Novel magnetic Fe₃O₄/g-C₃N₄/MoO₃ nanocomposites with highly enhanced photocatalytic activities: Visible-light-driven degradation of tetracycline from aqueous environment

Abstract

In the present work, a series of magnetically separable Fe₃O₄/g-C₃N₄/MoO₃ nanocomposite catalysts were prepared. The as-prepared catalysts were characterized by XRD, EDX, TEM, FT-IR, UV-Vis DRS, TGA, PL, BET and VSM. The photocatalytic activity of photocatalytic materials was evaluated by catalytic degradation of tetracycline solution under visible light irradiation. Furthermore, the influences of weight percent of MoO₃ and scavengers of the reactive species on the degradation activity were investigated. The results showed that the Fe₃O₄/g-C₃N₄/MoO₃ (30%) nanocomposites exhibited highest removal ability for TC, 94% TC was removed during the treatment. Photocatalytic activity of Fe₃O₄/g-C₃N₄/MoO₃ (30%) was about 6.9, 5, and 19.9-fold higher than those of the MoO₃, g-C₃N₄, and Fe₃O₄/g-C₃N₄ samples, respectively. The excellent photocatalytic performance was mainly attributed to the Z-scheme structure formed between MoO₃ and g-C₃N₄, which enhanced the efficient separation of the electron-hole and sufficient utilization charge carriers for generating active radials. The highly improved activity was also partially beneficial from the increase in adsorption of the photocatalysts in visible range due to the combinaion of Fe₃O₄. Superoxide ions (·O₂⁻) was the primary reactive species for the photocatalytic degradation of TC, as degradation rate were decreased to 6% in solution containing benzoquinone (BQ). Data indicate that the novel Fe₃O₄/g-C₃N₄/MoO₃ was favorable for the degradation of high concentrations of tetracycline in water.

Introduction

Tetracycline (TC) has been widely used to treat bacterial infections in humans and animals over the past few decades [1]. Besides for medical applications, TCs are also employed as a supplement in animal husbandry to promote animal growth [2]. However, due to the widespread use of TC, TC residues could be frequently detected in various environmental matrices [3,4]. The residual TC in the environment would seriously threaten the ecosystem and public health [5]. In general, TC cannot be effectively removed by conventional wastewater treatment processes, such as biological treatment [6]. Therefore, new techniques are required to remove TC in water. Recently, photocatalytic assays have received a lot of intensive research interest worldwide due to its high efficiency and reliability, and have emerged as highly effective techniques for TC degradation from water [7]. Some photocatalysts have the function of degrading pollutants while Excellent antibacterial activity [8,9]. Common traditional photocatalysts, such as ZnO, TiO₂ [10], have been confirmed can degrade TC under light irradiation. However, in practical applications, these photocatalysts couldn’t fully utilize solar energy, which causes them to be hindered in practical applications [11]. Therefore, the high efficient sunlight-driven photocatalysts are received lots of attention for the photocatalytic treatment of TC.

Graphitic carbon nitride (g-C₃N₄) has a strong visible-light response due to it easily produced electrons and holes under visible-light irritation [12]. Due to its advantages of low toxicity, low preparation cost and high stability, it has been applied to the removal of organic pollutants in water, which has aroused extensive research interest [13,14]. Unfortunately, g-C₃N₄ has low redox potential, and its photogenerated electron-hole pairs are easy to recombine [15]. These result in the limitation of its application as a self-sufficient semiconductor for the removal of contaminations by photodegradation [16]. Therefore, various methods have been evolved to enhance the photocatalytic activity of pure g-C₃N₄, including metal deposition [17,18], nonmetal doping [19], coupling with other materials [20], and using nano-sized structures [21]. By coupling g-C₃N₄ with other semiconductors to form a heterojunction structure, the shortcomings of high recombination rate of photogenerated electron-hole pairs of a single photocatalyst could be solved [22]. It should be noted that Z-scheme heterostructure formed by combining g-C₃N₄ with other semiconductors can efficiently separate the photogenerated electrons and holes, thereby improving the photocatalytic activity of g-C₃N₄ under visible light [23]. Yu et al., proposed a direct g-C₃N₄-TiO₂ Z-scheme photocatalyst, which increased the photocatalytic activity by 2.1 times compared to pure TiO₂ [24]. Hong et al., reported that the photocatalytic efficiency of a Z-scheme V₂O₅/ g-C₃N₄ heterojunction for the degradation RhB was as high as 7.3 and 13.0 times that of pure g-C₃N₄ and V₂O₅, respectively [25]. MoO₃ is a semiconducting material with wide gap, stable crystal structure, and photochromic sensitive nature [26]. It has been regarded as a promising candidate to form hybrid photocatalyst due to its special energetic and electrical properties [23]. Previous studies confirmed that combining with MoO₃, the photocatalitic activities of many photocatalyst, including TiO₂ [27], CdS [28], and polyimides [29], could be improved greatly. The composites possessed excellent photocatalytic activities by hindering charge recombination and improving charge transfer processes. Recently, researchers found that combining MoO₃ with g-C₃N₄ could produce Z-scheme photocatalyst. The photocatalytic performance was enhanced due to the suitable band gaps between the two semiconductors. Under light illumination, the photogenerated charge carrier can be efficiently separated and thus generated more reactive species [23,30]. However, most photocatalysts with high activity exist as nano-powders [31], and due to the small particle size, they can be easily dispersed in water thus couldn’t be separated effectively [32]. This characteristic makes it practically limited and prone to secondary pollution [33]. To overcome the above problem, some magnetic materials, such as Fe₃O₄ and CoFe₂O₄, have been achieved considerable attention [34,35]. Magnetic materials can transfer their magnetic properties to photacatalyst after being loaded, thus the photocatalyst can be separated effectively and easily from the treated solution using external magnetic field [36].

Spurred on by aforesaid information, after integration and envision, a novel ternary Z-scheme photocatalyst composites was presented combining g-C₃N₄ with MoO₃ and Fe₃O₄. The aim of this study was to develop an efficient photocatalyst by combining the interfacial connection of g-C₃N₄ and MoO₃ as well as the easy separation of magnetic materials. Their physical and chemical properties were investigated via a series of characterization. The TC-degrading ability of the prepared composites was studied. The influences of MoO₃ content on the photocatalytic performance of the composite were evaluated. The possible mechanisms for the photocatalytic activity enhancement and the TC degradation were presented.

1. Experimental

1.1. Material preparation

The g-C₃N₄ was prepared by direct heating of melamine to 520 ^oC for 3 h in a muffle furnace, and the resultant samples was milled into powder for further use.

The Fe₃O₄/g-C₃N₄ was prepared by the following steps: g-C₃N₄ was dispersed in ethanol/water (1:2) solution and then treated with an ultrasonic cleaner at 300 W for 6h to form an uniform solution with 62.5 mg/L g-C₃N₄. 20 ml of 175 mg/L FeCl₃ and 20 ml of 68 mg/L FeCl₂ added into 500 mL of the suspension of g-C₃N₄. The mixture was stirred and dispersed at 80 ^oC for 30 min prior to the quick injection of 10 mL of ammonia solution. The resultant mixture was stirred at 80 ^oC for another 30 min. The as-obtained precipitate was washed several times with ultrapure water and absolute alcohol before being dried in air at 80 ^oC for further use. The resultant sample was named Fe₃O₄/g-C₃N₄.

AHM (Ammonium heptamolybdate tetrahydrate) was added into ultrapure water with a little acetic acid. The resultant solution was adjusted to pH 3.5 with 36% acetic acid and stored at 80 ^oC for 12 h to obtain amount of white precipitation. The precipitation was washed by absolute ethanol for 5 times and consequently dried in air at 60 ^oC for 12 h (designed as secondary ammonium molybdate). After the obtained sample was ground for 30 minutes, it was sintered at 500 ^oC for 2 hours under the protection of nitrogen. The resultant sample was named MoO₃.

The Fe₃O₄/g-C₃N₄/MoO₃ nanocomposites were synthesized by calcination method. Secondary ammonium molybdate and Fe₃O₄/g-C₃N₄ were taken separately in mortars, grounded for 30 mins. Then the two samples were mixed and thoroughly grounded for another 30 mins before being sintered at 500 ^oC for 2 h under nitrogen atmosphere. After being cooled, the product was obtained. Following the same synthesis route different weight percentage of Fe₃O₄/g-C₃N₄/MoO₃ nanocomposites were obtained varying the wt% of secondary ammonium molybdate maintaining wt ratio 10, 20, 30 and 40 wt%. All the Fe₃O₄/g-C₃N₄/MoO₃ composites were denoted as Fe₃O₄/g-C₃N₄/MoO₃(10%), Fe₃O₄/g-C₃N₄/MoO₃(20%), Fe₃O₄/g-C₃N₄/MoO₃(30%), and Fe₃O₄/g-C₃N₄/MoO₃(40%).

1.2. Characterization

The XRD patterns were obtained by a Bruker D8 Advance X-ray diffractometer with CuKα radiation, employing scanning rate of 0.02^o/sec in the 2θ range from 5^o to 90^o. Surface morphology was studied by JSM-7500F SEM, using an accelerating voltage of 5 kV. The purity and elemental analysis of the products were obtained by EDX on JSM-7500F SEM. The microstructures were investigated by a JEM-2100F TEM with an acceleration voltage of 200 kV. HRTEM was conducted on a JEM-2100F. The UV-Vis DRS was performed by an UV270 spectrophotometer, utilizing BaSO₄ was the reflectance. The FT-IR spectra were studied by a Nicolet-iS10 instruction. XPS data was obtained by an Escalab 250Xi apparatus. The surface area and pore properties were estimated by BET and BJH models using the adsorption data collected by Micro for TriStar II Plus 2.02 apparatus at -196 ^oC. Thermo-gravimetric analysis (TGA) was carried out on a STA 449F3 thermal analyzer with a heating rate of 10 ^oC/min from room temperature to 1000 ^oC in an air flow. The photoluminescence (PL) spectra were obtained by a Fls980 fluorescence spectrophotometer with an excitation wavelength of 380 nm. Magnetic properties were investigated using a MPMS.

1.3. Photocatalytic activity measurement

The capacity of the synthesized catalysts to photodegrade TC was performed by a photochemistry reaction instrument (YM-GHX-V, Shanghai Yuming Instrument Co. Ltd, China) with a 1000 W Xe lamp applied as visible light source, as shown in S1 Fig. In the reaction system, the reaction solution is packed in a quartz tube with a capacity of 50 ml, and the quartz tube is fixed at a distance of 2 cm from the light source. An optical power meter (OPT-1A, China) was used to measure the intensity of the experimental lamp to be 37.5 mW/cm² (λ >400 nm). A water circulation system was utilized to keep the reaction system at 15 ^oC. In each experiment, 10 mg of the photocatalyst was added into 50 mL of TC solution (40 mg/L). Prior to illumination, the reaction solution was treated in dark for 30 min to achieve adsorption-desorption equilibium. Every 30 minutes, 0.5 mL was sampled from the reaction solution and centrifuged immediately at 5000 rpm for 7 min. The TC concentration was determined based on absorbance at 355 nm by Nano Drop 2000 spectrophotometer.

1.4. Active species trapping measurement

Radical scavenge experiments was performed to verify the role of active substances in the degradation of TC. Ethylenediaminetetraacetate (EDTA-2Na, 1 mM), potassium dichromate (K₂Cr₂O₇, 50 μM), isopropanol (IPA, 10 mM), and benzoquinone (BQ, 1 mM), were respectively applied as the trapping agent of h⁺, e⁻, ·OH, and ·O₂⁻ [37,38].

2. Results and discussion

2.1. Photocatalyst characterization

Fig 1 showed the typical XRD patterns of MoO₃, g-C₃N₄, Fe₃O₄, Fe₃O₄/g-C₃N₄ and Fe₃O₄/g-C₃N₄/MoO₃ composites. It could clearly observed that the (020), (110), (040), (021), (111), (060), and (200) peaks of MoO₃ were at 12.83^o, 23.46^o, 25.76^o, 27.40^o, 33.75^o, 39.07^o, and 46.04^o, which could be exactly indexed as the orthorhombic structure (α-MoO₃) (JCPDF 35–0609) [39]. Previous study reported that MoO₃ had three different crystalline structure, orthorhombic (α-MoO₃), monoclinic (β-MoO₃) and hexagonal (h-MoO₃) and α-MoO₃ was thermodynamically stable [40]. So it concluded that the proposed synthesis process benefit the growth of α-MoO₃ which was more thermodynamically stable than β-MoO₃. The (100) and (002) peaks of g-C₃N₄ appeared at 13.12^o and 27.52^o, which were in consistent with the characteristic interplanar staking peaks of the inter-layer structural packing and aromatic systems, respectively [41]. The main peaks of Fe₃O₄ appeared at 35.83^o, 43.18^o, 53.17^o, 57.43^o and 63.04^o, well presented to the lattice plane (311), (400), (422), (511) and (440), respectively [42]. The Fe₃O₄/g-C₃N₄ nanocomposites had the peaks corresponding Fe₃O₄ and g-C₃N₄, indicting Fe₃O₄ were successfully deposited on g-C₃N₄ surface. The patterns for Fe₃O₄/g-C₃N₄/MoO₃ nanocomposites were composed of the diffraction peaks corresponding to g-C₃N₄, MoO₃ and Fe₃O₄, confirming the coexistence of the three materials. Moreover, it was clearly that the intensity of the peaks for MoO₃ in Fe₃O₄/g-C₃N₄/MoO₃ nanocomposites increased with the increase of the weight percent of MoO₃. However, the peaks for g-C₃N₄ in the nanocomposites were not obviously observed as it overlapped with the peaks for MoO₃. The inset XRD patters for Fe₃O₄/g-C₃N₄/MoO₃ (30%) displayed the two deconvulation peaks at 27.32, suggesting the presence of both MoO₃ and g-C₃N₄. These results further verified that Fe₃O₄ and g-C₃N₄ combined with MoO₃ successfully.

Fig 1. XRD patterns for the MoO₃, g-C₃N₄, Fe₃O₄, Fe₃O₄/g-C₃N₄ and Fe₃O₄/g-C₃N₄/MoO₃ nanocomposites.

Inset image shows the deconvulation peaks for MoO₃ and g-C₃N₄.

Fig 2 exhibited the elemental mapping of the Fe₃O₄/g-C₃N₄/MoO₃ (30%) nanocomposites which was detected from a randomly selected area of the nanocomposite using EDX detector. It could be clearly found C, N, Fe, O and Mo (Fig 2B–2F) were all homogeneous indicating uniform distributions of Fe₃O₄, g-C₃N₄, and MoO₃ in the selected area of the corresponding SEM image (Fig 2A).

Fig 2.

(a) SEM images of Fe₃O₄/g-C₃N₄/MoO₃ (30%); (b-f) EDX mapping for the Fe₃O₄/g-C₃N₄/MoO₃ (30%) nanocomposite.

Fig 3 presented the morphology and microstructure of the MoO₃, g-C₃N₄, Fe₃O₄/g-C₃N₄, and Fe₃O₄/g-C₃N₄/MoO₃ (30%) samples investigated by TEM. It was obviously that MoO₃ possessed Flake-like structure with the size of about 200 nm (Fig 3(A)). Pure g-C₃N₄ (Fig 3(B)) shows lamellar-like and smooth morphology. In Fe₃O₄/g-C₃N₄ composites (Fig 3(C)), dark Fe₃O₄ nanoparticles with a particle size of 10–20 nm were deposited on the surface. For Fe₃O₄/g-C₃N₄/MoO₃ (30%) composites (Fig 3(D)), the composites of Fe₃O₄/g-C₃N₄ were well adhered on the surface of MoO₃. These results demonstrated the successful synthesis of the ternary Fe₃O₄/g-C₃N₄/MoO₃. To further verify the formation of Fe₃O₄/g-C₃N₄/MoO₃ ternary structure, HRTEM image it has been used to investigate the microstructure of 30% Fe₃O₄/g-C₃N₄/MoO₃ (Fig 3(E)). The HRTEM image illustrated that the heterostructure of Fe₃O₄/ g-C₃N₄/ MoO₃ composite material showed lattice fringes of 0.38 nm corresponded to the (110) plane of MoO₃, the fringes of 0.29 nm assigned to the (220) plane of Fe₃O₄. The interaction between MoO₃ and g-C₃N₄ could benefit a continuous flow of electrons between them due to the improvement of electron channelization through the interface [43], resulting in the improvement of photocatalytic efficiency.

Fig 3.

TEM of (a) g-C₃N₄; (b) Fe₃O_4; (c) Fe₃O₄/g-C₃N₄; (d) MoO₃; (e) Fe₃O₄/g-C₃N₄/MoO₃ (30%);(f) HRTEM images of the Fe₃O₄/g-C₃N₄/MoO₃ (30%) nanocomposite.

X-ray photoelectron spectroscopy (XPS) was carried out to further analyze the surface compositions and chemical states of Fe₃O₄/g-C₃N₄/MoO₃ (30%) sample (Fig 4). Fig 4(A) revealed the presence of Mo, C, N, O and Fe elements on the surface of the as-prepared sample. The photoelectron lines at binding energy of 233, 285, 363, 399, 531 and 712 eV were correspond to Mo 3d, C 1s, N 1s, O 1s and Fe 2p in the sample, respectively [44]. Fig 4(B) represented the XPS spectrum of Fe. The two Fe 2p_3/2 and 2p_1/2 peaks corresponding to binding energy 710.6 and 723.7 eV without shakeup satellite peak of Fe₂O₃, and their binding energy was consistent with that in pure Fe₃O₄ [45], suggesting the coexistence of dual iron oxidation states of Fe²⁺ and Fe³⁺ [46]. The binding energy spectrum of Mo was demonstrated in Fig 4(C), there were only two peaks existed at 232.1 and 235.3 eV corresponding to 3d_5/2 and 3d_3/2 of Mo atom in +6 oxidation states [47]. The C 1s signal could be divided into four peaks at 284.2, 285.8, 288.1, and 289.5 eV, implying the presence of chemically different carbon species in the sample (Fig 4(D)). The peaks located at 284.2 and 285.8 were attributed to C = C and C-O bonds, respectively [48]. The peak located at 288.1 eV was attributed to sp² hybridised C atoms in the triazine rings inside thearomatic structure, while the peak at 289.5 eV was corresponded to N = C-N group or -NH₂ group as originating from g-C₃N₄. The XPS peak of N 1s (Fig 4(E)) obviously centered at the binding energy of 398.0 eV, which could be assigned to the sp² hybridized nitrogen (C = N-C) whereas peak at and 401.2 eV represented the tertiary nitrogen (N-C₃). Based on Fig 4(F), there were two types of oxygen species, which should assign to the O 1s peak. The offering of the anionic oxygen in Fe₃O₄ centered at about 530.1 eV, and the oxygen in MoO₃ centered at 531.7 eV [49]. The XPS results strongly suggested the coexistence of Fe₃O₄, g-C₃N₄, and MoO₃.

Fig 4. XPS patterns of Fe₃O₄/g-C₃N₄/MoO₃ (30%) nanocomposite.

(a) Survey spectra_; (b)Fe 2p; (c) Mo 3d; (d) C 1s; (e) N 1s; (f) O 1s.

Chemical structures of the MoO₃, g-C₃N₄, Fe₃O₄, Fe₃O₄/g-C₃N₄ and various Fe₃O₄/g-C₃N₄/MoO₃ nanocomposites were studied by FT-IR spectra, and the results were exhibited in Fig 5. For pure g-C₃N₄, the absorption signal of 3165 cm⁻¹ was beneficial from the stretching vibrations of N-H. The strong absorption band in the range of 1240–1650 cm⁻¹ is correspond to typical skeletal stretching vibrations of C-N and C = N in s-triazine or tri-striazine [50]. Simultaneously, the band at 809 cm⁻¹ can be assigned to the typical breathing mode of the heptazine arrangement [51]. In case of pure Fe₃O₄ nanoparticles, two peaks at 566 and 421 cm⁻¹ were corresponded to the stretching vibrations of Fe-O [52]. Neat MoO₃ showed signals of 599 cm⁻¹ and 852 cm⁻¹ which were related to the stretching vibrational modes of O shared by three Mo and the Mo-O-Mo unit respectively in the crystalline α-MoO₃. In addition, a signal at 991 cm⁻¹ was due to Mo = O for the crystalline α-MoO₃ [53]. In the Fe₃O₄/g-C₃N₄/MoO₃ nanocomposites, the existence of the typical vibrational modes of g-C₃N₄, Fe₃O₄, and MoO₃ indicated the coexistence of these three contents in the nanocomposites.

Fig 5. FT-IR spectra of MoO₃, g-C₃N₄, Fe₃O₄, Fe₃O₄/g-C₃N₄ and Fe₃O₄/g-C₃N₄/MoO₃ nanocomposites.

S2 Fig displays TGA curves for the g-C₃N₄ and Fe₃O₄/g-C₃N₄/MoO₃ (30%) samples. As can be seen, the pristine g-C₃N₄ shows a weight loss of about 96% after heating up to 750 ^oC. Hence, it was concluded that the g-C₃N₄ decomposes almost completely heating up to 750 ^oC. It is evident that the thermal behavior of Fe₃O₄/g-C₃N₄ and Fe₃O₄/g-C₃N₄/MoO₃ (30%) samples are similar to that of g-C₃N₄. As can be seen, by loading Fe₃O₄ and MoO₃ on the g-C₃N₄ sheets, thermal degradation of the nano-composites starts from lower temperature relative to the pristine g-C₃N₄. Hence, similar to many g-C₃N₄-based nanocomposites, thermal stability of the pristine g-C₃N₄ decreases with depositing different particles [45,46]. The g-C₃N₄ contents of Fe₃O₄/g-C₃N₄ and Fe₃O₄/g-C₃N₄/MoO₃ (30%) nanocomposites were calculated from the weights remaining after heating the samples to over 650 ^oC. The g-C₃N₄ contents of the Fe₃O₄/g-C₃N₄/MoO₃ (30%) nanocomposite was about 8.2%, respectively. As can be seen, besides the weight loss of g-C₃N₄, another weight loss between 750 and 1000 ^oC in the Fe₃O₄/g-C₃N₄/MoO₃ (30%) composites, could be ascribed to the vaporization of MoO₃. The MoO₃ contents of the Fe₃O₄/g-C₃N₄/MoO₃ (30%) is about 16.4%. In addition, after calculation, The MoO₃ contents of the Fe₃O₄/g-C₃N₄/MoO₃ (30%) is about 75.4%. The results were listed in Table 1.

Table 1. Weight percentages of different compounds in the Fe₃O₄/g-C₃N₄/MoO₃ 30%) nanocomposite.

Compound	Weight percentage
g-C3N4	8.2
MoO3	16.4
Fe3O4	75.4

It was well known that the photoabsorptive capacity of a photocatalyst would greatly affect its photocatalytic activity [54]. Thus, UV-Vis DRS was used to investigate the photoabsorption ability of a series of as-prepared samples and the results were showed in Fig 6. As could be seen in Fig 6(A), both pristine g-C₃N₄ and MoO₃ possessed small absorption in visible region and had absorption edges at about 470 nm, which were compatible with the reported absorption edges for g-C₃N₄ and MoO₃ [55]. Fig 6(B) displayed the band gaps of g-C₃N₄ and MoO₃ were consistent with previous studies, which were 2.72 eV and 2.85 eV, respectively [20]. The band gap of all as-prepared photocatalysts were obtained by using Taucʼs equation (Eq 1).

$$\alpha h\nu =A{(hv-Eg)}^{n/2}$$ (1)

where, α, h, ν, and A were absorption coefficient, Planck's constant (eV. s), the light frequency (s⁻¹), and proportionality constant, respectively. Eg was the band gap, and n was the power which was assumed to be 1 and 4 for direct and indirect transitions, respectively [56,57]. As displayed in the figure, the addition of Fe₃O₄ to the pure g-C₃N₄ greatly enhanced the absorption in the visible range. Interesting, the addition of MoO₃ to the Fe₃O₄/g-C₃N₄ slightly decreased the visible light absorption when the weight percentages of MoO₃ were lower than 30%. The absorption would be significantly reduced when the content of MoO₃ was over the value. However, compared to pristine g-C₃N₄ and MoO₃, the visible light absorption of Fe₃O₄/g-C₃N₄/MoO₃ nanocomposites was considerably high. These facts possibly make Fe₃O₄/g-C₃N₄/MoO₃ to use more visible light, and produce more photoexcited charge carriers than pure g-C₃N₄ or MoO₃.

Fig 6.

(a) UV-Vis diffuse reflectance absorption spectra for the MoO₃, g-C₃N₄, Fe₃O₄, Fe₃O₄/g-C₃N₄ and Fe₃O₄/g-C₃N₄/MoO₃ nanocomposites_; (b) The corresponding calculated band gaps of g-C₃N₄ and MoO₃.

To analyze textural properties of the prepared MoO₃, g-C₃N₄, Fe₃O₄/g-C₃N₄, and Fe₃O₄/g-C₃N₄/MoO₃ (30%) photocatalysts, the results about N₂ adsorption-desorption isotherm were provided in Fig 7. As could be seen, the isotherm of each sample was of typical IV with H₃ hysteresis, indicating a characteristic of mesopores structure [58], which benefited to decreasing mass transfer limitations and harvesting light in the photocatalytic process [59]. BET and BJH models were used to investigate the specific surface areas and pore features of the four photocatalysts, respectively and the results were presented in Table 2. The surface areas of the MoO₃, g-C₃N₄, Fe₃O₄/g-C₃N₄, and Fe₃O₄/g-C₃N₄/MoO₃ (30%) were 73.1, 12.6, 97.4, and 72.7 m²g⁻¹, respectively. Compared to single-phase g-C₃N₄, Fe₃O₄/g-C₃N₄ had larger surface area, which might attribute to the formation of hierarchical structure after loading Fe₃O₄ on g-C₃N₄ [60]. However, after the Fe₃O₄/g-C₃N₄ being modifying with MoO₃, the surface area was decreased. This decrease probably caused by the covering of the Fe₃O₄/g-C₃N₄ surface by MoO₃, resulting in the blocking of some active sites on the surface [61]. Generally, a decreased in the specific surface area of a semiconductor was accompanied by a decrease in its photocatalytic activity. Hence, the highly improved photocatalytic activity of Fe₃O₄/g-C₃N₄/MoO₃ (30%) should be not described to its textural properties.

Fig 7. Nitrogen adsorption-desorption isotherms of MoO₃, g-C₃N₄, Fe₃O₄, Fe₃O₄/g-C₃N₄ and Fe₃O₄/g-C₃N₄/MoO₃ (30%) nanocomposite.

Table 2. The textural properties of g-C₃N₄, Fe₃O₄/g-C₃N₄, Fe₃O₄/g-C₃N₄/MoO₃ (30%) samples.

Photocatalyst	Surface area (m2 g-1)	Mean pore diameter (nm)	Total pore volume (cm3 g-1)
MoO3	73.0615	14.63057	0.267233
g-C3N4	12.6271	13.03849	0.041159
Fe3O4/g-C3N4	97.4179	11.24809	0.273941
Fe3O4/g-C3N4/MoO3 (30%)	72.6855	13.84626	0.251606

Fig 8 displayed the VSM curves for the Fe₃O₄ nanoparticles and Fe₃O₄/g-C₃N₄/MoO₃ (30%) photocatalyst at ambient temperature. Saturation magnetization of the Fe₃O₄ nanoparticles was 52.5 emu/g, while that of the Fe₃O₄/g-C₃N₄/MoO₃ (30%) nanocomposites decreased to 26.3 emu/g due to the existence of the non-magnetic g-C₃N₄ and MoO₃. However, both of the samples displayed super paramagnetic behavior. By pacing an external magnet beside the glass bottle containing the Fe₃O₄/g-C₃N₄/MoO₃ (30%) nanocomposite, the particles were rapidly attracted to the wall of the glass bottle, as shown in the top-left inset of Fig 8, suggesting an easy separation under external magnetic field.

Fig 8. Magnetization curves for the Fe₃O₄ nanoparticles and Fe₃O₄/g-C₃N₄/MoO₃ (30%) nanocomposite.

Inset of the figure shows separation of the nanocomposite from the treated solution using a magnet.

2.2. Photocatalytic activity and mechanism

Degradation of TC solution under visible light to evaluate the photocatalytic activity of the as-prepared catalyst, and the results were demonstrated in Fig 9. As shown in Fig 9(A), the blank experiments (in absence of any photocatalyst) revealed that the changes of TC concentration were negligible, that mean TC was quite stable under light irradiation, thus the self-degradation of TC was ruled out. The removal percentage of TC was denoted as C/C₀, in which C was the TC concentration after adsorption and light illumination for a certain time, and C₀ was the initial concentration of TC. For pristine MoO₃, there were 20% TC were adsorbed and only about 17% TC were photodegraded in 120 min. Single-phase g-C₃N₄ displayed almost no adsorption and moderate photocatalytic activity for TC, with a removal percentage of 28% after 120 min under visible light. It should be noted that, when the Fe₃O₄ loaded on g-C₃N₄, the photodegrading ability decreased, with a removal percentage of 10%, implying Fe₃O₄ had a negative effect on the photocatalytic activities. The incorporation of the MoO₃ boosted the overall activity, and the Fe₃O₄/g-C₃N₄/MoO₃ composites displayed remarkable enhancements in the photodegrading-abilities. After irradiated for 120 min, the removal percentages were about 48, 77, 94, and 44% for 10, 20, 30, and 40% Fe₃O₄/g-C₃N₄/MoO₃ nanocomposites, in which 18, 37, 46, and 9% were attributed to photodegradation, respectively. Obiviously, the nanocomposites with 30% MoO₃ possessed the best photocatalytic activity. Since the Fe₃O₄ had no positive effect on the photodegrading-ability, the improvements in the photacatalytic performance of the nanocomposites should attribute to the cooperation of g-C₃N₄ and MoO₃. Furthermore, when the weight percent of MoO₃ was over 30%, the degradation of TC decreased. That was to say, excess load of MoO₃ leaded to the lower photodegradating-ability, which implied that the superfluous MoO₃ could impede the interaction of g-C₃N₄ and MoO₃. The pseudo-first-order kinetic model (ln[TC] = ln[TC]₀−k_obst) was used to fit with the degradation process to quantify the activities of the resultant samples, in which the value of the observed first-order rate constant (k_obs) was equal to the corresponding slope of the straight line [62]. As shown in Fig 9(B), The k of MoO₃, g-C₃N₄, Fe₃O₄/g-C₃N₄, and Fe₃O₄/g-C₃N₄/MoO₃ (30%) nanocompositeis were 2.36×10⁻³, 3.25×10⁻³, 8.2×10⁻⁴, and 1.63×10⁻² min⁻¹, respectively. Thus, it could be concluded that activity of the Fe₃O₄/g-C₃N₄/MoO₃ (30%) nanocompositeis was about 6.9, 5 and 19.9-fold higher than those of MoO₃, g-C₃N₄, and Fe₃O₄/g-C₃N₄ composites, respectively. Fig 9(C)–9(F) displayed the UV-Vis spectral variation of TC solution during the adsorption and photodegradation over the MoO₃, g-C₃N₄, Fe₃O₄/g-C₃N₄, and Fe₃O₄/g-C₃N₄/MoO₃ (30%) nanocompositeis. For all the samples, the maximal absorbance at 355 nm decreased as the reaction progressed, suggesting gradual removal of TC. Comparison of Fe₃O₄/g-C₃N₄/MoO₃ (30%) with other similar reported systems of Fe₃O₄/g-C₃N₄ composites has been discussed in S1 Table.

Fig 9.

(a) Comparison of the photocatalytic activities of different samples. (A) MoO₃; (B) g-C₃N₄; (C) Fe₃O₄/g-C₃N₄; (D) Fe₃O₄/g-C₃N₄/MoO₃ (10%); (E) Fe₃O₄/g-C₃N₄/MoO₃ (20%); (F) Fe₃O₄/g-C₃N₄/MoO₃ (30%); (G)Fe₃O₄/g-C₃N₄/MoO₃ (40%) (H)No photocatalyst; (b) Pseudo-first-order kinetic curves of the corresponding samples; (c), (d), (e),and (f) Temporal evolutions of the spectra during the photocatalytic degradation of TC over MoO₃, g-C₃N₄, Fe₃O₄/g-C₃N₄,and Fe₃O₄/g-C₃N₄/MoO₃ (30%).

In generally, for most semiconductors the photo-induced e⁻/h⁺ pairs can recombine after being irradiated by light thus emit fluorescence, which can be indicated by PL. Higher PL intensity of a semiconductor indicates a higher recombination rate of its e⁻/h⁺ pair [63]. Fig 10, showed the PL spectroscopy of g-C₃N₄, Fe₃O₄/g-C₃N₄, and Fe₃O₄/g-C₃N₄/MoO₃ series samples. As seen in the figure, g-C₃N₄ displayed large PL signal due to the high recombination of photo-induced e⁻/h⁺ pairs and low quantum yield [16]. However, Fe₃O₄/g-C₃N₄ nanocomposites exhibited a stronger PL than that of the pure g-C₃N₄, indicating a lower separation rate of photo-induced e⁻/h⁺ pairs. Interestingly, the addition of MoO₃ to the Fe₃O₄/g-C₃N₄ nanocomposites followed by the formation of the Fe₃O₄/g-C₃N₄/MoO₃ obviously reduced the PL emission intensity due to the combination of MoO₃ and Fe₃O₄/g-C₃N₄, which suggested the fabrication of the nanocomposites efficiently enhanced the of separation of e⁻/h⁺ pairs on the surface. It should be noted that the PL signal increased significantly when the content of MoO₃ were over 30%, implying an easier recombination of photogenerated charge carriers. The incensement may attribute to the agglomeration of the overloaded MoO₃ on the surface of the nanocomposites, resulting in the reduction of the interface area between g-C₃N₄ and MoO₃.

Fig 10. PL spectra of g-C₃N₄, Fe₃O₄, Fe₃O₄/g-C₃N₄ and Fe₃O₄/g-C₃N₄/MoO₃ nanocomposites.

It had been reported that the ·O₂⁻, ·OH, e⁻ and h⁺ were the main active species attributed to the photodegradation contaminants during the photocatalytic reactions [64]. However, their contribution to the degradation of contaminants was not identical and could be investigated by utilizing the quenching experiments. In order to estimate the role of each radical in the TC photodegradation, EDTA-2Na, K₂Cr₂O₇, IPA, and BQ were respectively used as the quenchers for h⁺, e⁻, ·OH, and ·O₂⁻ in the TC degradation process in the Fe₃O₄/g-C₃N₄/MoO₃ (30%) system. As shown in Fig 11, the degradation percentage of TC after 120 min irradiation was 90% with free quencher and drastically decreased to about 6% when BQ was added into the system. In the same time, the addition of EDTA-2Na, K₂Cr₂O₇ and IPA resulted in 21, 39 and 83% photodegadation percentages of TC. These results indicated that ·O₂⁻ played vital role for the TC photodegradation, h⁺ and e⁻ had modest contribution to the TC decomposition, while ·OH was had the lowest contribution to the TC degradation. It should be noted that some other intermediates might be produced during the photodegradation reaction, which might take part in the degradation of TC.

Fig 11. Results of active species trapping experiments.

In composites of two semiconductors, the effective separation of e⁻/h⁺ pairs depends on the appropriate band-gap positions of them. The band positions of g-C₃N₄ and MoO₃ could be obtained using empirical equations (Eqs 2 and 3) [23]:

$$E_{CB}=X-E_C-\frac{1}{2}{E}_g$$ (2)

$$E_{VB}=E_{CB}+E_g$$ (3)

Where X is the absolute electronegativity of the atom semiconductor used to represent the geometric mean of the absolute electro-negativity of the constituent atoms, which is defined as the arithmetic mean of the atomic electron affinity and the first ionization energy; E_CB is the energy of free electrons of the hydrogenscale (4.5 eV); E_g is the band gap of the semiconductor; E_CB is the conduction band potential and E_VB is the valence band potential. According to previous studies, the absolute electronegativity X for g-C₃N₄ and MoO₃ were 4.73 eV and 6.40 eV [65,66], respectively. From the Taucʼs equation, E_g of g-C₃N₄ and MoO₃ were to be 2.72 eV and 2.85 eV, respectively.

Based on the above analysis, the conduction bands (CB) of g-C₃N₄ and MoO₃ respectively were -1.13 and 0.47. Accordingly, the valance bands (VB) of them were 1.59 and 3.33, respectively. The results were similar to other studies [23]. Based on the results obtained by PL experiments, for the Fe₃O₄/g-C₃N₄/MoO₃ (30%) photocatalyst, the photogenerated e⁻ and h⁺ could be effectively separated under visible light. According to the traditional mechanism, the e⁻ in the CB of g-C₃N₄ could transfer to the CB of MoO₃ while the h⁺ could migrate in the opposite direction. Generally, the reduction of O₂ with photoelectrons produced ·O₂⁻ (e⁻+ O₂→ ·O₂⁻, O₂/·O₂⁻ = -0.33 V vs. NHE) [67]. The ·OH⁻ could be obtained by photoholes oxidized H₂O directly (h⁺ + H₂O → ·OH + H⁺, ·OH/OH⁻ = 2.40 V vs. NHE) (Michael R. Hoffmann, 1995; Wen et al., 2017) or indirectly through ·O₂⁻ (·O₂⁻+ H₂O → H₂O₂ →·OH) [68]. In summary, the VB of MoO₃ and g-C₃N₄ are excited by visible light at the same time, and then the photoelectrons in the CB of MoO₃ and the holes in the solid-solid contact interface of the VB of g-C₃N₄ recombine, resulting in photoelectron retention In the CB of g-C₃N₄, holes are left in the VB of MoO₃. Therefore, g-C₃N₄ and MoO₃ could form Z-scheme and enhanced the separation of photogenetrated e⁻/h⁺ pairs at the interface of Fe₃O₄/g-C₃N₄/MoO₃ [64]. As shown in Fig 12, under light irradiation, the e⁻ in the CB of g-C₃N₄ had relative stability thus benefited to the continuous generation of O₂⁻ from O₂. The h⁺ in the VB of MoO₃ generating ·OH⁻ by oxidized H₂O. Some of h⁺ in the VB of MoO₃ took part in the oxidation of TC, while the rest h⁺ were reduce H₂O to^.OH, which was not the main reactive species for TC degradation.

Fig 12. Proposed mechanism for enhanced photocatalytic activity over Fe₃O₄/g-C₃N₄/MoO₃ nanocomposites.

4. Conclusions

In this study, a novel and easily separated ternary Fe₃O₄/g-C₃N₄/MoO₃ (30%) photocatalyst was presented using melamine, FeCl₃, FeCl₂, and AHM as materials. This catalyst provided enhanced photocatalytic activity toward the removal of TC in aqueous environment. The photocatalytic activity of the novel catalyst was approximately 6.9 times of MoO₃, 5 times of g-C₃N₄, and 19.9 times of Fe₃O₄/g-C₃N₄ on photodegradation of TC. The excellent photodegrading ability was due to the formation of Z-scheme structure between C₃N₄ and MoO₃, which could effectively separate the photogenerated e⁻/h⁺ pairs and efficiently utilize the e⁻ and h⁺. The highly improved TC-photodegrading ability was also beneficial from the wide range light absorption. This work indicated that the novel Fe₃O₄/g-C₃N₄/MoO₃ was beneficial in decreasing TC and other environmental pollutants with high-level concentration in water, and paved a new way to the development of photocatalytic technology.

Supporting information

S1 Fig. Schematic diagram of photocatalytic reaction device.

(DOC)

S2 Fig. TGA curves of pure g-C₃N₄ and Fe₃O₄/g-C₃N₄/MoO₃ (30%) photocatalysts.

(DOC)

S1 Table. Comparison of degradation performance of similar photocatalysts.

(DOC)

Data Availability

All the data supporting the ideas in this article are in the picture and table files.

Funding Statement

This work is supported in part by the Science and Technology Program of Changsha,China (kq1907097) and Central South University of Forestry and Technology Graduate Technology Innovation Fund (CX20190620).