzOH to BzH with H₂O₂ in H₂O in terms of the yield of BzH (5.6% vs. 28.2–43.7%). Among the four Al-promoted Fe₃O₄, Fe₃O₄-AIP exhibits the best catalytic performance (BzH yield: 43.7%). No conversion of BzOH was measured in the absence of a catalyst under the reaction conditions used in this study. Without the addition of H₂O₂, only 0.6% of BzOH was converted after refluxing at 100 °C for 1.5 h in the presence of Fe₃O₄-AIP-1.42%, which confirmed that the conversion of BzOH by reacting with O₂ in the air was minimal under the current reaction conditions. The contribution of leached iron ions to the conversion of BzOH should be limited since only 5.7% of BzOH was converted in the presence of pure Fe₃O₄ catalyst under otherwise identical conditions (Fig. 5). In a separate experiment, 8.0 mL of H₂O₂ was added in one step in the presence of the Fe₃O₄-AIP-1.42% catalyst. After running the reaction at 100 °C for 40 min, the Fe₃O₄-AIP-1.42% solid catalyst was magnetically removed. The liquid phase was allowed to continuously react in the presence of the leached iron ions at 100 °C. After reacting for another 50 min, the BzH yield increased marginally from 10.8% to 14.5%, which further supports that the contribution of the soluble iron ions to the conversion of BzOH should be limited. In other control experiments performed in homogeneous solutions containing 2.2–4.2 mM of Fe³⁺ under otherwise similar reaction conditions (Caution: Bumping may occur when the H₂O₂ solution is dropwise added to the Fe³⁺ solution under refluxing with BzOH at 100 °C), the conversions of BzOH were measured at 26.8–33.9% with lower BzH selectivities (ca. 70%) due to over-oxidation of BzOH by H₂O₂ in the homogeneous catalytic system, which generates more benzoic acid and benzyl benzoate as major by-products.

Fig. 5 Effect of different Al sources on catalytic performances of the resulting Al-promoted Fe₃O₄. The abbreviations of different Al sources were explained in Section 2.1. Reaction conditions: catalyst (0.200 g), BzOH (40 mmol), H₂O₂ (80 mmol, 8.0 mL, 30 wt%), water (8.0 mL), 100 °C, 1.5 h.

Next, a series of Fe₃O₄-AIP with varying Al content (0.08–5.53 wt%) were prepared and tested for catalytic activities (Fig. 6). Fig. 6 clearly reveals that the Al content of as low as 0.25 wt% significantly promotes the catalytic performance of Fe₃O₄ (BzH yield: 29.4% on Fe₃O₄-AIP-0.25% vs. 5.6% on Fe₃O₄-blank). However, the yield of BzH increases to a less extent or even decreases when further increasing the Al content (BzH yield: 34.8% on Fe₃O₄-AIP-5.53%). The optimized Al content is 1.42% (BzH yield: 43.7% on Fe₃O₄-AIP-1.42%). The effects of the residual carbon on the improved activity of Fe₃O₄-AIP-1.42% may not be significant since Fe₃O₄-blank with the compatible carbon content and same chemical state had much lower activity (BzOH conversion: 5.7% vs. 49.7%). As a result, Fe₃O₄-AIP-1.42% was selected in the following investigation.

Fig. 6 Effect of the Al content on catalytic performances of Fe₃O₄-AIP. Reaction conditions: catalyst (0.200 g), BzOH (40 mmol), H₂O₂ (80 mmol, 8.0 mL, 30 wt%), water (8.0 mL), 100 °C, 1.5 h.

3.2.2. Effects of reaction parameters. Four reaction parameters, including catalyst loading, H₂O₂ dosage, reaction temperature and time, were investigated to optimize the catalytic performance of Fe₃O₄-AIP-1.42%. As shown in Fig. 7a, the optimal loading of Fe₃O₄-AIP-1.42% is 0.200 g under otherwise identical conditions (BzH yield: 43.7%). Lower loading of Fe₃O₄-AIP-1.42% cannot effectively catalyze the decomposition of H₂O₂; on the other hand, higher loading of Fe₃O₄-AIP-1.42% leads to quicker generation of ˙OH via the decomposition of H₂O₂, which cannot be effectively used for the oxidation of BzOH to BzH. Similarly, the dosage of H₂O₂ has to be optimized. As shown in Fig. 7b, the BzOH conversion is lower when less amount of H₂O₂ was added, e.g., 23.6% when the molar ratio of H₂O₂ to BzOH (denoted as n(H₂O₂)/n(BzOH)) is 1. The BzOH conversion increased as the n(H₂O₂)/n(BzOH) value increased but the selectivity to BzH decreased (e.g., conversion and selectivity are 67.9% and 78.3%, respectively, as n(H₂O₂)/n(BzOH) increases to 3). As a result, the obtained product contains more benzoic acid as a major by-product when n(H₂O₂)/n(BzOH) increases. Therefore, the optimal molar ratio of H₂O₂ to BzOH is chosen to be 2.

Fig. 7 Effects of reaction parameters on catalytic performance of Fe₃O₄-AIP-1.42%. (a) Catalyst loading, (b) H₂O₂ dosage, (c) reaction temperature and (d) time. The amount of catalyst, BzOH and H₂O₂ in 8.0 mL water and the reaction temperature and time are respectively 0.2 g, 40 mmol, 80 mmol, 100 °C and 1.5 h, if not changed. Benzoic acid was detected as the major by-product.

Reaction temperature has a dramatic effect on the catalytic performance of Fe₃O₄-AIP-1.42%. As shown in Fig. 7c, as the reaction temperature reduces from 100 °C to 60 °C, the conversion of BzOH and yield of BzH decrease dramatically (e.g., only 2.1% yield achieved at 60 °C).

To investigate the effect of reaction time, 8 mL of H₂O₂ (30 wt%) was added in one step, instead of in two steps as described in Section 2.2. It should be noted that the addition of H₂O₂ in one step leads to the reduction in the BzOH conversion and BzH yield as compared to the two-step addition (conversion: 38.2% vs. 49.7%; yield: 34.6% vs. 43.7%). The reaction shows a sigmoid curve with time, as shown in Fig. 7d. There is an induction period at the initial stage (ca. 0.5 h) of the reaction. The reaction rate is slow during this period (BzOH conversion is only 6.5% at 0.5 h). The reaction accelerates between 0.5 h and 1 h (BzOH conversion is 37.8% at 1 h). Thereafter, the reaction rate tapers off (BzOH conversion is 38.2% at 1.5 h) due to the lack of H₂O₂ in the system (the concentration of H₂O₂ is only 0.029 wt% at 1.5 h, as indirectly titrated by Na₂S₂O₃ standard solution with I₂-starch as indicator).

Catalytic performances of different catalysts for the selective oxidation of BzOH to BzH with H₂O₂ in water or without solvent are compared in Table 2. As shown in entries 1–6, the overall catalytic performance of Fe₃O₄-AIP-1.42% is comparable to or better than those of other magnetic iron oxides reported in the literature.^13,15 However, in terms of the BzH yield, it is still lower than those of other homogeneous (e.g., CuSO₄ in entry 13) or heterogeneous catalysts with larger surface areas (e.g., zeolites ZSM-5 and TS-1 in entries 11–12 and supported iron, Cr(salen) and gold catalysts in entries 7–10). The lower activity of Fe₃O₄-AIP-1.42% may be related to its smaller surface area (31 m² g⁻¹), which limits the amount of accessible surface active sites.

Table 2 Best performances of BzOH oxidation to BzH with H₂O₂ in water or without solvent achieved over different catalysts

Entry	Catalyst	Conversion (%)	Yield (%)	Selectivity (%)	Recyclability (cycles)	Ref.
1	Fe3O4-AIP-1.42%	49.7	43.7	87.8	5	This work
2	Fe3O4-ECH-D	36.6	34.2	93.5	5	16
3	Fe3O4-L	17.4–53.4	13.3–47.3	76–88	NA	15
4	γ-Fe2O3-L	16.3–45.4	12.5–39.5	77–88	NA	15
5	γ-Fe2O3 (50 nm)	33	32	97	5	13
6	γ-Fe2O3 (3–5 nm)	85.7	30	35	NA	13
7	Fe/PICU	52.5	51.5	98	8	3
8	Cr(salen)–MCM-41(CH3)3	65	65	100	NA	1
9	Au/γ-Al2O3	44.5	56.1	25.0	3	12
10	Au/TS-1	67	56	84	5	11
11	TS-1	89.6	65.9	73.5	NA	39
12	ZSM-5	53–59	45	71–86	6	9 and 10
13	CuSO4	98	69.6	71	3	2

---

3.2.3. Stability and recyclability of Fe₃O₄-AIP-1.42%. Fe₃O₄-AIP-1.42% exhibits favorable magnetic properties. Its magnetization curve at room temperature is shown in Fig. 8. It has large saturated magnetization (92.1 emu g⁻¹) and small remanent magnetization (5.3 emu g⁻¹) and coercivity (35.6 Oe). These magnetic properties are suitable for magnetic separation with a magnet while avoiding aggregation after removal of the applied magnetic field.

Fig. 8 Room-temperature magnetization curve of Fe₃O₄-AIP-1.42%. Inset is the enlargement near the origin.

The XRD pattern and SEM image (Fig. S4†) of the magnetically recovered Fe₃O₄-AIP-1.42% catalyst after the first run reveal no obvious change in the crystal structure and particle morphology after the reaction, and confirm its structural stability under the current reaction conditions. However, the activity of the recovered Fe₃O₄-AIP-1.42% catalyst obviously decreases (BzH yield: 19.2%). It has been reported that the Fe²⁺ content of Fe₃O₄ has a significant impact on its catalytic activity, i.e., the higher the Fe²⁺ content, the higher the activity.¹⁷ The Fe³⁺/Fe²⁺ molar ratios in the fresh Fe₃O₄-AIP-1.42% particles and those recovered after the first run were determined to be 3.10 and 6.87, respectively. It indicates the oxidation of Fe²⁺ ions during the oxidation of BzOH with H₂O₂, leading to lower Fe²⁺ content on the surface of the recovered Fe₃O₄-AIP-1.42%. The over-presentation of Fe³⁺ ions on the surface requires more H₂O₂ and longer reaction time to regenerate enough active Fe²⁺ species for the oxidation of BzOH with H₂O₂, both of which may not be fulfilled under our current reaction conditions (see Section 2.2.), which explains low activity of the recovered Fe₃O₄-AIP-1.42%. Therefore, the recovered Fe₃O₄-AIP-1.42% was reactivated by reduction with 10 mL of NaBH₄ (0.3 M) at 40 °C for 0.5–1 h. The catalytic performance of the recovered Fe₃O₄-AIP-1.42% after the reactivation was shown in Fig. 9, which supports the good recyclability of Fe₃O₄-AIP-1.42%. It is noticed in Fig. 9 that the BzH yield was lower at the third cycle. This may be due to the shorter reduction time (usually 0.5–1 h) during the reactivation treatment with NaBH₄ of the second recovered Fe₃O₄-AIP-1.42%, which may not be enough to regenerate the surface Fe²⁺ ions, or because the re-oxidation of surface Fe²⁺ ions during the storage and handling after the reactivation treatment with NaBH₄ (please note that the catalytic performances of the recovered Fe₃O₄-AIP-1.42% were usually tested immediately after the reactivation treatment with NaBH₄, but the reactivated Fe₃O₄-AIP-1.42% at the third cycle has been exposed to the air for some time before the catalytic test). However, the stabile activity and selectivity of the recovered Fe₃O₄-AIP-1.42% were still achieved by careful control of experimental conditions during the reactivation treatment by NaBH₄, as shown at the fourth and fifth cycles (BzH yield: ca. 42%). It also can be seen in Table 2 that the recyclability of Fe₃O₄-AIP-1.42%, which is stable and reusable for at least 5 cycles without obvious reduction in activity and selectivity, is compatible to or better than those of reported catalysts under similar reaction conditions.

Fig. 9 Recyclability of Fe₃O₄-AIP-1.42%. The recovered catalysts were reactivated by NaBH₄ reduction before reuse.

3.3. Catalytic decomposition of H₂O₂

It is clear that modification by AlO_x significantly improves the catalytic activity of Fe₃O₄ microspheres. This is probably due to the enhancing effect of modification by AlO_x on Fe₃O₄-catalyzed decomposition of H₂O₂. To verify this hypothesis, the mixture of 20 mL of H₂O₂ (6 wt%) and 0.2 g of catalyst was refluxed at 100 °C. At given intervals, aliquots (0.5 mL) were sampled to analyze the concentration of H₂O₂ by KMnO₄ titration. As shown in Fig. 10a, H₂O₂ was stable after 1.5 h without a catalyst. In contrast, ca. 50% and 90% of the initial amount of H₂O₂ were decomposed after 0.5 h on Fe₃O₄-blank and Fe₃O₄-AIP-1.42%, respectively, which confirms that modification by AlO_x enhances the decomposition of H₂O₂ on Fe₃O₄.

Fig. 10 Catalytic decomposition of H₂O₂ on Fe₃O₄-blank and Fe₃O₄-AIP-1.42%. (a) Plots of C_t/C₀ versus time and (b) plots of ln(C₀/C_t) versus time.

Additional experiments were conducted at 60 °C, 70 °C and 80 °C to determine the apparent activation energy, E_a, of catalytic decomposition of H₂O₂ on blank and Fe₃O₄-AIP-1.42% microspheres, and the results were also included in Fig. 10a. The decomposition of H₂O₂ was fitted to a first-order kinetic model by plotting ln(C₀/C_t) versus time (Fig. 10b), and very good linear regression was generally observed (correlation efficient: 0.957–0.999, Table 3). This is in agreement with the first-order reaction kinetics reported earlier for the decomposition of H₂O₂ on iron oxides.^23,40 The first-order rate constant, k₁, was thus obtained from the slope of the regression line. As seen in Table 3, k₁ increases as the temperature increases. According to the Arrhenius equation, k = A exp(−E_a/RT), the values of A and E_a are obtained by plotting ln k₁ versus 1/T (Fig. 11). As shown in Table 4, very good linear regression of ln k₁–1/T was obtained for both blank and Fe₃O₄-AIP-1.42% microspheres (regression coefficients are 0.995 and 0.999, respectively). The pre-exponential factor, A, for blank Fe₃O₄ microspheres (3.25 × 10⁷ h⁻¹) is about 10 times larger than that for Fe₃O₄-AIP-1.42% (2.70 × 10⁶ h⁻¹), which indicates that the collision frequency between H₂O₂ and the surface of Fe₃O₄-AIP-1.42% is decreased upon the modification by AlO_x. However, the E_a value of the decomposition of H₂O₂ on Fe₃O₄-AIP-1.42% (41.1 kJ mol⁻¹) is obviously smaller than that on blank Fe₃O₄ (52.3 kJ mol⁻¹), which suggests that modification by AlO_x introduces new active sites on the surface of Fe₃O₄-AIP-1.42%. The smaller E_a value of Fe₃O₄-AIP-1.42% contributes to its higher activity than blank Fe₃O₄ in spite of the smaller A value of the former. It should be noted that both E_a values are much lower than the bond dissociation energy of H₂O₂ (213.8 kJ mol⁻¹)⁴¹ due to the catalytic effect of Fe₃O₄.

Table 3 Linear regression of first-order kinetics of the decomposition of H₂O₂ at 60–100 °C

Catalyst	Decomposition temperature (°C)	y = a + bx (R^2)	k1 (h^−1)
Fe3O4-AIP-1.42%	100	y = 4.605x − 0.138(0.979)	4.605
	80	y = 2.212x − 0.070 (0.993)	2.212
	70	y = 1.504x − 0.037 (0.995)	1.504
	60	y = 0.926x − 0.010 (0.999)	0.926
Fe3O4-blank	100	y = 1.463x − 0.054 (0.996)	1.463
	80	y = 0.627x − 0.019 (0.998)	0.627
	70	y = 0.365x − 0.005 (0.999)	0.365
	60	y = 0.190x − 0.003 (0.957)	0.190

---

Fig. 11 Arrhenius plots of H₂O₂ decomposition on Fe₃O₄-blank and Fe₃O₄-AIP-1.42%.

Table 4 Linear regression of Arrhenius plots of the decomposition of H₂O₂ on Fe₃O₄ catalysts

Catalyst	y = a + bx (R^2)	A (h^−1)	Ea (kJ mol^−1)
Fe3O4-AIP-1.42%	y = 14.791 − 4.945x (0.999)	2.70 × 10^6	41.1
Fe3O4-blank	y = 17.297 − 6.296x (0.995)	3.25 × 10^7	52.3

---

3.4. Discussion about the reaction mechanism

Iron-catalyzed decomposition of H₂O₂ for in situ generation of ˙OH has been extensively developed as advanced oxidation processes for environmental applications.^42,43 The activation of H₂O₂ on iron is generally accepted via a complex redox reaction sequence (eqn (2) and (3)). It is remarkable that the reaction rate between Fe³⁺ and H₂O₂ is at least three orders of magnitude slower than that between Fe²⁺ and H₂O₂. Since the Fe 3p and 2p XPS spectra show that iron exists mainly as Fe³⁺ on the surface of Fe₃O₄-AIP-1.42%, the initial induction period (ca. 0.5 h) in the oxidation of BzOH with H₂O₂ (Fig. 7d) is related to the generation of the active Fe²⁺ species by slow reduction of Fe³⁺ with H₂O₂.¹⁸

Fe²⁺ + H₂O₂ → Fe³⁺ + ˙OH + OH⁻ (k = 63–76 M⁻¹ s⁻¹) (2)

Fe³⁺ + H₂O₂ → Fe²⁺ + H⁺ + HO₂˙ (k = 0.001–0.01 M⁻¹ s⁻¹) (3)

Kinetic study revealed that modification by AlO_x reduces the E_a value of the decomposition of H₂O₂ from 52.3 kJ mol⁻¹ to 41.1 kJ mol⁻¹, which implies that the formation of new active sites on the surface of Fe₃O₄-AIP-1.42%. Al-catalyzed oxidation of BzOH to BzH with H₂O₂ has been reported in the literature.^9,19,44 Unlike the case of iron with both Fe²⁺ and Fe³⁺ states, aluminium has no multiple oxidation states. It is not possible to achieve electron transfer reactions between Al³⁺ and H₂O₂. The active intermediate in Al-catalyzed oxidation with H₂O₂ was believed to be Al peroxo complex (Al–OOH).^9,10 However, there has been no direct evidence so far to support the formation of active Al–OOH intermediates. Therefore, theoretical calculations were performed according to density functional theory (DFT) at the B3LYP/6-31+G(d,p) level to show the feasibility of the formation of the Al–OOH active intermediate via the interaction of the Al-containing species and H₂O₂.

To build the initial molecular model for the Al-containing species, the acidity and coordination state were considered. The total acid content of Fe₃O₄-AIP-1.42% was determined to be 405 μmol g⁻¹ by the acid–base titration, which is about four times that of Fe₃O₄-blank (110 μmol g⁻¹) and indicates the existence of the Al–OH group on the surface. The Al species usually exist in four- or six-coordinated state. The chemical state of Al in solid particles is usually determined by magic angle spinning nuclear magnetic resonance spectroscopy. However, this technique is not applicable to Al-promoted Fe₃O₄ microspheres due to their magnetic properties. Therefore, two molecular models, [Al(OH)(H₂O)₃]²⁺ (1A) and [Al(OH)(H₂O)₅]²⁺ (2A), were respectively adopted to simulate the initial Al-containing species.

In Fig. 12, molecular models of [Al(OH)(H₂O)₃]²⁺ (1A), [Al(OH)(H₂O)₅]²⁺ (2A), H₂O₂ (B), [Al(OH)(H₂O₂)(H₂O)₃]²⁺ (1C), [Al(OH)(H₂O₂)(H₂O)₄⋯H₂O]²⁺ (2C), [Al(OOH)(H₂O)₃⋯H₂O]²⁺ (1D) and [Al(OOH)(H₂O)₅⋯H₂O]²⁺ (2D) were respectively optimized with their energies calculated in their ground states (i.e., spin multiplicity s = 1 and overall charge q = 0 for H₂O₂ or +2 for the Al-containing species). It is clear from the energy diagrams in Fig. 12 that the interaction between 1A or 2A with H₂O₂ leads to the formation of the most stable 1D or 2D via intramolecular transformation (possibly involving hydrogen [1,3]-sigmatropic rearrangement) of the intermediate 1C or 2C, respectively (eqn (4) and (5)). The intermediate 1C is a five-coordinated Al species, and its formation from 1A and B is preferred over the formation of 2C from 2A and B by ca. 11 kJ mol⁻¹. The O–O bond lengths in both intermediates 1C and 2C are very close to the initial value in H₂O₂ (1.457 Å in Table 5). Therefore, H₂O₂ has not been activated in 1C and 2C due to weak interaction between the Al center and the peroxo group (–O̲OH), as indicated by the long bond lengths of Al–O̲OH (1.985 Å and 2.074 Å, respectively). In contrast, stronger interaction between Al and –O̲OH is expected in 1D as manifested by its shorter bond length (1.741 Å), which stretches the O–O bond length to 1.525 Å in 1D and is beneficial for its dissociation to form ˙OH. In comparison, the increase in the O–O bond length (1.475 Å) is marginal in 2D as the bond length of Al–O̲OH (1.801 Å) is still too long due to the steric hindrance of other five water ligands. It indicates that the four-coordinated configuration of 1A is a better structural model for the Al active sites. This is in agreement with the suggestion in the literature that the Al Lewis acid site predominantly reacts with H₂O₂ to form the Al–OOH complex.^9,10 It is noteworthy that one outer-sphere water ligand in model 1D, 2C or 2D is linked to one inner-sphere water ligand via strong hydrogen bond (111–152 kJ mol⁻¹). DFT calculations show that it is thermodynamically unfavorable to form [Al(OOH)(H₂O)₃]²⁺, [Al(OH)(H₂O₂)(H₂O)₄]²⁺ or [Al(OOH)(H₂O)₅]²⁺ by dissociation of the outer-sphere water ligand (Fig. S5†). Since both 1D and 2D contain the Al–OOH active species, it is reasonable to assume Al as the second type of active sites on the surface of Al-promoted Fe₃O₄ catalysts.

Fig. 12 Energy diagrams calculated at B3LYP/6-31+G(d,p) level by DFT. (a) Formation of [Al(OOH)(H₂O)₃⋯H₂O]²⁺ (1D) by interaction between [Al(OH)(H₂O)₃]²⁺ (1A) and H₂O₂ (B); (b) formation of [Al(OOH)(H₂O)₅⋯H₂O]²⁺ (2D) by interaction between [Al(OH)(H₂O)₅]²⁺ (2A) and H₂O₂ (B).

Table 5 Typical theoretical structural data in Al–OOH complexes

Model	1C	1D	2C	2D
a O–O bond length in H2O2: 1.457 Å.
O–O length (Å)^a	1.452	1.525	1.454	1.475
Al–O̲OH length (Å)	1.985	1.741	2.074	1.801
H-Bond length (Å)	—	2.413	2.529	2.495
H-Bond energy (kJ mol^−1)	—	152.02	109.87	111.10
H-Bond angle (°)	—	176.9	177.4	177.3

---

The significantly improved catalytic activity of Al-promoted Fe₃O₄ catalysts should be attributed to the synergistic effect between Fe and Al active sites. Very low conversions of BzOH (0.7–5.7%) were obtained by using pure Fe₃O₄ or γ-Al₂O₃ as catalysts (entries 2 and 3 in Table 6). The moderate conversions of BzOH over physical mixtures of Fe₃O₄ and γ-Al₂O₃ with mass ratio of 1–50 (16.3–18.0% in entries 5–7) are about 3 times the sum (6.4%) of the BzOH conversions over pure Fe₃O₄ and γ-Al₂O₃, but still much smaller than that (49.7% in entry 1) achieved over Fe₃O₄-AIP-1.42%. This indicates the stronger synergistic effect in Fe₃O₄-AIP-1.42% than in physical mixtures of Fe₃O₄ and γ-Al₂O₃ due to in situ modification of Fe₃O₄ by AlO_x. The Fe and Al active sites are in intimate contact in Fe₃O₄-AIP-1.42% via the bridging Fe–O–Al groups (Fig. 4e), which were believed to lead to the synergistic effect between Al and Fe.⁴⁵ It was also found that the catalytic activities of Al-promoted Fe₃O₄ catalysts were dependent on the Al content (e.g., BzH yield: 43.7% over Fe₃O₄-AIP-1.42% vs. 34.8% over Fe₃O₄-AIP-5.53% in Fig. 6). The XPS measurements revealed that the Fe₃O₄-AIP-5.53% particles might be encapsulated with the Al species. Its lower activity could be due to excessive surface Al species, which limits the accessibility of the Fe sites. Therefore, the synergistic effect in Al-promoted Fe₃O₄ is a delicate balance between both metal sites. It demonstrates that slight changes in the surface composition or structure of catalytic materials can have large impacts on the products and energetics of the chemical reaction.⁴⁶

Table 6 Catalytic performances of Fe₃O₄, γ-Al₂O₃ and their physical mixtures^a

Entry	Catalyst	Conversion (%)	Yield (%)	Selectivity (%)
a Reaction conditions: catalyst (0.200 g), BzOH (40 mmol), H2O2 (80 mmol, 8.0 mL, 30 wt%), water (8.0 mL), 100 °C, 1.5 h. For comparison, the catalytic results of Fe3O4-AIP-1.42% were also included.b γ-Al2O3 was prepared by calcination of pseudoboehmite in a muffle furnace at 600 °C for 2 h. The calcination temperature was increased from room temperature to 600 °C at a ramp rate of 5 °C min^−1.c The values in the parentheses indicate the mass ratios of Fe3O4 to γ-Al2O3.
1	Fe3O4-AIP-1.42%	49.7	43.7	87.8
2	Pure Fe3O4	5.7	5.6	98.8
3	Pure γ-Al2O3^b	0.7	0.7	100
4	Fe3O4 + γ-Al2O3 (1 : 10)^c	2.4	2.4	99.2
5	Fe3O4 + γ-Al2O3 (1 : 1)	16.9	16.7	98.9
6	Fe3O4 + γ-Al2O3 (10 : 1)	18.0	18.0	100
7	Fe3O4 + γ-Al2O3 (50 : 1)	16.3	16.3	100

---

The reaction mechanism of Al-promoted Fe₃O₄ catalysts was proposed as Scheme 2. The initiate state of the catalyst mainly contains Fe³⁺ and Al–OH groups. After the induction period (about 0.5 h), the active species, Fe²⁺ and Al–OOH, are formed via the interaction with H₂O₂. BzOH interacts with Al–OOH to form an active surface complex with a five-membered ring,¹⁰ and H₂O₂ is decomposed at the adjacent Fe²⁺ site to generate ˙OH radicals. Surface-bonded BzOH is then selectively oxidized to BzH via simultaneous abstraction of two hydrogen atoms, one from the methylene group (PhCH̲₂OH) by ˙OH radicals and the other from the hydroxyl group (PhCH₂OH̲) by Al–OOH. It is in this step that the synergistic effect between Fe and Al active sites may be involved. The catalytic cycle completes after the BzH desorption, which restores the initial state of the catalyst surface.

Scheme 2 The proposed reaction mechanism over Al-promoted Fe₃O₄ catalysts.

4. Conclusion

Al-promoted Fe₃O₄ magnetic catalysts with significantly enhanced catalytic properties were successfully prepared by in situ solvothermal synthesis in ethylene glycol and applied in the selective oxidation of benzyl alcohol to benzaldehyde with H₂O₂ in water. Theoretical and experimental investigations confirmed that the Al-modification introduced new Al active sites besides the primary Fe active sites on the surface of Fe₃O₄ microspheres, which led to the formation of Al–peroxo intermediates and reduced the apparent activation energy of the decomposition of H₂O₂ from 52.3 kJ mol⁻¹ to 41.1 kJ mol⁻¹. The synergistic effect between two metal sites contributes to much better catalytic activity of Al-promoted Fe₃O₄ as compared to γ-Al₂O₃, Fe₃O₄ or their physical mixtures as catalysts. The findings in this work could facilitate the direct application of natural magnetite as magnetic catalysts, or help the technology transfer of some existing Fe₃O₄-based catalysts (e.g., industrial iron catalysts for ammonia synthesis) for the selective oxidation of organic compounds with H₂O₂.

Acknowledgements

This work was supported by National Natural Science Foundation of China (No. 21571146) and Department of Education of Hubei Province (Project T201606 for Science and Technology Innovation Team of Outstanding Young and Middle-aged Scientists). We thank Dr Chuntao Zhang and Dr Congyun Shi from Wuhan University of Science and Technology for their assistance in DFT calculations.

References

1. X. Wang, G. Wu, J. Li, N. Zhao, W. Wei and Y. Sun, Catal. Lett., 2007, 119, 87–94.
2. J. U. Ahmad, M. T. Räisänen, M. Leskelä and T. Repo, Appl. Catal., A, 2012, 411–412, 180–187.
3. Y. Zhang, S. N. Riduan and J. Y. Ying, Chem.–Eur. J., 2009, 15, 1077–1081.
4. C. Ragupathi, J. Judith Vijaya, R. Thinesh Kumar and L. John Kennedy, J. Mol. Struct., 2015, 1079, 182–188.
5. V. Mahdavi, H. R. Hasheminasab and S. Abdollahi, J. Chin. Chem. Soc., 2010, 57, 189–198.
6. M. J. Beier, T. W. Hansen and J.-D. Grunwaldt, J. Catal., 2009, 266, 320–330.
7. J. Luo, F. Peng, H. Wang and H. Yu, Catal. Commun., 2013, 39, 44–49.
8. J. Luo, F. Peng, H. Yu and H. Wang, Chem. Eng. J., 2012, 204–206, 98–106.
9. S. Narayanan, J. J. Vijaya, S. Sivasanker, L. J. Kennedy and A. Ariharan, J. Porous Mater., 2014, 21, 633–641.
10. A. Jia, L.-L. Lou, C. Zhang, Y. Zhang and S. Liu, J. Mol. Catal. A: Chem., 2009, 306, 123–129.
11. G. Zhan, J. Huang, M. Du, D. Sun, I. Abdul-Rauf, W. Lin, Y. Hong and Q. Li, Chem. Eng. J., 2012, 187, 232–238.
12. N. Long and N. Quan, React. Kinet., Mech. Catal., 2015, 114, 147–155.
13. F. Shi, M. K. Tse, M. M. Pohl, A. Bruckner, S. Zhang and M. Beller, Angew. Chem., Int. Ed., 2007, 46, 8866–8868.
14. F. Shi, M. K. Tse, M.-M. Pohl, J. Radnik, A. Brückner, S. Zhang and M. Beller, J. Mol. Catal. A: Chem., 2008, 292, 28–35.
15. C. Kamonsatikul, T. Khamnaen, P. Phiriyawirut, S. Charoenchaidet and E. Somsook, Catal. Commun., 2012, 26, 1–5.
16. S. Xiao, C. Zhang, R. Chen and F. Chen, New J. Chem., 2015, 39, 4924–4932.
17. L. Gao, J. Zhuang, L. Nie, J. Zhang, Y. Zhang, N. Gu, T. Wang, J. Feng, D. Yang, S. Perrett and X. Yan, Nat. Nanotechnol., 2007, 2, 577–583.
18. S. Rahim Pouran, A. A. Abdul Raman and W. M. A. Wan Daud, J. Cleaner Prod., 2014, 64, 24–35.
19. Z. Lei and R. Wang, Catal. Commun., 2008, 9, 740–742.
20. L. Zheng, X. Cheng, D. Cao, G. Wang, Z. Wang, D. Xu, C. Xia, L. Shen, Y. Yu and D. Shen, ACS Appl. Mater. Interfaces, 2014, 6, 7014–7019.
21. C. Dupuis and G. Beaudoin, Miner. Deposita, 2011, 46, 319–335.
22. M. Appl, in Ullmann's Encyclopedia of Industrial Chemistry, Wiley-VCH Verlag, 2000, p. 156.
23. S.-S. Lin and M. D. Gurol, Environ. Sci. Technol., 1998, 32, 1417–1423.
24. Q. Jia, J. Zeng, R. Qiao, L. Jing, L. Peng, F. Gu and M. Gao, J. Am. Chem. Soc., 2011, 133, 19512–19523.
25. M. A. Blesa, H. A. Marinovich, E. C. Baumgartner and A. J. G. Maroto, Inorg. Chem., 1987, 26, 3713–3717.
26. W. Du, Y. Xu and Y. Wang, Langmuir, 2007, 24, 175–181.
27. N. Li, Y. Gao, X. Zhang, Z. Yu, L. Shi and Q. Sun, Chin. J. Catal., 2015, 36, 721–727.
28. F. Chen, R. Liu, S. Xiao and C. Zhang, Mater. Res. Bull., 2014, 55, 38–42.
29. Y. Zhu, L. Tian, Z. Jiang, Y. Pei, S. Xie, M. Qiao and K. Fan, J. Catal., 2011, 281, 106–118.
30. C.-J. Jia, L.-D. Sun, F. Luo, X.-D. Han, L. J. Heyderman, Z.-G. Yan, C.-H. Yan, K. Zheng, Z. Zhang, M. Takano, N. Hayashi, M. Eltschka, M. Kläui, U. Rüdiger, T. Kasama, L. Cervera-Gontard, R. E. Dunin-Borkowski, G. Tzvetkov and J. Raabe, J. Am. Chem. Soc., 2008, 130, 16968–16977.
31. T. Yamashita and P. Hayes, Appl. Surf. Sci., 2008, 254, 2441–2449.
32. C. S. Kuivila, J. B. Butt and P. C. Stair, Appl. Surf. Sci., 1988, 32, 99–121.
33. I. Iatsunskyi, M. Kempiński, M. Jancelewicz, K. Załęski, S. Jurga and V. Smyntyna, Vacuum, 2015, 113, 52–58.
34. S. Yang, N. Yan, Y. Guo, D. Wu, H. He, Z. Qu, J. Li, Q. Zhou and J. Jia, Environ. Sci. Technol., 2011, 45, 1540–1546.
35. F. F. H. Aragón, J. D. Ardisson, J. C. R. Aquino, I. Gonzalez, W. A. A. Macedo, J. A. H. Coaquira, J. Mantilla, S. W. da Silva and P. C. Morais, Thin Solid Films, 2016, 607, 50–54.
36. Y. Shi, S. Deng, H. Wang, J. Huang, Y. Li, F. Zhang and X. Shu, RSC Adv., 2016, 6, 15999–16009.
37. A. Zhang, W. Zheng, J. Song, S. Hu, Z. Liu and J. Xiang, Chem. Eng. J., 2014, 236, 29–38.
38. J. Shang, X. Guo, F. Shi, Y. Ma, F. Zhou and Y. Deng, J. Catal., 2011, 279, 328–336.
39. A. Bhaumik and R. Kumar, J. Chem. Soc., Chem. Commun., 1995, 349–350.
40. J. Abbot and D. G. Brown, Int. J. Chem. Kinet., 1990, 22, 963–974.
41. J. A. Dean, Lange's Handbook of Chemistry, McGraw-Hill, New York, 15th edn, 1998.
42. A. D. Bokare and W. Choi, J. Hazard. Mater., 2014, 275, 121–135.
43. F. Chen, S. Xie, X. Huang and X. Qiu, J. Hazard. Mater., 2016 DOI:10.1016/j.jhazmat.2016.02.073.
44. A. Z. Jia, L. L. Lou, C. Zhang, Y. Zhang and S. Liu, J. Mol. Catal. A: Chem., 2009, 306, 123–129.
45. A. Mariana Balu, A. Pineda, K. Yoshida, J. Manuel Campelo, P. L. Gai, R. Luque and A. Angel Romero, Chem. Commun., 2010, 46, 7825–7827.
46. F. Calle-Vallejo, J. Tymoczko, V. Colic, Q. H. Vu, M. D. Pohl, K. Morgenstern, D. Loffreda, P. Sautet, W. Schuhmann and A. S. Bandarenka, Science, 2015, 350, 185–189.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental details of Fe₃O₄-AIP-m; EDX analysis of Fe₃O₄-AIP-5.53%; SEM and XRD of Fe₃O₄-s synthesized with different Al sources; comprehensive energy diagrams calculated at B3LYP/6-31+G(d,p) level by DFT. See DOI: 10.1039/c6ra23019a

---