*2.1. Structural and Optical Characterization of LaFeO3*

The powder XRD patterns of the prepared LaFeO3 samples after calcination at various temperatures are shown in Figure 1. All the diffraction peaks belong to the orthorhombic LaFeO3 with ABO3-type perovskite structure (JCPDS card No. 88-0641) [24]. The main characteristic reflexes are located at 2θ of 22.6◦, 25.5◦, 32.2◦, 34.5◦, 39.7◦, 46.1◦, 47.7◦, 52.0◦, 54.0◦, 57.4◦, 64.0◦, 67.3◦ and 76.6◦, being indexed to the (101), (111), (121), (210), (220), (202), (230), (141), (240), (115), (242) and (204) diffraction planes, respectively [24]. This confirms the effective preparation of a single phase perovskite LaFeO3 without any crystalline impurities like La2O3 or Fe2O3. For calculating the crystallite sizes the reflexes of highest intensity at 2θ = 32.2◦ were selected. With increasing calcination temperature the diffraction peaks get sharper and more intense, indicating a better crystallization and growth of the grains. The average crystallite sizes of LaFeO3 *D* have been determined by using the Debye-Scherer formula [25]:

$$D = \frac{K\lambda}{\beta \cos \theta}$$

with *K* being the crystallite shape factor, *λ* the X-ray wavelength (1.5406 nm for Cu Kα), *β* is the width of the diffraction peak and *θ* is the Bragg angle. The crystallite sizes were 27.4 nm, and 45.7 nm for S-700 and S-900, respectively. In the smaller particles, less time was needed for the electrons and holes to diffuse from the inner part to the surface of the catalyst, where they could react. This typically leads to higher photocatalytic efficiency.

**Figure 1.** X-ray diffraction patterns of LaFeO3 obtained at different calcination temperatures.

Figure 2 shows the scanning electron micrographs of the prepared LaFeO3 samples at different calcination temperatures. Scanning electron microscopy (SEM) was used to determine the morphology of the perovskite LaFeO3 samples; as seen in Figure 2 both samples show a network structure with semi-spherical morphology. It was found that the particle sizes of S-700 were significantly smaller than those of S-900, consistent with the trend of the crystallite sizes determined from XRD.

**Figure 2.** SEM images of LaFeO3 calcined at different temperatures: (**a**) S-700, (**b**) S-900.

Energy dispersive X-ray spectroscopy (EDXS) was used to investigate the purity and chemical composition of synthesized LaFeO3 nanoparticles, the pattern of calcined LaFeO3 samples are shown in Figure 3. Besides a carbon signal appearing at 0.277 keV and resulting from the latex of the SEM sample holder due to the incomplete coverage of the sample [25], only lanthanum (La), iron (Fe) and oxygen (O) were present, confirming that the citric acid assisted sol-gel route leads to high purity LaFeO3 photocatalyst. The peaks at around 0.83 and 4.65 keV were related to La and the ones at around 6.399 and 0.704 keV to Fe; they proved that the formation of the LaFeO3 photocatalyst had a 1:1 molar ratio of metal ions as the atomic percentage obtained from EDXS was 0.35% for Fe and 0.34% for La in sample S-700 and 1.13% for Fe and 1.21% for La in S-900, respectively.

**Figure 3.** Energy dispersive X-ray (EDX) spectra of LaFeO3 at different calcination temperatures (S-700, S-900).

Figure 4 presents the IR spectra of different LaFeO3 samples prepared at different calcination temperatures in the wavenumber range 400–4000 cm−<sup>1</sup> in order to determine the possible functional groups in the sample. The FT-IR spectra were quite featureless, confirming again the purity of the synthesized samples. The peak at 556 cm−<sup>1</sup> can be attributed to the Fe–O stretching vibration being characteristic of the FeO6 octahedrons in perovskite-type LaFeO3 [26]. The band at 716 cm−<sup>1</sup> can be assigned bending vibrations of the La-O bonds [11]; the small peak at 2905 cm−<sup>1</sup> as well as the small peak at around 1600 cm−<sup>1</sup> resulted from small amounts of citric acid residues, they were accounted to C-H vibrations and the symmetric stretching of the carboxyl groups, respectively.

**Figure 4.** FT-IR spectra of LaFeO3 nanoparticles calcined at different temperatures.

The specific surface areas of the synthesized LaFeO3 samples were determined from a nitrogen adsorption-desorption isotherm using the BET approach. The isotherms of the samples can be classified into type III behavior (Figure 5), which is attributed to a weak adsorbate-adsorbent interaction [27]. The surface areas decreased with increasing calcination temperature. Although the surface areas were in general quite small, the area of S-700 exceeded even slightly the highest value reported by Tijare et al. of 9.5 m2/g [23]. In general, higher surface areas facilitated adsorption of organic pollutants, promoted charge carrier separation and enabled more light harvesting, resulting in total higher photocatalytic activity.

**Figure 5.** N2 adsorption-desorption isotherms for LaFeO3 samples and the resulting BET surface areas of the different samples.

Diffuse reflectance UV-Vis spectroscopy was employed to characterize the optical properties of the LaFeO3 nanoparticles, as shown in Figure 6a. In the perovskite-type oxide, the strong absorption edge at 520 nm was ascribed to the electronic transition from the O 2p orbitals forming the valence band to the Fe 3d orbitals in the conduction band [28]. The data showed that the sol-gel prepared LaFeO3 photocatalyst could serve as a potential visible-light-driven photocatalyst. In addition, the band gap energy of LaFeO3 catalysts can be determined from Kubelka−Munk equation [29] via a Tauc plot:

$$\boldsymbol{\alpha} = \mathbf{B} \left( h\boldsymbol{\upsilon} - \boldsymbol{E}\_{\mathcal{S}} \right)^{n} / h\boldsymbol{\upsilon}$$

with *α* being the absorption coefficient, *ν* the irradiation frequency, *Eg* the band gap, *B* being a constant (being usually 1 for semiconductors), h is the Planck constant and n is a constant depending on the type of semiconductor (direct transition: n = 1/2; indirect transition: n = 2). For the direct transition semiconductor LaFeO3, the band gap energy values were estimated by extrapolation of the linear part of the curves of the Kubelka–Munk function (α*hv*) 1/2 against the photon energy (*hv*), as displayed in Figure 6b. The S-700 sample absorbed slightly more light energy than the other ones showing that a decreasing particle size and an increased surface area led to a slight red shift.

**Figure 6.** Kubelka-Munk diffuse reflectance UV-Vis spectra (**a**) and resulting band gaps from Tauc plots (**b**) of the studied LaFeO3 samples.

X-ray photoelectron spectroscopy (XPS) was performed on the most promising LaFeO3 S-700 to determine the elemental composition and the chemical oxidation state of the sample surface. Figure 7 shows the X-ray photoelectron (XP) survey spectrum (Figure 7a) and the detailed spectra of the La 3d, Fe 2p and O 1s. In addition to different La, Fe and O lines, Figure 7a shows also the C 1s signal, which resulted from adventitious surface carbon and which was also referenced to a binding energy (284.8 eV) in order to exclude surface charge effects for all the other signals. The binding energies found in the XP spectra of La 3d (Figure 7b) and Fe 2p (Figure 7c) revealed that the iron and lanthanum ions were both present in the chemical valence state +III [30,31]. Figure 7d shows the O 1s signal with binding energies of about 529.9 eV, 531.1 eV and 532.0 eV which corresponded as the main signal to the contribution of the La-O and Fe-O crystal lattice bonds, some surface hydroxyl groups and chemisorbed water, respectively [32,33]. In line with the expected composition between La, Fe and O, an atomic ratio of about 1:1:3 was found by comparing the relative signal intensities in the different XP spectra.

**Figure 7.** X-ray photoelectron spectroscopy (XPS) for the S-700 sample survey (**a**), and core level XPS of La 3d (**b**), Fe 2p (**c**) and O1s (**d**).

In order to investigate the energy band structure including the conduction band (CB) and valence band (VB) position of LaFeO3 S-700 and S-900 photocatalysts, electrochemical flat potential measurements were performed, and the resulting data are plotted in Figure 8a,b using the Mott–Schottky (MS) relation in the dark [34]:

$$\frac{1}{C^2} = \left[\frac{2}{\varepsilon\varepsilon\_0\varepsilon N}\right] \left[E - E\_{fb} - \frac{kT}{c}\right].$$

with *C* being the capacitance of space charge, *ε* the dielectric constant, *ε*<sup>o</sup> the permittivity of free space, *N* the electron donor density, *E* the applied potential and *Efb* the flat band potential. Plotting 1/*C*<sup>2</sup> against *E* yields a straight line from which the slope of the donor density can be calculated and *Ef*<sup>b</sup> can be determined as the intercept of the abscissa by extrapolation to *C* = 0. The positive slopes of the MS plots confirmed LaFeO3 being an n-type semiconductor with electrons as the majority charge carriers. The flat band potentials of S-700 (Figure 8a) and S-900 (Figure 8b) at frequencies of 100 Hz and 1000 Hz were calculated to be −0.3 and −0.25 V (*E*Ag/AgCl(sat-KCl)) referenced to the KCl-saturated Ag/AgCl electrode, respectively. Thus, using the following equation:

$$E\_{NHE} = E^0\_{A\text{g}/A\text{gCl}(sat - KCl)} + E\_{A\text{g}/A\text{gCl}(sat - KCl)} + 0.059 \times pH$$

with *E*<sup>0</sup> Ag/AgCl(sat-KCl) = 0.199. For the pH value of 5.6 of the 0.1 M Na2SO4 electrolyte solution, potentials were 0.23 and 0.28 V versus the normal hydrogen electrode (NHE) result. For n-type semiconductors the *E*fb was strongly related to the bottom of the conduction band (CB); typically, it is assumed that CB is 0.1 V more negative than *E*fb [35], resulting in CB edges at about 0.13 V and 0.18 V vs. NHE, respectively. These values are close to the position of the conduction band edge for YFeO3 calculated by Ismael et al. [36]. The slightly positive CB potential explains that the LaFeO3 was not able to form hydrogen via water splitting under light irradiation. This was confirmed by respective experiments attempting photocatalytic H2 formation with our LaFeO3 samples on which platinum nanoparticles (0.5 wt. %, particle size <2 nm) were photodeposited as a co-catalyst. Even by the use of light with λ ≥ 320 nm and methanol as a sacrificial agent, no hydrogen was detected with all the Pt/LaFeO3 samples. This result stands in contrast to H2 formation reported earlier by Tijare et al. [23], Parida et al. [37] and Vaiano et al. [38], who, however, performed no analysis on conduction band positions. Thus, some doubts regarding the H2 production reported in their papers exist.

Xu et al. [39] reported hydrogen production activity over a LaFeO3/g-C3N4 composite in the presence of TEOA as a sacrificial reagent and Pt as a co-catalyst. Their results show that LaFeO3 alone had no activity due to the positive conduction band edge (0.11 V); hydrogen was only found if g-C3N4 (conduction band potential at −0.85 V vs. NHE) was added, which is in agreement of our results. Hydrogen production was observed for other ferrites, such as CuFe2O4 and NiFe2O4; for those the conduction band positions were found to be negative enough [40,41].

By taking into account the band gap energies of our LaFeO3 samples from the Tauc plots (Figure 6b), the valence band (VB) positions for S-700 and S-900 can be calculated according to the equation *Evb* = *Ecb* + *Eg* [42], resulting in about *Evb* = 2.51 V and 2.68 V respectively.

**Figure 8.** Mott-Schottky plots at 100 Hz and 1 kHz of (**a**) S-700 and (**b**) S-900, (**c**) Nyquist plots of S-700 and S-900 in 0.1 M Na2SO4 (pH = 5.6) at 0.4 V vs. Ag/AgCl and (**d**) transient photocurrent responses in 0.1 M Na2SO4/0.1 M Na2SO3 solution (pH = 5.6) under white LED illumination.

Electrochemical impedance spectroscopy (EIS) and transient photocurrent experiments were performed to investigate the electron-hole separation efficiency in the LaFeO3 photocatalysts. The electrode of S-700 shows the smaller arc size (Figure 8c). In general, a smaller arc size observed in EIS semicircular Nyquist plots documents smaller charge-transfer resistance on the electrode surface and accelerated interface transport of charge carriers, which results in an effective photo-induced

charge carrier mobility and separation [43,44]. Figure 8d indicates the transient photocurrent responses of S-700 and S-900. The photocurrent of S-700 sample was much higher than that of S-900 indicating the greatly improved charge transfer and separation ability [45,46]. The onset potential of the photocurrent indicates the flat band potential of the electrode [47]. In this case, the sulfate/sulfite electrolyte solution lowered the kinetic barrier for charge transport by trapping the photogenerated holes. Moreover, the onset of the photocurrent lies at about 0.2 V vs. NHE for S-700, being in good agreement with the flat band potential obtained from the Mott-Schottky plot.

#### *2.2. Photocatalytic Properties*

Photocatalytic activities of the prepared LaFeO3 samples were evaluated by degradation of RhB and 4-CP in aqueous solution under visible light irradiation using a 420 nm cut-off filter. Before irradiation, the suspensions were magnetically stirred in the dark for 40 min to ensure adsorption-desorption equilibrium between the organic substrate and the photocatalyst, after visible light irradiation the absorbance of RhB was noticeably reduced (Figure 9a), although there was very little decrease in absorption before irradiation. This indicates that RhB degradation occurred instead of further adsorption. Since the intensity of the absorption peaks gradually decreased without any change in their wavelength, it can be concluded that the degradation reaction takes place by an aromatic ring opening without formation of stable de-ethylated intermediates [48,49].

Figure 9b shows that without LaFeO3 being present the dye RhB was quite stable and no significant self-degradation under visible light took place. Also in the presence of SnO2, a semiconductor with a band gap of 3.0 eV, which can, thus, not be excited by light with λ ≥ 420 nm, only negligible degradation was found. Thus, sensitization effects can be ruled out as well. In the presence of the photocatalyst LaFeO3, the photodegradation efficiency decreased with increasing calcination temperature of the LaFeO3 due to the decreasing surface area and increasing particle size. Besides the highest surface area sample facilitating the adsorption of the organic dyes and possibly trapping more electrons and holes on the surface, the sample S-700 might also benefit from the slightly narrower optical band gap allowing for more visible light absorption.

**Figure 9.** (**a**) UV-Vis spectra for the degradation of rhodamine B (RhB) under visible light irradiation (<sup>λ</sup> <sup>≥</sup> 420 nm) on the LaFeO3 S-700 sample ([RhB] = 10−<sup>5</sup> M, catalyst weight = 0.1 g) at 25 ◦C. (**b**) Degradation of RhB as a function of irradiation time with visible light (λ ≥ 420 nm) in the presence of different LaFeO3 catalysts or SnO2 for comparison.

The degradation of RhB obeys a pseudo-first-order kinetics law of the type [50]:

$$\ln c\_0/c\_t = kt$$

with *c*<sup>0</sup> being the initial concentration of RhB, *c*<sup>t</sup> the concentration of RhB at any time *t*, *t* the illumination time (min) and *k* is the first order rate constant (min−1). Figure 10 shows the linear relationships between ln(*c*0/*c*t) and *t*, the rate constant for S-700, S-750, S-850 and S-900 were 0.0062, 0.0032, 0.0026 and 0.0013 min<sup>−</sup>1, respectively.

In general, it is known that the photocatalytic degradation reaction of organic contaminants proceeds mainly by the contribution of oxygen-containing reactive species such as superoxide (**.** O2 −) and hydroxyl radicals (**.** OH) [51,52]. Thus, in order to explore the reactive species for RhB degradation different scavengers were tested in the photocatalytic process. The hydroxylation test was done using terephthalic acid (TA) as a probe molecule, in this test TA reacts with OH to produce highly fluorescent 2-hydroxyterepthalic acid (fluorescence maximum at 426 nm [53]).

**Figure 10.** Kinetics curves of LaFeO3 perovskites calcined at different temperatures.

As seen from Figure 11a for S-700 very low fluorescence intensity at 426 nm was observed after 6 h of visible light irradiation suggesting very low hydroxyl radical formation on the surface of the catalyst. Usually, the photoluminescence (PL) intensity at about 425 nm is proportional to the amount of the produced hydroxyl radical on the surface of the catalyst. In good agreement to that, isopropanol (0.01 M), a known hydroxyl radical (**.** OH) quencher [54], showed only little effect on the RhB degradation reaction (Figure 11b). However, the addition of benzoquinone (0.01 M) [55] as a superoxide radical (**.** O−2) quencher strongly decreased the degradation of RhB, indicating that degradation proceeds via superoxide radicals, which can only be produced via the reduction of dissolved oxygen by the excited electrons in the conduction band (CB) of LaFeO3. This is surprising since the CB of LaFeO3 S-700 was detected to be at about 0.1 V, being more positive than the standard potential for the superoxide radical formation from adsorbed oxygen *<sup>E</sup>*0(O2/•O2 <sup>−</sup>) = −0.046 V [56]. Thus, the superoxide radical formation should not be possible. However, in the photocatalytic experiment, the electrochemical standard conditions were not given, thus due to potential shifts depending on the Nernst law, the superoxide radical formation might become possible to some extent. An indication for potential changes during the photocatalysis experiment might become visible in the increasing degradation of RhB in the presence of benzoquinone, which occurs with longer irradiation time (Figure 11b). Approximately the same decrease in activity for RhB degradation is obtained when 10 vol. % methanol [57] were added as a hole (h+) quencher. The photogenerated holes can oxidize OH<sup>−</sup> ions to **.** OH radicals because the valence band position of S-700 (2.46 V) is more positive than the redox potential of **.** OH/*−*OH (*E*<sup>0</sup> = 1.99 V) [58]. These results for the active species being responsible for degradation agree with the results for the perovskite YFeO3 studied earlier by us [36]. In that paper, it was concluded that superoxide radicals (**.** O2) and (**.** OH) have an effect but the holes (h+) are the main species on catalyst surfaces responsible for the photocatalytic activities.

**Figure 11.** (**a**) Fluorescence spectra of the 2-hydroxyterepthalic acid solution in the presence of LaFeO3 S-700 and (**b**) the reactive species in trapping experiments during the degradation of RhB.

Figure 12 shows that the LaFeO3 S-700 was not only able to degrade dyes like RhB under visible-light irradiation at λ > 420 nm, but also compounds like 4-CP which does not absorb lights themselves in that spectral range. Thus, a light-induced self-degradation can be ruled out. With LaFeO3 S-700 the degradation efficiency on 4-CP was lower than that on RhB, but still more than 60% of the 4-CP were transformed by destroying the aromatic ring system, which is responsible for the absorption at 315 nm, which was recorded and followed with time in Figure 12. In a HPLC analysis performed with the reaction mixture after 5 h of illumination, no significant amounts of decomposition products were found. Total organic content (TOC) analysis after five hours illumination time confirmed the 4-CP degradation; a reduction of the TOC by 55% was found. Our LaFeO3 photocatalyst showed higher activity for 4-CP degradation compared to that reported by Pirzada et al. [59] and Hu et al. [60].

**Figure 12.** Degradation of 4-CP as a function of irradiation time without a photocatalyst and in the presence of the LaFeO3 catalyst S-700.

## **3. Experimental**
