Influence of pretreatment on the properties of α-Fe₂O₃ and the effect on photocatalytic degradation of methylene blue under visible light

Currently, energy and environmental needs are directing research toward compounds preferably made up of abundant, inexpensive, and non-toxic elements in order to obtain materials with interesting functionalities that can be produced by reliable and clean methods on an industrial scale (Lewis & Nocera 2006). This investigation is a contribution to thin films through the attempt to improve the properties of hematite nanostructures. Stoichiometric hematite (α-Fe₂O₃) is a narrow band gap semiconductor with a red color. It most generally occurs in n-type (Sivula et al. 2011; Elda et al. 2017) and its physical properties depend on the preparative conditions (Cornell & Schwertmann 2003). The optical gap is between 1.9 and 2.2 eV, depending on the synthesis method and such band gap allows it to absorb ∼ 30% of solar light. Moreover, the hematite is the most stable iron oxide in the binary system Fe-O, and it shows a good stability against dissolution over a wide pH range (pH > 4) owing to its basic character (Cornell & Schwertmann 2003), in addition to its availability and easy elaboration. α-Fe₂O₃ is widely used in magnetic recording and in the medical field; the nanostructures prove to have various applications in many fields such as Li-ion batteries and gas sensors (Sun et al. 2010), photoanodes for dye solar cells, water splitting and photodegradation of pollutants (Chirita & Grozescu 2009).

For photocatalytic applications, an ideal material must be stable, inexpensive, non-toxic and possess an appreciable photoactivity (Chirita & Grozescu 2009). Hematite largely satisfies these conditions and intensive research has been carried out to assess its photocatalytic efficiency, primarily through adequate engineering of the solid-liquid junction. With a clear understanding of the underlying mechanisms of absorption of photons to final catalytic reactions. The hematite is characterized by a short diffusion length (2–4 nm) and a high recombination rate of photoinduced electron-holes (e⁻/h⁺). Researches on α-Fe₂O₃ focus on the elaboration of nanostructures in the forms of nanosheets, nanowires, nanocrystals, etc., to improve the catalytic performances by ensuring good charge transfer, thus minimizing the recombination rate and increasing the electrical conductivity. Hence, it is largely preferable to use nanostructures with a low thickness to allow the photogenerated charge carriers to reach the liquid interface before their recombining and to contribute to photo redox reactions. In addition, nanostructures offer more semiconductor-liquid contact surface while increasing light harvesting (Steier et al. 2014).

Several techniques are currently used to synthesize nanostructured hematite such as chemical vapor deposition, the hydrothermal route, chemical precipitation, the sol-gel method, electrochemical anodization, pulsed laser deposition (Elda et al. 2017) and thermal oxidation of iron (Wen et al. 2005). The latter is simple and low cost and can take place under various oxidizing atmospheres. The morphology of the nanostructures and the properties can be modulated by modifying the experimental conditions during oxidation (Wen et al. 2005). Thermal oxidation can be done on iron sheets (Yuan et al. 2013) or onto thin iron films (Hiralal et al. 2011); it can be realized in a furnace (Yuan et al. 2013), on a hot plate (Hiralal et al. 2011) or by passing a strong electric current through iron wires (Nasibulin et al. 2009). Many factors can influence the characteristics of thermally developed oxides. Among these factors, surface pretreatment is the one whose effect is most often underestimated. Reports related to the surface preparation effects on thermally oxidized iron are not abundant, most date back between 1947 and 1997 (Bénard & Coquelle 1947; Graat et al. 1997) and clearly showed the significant influence of surface pretreatment on the oxidation rate and oxide properties. Since the advent of nanotechnology in the last two decades, we have not found works really dedicated to the study of the pretreatment effects on the properties of iron oxide nanostructures obtained by thermal oxidation. In this work, we try to fill this gap by investigating the effect of electrochemical pretreatment, in comparison with the classical cleaning process, on the properties of hematite nanosheets.

The synthesis of α-Fe₂O₃ nanosheets was carried out by oxidation in air at 500 °C during 2 h for both protocols. Iron sheets (99.5% purity, Good Fellow 0.1 mm × 5 mm × 18 mm) were used as substrates. The samples were divided into two categories. In the first one, the samples were submitted only to cleaning procedure before treatment (protocol a), consisting of immersing the samples in an ultrasonic bath successively in acetone (10 min), ethanol (10 min), distilled water (10 min), and finally drying with hot air (∼45 °C). In the second protocol (b), the samples were submitted to an electrochemical pretreatment (0.5 M H₂SO₄, 6 V, 5 min, Fe anode, Pt cathode) followed by the same cleaning procedure. The thermal annealing is carried out at 500 °C at a heating rate of 15 °C/min under an air humidity of 65%.

The crystalline phase was confirmed by X-ray diffraction (XRD) using a Panalytical X'Pert diffractometer with a copper anticathode (λ = 1.540598 Å) in the 2θ range (20–80°). The surface morphology of the films was characterized by Environmental Scanning Electron Microscope (ESEM) Philips XL30 at 20 kV using secondary electron mode (SE). The Raman spectroscopy Horiba LabRAM HR (633 nm He-Ne laser,700 μW) was used for analyzing the different phonon modes of the hematite. The diffuse reflectance spectrum was plotted in the range (200–800 nm) with a Jasco V-530 spectrophotometer equipped with an MgO-coated integrating-sphere, the diffuse part of the reflected light was collected. The Mott–Schottky characteristics were plotted in alkaline solution (NaOH, 0.5 M) in a standard electrochemical cell containing the working electrode, Pt auxiliary electrode and a saturated calomel electrode (SCE).

The photocatalytic study was performed on the films (4 cm²) irradiated by a LED lamp (20 W), emitting mainly in the visible region with a maximum at 520 nm. The distance between the reactor containing the MB solution and the light source was fixed at 5 cm, providing a flux of 26 mW cm⁻², and the film is immersed horizontally. All experiments were conducted at room temperature. Typically, a piece of a sample made up of the as-grown hematite on iron foil (Fe/Fe₂O₃) was put in a Pyrex beaker containing 30 mL of MB (4.0 ppm). The solution was left in the dark for 60 min for adsorption equilibrium of MB on the samples before illumination. Aliquots (3 mL) were withdrawn from the reaction solution at regular times (20 min), and the MB concentration was measured with a Shimadzu UV-2550 spectrophotometer according to Beer Lambert's law. The absorption of the solution was determined by linear interpolation of a calibration curve.

Figure 1 shows the surface morphologies of the films treated with protocols (a) and (b). One can see that the electrochemical pretreatment significantly affects the morphology of the grown film. Nanosheets were formed on both surfaces, we can notice that in the image corresponding to protocol (a) (Figure 1 left) the nanosheets are smaller and numerous while those of protocol (b) (Figure 1 right) are larger and well separated with almost flat spaces.

According to the SEM analysis, we can presume that with protocol (a), the nucleation and growth of nanosheets continue to occur during the entire annealing step; this is not the case for protocol (b) where large nanosheets preferably develop during annealing, giving rise to a reduced amount of new nucleated nanostructures. This result can be linked to the predominance of the grain-boundary diffusion over the lattice diffusion for protocol (b) because the former particularly favors the development of protuberances in the superficial film, inducing different types of nanostructures.

Figure 2 shows the XRD patterns of α-Fe₂O₃ annealed after processing with protocols (a) and (b). All peaks are indexed in a rhombohedral symmetry (space group: R-3c), in agreement with the JCPDS Card N° 33-0664, indicating a good crystallinity with lattice constants a= 0.5034 nm and c= 1.375 nm. α-Fe₂O₃ has three predominant peaks (110), (214) and (104); a small amount of magnetite (Fe₃O₄) is also detected as a secondary phase (2θ 43.51° and 56.87°, JCPDS Card N° 96-900-9769).

The peak (110) presents stronger intensity for the hematite elaborated with protocol (a), indicating that under conventional cleaning a higher amount of crystallite grows with the (100) planes parallel to the substrate surface. We also observe a clear appearance of the peak (012), not clearly distinguished in protocol (b). Moreover, a slight shift to lower angles is observed for samples treated by protocol (a), indicating more compressed crystallites in agreement with others (Bora et al. 2012).

Table 1 summarizes the results of the XRD microstructural analysis; the values of D_FWHM are almost the same for both protocols, while D_W−H shows a net increase when the electrochemical pretreatment was used. The negative sign of the microstrain (ɛ) reveals that the α-Fe₂O₃ crystalline structure in both cases is under compressive stress, which is more pronounced with protocol (a). Hence, the surface pretreatment by electrochemical quenching diminishes in a non-negligible manner the compressive stress of the α-Fe₂O₃ structure, reduces the dislocation density, and enhances the crystallite size (Table 1).

The Raman spectra of α-Fe₂O₃ samples obtained with or without electrochemical pretreatment, respectively protocol (b) and protocol (a), are displayed in Figure 3. The hematite, which belongs to the R-3c space group, has seven symmetrical Raman active modes: 2 A_{1 g} and 5 E_g, corresponding to transversal optical vibrations. These expected bands are actually observed at 225 (A_{1 g}), 245 (E_g), 291 (E_g), 301 (E_g), 410 (E_g), 498 (A_{1 g}) and 613 cm⁻¹ (E_g), which is consistent with the results reported in the literature (Lassoued et al. 2017). Additionally, a weak band at 662 cm⁻¹ is also detected, whose origin is assigned to an IR-active phonon mode activated by the disorder in the hematite crystal (Jubb & Allen 2010). Moreover, a global blue shift of Raman lines is observed for samples not treated electrochemically and this is indicative of a smaller crystallite size (Jubb & Allen 2010) and/or a higher compressive stress (McNally 2012), in agreement with previous microstructural results.

Figure 4(b) and 4(c) show the plot of (αhν)² versus hν; the energy gap can be obtained from the intercept of the linear part of the curve with the energy axis. The optical transition is clearly affected by the pretreatment, as E_g is 1.97 eV for the sample treated electrochemically (protocol b) and increases up to 2.12 eV for the conventional cleaning (protocol a). The smaller gap obtained after the electrochemical treatment is presumably related to the larger crystallite size (Pandey et al. 2014), as mentioned above in the microstructural analysis.

The values of N_A (6.77 × 10¹⁹ cm⁻³) for protocol (a) and (3.01 × 10²⁰ cm⁻³) for protocol (b), correspond to an increase by one order of magnitude when electrochemical pretreatment is used. Hematite is commonly an n-type semiconductor, the p-type semiconductivity of as-elaborated α-Fe₂O₃ is very likely related to the midgap states at the surface. Indeed, near-surface layers generally exhibit a higher deviation from stoichiometry than the bulk; besides, a segregation of intrinsic lattice defects at the surface can occur (Gleitzer et al. 1991). When these midgap surface states are sufficiently high to bring the valence band closer to the Fermi level than the conduction band, the region near the surface turns to p-type and forms an inversion layer (Baca & Ashby 2005).

By eliminating in Equation (7) the term k_BT (∼26 meV at 300 K), since it can be neglected in front of 930 and 770 mV, the straight part of the curve crosses the potential-axis (C⁻² = 0) at the following flat band potentials: E_fb = 0.93 V_/SCE (Figure 5(a)) for protocol (a) and E_fb = 0.77 V_/SCE (Figure 5(b)) for protocol (b). The cathodic shift of E_fb toward lower potentials can be related to the increased crystallite size obtained by the electrochemical processing, as observed by Iandolo et al. (2015).

In summary, these results show that the electrochemical pretreatment before thermal oxidation induces changes in the carriers’ density as well as on the flat band potential, and thus on the band bending. These electrochemical changes will inevitably impact the photocatalytic properties as will be demonstrated in the next section.

Figure 6 shows a decrease in the concentration of MB as a function of time, which indicates the degradation of dye for both protocols (a) and (b). We notice that 78% of MB was degraded within 180 min of visible light irradiation for samples not pretreated (Figure 6(a)) while this degradation reached 84% for the sample electrochemically treated (Figure 6(b)). And it should be pointed out that these MB elimination percentages are linked to photodegradation and not to adsorption, since preliminary tests have shown that elimination by adsorption process does not exceed 8%.

The kinetics of degradation show non-linear variations of C_t/C₀ as a function of time (Figure 7), confirming that the photodegradation kinetic is of the first order. The rate constant k_app is equal to 7.37 × 10⁻³ min⁻¹ for the untreated sample and increases up to 8.42 × 10⁻³ min⁻¹ for the electrochemically processed sample.

These results confirm on the one hand that the as-grown nanosheets are effective for the MB degradation if we compare with the literature (Vu et al. 2019). On the other hand, the samples submitted to electrochemical pretreatment are more efficient. This better photocatalytic ability is probably due to the higher carrier density, the lower band gap value and the better crystallinity.

Moreover, it is important to mention that the performance of the films does not decrease after the third cycle for both protocols, as is illustrated in Figure 8.

Figure 9 illustrates the photocatalytic process; the efficient photons (energy > E_g) are absorbed, causing the electron excitation to the conduction band (CB) leaving a hole in the valence band (VB). Under favorable internal conditions the electrons and holes move in opposite directions under the effect of the interfacial electric field. They reach the interface electrolyte surface and react respectively with adsorbed O₂ and H₂O, generating two principal radicals, O⁻_2^· and ·OH, which mineralize the organic molecule (MB) to CO₂ and H₂O. This process occurs only by favorable conditions means: high photon absorption, low bulk recombination, reduced surface traps, efficient separation of (e⁻–h⁺) pairs and efficient surface charge injection (Tamirat et al. 2016).

In this work, nanosheets of the hematite α-Fe₂O₃ were prepared by thermal oxidation of iron foils in air at 500 °C. The iron surface electrochemically pretreated before annealing showed a remarkable influence on the crystallite size, internal stress, surface morphology and optical properties of the hematite. The diffuse reflectance showed that the band gap varies according to the surface pretreatment protocol, from 2.12 eV for a conventional cleaning to 1.97 eV for electrochemical pretreatment. Additionally, the capacitance measurements indicated p-type behavior, with a high hole concentration and a cathodic shift of the flat band for the samples treated electrochemically predicting a better photocatalytic effect. Finally, the photocatalytic study for the methylene blue degradation demonstrated the ability of electrochemical pretreatment to enhance appreciably the photoactivity of α-Fe₂O₃ nanosheets.

All relevant data are included in the paper or its Supplementary Information.