Elimination of nitrate in secondary ef ﬂ uent of wastewater treatment plants by Fe 0 and Pd-Cu/diatomite

Because total nitrogen (TN), in which nitrate (NO 3 – ) is dominant in the ef ﬂ uent of most wastewater treatment plants, cannot meet the requirement of Chinese wastewater discharge standard ( < 15 mg/L), NO 3 – elimination has attracted considerable attention. In this research, the novel diatomite-supported palladium-copper catalyst (Pd-Cu/diatomite) with zero-valent iron (Fe 0 ) was tried to use for catalytic reduction of nitrate in wastewater. Firstly, speci ﬁ c operational conditions (such as mass ratio of Pd: Cu, catalyst amounts, reaction time and pH of solution) were optimized for nitrate reduction in arti ﬁ cial solution. Secondly, the selected optimal conditions were further employed for nitrate elimination of real ef ﬂ uent of a wastewater treatment plant in Beijing, China. Results showed that 67% of nitrate removal and 62% of N 2 selectivity could be obtained under the following conditions: 5 g/L Fe 0 , 3:1 mass ratio (Pd:Cu), 4 g/L catalyst, 2 h reaction time and pH 4.3. Finally, the mechanism of catalytic nitrate reduction was also proposed.


INTRODUCTION
Contamination of nitrate in water resources has become a severe environmental problem. Excess nitrate in water can cause water pollution (such as river eutrophication and water quality deterioration). On the other hand, nitrate can also be reduced to nitrite, which poses several health threats to humans, such as liver damage and even cancer (Hosseini et al. ). Nitrate in wastewater mainly comes from agricultural activities, domestic and industrial sewage (Hwang et al. ). With aggravation of water pollution, efficient technologies for nitrate elimination have attracted much attention.
Among the existing technologies for nitrate removal, physico-chemical denitrification, biological and chemical reductions have been widely used (Kim et al. ).
However, physico-chemical methods like ion exchange, reverse osmosis and electrodialysis, require frequent regeneration of the medium and further treatment for the secondary waste produced (Soares et al. ). Biological denitrification can achieve high nitrate removal.
However, this process is complex and requires monitoring of the carbon source (Subramanyan et al. ). To date, chemical catalytic reduction of nitrate has been regarded as one of the promising techniques to reduce nitrate in wastewater. Vorlop & Tacke () proposed the new method of catalytic reduction to reduce nitrate in water. Since then, this potential technology has been gradually accepted by researchers. In previous research, H 2 and organic acids (e.g. HCOOH) were employed as a reductant for catalytic nitrate reduction. However, the potential risk of explosion, low solubility of H 2 in aqueous media, difficulty in operational conditions and potential health problems limit

Materials
Diatomite with a specific surface area of 285 m 2 /g and a mean particle diameter of 62 nm was used as the support.

Catalyst preparation
A wet impregnation method was adopted to obtain Pd-Cu/ diatomite catalyst (Saada et al. ). PdCl 2 and CuCl 2 ·2H 2 O were used as metal precursors to obtain the desired amount of metals coated on diatomite. Five grams of the catalyst was prepared as follows: (1) PdCl 2 (0.5 g) and CuCl 2 ·2H 2 O (0.3 g) solution were added into the diatomite (4.8 g) suspension solution and continuously stirred for 10 min (400 rpm); (2) the above mixed solution (400 mL Pd-Cu/diatomite catalyst was separated using a centrifuge (5,000 rpm, 10 min per cycle); (6) the prepared catalyst was washed twice with deionized water and dried in a vacuum oven at 60 W C for 24 h.

Catalyst characterization
The specific surface area of the catalyst was determined by a Brunauer-Emmett-Teller (BET) surface area analyzer (F-Sorb  All experiments were performed in necked flasks with a total volume of 150 mL. One hundred mL of NaNO 3 solution or wastewater was added to each flask with 0.5 g of Fe 0 . All flasks were placed in an electronic oscillator stirring under 250 rpm at room temperature.

Laboratory analyses
Samples were collected periodically to determine nitrate, nitrite and ammonium concentrations in accordance with standard methods. Nitrate removal and catalytic selectivity to N 2 were calculated as: where C 0 is the initial concentration of nitrate in solution (mg/L), C t is the nitrate concentration (mg/L) at time t (min) and C N2 is the amount of N 2 produced (mg/L).

Optimization of operational condition
Pd:Cu mass ratio By-products were analyzed after catalytic reduction.
Results indicated that the removed nitrate was mostly converted to N 2 , whilst ammonium and nitrite were only minor parts under the optimal Pd:Cu mass ratio (3:1). EDS analysis shown in Figure 2(a) indicates that Pd and Cu were both detected on diatomite and the mass ratio was nearly 3:1.   The results are shown in Table 1.
Based on the results analyzed above, 64% of nitrate removal efficiency and 60% of N 2 selectivity were reached.
According to the related research, the catalytic performance under the catalysis of different Pd-Cu catalysts (e.g. CeO 2 , TiO 2 , MnO 2 , Al 2 O 3 , SiO 2 ) with H 2 remained lower, the   nitrate removal ranging from 20 to 40%, while the N 2 selectivity ranged from 15 to 70% (Soares et al. ). It is clear to see that the Fe 0 and Pd-Cu/diatomite showed a better catalytic performance.
Additionally, it is obvious that after treatment by catalytic reduction, the effluent total nitrogen (TN) concentration greatly reduced to 11.2 mg/L (<15 mg/L). Indicators in Table 1

Mechanism
Many studies have been conducted on nitrate reduction with the reductant H 2 and different kinds of bimetallic catalysts.
In our research, instead of H 2 , Fe 0 with the catalyst (Pd-Cu/ diatomite) was used for the catalytic reduction of nitrate.
The nitrate reduction mechanism is depicted as a conceptual model in Figure 6.
The catalytic reduction of nitrate has been regarded as a typical heterogeneous catalysis process. In this redox process, Fe 0 mainly served as the electron donor (see Fe 0 À 2e À , Fe 2þ φ θ Fe 0 Fe 2þ ¼ 0:44 V; (3) The standard electrode potential formula is E θ ¼ φ θ (oxidiser) -φ θ (reductant). The standard electrode potentials (E θ ) of NO 3 À to NO 2 À , NO 2 À to NH 4 þ and NO 2 À to N 2 by Fe 0 are 1.27, 1.33 and 1.96, respectively (>0.2 V). In Based on the experiments and related literature, H þ concentration in solution is also another critical factor for nitrate reduction. As demonstrated in Figure 6, when H þ was sufficient in solution, the catalytic process may follow paths 1 and 2, and then more ammonium could be obtained, dominating the products that have been proved by the data in Figure 5. Actually, N 2 is the harmless gas, which is the desirable product. Therefore, appropriate H þ concentration in solution could be controlled by adjusting pH to make this catalytic reaction follow path 1 and lower the occurrence of path 2. Additionally, appropriate supported materials could also boost the nitrate removal and transformation to N 2 (Aristizábal et al. ).
The bi-metals Pd and Cu coated on diatomite are also crucial in catalytic reduction. On the one hand, it is generally believed that metallic copper, as a promoting metal, could reduce nitrate by producing cupric oxide (CuO) or cuprous oxide (Cu 2 O) and then be regenerated by hydrogens adsorbed on Pd active site, as shown in Figure 7 (Zhao et al.

).
However, based on the results, we assume that Pd played the leading role in catalytic nitrate reduction, strengthening the electron transformation from Fe 0 to H þ in solution, forming active H and promoting NO 3 À conversion to N 2 , as shown in Figure 8.

Kinetic modeling
Different initial nitrate concentrations (100, 50, 40, 30 and 20 mg/L) were set to study the kinetics of catalytic reduction of nitrate by Fe 0 with Pd-Cu/diatomite catalyst. The reaction can be described by the first-order and zero-order reaction as given below (Pintar et al. ).
The first-order equation: and then the following equation can be obtained: The zero-order equation: the equation can be transformed: where ν is the rate of reaction (mg/(L·min)); [NO 3 -] is nitrate concentration at time t (mg/L), [NO 3 -] 0 is the initial concentration of nitrate (mg/L), k obs is the reaction rate constant.
The results in Table 2 indicate that the correlation coefficients (R 2 ) obtained from the first-order equations were in the range of 0.9983-0.9995, while R 2 of zero-order equations ranged from 0.8775 to 0.9144. Therefore, it is obvious to