Photocatalytic oxidative desulfurization of dibenzothiophene under simulated sunlight irradiation with mixed-phase Fe₂O₃ prepared by solution combustion

Abstract

Mixed-phase Fe₂O₃ was prepared from Fe(NO₃)₃ and (C₂H₅)₃NHCl via solution combustion. Photocatalytic oxidative desulfurization of dibenzothiophene (DBT) under simulated sunlight irradiation using mixed-phases of α- and β-Fe₂O₃ as catalysts was investigated. The DBT in the oil phase was extracted into water phase and then photooxidized to CO₂ and SO₄²⁻ by Fe₂O₃ and O₂ dissolved in water. The Fe₂O₃ containing 36.6% β-Fe₂O₃ and 63.4% α-Fe₂O₃ exhibited the highest photocatalytic activity. Sulfur removal of DBT in n-octane was 92.3% for 90 min irradiation under the conditions of V(water) : V(n-octane) = 1 : 1, air flow rate at 150 mL min⁻¹, and Fe₂O₃ addition at 0.05 g. The kinetics of photooxidative desulfurization of DBT was a pseudo-first-order with an apparent rate constant of 0.0287 min⁻¹ and half-time of 24.15 min. The sulfur content of the actual diesel could be reduced from 478 μg mL⁻¹ to 44.5 μg mL⁻¹ after 90 min. The radical scavengers experiments and terephthalic acid fluorescence technique indicated that ˙OH and O₂˙⁻ were the main reactive species in the DBT photocatalytic degradation. A potential photocatalytic desulfurization mechanism using mixed-phase Fe₂O₃ was proposed.

---

1. Introduction

Deep desulfurization of fuel oils has received increasing global attention because of the emergence of stringent sulfur content protocols. Conventional catalytic hydrodesulfurization is highly efficient in removing thiols, sulfides, and disulfides. However, this approach is less effective for dibenzothiophene (DBT), and DBTs having alkyl substituents on their 4- and/or 6-positions.^1,2

Oxidative desulfurization (ODS) has attracted much attention as an alternative technology due to its high efficiency, low cost, and moderate operation conditions.^3–8 As a kind of ODS technique, photocatalytic ODS is also attracting interest. Under this photocatalytic ODS process, sulfur compounds can be photooxidized to CO₂, SO₄²⁻, sulfoxides, and sulfones. These are highly polarized compounds, which can be removed from nonpolar oils into polar extractants, including water,⁹ acetonitrile,¹⁰ or ionic liquids (ILs).¹¹ Hirai and Shiraishi et al. conducted numerous studies on the photochemical desulfurization for light oils by combining ultraviolet (UV) irradiation and liquid–liquid extraction. Their works include the study on product identification and the effect of photosensitizer benzophenone on sulfur removal.^12–16 To improve sulfur removal, researchers also used TiO₂ as photocatalyst^17–22 because of its nontoxicity, chemical and photochemical stability, low cost, and excellent photocatalytic activity.^23,24

Above mentioned studies were all performed by UV light irradiation because the sulfur-containing compounds cannot be photodegraded itself under visible light. Another reason is that the photocatalyst TiO₂ also can only absorb UV light (λ < 387 nm), which accounts for only a small fraction (3–5%) of the energy of the sun, because of its wide band gap of 3.2 eV.^25,26 Due to the difficulty of desulfurization under UV light in industry, Shiiraishi et al.²⁷ investigated the visible light-induced DBT desulfurization using 9,10-dicyanoanthracene (DCA) as photosensitizer. However, during experimentation, the expensive DCA was dissolved in both oil and extractants, and required recovery by subsequent reextraction or adsorption processes. Using photocatalysts with narrow band gaps that can absorb visible light is a simple approach to desulfurize under visible light irradiation. For instance, the single crystal of α-Fe₂O₃ has been used as a photocatalyst in the visible region.^28–31 Such photocatalysts, however, has the disadvantage of a high recombination rate of photo-generated electrons and photo-generated holes, which reduces their photocatalytic efficiencies.

To prevent an electron–hole pair from recombining, hybrid materials have been developed. An outstanding example is Degussa P25 TiO₂, which consists of ∼80% anatase and ∼20% rutile, and displays high photocatalytic activity in the UV region. Employing the synergistic effect of mixed phases TiO₂, Ye et al. prepared mixed phases of ilmentite- and pyrochlore-type AgSbO₃ which showed promising visible light sensitive activity.³²

Fe₂O₃, including α-, β-, ε-, and γ-Fe₂O₃, is similar to TiO₂ and AgSbO₃. After thorough inquiry, there seems to be no investigation conducted yet on the photocatalytic activity of mixed-phase Fe₂O₃. The present study considered mixed-phase Fe₂O₃ with α- and β- structures, obtained by solution combustion, as candidates for visible light-induced photocatalyst. Solution combustion synthesis is a fast and easy approach that allows the formation of homogeneous solid solutions in preparing nanomaterials.³³ Using this approach to synthesize iron oxides, ferric nitrate was used as the mother salt, while urea, glycine, ammonium nitrate, or hydrazine were used as fuels.^34–38 Due to the nitrate and fuel having an exothermic combustion reaction, the heat can cause the formation of iron oxide nanopowders. Herein, we employed a new fuel, triethylamine hydrochloride ((C₂H₅)₃N·HCl), to react with Fe(NO₃)₃ to obtain Fe₂O₃. The ratio of the α-Fe₂O₃ in the mixed-phase Fe₂O₃ was primarily controlled by the fuel addition.

The present study also investigated the photocatalytic oxidative desulfurization of DBT through simulated sunlight irradiation with mixed-phase Fe₂O₃. The relationship between the sulfur removal and the ratio of α-Fe₂O₃/β-Fe₂O₃ were also examined for development of the photocatalyst with higher activity. Terephthalic acid fluorescence probing technique was used to observe the formation of the ˙OH radicals during the photooxidation process. The roles of radicals and hole scavengers on the photocatalytic oxidation of DBT were also observed. A major reaction pathway and visible light photocatalytic mechanism were inferred according to the experimental results. The current paper presents a highly effective approach in oxidizing DBT under simulated sunlight. It also provides a new concept for the development of visible-light-driven simple oxide photocatalyst with high activity.

2. Experimental

2.1. Preparation of catalysts

Fe(NO₃)₃·9H₂O and (C₂H₅)₃N·HCl were procured from Alfa Aesar and used as received. A mixture of 0.01 mol of Fe(NO₃)₃·9H₂O and (C₂H₅)₃N·HCl (0.002–0.004 mol) was prepared and heated to ca. 80 °C for 2 min in air. Clear brown aqueous concentrated salt solutions were then obtained, which were still in liquid form at room temperature. The liquids were then heated continuously up to 200 °C for 1 min in a tube furnace with a heating rate of 10 °C min⁻¹ and O₂ flow rate of 100 mL min⁻¹. Finally, the products were obtained and taken out, which were reddish-brown fine solid powders.

2.2. Characterization and analysis

Electrospray ionisation-mass spectrometry (ESI-MS) of solution was performed on a Shimadzu 2010 EV LC-MS to validate the solution structure. X-ray diffraction (XRD) patterns were obtained on a Rigaku D/MAX 2500 X-ray diffractometer using Cu-Kα radiation at a scanning rate of 0.02° s⁻¹ for 2θ from 10° to 80°. XRD was employed to examine the iron oxide crystalline structure and estimate crystal grain size. Mössbauer spectra were measured by a Fast Comtec MA-250/mr-351 Mössbauer spectrometer to determine the α-Fe₂O₃/β-Fe₂O₃ mass ratio. The measurements were conducted at room temperature by a ⁵⁷Co(Pd) source. To examine the residue elements of the samples after combustion, energy dispersive X-ray (EDX) measurements were conducted on a JEOL JSM-6510 scanning electron microscopy system equipped with an EDAX energy spectrometer. The Brunauer–Emmett–Teller (BET) surface area was measured using a Quantachrome NOVA2000 surface area analysis meter. A Shimadzu IRPrestige 21 spectrometer was employed to obtain Fourier transform infrared (FTIR) spectra. The UV-visible diffuse reflection spectra (UV-Vis DRS) were recorded using a Hitachi U-4100 spectrophotometer in the range 200 nm to 800 nm at room temperature. To measure the total Fe concentration, an AA240FS+AA240Z Varian atomic absorption spectrometer was utilized. Sulfate ions produced during the reactions, were examined using ionic chromatography with a DIONEX DX 100 instrument, equipped with an IonPacAS4A-SC Analytical Column (4 × 250 mm). CO₂ was identified by apparatus for measurement of CO₂.

2.3. Photocatalytic oxidative desulfurization

The photocatalytic reaction was performed in a photochemical reactor (Fig. S1†). A quartz trap with a circulating water jacket was placed in a glass vessel. To simulate solar radiation, a 350 W Xe-lamp with light intensity of 7.86 W m⁻² at 420 nm was placed inside the quartz trap. Water was introduced through the water jacket to cool the lamp and solution during reaction. A glass vessel with an inlet tube for air entry and an outlet tube for air and gaseous products release was placed outside the jacket. The model oil for photocatalytic reaction was prepared by dissolving DBT in n-octane with 500 μg mL⁻¹ sulfur content. The given amount of catalyst and 50 mL of model oil or actual diesel oil (478 μg mL⁻¹ S) were mixed with 50 mL of water in the vessel. The mixed solution was photoirradiated for 0 min to 90 min with air bubbling (150 mL min⁻¹). Air exiting the outlet was introduced into distilled water or a clear Ca(OH)₂ solution. The reaction system cannot be stirred using a magnetic stirrer because mixed-phase Fe₂O₃ has magnetism properties. Thereby, the effect of air bubbling was not only to dissolve O₂ in the water, but also to blend the solution. For comparison, the photocatalytic activity of commercial P25 TiO₂ purchased from Degussa was also measured following the same procedure. The total sulfur content of the samples was determined by a micro coulometer.

3. Results and discussion

3.1. ESI-MS spectra of solution 0.3(C₂H₅)₃NHCl·Fe(NO₃)₃

Fig. S2† shows the ESI-MS spectra of solution 0.3(C₂H₅)₃NHCl·Fe(NO₃)₃. Intensive peaks in the positive ion spectrum at m/z 102 and 266 correspond to (C₂H₅)₃NH⁺ and [(C₂H₅)₃NH⁺](NO₃⁻)[(C₂H₅)₃NH⁺], respectively. Peaks at m/z 250, 277, and 304 are observed in the negative ion spectrum, which correspond to Fe(NO₃)₂Cl₂⁻, Fe(NO₃)₃Cl⁻, and Fe(NO₃)₄⁻, respectively. The hypergolic ability of nitrogen-containing salts have been proved by previous researchers.^39,40 The ionic structure also ensures the stability and the homogeneity of the obtained mixture.

3.2. XRD, Mössbauer spectra, BET surface area measurements, and EDX spectrometry

When the solutions x(C₂H₅)₃NHCl·Fe(NO₃)₃ (x = 0.2−0.4) were heated continuously until the liquids boiled, a flame emerged, smoke was emitted, and powders were obtained. These solutions were stable liquid systems and homogeneity was achieved simply because the solutions were mixed using an ionic scale. This process ensures the formation of homogeneous solid materials. As the solutions were being combusted, large amounts of gaseous products were quickly emitted from the system, which prevented the combination of the obtained materials and facilitated the formation of nanopowders. Fig. 1 presents the XRD patterns of these samples. There are mainly amorphous products when the fuel was low (x = 0.2, sample F0). When the fuel addition was increased, the obtained samples F1, F2, F3, and F4 displayed structural characteristics with mixed rhombohedral α-Fe₂O₃ (space group: R3c; JCPDS No. 89-0596) and cubic β-Fe₂O₃ (space group: Ia3; JCPDS No. 39-0238) phases. No peaks from other iron oxide phases, such as γ-Fe₂O₃ and Fe₃O₄, were found.

Fig. 1 XRD patterns of Fe₂O₃ samples.

Fig. S3† presents the Mössbauer spectra of samples and the Mössbauer parameters are shown in Table S1.† The doublet and sextet are attributed to β- and α-Fe₂O₃, respectively.⁴¹ The phase composition of β/α-Fe₂O₃ ratio measured from the Mössbauer parameters is also shown in Table 1. Besides this, Table 1 shows the crystallite size measured from XRD data using Scherer equation and BET surface area of all the samples. It was obvious that the β/α-Fe₂O₃ ratio have decreased with the increase in the fuel addition. The cause is associated with heat release. Combustion reactions are represented in proposed eqn S1 to S5.† Herein, the enthalpy change of standard reactions (ΔH°, 298.15 K) are −80.12, −273.39, −466.66, −659.93, and −853.20 kJ mol⁻¹ using the data of standard formation enthalpy listed in Table S2.†

Table 1 Physical properties of the Fe₂O₃ samples

Sample	Molar ratio of fuel to Fe(NO3)3	Phase composition/wt%		XRD crystallite size/nm		BET surface area/m^2 g^−1	Band gap/eV
		β-Fe2O3	α-Fe2O3	β-Fe2O3d(211)	α-Fe2O3d(110)
F0	0.20	amorphous
F1	0.25	57.0	43.0	42.5	43.9	55.6	1.84
F2	0.30	36.6	63.4	43.1	45.3	49.5	1.82
F3	0.35	16.2	83.8	44.2	46.1	41.6	1.85
F4	0.40	12.1	87.9	45.6	53.7	32.2	1.97

---

Previous studies^42,43 indicated that β-Fe₂O₃ transformed to α-Fe₂O₃ within 400 °C to 600 °C, which depended on the crystallite size. In the current research, with the increase of the fuel and the corresponding heat release, more β-Fe₂O₃ became α-Fe₂O₃, and the crystallite size increased, including reductions in the BET surface area.

The elements were examined by EDX technique because the solution combustion method typically produces residues in the samples.⁴⁴ Results of the EDX spectra and the elements analysis are shown in Fig. S4 and Table S3.† From the spectra, no N element was present in the samples, suggesting that the combustion should be complete. The presence of Cl element in the samples may likely be due to the HCl adsorbed in the powders. With the increase of the additional (C₂H₅)₃N·HCl amount, the Cl element content also increased. The large amount of C element may be due to the pollution of conductive adhesive used in the test or the residual fuel.

3.3. FTIR spectroscopy

EDX analysis can not exclude the presence of residual carbon moieties, for this reason we have performed a FTIR analysis of the solids in order to determinate the presence of carbonaceous species on the solid. As shown in Fig. 2, the vibration bands at around 548 cm⁻¹ and 470 cm⁻¹ correspond respectively to the lattice vibration of Fe–O stretching in α-Fe₂O₃⁴⁵ and β-Fe₂O₃.⁴⁶ It is seen that the intensity of the peak at 548 cm⁻¹ increased and that of the peak at 470 cm⁻¹ decreased, indicating the change of the β/α-Fe₂O₃ ratio, which is consistent with XRD and Mössbauer measurements. The absorption band at 1636 cm⁻¹ is attributed to free or crystal water. The common peak at 2361 cm⁻¹ may be due to the proliferation of the IR beam through air.⁴⁷ Peaks of potential undecomposed groups, including NO₃⁻ and alkyl, were not detected. This also suggests that combustion was almost finished.

Fig. 2 FTIR spectra of samples (a) F1, (b) F2, (c) F3, and (d) F4.

3.4. Effect of β/α−Fe₂O₃ ratio on the sulfur removal of DBT

When 50 mL of model oil and an equal volume of water were mixed for 5 min, sulfur removal was 8.5% without photoirradiation. This indicates that some DBT were extracted from the nonpolar oil phase to the polar water phase. When the 0.05 g of Fe₂O₃ was introduced into the system, the sulfur removal was 8.9% after 90 min, which is attributed to the DBT adsorption on the Fe₂O₃ surface. If the light was turned on, DBT can be oxidized, while the remaining DBT in n-octane phase was further extracted into the water phase.

Fig. 3 shows the effect of the various Fe₂O₃ on the DBT sulfur removal after photoirradiation for 90 min in the system with V(H₂O) : V(oil) = 1 : 1. The sulfur removal of DBT with sample F4 containing only 12.1% β-Fe₂O₃ was only 48.5%, while the visible light photocatalytic efficiency of the samples increased with the increase of β-Fe₂O₃ content in the mixed-phase. When the β-Fe₂O₃ content exceeded 36.6%, a slight decrease of the photocatalytic efficiency was observed. Therefore, sample F2 with 36.6% β-Fe₂O₃ and 63.4% α-Fe₂O₃ exhibited the highest photocatalytic effect with the sulfur removal of 92.3%.

Fig. 3 The sulfur removal of DBT with Fe₂O₃ samples. Reaction conditions: model oil = 50 mL, water = 50 mL, t = 90 min, Fe₂O₃ amount = 0.05 g, room temperature.

The UV-Vis DRS was measured because the UV-Vis absorption edge is related to the energy band gap of a semiconductor catalyst. From Fig. 4, the four Fe₂O₃ photocatalysts have noticeably similar and strong visible light absorption capabilities. The band gaps of these photocatalysts can be identified by the following equation: α(hν) = B(hν − E_g)^1/2, where α represents the absorption coefficient, h is Planck’s constant, ν is light frequency, and B is an energy-independent constant.⁴⁸ Based on the equation, the band gaps of the samples can be deduced from the tangent lines that are extrapolated to (α(hν))² = 0. Results are shown in Table 1. The small band gap is seen to correspond to high photocatalytic activity.⁴⁹ Conversely, while the light absorption of sample F3 in the wavelength ranging from 350 nm to 560 nm was evidently higher than that of others, it did not exhibit the highest activity. This suggests that other factors influence photocatalytic efficiency.

Fig. 4 UV-Vis DRS of Fe₂O₃ samples.

3.5. Photocatalytic desulfurization mechanism over mixed-phase Fe₂O₃

In the semiconductor photocatalytic process, photoinduced active species, including trapped hole (h⁺), ˙OH radical, and superoxide anion (O₂˙⁻), were considered as the main oxidizing species. Terephthalic acid fluorescence approach, which facilitates the detection of ˙OH radical species, was employed in the present study because terephthalic acid can react with ˙OH to form highly fluorescent 2-hydroxyterephthalic acid that can be detected by fluorescence spectroscopy.⁵⁰ As shown in Fig. 5, the intensity of the peaks with F2 gradually increased over time, indicating that the free ˙OH radical may likely be the main active species. To further validate the function of ˙OH, h⁺, and O₂˙⁻, the experiments were performed in the presence of radical scavengers. Herein, KI, CH₃OH, and 1,4-benzoquinone (BQ) were employed as the scavengers for h⁺, ˙OH, and O₂˙⁻, respectively.^51–53

Fig. 5 ˙OH radical trapping fluorescence spectra on terephthalic acid solution with F2. Reaction conditions: terephthalic acid solution = 100 mL, terephthalic acid concentration = 2 mM, Fe₂O₃ amount = 0.05 g, room temperature.

Fig. 6 shows that the addition of 2 mM of KI slightly improved the sulfur removal, suggesting that h⁺ radicals were not the primary oxidative species involved in this process. They also did not cause the photooxidation of DBT. Since KI was utilized as h⁺ scavenger to minimize the h⁺ and electron (e⁻) recombination, the introduction of KI would have caused more electrons to survive from the recombination.⁵⁴ This enabled them to form other oxidative species that caused the DBT degradation.

Fig. 6 Effects of radical scavengers on the sulfur removal of DBT over F2 photocatalyst.

Conversely, the sulfur removal appeared to have been reduced from 89.6% to 48.4% and 15.6% after 80 min by the addition of 2 mM of BQ and 50 mM of CH₃OH, which acted as ˙OH and O₂˙⁻ radical scavengers, respectively. This indicates that ˙OH and O₂˙⁻ are the main oxidative species. To further examine the influence of oxygen, N₂ was added instead of air for 30 min during illumination. The sulfur removal of the model diesel was 8.3%, which was almost the same as the extraction yield, suggesting the need for oxygen in DBT photocatalytic degradation. When air was introduced, the sulfur removal was 8.8% in the absence of Fe₂O₃. This suggests the DBT could not be oxidized by a singlet oxygen (¹O₂) excited by O₂. O₂ acted not only as reactant, but also as an electron scavenger, which minimized the recombination of photogenerated hole–electron pairs, and thus improved the photocatalytic efficiency.⁵⁵

To examine the cause for the difference of the sulfur removal using different photocatalysts, the order of ˙OH concentrations among the samples is shown in Fig. 7. The order of ˙OH concentrations was F2 > F1 > F3 > F4, which is consistent with the order of photocatalytic activities of these samples. Fig. 7 shows that F2 had the highest separation efficiency of electron–hole pairs because of its appropriate β/α-Fe₂O₃ ratio and energy band structure. This occurrence can be described based on the models of other mixed-phase materials, such as TiO₂^56,57 and AgSbO₃,³² which have a similar energy band structure. Scheme 1 presents the proposed charge separation mechanism for mixed-phase Fe₂O₃. This is based on the band gaps of β- and α-Fe₂O₃ (about 1.7 eV to −2.0 eV and 2.1 eV to −2.2 eV, respectively⁵⁸) and a previously proposed model.³² Electron transfer from the α-Fe₂O₃ conductive band (CB) to the β-Fe₂O₃ CB can effectively separate the photo-generated electrons and holes, which thereby reduces their recombination. The α-Fe₂O₃-originating holes will then move from the valence band (VB) of α-Fe₂O₃ to β-Fe₂O₃ phase and then react with OH⁻ to form ˙OH radical to participate in the oxidation reaction.⁵⁶ However, when the β-Fe₂O₃ phase content exceeded 36.6%, photocatalytic activity slightly decreased because its band gap is narrower than that of the α-Fe₂O₃ phase and the recombination rate of the electron–hole pairs during the β-Fe₂O₃ phase is faster than that in the α-Fe₂O₃ phase. This phenomenon is similar to that in P25 TiO₂.

Fig. 7 ˙OH-trapping fluorescence spectra of various Fe₂O₃ samples in a terephthalic acid solution. Reaction conditions: terephthalic acid solution = 100 mL, terephthalic acid concentration = 2 mM, Fe₂O₃ amount = 0.05 g, room temperature.

Scheme 1 The proposed photoexcitation mechanism of Fe₂O₃ with two-phase structure

3.6. Photocatalytic desulfurization products with mixed-phase Fe₂O₃

To verify the photooxidation products of DBT, the air of the outlet was introduced into distilled water. Its pH was 6.89 prior to the reaction, which went down to 3.51 after reaction. This was almost similar to the pH of saturated H₂CO₃ at 298 K (3.49). When KMnO₄ solution was added into the water, the KMnO₄ did not fade, suggesting the absence of SO₂ in the DBT products. When the outlet of air was added into a clear Ca(OH)₂ solution, the solution became turbid. When the outlet was introduced to apparatus for measurement of CO₂, the apparatus showed that the flow of CO₂. Therefore, CO₂ was one of the final products.

When BaCl₂ solution was introduced into the lower layer water after reaction, a white precipitate has formed, which HCl could not be dissolved. This shows that SO₄²⁻ was present in the water. Ion chromatography also validated the presence of SO₄²⁻. The obtained final products of CO₂ and SO₄²⁻ agree with the results of Vargas et al.²⁰

3.7. The influence of Fe₂O₃ amount on the sulfur removal of DBT and its kinetics

Fig. 8 presents the effect of photocatalyst Fe₂O₃ F2 amount on sulfur removal of DBT. The DBT removal at 90 min increased from around 81% with a dosage of 0.025 g Fe₂O₃ to 92.3–93.1% with 0.05–0.075 g Fe₂O₃; then slightly decreased. This can be explained in terms of the availability of active sites on the Fe₂O₃ surface and the light penetration into the reaction system. As the availability of active sites increases, the catalyst amount also increases.⁵⁹ At higher catalyst amount, light scattering by photocatalyst particles and reduction of the light penetration reduce the light absorption inside the solution.⁶⁰ As shown in Fig. 8, the optimum Fe₂O₃ amount is 0.05 g to prevent the use of unnecessary additional catalyst.

Fig. 8 Effect of the Fe₂O₃ amount on the sulfur removal of DBT. Reaction conditions: model oil = 50 mL, water = 50 mL, t = 90 min, room temperature.

To investigate the photooxidation kinetics of DBT, samples (0.1 mL) were obtained from the oil phase and the sulfur content was measured every 10 min in the reaction process. Based on previous studies,^19–21 the photocatalytic oxidation rates of sulfur compounds are described by a pseudo-first-order kinetics. The sulfur content of the model gasoline will address eqn (1):

ln(C₀/C_t) = kt (1)

where C₀ and C_t are respectively the sulfur content of model gasoline at time zero and time t (min), t is the illumination time, and k is the reaction rate constant (min⁻¹). A plot of ln(C₀/C_t) versus time represents a straight line, the slope upon which linear regression equals the apparent first-order rate constant. The half-life (t_1/2 (min)) is measured using eqn (2), which is derived from eqn (1) by replacing C_t with C₀/2:

t_1/2 = ln2/k = 0.693/k (2)

Fig. 9 presents the time-course variation of ln(C₀/C_t). Table 2 lists the calculated k, correlation coefficient r, and t_1/2. The k is 0.0287 min⁻¹ and half-life t_1/2 is only 24.15 min with 0.05 g of Fe₂O₃. By comparison, k is only 0.017 min⁻¹ with 0.05 g P25 TiO₂, showing the photocatalytic activity of F2 under simulated sunlight irradiation is about 17 times that of P25. This suggests that the mixed-phase Fe₂O₃ is an effective visible light photocatalyst.

Fig. 9 Time-course variation of ln(C₀/C_t). Reaction conditions: model oil = 50 mL, water = 50 mL, t = 90 min, room temperature.

Table 2 DBT degradation apparent pseudo-first-order rate constants (min⁻¹) and associated half time (min), obtained from the slope of the ln(C₀/C_t) vs. time plot

Fe2O3 amount (g)	0.025	0.05	0.075	0.10	0.125
k (min^−1)	0.0193	0.0287	0.0301	0.0276	0.0213
Correlation coefficient r	0.9925	0.9944	0.9929	0.9950	0.9933
t 1/2 (min)	35.91	24.15	23.03	25.11	32.54

---

3.8. Leaching of iron in water

Light-induced dissolution of iron oxides is a common issue and has been explored in some studies.^61–63 This process involves the following steps.^61,62 (1) The CB electron generation by photoexcitation of iron oxides (eqn (3)). (2) Trapping of electron in the surface Fe(III) sites (Fe(III)_surf) to Fe(II) (eqn (4)). (3) Desorption of photochemically formed Fe(II) at the surface of iron oxides (Fe(II)_surf) into the solution (eqn (5)). The Fe(II)_aq can also be oxidized to Fe(III)_aq in the solution.

Fe^III₂O₃ + hν → e_cb⁻ + h_vb⁺ (3)

Fe(III)_surf + e_cb⁻ → Fe(II)_surf (4)

Fe(II)_surf → Fe(II)_aq (5)

When 0.05 g of sample F2 was used as photocatalyst in 50 mL water, the total Fe ion concentration was 3.4359 mg L⁻¹ after 90 min reaction. The leaching of iron in water accounts for ca. 0.49 at% of the used Fe₂O₃. This suggests that the Fe₂O₃ is a type of a relatively stable photocatalyst.

On the other hand, the leaching Fe ions may be responsible for the desulfurization of DBT. To identify its effect, 0.5 g of nano α-Fe₂O₃ purchased from Aladdin was added into 50 mL of water, irradiated by light for 90 min. Then the α-Fe₂O₃ was removed through a 0.22 μm Millipore filter. The total Fe ions concentration of obtained solution was 3.7985 mg L⁻¹. This solution was mixed with model oil and irradiated with air flow rate of 150 mL min⁻¹ for 90 min. The desulfurization yield is 11.38%, which is slightly higher than the extraction yield of 8.5%, indicating Fe ions have little contribution to the desulfurization of DBT.

3.9. Photocatalytic desulfurization of actual diesel

To desulfurize actual diesel, the photocatalytic process was employed. The sulfur content of the diesel can be reduced from 478 μg mL⁻¹ to 44.5 μg mL⁻¹ after 90 min. This corresponds to 90.7% sulfur removal, which is lower compared with that of the model oil. The result agrees with other photocatalytic desulfurization research,^11,15,16 where two conflicting factors emerged. On one hand, the diesel contains two-ring aromatics, such as naphthalene (NP), which subdues the photoreaction of DBT since the triplet energy of NP is inferior than that of DBT. The energy transfer from photoexcited DBT to ground-state NP could possibly have taken place during the reaction.¹⁶ On the other hand, the photooxidation of alkyl-substituted DBTs in actual diesel have continued faster than that for nonsubstituted DBT.¹⁵ This is due to the increase in electron density values for both DBTs and benzothiophenes (BTs) along with the growing carbon number of the alkyl substituents. Afterwards, DBTs and BTs with alkyl substituents of higher carbon number will be more easily desulfurized.⁶⁴

Conclusions

Mixed-phase Fe₂O₃ photocatalysts were prepared through solution combustion. With an increase in fuel, the β/α-Fe₂O₃ ratio decreases, the crystallite size increases, and the BET surface area decreases. The mixed-phase Fe₂O₃ were employed as photocatalyst to desulfurize DBT through simulated solar irradiation. The highest photocatalytic effect was demonstrated by Fe₂O₃ with 36.6% β-Fe₂O₃ and 63.4% α-Fe₂O₃. The DBT sulfur removal in n-octane reached 92.3% with 90 min irradiation. The kinetics of photooxidative desulfurization of DBT was a pseudo-first-order with an apparent rate constant of 0.0287 min⁻¹ and half-time of 24.15 min. The sulfur content of actual diesel could be reduced from 478 μg mL⁻¹ to 44.5 μg mL⁻¹ at 90 min. The photocatalytic desulfurization mechanism of DBT with mixed-phase Fe₂O₃ was also proposed.

Acknowledgements

We are grateful for the financial support from the National Natural Science Foundation of China (Nos. 21076060 and 20806021), the Key Project of Chinese Ministry of Education (No. 210018), the One-Hundred Outstanding Innovative Talents Scheme of Hebei Province Education Department (No. CPRC022), the Natural Science Foundation of Hebei Province (No. B2009000678), and the Training Funds for Talents Project of Hebei Province. We also thank Dr M. Liu for Mössbauer measurements and analysis.

Notes and references

1. J. T. Zhang, W. S. Zhu, H. M. Li, W. Jiang, Y. Q. Jiang, W. L. Huang and Y. S. Yan, Green Chem., 2009, 11, 1801–1807.
2. M. C. Capel-Sanchez, J. M. Campos-Martin and J. L. G. Fierro, Energy Environ. Sci., 2010, 3, 328–333.
3. B. Pawelec, R. M. Navarro, J. M. Campos-Martin and J. L. G. Fierro, Catal. Sci. Technol., 2011, 1, 23–42.
4. M. Feng, Recent Pat. Chem. Eng., 2010, 3, 30–37.
5. J. M. Campos-Martin, M. C. Capel-Sanchez, P. Perez-Presas and J. L. G. Fierro, J. Chem. Technol. Biotechnol., 2010, 85, 879–890.
6. A. Anisimov and A. Tarakanova, Russ. J. Gen. Chem., 2009, 79, 1264–1273.
7. F. T. Li, R. H. Liu, J. H. Wen, D. S. Zhao, Z. M. Sun and Y. Liu, Green Chem., 2009, 11, 883–888.
8. F. T. Li, C. G. Kou, Z. M. Sun, Y. J. Hao, R. H. Liu and D. S. Zhao, J. Hazard. Mater., 2012, 205–206, 164–170.
9. T. Hirai, Y. Shiraishi, K. Ogawa and I. Komasawa, Ind. Eng. Chem. Res., 1997, 36, 530–533.
10. F. T. Li, D. S. Zhao, H. X. Li and R. H. Liu, Ann. N. Y. Acad. Sci., 2008, 1140, 383–388.
11. D. S. Zhao, R. Liu, J. L. Wang and B. Y. Liu, Energy Fuels, 2008, 22, 1100–1103.
12. T. Hirai, K. Ogawa and I. Komasawa, Ind. Eng. Chem. Res., 1996, 35, 586–589.
13. Y. Shiraishi, T. Hirai and I. Komasawa, Ind. Eng. Chem. Res., 2001, 40, 293–303.
14. Y. Shiraishi, T. Hirai and I. Komasawa, Ind. Eng. Chem. Res., 1999, 38, 3300–3309.
15. Y. Shiraishi, Y. Taki, T. Hirai and I. Komasawa, Ind. Eng. Chem. Res., 1999, 38, 3310–3318.
16. T. Hirai, Y. Shiraishi, K. Ogawa and I. Komasawa, Ind. Eng. Chem. Res., 1997, 36, 530–533.
17. A. M. A. Abdel-Wahab and A. E. M. Gaber, J. Photoch. Photobio., A, 1998, 114, 213–218.
18. Y. Shiraishi, T. Hirai and I. Komasawa, J. Chem. Eng. Jpn., 2002, 5, 489–492.
19. S. Matsuzawa, J. Tanaka, S. Sato and T. Ibusuki, J. Photoch. Photobio. A, 2002, 149, 183–189.
20. R. Vargas and O. Núnez, J. Mol. Catal. A: Chem., 2008, 294, 74–81.
21. J. Zhang, D. S. Zhao, J. L. Wang and L. Y. Yang, J. Mater. Sci., 2009, 44, 3112–3117.
22. R. Vargas and O. Nunez, Sol. Energy, 2010, 84, 345–351.
23. N. Shi, X. H. Li, T. X. Fan, H. Zhou, J. Ding, D. Zhang and H. X. Zhu, Energy Environ. Sci., 2011, 4, 172–180.
24. F. T. Li, Y. Zhao, Y. Liu, Y. J. Hao, R. H. Liu and D. S. Zhao, Chem. Eng. J., 2011, 173, 750–759.
25. J. L. Zhang, Y. M. Wu, M. Y. Xing, S. A. K. Leghari and S. Sajjad, Energy Environ. Sci., 2010, 3, 715–726.
26. T. Mishra, M. Mahato, N. Aman, J. N. Patel and R. K. Sahu, Catal. Sci. Technol., 2011, 1, 609–615.
27. Y. Shiraishi, Y. Taki, T. Hirai and I. Komasawa, Chem. Commun., 1998, 2601–2602.
28. B. C. Faust, M. R. Hoffmann and D. W. Bahnemann, J. Phys. Chem., 1989, 93, 6371–6381.
29. T. Ohmori, H. Takahashi, H. Mametsuka and E. Suzuki, Phys. Chem. Chem. Phys., 2000, 2, 3519–3522.
30. S. W. Cao and Y. J. Zhu, J. Phys. Chem. C, 2008, 112, 6253–6257.
31. S. K. Mohapatra, S. E. John, S. Banerjee and M. Misra, Chem. Mater., 2009, 21, 3048–3055.
32. T. Kako and J. H. Ye, J. Mol. Catal. A: Chem., 2010, 320, 79–84.
33. B. Murugan and A. V. Ramaswamy, J. Phys. Chem. C, 2008, 112, 20429–20442.
34. K. Deshpande, A. Mukasyan and A. Varma, Chem. Mater., 2004, 16, 4896–4904.
35. X. Wang, L. Zhang, Y. Ni, J. Hong and X. Ca, J. Phys. Chem. C, 2009, 113, 7003–7008.
36. J. Toniolo, A. S. Takimi, M. J. Andrade, R. Bonadiman and C. P. Bergmann, J. Mater. Sci., 2007, 42, 4785–4791.
37. S. Asuha, S. Zhao, X. H. Jin, M. M. Hai and H. P. Bao, Appl. Surf. Sci., 2009, 255, 8897–8901.
38. M. Tadic, V. Kusigerski, D. Markovic, N. Citakovic, M. Remskar and V. Spasojevic, J. Alloys Compd., 2009, 486, 839–843.
39. H. X. Gao, Y. H. Joo, B. Twamley, Z. Q. Zhou and J. M. Shreeve, Angew. Chem., Int. Ed., 2009, 48, 2792–2795.
40. M. Smiglak, W. M. Reichert, J. D. Holbrey, J. S. Wilkes, L. Y. Sun, J. S. Thrasher, K. Kirichenko, S. Singh, A. R. Katritzky and R. D. Rogers, Chem. Commun., 2006, 2554–2556.
41. R. Zboril, M. Mashlan, D. Krausova and P. Pikal, Hyperfine Interact., 1999, 120–121, 497–501.
42. C. W. Lee, S. S. Jung and J. S. Lee, Mater. Lett., 2008, 62, 561–563.
43. S. Sakurai, A. Namai, K. Hashimoto and S. I. Ohkoshi, J. Am. Chem. Soc., 2009, 131, 18299–18303.
44. M. Mapa and C. S. Gopinath, Chem. Mater., 2009, 21, 351–359.
45. S. Music, I. Czako-Nagy, I. Salaj-Obelic and N. Ljubesic, Mater. Lett., 1997, 32, 301–305.
46. A. Kumar and A. Singhal, Nanotechnology, 2007, 18, 475703.
47. M. Huuhtanen, K. Rahkamaa-Tolonen, T. Maunula and R. L. Keiski, Catal. Today, 2005, 100, 321–325.
48. W. F. Yao, C. P. Huang and J. H. Ye, Chem. Mater., 2010, 22, 1107–1113.
49. L. Zhao, M. D. Han and J. S. Lian, Thin Solid Films, 2008, 516, 3394–3398.
50. K. I. Ishibashi, A. Fujishima, T. Watanabe and K. Hashimoto, Electrochem. Commun., 2000, 2, 207–210.
51. J. H. Kou, Z. S. Li, Y. P. Yuan, H. T. Zhang, Y. Wang and Z. G. Zou, Environ. Sci. Technol., 2009, 43, 2919–2914.
52. X. Q. Zhu, J. L. Zhang and F. Chen, Appl. Catal. B: Environ., 2011, 102, 316–322.
53. Y. Q. Yang, G. K. Zhang, S. J. Yu and X. Shen, Chem. Eng. J., 2010, 162, 171–177.
54. L. F. Yin, J. F. Niu, Z. Y. Shen and J. Chen, Environ. Sci. Technol., 2010, 44, 5581–5586.
55. M. H. Habibi and H. Vosooghian, J. Photoch. Photobio. A, 2005, 174, 45–52.
56. Z. F. Zheng, H. W. Liu, J. P. Ye, J. C. Zhao, E. R. Waclawik and H. Y. Zhu, J. Mol. Catal. A: Chem., 2010, 316, 75–82.
57. D. C. Hurum, A. G. Agrios, K. A. Gray, T. Rajh and M. C. Thurnauer, J. Phys. Chem. B, 2003, 107, 4545–4549.
58. Y. Q. Liang and R. Krol, Chem. Phys. Lett., 2009, 479, 86–90.
59. I. K. Konstantinou and T. A. Albanis, Appl. Catal. B: Environ., 2004, 49, 1–14.
60. M. R. Sohrabi and M. Ghavami, Desalination, 2010, 252, 157–162.
61. E. Rodriguez, G. Fernandez, B. Ledesma, P. Alvarez and F. J. Beltran, Appl. Catal. B: Environ., 2009, 92, 240–249.
62. K. Kim, W. Choi, M. R. Hoffmann, H. I. Yoon and B. K. Park, Environ. Sci. Technol., 2010, 44, 4142–4148.
63. W. H. Casey and C. Ludwig, Nature, 1996, 381, 506–509.
64. Y. Shiraishi, K. Tachibana, T. Hirai and I. Komasawa, Ind. Eng. Chem. Res., 2002, 41, 4362–4375.

---

Footnote

† Electronic supplementary information (ESI) available: Schematic diagram of photochemical reactor (Fig. S1), ESI-MS spectra of solution (Fig. S2), room temperature Mössbauer spectra (Fig. S3), Mössbauer parameters (Table S1), combustion reaction equations (eqn (S1 to S5)), standard formation enthalpy of materials (Table S2), EDX spectra (Fig. S4), and elements analysis results (Table S3). See DOI: 10.1039/c2cy00485b

---