Adsorption/reduction of N-dimethylnitrosamine from aqueous solution using nano zero-valent iron nanoparticles supported on ordered mesoporous silica

As the National Toxicology Program reported in 2011, N-dimethylnitrosamine (NDMA) is reasonably anticipated to be a human carcinogen. It is detected in many drinking water sources and rivers in some countries (such as China, USA and Canada), and also in treated drinking water and waste water (Mitch et al. 2003; Han et al. 2013). Subsequent studies have demonstrated that NDMA is one disinfection by-product of chlorination and chloramination of water and waste water (Choi & Valentine 2002; Mitch & Sedlak 2002). Besides, it can be formed during ozonation and other treatments with strong oxidants (Andrzejewski et al. 2008; Yang et al. 2009; von Gunten et al. 2010). It arouses increasing concern among public health agencies. The US Environmental Protection Agency classifies it in group B2 as a ‘Probable human carcinogen’. So it is necessary to develop some effective methods to remove it from drinking water.

Reductive processes with or without the augment of hydrogen have been investigated as alternative treatment methods for NDMA removal. Catalysts, including Pd, Pd-Cu, Pd-In, Ni and NiB, have shown the ability to reduce NDMA with hydrogen (Davie et al. 2008; Frierdich et al. 2009). Zero-valent iron and nickel-enhanced iron both transform NDMA to dimethylamine (DMA) and NH₄⁺ without hydrogen (Odziemkowski et al. 2000). Like many other nanomaterials, the zero-valent iron nanoparticle (ZVINP) is easy to agglomerate into large particles, resulting in the decrease of surface area and reactivity performance, which limits its application in water treatment (Phenrat et al. 2006). Therefore, researchers have paid attention to the modification of nano zero-valent iron (He & Zhao 2005; Chun et al. 2010; Wang et al. 2010; Zhang et al. 2011). Immobilization of ZVINP in mesoporous material is a promising approach to solve the problem above (Li et al. 2011; Lv et al. 2011). Mesoporous materials with large specific surface area, uniform aperture adjustment, morphology and active surface groups come onto people's horizons. It is widely used in macromolecule adsorption and separation, biological medicine, chemical catalysis, chemical sensors, and the synthesis of nanomaterials (Wakayama et al. 2003; Fan et al. 2005). Mesoporous silica has attracted intense interest due to its adsorption capacity, high surface area, porous structure, and relatively low cost. Mesoporous silica MCM-41 has been proven to be generally quite effective for iron loading (Liu et al. 2014). The combination of MCM-41 and ZVINP would take advantage of the two materials. It is interesting to evaluate the efficiency of NDMA adsorption/reduction by ZVINP supported on MCM-41.

In this study, we designed and prepared a new kind of nanocomposite by loading nano zero-valent iron onto ordered mesoporous silica materials (ZVINP/MCM-41), which could reduce contact with oxygen. It could avoid zero-valent iron oxide and increase adsorption at the same time. The prepared ZVINP/MCM-41 showed excellent reductive performance towards NDMA.

All chemicals, obtained from Sinopharm Chemical Reagent Beijing Co Ltd, China, are of analytical reagent grade or better quality. Ultrapure water was used throughout the experiment (18.2 MΩ cm).

MCM-41 was synthesized according to the literature as by Dimos et al. (2009). An amount of 50 g tetraethylorthosilicate was added into a polyethylene bottle containing 417.5 g H₂O, 268.5 g NH₃ (25% wt) and 10.5 g cetyltrimethylammonium bromide, and stirred for 30 min. After heat treatment at 80 °C for 96 h, the product was retrieved. It was filtered, rinsed with cold ethanol and finally placed on a plate for air-drying.

ZVINP/MCM-41 was prepared by the liquid phase reduction method. Typically, FeSO₄·7H₂O (8.93 g) was dissolved in ethanol/water (4/6) solution (100 mL), followed by the addition of polyethylene glycol-4000 (0.125 g/g). MCM-41 (20 g) was added into the above solution and stirred well. Then, NaBH₄ solution (0.45 mol/L) was added into the above mixed solution drop by drop. All of the reactions were conducted in an anaerobic environment. After reaction, the products were washed three times with ethanol, acetone, and anaerobic water (oxygen was removed from the water by passing nitrogen through the water) respectively. The resulting precipitates (ZVINP/MCM-41) were ready for characterization.

The morphological structure of ZVINP/MCM-41 was observed by ZEISS scanning electron microscopy. All SEM specimens were sputter-coated with a thin layer of gold palladium under vacuum in an argon atmosphere prior to examination. X-ray diffraction (XRD) patterns of the prepared samples were acquired with a Tokyo Rigaku D/MAX 2200 X-ray diffractometer using CuKα radiation (40 kV, 300 mA) of wavelength 0.154 nm to confirm the structure of the materials. Surface area measurements were performed on a Micromeritics ASAP 2020 surface area and porosity analyzer. Fourier transform infrared (FTIR) spectra were recorded in the spectral range of 4,000–400 cm⁻¹ on a Perkin–Elmer Spectrum One FTIR spectrometer.

In a typical batch experiment, 20 mg of the as-prepared ZVINP/MCM-41 was added into a 1,000 mL 1.0 μg/L of NDMA solution, the mixture was adjusted to pH 7.6 using HCl and NaOH and stirred for 120 min. Thereafter, ZVINP/MCM-41 was separated by filter separation. In order to obtain the reaction kinetics of the NDMA solutions, different reaction times (15–240 min) were treated with the same procedure as above. In order to obtain the removal isotherms of the NDMA solutions, varying initial concentrations of individual NDMA solutions (1.0–6.0 μg/L) were treated with the same procedure as above at room temperature. Blank samples (containing only deionized water and corresponding ZVINP/MCM-41) were prepared and monitored for the duration of the experiment as a control.

The NDMA in the solution was analyzed using an Agilent 7890A triple quadrupole gas chromatography–mass spectrometer. It was equipped with an HP-5MS column (30 m × 0.25 mm, 0.25 μm) using helium (99.999%) as carrier gas (1 mL/min flow rate). The column temperature was held at 40 °C for 1 min, increased to 150 °C at 5 °C/min, and then increased to 280 °C at 20 °C/min. Temperatures of the transfer line and source of ionization were 280 °C and 230 °C, respectively. The injection port temperature was 250 °C. The retention time of NDMA was 3.95 min.

For unsymmetrical dimethylhydrazine (UDMH) measurement, the liquid samples were pretreated by 4-nitrobenzaldehyde, separated by the C18 column and detected by the ultraviolet detector (λ = 390 nm). The acetonitrile and water with a volume ratio of 4:1 were taken as the mobile phases. DMA concentration was measured by an Agilent GC7890/FID after derivatization with benzenesulfonyl chloride using the improved method by Yang et al. (2010). It was equipped with an HP-5 column (30 m × 0.32 mm, 0.25 μm) using nitrogen (99.999%) as carrier gas (2 mL/min flow rate). The column temperature was held at 120 °C, increased to 195 °C at 40 °C/min, held for 1 min, and then increased to 290 °C at 60^oC/min. The injection port temperature was 280 °C. The retention time of DMA was 6.56 min.

The XRD patterns of MCM-41 and ZVINP/MCM-41 are presented in Figure 1(c) and 1(d). Figure 1(c) compares the X-ray diffraction patterns of MCM-41 before and after ZVINP loading. In Figure 1(c), there is an obvious zero-valent iron characteristic peak at 2θ = 43°–45° (α-Fe⁰), which is in accordance with the report by Choi & Valentine (2002). It indicates that ZVINP was successfully loaded into MCM-41.

The FTIR spectra of MCM-41 and ZVINP/MCM-41 in the range of 4,000–400 cm⁻¹ are shown in Figure 1(d). A band at 3,445 cm⁻¹ in the spectra shows H–O–H hydrogen-bonded water. The intensive band at around 1,040 cm⁻¹ could be assigned to Si–O stretching vibrations. After loading ZVINP, the Fe–O and Fe–OH vibrations appeared at 1,400 and 1,020 cm⁻¹, respectively.

According to the Brunauer–Emmett–Teller (BET) analysis, after loading ZVINP, Barrett–Joyner–Halenda (BJH) desorption cumulative volume of pores was 0.1785 cm³/g, which was smaller than that before loading ZVINP (0.2037 cm³/g) and the BET surface area was 30.58 m²/g, which was larger than that before loading ZVINP (27.64 m²/g). This may be due to the loading of ZVINP on the surface of the MCM-41, which could increase the surface area.

In order to determine the effect of ZVINP/MCM-41 dose, the 2 h removal experiments were carried out at 25°C within the ZVINP/MCM-41 dosage range from 5 to 40 mg in 1,000 mL 1.0 μg/L NDMA solution. As shown in Figure 2(b), the removal of NDMA increased with the increase of ZVINP/MCM-41 dose, and reached a plateau at the appropriate dose of 20 mg. A dose of 20 mg of the nanocomposite was chosen for the research.

The pH value of the solution was an important controlling parameter in the removal process. To determine the effect of pH, the 2 h removal experiments were carried out at 25 °C with the pH ranging from 4.3 to 9.2 in 1,000 mL 1.0 μg/L NDMA solution. The removal of NDMA using ZVINP/MCM-41, ZVINP and MCM-41 increased significantly with increasing pH from 4.3 to 7.6, and decreased obviously at pH 9.2. It can be concluded that too much acid or alkaline could affect the structure of ZVINP/MCM-41, which may reduce the NDMA removal rate of ZVINP/MCM-41 (Figure 2(c)). Under pH 7.6, the removal efficiency of NDMA using ZVINP/MCM-41 was better.

A dose of 20 mg ZVINP/MCM-41 was added into 1,000 mL 1.0 μg/L (0.0135 mmol/L) of NDMA solution at pH 7.6 and the remaining amount of NDMA in solution was detected at 15, 30, 45, 60, 75, 90, 120, 150, 180, 210, and 240 min, respectively. The results showed that the removal of NDMA by ZVINP/MCM-41 increased with time and reached equilibrium at about 2 h (Figure 2(d)).

In Figure 2(d), the removal efficiency of ZVINP was low in the first 90 min, increased after 90 min and got to equilibrium after 210 min, and it is also shown that the removal rate of ZVINP/MCM-41 was faster than ZVINP. Reaction kinetics were needed to confirm it.

As NDMA degraded, DMA and UDMH formed accordingly. The nitrogen mass balance varied with the degradation of NDMA, and the final nitrogen mass balance was only 70.4%. UDMH accounted for 48.9% of the nitrogen mass, while DMA accounted for 21.5%. The loss of nitrogen mass was about 29.6%. The most detected products in the previous studies on NDMA reduction have been DMA and ammonium (NH₄⁺) (Davie et al. 2008; Frierdich et al. 2009). This can explain some nitrogen loss. Others may be adsorbed into ZVINP/MCM-41.

From the whole process we could infer that the removal of NDMA from water involves three main steps: (1) ZVINP/MCM-41 adsorbed NDMA onto the surface; (2) ZVINP/MCM-41 reduced NDMA to DMA and ammonium; (3) DMA was desorbed into the solution.

In the study, ZVINP/MCM-41 was prepared, and batch equilibrium removal of NDMA onto ZVINP/MCM-41 was carried out. Removal of the NDMA onto ZVINP/MCM-41 reached equilibrium within 120 min at pH 7.6, which agrees well with the second-order kinetic model. The product analysis found that in the process of removal, adsorption and reduction exist. Furthermore, the reduction reaction occurred after the adsorption reaction. It showed that this removal process was one of adsorption and reduction synergies.

This study was supported by National Major Projects on Water Pollution Control and Management Technology (No. 2012ZX07404-003) and the Natural Science Foundation of Shandong Province (No. ZR2014EEP008).