1 Introduction

The development of urbanization and industrialization has aggravated the discharge of phosphorus-containing wastewater, resulting in an increase in the eutrophication of water bodies. The occurrence of eutrophication has posed a threat for both human health and the ecological environment. Therefore, various technologies (i.e., membrane dialysis, chemical immobilization, biological remediation, and adsorption) were used to purify P-rich water bodies worldwide (Bellier et al. 2006; Daniels and Parker 1973; Eliassen and Tchobanoglous 1969; Erickson 2003; Morton et al. 2005; Zheng et al. 2011a, b). Among these technologies, adsorption has been demonstrated to be one of the most effective means for removing phosphate from an aqueous solution due to the low cost and high efficiency (Chen et al. 2015, 2016; Jung and Ahn 2016; Jung et al. 2016; Lalley et al. 2016; Li et al. 2016; Liu and Zhang 2015; Ma et al. 2016; Naghash and Nezamzadeh-Ejhieh 2015; Rajeswari et al. 2015; Ren et al. 2016; Sakulpaisan et al. 2016; Seliem et al. 2016; Waiman et al. 2016; Wang et al. 2015, 2016a, b; You et al. 2016; Zheng et al. 2016). For instance, the capacity of active red mud, calcite, and activated carbon for phosphate was determined to be 9.8, 3.1, and 3.02 mg/g, respectively (Huang et al. 2008; Hussain et al. 2011; Karageorgiou et al. 2007; Yue et al. 2010; Zhao et al. 2009). However, the limited adsorption capacity of these natural adsorbents hindered practical application.

Nanoscale zero valent iron (NZVI), as a new nanoparticle, has been extensively investigated in the role of removal of various environmental pollutants (Burris et al. 1995; Matheson and Tratnyek 1994; Noubactep 2009; Reynolds et al. 1990; Weber 1996). Moreover, many contaminants, such as organic compounds, nitrate nitrogen, metal, metalloid, and radionuclides, were successfully removed or degraded from an aqueous solution in labs by using NZVI (Adeleye et al. 2013; Bilardi et al. 2013; Choe et al. 2001; Crane and Scott 2012; El-Temsah et al. 2012; Lavine et al. 2001; Liendo et al. 2013; Liu et al. 2012; Mantha et al. 2001; Mukherjee et al. 2016; Noubactep et al. 2003; O’Carroll et al. 2013; Yan et al. 2013; Yun et al. 2013). However, few reports on the removal of phosphate using NZVI were available (Almeelbi and Bezbaruah 2012).

Generally, the NZVI was synthesized by the reduction of Fe3+ with NaBH4 under N2 conditions (Chang et al. 2014; Crane et al. 2011; Crane and Scott 2012; Hoag et al. 2009; Hoch et al. 2008; Liu et al. 2012; Scott et al. 2009). However, as-prepared NZVI was easily aggregated by using this method. Recently, we synthesized NZVI with a method of reduction of natural limonite using hydrogen, which was low cost, offered abundant storage, and was highly efficient. It has been demonstrated that as-prepared NZVI presented the highest removal capacity for Cr(VI) and phosphate (Chang et al. 2014; Liu et al. 2013). However, to remove phosphate, many parameters still need to be taken into consideration, such as annealing, temperature, pH, interferences, and ion strength.

The objectives of this study are as follows: (1) to synthesize NZVI by reduction of natural limonite under H2 conditions and characterize it by XRD, SEM, TEM, and N2-BET surface area; (2) to investigate the effect of environmental factors (i.e., synthesized conditions, pH, ionic strength, oxygen content, and initial P concentration) on the removal of phosphate onto NZVI by batch techniques; and (3) to explore a new application of natural limonite in environmental cleanup.

2 Experiment

2.1 Preparation and Characterization of NZVI

NZVI was prepared by reduction of natural limonite (from the Tongling, Anhui Province, China) with a particle size of less than 74 μm under hydrogen conditions at different temperatures (i.e., 300, 400, 450, 500, 550, and 600 °C). The detailed preparation procedure was described in a previous report (Liu et al. 2013). The as-prepared NZVI, before and after P removal, was characterized using X-ray diffraction (XRD; Dandong DX-2700), transmission electron microscopy (TEM; JEOL JEM-2100), scanning electron microscopy (SEM; JEOL JSM-7100F), N2-BET surface area (Novawin 3000e), and X-ray photoelectron spectroscopy (XPS; Thermo Escalab 250).

2.2 Batch Removal Experiments

To evaluate the removal performance of NZVI prepared at different temperatures, the effects of pH, ion strength, initial PO4 3−-P (P) concentration, and oxygen amount upon removal of phosphate were investigated at ambient temperature using batch experiments. The effect of pH on the removal of P by NZVI (C P  = 10.0 mg/L, m/v = 0.5 g/L) was conducted under ambient conditions. The removal isotherms were carried out at a pH of 6.3 under the different P concentrations, whereas removal kinetics was performed at both oxic and anaerobic conditions.

The effect of foreign anions (Cl−, NO3 −, SO4 2−, S2O3 2−) on P removal was commenced under the fixed condition (C P  = 10.0 mg/L, pH 6.3, m/v = 0.5 g/L).

The column experiment (12-mm width × 300-mm height) was carried out at an influent concentration of P (5 mg/L). After reaction, P was extracted by filtration, and the concentration of phosphate in supernatant was measured using the method proposed by Murphy and Riley (1962). The uptakes of phosphate were calculated from the decrease of initial and equilibrated concentrations.

3 Results and Discussion

3.1 Characterization

The phase composition of as-prepared NZVI was identified by XRD patterns. As shown in Fig. 1, the main constituents of the natural limonite were goethite, whereas the occurrence of magnetite was observed at a heating temperature of 300 °C, indicating the reductive transformation of goethite into magnetite under hydrogen conditions. However, the reflection intensities of magnetite obviously decreased with the increase of temperature. Meanwhile, new reflections of NZVI were found at a heating temperature of 550 °C. The results of XRD patterns indicated that NZVI can be obtained by reduction of limonite under hydrogen conditions at a heating temperature of 550 °C. Therefore, the as-prepared NZVI obtained at 550 °C was used in the following removal experiments.

Fig. 1
figure 1

XRD patterns and surface area of limonite before and after annealing

Full size image

The morphology of as-prepared NZVI was characterized by SEM and TEM images. As shown in Fig. 2a, the crystals of goethite presented a virgulate or circulate shape, with a size ranging from nanoscale to microscale and some irregular substances. In addition, the aggregation consisting of goethite crystals exhibited fasciculation and concentric circles. After reduction at 550 °C, a large number of particles at the nanoscale size were found in the TEM image and identified as zero valent iron, based on the results of XRD pattern and EDS. Furthermore, the SEM images exhibited two kinds of crystal shapes, consistent with the original morphology. However, the prepared material displayed a nanoscale size that was significantly smaller than the original size. This result indicated that the prepared NZVI was of nanoscale size, which was similar to the results of the previous report (Liu et al. 2013).

Fig. 2
figure 2

TEM, SEM, and EDS of limonite (a, b) and reduced limonite at 550 °C (c–f)

Full size image

Additionally, the specific surface area of samples was measured as shown in Fig. 1. The surface area of samples increased to 37.52 m2/g when the temperature reached 300 °C and reduced to 17.24 m2/g when the temperature increased to 600 °C. To the best of our knowledge, NZVI prepared in the liquid phase using potassium/sodium borohydride had a specific surface area of approximately 20 m2/g, which was similar to the findings in previous work (Ponder et al. 2000; Zhang et al. 2012). On the other hand, these surface areas were considerably higher than the amount of iron metal reported previously (from 0.5 to 1.8 m2/g) (Liu et al. 2008; Matheson and Tratnyek 1994; Su and Puls 2004). In brief, NZVI was prepared by the reduction of natural limonite.

3.2 Effect of pH

Figure 3 shows the effect of pH on phosphate removal by various NZVI preparations of different temperatures. NZVI was much more easily dissolved in the aqueous solution, especially in the acidic solution. Furthermore, the presence of dissolved oxygen promoted the further oxidation of Fe2+, resulting in the formation of Fe3+, which can bind with phosphate. This behavior improved the decrease in P concentration in the supernatant solution by forming FePO4 due to a much lower K sp constant of 1.3 × 10−22. In addition, the redox reaction between NZVI and water markedly increased the pH of the solution, promoting the formation of Fe(OH)2 and Fe(OH)3, which can adsorb phosphate by incorporation and electrostatic interaction. Therefore, both precipitation and adsorption contributed to the increase in the amount of P removed.

Fig. 3
figure 3

The effect of pH on P removal by as-prepared NZVI at different heating temperatures (C P  = 10 mg/L, m/v = 0.5 g/L, T = 293 K)

Full size image

3.3 Removal Kinetics

The removal kinetic was analyzed according to the variation of phosphate concentration in an aqueous solution, with different amounts of oxygen. Three different amounts of oxygen were designed; the first was oxic (the reactor was exposed to air), the second was anoxic (the reactor was sealed), and the third was anaerobic (the oxygen in the prepared solution was driven away by boiling).

The removal kinetics of P on NZVI at different initial P concentrations in an oxic system is shown in Fig. 4. It was observed that phosphate was removed completely within 1 h of the reaction time at an initial P concentration of less than 2 mg/L. The removal of P significantly decreased with an increase in initial P concentration, which could be attributed to the competitive removal of more adsorbate (PO4 3−) on limited adsorbent at high initial P concentrations (Reardon 1995; Su and Puls 2001, 2004; Xie and Shang 2007). At pH >4.0, the phosphate can create a complex with the surface groups of corrosion products by a bridging action (Harms et al. 2003). Therefore, an increase in phosphate gradually occupied the surface adsorption sites of the NZVI, which decreased the amount of phosphate removal. This difficulty can be resolved by increasing the mass transport rate and/or the amount of adsorbent released.

Fig. 4
figure 4

Effect of reaction time on P removal in an oxic system (pH = 6.3, m/v = 0.5 g/L, T = 293 K)

Full size image

The data are plotted as P removal against reaction time for different initial concentrations in an anoxic system. This is shown in Fig. 5. The variation of P removal was almost the same as that in an oxic system when the initial P concentration was 5 mg/L or less. Nevertheless, when the initial P concentration was 10 mg/L or more, the P removal was considerably lower than that in an oxic system. Therefore, it was speculated that the amount of oxygen was responsible for the removal of phosphate and to some degree, was positively related to the P removal.

Fig. 5
figure 5

Effect of initial P concentration on P removal in an anoxic system (pH = 6.3, m/v = 0.5 g/L, T = 293 K)

Full size image

Figure 6 shows the removal kinetics of P on NZVI at anaerobic conditions. The P removal at anaerobic conditions was significantly lower than that of the oxic conditions, indicating that a decrease in the amount of oxygen weakened the adsorption capacity of NZVI. NZVI was oxidized into Fe2+ and then transformed into Fe3+ in the presence of oxygen in the acidic condition. The consummation of H+ in this process improved the pH in the aqueous solution, which also favored the removal of phosphate. The effect of the amount of oxygen can also be observed from the adsorption amount of P by NZVI under different systems, as displayed in Fig. 7. The results indicated that the adsorption amount of P can reach 22.7, 16, and 11.1 mg/g under the oxic, anoxic, and anaerobic systems, respectively. This was comparable to the NZVI prepared by the sodium borohydride reduction method (Almeelbi and Bezbaruah 2012) and the NZVI prepared with iron slag (5.3 mg/g) (Xiong et al. 2008). Therefore, it can be concluded that as-prepared NZVI can be used as an excellent adsorbent in the removal of phosphate (99.9%) from aqueous solutions.

Fig. 6
figure 6

Effect of initial P concentration on P removal in an anaerobic system (pH = 6.3, m/v = 0.5 g/L, T = 293 K)

Full size image
Fig. 7
figure 7

The adsorption amount of P by the NZVI under oxic, anoxic, and anaerobic systems (pH = 6.3, m/v = 0.5 g/L, T = 293 K)

Full size image

The pseudo-second-order kinetic equation was used to describe the kinetic data (the linear relationship of the pseudo-first-order kinetic equation was not effective for this study, so it is not listed here),

$$ \frac{t}{q_t}=\frac{1}{{kq_e}^2}+\frac{t}{q_e} $$

where q t (mg/g) and q e (mg/g) are the adsorption amount at the time of t and equilibrium, respectively, and k is a rate constant. As shown in Table 1, the adsorption process of P on NZVI can be satisfactorily fitted by pseudo-second-order kinetic models with the correlation coefficients (R 2 > 0.99). As shown in Table 1, the observed rate of removal decreased with an increase in the initial P concentration, indicating that the increase of initial P concentration hampered the adsorption process. The limited reaction/adsorption sites restricted the adsorption and reduced the observed rate of removal. Furthermore, the oxygen amount in the adsorption system also considerably influenced the rate of removal, which basically experienced a remarkable decrease as the amount of oxygen decreased. This is displayed in Table 1. As discussed above, the oxygen notably affected the corrosion of Fe0, the further oxidation of Fe2+, and the formation of Fe3+, which played a crucial role in the removal of phosphate. The decrease in oxygen completely limited the redox process, resulting in the decrease of reaction/adsorption sites for phosphate. Therefore, the observed rate of removal decreased as the amount of oxygen decreased in this adsorption system, especially in a relatively low initial P concentration.

Table 1 Rate constants and correlation coefficients for removal of phosphate by NZVI with different initial P concentrations
Full size table

Collectively, high P concentration or a low amount of oxygen significantly decreased the observed rate of removal. However, the presence of dissolved oxygen was enough to remove all of the phosphate from the aqueous solution when the P concentration was 5 mg/L or less. It indicated that the NZVI had great potential for the removal of phosphate from the effluent, especially in a deep treatment.

3.4 Effect of Foreign Anions

The effect of various foreign anions (SO4 2−, S2O3 2−, Cl−, and NO3 −) on P removal in NZVI is shown in Fig. 8. It is observed that removal of P increased to 95.8 and 99.9%, with the increase of the concentration of SO4 2− and S2O3 2− from 0 to 0.2 mmol/L, respectively. This indicates that SO4 2− and S2O3 2− both significantly improved the removal of phosphate. It was reported that the presence of SO4 2− and S2O3 2− accelerated the corrosion of iron and promoted the formation of a secondary phase (such as green rust). This could be attributed to the complexation of a divalent anion (SO4 2− and S2O3 2−) with more iron ions (Gui and Devine 1994). The formation of corrosion products of NZVI generated the abundant reactive sites for the P removal.

Fig. 8
figure 8

Effect of foreign anions on P removal in NZVI (m/v = 1.0 g/L, C P  = 10 mg/L, pH 6.3)

Full size image

For Cl− and NO3 −, P removal displayed a fluctuation (Fig. 8). The enhanced removal of phosphate on NZVI was observed at concentrations of Cl− and NO3 − greater than 0.2 mmol/L. The role of Cl− in the corrosion of Fe0 seems to be controversial. Reardon et al. reported that an increase in Cl− concentration from 0.02 to 3 mmol/L decreased the corrosion rate of iron (Reardon 1995). However, Ruangchainikon et al. indicated that Cl− favored the formation of green rust (GR(Cl−)) (Ruangchainikom et al. 2006). Thus, it was speculated that the restriction of the corrosion rate of NZVI in the presence of Cl− was higher than the promotion of the formation of GR(Cl−) at a Cl− concentration of lower than 0.2 mmol/L (Barthélémy et al. 2012). However, the formation of green rust remarkably favored the increase of P removal at a Cl− concentration of more than 0.2 mmol/L. The effect of NO3 − on P removal had a similar tendency as that of Cl−. Fe0 was successfully utilized to decompose nitrate, and the performance was affected by coexistence ions (e.g., phosphate and sulfate) (Su and Puls 2004). Therefore, it was speculated that nitrate likely competed with phosphate for active sites. On the other hand, P removal increased to 3.1% with the NO3 − concentration increasing to 2.77% (Almeelbi and Bezbaruah 2012). The reaction between NO3 − and Fe0 should be accounted for, as the effective corrosion of Fe0 and the formation of Fe2+ provided more adsorption sites for phosphate. Collectively, the promotion of the corrosion of NZVI in the presence of NO3 − at a high concentration was higher than its restriction, which resulted in the slight increase of P removal.

3.5 Column Experiment

The effect of residence time on the removal of phosphate in the column experiment is displayed in Fig. 9. When the residence time was 80 min, the effluent concentration decreased to 0.04 mg/L from the influent concentration of 4.92 mg/L. When the residence time was adjusted to 50 min after running for 10 days, the effluent concentration had a slight increase to 0.16 mg/L. However, the effluent concentration dropped back to 0.04 mg/L when the residence time was changed into 80 min again, after running for 35 days. Afterwards, the effluent concentration varied between 0.03 and 0.11 mg/L, up to 200 days (and still running). It indicated that this flowing column significantly decreased the P concentration to less than 0.2 mg/L, as the residence time was 80 min despite a relatively high concentration of 5 mg/L. Therefore, it can be concluded that the prepared NZVI can be utilized as filler for the removal of phosphate from an aqueous solution. Furthermore, this NZVI has proven to be a promising material due to its low cost, abundant storage, and high efficiency. Figure 10 shows the XRD and XPS patterns of P-adsorbed NZVI, after running for 200 days. The samples were obtained from the bottom, middle, and top, respectively. As seen in Fig. 10a, the reflections of iron disappeared and were replaced by that of magnetite, indicating that the NZVI was oxidized into magnetite. The peak, which was centered at 133.5 eV with a width between 132.2 and 135.4 eV, was found and ascribed to the combination of Fe2+/Fe3+ and phosphate (Liu et al. 2013). The XPS results showed that the peak height of P-binding energy had an obvious increase from top to bottom, implying that more P was detected on the surface of the P-adsorbed NZVI than at the bottom of the column. Combined with the result of Fig. 9, it seems apparent that the bottom of the NZVI adsorbed a large amount of phosphate, whereas the top of the NZVI adsorbed almost no phosphate. The NZVI was oxidized into magnetite, which still had an exceptional impact on the adsorption of phosphate, despite running for 200 days.

Fig. 9
figure 9

Column experiments of the removal of phosphate in NZVI

Full size image
Fig. 10
figure 10

XRD patterns (a) and XPS patterns (b) of NZVI, before and after P adsorption. The top, middle, and bottom samples collected from the different sites of column experiments

Full size image

4 Conclusions

The batch experiment results indicated the NZVI was successfully synthesized by the reduction of natural limonite at 550 °C. The batch experiments indicated a slight effect of pH on the P removal in the acidic condition. The kinetic studies indicated that the removal of P from as-prepared NZVI can be satisfactorily fitted by the pseudo-second-order kinetic model. In addition, an increase in oxygen content remarkably enhanced P removal. The removal capacity of as-prepared NZVI was determined to be 16 mg/g for phosphorus at a pH of 6.3. These findings revealed that as-prepared NZVI can be used as an effective adsorbent in the removal of phosphate from an aqueous solution during environmental cleanup.