Fabrication of Adsorbed Fe(III) and Structurally Doped Fe(III) in Montmorillonite/TiO 2 Composite for Photocatalytic Degradation of Phenol

: The Fe(III)-doped montmorillonite (Mt)/TiO 2 composites were fabricated by adding Fe(III) during or after the aging of TiO 2 /Ti(OH) 4 sol–gel in Mt, named as x Fe-Mt/(1 − x )Fe-TiO 2 and Fe/Mt/TiO 2 , respectively. In the x Fe-Mt/(1 − x )Fe-TiO 2 , Fe(III) cations were expected to be located in the structure of TiO 2 , in the Mt, and in the interface between them, while Fe(III) ions are physically adsorbed on the surfaces of the composites in the Fe/Mt/TiO 2 . The narrower energy bandgap ( Eg ) lower photo-luminescence intensity were observed for the composites compared with TiO 2 . Better photocatalytic performance for phenol degradation was observed in the Fe/Mt/TiO 2 . The 94.6% phenol degradation was due to greater charge generation and migration capacity, which was conﬁrmed by photocurrent measurements and electrochemical impedance spectroscopy (EIS). The results of the energy-resolved distribution of electron traps (ERDT) suggested that the Fe/Mt/TiO 2 possessed a larger amorphous rutile phase content in direct contact with crystal anatase than that of the x Fe-Mt/(1 − x )Fe-TiO 2 . This component is the fraction that is mainly responsible for the photocatalytic phenol degradation by the composites. As for the x Fe-Mt/(1 − x )Fe-TiO 2 , the active rutile phase was followed by isolated amorphous phases which had larger ( Eg ) and which did not act as a photocatalyst. Thus, the physically adsorbed Fe(III) enhanced light adsorption and avoided charge recombination, resulting in improved photocatalytic performance. The mechanism of the photocatalytic reaction with the Fe(III)-doped Mt/TiO 2 composite was proposed.


Introduction
Mt and Fe(III) are abundant in nature, while Ti is one of the most abundant elements (the ninth most abundant element (0.63% by mass)), and the seventh most abundant metal in Earth's crust. Ti can be found in almost all living things, as well as in bodies of water, rocks, and soils. The most common compound is titanium dioxide (TiO 2 ), a popular photocatalyst that has long been studied and applied to scientific research and industry. TiO 2 is highly photocatalytic-active, non-toxic, chemically stable in most conditions, and has a low cost. However, the relatively large energy bandgap (Eg) and the recombination of electron (e − ) and hole (h + ) reduce the photocatalytic activity of TiO 2 . In addition, the nanoparticles of TiO 2 tend to aggregate, making it difficult to separate from water. To overcome these difficulties, numerous strategies have been applied, such as morphology modification [1,2], surface sensitization [3,4], metal nanoparticle deposition [5,6], element doping [7,8], and the preparation of composite materials with other materials [9,10].
Montmorillonite (Mt) is a typical supporting material to make composites with TiO 2 to avoid aggregations of TiO 2 particles, because the TiO 2 particle sizes were reduced and had a better distribution in the Mt layers. Moreover, the significantly reduced fine particle sizes can improve the photocatalytic performance of TiO 2 [11][12][13]. Mt is a 2:1 clay mineral, Na-Mt (Kunipia-F) with a cation exchange capacity (CEC) of 1.114 mmol/g (Bergaya and Vayer, 1997) was bought from Kunimine Industries Co. Ltd. (Tokyo, Japan). Through an elemental analysis (method of ME-ICP61, ALS Global Ltd., North Vancouver, Canada), the chemical formula of Kumipia-F is given by (Na 0.97 Ca 0.08 ) +1 and Millipore water with pH 6.31 were purchased from Kishida Chemical Co., Ltd. (Osaka, Japan) and Synergy UV (Merck, Darmstadt, Germany), respectively.

Preparation of Iron-Modified Mt and Clay/TiO 2 Composites
To synthesize protonated Mt (H-Mt) and Fe(III)-doped Mt (Fe-Mt), Mt was added to HNO 3 or Fe(III) (Fe (NO 3 ) 3 ·9H 2 O, 99%) solution which has a pH of 2 and was vigorously stirred at room temperature for 1 h (S/L = 10 g/L). Next, the H-Mt and Fe-Mt were centrifuged, rinsed, collected, lyophilized, ground, and sieved (<149 µm). Leveraging the sol-gel technique, Ti (OH) 4 /TiO 2 nanoparticles were generated in aqueous media, following which TiCl 4 and HCl were then added to Milli-Q water. The resultant solution contained 0.83 M TiCl 4 and 1.0 M HCl, which was further stirred for 1 h and left to stand still for 6 h. Through hydrolysis, a nearly transparent, stable Ti (OH) 4 /TiO 2 sol-gel was attained.
Before heterocoagulation, the acidity of Ti (OH) 4 /TiO 2 sol was altered to pH = 2 with the addition of a NaOH solution since it has an original pH value of around 0. Next, the Sol-gel was added to 1% (g clay/100 mL of water) H-Mt suspension. Meanwhile, the Fe(III) eqv. to 0.274 mmol/g Mt was added for xFe-Mt/(1 − x)Fe-TiO 2 samples exclusively. Both kinds of samples were vigorously stirred for 30 min. The theoretical concentration of TiO 2 would be 30% (m/m). Heterocoagulated Fe/Mt/Ti(OH) 4 -TiO 2 (Fe/Mt/TiO 2 ) or xFe-Mt/(1 − x)Fe -Ti(OH) 4 -TiO 2 (xFe-Mt/(1 − x)Fe-TiO 2 ) precipitated after 20 h of sedimentation at 70 • C, 80 • C, and 90 • C. Fe(III) was added for Fe/Mt/TiO 2 samples immediately after sedimentation. All the suspensions were stirred for another 1 h, followed by centrifugation at 8300 rpm for 30 min. The obtained solid substance was rinsed with Milli-Q water, freeze-dried for 24 h, ground, and sieved (<149 µm). Residual Fe(III) in the supernatant was detected by the inductively coupled plasma-optical emission spectroscopy (ICP-OES, Perkin Elmer 8500, Waltham, MA, USA). The samples synthesized at 70 • C were the main focus for comparison because the synthesis temperature did not influence the photocatalytic performance of the Fe/Mt/TiO 2 composites ( Figure S1).

Solid Characterization
The resultant composites were inspected with X-ray fluorescence (XRF, Shimadzu-EDX800, Kyoto, Japan). Powder X-ray diffraction (XRD) patterns of materials were recorded with an Ultima IV X-ray diffractometer (Rigaku, Akishima, Japan) by Cu Kα radiation produced at 40 kV and 40 mA, a divergence slit of 1.0 mm, an anti-scatter slit of 10 mm, and a receiving slit of 0.15 mm over a 2θ of 2.0-10 • , with a step size of 0.02 • and a scanning speed of 2 • /min. X-ray photoelectron spectroscopy (XPS) was carried out on an ESCA 5800 system (Ulvac-PHI, Kanagawa, Japan) with a monochromatic Al Kα X-ray source at 200 W. A survey scan was carried out from 0 to 1000 eV with passing energy of 187.85 eV, narrow scans of N 1 s, as well as C 1 s orbitals, which were acquired using passing energies of 23.5 and 58.7 eV. Peak separation was performed with Casa XPS software (Version 2.3.16 PR 1.6) following the subtraction of a Shirley baseline. Binding energies (EB) of peaks were calibrated using EB [C 1s] = 284.6 eV. The scanning electron microscopy (SEM) and energy dispersive X-ray analysis (EDX) investigations were performed on a VE-9800 scanning electron microscope (Keyence, Osaka, Japan) with an accelerating voltage of 20 kV. The transmission electron microscopy (TEM) and EDX were applied using a JEM-2100HCKM, JEOL (Akishima, Japan). Diffuse reflectance spectroscopy (DRS) was conducted with a range of 200-800 nm on a UV-2450 spectrophotometer which has a diffuse-reflectance attachment with BaSO 4 as a reference. Solid-state photoluminescence spectroscopy (PL) was conducted with a JASCO F-6600 spectrofluorometer. Electrochemical properties, such as photocurrent response and the electrochemical impedance spectroscopy (EIS) of composites, were examined by the 1280c AMETEK advanced measurement technology (Berkshire, United Kingdom). Later, 200 mg of Fe/Mt/TiO 2 and xFe-Mt/(1 − x)Fe-TiO 2 composites were included in a periodic acid-Schiff (PAS) cell which contained an electret condenser microphone and a quartz window on top under N 2 flow saturated with methanol vapor for a minimum of 30 min. The light beam from a Xe lamp with a grating monochromator modulated at 80 Hz using a light chopper was emitted in the range of 650 nm-350 nm through the cell window to identify the PAS signal with a digital lock-in amplifier. The energy-resolved distribution of electron traps (ERDT) was obtained by figuring out the amount of photo absorption change for the accumulated electrons.

Photocatalytic Activity Test
The photocatalytic activities of synthesized products were tested for phenol degradation under the irradiation of a 300 W xenon lamp. Before the irradiation, 20 mg of the photocatalyst was suspended in 100 mL of 10 mg/L phenol aqueous solution in a photoreactor with a cooling water jacket outside. The suspensions were stirred at 500 rpm in darkness for 5 min to arrive at an adsorption-desorption equilibrium. In the irradiation process, approximately 1 mL of suspension was collected from the reaction cell at proper intervals and was subsequently filtered by a 0.45-µm cellulose acetate membrane filter for removal of the photocatalyst. Afterward, the solutions were examined by high-performance liquid chromatography (HPLC) [38]. The concentrations of phenol, together with its degradation products, were estimated by evaluating the absorption of these substances at a wavelength of 270 nm. Scavenger tests and the reusability test were carried out following the aforementioned procedures. For the scavenger test, 50 mmol of ethanol, isopropanol, and para-benzoquinone was added before the addition of photocatalytic materials.

Characterizations
After the filtration of the synthesized Fe/Mt/TiO 2 and xFe-Mt/(1 − x)Fe-TiO 2 , the remaining Fe(III) was detected by ICP-OES. The Fe(III) concentration was below the detection limit of the ICP, suggesting that almost all the added Fe(III) was doped in the composites. Therefore, the doped amount of Fe(III) in all the composites were the same and should not be the impact factor influencing the photocatalytic character of the samples. XRF was applied to calculate the weight percentage of the TiO 2 phase in the whole composite (Table 1). When comparing the Fe/Mt/TiO 2 and xFe-Mt/(1 − x)Fe-TiO 2 , the Fe/Mt/TiO 2 always kept a slightly higher amount of TiO 2 than xFe-Mt/(1 − x)Fe−TiO 2 , but the difference was not significant. The highest TiO 2 content was obtained when the product was synthesized at 80 • C. When compared with clay/TiO 2 composites synthesis without Fe(III) in previous research [39][40][41], the optimal temperature for the highest TiO 2 yield was at 70 • C. The presence of Fe(III) in composites required a higher synthetic temperature for the highest TiO 2 yield. The XRD patterns of the original Mt, Fe(III) cation-doped Mt (Fe-Mt, synthesized at ambient temperature), TiO 2 , Fe/Mt/TiO 2 , and xFe-Mt/(1 − x)Fe-TiO 2 synthesized at 70 • C were analyzed, and the XRD pattern of these samples was shown in Figure 1. The 001 peaks from Mt included the interlayer distance and the thickness of one layer (0.96 nm) [42]. The interlayer distances in the Fe-Mt, Fe/Mt/TiO 2 , and xFe-Mt/(1 − x)Fe-TiO 2 were almost the same value (0.42/0.43 nm) when compared with the original Mt (0.25 nm), which suggested that only ion exchange of Fe(III) with the interlayer cations in Mt occurred during the synthesis of the composites. Additionally, the TiO 2 should only be distributed on the surface of Mt. The hydrated ionic sizes of Fe(III) are 0.45 nm [42]. The smaller interlayer distances of the Mt in the composites, rather than the ionic sizes, may be distributed from the Van der Waals force among the aluminosilicate layers extruding from the hydrated shell of Fe(III), which are imperfectly shaped [42].  [43]. The formation of TiO 2 in the composites by different synthetic temperatures was elucidated, and their XRD patterns were presented in Figure S2. Peak separation was utilized to further analyze the XRD patterns in more detail in Figure 2. The typical 101 diffraction of anatase at 25.4 • and the 110 diffraction of rutile between 28.2-28.4 • were well fitted. The parameters after peak fitting were listed in Table S1. The large full width at half-maximum (FWHM) values for anatase and rutile related to the broad peaks of all the samples suggested a low crystallinity of the TiO 2 phase on the Mt layers. The TiO 2 content in xFe-Mt/(1 − x)Fe-TiO 2 at three synthesized temperatures was higher than in Fe/Mt/TiO 2 , and the Fe(III) in TiO 2 can improve the crystallization of the anatase phase. Increasing the temperature can improve the crystallization of anatase in Fe/Mt/TiO 2 but depress it in xFe-Mt/(1 − x)Fe-TiO 2 from 80 • C to 90 • C. As proven by ICP analysis, only the location of Fe(III) can dominate the growth of TiO 2 phases on the aluminosilicate layers of Mt. For the following characterizations, the Fe/Mt/TiO 2 and xFe-Mt/(1 − x)Fe-TiO 2 synthesized at 70 • C were applied to compare with previous studies.  [47][48][49]. The peak raised at 102.5 in both composites, as is shown in Figures S3f and S4f, are both assigned to tetrahedral Si-O [50,51]. The valence band (VB) on top of the Fe/Mt/TiO 2 and xFe-Mt/(1 − x)Fe-TiO 2 was estimated by drawing the straight tangent line to the VB spectrum of the samples in Figure 5 [52,53], and the values are 1.72 and 1.95 eV, respectively.     The recombination rate of e − -h + pairs was measured using PL ( Figure 7). The higher PL intensity of TiO 2 compared to the other two composites suggested that the better charge separation efficiency on the composite was mainly due to the effects of Mt. In comparison with the Fe/Mt/TiO 2 , xFe-Mt/(1 − x)Fe-TiO 2 showed a decreased luminous intensity, which may be because the doped Fe(III) in the structure of the TiO 2 generated the middle-gap state to avoid the recombination of the e − -h + pairs in the TiO 2 . Furthermore, with the transient photocurrent response measurements and EIS to relate with the photocatalytic phenol degradation activity, the transfer and production of electrons in the TiO 2 , Fe/Mt/TiO 2 , and xFe-Mt/(1 − x)Fe-TiO 2 were investigated. As exhibited in Figure 8a, the photocurrent density was ranked as Fe/Mt/TiO 2 > xFe-Mt/(1 − x) Fe-TiO 2 > TiO 2 . Moreover, the EIS spectra of the TiO 2 , Fe/Mt/TiO 2 , and xFe-Mt/(1 − x)Fe-TiO 2 were measured, as shown in Figure 8b. According to the results, a smaller radius of the EIS Nyquist diagram was discovered in the Fe/Mt/TiO 2 , which suggested a lower charge transfer resistance to improve the photocatalytic activity.  − ln(C/C 0 ) = kt (2) where C and C 0 (mmol/g) refer to the residual and initial concentrations of phenol and parameter k (min −1 ) marks the rate constant of the pseudo-first-order model. The parameters for the pseudo-first-order were listed in Table S1. The correlation coefficient (R 2 ) for the pseudo-first-order is more than 0.980 for all three samples, and the chi-square for all the samples is of −4 order of magnitude. Therefore, the pseudo-first-order model can predict kinetics. The rate constants are 0.015, 0.022, and 0.037 min −1 for the TiO 2 , Fe/Mt/TiO 2 , and xFe-Mt/(1 − x)Fe-TiO 2 , respectively. The Fe/Mt/TiO 2 and xFe-Mt/(1 − x)Fe-TiO 2 showed superb efficiency. Most importantly, the xFe-Mt/(1 − x)Fe-TiO 2 had a higher capacity than the Fe/Mt/TiO 2 , which is consistent with the photocurrent, but the impedes the results and goes against the DRS and PL results. The reusability test was applied ( Figure 10) to investigate the chemical stability of the Fe/Mt/TiO 2 , which possessed the better photocatalytic performance during the photocatalytic oxidation of phenol. High photocatalytic degradation of phenol in the Fe/Mt/TiO 2 was maintained over 93.1% for three cycles. The slight reduction of the photocatalytic performance after the first cycle should be due to the mass loss of the samples during washing.

ERDT Patterns of TiO 2 , Fe/Mt/TiO 2 , and xFe-Mt/(1 − x)Fe-TiO 2
Reversed double-beam photoacoustic spectroscopy (RDB-PAS) was utilized to investigate the surface electronic properties of Fe-Mt, TiO 2 , Fe/Mt/TiO 2 , and xFe-Mt/(1 − x)Fe-TiO 2 . The obtained ERDT could indicate the surface photocatalytic properties of the measured samples [61]. Figure 11 shows the ERDT patterns for Fe-Mt, TiO 2 , Fe/Mt/TiO 2 , and xFe-Mt/(1 − x)Fe-TiO 2 , and the parameters are listed in Table 2. This technique can be applied only to n-type semiconductors; thus, Fe-Mt had no electron trap (ETs) on ERDT because Fe(III) generated a Fermi level as an electron donor which is a p-type characteristic of semiconductor. The energy-reversed distribution of electron traps emerged at energy ranges of 2.20-4.00 eV for the TiO 2 , Fe/Mt/TiO 2 , and xFe-Mt/(1 − x)Fe-TiO 2 . This suggested that there were TiO 2 phases covering the surfaces of the composites. The electron trapping patterns of TiO 2 were fitting using peak separation. Three peaks at 2.96, 3.22, and 3.44 eV were assigned as crystal rutile, amorphous rutile, and amorphous rutile covered with crystal anatase, respectively. The dominance of the amorphous phase was consistent with the low crystallinity of the TiO 2 which was suggested in XRD results (Figure 2 and Table S1). The ERDT patterns of the Fe/Mt/TiO 2 and xFe-Mt/(1 − x)Fe-TiO 2 were also separated. The peak at 3.44 eV represents the amorphous rutile covered in crystal anatase which also emerged in the pattern of the TiO 2 . Crystal TiO 2 should be the dominant fraction for photocatalytic reaction, and the amorphous phase may gain the photocatalytic activity the presence of crystal phase in the TiO 2 [62][63][64]. The crystal anatase in both composites was covered by amorphous TiO 2 . Only the amorphous TiO 2 directly in contact with the crystal phase should be responsible for photocatalytic activity. The intensity of this kind of rutile peak in the Fe/Mt/TiO 2 (85.4%) was twice of that in the xFe-Mt/(1 − x)Fe-TiO 2 (43.5%). Therefore, this result can explain why the xFe-Mt/(1 − x)Fe-TiO 2 showed lower photocatalytic activity for the degradation of phenol than the Fe/Mt/TiO 2 . Both electron accumulation peaks at 3.62 and 3.76 eV were attributed to isolated amorphous TiO 2 that will be not in contact with the crystal phase of the TiO 2 . This fraction of the TiO 2 on the composite was expected to have negligible photocatalytic activity. Besides, the total intensity of the isolated TiO 2 phase on the xFe-Mt/(1 − x)Fe-TiO 2 was much higher than on the Fe/Mt/TiO 2 and may overlap with the photocatalytic active amorphous rutile, depressing the charge generating and transferring. Table 2. Parameters of ERDT peak fitting of Fe/Mt/TiO 2 and xFe-Mt/(1 − x)Fe-TiO 2 . c-R, a-R, a-R on c-A, and iso-TiO 2 are the abbreviation of crystal rutile, amorphous rutile, amorphous rutile covered on crystal anatase, and isolated amorphous TiO 2 , respectively.

Radical Scavenger Tests
During the progress of the light irradiation, the radicals generated are responsible for the photocatalytic reaction. To detect which radical can be predominantly generated throughout the photocatalytic reaction, radical scavengers were added during the photocatalytic degradation of phenol in the Fe/Mt/TiO 2 , which is the best photocatalyst ( Figure 12). Ethanol is the h + scavenger, and isopropanol is the ·OH − scavenger. The addition of both ethanol and isopropanol had a slight impact on the photocatalytic performance of the Fe/Mt/TiO 2 , with a rate constant of 0.014 for both. It means that the h + and ·OH − should not be the main radical sources for the photocatalytic reaction. A significant depression for the photocatalytic degradation of phenol with the rate constant of 0.010 was observed with the addition of para-benzoquinone, which is the ·O 2 − radical scavenger. Therefore, ·O 2 − is the dominant radical for the photocatalytic degradation of phenol. This phenomenon is reasonable because the VB calculated from XPS results was higher than the reduction potential of the ·OH − production. Thus, the ·OH − radical should not be directly generated by the Fe/Mt/TiO 2 .

Photocatalytic Mechanisms
Based on the above results, the mechanisms of both the Fe/Mt/TiO 2 and xFe-Mt/(1 − x) Fe-TiO 2 could be proposed (Scheme 1). In both composites, the crystal anatase will not be directly exposed on the surface but in contact with the amorphous rutile, which is performed as the main photocatalyst. The amount of amorphous rutile covered in crystal anatase in the Fe/Mt/TiO 2 was approximately twice that in the xFe-Mt/(1 − x)Fe-TiO 2 . On the contrary, the amount of isolated amorphous anatase and rutile overlapping with the photocatalytic active sites in xFe-Mt/(1 − x)Fe-TiO 2 was much higher. These two factors resulted in a higher photocatalytic activity for the Fe/Mt/TiO 2 than the xFe-Mt/(1 − x)Fe-TiO 2 . Based on DRS and Eg results, the TiO 2 phase in the Fe/Mt/TiO 2 has an Eg of 3.24 eV, with the 1.75 eV energy level of VB and the 1.49 eV energy level of CB. Additionally, the TiO 2 phase in the xFe-Mt/(1 − x)Fe-TiO 2 has an Eg of 3.19 eV, with the 1.95 eV energy level of VB and -1.24 eV energy level of CB. The VBT of both the Fe/Mt/TiO 2 and the xFe-Mt/(1 − x)Fe-TiO 2 are lower than the reduction potential of ·OH − production [25]. So, the h + and OH − should have little influence on the photocatalytic degradation of phenol. In addition, the radical scavenger test for the Fe/Mt/TiO 2 also proved that the O 2 − should be the predominant radical during the photocatalytic reaction. The Fe(III) located on the interlayer and surface of the Mt in both composites may generate a new Fermi level [25] and establish heterojunctions with the bulk TiO 2 in the composites to reduce the charge recombination and enhance electron transfer.

Conclusions
The . Both composites' optical and electrochemical properties and photocatalytic phenol degradation activity were better than that of pristine TiO 2 , due to the formation of the heterojunction between TiO 2 and Mt, and the introduction of Fe(III) into the composites. Compared between two composite samples, the xFe-Mt/(1 − x)Fe-TiO 2 composite showed better light absorption abilities (Eg: 3.19 eV) than the Fe/Mt/TiO 2 composite (Eg: 3.24 eV), confirmed by DRS results because the doped Fe(III) in the structure of the TiO 2 generates a new electronic level to reduce the Eg. However, the Fe/Mt/TiO 2 exhibited a better charge generation ability, which was proven by the higher peak intensity in the photocurrent and a smaller radius in the EIS. The photocatalytic degradation of phenol was 94.6% for the Fe/Mt/TiO 2 , which was higher than that of the xFe-Mt/(1 − x)Fe-TiO 2 at 85.1%. These are due to the higher active fractions of amorphous rutile covered in crystal anatase on the surface of the composite (two-times more than the xFe-Mt/(1 − x)Fe-TiO 2 ) confirmed by the ERDT results. Moreover, the ·O 2 − radical played an important role in the photocatalytic degradation of phenol. Thus, the modification of Mt/TiO 2 composites by Fe(III) can be an alternative material for the photocatalytic treatment of water pollutants.