A. Characterization of the synthesized Fe₃O₄ NPs

X-ray diffraction (XRD) was carried out to identify the structures and crystallinity of the synthesized Fe₃O₄ NPs. As shown in Fig. 1, all of the obtained products show peaks at 18.3, 30.1, 35.5, 43.1, 53.5, 57.0, and 62.6°, which can be indexed to the reflections from crystal planes (111), (220), (311), (400), (422), (511), and (440) of the cubic phase of Fe₃O₄ according to the standard JCPDS data (card no. 01–1111). ^{Reference Liu, Lu, Xing and Zhu31} The absence of any unidentified peaks in Fig. 1 indicates that the synthesized NPs are pure and crystalline. Additionally, there was a difference in XRD peak intensity among the XRD patterns of the four Fe₃O₄ NPs samples. The application of Scherrer equation to the (311) reflection peak indicated that the crystal sizes were 21.7, 39.5, 60.8, and 80.5 nm for the four Fe₃O₄ NPs samples, as listed in Table I.

FIG. 1.

XRD patterns of the synthesized Fe₃O₄ NPs: (a) 30 nm, (b) 70 nm, (c) 250 nm, and (d) 600 nm.

TABLE I.

Comparison of the structural characteristics of the synthesized Fe₃O₄ NPs.

^a S _BET is the specific surface area evaluated using the BET model.

^b q _m is the monolayer adsorption capacity evaluated using the Langmuir model.

^c Bulk ratio between Fe^II and Fe^III in Fe₃O₄.

The size and morphology of the four Fe₃O₄ NPs samples were further studied with SEM and TEM. As can be seen from Fig. 2, the four Fe₃O₄ NPs samples exhibit spherical morphologies with average particle sizes of approximately 30, 70, 250, and 600 nm, respectively, which are much larger than the crystal size obtained from XRD analysis. The main reason is that the Fe₃O₄ NPs were the results of secondary growth of crystal grain. The crystal sizes obtained from XRD results represent the sizes of Fe₃O₄ crystal grains, while the average particle sizes observed from SEM and TEM show the sizes of Fe₃O₄ NPs. This may be due to the synthesis method of Fe₃O₄ NPs. ^{Reference Yu, Zhang, Zhang, Qiao, Yang and Liu32} As reported, the formation of Fe₃O₄ follows the nucleation-growth mechanism through oxidation–precipitation method. The nucleation of magnetite particles occurs when the molar ratio of Fe(OH)₂/Fe(OH)₃ becomes close to 1:2, which is the characteristic value for the magnetite structure. Then the core of Fe₃O₄ begins to grow up. ^{Reference Yu, Zhang, Zhang, Qiao, Yang and Liu32} Therefore, the sizes of the product were strongly depend on reaction conditions such as reaction time, molar ratio of FeSO₄ and oxidizing agent, and the concentration of FeSO₄, etc. The possible mechanism of formation of Fe₃O₄ can be described as follows ^{Reference Yang, He, Wu, Chen, Liang, Qin, Fan, Zhu and Yuan33} :

FIG. 2.

SEM and TEM images of the synthesized Fe₃O₄ NPs: (a-1, a-2) 30 nm, (b-1, b-2) 70 nm, (c-1, c-2) 250 nm, and (d-1, d-2) 600 nm, and HR-TEM images of 30 nm Fe₃O₄ NPs (e-1, e-2).

In the procedures of fabrication of 600 nm Fe₃O₄ NPs [Fig. 2(d)], the FeSO₄ was first added into the NaOH solution to form a milk white Fe(OH)₂ suspension. After the addition of NaNO₃, the suspension gradually turned dark green then black in color, which indicated the formation of Fe₃O₄. However, when we increased 3 times of the NaNO₃ dosage and changed the order of adding reagents (adding NaOH into the mixture of FeSO₄ and NaNO₃), the ratio of ferrous to ferric ions in the metal salt solution was decreased. Thus, the smaller hydroxide particles would be generated readily, resulting in the formation of small Fe₃O₄ NPs. ^{Reference Li, Ding and Xue23} On the other hand, the reaction time was shortened from 4 to 2 h, which further made the Fe₃O₄ NPs grow smaller. ^{Reference Yan, Liu, Qiu, Wu, Yi, Zhang and Xua24} Thus, the Fe₃O₄ NPs with smaller dimension [250 nm, Fig. 2(c)] were synthesized using a lower molar ratio of FeSO₄ and oxidizing agent and a shorter growth time. When we further changed the order of adding reagents (adding FeSO₄ into the mixture of NaOH and NaNO₃), a black precipitate of Fe₃O₄ NPs was formed in a very short time after mixing the iron salt solution with the base solution, which indicated that little intermediates were formed prior to the formation of Fe₃O₄. This observation demonstrates that the reaction mechanism for the formation of Fe₃O₄ NPs may be different between the two cases, which may influence the size of Fe₃O₄ NPs. The formation of Fe₃O₄ NPs in this case is more likely by the “coprecipitation” reaction, ^{Reference Li, Ding and Xue23} which is supposed to be responsible for the formation of smaller Fe₃O₄ NPs [70 nm, Fig. 2(b)]. Several papers have reported that the size of the synthesized Fe₃O₄ NPs increases with the increase of the initial ferric salt concentration. ^{Reference Meng, Zhang, Zhao, Qiu and Yang25,Reference Tao, Dou and Sun34} In the present study, we have similar results. When the ferric salt concentration was reduced 3-fold, Fe²⁺ was kept at a low concentration and could be easily oxidized to Fe³⁺. As a result, more crystal nuclei were generated in the solution, which led to a small particle size [30 nm, Fig. 2(a)]. Moreover, further characterization of crystallinity by HR-TEM was conducted on the 30 nm Fe₃O₄ NPs. The HR-TEM images shown in Fig. 2(e) demonstrate that highly crystalline Fe₃O₄ NPs with lattice fringes were formed in the synthesis process, which correspond to (111) plane of Fe₃O₄ crystal with lattice spacing of 0.485 nm. In general, we are able to prepare satisfying Fe₃O₄ NPs with controllable sizes by varying the experimental conditions through a simple oxidation–precipitation method with any surfactant.

Magnetic properties of the obtained Fe₃O₄ NPs were characterized by measuring the applied field dependence of magnetization (Fig. 3). The values of saturation magnetization (M _s) and coercivity (H _c) read from the magnetization curves are tabulated in Table I. As can be seen, the M _s and H _c of the obtained Fe₃O₄ NPs decreased with the decrease of particle sizes, which may be due to the formation of the magnetically disordered surface as a result of the large surface-to-volume ratio. ^{Reference Goya, Berquó, Fonseca and Morales35} However, the value of M _s of the 30 nm Fe₃O₄ NPs was still as high as 80.9 emu/g, which are found to be 87% of the value of bulk Fe₃O₄ (93 emu/g). ^{Reference Li, Ding and Xue23} Moreover, room-temperature magnetic response of the 30 nm Fe₃O₄ NPs was also examined by an external magnet (Fig. 3 inset). Upon placement of an external magnet beside the sample, the Fe₃O₄ NPs quickly concentrated on the side, leaving the solution transparent, which demonstrated the excellent magnetic response of the synthesized Fe₃O₄ NPs to a magnetic field.

FIG. 3.

Hysteresis loop of the synthesized Fe₃O₄ NPs.

Nitrogen adsorption/desorption isotherms of the obtained Fe₃O₄ NPs showed in Fig. 4 display type II isotherms. The hysteresis area of N₂ adsorption/desorption isotherms suggest the occurrence of capillary condensation and the wide distributions of pores in Fe₃O₄ NPs. The specific surface areas obtained from N₂ adsorption measurements for the corresponding samples were 42.64, 39.51, 6.39, and 2.86 m²/g, respectively, as presented in Table I. Notably, the sample exhibited a smaller crystal size showed larger specific surface area, which was also found in other reports. ^{Reference Wang, Zhu, Wang, Wang, Lin and Tang14,Reference Hou, Zhang, Jérôme, Duprez, Zhang and Royer18} In addition, Fe^II/Fe^III ratios of the obtained Fe₃O₄ NPs were also measured in the bulk by chemical analysis, which are presented in Table I. As can be seen, the Fe^II/Fe^III ratios obtained over the synthesized Fe₃O₄ NPs were close to the theoretical Fe^II/Fe^III ratio (0.5) for Fe₃O₄ materials.

FIG. 4.

Nitrogen adsorption/desorption isotherms for the synthesized Fe₃O₄ NPs.

B. Adsorption isotherm

According to previous studies, the adsorption performance of the catalysts has a strong effect on their catalytic activity. ^{Reference Wan, Li, Wang, Chen, Lu and Hu36} Thus, the adsorption isotherms were carried out to investigate the contaminant sorption on the obtained Fe₃O₄ NPs (Fig. 5). RhB aqueous solutions with different initial concentrations varying from 5 mg/L to 400 mg/L were used for the adsorption experiment. The experimental isotherm data were fitted to the equations of Langmuir and Freundlich by applying linear regression analysis. On the basis of the statistic analysis, the isotherms can be well described with the Langmuir model which assumed that the single adsorbate binds to a single site on the adsorbent and that all surface sites on the adsorbents have the same affinity for the adsorbate and no interactions between the adsorbates. The linear form of the Langmuir equation is given by:

where q _m (mg g⁻¹) is the monolayer adsorption capacity, and K _a (L mg⁻¹) is the Langmuir constant. The fitted Langmuir equations are q _e = 34.64 × 0.029C _e/(1 + 0.029C _e) (R ² = 0.9962), q _e = 27.79 × 0.033C _e/(1 + 0.033C _e) (R ² = 0.9849), q _e = 5.44 × 0.027C _e/(1 + 0.027C _e) (R ² = 0.9908) and q _e = 3.49 × 0.032C _e/(1 + 0.032C _e) (R ² = 0.9898) for the 30, 70, 250, and 600 nm Fe₃O₄ NPs, respectively. Good correlations imply that the adsorption of RhB on the catalysts is monolayer adsorption. The adsorption capacities of RhB onto the obtained Fe₃O₄ NPs are listed in Table I. It is obviously shown that the adsorption capacity of Fe₃O₄ NPs increased with decreasing particle size or increasing specific surface area. As the average particle size decreased to 30 nm, the specific surface area and adsorption capacity reached as high as 42.64 m²/g and 34.64 mg/g, respectively, which are almost fifteen and ten times higher than that for the 600 nm Fe₃O₄ samples.

FIG. 5.

Isothermal adsorption of RhB on the synthesized Fe₃O₄ NPs ([adsorbent] = 1 g/L; T = 25 °C).

C. The catalytic activity of the synthesized Fe₃O₄ NPs

The catalytical degradation efficiencies of RhB by different sizes of Fe₃O₄ NPs were shown in Fig. 6(a). The ability of H₂O₂ for the elimination of RhB without any heterogeneous catalyst was also evaluated. As shown in Fig. 6(a), the RhB conversion is almost negligible in the presence of H₂O₂ only, which may be due to the low oxidation potential of H₂O₂ compared with ˙OH. ^{Reference Nidheesh2} When the catalyst was added to the RhB solution, a significant RhB reduction was achieved. The RhB conversion rates of 30, 70, 250, and 600 nm Fe₃O₄ NPs were 96.3, 94.3, 73.5, and 49.0% after 300 min, respectively. Many reports have indicated that ˙OH could be generated by the reaction between H₂O₂ and Fe₃O₄ leading to the degradation of RhB, which has been described in Eqs. (1)–(6). The degradation of RhB in the Fe₃O₄/H₂O₂ system follows the pseudo-first-order kinetics, as given below:

where C ₀ and C _t are the concentrations of RhB at the beginning and time t, respectively, and k is the apparent rate constant (min⁻¹). The representative plots of −ln(C _t/C ₀) versus t are presented in Fig. 6(b) and good correlation coefficients (R ² ∼ 0.99) are obtained in our system. As evident in Fig. 6 and Table II, it is obvious that the 30 nm Fe₃O₄ NPs exhibited superior catalytic activity in the Fe₃O₄/H₂O₂ process compared with the other sizes of Fe₃O₄. The results in Fig. 6, Table I, and Table II then indicated that a greater specific surface area or adsorption capacity was beneficial to the heterogeneous Fenton-like reaction, leading to a high degradation degree of RhB. Figure 7 shows the evolution of the apparent rate constant with the specific surface area. It can be obviously seen that the catalytic activity of Fe₃O₄ NPs increased with increasing specific surface area. The results indicated that the Fenton-like process in Fe₃O₄/H₂O₂ system was a surface reaction since a larger specific surface area provided more active sites, which lead to more efficient contact with the substrate. Higher content of surface ˙OH could be produced over high specific surface area Fe₃O₄ NPs, enhancing the catalytic oxidation of RhB. On the other hand, the Fe₃O₄ NPs with smaller particle size could adsorb more RhB molecules on the surface (Table I), which were easily attacked by the generated ˙OH, resulting in higher removal efficiency. Similar situations have also been observed in several works that it is the adsorption property of heterogeneous catalyst caused the improvement in the rate of substrate degradation. ^{Reference Song, Yang, Rao, Liu and Zhang7,Reference Ucoski, Nunes, DeFreitas-Silva, Idemori and Nakagaki8,Reference Wan, Li, Wang, Chen, Lu and Hu36} It is well accepted that the Fenton-like oxidation by Fe₃O₄ NPs is controlled by surface mechanism reaction and adsorption of H₂O₂ or RhB on the surface of magnetite can affect the whole degradation reaction rate. ^{Reference Xue, Hanna, Abdelmoula and Deng11} Compared with the 200 or 600 nm Fe₃O₄ NPs, there are more active sites (such as Fe^II and Fe^III) exposed on the surface of 30 nm Fe₃O₄ NPs, resulting in more active radicals generated according to Eqs. (1)–(4). Meanwhile, more RhB molecules were adsorbed on the surface of 30 nm Fe₃O₄ NPs. Thus, the reaction between active radicals and RhB on the surface of 30 nm Fe₃O₄ NPs was accelerated rapidly. Therefore, the significant higher catalytic activity of the fine Fe₃O₄ NPs mainly originated from the higher specific surface area, as demonstrated by Fig. 7, due to the increase in exposed active site number and adsorption capacity.

FIG. 6.

RhB removal efficiency (a) and pseudo-first-order rate (b) of the synthesized Fe₃O₄ NPs of different sizes (C ₀ = 40 mg/L, [catalyst] = 1 g/L, [H₂O₂] = 50 mM, T = 25 °C, pH = 3).

TABLE II.

Selected reaction results and catalyst stability for the RhB degradation (C ₀ = 40 mg/L, [catalyst] = 1 g/L, [H₂O₂] = 50 mM, T = 25 °C, pH = 3).

^a X _RhB is the RhB conversion after 300 min of reaction.

^b X _COD is the COD abatement after 300 min of reaction.

^c Fe leaching is the total iron content in the solution at the end of the reaction.

^d (a)-1, (a)-2 and (a)-3 are the catalysts used for one, two and three successive tests.

FIG. 7.

The relationship between the constant rate and specific surface area of the Fe₃O₄ NPs of different sizes.

Furthermore, we also examined the contribution of homogeneous Fenton reaction catalyzed by dissolved iron to the whole degradation process. Hence, a special experiment was conducted after the degradation experiment. The solid catalysts were removed from the reaction solution by a Millipore filter (pore size 0.22 μm), then RhB and H₂O₂ were added to the filtrate. The operation of homogeneous Fenton reaction catalyzed by dissolved iron was similar to that of heterogeneous reaction mentioned in Sec. II. The results showed in Fig. S1 in the supplementary data revealed that the RhB conversion through homogeneous Fenton reaction was below 10% after 300 min. Thus, it indicated that the effect of dissolved iron on the degradation process is limited, which further demonstrated that the whole degradation process of RhB was a surface Fenton-like reaction. Moreover, the dissolved iron concentration was also measured after the reaction (Table II). The Fe leaching did not exceed 0.3 mg/L of each sample, which were negligible.

D. Reusability and stability of the synthesized Fe₃O₄ NPs

To investigate the reusability of the synthesized Fe₃O₄ NPs as heterogeneous catalysts, the recycling catalysis experiments of the 30 nm Fe₃O₄ NPs were carried out and the catalytic activity was examined in each cycle. After each recycling, the catalyst was treated by magnetic separation, dried and reused. As shown in Fig. 8 and Table II, only a small decrease in the RhB conversion and COD removal were observed after three successive cycles, with no significant reduction as a whole. XPS was used to quantify the changes of surface Fe^II/Fe^III ratio. The details of the Fe 2p peaks of the 30 nm Fe₃O₄ before and after use during the RhB degradation are presented in Fig. 9. The peaks located at about 710.9 and 724.1 eV represented the binding energies of Fe 2p _3/2 and Fe 2p _1/2, respectively, and the binding energies of 709.9 and 711.2 eV indicated the presence of Fe^II and Fe^III, consistent with the literature data for magnetite. ^{Reference Wang, Zhu, Wang, Wang, Lin and Tang14,Reference Yamashita and Hayes37} With the help of peak shape parameters, the Fe 2p _1/2 peak for Fe₃O₄ was deconvoluted into Fe^II and Fe^III component, which is found to be 37.2 and 62.8% of the total iron surface atoms, respectively. For the sample after RhB degradation, the Fe^II is found to be 27.5% and Fe^III 72.5%. It is obvious that parts of the Fe^II in the outermost layer of Fe₃O₄ are oxidized into Fe^III during Fenton oxidation reaction, which may lead to the slight decrease in the removal efficiency of RhB between the three successive cycles. However, the RhB conversion and COD removal still remained high (Table II), which indicated that satisfying catalytic activity was maintained after three runs even if the Fe₃O₄ was observed to slightly dissolve at each cycle (the Fe leaching was 0.17–0.26 mg/L at the end of each test, Table II). In addition, TEM images (Fig. S2) and XRD patterns (Fig. S3) of the Fe₃O₄ NPs after reaction in the supplementary data showed that the morphology, dimension, and structure of the Fe₃O₄ NPs remained unchanged, and no clear agglomeration was observed. These results indicate that the synthesized Fe₃O₄ NPs were stable and exhibited excellent reusability in the catalysis process.

FIG. 8.

Effect of recycling the catalyst on the removal of RhB. Inset: fitted by pseudo-first-order rate law (C ₀ = 40 mg/L, [30 nm Fe₃O₄] = 1 g/L, [H₂O₂] = 50 mM, T = 25 °C, pH = 3).

FIG. 9.

XPS spectrum of Fe on 30 nm Fe₃O₄ NPs surface before and after degradation of RhB (Fe 2p line).