Characterization of BMS@TiO2
Surface morphology and chemical composition of BMS@TiO2
The morphology of BMS and BMS@TiO2 were characterized by SEM and TEM shown in Figs. 2(a) ~ 2(d). Both BMS and BMS@TiO2 particles are mainly composed of irregular-crisscross whiskers, and their morphology is consistent with that of the 5·1·7 phase (5Mg(OH)2·MgSO4·7H2O) formed in the vent of the basic magnesium sulfate cement block [43]. Additionally, TiO2 nanoparticle is combined with 5·1·7 phase, as evidenced by the lattice stripes of 5·1·7 phase and TiO2 corresponding to the (222) and (101) planes respectively shown in Fig.2(d), which also demonstrates that TiO2 can be firmly embedded in BMS substrate by the cement reactions between magnesium oxide and magnesium sulfate at room temperature. Figs. 2(e) ~ 2(h) display the element distribution of BMS and BMS@TiO2, and element composition is listed in Table S2. Obviously, TiO2 is widely distributed in 5·1·7 phase matrix and enriched in some regions probably because of the inevitable agglomeration of TiO2 nanoparticles during the synthesis process of BMS@TiO2. The content of titanium (Ti) in BMS@TiO2 is 1.4% and the mass fraction of TiO2 in BMS@TiO2 is further calculated to be 2.3% approximately.
Physical properties and structural characteristics of BMS@TiO2
Figs. 3(a) and 3(b) show the FTIR spectra of BMS, TiO2, and BMS@TiO2. In the spectrum of TiO2, the vibration band at 3438 cm-1indicates the presence of residual H2O molecules adsorbed on TiO2. Meanwhile, the stretching vibration peak at 1618 cm-1 corresponds to the bending vibration of O-H groups [44]. The strong band at 1012 cm-1 indicates the vibration of the Ti-O-Ti or Ti-O bond [45]. Furthermore, the vibration peak of Ti-O-Ti at 450 ~ 750 cm-1 is the anatase phase of TiO2 [46, 47]. In BMS and BMS@TiO2, the peak at 3700 cm-1 occurs the stretching vibration of OH-, and the broad band peak at 3400 cm-1 is caused by the stretching vibration of crystal water (H-O), while the peak at 1636 cm-1 is due to the bending vibration of crystal water. The absorption band at 1450 cm-1, which also appears in the infrared spectrum of magnesium hydroxide [48], probably corresponds to the asymmetric stretching vibration peak of Mg-OH. Likewise, the peak at 1103 cm-1 corresponds to the O3S-O stretching vibration peak of SO42-. The peak at 617 cm-1 represents the stretching vibration peak of the S-O bond, while the subtle peak at 443 cm-1 corresponds to the stretching vibration peak of MgO-H [49-51], as well as the distinctive stretching vibration peak of Ti-O appears at 530 cm-1 in the BMS@TiO2 spectrum. Furthermore, XRD patterns of BMS, TiO2, and BMS@TiO2 in Fig. 3(c) also indicate that the predominant component both in BMS and BMS@TiO2 is 5·1·7 phase, and the characteristic peaks at 9.44º, 17.80º, 30.83º, 36.15º and 37.34º belong to the distinctive features of 5·1·7 phase [52-54]. Besides, TiO2 in BMS@TiO2 exhibits distinct anatase features with characteristic peaks at 25.33º, 36.95º, 37.88º, 38.59º, 48.07º, 53.88º, 55.15º, 62.71º, as described in previous studies [45, 55].
The hysteresis characteristic and pore properties of BMS and BMS@TiO2 were assessed using the N2 adsorption-desorption method, and the results are depicted in Figs. 3(d) and 3(e). It can be proved that the hysteresis loops of BMS and BMS@TiO2 exhibit typical IV isotherms with H3 hysteresis loops. This phenomenon possibly arises from the formation of slit-like pores among 5·1·7 phase whisker clusters [56]. The pore size distribution curves show that the major pore at 2.5 ~ 3.5 nm accompanied at approximately 20 nm appear in BMS and BMS@TiO2, and the specific surface area of BMS@TiO2 is about 37 m2·g-1 similar to that of BMS presented in Table S3. It is conceivable that the macroscopical gas hole formed during the foaming step by hydrogen peroxide and the microcosmic slit-like pores formed by 5·1·7 whisker clusters endow BMS@TiO2 with a large specific surface area, which is beneficial to the adsorption and photocatalysis processes. The point of zero charges (PZC) of BMS, TiO2, and BMS@TiO2 were measured by comparing the solution initial pH and final pH shown in Fig. 3(f), and the pHPZC values of BMS, TiO2, and BMS@TiO2 are 11.54, 7.05, and 11.54, respectively. This is mainly because Mg-OH in basic magnesium sulfate is alkaline, making the interface negatively charged. Consequently, the physical properties and structural characteristics of basic magnesium sulfate material exhibited negligible alteration following the incorporation of TiO2.
Optical properties of BMS@TiO2
The optical properties of BMS, TiO2, and BMS@TiO2 were evaluated using UV diffuse reflection spectroscopy (UV-DRS) and photoelectric signal detection, and the results are presented in Fig. 4. Compared with TiO2, BMS@TiO2 also exhibits enhanced light absorption responses within 250 ~ 450 nm, and the absorbance exceeds 0.6, which is up to the half of pure TiO2. The band gap between the conduction and valence bands of the material is determined by the Tauc plots, as shown in formula (6) [57].
Where α is the absorbance value (a.u.), h is Planck's constant, v is the optical frequency, Eg is the band gap energy (eV), and A is the constant. The value of n depends on the type of semiconductor material. TiO2 (anatase) is an indirect transition semiconductor material, so the n value is 1/2.
Fig. 4(b) illustrates that BMS@TiO2 has a band gap energy at about 3.05 eV, similar to TiO2, and can be excited by ultraviolet light. Meanwhile, the steady-state surface photovoltage and transient photocurrent density are used to study the photoelectric conversion efficiency, that is, the separation and transfer efficiency of photogenerated charge and hole. In general, the higher the photovoltage and photocurrent, the stronger the photogenerated carrier transfer ability [58]. Fig. 4(c) and 4(d) show that all BMS, TiO2, and BMS@TiO2 exhibits a certain photoelectric conversion ability, with the order of strength being TiO2>BMS@TiO2 >BMS. As evidenced by Fig. 4, adding a small amount of TiO2 (2.3%) significantly improved the photoelectric properties of basic magnesium sulfate, and thus BMS@TiO2 has an appropriate photocatalytic ability.
Adsorption study
Fig. 5(a) shows the kinetics of DMP adsorption by BMS@TiO2. The adsorption equilibrium is reached after 6 hours, and the adsorption capacity is maintained at approximately 5.32 mg·g-1 in 1 mol·L-1 NaCl solution with an initial DMP concentration of 30 mg·L-1. The model fitting parameters are shown in Table S4. The R2 values for the pseudo-first-order and pseudo-second-order fitting were 0.9349 and 0.9907 with the theoretic maximum adsorption capacity to 5.51 and 5.49 mg·g-1, respectively, suggesting the adsorption kinetics of BMS@TiO2 for DMP can be accurately described by the pseudo-second-order model. Fig. 5(b) depicts the adsorption isotherm of DMP by BMS@TiO2, and the fitting parameters for Langmuir and Freundlich adsorption isotherm models are shown in Table S5. The results show that the adsorption data of BMS@TiO2 are in good accord with the Langmuir isothermal model. Furthermore, the Langmuir equilibrium constants in milligrams are used in the Van't Hoff equation, and the values of ΔG, ΔHo, and ΔSo are obtained from Fig. 5(c) and summarized in Table S6. The negative ΔG values for different temperatures and the positive ΔHo value indicates that the adsorption process is endothermic and spontaneous.
Furthermore, the influence of different co-existing salts and ionic strengths on the adsorption capacity are illustrated in Fig. 5(d). The general trend is that the adsorption capacity decreases with the increase of ionic strength. With the same ionic strength, the adsorption capacity of potassium chloride and sodium chloride changes to a similar extent. According to XRD patterns in Fig. S2, the phase composition of BMS@TiO2 remains unchanged in both pure water and brine solution, exhibiting certain resistance to water and salt [59, 60]. There seems to because co-existing salts may affect the adsorption process, which in turn leads to a decrease in adsorption capacity.
During the adsorption process, the adsorption solution pH value is approximately 6, so DMP is protonated in acidic condition shown as the chemical reaction (R1) [61] and becomes positively charged, while BMS@TiO2 exhibits a negative surface charge known from the pHPZC value of 11.54. Further taking into account the structural characteristics of BMS@TiO2, the adsorption mechanism of DMP onto BMS@TiO2 can be inferred as electrostatic interaction and pore interception.
(R1)
Consequently, the electrostatic interaction between DMP and BMS@TiO2 is weakened due to the cationic shielding effect when the concentration of co-existing salts increase, resulting in the decrease of adsorption capacity [62]. Especially, the charge of Mg2+ is higher than that of Na+ and K+, resulting in a stronger charge shielding effect. On the other hand, cationic ions compete with DMP for water molecules, thereby reducing the affinity of DMP and water molecules, resulting in DMP easier to agglomerate and ultimately decreasing the critical micelle concentration (CMC) value of DMP. Therefore, the agglomerated DMP is not easily absorbed into BMS@TiO2 by pore interception. Additionally, Mg2+ has a stronger hydration capacity compare to Na+ and K+, resulting in a more pronounced salting out effect, and the CMC value of DMP decreases more significantly when adding Mg2+ [63, 64].
Fig. S3 shows SEM images, XRD patterns, FTIR spectra, and XPS full-scan spectra for BMS@TiO2 before and after adsorption. The results indicate that in BMS@TiO2 after adsorption, the 5·1·7 phase structure is still maintained, and regular whisker shape in SEM images are no longer present, which may be caused by the absorption of DMP onto 5·1·7 phase. Meanwhile, the corresponding FTIR spectra characteristic peaks of 5·1·7 phase do not significantly change before and after absorption, and the peaks at 2927 cm-1 and 2851 cm-1 appeared in the sample after adsorption are respectively assigned to the asymmetric (νa) and symmetric (νs) stretching modes of -CH2- groups in the organic adsorbate [65]. However, no new covalent bond signal is observed in FTIR spectra, indicating that the absorption of DMP by BMS@TiO2 is a physical adsorption process. According to the XPS full-scan spectra of before and after absorption shown in Fig. 3S(d) and Table S7, it can be observed that the binding energy of Mg, O, Ti, and S on the surface of BMS@TiO2 shift to varying degrees after adsorption, and especially the orbital binding energy offset of Mg, O, Ti, and S after adsorption is not enough to the extent of a chemical reaction, further indicating that the physical adsorption occurs on BMS@TiO2 for DMP [66]. Additionally, the binding energy changes of Mg1s and O1s are 0.52 eV and 0.20 eV, respectively and comparatively significant, which indicate that DMP may interacts with MgO6 octahedral skeleton of 5·1·7 phase in BMS@TiO2 [67].
Photocatalytic degradation behavior and kinetic evaluation of DMP
Fig. 6(a) demonstrates the adsorption capacity and photodegradation efficiency of DMP on BMS, TiO2, and BMS@TiO2. Compared with TiO2 nanoparticles, 20-60 mesh BMS@TiO2 particles with 2.3% TiO2 exhibits an equally excellent adsorption and photocatalytic performance. Additionally, the photodegradation efficiency of DMP adsorbed on BMS particles reaches about 12%, which reveals that BMS also has weak photocatalytic activity, and this may be because BMS has weak light absorbance and some degree of photoelectric conversion efficiency shown in Fig. 4. Pseudo-first-order and pseudo-second-order kinetic models are employed to fit the photodegradation kinetics data of DMP on BMS@TiO2 in Fig. 6(b), and the corresponding fitting parameters are listed in Table S8. The results indicate that the photodegradation equilibrium of DMP is reached after 16 hours and the photodegradation efficiency of DMP is approximately 92%. And photocatalytic kinetics of DMP on BMS@TiO2 can be accurately represented by a pseudo-first-order kinetic model. In addition, as depicted in Fig. 6(c), FTIR spectra of BMS@TiO2 before and after photodegradation at different time are obtained to verify the photodegradation kinetics behavior of DMP. The intensity of the peaks at 2927 cm-1 and 2851 cm-1 assigned to the asymmetric (νa) and symmetric (νs) stretching of -CH2- groups of the organic adsorbate gradually decreases with the extension of photodegradation time, adequately manifesting the photodegradation and removal of DMP adsorbed on BMS@TiO2. The total organic carbon (TOC) removal efficiency and GC-MS analysis of adsorbates on BMS@TiO2 before and after photodegradation at different times shown in Fig. S4 are employed to further understand the photodegradation behavior of DMP. As the photodegradation time extends, TOC removal efficiency is up to 52% within 16 hours, distinctly lower than the photodegradation efficiency of DMP. According to the GC-MS analysis results of adsorbates on BMS@TiO2 before and after photodegradation at 12 and 16 hours, there are at least three intermediates in industrial DMP, herein identified as I1, I2, and I3, respectively [68]. Moreover, DMP on BMS@TiO2 thoroughly vanished by ultraviolet radiation at 16 hours, while the contents of these impurities slowly diminish. It is reasonable to conclude that DMP adsorbed on BMS@TiO2 can be degraded and a small portion of impurities can also be adsorbed and photodegraded to varying degrees may because of their different degradability, limiting the apparent TOC removal efficiency.
Stability and reusability of BMS@TiO2
The consecutive adsorption and photodegradation tests for DMP on BMS@TiO2 are conducted, and BMS@TiO2 samples before and after cycle tests are analyzed by XRD, FTIR and N2 adsorption/desorption method. The results obtained in Fig. 7 reveal that BMS@TiO2 maintains a consistent adsorption performance with an average adsorption capacity of 5.33 mg·g-1 after five cycle tests, while the photocatalytic degradation efficiency gradually decreases from 92% at the first cycle to 81% at the fifth cycle. XRD patterns and pore characteristics of BMS@TiO2 before and after cycle tests evidence that all of 5·1·7 phase, TiO2 and the slit-pores at 2.5 ~ 3.5 nm accompanied at approximately 20 nm always exists, demonstrating that electrostatic interaction, pore interception and later photocatalysis of BMS@TiO2 interacted with DMP steadily accomplished during cycle tests. Fig. 7(d) shows the FTIR spectra of BMS@TiO2 before and after cycle tests. The corresponding characteristic peaks of BMS@TiO2 remain unchanged, and peaks at 2927 cm-1 and 2851 cm-1 attributed to the asymmetric (νa) and symmetric (νs) stretching patterns of -CH2- groups appear after cycle tests. This is because, the residual impurities adsorbed on BMS@TiO2 compete to consume active radicals in the next photodegradation, resulting in a decrease in the next apparent photocatalytic degradation efficiency. The degradation efficiency remained at approximately 81% in the third to fifth cycles shown in Fig. 7(a), which may be due to the residual quantity of impurities on BMS@TiO2 reaching adsorption-degradation equilibrium at the cycle experimental conditions.
Photodegradation mechanism of DMP by BMS@TiO2
EPR spectra are conducted using 5,5-dimethyl-1-pyrroline N-oxide (DMPO) as the spin trapping agent in order to track the generation process of active species for degrading DMP adsorbed on BMS@TiO2. The remarkable characteristic peaks in Figs. 8(a) and 8(b) indicate respectively the formation of •OH and •O2- in BMS@TiO2 photocatalytic system, which illustrates that the photodegradation removal of DMP adsorbed on BMS@TiO2 is mainly accomplished by •OH and •O2-. And the signal intensity of all peaks gradually increases with the extension of exposure time, indicating the continuous generation of active species. The charge distribution and reactive sites of DMP are revealed by the electrostatic potential. This potential predicts the nucleophilic and electrophilic regions of the molecule, and negative and positive electrostatic potential regions favoring the occurrence of electrophilic attacks and nucleophilic attacks, respectively [69, 70]. As shown in Fig. 8(d), the electrostatic potential of DMP is visualized as red and blue surfaces surrounding the molecule. The blue color represents negative electrostatic potential values and the red color represents positive electrostatic potential values. The negative electrostatic potential region (blue colored) can be observed for the nitrogen and oxygen atoms located on the morpholine ring, and the positive electrostatic potential region (red colored) is located on the morpholine ring and the hydrogen atoms on C10 and C11 (Fig. 8(c)). Therefore, the nitrogen and oxygen atoms on the morpholine ring are susceptible to react with photogenerated holes and radicals, while the hydrogen atoms on the ring are susceptible to nucleophilic reaction.
Based on the electrostatic potential of DMP and the intermediates, the possible reaction mechanism and degradation pathways of DMP by BMS@TiO2 are proposed in Fig. 8(e). It is speculated that the hydrogen atoms on C17 of DMP are oxidized by the photogenerated radicals through nucleophilic reaction. As a result, the -CH2- group on the ring was oxidized and transformed into the C=O group [71]. Next, the generated P2 is oxidized by radicals h+, •OH, and •O2- to produce P3, which underwent rapid transformation to P4 through tautomerization. Groups -C4OH7 in P4 are removed by radicals to form a long-chain alkane primary amine (P5) [72]. The primary amine (P5) is then progressively deaminated and demethylated to form P6 by radicals. Ultimately, these intermediate products may be broken down into small molecules and completely mineralize into CO2 and H2O. The lack of inclusion of intermediates in the GC-MS at 12 and 16 hours is probably owing to their poor stability and susceptibility to degradation in the experimental conditions.
As shown in Fig. 9, acute toxicity (as measured by the fathead minnow 50% lethal dose (LC50-96 h) and bioaccumulation factor are employed to evaluate the toxicities of DMP and the speculative photodegradation intermediates through the Toxicity Estimation Software Tool (T.E.S.T.) using the consensus method based on Quantitative Structure Activity Relationship (QSAR) prediction. Compared to DMP, the acute toxicities of intermediates tend to decrease when generating P1 and P2, then increase up until ensuing P3-P6, signifying that most intermediates are unfavored for reducing the toxicity and potential danger to the aqueous environment. The bioaccumulation factor exhibits similar pattern as acute toxicity, implying that only sufficient photocatalytic degradation and entire mineralization of DMP to CO2 and H2O could alleviate the bioconcentration effect of DMP on the environment. Simultaneously, the impurities in industrial DMP, except I1, are more toxic and tendentious than DMP to bioconcentration effect. It is conceivable that the treatment procedures for adsorption of DMP and coexisting harmful impurities from brine and then photocatalytic degradation by BMS@TiO2 in air could be an efficient and environmentally friendly approach to removing micropollutants, avoiding harmful intermediates enter the brine and cause secondary pollution. Fig. 10 summarizes the process of adsorption and photodegradation of DMP using BMS@TiO2.