Study of phosphate removal from aqueous solution by zinc oxide

As a principal nutrient, phosphate (P) plays a vital role in ecosystems. However, if excessive amounts of phosphate, released from agricultural fertilizer runoff, municipal wastewater, or effluent from chemical and food processing industries, are discharged into water bodies, there might be overgrowth of aquatic plants, depletion of dissolved oxygen and subsequent deterioration of water quality. Accordingly, it is necessary to reduce the phosphate concentration and control the discharge of phosphate.

To achieve this goal, various techniques, including chemical precipitation, biological removal, adsorption, crystallization and anion exchange, have been studied (Morse et al. 1998). However, the biological process is not steady and relies heavily on water quality, and chemical precipitation can be subject to problems with sludge treatment and disposal (Yeon et al. 2008). Anion exchange also has drawbacks, e.g., low selectivity in the presence of competing anions (Tsuji 2002). Compared with the traditional methods mentioned above, adsorption provides faster phosphate removal (Chouyyok et al. 2010), and can be applied to different phosphate concentrations under various environmental conditions (Wang et al. 2007; Haghseresht et al. 2009; Rentz et al. 2009; Pitakteeratham et al. 2013). It is considered a promising technique for removing phosphate. For this reason, a great many low-cost and easily available materials, including industrial by-products, natural materials, metal-modified minerals, man-made products, inorganic–organic hybrid materials and metal oxide/multi-metal composite oxides have been tested (Liao et al. 2006; Mohan & Pittman 2007; Vohla et al. 2011; Jiang & Ashekuzzaman 2012).

Zinc oxide (ZnO), as an environment-friendly material (Wan 2006), has been used for diverse applications (Hariharan 2006; Wang et al. 2011; Moezzi et al. 2012) and is considered a mature engineering material with annual production of one and a half million tons (Moezzi et al. 2012). Kikuchi et al. (2006) have reported its sorption performance toward metal ions and Liufu et al. (2004) investigated the effect of polyethylene glycol on the adsorption onto ZnO/polymer solution interface, and found that the blockage of the adsorption sites played a major role in the change of zeta potential. Regarding phosphorus, Namasivayam & Sangeetha (2004) found that ZnCl₂-activated carbon developed from coir pith is a good sorbent for phosphate adsorption over a wide pH range (pH ∼ 3–10). Moreover, Cheng et al. (2009) and Zhou et al. (2011) reported that zinc-aluminum hydroxides (oxides) can effectively remove phosphate with a high content of ZNO at a Zn/Al molar ratio of 2, sintered at 573 K. He et al. (2010) reported that ZnAl layered double hydroxides show selective adsorption of phosphate and the calcined samples possibly proceed through ligand complexation or electrostatic attraction between phosphate ions and hydrated ZnO formed after calcination. However, up to now, phosphate adsorption by more stable ZnO particles has not been reported.

In the present study, ZnO was synthesized and used to investigate the mechanism of phosphate removal. The effect of pH and co-existing anions on phosphate capture, as well as the adsorption isotherm and kinetics, were evaluated. X-ray diffraction (XRD), scanning electron microscope (SEM) and Fourier transform infrared spectroscopy (FT-IR) were also performed to preliminarily elucidate the underlying sorption mechanism of ZnO. With this information, a better understanding of adsorption behaviors and properties could be obtained, and the feasibility of the use of ZnO in water/waste water will become clearer.

All chemicals of analytical grade were purchased from the Tianjin Fuchen Chemical Reagents Factory (Tianjin, China), and used without any further purifications. ZnO particles were fabricated by the chemical precipitation-thermal decomposition method with minor modifications. In brief, 0.1 M NaOH solution was dropwise added into 80 mL distilled water containing 10 g of ZnSO₄•7H₂O with continuous stirring, until the solution pH reached around 8 ± 0.1. The suspended solution was stirred for 3 h at room temperature. The obtained pastes were harvested by repeating the following procedure three times: centrifuging the solution at 4,000 rpm for 5 min, and then washing with distilled water. The pastes were then put into the Muffle furnace and kept at 573 K for 6 h (Zhou et al. 2011) to produce the ZnO particles. The obtained ZnO particles were ground into powder with an agate mortar and kept in a dryer for further use.

All the samples were filtered through 0.45 μm membranes before determination. Phosphate concentrations were determined by a Metrohm 881 ion chromatograph coupled with a Metrosep A Supp 4 column. A solution comprising 1.8 mM of Na₂CO₃ and 1.7 mM of NaHCO₃ was used as a mobile phase at a flow rate of 1.2 mL/min.

The surface morphology and element composition of the ZnO particles were obtained by a SUTW-Sapphire system with energy dispersive X-ray analysis (EDAX) (PHILIPS XL30 ESEM, Eindhoven, The Netherlands). The XRD pattern was used to characterize the relative intensity of ZnO particles, which were fabricated as described in the ‘Materials' section above, and recorded by a Bruker D8 Advance (Bruker, Bremen, Germany) using Cu Kα radiation within the 2θ range of 4–90°. The FT-IR spectroscopy was obtained using a Vertex 70 with a DTGS (deuterated triglycine sulfate) detector (Bruker, Bremen, Germany) at 4,000–400 cm⁻¹, with a pellet of powdered potassium bromide. For comparison, the XRD, SEM and FT-IR of the ZnO particles after adsorption were also conducted.

A series of batch experiments was conducted to investigate the characteristics of phosphate adsorption onto ZnO. Kinetic experiments were performed by sampling a 2 mL solution from the 1,000 mL working solution containing 400 ± 0.03 mg of ZnO particles and the desired concentration of working solution (10, 20 and 30 mg/L) at various shaking times (0, 5, 10, 20, 30, 45, 60, 90, 150, 270, 390 510, 690 min) in an incubator shaker (WR-1, Shanghai, China) at 298 K.

Sorption isotherm experiments were conducted in 50 mL sealed Erlenmeyer flasks containing 25 mL working solution and 10 ± 0.03 mg of ZnO particles with various phosphate concentrations (0–100 mg/L) and a shaking time of 690 min at desired temperatures.

To determine the effect of pH, 10 ± 0.03 mg of ZnO particles was added into 50 mL Erlenmeyer flasks containing 25 mL working solution (20 mg/L) at desired pH values (3–11) adjusted by 0.1 M HCl and 0.1 M NaOH solution at 298 K. These flasks were sealed with parafilms (Pechiney Plastic Packaging Co., Washington, NJ, USA), and then transferred into the incubator shaker for 1,440 min. The solutions were then filtered through 0.45 μm membranes to determine the equilibrium pH. The concentrations of the releasing Zn²⁺ were determined by inductively coupled plasma (ICP) (PLASMA-SPEC I, Shelton, CT, USA).

The effects of co-existing anions (sulfate, carbonate, nitrate and chloride ion) on phosphate adsorption were investigated in 25 mL working solution (20 mg/L and 10 ± 0.03 mg of ZnO particles), which contained 10 or 20 mg/L of ⁠, ⁠, ⁠, or Cl⁻ ions at 298 K.

All working solutions were directly diluted from phosphate stock solution (counted as P, 1,000 mg/L) and each of the experiments was conducted twice.

According to the SEM photographs shown in Figure 1, ZnO particles exhibited individual, sheet morphologies and the particles formed aggregates of different sizes. The EDAX spectrum of ZnO particles after phosphate adsorption was also analyzed and the P element occurred in the EDAX spectrum (Figure S1, in the supporting information, available online at http://www.iwaponline.com/jwh/013/210.pdf), which confirmed the presence of phosphate on the surface of the ZnO particles after adsorption.

Figure 2 shows the XRD results of fresh ZnO samples and ZnO samples obtained from the kinetic experiments. The strong and narrow peaks indicate that the adsorbent was well crystallized during the calcination process. The diffraction peaks of ZnO samples before and after adsorption correspond well with the standard data for ZnO (Zincite, syn. JCPDS card #36-1451).

The FT-IR spectra of fresh and phosphate-loaded ZnO particles are presented in Figure S2 (available online at http://www.iwaponline.com/jwh/013/210.pdf). The broad peaks apparent around the 3,400–1,630 cm⁻¹ region might be attributed to the existence of the O-H stretching and bending vibrations, which were mainly caused by the adsorbed H₂O molecules (Lv et al. 2013; Zhang et al. 2013). The peaks appearing at around 1,114, 1,069 and 965 cm⁻¹ might be assigned to the vibration of (Wang et al. 2006) from the raw materials (ZnSO₄·7H₂O). The peaks presenting at around 415–425 cm⁻¹ might correspond to the stretching frequency of Zn-O (Musić et al. 2002; Prasad et al. 2006). After adsorption, marked peaks appearing at 1,026, 1,006 and 945 cm⁻¹ might be attributable to the symmetrical stretching vibration of phosphate (Arai & Sparks 2001; Ning et al. 2013). The peak at 965 cm⁻¹ disappearing after phosphate adsorption might be covered by that of phosphate. It is also noteworthy that the frequency of Zn-O vibration moved from 415 to 425 cm⁻¹ after uptake of phosphate, which might be due to the fact that the active sites reacted with phosphate.

As can be seen from Figure 3(a), the phosphate adsorption capacity of ZnO increased rapidly in the first 150 min, and then gradually increased until the equilibrium was reached at 510 min. The further extension of contact time did not affect the adsorption amounts of ZnO noticeably. The phosphate adsorption amount increased with the increasing initial phosphate concentration (from 23.6 mg P/g in 10 mg/L to 60.5 mg P/g in 30 mg/L).

To further study the phosphate adsorption onto ZnO, the adsorption kinetic data were fitted to an integrated first (or second)-order kinetic expression, and are presented in Figure 3(b) and 3(c); the fitting parameters are listed in Table 1. According to the first-order kinetic model, the adsorption process could be divided into two parts by the adsorption rate constant (k_p1 > k_p2, k_p1 and k_p2 were the rate constant of fast and slow adsorption phase, respectively, and the values are listed in Table 1), i.e., fast adsorption and slow adsorption, which were also verified by the effect of contact time on the adsorption process shown in Figure 3(a). The fast adsorption process might be attributed to electrostatic attraction, or the driving force of a concentration gradient at higher initial phosphate concentration, which resulted in the rapid transportation of phosphate to the surface of the adsorbent, with respect to that of the adsorbent surface. The following slow adsorption process suggests that the driving force of the concentration gradient became weaker or the intraparticle diffusion dominated the adsorption process at that stage.

In addition, the pseudo-second-order model was also introduced to fit the kinetic data, and the results suggest that the adsorption of phosphate onto ZnO was chemisorption, or chemical bonding between adsorbent-active sites (Wang et al. 2006). That conclusion could also be supported by the close values of q_e(exp) and q_e(cal) of the pseudo-second-order model.

The adsorption of phosphate onto ZnO at different temperatures was fitted by Langmuir and Freundlich isotherm models and is depicted in Figure 4. According to Table 2, at 298 K, the isotherm data could be better fitted by the Langmuir model rather than the Freundlich model, indicating that the production of zinc phosphate might be neglected because the Langmuir model could have the optimum performance to characterize the adsorption processes without precipitation (Del Bubba et al. 2003). Comparing the values of parameter R_L (Figure S3, available online at http://www.iwaponline.com/jwh/013/210.pdf), it is suggested that the adsorption of phosphate was more favorable at higher initial concentrations than at lower ones (Foo & Hameed 2010).

Simultaneously, the n values obtained from the Freundlich model ranged from 1 to 10, which also implies favorable adsorption of phosphate onto ZnO. In addition, at 308 K, the correlation coefficient R² (0.970) suggests that high temperature strengthened the adsorption process, which led to disordered, overlapped distributions of phosphate ions on the surface of the ZnO particles, or chemical reaction between ZnO and phosphate to form crystals of zinc phosphate.

The q_m values might not provide an accurate estimation of the long-term sorption capacity, but could still be useful for comparing alternative materials. As shown in Table 3, the adsorption capacity was 168.4 mg P/g for ZnO, which is much higher than most of the other adsorbents reported before, and suggests that ZnO has a relatively high potential for use as a phosphate adsorbent.

As shown in Figure 5 and Table 4, the values of ΔG°, which indicate the spontaneity degree of the sorption process, are negative (−3.15 to −1.87 kJ/mol) and decreased gradually with the increasing temperature, suggesting that the adsorption process was spontaneous and was enhanced with the increasing temperature. The capacity for ZnO particles to adsorb phosphate also increased with increasing temperature (see Figure 4), which also supports this conclusion. The positive value of ΔH° (16.65 kJ/mol) demonstrates the endothermic nature of the reaction. The positive ΔS° value obtained (64.46 J/(mol·K)) indicates the degree of disorder, and the increase of randomness due to the increase in the number of species at the solid/liquid interface when the phosphate moves from the hydrous phase to the surface of the adsorbent (Wang et al. 2006).

The effect of pH on phosphate removal by ZnO was investigated at pH values ranging from 3 to 11. It is clear from Figure 6 that the adsorption is strongly dependent on pH value, and the adsorption capacity decreased significantly with the increasing pH value. The maximum adsorption capacity was about 50 mg P/g at pH ∼ 3.1, which was over 15 times that at pH ∼ 10.6. However increased pH value was found in the final solution after 690 min adsorption at an initial pH from 3 to 8.4 but this decreased at initial pH from 8.4 to 11.

Phosphate exists in aqueous solution as different species, i.e., H₃PO₄///⁠. The surface group properties of ZnO are associated with the point of zero charge (pH_pzc), and variations of speciation distributions are attributed to the changes of solution pH. pH_pzc values were determined by the pH-drift method described by Yang et al. (2004) and Liu et al. (2011). As can be seen from Figure 6, the pH_pzc for ZnO is around 8.4. When solution pH is lower than 8.4, the surface of ZnO could be protonated as ≡Zn-OH + H⁺↔ ≡Zn-⁠, where ≡Zn-OH represents the uncharged surface group, and ≡Zn-OH₂⁺ represents the positively charged surface group. The positively charged surface groups are easier to replace at a lower pH from the metal binding sites than hydroxyl groups, which could facilitate the ligand exchange process (Chubar et al. 2005; He et al. 2010). Moreover, lower pH is favorable for phosphate adsorption through enlarging protonation to boost the positively charged sites, and thus leads to higher phosphate adsorption capacities. Simultaneously, anionic phosphate could interact with the positively charged surface group by electrostatic forces. However, with increasing pH value, becomes the dominant species instead of ⁠, progressively, and the capacity for continuous protonation is weakened, which leads to the decline of positively charged sites, and contributes to the decline in the ability of ZnO to adsorb for phosphate.

When solution pH is higher than 8.4, the dominant species are and ⁠, where the deprotonation of ≡Zn-OH, i.e., ≡Zn-OH – H⁺ ↔ ≡Zn-O⁻ + H⁺ exists, where ≡Zn-O⁻ represents the negatively charged surface group. The strong electrostatic repulsion between the two negative species Zn-O⁻ and become more dominant, which prevents from adsorbing onto the surface of ZnO. At the same time, the ligand exchange becomes weaker (He et al. 2010) because of deprotonation of the adsorbent, and the Lewis acid-base interaction is strengthened due to the increased numbers of oxygen anions in the different ionic forms of phosphate (Zhang et al. 2012). With increasing pH, the Lewis acid-base interaction dominates the adsorption process. There existed strong electrostatic repulsion between the surface of the adsorbent and the more negative-charged ⁠. At the same time, there was competition between the negatively charged phosphate species and the hydroxide groups adsorbed onto the more negatively charged adsorbent (Huang et al. 2013), and phosphate adsorption by electrical force might not occur. All these reasons led to the capacity for phosphate adsorption to drop to the minimum point, i.e., 2.9 mg P/g.

Competitive sorption is an important factor that influences the removal efficiency. Anions such as ⁠, ⁠, and Cl⁻ commonly exist in water and wastewater, and might interfere with the adsorption of phosphate by competing for sorptive sites on the surface of adsorbents. As shown in Figure 7, ⁠, and Cl⁻ do not show any negative effects on phosphate adsorption at concentrations from 0 to 20 mg/L. Anions such as are often weakly bounded with surface sites of metal hydroxides to form outer-sphere surface complexes, but phosphates are relatively strongly bounded with the surface sites forming inner-sphere surface complexes (Zhang et al. 2008), so the phosphate adsorption process is barely affected by ⁠, and Cl⁻. However, Rahnemaie et al. (2007) suggested that carbonate was able to act as a competitor for phosphate at a relatively high concentration. The removal efficiency decreased from 96.5% to 53.5% when the concentration of was increased from 0 to 20 mg/L, which was probably due to the higher affinity of to the surface of ZnO to form or ⁠. According to the solubility product constants of zinc carbonate (K_sp = 1.46 × 10⁻¹⁰) and zinc phosphate (K_sp = 9.1 × 10⁻³³) (Lide 1997-1998), zinc phosphate was easier to precipitate and more settlable in water solution. However, the addition of carbonate decreased the phosphate removal efficiency, suggesting that phosphate adsorption onto ZnO might not be caused by chemical precipitation but rather by chemical adsorption through exchange with OH⁻ on the hydroxylation surface of ZnO. However, carbonate addition during the adsorption process that consumed some H⁺ and improved solution pH values also decreased the phosphate removal efficiency.

In the present study, ZNO was synthesized using the chemical precipitation-thermal decomposition method to investigate its capacity for phosphate adsorption. The obtained ZnO particles were well crystallized and exhibited individual, sheet morphologies. ZnO has a strong phosphorus removal ability of 163.4 mg/g and the dominant phosphate removal mechanism was probably due to adsorption rather than precipitation. Adsorption of phosphate was highly pH-dependent, and the adsorption capacity decreased with increasing pH. At the same time, Lewis acid-base interaction took the place of ligand exchange to be the dominant mechanism. significantly affected phosphate adsorption due to the higher affinity of for the surface of ZnO. Adsorption performance analysis can provide valuable information, representing practical value for the technological applications of phosphorus removal from aqueous solutions.

The authors wish to thank the Natural Science Foundation of China (No. 51238001) and the China Scholarship Council (CSC No. 201404480081) for their financial support.