Synthesis and Characterization of Iron-Doped TiO2 Nanoparticles Using Ferrocene from Flame Spray Pyrolysis

Iron-doped titanium dioxide nanoparticles, with Fe/Ti atomic ratios from 0% to 10%, were synthesized by flame spray pyrolysis (FSP), employing a single-step method. Ferrocene, being nontoxic and readily soluble in liquid hydrocarbons, was used as the iron source, while titanium tetraisopropoxide (TTIP) was used as the precursor for TiO2. The general particle characterization and phase description were examined using ICP-OES, XRD, BET, and Raman spectroscopy, whereas the XPS technique was used to study the surface chemistry of the synthesized particles. For particle morphology, HRTEM with EELS and EDS analyses were used. Optical and magnetic properties were examined using UV–vis and SQUID, respectively. Iron doping to TiO2 nanoparticles promoted rutile phase formation, which was minor in the pure TiO2 particles. Iron-doped nanoparticles exhibited a uniform iron distribution within the particles. XPS and UV–vis results revealed that Fe2+ was dominant for lower iron content and Fe3+ was common for higher iron content and the iron-containing particles had a contracted band gap of ~1 eV lower than pure TiO2 particles with higher visible light absorption. SQUID results showed that doping TiO2 with Fe changed the material to be paramagnetic. The generated nanoparticles showed a catalytic effect for dye-degradation under visible light.


Introduction
Flame spray pyrolysis (FSP) has emerged as a cost-effective method for large-scale synthesis of nanoparticles [1] and a number of large-scale pilot projects based on FSP have been recently demonstrated [2]. The Johnson-Matthey Company announced the use of FSP for the exploration of next-generation materials related to catalytic applications [3]. Wegner et al. [2], demonstrated that FSP can be used to produce nanoparticles at a rate of a few kilograms per hour at a cost below 100 EUR/kg, where chemical precursors account for the majority of the cost. It is critical to explore research in different precursors for use in FSP, where future innovation may occur at the juncture of combustion, material, and aerosol engineering.
Titanium dioxide (TiO 2 ) is widely produced using flame-based methods [4][5][6] for various applications in catalysis, water-treatment, and solar cells. Doping TiO 2 nanoparticles with transition metals, such as Fe and Co, can result in better photocatalytic conversions due to a reduction in the band-gap and the recombination rate of the electron hole pair [7][8][9].
Additionally, such transition metals can be used to produce TiO 2 nanoparticles with dilute magnetic semiconductor properties [10] that are important for spintronics applications.
Wet-chemistry methods have been utilized to synthesize iron-doped titania nanoparticles [11][12][13][14]. To obtain crystallized nanoparticles from such methods, many steps are involved during the synthesis process in addition to post annealing, which may result in phase segregation at higher iron content (Fe 3 at.% and more) [13]. Single-step production of Fe-TiO 2 nanoparticles is the main advantage in adopting gas-phase techniques of nanoparticle synthesis. Wang et al. [13] produced iron-doped titania with iron content from 0% to 20% using an radiofrequency (RF) plasma. Their major finding was that irondoping advanced the change from anatase to rutile nanoparticles. Teoh et al. [15] utilized a flame spray pyrolysis method in generating iron-doped titania. They used titanium tetraisopropoxide as the titanium precursor and iron naphthenate as the iron precursor. FSP increased the iron-doping up to five times higher than conventional methods with annealing. Increasing Fe concentration results in the transformation from anatase to rutile phase nanoparticles. Fe-TiO 2 nanoparticles were active under visible light and the rate of oxalic acid mineralization is 6.4 times greater when compared to pure TiO 2 and Degussa P25. Magnetic properties were not reported for the FSP-made Fe-TiO 2 nanoparticles.
Ferrocene has many properties, which make it a suitable source of iron for iron-doping in titania nanoparticles such as nontoxicity, stability, and solubility in liquid hydrocarbons [16]. Ferrocene, as a fuel additive for soot reduction, has been broadly considered in the combustion community [16,17]. Ferrocene, as a catalyst, has been utilized in synthesis of carbon nanotubes [18,19] and silica nanoparticles with a controlled surface area in diffusion flames [20]. Ferrocene was used in FSP as a co-precursor for the production of carbon nano-tubes (CNTs) [21,22], boron nitride/carbon nano-tube composite (BN/CNTs) [23], and metal oxides as Al 2 O 3 [24], Fe 2 O 3 [25], and thin films [26]. However, as far as the authors' aware, ferrocene has not been used as an iron source in generating iron-doped titania from FSP technique.
In this work, ferrocene was utilized as an iron precursor and titanium tetraisopropoxide (TTIP) was utilized as a titanium precursor to synthesize iron-doped TiO 2 nanoparticles. Various characterization techniques were performed to study the phase, morphology, and surface structure of the nanoparticles. Optical and magnetic properties were measured. Finally, the doped-particles were used as a photocatalyst for dye-degradation experiments under visible light.

General Nanoparticle Characterization
To estimate the accurate amount of iron doped in TiO 2 samples, inductively-coupled plasma-optical emission spectrometry (ICP-OES) was implemented (Table 1). Direct proportionality was noticed between amounts of ferrocene added to the iron content found in each Fe-TiO 2 sample. The percentage of iron in the doped samples ranged from 0.87% to 5.24 at.%. The specific surface area results utilizing Brunauer-Emmett-Teller (BET) nitrogen adsorption experiments are shown in Table 1. A specific surface area (SSA) for the iron-free TiO 2 sample was the highest and had a value of 151 m 2 /g, while it was dropped to 118 m 2 /g when the Fe/Ti atomic ratio was 1%. Then it was raised monotonically with iron content in the doped-sample. The average primary particle diameter d BET was calculated from BET results while assuming spherical nanoparticles [27]. It was found that d BET was around 10 nm for most samples. Nanoparticle structure and crystallinity was identified using the X-ray diffraction (XRD) technique. Figure 1 illustrates the XRD pattern for Fe-doped samples compared to pure TiO 2 nanoparticles, revealing peaks at 2θ ∼ = 25.4 (for anatase 101) and 27.5 (for rutile 110) crystalline phases. The dominant phase found in the pure TiO 2 nanoparticles was anatase, which is typical for TiO 2 flame synthesis under lean conditions [28], while Fedoping tended to promote the rutile phase percentage and reduces the anatase percentage with increasing iron content in the sample [29]. At 10% of iron-doping, the rutile phase was dominant. By integrating the respective XRD peak intensities, anatase content could be calculated [30]. A calculated anatase percentage for various samples are listed in Table 1. Fe can occupy the Ti position in the TiO 2 lattice because Fe 3+ and Ti 4+ have similar ionic radii [31]. These results are in agreement with previously published data for Fe-TiO 2 using gas-phase methods [13,15]. Nanoparticle structure and crystallinity was identified using the X-ray diffraction (XRD) technique. Figure 1 illustrates the XRD pattern for Fe-doped samples compared to pure TiO2 nanoparticles, revealing peaks at 2θ ≅ 25.4 (for anatase 101) and 27.5 (for rutile 110) crystalline phases. The dominant phase found in the pure TiO2 nanoparticles was anatase, which is typical for TiO2 flame synthesis under lean conditions [28], while Fedoping tended to promote the rutile phase percentage and reduces the anatase percentage with increasing iron content in the sample [29]. At 10% of iron-doping, the rutile phase was dominant. By integrating the respective XRD peak intensities, anatase content could be calculated [30]. A calculated anatase percentage for various samples are listed in Table  1. Fe can occupy the Ti position in the TiO2 lattice because Fe 3+ and Ti 4+ have similar ionic radii [31]. These results are in agreement with previously published data for Fe-TiO2 using gas-phase methods [13,15]. Scherrer formula [32] is another method that gives a good indication of the crystallite size dXRD by estimating the position of main anatase (101) and rutile (110) peaks. It was observed that dXRD for all analyzed samples ranges from 12 to 19 nm, as shown in Table 1. To emphasize the various phases existing in the synthesized TiO2 nanoparticles, Raman spectroscopy was carried out, as shown in Figure 2. Anatase was the dominant phase for iron-free TiO2 particles, whereas the rutile phase was the dominant phase for iron-doped particles. The Raman data were consistent with the XRD results. Scherrer formula [32] is another method that gives a good indication of the crystallite size d XRD by estimating the position of main anatase (101) and rutile (110) peaks. It was observed that d XRD for all analyzed samples ranges from 12 to 19 nm, as shown in Table 1. To emphasize the various phases existing in the synthesized TiO 2 nanoparticles, Raman spectroscopy was carried out, as shown in Figure 2. Anatase was the dominant phase for iron-free TiO 2 particles, whereas the rutile phase was the dominant phase for iron-doped particles. The Raman data were consistent with the XRD results.
Scherrer formula [32] is another method that gives a good indication of the crystallite size dXRD by estimating the position of main anatase (101) and rutile (110) peaks. It was observed that dXRD for all analyzed samples ranges from 12 to 19 nm, as shown in Table 1. To emphasize the various phases existing in the synthesized TiO2 nanoparticles, Raman spectroscopy was carried out, as shown in Figure 2. Anatase was the dominant phase for iron-free TiO2 particles, whereas the rutile phase was the dominant phase for iron-doped particles. The Raman data were consistent with the XRD results.

Fe-TiO 2 Morphology
Bright-field TEM (BF-TEM) along with energy-dispersive X-ray spectroscopy (EDS) analyses were carried out to examine the overall elemental composition and quality of the iron-doped TiO 2 particles. Several electron micrographs at low magnification and EDS spectra were acquired from various locations on the grid for a good representation of size and composition. Figure 3a-c show typical electron micrographs for pure TiO 2 and Fe-TiO 2 nanoparticles. Iron-free particles have an average size of 12 nm. The particle morphology in iron-doped cases (Fe/Ti ratio of 3% and 10%) is similar to that of the iron-free TiO 2 case. The selected area electron diffraction (SAED) micrographs (Figure 3a-c insets) contain diffraction rings that confirm the synthesized nanoparticles were of anatase and rutile phases (also confirmed by XRD and Raman results). Figure 3d-f illustrates the high resolution TEM (HRTEM) electron micrographs for pure and iron-doped TiO 2 nanoparticles. All analyzed samples showed high crystallinity as well as an ordered lattice structure.

Fe-TiO2 Morphology
Bright-field TEM (BF-TEM) along with energy-dispersive X-ray spectroscopy (EDS) analyses were carried out to examine the overall elemental composition and quality of the iron-doped TiO2 particles. Several electron micrographs at low magnification and EDS spectra were acquired from various locations on the grid for a good representation of size and composition. Figures 3a-3c show typical electron micrographs for pure TiO2 and Fe-TiO2 nanoparticles. Iron-free particles have an average size of 12 nm. The particle morphology in iron-doped cases (Fe/Ti ratio of 3% and 10%) is similar to that of the iron-free TiO2 case. The selected area electron diffraction (SAED) micrographs (Figures 3a-3c insets) contain diffraction rings that confirm the synthesized nanoparticles were of anatase and rutile phases (also confirmed by XRD and Raman results). Figures 3d-3f illustrates the high resolution TEM (HRTEM) electron micrographs for pure and iron-doped TiO2 nanoparticles. All analyzed samples showed high crystallinity as well as an ordered lattice structure.   Figure 4a-c show higher magnification of HRTEM images and illustrates the interplanar spacing (d-spacing) of the planes for TiO 2 phases of selected samples. It indicates the decrease in d-spacing with an increase of the iron content in the sample. For pure TiO 2 nanoparticles, the d-spacing of 0.354 nm is related to (101) anatase planes, whereas the value of 0.327 nm for iron doped sample with Fe/Ti ratio of 10% is related to (110) rutile planes [33]. This phase change from anatase to rutile when increasing the iron content is in good agreement with the previous results. The statistical analysis of particle size distribution obtained from TEM results was based on more than 150 nanoparticles and the size distribution histograms are provided in Figure 4d-f. It could be concluded that more than 95% of the particles have a diameter between 7 and 19 nm. The particle size distribution does not change much with doping iron in the TiO 2 samples.   Figure 5 shows the data revealed from EDS spectra, which confirms the existence of both Ti and O elements in the iron-free TiO2 sample. For the 3% (Fe/Ti atomic ratio) sample, the Fe peak started to appear and its intensity increased for higher iron-doping (Fe/Ti = 10%) with an additional Fe peak, which indicates a higher iron concentration in the nanoparticles. In addition, a small Cu peak is observed in the sample with higher iron concentration. Additional Cu peaks observed in the spectrum are attributed to the Cu-grid. The presented electron energy-loss spectroscopy (EELS) spectra in Figure 6 contain the energy loss edges of Ti, O, and Fe elements. The energy loss edge of Ti at the energy loss value of 456 eV is Ti-L23 and it represents electron beam-induced electronic transitions in Ti atoms from the initial states of 2P1/2 and 2P3/2 to final states of 3S or 3P orbitals. Similarly, the energy of edge of Fe at the energy loss value of 710 eV is Fe-L23 and it also represents the same electronic transitions in Fe atoms as of Ti atoms. Whereas the energy  Figure 5 shows the data revealed from EDS spectra, which confirms the existence of both Ti and O elements in the iron-free TiO 2 sample. For the 3% (Fe/Ti atomic ratio) sample, the Fe peak started to appear and its intensity increased for higher iron-doping (Fe/Ti = 10%) with an additional Fe peak, which indicates a higher iron concentration in the nanoparticles. In addition, a small Cu peak is observed in the sample with higher iron concentration. Additional Cu peaks observed in the spectrum are attributed to the Cu-grid.  Figure 5 shows the data revealed from EDS spectra, which confirms the existence of both Ti and O elements in the iron-free TiO2 sample. For the 3% (Fe/Ti atomic ratio) sample, the Fe peak started to appear and its intensity increased for higher iron-doping (Fe/Ti = 10%) with an additional Fe peak, which indicates a higher iron concentration in the nanoparticles. In addition, a small Cu peak is observed in the sample with higher iron concentration. Additional Cu peaks observed in the spectrum are attributed to the Cu-grid. The presented electron energy-loss spectroscopy (EELS) spectra in Figure 6 contain the energy loss edges of Ti, O, and Fe elements. The energy loss edge of Ti at the energy loss value of 456 eV is Ti-L23 and it represents electron beam-induced electronic transitions in Ti atoms from the initial states of 2P1/2 and 2P3/2 to final states of 3S or 3P orbitals. The presented electron energy-loss spectroscopy (EELS) spectra in Figure 6 contain the energy loss edges of Ti, O, and Fe elements. The energy loss edge of Ti at the energy loss value of 456 eV is Ti-L23 and it represents electron beam-induced electronic transitions in Ti atoms from the initial states of 2P 1/2 and 2P 3/2 to final states of 3S or 3P orbitals. Similarly, the energy of edge of Fe at the energy loss value of 710 eV is Fe-L23 and it also represents the same electronic transitions in Fe atoms as of Ti atoms. Whereas the energy loss edge of O at the energy value of 532 eV is the O-K edge, it represents the electronic transitions from 1S 1/2 initial states to 2P final states in O atoms. The intensity of signals of these energy loss edges is proportional to the elemental composition of nanomaterials and, therefore, have been utilized to map the spatial distributions of these elements in the samples. More detailed analysis for EDS and EELS results can be found in Reference [34]. The acquired elemental maps of the 10% sample are presented in Figure 7. The results demonstrated that Ti, O, and Fe were present and reasonably uniform, implying that Fe was uniformly doped in the TiO 2 nanoparticles and did not contain segregated iron oxide particles. samples. More detailed analysis for EDS and EELS results can be found in Reference [34].
The acquired elemental maps of the 10% sample are presented in Figure 7. The results demonstrated that Ti, O, and Fe were present and reasonably uniform, implying that Fe was uniformly doped in the TiO2 nanoparticles and did not contain segregated iron oxide particles.

Surface Chemistry
Chemical composition was analyzed using X-ray photoelectron spectroscopy (XPS). Survey spectra from pure TiO2 and from Fe-doped TiO2 samples are shown in Figure 8. Ti, O, C, and Fe elements are detected for the iron-doped samples whereas Fe was not detected from pure TiO2 sample. It should be noted that the synthesized samples do not contain any carbon while the carbon peak in the survey scan comes from the background. High resolution XPS spectra of Ti 2p and O 1s core levels from pure TiO2 are shown in Figure 9. The Ti 2p3/2 core level is fitted using two components. The dominant peak centered at 458.6 eV is associated with Ti ions with a formal valence 4+ (Ti 4+ ), while the peak at the lower binding energy of ~456.6 eV is associated with Ti ions with a reduced charge state (Ti 3+ ) [35]. The O 1s core level is centered at 529.8 eV, corresponding to the metal oxide (TiO2) [35]. These Ti 2p and O 1s peaks are also observed for all the Fe-doped TiO2 samples. More detailed analysis for EDS and EELS results can be found in Reference [34].
The acquired elemental maps of the 10% sample are presented in Figure 7. The results demonstrated that Ti, O, and Fe were present and reasonably uniform, implying that Fe was uniformly doped in the TiO2 nanoparticles and did not contain segregated iron oxide particles.

Surface Chemistry
Chemical composition was analyzed using X-ray photoelectron spectroscopy (XPS). Survey spectra from pure TiO2 and from Fe-doped TiO2 samples are shown in Figure 8. Ti, O, C, and Fe elements are detected for the iron-doped samples whereas Fe was not detected from pure TiO2 sample. It should be noted that the synthesized samples do not contain any carbon while the carbon peak in the survey scan comes from the background. High resolution XPS spectra of Ti 2p and O 1s core levels from pure TiO2 are shown in Figure 9. The Ti 2p3/2 core level is fitted using two components. The dominant peak centered at 458.6 eV is associated with Ti ions with a formal valence 4+ (Ti 4+ ), while the peak at the lower binding energy of ~456.6 eV is associated with Ti ions with a reduced charge state (Ti 3+ ) [35]. The O 1s core level is centered at 529.8 eV, corresponding to the metal

Surface Chemistry
Chemical composition was analyzed using X-ray photoelectron spectroscopy (XPS). Survey spectra from pure TiO 2 and from Fe-doped TiO 2 samples are shown in Figure 8. Ti, O, C, and Fe elements are detected for the iron-doped samples whereas Fe was not detected from pure TiO 2 sample. It should be noted that the synthesized samples do not contain any carbon while the carbon peak in the survey scan comes from the background. High resolution XPS spectra of Ti 2p and O 1s core levels from pure TiO 2 are shown in Figure 9. The Ti 2p 3/2 core level is fitted using two components. The dominant peak centered at 458.6 eV is associated with Ti ions with a formal valence 4+ (Ti 4+ ), while the peak at the lower binding energy of~456.6 eV is associated with Ti ions with a reduced charge state (Ti 3+ ) [35]. The O 1s core level is centered at 529.8 eV, corresponding to the metal oxide (TiO 2 ) [35]. These Ti 2p and O 1s peaks are also observed for all the Fe-doped TiO 2 samples.   Figure 10 illustrates high resolution XPS spectra for iron-doped TiO2 nanoparticles resulting from the Fe 2p core level. High resolution data was fitted using eight components located at 710.8, 709.1, 718.3, 719.4, 722.3, 724.2, 728.5, and 732.9 eV. The Fe 2p3/2 doublet peak at 710.8 eV, and its corresponding satellite peak at 719.4 eV, together with the Fe 2p1/2 doublet peak at 724.3 eV, and its corresponding satellite peak at 732.9 eV, are the main signature of the Fe 3+ oxidation state of Fe [36,37]. Another signature of Fe 3+ ions are the satellite peaks located at 8.6 eV from the main peaks.
The Fe 2p3/2 doublet peak at 709.1 eV and its corresponding satellite peak at 715.3 eV, together with the Fe 2p1/2 doublet peak at 722.3 eV and its corresponding satellite peak at 728.5 eV are signatures of the Fe 2+ oxidation state of Fe [37,38]. Other signatures of Fe 2+ ions are the satellite peaks located at 6.2 eV from the main peaks. The area ratio of Fe 2p3/2-Fe 2p1/2 components equals 2:1 with their doublet separation of 13.5 and 13.2 eV for Fe 3+ and Fe 2+ , respectively. The same values are used for their corresponding satellites. Increasing the percentage of the Fe dopant leads to an increase of the Fe 3+ /Fe 2+ ratio from 0.5 at 3% (Fe/Ti ratio) to 1.8 at 10% (Fe/Ti ratio).   Figure 10 illustrates high resolution XPS spectra for iron-doped TiO2 nanoparticles resulting from the Fe 2p core level. High resolution data was fitted using eight components located at 710.8, 709.1, 718.3, 719.4, 722.3, 724.2, 728.5, and 732.9 eV. The Fe 2p3/2 doublet peak at 710.8 eV, and its corresponding satellite peak at 719.4 eV, together with the Fe 2p1/2 doublet peak at 724.3 eV, and its corresponding satellite peak at 732.9 eV, are the main signature of the Fe 3+ oxidation state of Fe [36,37]. Another signature of Fe 3+ ions are the satellite peaks located at 8.6 eV from the main peaks.
The Fe 2p3/2 doublet peak at 709.1 eV and its corresponding satellite peak at 715.3 eV, together with the Fe 2p1/2 doublet peak at 722.3 eV and its corresponding satellite peak at 728.5 eV are signatures of the Fe 2+ oxidation state of Fe [37,38]. Other signatures of Fe 2+ ions are the satellite peaks located at 6.2 eV from the main peaks. The area ratio of Fe 2p3/2-Fe 2p1/2 components equals 2:1 with their doublet separation of 13.5 and 13.2 eV for Fe 3+ and Fe 2+ , respectively. The same values are used for their corresponding satellites. Increasing the percentage of the Fe dopant leads to an increase of the Fe 3+ /Fe 2+ ratio from 0.5 at 3% (Fe/Ti ratio) to 1.8 at 10% (Fe/Ti ratio).  Figure 10 illustrates high resolution XPS spectra for iron-doped TiO 2 nanoparticles resulting from the Fe 2p core level. High resolution data was fitted using eight components located at 710.8, 709.1, 718.3, 719.4, 722.3, 724.2, 728.5, and 732.9 eV. The Fe 2p 3/2 doublet peak at 710.8 eV, and its corresponding satellite peak at 719.4 eV, together with the Fe 2p 1/2 doublet peak at 724.3 eV, and its corresponding satellite peak at 732.9 eV, are the main signature of the Fe 3+ oxidation state of Fe [36,37]. Another signature of Fe 3+ ions are the satellite peaks located at 8.6 eV from the main peaks. The valence band spectra from pure and Fe-doped TiO2 nanoparticles are shown in Figure 11. The spectrum of undoped TiO2 contained three major features (marked as A, B, and C) at binding energy positions around ~7.5, 5, and 1 eV, respectively. This is consistent with the literature [39]. Features A and B are related to the O-2p derived states and are referred to as 'bonding' and 'non-bonding orbital emissions,' respectively. Feature C corresponds to the Ti 3+ 3d defect state. After Fe doping, feature D (~2.5 eV) began to appear. Such a shape, based on the literature, is assigned to a mixed Fe 3d and Ti 3d derived states [39]. Note that the valence band edge from the Fe-doped TiO2 shifted by ~1 eV to a lower binding energy, compared to undoped TiO2, which could enhance photo-catalytic properties in the region of visible light [40].

Magnetic Properties
To study the intrinsic magnetic property, temperature variations of magnetization The Fe 2p 3/2 doublet peak at 709.1 eV and its corresponding satellite peak at 715.3 eV, together with the Fe 2p 1/2 doublet peak at 722.3 eV and its corresponding satellite peak at 728.5 eV are signatures of the Fe 2+ oxidation state of Fe [37,38]. Other signatures of Fe 2+ ions are the satellite peaks located at 6.2 eV from the main peaks. The area ratio of Fe 2p 3/2 -Fe 2p 1/2 components equals 2:1 with their doublet separation of 13.5 and 13.2 eV for Fe 3+ and Fe 2+ , respectively. The same values are used for their corresponding satellites. Increasing the percentage of the Fe dopant leads to an increase of the Fe 3+ /Fe 2+ ratio from 0.5 at 3% (Fe/Ti ratio) to 1.8 at 10% (Fe/Ti ratio).
The valence band spectra from pure and Fe-doped TiO 2 nanoparticles are shown in Figure 11. The spectrum of undoped TiO 2 contained three major features (marked as A, B, and C) at binding energy positions around~7.5, 5, and 1 eV, respectively. This is consistent with the literature [39]. Features A and B are related to the O-2p derived states and are referred to as 'bonding' and 'non-bonding orbital emissions,' respectively. Feature C corresponds to the Ti 3+ 3d defect state. After Fe doping, feature D (~2.5 eV) began to appear. Such a shape, based on the literature, is assigned to a mixed Fe 3d and Ti 3d derived states [39]. Note that the valence band edge from the Fe-doped TiO 2 shifted by~1 eV to a lower binding energy, compared to undoped TiO 2 , which could enhance photo-catalytic properties in the region of visible light [40].
with the literature [39]. Features A and B are related to the O-2p derived states and are referred to as 'bonding' and 'non-bonding orbital emissions,' respectively. Feature C corresponds to the Ti 3+ 3d defect state. After Fe doping, feature D (~2.5 eV) began to appear. Such a shape, based on the literature, is assigned to a mixed Fe 3d and Ti 3d derived states [39]. Note that the valence band edge from the Fe-doped TiO2 shifted by ~1 eV to a lower binding energy, compared to undoped TiO2, which could enhance photo-catalytic properties in the region of visible light [40].

Magnetic Properties
To study the intrinsic magnetic property, temperature variations of magnetization measurements (moment M vs. temperature T) were performed. Initially, the sample was cooled to 5 K under a nominal zero field. Then a field of H = 1 kOe was applied. Subsequently, M vs. T data was measured under the constant magnetic field, while warming the sample from 5 K to 300 K ( Figure 12). Results showed that all the samples were typical paramagnetic and none showed any Curie transition (TC), within the temperature range,

Magnetic Properties
To study the intrinsic magnetic property, temperature variations of magnetization measurements (moment M vs. temperature T) were performed. Initially, the sample was cooled to 5 K under a nominal zero field. Then a field of H = 1 kOe was applied. Subsequently, M vs. T data was measured under the constant magnetic field, while warming the sample from 5 K to 300 K ( Figure 12). Results showed that all the samples were typical paramagnetic and none showed any Curie transition (TC), within the temperature range, which implies that Fe doping did not induce a long range of a ferromagnetic interaction in these samples.  Isothermal magnetization measurements (magnetic moment M vs. magnetic field H) were carried out. A necessary diamagnetic correction corresponding to the bare sample holder was applied after magnetization measurements on each sample. Figure 13 shows the magnetic hysteresis (M-H) curves of the undoped and Fe-doped samples taken at 5 and 300 K. The results showed that magnetization increases monotonically with increased Fe content. At room temperature, most of the doped samples have small moment values, which is similar to the undoped sample. However, for the 10% case, a higher moment was observed. The inset in Figure 13b shows a narrow opening of the hysteresis loop near the origin for the 10% case, indicating a paramagnetic behavior. Earlier reports attributed the absence of ferromagnetism (or paramagnetism in Fe-TiO2) to sample preparation methods and/or dopant concentrations [41] where dopant Fe ions remain as isolated ions weakly interacting in the host TiO2. The lack of magnetic properties in previous works was also attributed to the presence of secondary phases, such as α-Fe2O3 or Fe3O4, since iron ions may coexist in different valance states [42,43]. For magnetic behavior, oxygen defects [41,44,45] could induce magnetic interactions. Since only paramagnetic properties were observed in the present study, it was attributed to oxygen defects and Fe 3+ ions in the TiO2 crystal. Isothermal magnetization measurements (magnetic moment M vs. magnetic field H) were carried out. A necessary diamagnetic correction corresponding to the bare sample holder was applied after magnetization measurements on each sample. Figure 13 shows the magnetic hysteresis (M-H) curves of the undoped and Fe-doped samples taken at 5 and 300 K. The results showed that magnetization increases monotonically with increased Fe content. At room temperature, most of the doped samples have small moment values, which is similar to the undoped sample. However, for the 10% case, a higher moment was observed. The inset in Figure 13b shows a narrow opening of the hysteresis loop near the origin for the 10% case, indicating a paramagnetic behavior. Earlier reports attributed the absence of ferromagnetism (or paramagnetism in Fe-TiO 2 ) to sample preparation methods and/or dopant concentrations [41] where dopant Fe ions remain as isolated ions weakly interacting in the host TiO 2 . The lack of magnetic properties in previous works was also attributed to the presence of secondary phases, such as α-Fe 2 O 3 or Fe 3 O 4 , since iron ions may coexist in different valance states [42,43]. For magnetic behavior, oxygen defects [41,44,45] could induce magnetic interactions. Since only paramagnetic properties were observed in the present study, it was attributed to oxygen defects and Fe 3+ ions in the TiO 2 crystal. and/or dopant concentrations [41] where dopant Fe ions remain as isolated ions weakly interacting in the host TiO2. The lack of magnetic properties in previous works was also attributed to the presence of secondary phases, such as α-Fe2O3 or Fe3O4, since iron ions may coexist in different valance states [42,43]. For magnetic behavior, oxygen defects [41,44,45] could induce magnetic interactions. Since only paramagnetic properties were observed in the present study, it was attributed to oxygen defects and Fe 3+ ions in the TiO2 crystal.

Optical Properties
The UV-vis absorption spectroscopy was adopted to determine light absorption from iron-free and iron-doped TiO2 nanoparticles. The spectra are shown in Figure 14.

Optical Properties
The UV-vis absorption spectroscopy was adopted to determine light absorption from iron-free and iron-doped TiO 2 nanoparticles. The spectra are shown in Figure 14. The pure TiO 2 sample did not show any absorption in the region of visible light (>400 nm) whereas it manifested an absorption edge toward the UV region (<400 nm). This absorption peak could be assigned to anatase band-gap excitation, which equals 3.2 eV [31]. The visible light absorption (350-550 nm) was significantly increased by iron doping in TiO 2 nanoparticles and monotonically increased with iron content in the sample. This observation is consistent with data found in literature [46,47]. The powder color converted from white (pure TiO 2 ) by passing through yellow (low iron content) to brown (high iron content), indicating that doping TiO 2 nanoparticles with iron enhanced the visible light absorption. The pure TiO2 sample did not show any absorption in the region of visible light (>400 nm) whereas it manifested an absorption edge toward the UV region (<400 nm). This absorption peak could be assigned to anatase band-gap excitation, which equals 3.2 eV [31]. The visible light absorption (350-550 nm) was significantly increased by iron doping in TiO2 nanoparticles and monotonically increased with iron content in the sample. This observation is consistent with data found in literature [46,47]. The powder color converted from white (pure TiO2) by passing through yellow (low iron content) to brown (high iron content), indicating that doping TiO2 nanoparticles with iron enhanced the visible light absorption.

Photocatalytic Activity Test
A photocatalytic reactor system was utilized to carry out the photodegradation tests of Rhodamine B (Rh-B) solution with iron-doped TiO2 nanoparticles under visible-light irradiation. A mass equals 40 mg of iron-doped TiO2 nanoparticles (as a catalyst) was added to 100 mL of organic dye solution (20 mg/L). Then the mixture was stirred in dark for 60 min to reach an adsorption-desorption equilibrium before performing the photodegradation experiments. The reactor was loaded up with the mixture of catalyst and organic dye solution. Water cooling was used to maintain the temperature constant at 298 K. Then, a Xenon lamp (CX-04E) with an optical filter (>420 nm) passing only visible light was used to irradiate the suspension. At a time interval of 15 min, about 2 mL of the suspension was extracted and centrifuged at 14,000 rpm for a period of 5 min in order to get rid of the catalyst. To calculate the new dye concentration in the solution, absorbance at λmax = 553 nm (maximum absorption peak of Rh-B) was measured and given the symbol C. The rate of dye degradation is recorded as C/C0 where C0 is the initial concentration after reaching the adsorption-desorption equilibrium.

Photocatalytic Activity Test
A photocatalytic reactor system was utilized to carry out the photodegradation tests of Rhodamine B (Rh-B) solution with iron-doped TiO 2 nanoparticles under visible-light irradiation. A mass equals 40 mg of iron-doped TiO 2 nanoparticles (as a catalyst) was added to 100 mL of organic dye solution (20 mg/L). Then the mixture was stirred in dark for 60 min to reach an adsorption-desorption equilibrium before performing the photodegradation experiments. The reactor was loaded up with the mixture of catalyst and organic dye solution. Water cooling was used to maintain the temperature constant at 298 K. Then, a Xenon lamp (CX-04E) with an optical filter (>420 nm) passing only visible light was used to irradiate the suspension. At a time interval of 15 min, about 2 mL of the suspension was extracted and centrifuged at 14,000 rpm for a period of 5 min in order to get rid of the catalyst. To calculate the new dye concentration in the solution, absorbance at λ max = 553 nm (maximum absorption peak of Rh-B) was measured and given the symbol C. The rate of dye degradation is recorded as C/C 0 where C 0 is the initial concentration after reaching the adsorption-desorption equilibrium.

Degradation of Rhodamine B
It was clear from the previous characterization of Fe-TiO 2 nanoparticles that iron doping shifted the optical response of TiO 2 nanoparticles from UV to the visible light region. The optical response was directly proportional to iron concentration in the irondoped samples. This implied that these iron-doped TiO 2 nanoparticles, unlike iron-free TiO 2 , could be active in the visible-light region and considered to be catalysts.
Since Rhodamine B is considered a major source of pollution for industrial waste water, its degradation in an aqueous solution has been studied. Figure 15 illustrates that, in the presence of iron-doped nanoparticles with Fe/Ti = 10%, the absorbance (at λ max = 553 nm) of the Rh-B dye monotonically decreases with time. The degradation efficiency of Rh-B dye in the presence of iron-doped TiO 2 catalysts have been calculated from the following equation.
where C 0 is the initial concentration of dye and C is the concentration of dye after irradiation in a selected time interval. where C0 is the initial concentration of dye and C is the concentration of dye after irradiation in a selected time interval. The Rh-B dye concentration in the solution and its absorbance at λmax =553 nm are directly proportional according to Beer's law. C0 was considered to be the initial concentration of Rh-B dye in the solution to make sure that only a chemical reaction changes the dye concentration and removes any effect of dark adsorption. Figure 16 illustrates the RH-B dye degradation rates in the existence of iron-free TiO2 and iron-doped TiO2 nanoparticles under visible-light radiation. When using pure TiO2 nanoparticles as a catalyst for two hours of radiation, the dye was degraded by ~12% and the degradation efficiency was raised to 20% when using an iron-doped sample with Fe/Ti = 1%. Then it was monotonically increased up to 53% when using an iron-doped sample with Fe/Ti = 10%. The degradation efficiencies of Rh-B dye in an aqueous solution in the current study are consistent with values in literature [14,48,49]. After absorbing light, the excited dye (Rhodamine B) injected an electron into the conduction band of TiO2 where it was captured by surface-adsorbed O2 to form O2 •− . Then, the dye cation radicals were degraded via attack by oxygen active species [50]. In addition, it was noticed that no Rh-B dye degradation took place under visible-light radiation without catalysts.  The first-order rate constant k (min −1 ) for photodegradation of dyes has been calculated by employing the following equation: where C0 and C are dye concentrations initially and after time t, respectively. The photodegradation rate constant (k) could be calculated from the above kinetic model and found to have the value of 0.001, 0.0019, 0.0039, 0.0047, and 0.006 min −1 for samples Fe/Ti = 0%, 1%, 3%, 5%, and 10%, respectively. Previous studies showed that the rate constant of Rh- The Rh-B dye concentration in the solution and its absorbance at λ max = 553 nm are directly proportional according to Beer's law. C 0 was considered to be the initial concentration of Rh-B dye in the solution to make sure that only a chemical reaction changes the dye concentration and removes any effect of dark adsorption. Figure 16 illustrates the RH-B dye degradation rates in the existence of iron-free TiO 2 and iron-doped TiO 2 nanoparticles under visible-light radiation. When using pure TiO 2 nanoparticles as a catalyst for two hours of radiation, the dye was degraded by~12% and the degradation efficiency was raised to 20% when using an iron-doped sample with Fe/Ti = 1%. Then it was monotonically increased up to 53% when using an iron-doped sample with Fe/Ti = 10%. The degradation efficiencies of Rh-B dye in an aqueous solution in the current study are consistent with values in literature [14,48,49]. After absorbing light, the excited dye (Rhodamine B) injected an electron into the conduction band of TiO 2 where it was captured by surface-adsorbed O 2 to form O 2 •− . Then, the dye cation radicals were degraded via attack by oxygen active species [50]. In addition, it was noticed that no Rh-B dye degradation took place under visible-light radiation without catalysts.  The first-order rate constant k (min −1 ) for photodegradation of dyes has been calculated by employing the following equation: where C0 and C are dye concentrations initially and after time t, respectively. The photodegradation rate constant (k) could be calculated from the above kinetic model and found to have the value of 0.001, 0.0019, 0.0039, 0.0047, and 0.006 min −1 for samples Fe/Ti = 0%, 1%, 3%, 5%, and 10%, respectively. Previous studies showed that the rate constant of Rh- The first-order rate constant k (min −1 ) for photodegradation of dyes has been calculated by employing the following equation: where C 0 and C are dye concentrations initially and after time t, respectively. The photodegradation rate constant (k) could be calculated from the above kinetic model and found to have the value of 0.001, 0.0019, 0.0039, 0.0047, and 0.006 min −1 for samples Fe/Ti = 0%, 1%, 3%, 5%, and 10%, respectively. Previous studies showed that the rate constant of Rh-B photodegradation had reported a value of 0.0029 min −1 when using iron-doped titania and 0.00016 min −1 when using P25 [48]. Higher rate constants for the decomposition of Rh-B of the synthesized materials than standard P25 and Fe-TiO 2 NPs prepared using different methods suggested that the present materials are more active. Hence, the above findings clearly demonstrate that a higher amount of iron incorporation into the framework of TiO 2 makes the catalyst more efficient for photocatalytic degradation of organic dyes.

Experiment
FSP was applied to synthesize TiO 2 and Fe-TiO 2 nanoparticles. Figure 17 shows the schematic of experimental setup, composed of a spray system with an annular pilot flame, a precursor delivery system, and a particle collection system. The spray system was an air-assisted spray nozzle. The precursor flowed through a capillary tube while the dispersion gas passed through the annular gap. Dispersion gas of oxygen (10.6 L/min) was supplied to the spray nozzle for precursor atomization with a pressure drop across the nozzle of 2.5 bar. An annular slit (15.88 mm inner diameter and 1 mm thickness), surrounding the nozzle provided the supporting premixed pilot flame. Methane (CH 4 with purity 99.995%) with a flow rate of 1.25 L/min was mixed with 2.5 L/min of oxygen (O 2 with purity 99.9995%) for the pilot flame.
Various precursors were used to synthesize pure and doped-TiO 2 nanoparticles. Titanium tetraisopropoxide (Sigma-Aldrich, 97% purity) for TiO 2 nanoparticles (NP), and dissolving ferrocene (Sigma-Aldrich, 98%) in m-xylene (Sigma-Aldrich, 99%) for Fe doping. Table 1 shows the Fe/Ti atomic ratios in synthesizing nanoparticles resulting from changing ferrocene concentration. A syringe pump was used to inject the precursor into the spray nozzle with a flow rate of 450 mL/hr. Synthesized nanoparticles were collected on a glass-fiber filter with the aid of a vacuum pump.
air-assisted spray nozzle. The precursor flowed through a capillary tube while the dispersion gas passed through the annular gap. Dispersion gas of oxygen (10.6 L/min) was supplied to the spray nozzle for precursor atomization with a pressure drop across the nozzle of 2.5 bar. An annular slit (15.88 mm inner diameter and 1 mm thickness), surrounding the nozzle provided the supporting premixed pilot flame. Methane (CH4 with purity 99.995%) with a flow rate of 1.25 L/min was mixed with 2.5 L/min of oxygen (O2 with purity 99.9995%) for the pilot flame. Various precursors were used to synthesize pure and doped-TiO2 nanoparticles. Titanium tetraisopropoxide (Sigma-Aldrich, 97% purity) for TiO2 nanoparticles (NP), and dissolving ferrocene (Sigma-Aldrich, 98%) in m-xylene (Sigma-Aldrich, 99%) for Fe doping. Table 1 shows the Fe/Ti atomic ratios in synthesizing nanoparticles resulting from changing ferrocene concentration. A syringe pump was used to inject the precursor into the spray nozzle with a flow rate of 450 mL/hr. Synthesized nanoparticles were collected on a glass-fiber filter with the aid of a vacuum pump.
The nanoparticle morphology and size were extracted from transmission electron microscope results (TEM, FEI Com., Titan 80-300ST) by operating it at an accelerating voltage of 300 kV. The microscope was set in various modes, such as bright-field TEM (BF-TEM), high resolution TEM (HRTEM), selected area electron diffraction (SAED), core-loss electron energy-loss spectroscopy (EELS), and energy-filtered TEM (EFTEM) modes to obtain specific information about the physical properties of Fe-doped TiO 2 nanoparticles. The entire TEM data were obtained and resolved by Digital-Micrograph Software Package (Gatan Inc., Version GMS1.85).

Conclusions
Flame spray pyrolysis was utilized to synthesize pure and Fe-doped TiO 2 nanoparticles with various iron concentrations in a single-step process. Ferrocene was mixed with TTIP and used as the precursor for iron-doped TiO 2 . XRD and Raman confirmed that the anatase phase was dominant in pure TiO 2 nanoparticles (>90% anatase), with an average particle diameter of about 12 nm (based on HRTEM). The addition of iron substantially changed the anatase phase to rutile with the anatase percentage decreasing from 86% to 21% when the Fe/Ti atomic ratio in the sample was changed from 1% to 10%, respectively. Morphology of the generated nanoparticles was studied using HRTEM, which revealed the crystallinity of the spherical particles and the uniformity of Fe doping distribution across the samples. XPS confirmed the formation of Fe 3+ and Fe 2+ within the Fe-doped TiO 2 samples. The ratio of Fe 3+ /Fe 2+ varied from 0.5 at Fe/Ti ratio of 3% to 1.8 at a 10% ratio. The valence band edge for the Fe-doped TiO 2 was also lowered by~1 eV when compared to the pure TiO 2 sample. Regarding magnetic properties, Fe doping of TiO 2 nanoparticles changed the material from nonmagnetic to paramagnetic. Doping TiO 2 particles with Fe remarkably increased light absorption and expanded it to the visible-light region due to their narrower band gap. The iron-doped samples showed much higher catalytic activity for dye degradation under visible-light radiation when compared to pure TiO 2 nanoparticles. Non-toxic, cost-effective photo-catalysts are vital for industrial waste water treatment.
In summary, FSP is an attractive method for the production of Fe-doped TiO 2 with ferrocene as the source of iron. The resulting material has desirable magnetic and optical properties, which can be used in various photocatalytic and electronic applications.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.