Improvement of water resistance by Fe2O3/TiO2 photoelectrocatalysts for formaldehyde removal: experimental and theoretical investigation

TiO2-based photocatalysts are a potential technology for removing indoor formaldehyde (CHOH) owing to their strong photooxidation ability. However, their photooxidation performance is generally weakened when suffering from the competitive adsorption of H2O. In a method inspired by the oxygen evolution reaction (OER) to generate intermediates with hydroxyl radicals on the anode electrode catalysts, an electric field was employed in this research and applied to the photooxidation of CHOH to prevent the competitive adsorption of H2O. Additionally, 0.5–5% Fe2O3 decorated TiO2 was employed to improve the photoelectrocatalytic activity. The influence of an electric field on hydroxyl-radical production was investigated by both density functional theory (DFT) with direct-imposed dipole momentum and photoelectrocatalytic experimental tests. The surface characterization of the photocatalysts, including transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), and electron paramagnetic resonance (EPR), was conducted. DFT results show that a positive electric field with a strength of 0.05 Å/V was more favorable to produce hydroxyl on Fe2O3/TiO2(010) than was a negative electric field. Fe2O3 decoration can significantly boost hydroxyl formation, resulting from a decrease in the binding energy between the Fe of Fe2O3 and the oxygen and hydrogen atoms of H2O. The dissociated hydrogen atom of the H2O preferentially remained on the catalysts’ surface rather than being released into the gas flow. The experimental results demonstrated that applying 150 V could not directly enhance the photooxidation of CHOH by either TiO2 or Fe2O3/TiO2 but that it could relieve the H2O inhibitory effect by more than 10% on the Fe2O3/TiO2.


Introduction
Formaldehyde (CHOH) is one of China's most harmful indoor air pollutants and poses a dangerous threat to human health (Kim et al. 2011;Tang et al. 2009). Various removal technologies including adsorption (Yang et al. 2017), solvent absorption (Yang et al. 2017), catalytic conversion (Guo et al. 2019;Li et al. 2020;Miao et al. 2019;Nie et al. 2016), and photooxidation can effectively address indoor CHOH pollution. Due to their photoinduced holes and electrons, photocatalysts can produce highly oxidative hydroxyl or superoxide radicals, thus being good candidates for the photooxidation of CHOH Yu et al. 2013;Zhang et al. 2017). However, because of the competitive adsorption of the oxygen in H 2 O and the carbonyl group in CHOH, photocatalysts suffer from H 2 O interference with severalorder concentrations higher than CHOH under a specific humidity, which would enormously undermine their photocatalytic activity. A key to solving the inhibitory effect of gaseous H 2 O is activating the H 2 O molecule and producing more surface hydroxyl radicals. However, H 2 O splitting for hydroxyl production cannot proceed spontaneously and requires an enormous amount of energy to overcome the energy barrier of hydroxyl formation (Chen et al. 2014;Ding et al. 2018). It is well known that under overpotentials in an electrochemical cell, the oxygen evolution reaction (OER) on anode catalysts produces an intermediate of hydroxyl radicals by abstracting one hydrogen atom of H 2 O. Then, the released electrons are transported to the anodes, following Eq. (1) below, where * represents the coordinatively unsaturated sites of the catalyst (Valdés et al. 2008). Although the OER commonly occurs in an aqueous environment, it is vital to experimentally and theoretically investigate whether it can proceed between a gas-solid interface. Another question was whether photooxidation of gaseous H 2 O to produce hydroxyls could be strengthened under an external electric field, thus substantially enhancing the removal of CHOH. Theoretical calculation methods are currently employed to simulate the electrochemical catalysis, including (a) The computational hydrogen electrode (CHE) method, proposed by Nøskov et al. (Nørskov et al. 2004); (b) A double reference method with a modification of the electrical double layer (EDL) (Filhol and Neurock 2006;Lozovoi and Alavi 2003); (c) Direct-imposed dipole momentum on a periodic surface slab representing an electric field (Deshlahra et al. 2009;Jörg and Matthias 1992).
The CHE model can successfully analyze the overpotential of the OER on a none charged catalyst slab, but it is difficult to simulate the dependence of physiochemical adsorption on an applied external field. The EDL model could well reflect the dependence by adding one or more extra charges and identical compensating charges, which is suitable for analyzing a liquidsolid slab system (Duan and Henkelman 2019;Filhol and Neurock 2006). The third method is widely used to simulate a charged gas-solid catalytic reaction system by introducing a dipole momentum on a slab + adsorbates to embody an external electrostatic field (Deshlahra et al. 2009;Jörg and Matthias 1992). Herein, the third method of introducing a dipole moment as implemented in Vienna Ab-initio Simulation Package (VASP) was employed to investigate the synergetic effect of photoelectrocatalytic oxidation on the formation of *−OH. It is worth noting that unlike when a hydrogen atom diffuses into the electrolyte after producing a hydroxyl radical (·OH) on a liquidsolid interface, hydrogen would be attached to catalysts or transported away by carrier gasses on a gas-solid interface. Accordingly, the free energy of the formation of the two terminates was calculated in this research, where (a) one remains on the catalyst's surface as *−·OH + *−H and (b) the other leaves to form *−·OH+H + (gas). TiO 2 , due to its strong photoredox ability, excellent chemical stability, low cost, and few secondary derivatives, has been widely investigated for photodegrading formaldehyde. Under UV irradiation, strong oxidative free radicals, ·O 2 − and ·OH, are produced at the surface of TiO 2 and can effectively react with CHOH to form formate and the final products of CO 2 and H 2 O (Sun et al. 2010;Tasbihi et al. 2015). Thus, it has great potential to be employed as a commercial photocatalyst inside air cleaners to treat indoor CHOH (Cremer et al. 2014;Liang et al. 2012). However, because of the rapid recombination of photoinduced electron-hole pairs for TiO 2 , the photocatalytic activity of TiO 2 is not high enough to degrade CHOH efficiently (Yu et al. 2013). How to slow the recombination of photogenerated electron-hole pairs plays a vital role in enhancing the photocatalytic activity of TiO 2 .
Thus, a material with low resistance or impedance to modify TiO 2 is needed to enhance photodegrading CHOH assisted with an electric field. Fe 2 O 3 , a cheap ferromagnetic material, possesses good electrical properties, such as high capacity, high charging rate, long-life usage, and low resistance, which can effectively decrease the insulating behavior of TiO 2 (Piva et al. 2016;Ding et al. 2017). Additionally, the interface of Fe 2 O 3 and TiO 2 can accelerate the transfer of photoinduced electrons from the conduction band of Fe 2 O 3 to that of TiO 2 (Moniz et al. 2014;Wang et al. 2012Wang et al. , 2014. Thus, decorating Fe 2 O 3 on TiO 2 was employed to enhance the photoelectrical activity. Therefore, the photoelectrical catalytic oxidation of H 2 O to produce hydroxyl on TiO 2 and Fe 2 O 3 decorated TiO 2 was explored using DFT calculations. The surface characterization of photocatalysts, including transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), electron paramagnetic r esonance (EPR), was conducted. Furthermore, this study experimentally investigated the photoelectrical catalytic oxidation of gaseous CHOH by Fe 2 O 3 -decorated TiO 2 under an external electric field with and without humidity.

Experimental methods
Preparation for Fe 2 O 3 /TiO 2 Fe 2 O 3 /TiO 2 was prepared by the impregnation method. A weight of 1.00 g commercial TiO 2 (Tianhe Environmental Engineering Co., Ltd., Baoding) was mixed into 70 mL distilled water and then stirred continuously for 30 min at room temperature. Then, 0.5%, 3.0%, and 5% Fe 2 O 3 /TiO 2 samples were prepared by adding 0.025, 0.150, and 0.250 g FeCl 3 · 6H 2 O into the TiO 2 solution, respectively. After continuous stirring for 90 min at room temperature, the mixtures were dried at 105°C in a rotary evaporator at a reduced pressure of 0.01 kPa to dry the residue H 2 O. Then, the dried samples were calcinated at 500°C for 12 h in an air atmosphere to form Fe 2 O 3 /TiO 2 . Before each CHOH degrading test, the prepared Fe 2 O 3 /TiO 2 s catalysts were loaded onto glass fiber as follows: 1.0 g photocatalysts were dispersed in 1 L of distilled water with ultrasonic treatment for 10 min. Cylindrical glass fiber was immersed in the Fe 2 O 3 /TiO 2 solution for at least 5 h and then dried for 2 h at 105°C. The filled amount of Fe 2 O 3 /TiO 2 was the weight difference of the glass fiber before and after the loading treatment.
Characterization of Fe 2 O 3 /TiO 2 A scanning electron microscope (SEM) was used to obtain the morphologies of the Fe 2 O 3 /TiO 2 s with a beam accelerating voltage and current of 20 kV and 20 μA, respectively. Highresolution transmission electron microscopy (HR-TEM) was performed on a Tecnai G2 F20 (FEI, USA) with an accelerating voltage of 200 kV. X-ray diffraction (XRD) analyzer was used to analyze the crystal pattern of TiO 2 and Fe 2 O 3 /TiO 2 (s) with a Shimadzu X-ray beam with Cu-Ka radiation. Raman spectra of the specimens were obtained using a SenterraII (BRUCK, USA) with a scanning range of 100-1000 cm −1 . The photoluminescence spectra (PL) of the TiO 2 and Fe 2 O 3 / TiO 2 s catalysts were recorded on a Varian Cary Eclipse spectrometer with an excitation wavelength of 325 nm. UV-vis spectra of the photocatalysts were recorded in the range of 200-800 nm by a Shimadzu UV-vis(UV-2550) spectrometer equipped with a diffuse reflection accessory. BaSO 4 was employed as the reference substance. Electrochemical impedance spectroscopy (EIS) and linear sweep voltammetry (LSV) were performed using an electrochemical workstation (CHI604E, C.H. Inc.). The EIS and LSV were measured at a constant A.C. amplitude of 5.0 mV with a sweep frequency of 10 5 -0.1 Hz. The voltage ranged from −0.3 to 0.2 V for the LSV measurement with 0.1 M Na 2 SO 4 solution as the electrolyte. The Mott-Schottky plots were established with the A.C. amplitude of 10 mV at 100 Hz. The sweep voltage ranged from −1.2 to 0.5 V. The EPR measurements were performed using a Bruker ER200D operated in the X-band. Approximately 10-50 mg of the photocatalyst was placed into a particular spectroscopically pure quartz cell.

DFT calculation methods
The commercial TiO 2 exhibits a predominantly exposed (010) facet (Huang et al. 2019), which was also illustrated in the following TEM characterization results. Thus, an anatase TiO 2 (010) slab was constructed to simulate the TiO 2 base. Before constructing the slab, we first optimized the unit cell of anatase TiO 2 . The crystal lattice spacing was optimized to be a = b = 3.80 and c = 9.50 Å; this optimization was in agreement with previous research (Shen et al. 2021). The TiO 2 (010) slab consisted of four layers and 48 atoms (Ti 16 O 32 ), and a vacuum height of 20 Å oriented along the c-axis. The Fe 2 O 3 decoration was assembled by loading one (Fe 2 O 3 ) cluster consisting of 2 Fe and 3 O atoms on the TiO 2 (010), as shown in Fig. 9. An external electric field was applied vertically to the slab by an approach proposed by Naugebeaur and Scheffler. Positive and negative electric fields with strengths of 0.05 and −0.05 eV/Å were imposed on the supercell to simulate different electric-field directions. A positive external electric field led to a distribution of positive charges on the upper side of the slab, and the negative external electric field led to opposite results (Deshlahra et al. 2009). The formation of hydroxyl radicals with (or without) the dissociated hydrogen atom of H 2 O connected on TiO 2 (010) and Fe 2 O 3 /TiO 2 (010) was constructed as follows: for the case of hydrogen atoms anchored at the surface, the hydroxyl was connected to an unsaturated five-coordinate Ti atom (Ti 5c ) and an atop four-coordinate Fe atom (Fe 4c ), for TiO 2 (010) and Fe 2 O 3 /TiO 2 (010), respectively. The hydrogen atom was then linked to the adjacent oxygen atom of the Ti 5c (Fe 4c ) on TiO 2 (010) (Fe 2 O 3 /TiO 2 (010)). In the absence of hydrogen, which was taken away from slabs, only a hydroxyl was bonded to the Ti 5c (Fe 4c ) for the TiO 2 (010) (Fe 2 O 3 /TiO 2 (010)). The free energies of hydroxyl formation with (△G *−OH + *−H ) and without hydrogen (△G *−OH ) under an external electric field were determined by Eqs.
(2)-(3), respectively, where G *−OH + *H and G *−OH are the free energies of hydroxyl (with or without hydrogen atoms) connected to charged TiO 2 (010) and Fe 2 O 3 /TiO 2 (010) and calculated within the applied electric field. G bare* is the free energy of the charged catalysts within the electric field without any hydroxyl or hydrogen atom from H 2 O. G H2O(g) , and G H+ are the free energies of H 2 O and hydrogen atoms without any charges, respectively, which were calculated in the absence of an external electric field. U is the overpotential relative to the standard potential of a hydrogen electrode. Additionally, the difference in zeropoint energies and the entropy change in T△S were corrected by determining the vibrational frequencies of the adsorbed −OH with and without hydrogen (Valdés et al. 2008). The reaction temperature was set to 298 K. The correction for △G *−OH + *H or △G *−OH at a neutral pH and in an ambient environment was also carried out based on Nernst's equation by Eq. (4). Herein, the photocatalysts were enveloped under neutral conditions without any acid or alkaline components in the continuous injection of gaseous H 2 O. Thus, the pH was set to 7.0 to simulate the neutral environment on the gas-solid interface.
The model calculation was determined based on the generalized gradient approximation with the Perdew-Burke-Ernzerhofexchange-correlation function (GGA-PBE).
Hubbard U values (U eff ) of 4.20 and 5.00 eV were applied for Ti 3d and Fe 3d atomic orbitals, respectively, referring to previous research (Shim et al. 2009). The Brillouin zone was determined automatically with the gamma center in the Monkhorst-Pack scheme. The energy cutoff was set to 400 eV. The convergence criteria for the energy (EDIFF) and force (EDIFFG) were 10 −4 eV and −0.02 eV/Å, respectively. All atoms in the slabs were relaxed in all the structure optimizations. All the determinations mentioned above were performed with the VASP.

Photoelectrocatalytic measurements
The photoelectrocatalytic reaction system consisted of a CHOH generation system, a photoelectric catalytic reactor, and an online detector, as shown in Fig. 1. Gaseous CHOH was produced from a permeable tube enclosed in a U-shaped glass tube filled with a few glass beads to keep the influent gas cross-sectionally uniform. The U-shaped tube was placed in a water bath at a constant temperature of 80°C. Pure N 2 (99.99%) was used as the carrier gas. The influent concentration of CHOH was set to 1.50 ppm for each test. Additionally, O 2 (99.9%) and H 2 O (g) were also injected into the reactor. Different humidities of the influent gas flow were adjusted by changing the flow rate of N 2 purged into an impinger bottle containing water, which was measured by a humidity detector.
The overall influent flow rate was set to 700 mL/min. Before entering the reactor, all the gas components were thoroughly mixed in a chamber, which contained a magnetic stirrer, at a stirring rate of 200 r/min. All pipelines were made of polytetrafluoroethylene to prevent the adhesion of the CHOH. The photoelectrocatalytic reactor had a cylindrical shape with a height of 600 mm and an interior diameter of 150 mm. A quartz cover on the top of the reactor was used to receive UV irradiation. A UV lamp (λ=254 nm, P = 15 W) was placed directly above the reactor at a distance of 15 cm from the fixed catalysts, and the fluence rate (ψ) was evaluated to be 5.3 mW/cm 2 . Two pieces of round woven mesh made of stainless steel were used not only as a support for catalystcoated glass fibers but also as the positive and negative electrodes, which were connected with a self-designed direct-current (DC) voltage output device. To avoid the consumption of energy, the supplied power (P) was as low as 5 W. The output voltage between the two electrodes was measured to be approximately 150 V. Correspondingly, the strength of the electric field was determined to be 5000 V/m since the distance between the two pieces of woven mesh was approximately 3.0 cm. For each test, the fiberglass filter coated with Fe 2 O 3 /TiO 2 was sandwiched between the woven mesh. The superficial velocity for the photoelectric reaction was determined to be 3.96 cm/min. The reaction temperature inside the reactor was approximately 25°C. A working air conditioner was used to cool the reactor. A CHOH detector (4160−19.99 m, Interscan) was used to monitor the effluent concentration of CHOH online.
The removal efficiency was determined as the ratio of the difference in the influent and effluent concentrations of CHOH over the influent concentration, as shown in Eq. (5) Fig. 1 Diagram of the photoelectrical catalytic reaction system described, where C 0 (ppm) is the influent concentration of CHOH, and C i (ppm) is the effluent concentration of CHOH. The removal rate r was calculated as the mass of CHOH reduced per unit time (min) and unit mass catalyst (g) described below: where M CHOH is the molar mass of CHOH (30 g/mol), Q is the overall influent flow rate (700 mL/min), and m is the catalyst mass (g) loading on the glass fibers.

Results and discussion
Surface characterization of TiO 2 and Fe 2 O 3 /TiO 2 Morphology of TiO 2 and Fe 2 O 3 /TiO 2 Figure 2 shows the SEM and TEM morphologies of the commercial TiO 2 and 5%Fe 2 O 3 /TiO 2 . The morphologies of TiO 2 and Fe 2 O 3 /TiO 2 exhibited similar granule shapes according to SEM ( Fig. 2(a) and (e)). The TEM images of TiO 2 clearly showed that most of these granules displayed elongated tetragonal cuboids with a d-space of 0.37 nm in Fig. 2(b), which highly approached that of the exposed facet of the (010) plane (Pan et al. 2011). Thus, we inferred that the (010) plane accounted for a considerably large proportion of the exposed facet of the TiO 2 . Additionally, a minor portion of granules with a d-space of 0.24 nm was attributed to the (103) plane (Ciancio et al. 2012). The average diameter of the TiO 2 granules was statistically estimated to be 12.9±4.9 nm, as shown in Fig. 2(d). After the decoration of Fe 2 O 3 , the 3.0% Fe 2 O 3 /TiO 2 also presented a primarily cuboid shape similar to that of TiO 2 (Fig. 2(f)-(g)). The primary d-space of the 3.0% Fe 2 O 3 /TiO 2 cuboid granule was measured to be 0.37 nm, corresponding to the (010) plane, and 0.25 nm for the minor interplane space, assigned to the (110) anatase plane. Thus, the morphology of the prepared Fe 2 O 3 /TiO 2 was not substantially altered compared with that of pristine TiO 2 . Nevertheless, the average size of the 3.0% Fe 2 O 3 /TiO 2 was estimated to be approximately 14.6 ± 4.6 nm, resembling TiO 2 .
Crystal structure and raman spectra  (101), (004), (Jørgensen et al. 2007) in 0.5-5%Fe 2 O 3 /TiO 2 , implying that Fe 2 O 3 was mainly present in a highly dispersed amorphous phase, or as tiny crystals on the surface of TiO 2 that could not be detected by XRD (Fang et al. 2007). However, the decoration of Fe 2 O 3 substantially influenced the diffraction profiles of the TiO 2 s. For instance, the (020) peak intensity was much higher for Fe 2 O 3 /TiO 2 s than TiO 2 , as shown in Fig. 3(a) and Supplementary Table 1. Additionally, the (020) peak of Fe 2 O 3 /TiO 2 s was much sharper than that of TiO 2 , which was probably associated with the 14.2 ± 4.7 nm  (Liu et al. 2017), suggesting that Fe 2 O 3 loading could break the strain balance between the top and sublayers of TiO 2 , thus influencing the epitaxy processes of TiO 2 during calcination. Additionally, the full width at half maximum (FWHM) at the 25.3°peak representing the (101) plane was 0.646°, 0.593°, 0.585°, and 0.520°for TiO 2 and 0.5-5%Fe 2 O 3 /TiO 2 , respectively. The FWHM was determined according to the Debye-Scherrer equation (Eq. (7)), where D (nm) is the average particulate size and β (without the value) is the FWHM. The particle size of 0.5-7%Fe 2 O 3 / TiO 2 increased slightly, since the β for the peak corresponding to the Fe 2 O 3 /TiO 2 's (101) plane was larger than that of TiO 2 , unveiling that decoration of Fe 2 O 3 slightly promoted the growth in TiO 2 size, highly concurring with the size distribution as shown in the TEM images. Figure 3(b) shows the Raman spectra of TiO 2 and 0.5-5% Fe 2 O 3 /TiO 2 . The Raman peaks at 144, 396, 515, and 640 cm −1 corresponded to the Eg, B 1 g, B 2 g, and Eg modes of TiO 2 . The characteristic peaks of Fe 2 O 3 were not present in the Raman patterns of 0.5-5% Fe 2 O 3 /TiO 2 such as those for α-Fe 2 O 3 occurring at 220 (A 1g ), 243 (E g ), 290 (E g ), 408 (E g ), 611 (E g ), 638 (E g ), and 1318 cm −1 (E g ) (Kodan et al. 2019), as well as those for γ-Fe 2 O 3 at 350 (E g ), 395 (T 2g ), 507 (T 2g ), 670 (A 1g ), and 708 cm −1 (A 1g ) (Pawan et al. 2014). The similar Raman patterns between TiO 2 and 0.5-5%Fe 2 O 3 /TiO 2 agreed with the XRD results, revealing that the decorated Fe 2 O 3 on TiO 2 formed an amorphous phase or tiny crystallite cell. Nevertheless, the intensities of the 144 and 130 cm −1 peaks assigned to the symmetric stretching vibration of the O-Ti-O bonds were relatively lower for Fe 2 O 3 /TiO 2 than for the TiO 2 s, particularly that of 144 cm −1 in 5% Fe 2 O 3 /TiO 2 spectrum, which dropped notedly. The lower intensity of the E g vibration mode of the Fe 2 O 3 /TiO 2 s probably resulted from the decrease in the O-Ti-O number on the exposed facet of anatase TiO 2 due to the surface cover of amorphous Fe 2 O 3 .   Fig. 3 a XRD and b Raman patterns of TiO 2 and 0.5-5%Fe 2 O 3 /TiO 2 were assigned to Ti 4+ 2p 1/2 and 3/2 of TiO 2 , respectively (Tsai et al. 2013). After the surface loading of Fe 2 O 3 , two small subpeaks emerged at lower binding energies of 463.4 and 457.7 eV, suggesting that the electron density of the Ti 4+ 2p orbital increased with the loading of Fe due to the formation of Fe-O-Ti. The intensity of the Fe 2p peak, as shown on the Y-axis scale in Fig. 4(b), was much smaller than that of Ti 4+ , indicating that the content of surface Fe atoms was far lower than that of Ti. The main peaks at approximately 710.2 and 724.6 eV were ascribed to the Fe 3+ 2p1/2 and 2p3/2 orbitals of Fe 2 O 3 , respectively, illustrating that Fe 3+ dominated most surface Fe atoms. Additionally, minor shake-up peaks were present in the range of 713-720 eV, which resulted from an electron transition from a 3d orbital to an empty 4s orbital as the ejection of the Fe 2p electrons (Yin et al. 1974).  of the 0.5%, 3%, and 5%Fe 2 O 3 /TiO 2 dropped to 3.14, 2.99, and 2.87 eV, respectively. The UV-vis spectrum of Fe 2 O 3 illustrated a noticeable redshift compared with those of TiO 2 and 0.5-5% Fe 2 O 3 /TiO 2 . The adsorption edge occurring at 530-650 nm of the pristine Fe 2 O 3 is probably related to the ligand field transition of 2( 6 A 1 )→( 4 T 1 ) (Huang et al. 2017). The band gap was approximately 1.84 eV, approaching 1.88 eV, as reported by Wang et al. (Mansour et al. 2020). Thus, the redshift for 0.5-5.0% Fe 2 O 3 /TiO 2 compared with TiO 2 was principally attributed to the loading of the Fe 2 O 3 cluster.

UV-Visible spectra
Photoluminescence spectra and electric properties   Fig. 6(b). The simulated EIS radius of TiO 2 was much larger than those of Fe 2 O 3 /TiO 2 s, illustrating that the decoration of Fe 2 O 3 greatly reduced the interface charge transfer resistance between the electrode and electrolyte, which was probably attributed to the lower electronic resistivity of Fe 2 O 3 (Chang et al. 2014). Moreover, the simulated EIS decreased with the content of surface Fe 2 O 3 , probably owing to an increase in tiny crystals within the Fe 2 O 3 , which could significantly reduce the electronic resistivity, although the XRD results did not discern the αor γ-Fe 2 O 3 crystalline phase (Piva et al. 2016).
The LSV spectra of TiO 2 and Fe 2 O 3 /TiO 2 s were measured under UV illumination, as illustrated in Fig.  6(c). The photocurrent density of TiO 2 was much lower than that of Fe 2 O 3 /TiO 2 in −0.25-0.3 V. Additionally, the photocurrent density of the specimens also increased   Fig. 6 a PL spectra, b simulated EIS, and c LSV patterns of TiO 2 and 0.5-5%Fe 2 O 3 /TiO 2 with the content of Fe 2 O 3 . The EIS and LSV results confirmed that the surface decoration of Fe 2 O 3 could promote the separation efficiency of photogenerated electron-hole pairs of TiO 2 .

Electronic structure
The Mott-Schottky plots for TiO 2 and 0.5-5%Fe 2 O 3 /TiO 2 were constructed at 100 Hz using an Ag/AgCl electrode at pH=7.0, as shown in Fig. 7(a). The results illustrated that the slope of 1/C 2 versus the applied potential was positive for both TiO 2 and 0.5-5%Fe 2 O 3 /TiO 2 , demonstrating that they were attributed to n-type semiconductors. Thus, the edges of the flat band (FB) of TiO 2 and 0.5-5%Fe 2 O 3 /TiO 2 are nearly equal to those of the conduction band (CB), as shown in Eq. (8) The CBs of TiO 2 and 0.5-5%Fe 2 O 3 /TiO 2 were −0.86, −0.91, −1.01, and −1.07 eV, respectively. The corresponding edges of the valence band (VB) could be obtained from the band gap (Eg) and CB as described by Eq. (9), The band gaps of TiO 2 and 0.5-5%Fe 2 O 3 /TiO 2 were 3.20, 3.14, 2.99, and 2.87 eV, respectively. Their VBs were determined to be 2.34, 2.23, 1.98, and 1.80 eV, respectively, as shown in Fig. 7(b). The CBs of 0.5-5%Fe 2 O 3 /TiO 2 were higher than that of TiO 2 (−0.86 eV). This result is possibly related to the realignment of the two Fermi levels of Fe 2 O 3 and TiO 2 after the Fe 2 O 3 and TiO 2 were contacted (Liu et al. 2015). As shown in the schematic diagram of the electronic structure of hybrid Fe 2 O 3 /TiO 2 (see Fig. 7(c)), the photogenerated electrons transferred from the conduction band of Fe 2 O 3 to that of TiO 2 , while the holes moved from the valence band of TiO 2 to Fe 2 O 3 . Thus, Fe 2 O 3 /TiO 2 is beneficial for separating the photogenerated electron-hole pairs.

EPR
The typical feature of the •OH signal displays a quartet pattern with an intensity ratio of 1 (g ≈ 1.990): 2 (g ≈ 2.003): 2 (g ≈ 2.012): 1 (g ≈ 2.021) for the TiO 2 and 3%Fe 2 O 3 /TiO 2 , as depicted in Fig. 8. The results indicated that a higher intensity was ascertained on the 3%Fe 2 O 3 /TiO 2 than on TiO 2 , implying that the decoration of Fe 2 O 3 was more conducive to increasing •OH, which was attributed to the efficient photoinduced electron-hole pairs on the interface of Fe 2 O 3 and TiO 2 (Sun et al. 2012).

DFT calculations
Figure 9(a) shows the effect of the electric field on the free energy of formation of hydroxyl and hydrogen on TiO 2 (010) and Fe 2 O 3 /TiO 2 (010). In the absence of an electric field, energies of 1.56 and 0.70 eV were required for the formation of hydroxyl and hydrogen on TiO 2 (010) and Fe 2 O 3 /TiO 2 (010), respectively, indicating that the decomposition of H 2 O to produce hydroxyl and hydrogen could not proceed spontaneously on these slabs; this decomposition reaction resembled the first reaction step of the OER for anatase TiO 2 in aqueous conditions (Malik et al. 2020). Nevertheless, for Fe 2 O 3 /TiO 2 (010), the free energy of *−OH + *−H formation decreased by approximately half compared with TiO 2 . The top Ti-O bond length (2.427 Å) was also reduced to 2.123 Å due to the decoration of amorphous Fe 2 O 3 , meaning that atop Fe 2 O 3 on TiO 2 (010) was more active for splitting H 2 O than TiO 2 (010). When a positive external electric field with a strength of 0.05 eV/Å was applied for TiO 2 (010), the free energy (1.44 eV) of *−OH + *−H formation decreased slightly. It is worth noting that the free energy of formation on Fe 2 O 3 /TiO 2 (010) decreased dramatically to −1.04 eV with a positive electric field, thus spontaneously driving the splitting of H 2 O into hydroxyl radicals.
To better understand the effect of the electric field on the formation of hydroxyl and hydrogen, the variation in dipole moment (μ 0 ) of hydroxyl and hydrogen as a whole adsorbed on TiO 2 (010) and Fe 2 O 3 /TiO 2 (010) was compared. μ 0 is equivalent to the difference between the dipole moment of *−OH + *−H (*−OH) and that of the bare catalyst slab (Deshlahra et al. 2009), which were both obtained from the resulting DFT file. Without an electric field, the dipole moments of *−OH + *−H for TiO 2 (010) and Fe 2 O 3 /TiO 2 (010) were 0.019 and 0.171 eÅ, respectively, which suggested the more negative charges were distributed on the oxygen atoms of the hydroxyl and on the hydrogen atoms of the Fe 2 O 3 / TiO 2 (010) than those of the TiO 2 (010). The change in free energy (ΔE, eV) of *−OH + *−H formation due to exposure to an electric field can be predicted by the first-order Stark effect as where μ 0 (eÅ) is the dipole moment of the adsorbate without an electric field and F (V/Å) is the strength of an external electric field. The ΔE was −0.00865 and −0.00097 eV for Fe 2 O 3 /TiO 2 (010) and TiO 2 (010), respectively, according to Eq. (10), when the electric field was set to 0.05 eV/Å. Although both ΔE values obtained by Eq. (10) were much smaller than those calculated based on the structure optimization results, the ΔE of Fe 2 O 3 /TiO 2 (010) was higher than that of TiO 2 (010); this trend demonstrated that the external electric field in the positive direction favors the formation of hydroxyl and hydrogen on the surface of Fe 2 O 3 /TiO 2 (010  Fig.  9(a), suggesting that the hydroxyl was bonded more closely to the Fe 2 O 3 /TiO 2 (010) under a positive electric field. It is noted that with the positive electric field, the dissociated hydrogen atom from H 2 O preferred to interact with one Fe atom at a shorter distance of 2.392 Å, while before the electric field was added, the hydrogen atom was physically connected to one oxygen atom of Fe 2 O 3 at a distance of 2.994 Å. In contrast, a more stable structure was assigned to an unbroken H 2 O molecule adsorbed on TiO 2 (010) without the formation of hydroxyl and hydrogen moieties under a positive electric field. The distance between the surface oxygen and surface Ti atom was elongated to 3.208 Å, suggesting that H 2 O was more difficult to oxidize and split into hydroxyl on TiO 2 (010) than on Fe 2 O 3 /TiO 2 (010) under a positive electric field. The effect of a negative electric field on the formation of hydroxyl was also investigated. According to Eq. (10), the predicted ΔE for the change in the formation energy would be 0.00865 and 0.00097 eV for TiO 2 (010) and Fe 2 O 3 / TiO 2 (010) under a field strength of −0.05 V/Å, indicating that an electric field in the negative direction was not beneficial for splitting H 2 O compared with the positive electric field. The formation free energies of the hydroxyl and hydrogens were also determined based on structure optimization, as shown in Fig. 9(a), which were 1.01 eV for TiO 2 (010) and −0.22 eV for Fe 2 O 3 /TiO 2 (010). They were much lower than those under a positive electric field, demonstrating that a positive electric field was more effective than a negative electric field to promote hydroxyl production. Additionally, the atomic distance between the oxygen atom of the hydroxyl and the Fe or Ti atom, and the distance between the dissociated hydrogen atom and the Fe atom became longer under a negative electric field, as shown in Fig. 9(a).
When the detached hydrogen atom of H 2 O was lost under the gaseous flow, the production of hydroxyl could not still proceed spontaneously regardless of the direction in which the electric field was applied, as illustrated in Fig. 9(b). For instance, the free energy of *−OH formation for TiO 2 (010) was 1.759 eV, while it was 1.445 eV for Fe 2 O 3 /TiO 2 (010). The free energy of *−OH formation alone increased to 2.927 eV and 2.423 eV for TiO 2 (010) and Fe 2 O 3 /TiO 2 (010), respectively, when a positive electric field was added. Likewise, the  Fig. 9 Free energy of formation of a slab + hydroxyl + H and b slab + hydroxyl negative electric field resulted in free energies of 1.445 eV and 0.7817 eV. The results implied that the hydrogen atom that was dissociated from the H 2 O molecule was more likely to remain on the TiO 2 (010) and Fe 2 O 3 /TiO 2 (010) than to be taken away by flow gasses. Figure 10 illustrates the electric field's influence on the density of states of Fe 2 O 3 /TiO 2 and Fe 2 O 3 /TiO 2 (010) with the adsorbed hydroxyl and hydrogen. The band gap of Fe 2 O 3 /TiO 2 was approximately 2.30 eV, as shown in the total density state Fe 2 O 3 /TiO 2 in Fig. 10(a). No distinct energy state emerged within the band gap of Fe 2 O 3 /TiO 2 , concurring with the PL profile of Fe 2 O 3 /TiO 2 . Additionally, the projected density state showed that the profile of Fe 3d mainly resided on the edge of the conduction band. Figure 10(b)-(d) shows that the band gaps of Fe 2 O 3 /TiO 2 (010) with adsorbed hydroxyl and hydrogen with and without an electric field were all 2.30 eV, suggesting the external electric field did not change the magnitude of the band gap of Fe 2 O 3 /TiO 2 (010). However, without the electric field, a few energy states were distributed within the band gap for the Fe 2 O 3 /TiO 2 (010) that had adsorbed hydroxyl and hydrogen; these energy states were mainly derived from the interaction between Fe 3d and the sp hybrid orbitals of the surface hydrogen and oxygen atoms, indicating a strong bond formed between the Fe and atom oxygen and hydrogen, as shown in Fig. 10(b). However, once the positive electric field was applied to the Fe 2 O 3 /TiO 2 (010) with adsorbed hydroxyl and hydrogen, most of the band gap density states vanished, particularly at the higher energy level. Only two states closer to the valance band margin remained, revealing that the bond energy between the Fe3d and hybrid OH orbitals became more stable under a positive electric field, thus promoting the bond to be much stronger (see Fig. 10(c)). In contrast, if a negative electric field was added, the location and number of the energy state remained almost unchanged, suggesting that a negative direction less influenced the bond of the Fe 3d and OH hybrid orbitals, as illustrated in Fig. 10(d).
The Fermi level of Fe 2 O 3 /TiO 2 , which is defined herein as the top of the valence band, was −0.72 eV, while that of Fe 2 O 3 /TiO 2 + hydroxyl + hydrogen was 1.12 eV, as shown in Fig. 10(a) and (b). The remarkable shift of the Fermi level toward the conduction band margin for Fe 2 O 3 /TiO 2 + hydroxyl + hydrogen resulted from the new mid-gap state of the interaction between Fe 3d and atop hydrogen and oxygen atoms. When an external positive electric field was applied to the slab, the Fermi level (0.09 eV) decreased markedly in comparison to that without an electric field, as shown in Fig. 10(c). This implied that the positive electric field attracted more positive charges that accumulated on the surface of Fe 2 O 3 /TiO 2 , thus decreasing the Fermi level (Tiewcharoen et al. 2017), which was beneficial for splitting H 2 O. In contrast, under the negative electric field, the Fermi level shifted toward a much higher energy level, reaching 1.40 eV, suggesting that the negative electric field resulted in more negative electric charges distributed on the surface, thus notedly lifting the Fermi level, but decreasing the formation of hydroxyl and hydrogen.

Photoelectric activity
The photocatalytic and photoelectric removal efficiencies of CHOH by neat TiO 2 , and 0.5-5.0%Fe 2 O 3 /TiO 2 were determined without the addition of H 2 O, as illustrated in Fig. 11. Figure 11(a) displays that without UV irradiation, the removal efficiency by neat TiO 2 was approximately 64%, suggesting that even if no UV was exploited, TiO 2 could adsorb CHOH via the interactions between the surface Ti of TiO 2 and the  Fig. 10 Projected density of state for bare Fe 2 O 3 /TiO 2 a, Fe 2 O 3 / TiO 2 + hydroxyl + hydrogen b without an electric field, c Fe 2 O 3 / TiO 2 + hydroxyl + hydrogen with a positive electric field, and d with a negative electric field, respectively. Dashed lines denote the Fermi level oxygen atom of CHOH, as well as the surface oxygen and carbon atom of CHOH. Then, the two hydrogen atoms of CHOH were further separated (Wu et al. 2018). The isolated fragments of C-O and H then reacted with O 2 to produce CO 2 and H 2 O. The Fe 2 O 3 /TiO 2 s had a much higher removal efficiency than neat TiO 2 , reaching more than 80%, particularly for 3%Fe 2 O 3 /TiO 2 , which had the highest removal efficiency of 86% among these catalysts. The enhancement of the removal efficiency by Fe 2 O 3 /TiO 2 was probably due to the stronger interaction between the surface Fe 2 O 3 /TiO 2 and CHOH than between TiO 2 and CHOH, whose adsorption energies were determined to be −1.49 and −1.08 eV, respectively, as illustrated in Figure S-1. When an external voltage of 150 V was introduced into the reactor, the removal efficiencies were nearly the same as those without the addition of applied voltage for both TiO 2 and Fe 2 O 3 /TiO 2 s, conveying that an external voltage of 150 V on the catalysts could not directly influence the interaction between CHOH and TiO 2 (Fe 2 O 3 /TiO 2 s). Figure 11(b) shows that when the UV irradiation was added, the photocatalytic removal efficiencies of CHOH were approximately 68% and 85-97% by TiO 2 and Fe 2 O 3 /TiO 2 s, respectively, an overall increase of 5-10% compared with their removal efficiency without UV irradiation. The increase in the removal efficiencies was attributed to the oxidation of CHOH by hydroxyl radicals derived from the photoinduced holes with surface adsorbed H 2 O, as described by Eqs. (11)-(12), Additionally, the increment contributed by Fe 2 O 3 /TiO 2 s was 7-10%, much higher than 5% by TiO 2 , owing to the interfacial transportation of photoinduced electrons and a lower electrical impedance of Fe 2 O 3 /TiO 2 . When the electric field and UV irradiation were simultaneously applied to the reaction system, the removal efficiencies of CHOH were not further enhanced, suggesting that the interaction between the surface Ti and Fe atoms and CHOH could not be influenced by such a relatively low electric field, even if electrons had moved freely within their conduction band.
When a mixed flue gas with a relative humidity of 50% was supplied into the reactor, the removal efficiencies of CHOH by TiO 2 and 0.5-5%Fe 2 O 3 /TiO 2 under irradiation or not were all depressed compared to those without H 2 O interference, as illustrated in Fig. 12(a) and (b), indicating that H 2 O caused competitive adsorption with CHOH. Nevertheless, either the catalytic (41%) or photocatalytic (51%) removal efficiencies of CHOH by TiO 2 were lower than 54-60% and 70-80% by 0.5-5%Fe 2 O 3 /TiO 2 , implying that the decoration of Fe 2 O 3 can relieve H 2 O inhibition to some degree. When an external voltage of 150 V was introduced, the catalytic removal efficiencies of CHOH by TiO 2 and Fe 2 O 3 /TiO 2 were still not enhanced without irradiation. Noticeably, their photocatalytic activity was markedly promoted once UV irradiation was added, and the removal efficiencies had increased by approximately 5% and 10% for TiO 2 and Fe 2 O 3 /TiO 2 , respectively. Consistent with the DFT results, the combined effect of the applied electric field and UV irradiation on the Fe 2 O 3 /TiO 2 was more beneficial for enhancing the removal efficiency of CHOH than TiO 2 . Additionally, the pH of the effluent gas during the photoelectrical catalytic degradation of CHOH under a relative humidity of 50% was sampled at the outlet and measured to be neutral, experimentally revealing that the protons were not taken away by the carrier gasses with the external electric field. Additionally, after the reaction finished, we observed that the Fe 2 O 3 /TiO 2 retained the same faint, slightly yellowish as it had when it was fresh, revealing that the electric field-assisted photooxidation of CHOH on Fe 2 O 3 /TiO 2 does not cause an accumulation of photogenerated electrons or holes on the Fe 2 O 3 /TiO 2 , therefore avoiding drastic changes in its structure. The photogenerated holes were principally distributed on Fe 2 O 3 , which finally participated in the formation of hydroxyl radicals to oxidize CHOH. The photogenerated electrons trapped by the Ti 3+ had different reaction pathways, dominantly including (a) reacting with O 2 to produce ·O 2 − for CHOH oxidation or directly participating in the attachment of the carbonyl of CHOH to  Ti 3+ (Cremer et al. 2014), or (b) moving toward the positive electrode of the input DC device, particularly for those produced by the TiO 2 or Fe 2 O 3 /TiO 2 nanogranules having physical contact with the mesh plates.
Additionally, the photoelectrical removal efficiencies of CHOH under different relative humidity (20-80%) conditions were much higher than their photooxidation efficiency of CHOH, as illustrated in Figure S-2. Therefore, combining the DFT and EPR results, the resistance to the competitive adsorption of H 2 O was attributed to the more hydroxyl radicals were photoelectrical splitting of H 2 O molecules.
By adjusting the distance between the two mesh plates to change the electric field's strength, we investigated the relationship between the applied electric field's strength and the incremental change in removal efficiency and rate compared to those of the samples without an applied electric field, as shown in Fig. 13(a). The results illustrated that the removal efficiencies of CHOH by 3%Fe 2 O 3 /TiO 2 were 74.3±1.83%, 76.8±1.57%, 86.8±1.30%, and 89.9±1.57%, under electric field strengths of 0, 2500, 5000, and 7500 V/m, respectively, suggesting an increase in the removal efficiencies with the electric field strength. However, the relationship between the increments of removal rate and the electric field magnitude was not a good linear relationship, as illustrated in Fig.   13(b). The results implied that the formation of free energy of hydroxyl depends on the electric field strength, and other reaction parameters such as the energy barrier are possibly influenced.
The removal rates of CHOH with various low levels of power input were compared in Fig. 14. Without competitive adsorption of H 2 O, 15-W lump irradiation could improve the removal rates by 7% and 10%, reaching maximum removal rates of 2954 and 4129 ng/g/min for TiO 2 and 3% Fe 2 O 3 / TiO 2 , respectively, while the 5-W electric field contributed little to the enhancement. Under a relative humidity of 50%, the removal rates of CHOH significantly dropped for both TiO 2 and Fe 2 O 3 /TiO 2 without any input energy. However, after introducing 15-W irradiation, the removal rates of CHOH increased by 26% and 24% for TiO 2 and 3% Fe 2 O 3 / TiO 2 , respectively. Moreover, the 5-W electric field was further added, the removal rates increased further by 13% and 23%, particularly for 3% Fe 2 O 3 /TiO 2 , which had nearly the same efficacy as the 15-W input power. Additionally, the final photoelectrocatalytic oxidation rate reached 3747 ng/g/min for 3% Fe 2 O 3 /TiO 2 , which approached its highest H 2 O-free removal rate of 4130 ng/g/min. Therefore, a total of 20-W low-power input was verified to effectively resist the inhibition of H 2 O on the CHOH removal by 3%Fe 2 O 3 /TiO 2 .