TGA analysis of gallium-based materials was carried out under an oxygen atmosphere during the whole characterization, from ambient temperature up to 850°C. Using oxygen allowed us to identify the phase transformation to its oxidation state and determine the thermal treatment conditions of samples. Figure 2A shows the weight percentage loss and weight derivative as a function of the sample temperature for GO and GOFe. From the weight loss, it is possible to distinguish three temperature ranges: 1) a ≈ 4 % weight loss from 28 to 256°C attributed to superficial and adsorbed water in the structure, 2) a steeper weight loss (about 9 – 10%) from 256 to 380°C attributed to the detachment of hydroxyl groups which indicates the elimination of the adsorbed moisture while the material is exposed to oxygen atmosphere, and 3) at a temperature higher than 380°C a weight loss ( ≈ 1 − 2 % over 450°C) due to the oxidation process of GaO(OH) to Ga₂O₃ and crystallization of Ga₂O₃. The weight derivative plots can confirm this. The thermal behavior is consistent with reported gallium-based materials (Quan et al., 2010; Reddy et al., 2015).

From the heat flow plots for GO and GOFe, Figure 2B, it is possible to distinguish two endothermic peaks at 315 and 375°C, respectively. These peaks can be attributed to the transformation of GaO(OH) into α-Ga₂O₃ (Reddy et al., 2015; Shan et al., 2017; Yuwen et al., 2020). In addition, a less pronounced endothermic peak around 730°C can be observed. This peak is related to the phase transformation from α-Ga₂O₃ to β-Ga₂O₃, being more pronounced for GOFe material (Reddy et al., 2015; Shan et al., 2017; Yuwen et al., 2020).

XRD patterns for all GO materials are shown in Figure 3A. The diffraction peaks in the XRD pattern of GOH can be indexed to GaOOH according to JCPDS 06-0180, which has a crystallized orthorhombic structure (Reddy et al., 2015; Huang et al., 2017; Sharma et al., 2020). In this case, the representative peaks to crystalline planes are (110), (130), (111) and (240) at diffraction angles, 2θ = 21.50, 33.78, 37.29 and 54.02°, respectively. In turn, the XRD patterns for GO materials show the formation of α and β phases, depending on the heat treatment. The samples GO 500°C and GO 650°C exhibit the same peaks, which correspond to the rhombohedral structure of the α phase (Huang et al., 2017; Sharma et al., 2020). The representative peaks of the α phase are 2θ = 33.90 and 36.16° for (104) and (110) crystalline planes. From these two patterns, it can be observed that increasing the temperature treatment favors the crystallization process of the sample, leading to more defined signals corresponding to the α phase according to JCPDS 158 06-0503. GO 800°C and GO 950°C samples have higher order diffraction lattice planes, which implies a more crystalline material than can be readily indexed to the monoclinic structure of β-Ga₂O₃ phase according to JCPDS 76-0573 (Quan et al., 2010; Guo et al., 2017; Cheah et al., 2020; Sharma et al., 2020). The representative peaks corresponding to the crystalline planes are (401), (002), (111) and (512) at diffraction angles 2θ = 30.47, 31.87, 35.43 and 64.74, respectively. These results are consistent with the thermogravimetric analysis, where phase transformations are observed. The β-Ga₂O₃ has been reported to be the more stable phase among the different gallium oxides phases (Stepanov et al., 2016; Tak et al., 2021). According to previous results, in this work GO was treated thermally at 950°C and was used as photocatalyst.

XRD patterns of Go and GOFe ion-doped gallium oxides are presented in Figure 3B, where for the sake of comparison the β-Ga₂O₃ pattern (GO 950°C or simply GO) is also included. The XRD pattern of GOFe has representative peaks of iron gallium oxide (Fe_2.1Ga_0.9O₄) for the cubic structure (JCPDS 01-074-2226) and iron oxide (Fe₂O₃) for rhombohedral structure (JCPDS 01-084-0311). These peaks overlap with those of the gallium oxides (GO). However, the height of peaks indicates the presence of iron gallium oxide and iron oxides. The peak at 2θ = 30.23° corresponds to the crystalline plane (220) for iron gallium oxide material. For iron oxides, the peak at 2θ = 62.4°, crystalline plane (214) is observed.

The FT-IR spectra for the gallium-based materials synthesized at pH of 10: GOH, GO heat treatment at 500, 650, 800 and 950°C, as well as GOFe (which is synthesizing at 950°C) are presented in Figure 4. For GOH, the broad band at 2,960 and 3,210 cm⁻¹ can be attributed to the stretching vibration of H–O–H band and O–H stretching of GaO(OH). The band at 1,980 cm⁻¹, as well as the bands at 1,020 cm⁻¹ and 942.1 cm⁻¹, can be assigned to constitutional Ga–OH bending bands and their overtones, respectively. The stretching vibration bond of the O–H group is also observed around 1,360 cm⁻¹ owing to the absorption of water molecules (Quan et al., 2010). In turn, the FT-IR spectra of remaining materials, GO treated at 500–950°C and GOFe at 950°C, reveal a broad H–O–H stretching band at around 3,395 cm⁻¹. In addition, heat treated GO and GOFe materials show a band at 600–700 cm⁻¹, which is assigned to the valence vibrations of Ga–O in the lattice formed by GaO₆ octahedra and GaO₄ tetrahedra, corresponding to corundum and monoclinic structures of α and β crystalline phases, respectively. These results are consistent with those reported by Quan et al. (2010) and Girija et al. (2013). In FT-IR spectra for GOFe, species associated with Fe ions are not identified.

Evaluating the photocatalytic activity of synthesized materials requires determining the experimental conditions for pH, and the concentrations of catalyst, pollutant, and oxidizing agent. This was achieved by carrying out several experiments using GO as the catalyst at a pH = 3.00 under UV-Visible light. It is noteworthy to mention that the fast rates of electron-hole recombination exhibited by Ga₂O₃ require an oxidizing agent (Sun et al., 2013), which in this case is H₂O₂. In the photocatalytic process, hydroperoxyl ions are reduced by the electrons in the conduction band, yielding hydroperoxyl radicals, which can react with RhB molecules and carries out the discoloration process (Hidaka and Tsukamoto, 2019). Therefore, it is important to ensure that the concentration of H₂O₂ does not mask the photocatalytic activity of the interest materials. This is shown in Figure 5A. Photolysis of RhB dye, at C ₀ = 0.022 mM, was verified by running an experiment without catalyst in the presence of 7 and 48 mM of H₂O₂ and under Visible-UV. Results show that about 9% of discoloration is obtained for 7 mM after 300 min of reaction. When 48 mM of H₂O₂ is used, a complete discoloration of RhB dye is observed during the first 150 min. This can be attributed to reactions with the hydroperoxyl radicals (HO₂ ⁻), the radical which are formed by the interaction of UV radiation with the oxidizing agent. In addition, it is important to consider the adsorption/desorption process, which is crucial to carry out the pollutant degradation by photocatalysis. The adsorption/desorption process of RhB dye molecules on GO catalyst was analyzed to determine the time in which the adsorption equilibrium is reached. This experiment was performed under dark conditions at C ₀ = 0.022 mM of RhB, a pH = 3.00 and in absence of H₂O₂. It can be observed that adsorption equilibrium is reached within the first 30 min of the contact between GO catalyst and RhB dye. After this time, changes in RhB concentration are not significant. Similar results (30 min) have been reported by several authors when working with Ga₂O₃ catalysts (Sun et al., 2013; Reddy et al., 2015; Magdalane et al., 2017; Du et al., 2021).

The individual effect of concentration of H₂O₂, RhB, and GO catalyst on the discoloration process under UV-Visible light can be observed in Figure 5B. Increasing the oxidizing agent concentration while keeping other parameters constant leads to a faster discoloration process. Nevertheless, this decrease in the discoloration time can be attributed mainly to the presence of H₂O₂ and not to the GO photocatalytic activity. Using the oxidizing agent aims to slow down the recombination process of the photogenerated charges and not contribute to the discoloration process. A H₂O₂ concentration of 7 mM, 2:1 catalyst to oxidizing agent ratio, has been used, as reported by (Palanivel et al., 2021). In turn, increasing RhB concentration from 0.022 to 0.045 mM leads to a longer discoloration time. To keep discoloration time in 300 min as the reference, RhB concentration is fixed to 0.022 mM, which is similar to that reported in other works (Reddy et al., 2015; Das et al., 2019; Du et al., 2021). Finally, three GO catalyst concentrations have been investigated. The higher the catalyst concentration, the faster the discoloration process. Note that in this case, there is no linear decrease in discoloration time with catalyst concentration, which indicates that high catalyst concentrations inhibit light transmission through the heterogeneous dispersion (Hidaka and Tsukamoto, 2019). It has been reported Ga₂O₃ concentrations ranging from 0.5 g L⁻¹ (Sun et al., 2013) to 1 g L⁻¹ (Reddy et al., 2015; Das et al., 2019; Du et al., 2021). Therefore, according to the results obtained in this section, the catalyst concentration used to continue the experiments was 0.5 g L⁻¹.

The photocatalytic activity of GO and GOFe catalysts, using RhB dye as a model pollutant, was tested under three different pH values and two light sources. Results reported in this section corresponds to GO and GOFe synthesized via precipitation with NH₄OH, at pH = 10.00, drying temperature at 50°C and heat treatment at 950°C. A three factor 3 × 2 × 2 experimental design has been used in this work with RhB discoloration as the response variable. The independent variables consisted on pH (3.00, 5.00 and 9.00), catalyst (GO and GOFe) and light source (Visible and UV-Visible) leading to 12 distinct conditions. All experiments were carried out at the following concentrations: 0.022 mM of RhB, 7 mM of H₂O₂ and 0.5 g L⁻¹ of catalyst, as established in Section 3.2.1.

Solutions of RhB dye were prepared at an initial concentration of 0.022 mM. The discoloration of the RhB dye was followed by measuring the evolution of the absorbance peak of the solution in the visible region (chromophore group). RhB dye concentration is proportional to the measured absorbance caused by the chromophore group of the molecule. The chromophore group containing the structure amino (NH) absorbs in the visible region (λ _char = 0.556 μm for a pH = 3.00 and at λ _char = 0.554 μm for pH = 6.00, 7.00 and 9.00), while the aromatic rings absorb in the UV region. The chromophore group shift is attributed to the charge separation of the carboxylic anion (R⁻) and proton (H⁺), which accepts the same resonant structures. Therefore, it exhibits a similar absorption spectrum with a shift in the chromophore group. Figure 6 shows the temporal evolution of the dye absorption spectra at concentrations of 0.022 mM of RhB, 7 mM of H₂O₂, 0.5 g L⁻¹ of GOFe catalyst and a pH = 9.00, under Visible-UV light. In this figure, the disappearance of the chromophore group can be observed as the photocatalytic reaction proceeds.

The discoloration profiles of the RhB under Visible-UV and Visible light, using GO as photo catalyst, are presented in Figure 7A. Photolysis of RhB dye was verified by running an experiment without catalyst, in presence of 7 mM of H₂O₂ and under Visible illumination. RhB discoloration by photolysis was not observed under visible light. These later results complement those presented in Figure 5A.

The discoloration process was performed in two stages: adsorption (the first 30 min) followed by the photocatalytic reaction during the next 300 min. In Figure 7 these processes are indicated by a different background color. In Figure 7A, it can be observed that acid conditions favor the adsorption process. At a pH = 3.00 more than 5% of the dye adsorbs, while at pH = 5.00 and pH = 9.00 the amount of adsorbed RhB molecules is lower. This behavior can be explained by the interactions of the catalyst’s surface charge and ionization of RhB. It has been reported that the isoelectric point for G₂O₃ is given at pH = 9.00 (Kosmulski, 2001). Therefore, below this pH value, the surface of GO is charged positively, favoring the attraction of the carboxylic anion (R⁻) of the RhB dye. The adsorption process influences discoloration since a larger amount of RhB dye molecules adsorbed on the catalyst’s surface, increasing the discoloration rate of the RhB dye. Discoloration process is reached after 300 min with a pH = 3.00 under Visible-UV light for GO, Figure 7A. In contrast, at pH = 5.00 and 9.00, about 40 and 50% of color remain, respectively, for the same period. Under visible light, 33% of discoloration was obtained with a pH = 3.00 after 300 min of reaction but no significant discoloration was observed for pH = 5.00 and 9.00. In literature it has been reported a 90% discoloration after 180 min of reaction using β-Ga₂ O ₃ nanoblocks and nanospindles Girija et al. (2013). Nevertheless, they used a 150 W Xenon lamp, in contrast to the results presented in this work in which about 80% of discoloration is observed but under a 13 W ( ∼ 11.5 times less power) UV-Visible light source. Du et al. (2021) reported a 50% discoloration in the 60 min with Ga₂O₃ nanofibers using a 300 W Xenon lamp, while a 25% is obtained in this work with ∼ 23 times less power.

Now we turn our attention to the discoloration process when GOFe is used as the photocatalyst, shown in Figure 7B under Visible-UV and Visible light sources. Experimental conditions are the same as in previous section (for GO) except for the catalyst which is GOFe. Note that now the discoloration process is completed after 300 min of reaction at a pH = 3.00 with either light source. Results comparable to those obtained for GO (see Figure 7A) were observed. It is noteworthy to mention that the discoloration process is achieved with Ga₂O₃ catalyst doped with iron ions under Visible illumination conditions. This suggests that the incorporation of iron ions promotes a shift of the catalyst absorption band towards lower values (Li et al., 2018; Wang et al., 2018). Additionally, for those cases in which a complete discoloration is obtained, two behaviors are distinguished: (1) a slow discoloration rate in the first instances followed by (2) a faster discoloration process. This behavior can be attributed to the surface reactions between the charge pair (e ⁻-h ⁺) generated in the catalyst and its reaction with H₂O₂ and RhB dye.

The statistical analysis has been performed by a two-way ANOVA with interaction considering pH (3.00, 5.00, and 9.00), catalyst (GO and GOFe) and light source (Vis and Vis-UV) as independent variables, and dye (RhB) concentration as the dependent variable at all reported experiment instants (t = 30, 60, 90, 150, 210, 270, and 330 min). The F-statistic and p-values are presented in Table 1. The significance level is set at 0.05. In this Table, all those cases in which p ≤ 0.05 are indicated in bold font. It was found a statistically-significant difference in dye concentration by pH levels through all photocatalytic process. At the first instances of the process (t = 30 and 60 min) only pH exhibited a significant difference. Light source and the interactions between the three factors (Catalyst*pH, Catalyst*Light and pH*Light) are significant t = 90–210 min, where 50% of the discoloration occurs. At the latter times (t = 270 and 330 min) pH and light source are significant. These results are consistent with observations on the discoloration profiles. Moreover, a Scheffe post-hoc analysis revealed that pH = 3.00 and pH = 9.00 showed a significant difference, but pH = 3.00 and pH = 5.00 and pH = 5.00 and pH = 9.00 did not.

Figure 8 presents the two-way interactions at those sampling times where these effects are significant (t = 90, 150 and 210 min). All interactions are ordinal. Although lines cross in Catalysts*pH and Catalyst*Light interactions, their slopes have the same sign. Catalysts*pH and pH*Light interaction plots confirm that acid conditions favor the discoloration process (steeper slopes), while no strong interaction for pH = 9.00 and 5.00 suggest their weak influence. A synergistic effect is observed in pH*Light plots, from which a pH = 3.00 under Vis-UV illumination enhances the discoloration rate. Statistical analysis corroborates the results described above.

Practical applications of catalysts require materials to be stable under illumination and reuse (cycling) (Sun et al., 2013; Magdalane et al., 2017; Nguyen et al., 2022). Figure 9A presents the discoloration process of RhB dye under Visible-UV illumination for GO. The stability study was performed for three cycles with a reaction time of 300 min each. The experiments were carried out at pH = 3.00, 0.5 g L⁻¹ of catalyst concentration and 7 mM of H₂O₂. Section 2.5 details the procedure for each experiment. At the end of the photocatalytic reaction, the photocatalyst was recovered, dried at room temperature, and used for the next experiment cycle. This procedure was repeated for each cycle.

The differences in the process can be retrieved from the discoloration rate constants obtained from fitting models, which for this case is done with a pseudo zero-order kinetic model (Eq 3). According to the results, the discoloration rate constants k ₀ for cycles one, two and three were 0.0001035 ± 0.0000055 mM/min (R ² = 0.9775 ± 0.0032), 0.0000925 ± 0.0000045 mM/min (R ² = 0.9956 ± 0.0017), and 0.000096 ± 0.000005 mM/min (R ² = 0.9914 ± 0.0014) respectively. These results indicate that GO catalyst retains its photocatalytic performance even after third cycle, which is consistent with other studies (Das et al., 2019).

An analog procedure can be done for the discoloration process of RhB using GOFe, presented in Figure 9B. Results demonstrate that the kinetic constants k ₀ for the discoloration rate of cycles one and two are similar, namely, 0.0000655 ± 0.0000055 mM/min (R ² = 0.9735 ± 0.0056) and 0.0000645 ± 0.0000065 mM/min (R ² = 0.9496 ± 0.0053), respectively. In turn, for cycle three this constant is 0.000125 ± 0.000012 mM/min (R ² = 0.9887 ± 0.0056), which represents a two-fold increase of the discoloration rate. This increase can be attributed to a change in the surface charge of the iron-doped gallium oxides, that promotes a bigger number of RhB dye molecules to be absorbed.