Fe₃O₄ (0.005 g, 0.0125 g, 0.025 g, 0.05 g, 0.25 g and 0.5 g, respectively) was weighed into six beakers. Anhydrous ethanol (40 mL), tetraethyl orthosilicate (8 mL), ammonia (10 mL) and distilled water (1 mL) were added to each beaker. The resulting suspensions were stirred at room temperature for 5 h, and then dried at 105 °C. The resultant particles were Fe₃O₄/SiO₂.

The Fe₃O₄/SiO₂ particles prepared with different amounts of Fe₃O₄ were suspended in anhydrous ethanol (40 mL) in one of six beakers and then acetic acid (15 mL), tetra-n-butyl titanate (21 mL) and distilled water (8.6 mL) were added to each of these six beakers. The mixtures were further stirred for 2 days, and then dried at 105 °C. The solid particles were ground to pass 500 mesh, and calcined at 500 °C (heating rate is 2 °C/min) in a muffle furnace. The resultant particles were Fe₃O₄/SiO₂/TiO₂, with theoretical Fe₃O₄ contents of 0.1%, 0.25%, 0.5%, 1%, 5% and 10%, respectively. TiO₂ was prepared by the same method without any Fe₃O₄/SiO₂.

X-ray diffraction (XRD) patterns were collected in a XD-2 instrument (Persee, China) using Cu Kα radiation. Scanning electron microscopy (SEM) images were collected on an S-4800 field emission scanning electron microscope (Hitachi, Japan). Brunauer-Emmett-Teller (BET) surface areas were measured by nitrogen adsorption at 77.35 K on an ASAP-2010 adsorption apparatus (Micromeritics, USA).Fourier-transform infrared (FTIR) spectra were collected on an IRPrestige-21 type instrument (Shimadzu, Japan) running at 2 cm⁻¹ resolution. The UV-vis spectra were collected on a UV-3010 UV-visible spectrometer (Hitachi, Japan) using BaSO₄ as a reference. Glyphosate concentration was measured on a F-7500 fluorescence spectrometer (Hitachi, Japan).

The concentration of glyphosate standard stock solution was 1,000 mg/L. Soil for experiments was typical red loam, which was collected from the Banan District, Chongqing, China. The crushed soil was dried at 105 °C. Soil samples (50 g) and 4 mL/100 g (soil) of glyphosate solution were weighed into a glass dish (diameter 120 mm). Moisture content of this mixture was adjusted to a predetermined value using distilled water. Then a certain amount of photocatalyst was added and mixed well. The dish was placed under sunlight for 2 h. To keep the soil moisture constant, a certain amount of distilled water was added every 15 min by weighing the dish to calculate the water volume, while mixing the sample at the same time. After 2 h irradiation all the soil was suspended in distilled water, and separated by centrifugation after 10 min washing. This process was repeated four times, and then the solution volume was adjusted to 500 mL sing distilled water. The concentration of glyphosate was determined by a fluorescence quenching method [18]. Three parallel experiments were carried out for each experiment, and the results were the average of these three experiments. The glyphosate degradation by photolysis used for comparison was carried out without any photocatalyst.

The XRD patterns of photocatalysts are compared in Figure 1. Only the diffraction peaks of typical anatase TiO₂ were observed with 2θ at 25.3°, 37.9°, 48.2°, 54.0°, 55.1°, 62.6°, 68.8°, 70.3°and 75.2°. Diffraction peaks of titanate and Fe₃O₄ crystal didn’t appear. This indicates that the Fe₃O₄ is well encapsulated by SiO₂, and no interaction between Fe₃O₄ and TiO₂ occurs.

Figure 2 shows the scanning electron microscopy (SEM) images of TiO₂ and Fe₃O₄/SiO₂/TiO₂ with 0.1%, 0.5%, 10% Fe₃O₄, respectively. As shown in the figure, the crystal sizes of these four catalysts were all about 1 μm, but agglomeration of catalysts ocurred when Fe₃O₄/SiO₂ was added, and this phenomenon became more apparent as the Fe₃O₄/SiO₂ content increased. The Fe₃O₄/SiO₂ particles play a vital role and are encapsulated by Ti(OH)₄ constantly during hydrolysis, so the particle size of the Fe₃O₄/SiO₂/TiO₂ catalyst ultimately increases and its dispersion properties weaken. SEM characterization shows that excess Fe₃O₄/SiO₂ in the Fe₃O₄/SiO₂/TiO₂ photocatalyst may lead to a sharp decrease of photocatalyst dispersion properties.

The BET surface area, average pore diameter and total pore volume of all catalysts are listed in Table 1. Characterization results showed that Fe₃O₄ lowered BET surface area and total pore volume while the average pore diameter of Fe₃O₄/SiO₂/TiO₂ photocatalyst increased. These changes are also due to dispersion property variation caused by the addition of Fe₃O₄.

Functional groups on different photocatalysts were confirmed by Fourier-transform infrared spectroscopy. The FTIR characterization results are presented in Figure 3. As shown in the spectra the Si-O-Si asymmetric stretching vibration peaks [19] around 1,091.71 cm⁻¹ and 1,222.67 cm⁻¹ appeared when the Fe₃O₄ load reached 0.5%. This demonstrated that a certain amount of SiO₂ had formed on the catalysts, and these peaks became increasingly apparent as the Fe₃O₄ load continued to increase. The asymmetric stretching vibration [20] of Ti-O-Si around 936.19 cm⁻¹ appeared when the Fe₃O₄ loading reached 1%. This peak indicated that certain amount of Si had replaced Ti in the TiO₂ lattice. The peak at 480.28 cm⁻¹ was assigned to Ti-O stretching vibrations. There was no evidence for any interaction between Fe and Ti in the spectra. It shows that Fe₃O₄ is encapsulated by SiO₂ very well, therefore the photocatalytic activity of Fe₃O₄/SiO₂/TiO₂ is not reduced by this interaction. Because of the role of capturing electrons, less Ti-O-Si will promote isolation of hole-electron pairs, enhancing the photocatalytic activity. However, excessive Ti-O-Si will produce recombination centers for holes and electrons, and reduce the photocatalytic activity.

The UV-Vis absorption properties of different catalysts are shown in Figure 4. In this figure a clear absorption peak can be found in the visible region (λ ≥ 420 nm) on every catalyst spectrum with Fe₃O₄. All spectra still maintained the absorption characteristics of TiO₂ in the ultraviolet region, indicating that there is no interaction between TiO₂ and Fe₃O₄, because of the separation effect of SiO₂. UV-Vis characterization demonstrates that photocatalysts with Fe₃O₄ have excellent photoluminescence capability in both the visible and ultraviolet regions. These photocatalysts may exhibit higher photocatalytic activity than TiO₂ without Fe₃O₄/SiO₂. It can also be seen from this figure that the Fe₃O₄ loading affected the visible light absorption properties of the catalysts significantly. Absorption capacity was enhanced by increasing Fe₃O₄ loading when the Fe₃O₄ loading was less than 0.5%, then it weakened when the Fe₃O₄ loading was more than 0.5%. The 0.5% Fe₃O₄/SiO₂/TiO₂ has the strongest absorption capacity and thus may have the best photocatalytic activity.

Six samples were prepared with 50 g soil, 0.2 g photocatalysts and 2 mL glyphosate standard stock solution. The photocatalyst in each sample had different Fe₃O₄ loading. The moisture content of samples was adjusted to 40%. Experimental results after 2 h irradiation (sunlight intensity was 11.88 mW/cm²) are shown in Figure 5. It can be seen that catalysts with Fe₃O₄ had better activity than TiO₂. Fe₃O₄ greatly enhances the efficiency of visible light absorption, so Fe₃O₄/SiO₂/TiO₂ catalysts show excellent photocatalytic activity under sunlight irradiation. Using Fe₃O₄ as photocatalyst, the degradation of glyphosate was only 53.04%, which was much lower than with 0.5% Fe₃O₄/SiO₂/TiO₂ where the degradation was 78.84%. It indicates that Ti-O-Si bond in Fe₃O₄/SiO₂/TiO₂ catalyst can reduce the probability of the hole-electron pair recombination, thus improving the photocatalytic activity of the catalyst.

Degradation of glyphosate increased from 34.42% to 78.84% when the Fe₃O₄ load was varied from 0 to 0.5%, correspondingly. Degradation efficiency declined when more than 0.5% Fe₃O₄ was loaded. The Fe₃O₄ in the catalysts has two kinds of effects. On one hand, it can enhance the visible sunlight absorption properties of the catalyst. On the other hand, the interaction between Si and Ti has a negative effect on the photocatalytic activity when the Fe₃O₄ loading is excessive. Which aspect is more important depends on the amount of Fe₃O₄. Characterization through FTIR shows that Ti-O-Si bonds become apparent in the catalyst with 1% Fe₃O₄ loading, which illustrates that Ti-O-Si has become a recombination center for holes and electrons, so photocatalyts with 1%, 5% and 10% Fe₃O₄ decrease the photocatalytic activity. 0.5% Fe₃O₄/SiO₂/TiO₂ was thus chosen for the subsequent experiments.

Eight samples were prepared with 50 g soil, 0.2 g 0.5% Fe₃O₄/SiO₂/TiO₂ photocatalyst and 2 mL glyphosate standard stock solution. The moisture content was adjusted to a reference value. Experimental results are shown in Figure 6 (sunlight intensity was 11.73 mW/cm²). In view of Figure 6, degradation of glyphosate was not ideal when the moisture content was lower than 30% or higher than 50%. Degradation was basically at a higher level with a moisture content ranging from 30% to 50%. The moisture content of samples affects the glyphosate distribution among the water, soil and air phases in soil. In addition, the moisture content also affects the amount of free radical formation on the catalyst surface. Glyphosate cannot diffuse to the catalyst surface fast when the moisture content is lower than 30%. Only if the glyphosate is adsorbed on the surface of photocatalysts, can it be degraded because the lifetime of free radicals is reportedly very short, and these free radicals are practically produced on surface of photocatalysts. Therefore, a low moisture content hinders the degradation of glyphosate. On the contrary, O₂ cannot diffuse to the surface of photocatalysts, when the moisture content is higher than 50%. This effect results in a significant decrease in the production of free radicals, because O₂ is one of the most important substances in the free radical generation process. Therefore too much water in soil also leads to a lower removal rate. The most efficient degradation of glyphosate ocurrs when the moisture content is between 30% and 50% for the diffusion of pollutants and O₂ balance. The moisture content in subsequent experiments was set at 40%.

A certain amount of 0.5% Fe₃O₄/SiO₂/TiO₂ catalyst and 2 mL glyphosate standard stock solution were separately added to nine samples. The moisture content of all samples was adjusted to 40%. Degradation of glyphosate irradiated 2 h by sunlight is shown in Figure 7 (sunlight intensity was 11.79 mW/cm²). As seen in the figure, it can be concluded that an increase in photocatalyst dosage accelerated the rate of degradation of glyphosate. Degradation was proportional to the photocatalyst dosage when the latter was less than 0.2 g. It increased rapidly. Nevertheless, further photocatalysts led to a slow increase of degradation. There is a maximum value of glyphosate diffusions rate because its concentration in the sample is limited. When this maximum value is reached, excess photocatalyst is useless. Glyphosate concentration is the limiting factor in the reaction under these conditions, so the reaction rate tends to remain constant.

Soil samples (25 g, 50 g, 75 g, 100 g and 150 g) were placed separately in five glass dishes (120 mm diameter). Photocatalyst (0.5% Fe₃O₄/SiO₂/TiO₂) dosage and glyphosate standard stock solution amounts were 0.4 g/100 g (soil) and 4 mL/100 g (soil), respectively. Moisture content of soil was raised to 40%. The thicknesses of these samples were 2.2 mm, 4.1 mm, 6.2 mm, 8.4 mm and 10.5 mm after tiling. Degradation of glyphosate is shown in Figure 8(a) (sunlight intensity was 11.97 mW/cm²). In view of this figure, we can simply conclude that the thicker the sample was, the lower the degradation rate was, as 87.33% glyphosate was removed after 2 h irradiation when the soil thickness was 2.2 mm., whereas on the contrary, the removal rate was reduced to 38.21% when thickness was 10.5 mm. This is because the sunlight cannot reach deep inside the soil, so the necessary conditions for the photocatalytic degradation in this part of soil are absent and suggesting that the photocatalytic reaction only occurs in the surface part of soil and degradation decreases as the soil layer becomes thicker.

However, Figure 8 (b) shows that the total mass degradation of glyphosate in 2 h kept increasing despite the fact the degradation went down. Photocatalysis occurs in all levels of the soil when the soil thickness declines, because sunlight can penetrate all the layers. Under these conditions, increasing the thickness can promote the amount of glyphosate involved in photocatalysis, so the mass degradation increases rapidly. When the soil thickness is beyond sunlight exposure capacity, no more soil and glyphosate are involved in photocatalytic reaction, so glyphosate degradation mass in 2 h tends to a fixed quantity. It is shown in Figure 8 (b) that the maximum degradation amount within 2 h in this system is about 1.9 mg.

All samples were prepared with 50 g soil, 0.2 g 0.5% Fe₃O₄/SiO₂/TiO₂ photocatalyst and 2 mL glyphosate standard stock solution. Moisture content of all samples was adjusted to 40%. These samples were irradiated under different weather conditions. According to the light intensity data, the experimental results shown in Figure 9 suggest that the glyphosate degradation rate increased slowly with sunlight intensity. The removal rate was 67% when the light intensity was about 6 mW/cm², and it reached 86% when the light intensity was about 16 mW/cm². Glyphosate was degraded faster when the light intensity exceeded 10 mW/cm². We tentatively interpret this phenomenon as a consequence of the temperature change of the soil. Generally, when it is cloudy and the temperature is lower when the irradiation intensity is below 10 mW/cm². Higher sample temperatures are conductive to the photocatalytic degradation.