1 Introduction

Excessive application of fertilisers and chemicals in arable soils, intensive livestock operations and disposal of high volumes of untreated wastewater and solids have resulted in an increasing number of surface water, groundwater and soil contamination problems (Dou et al. 2009; Shin et al. 2004; Valsami-Jones 2001). High levels of P, along with N, due to farm effluents and fertilisers application to soils, pose a significant threat to many water bodies because of the eutrophication risk (Daniel et al. 1998). Although P is an essential element for plant growth and necessary to maintain profitable crop yield, it also increases the biological productivity and thus it can cause adverse environmental effects on surface waters (Correll 1998). Water-quality criteria for P have been established and translated to legislated limits, below which environmental risks are unlikely or risk is within an acceptable range (Council Directive 91/271/EC 1991; USEPA 1986).

Based on the above, the removal of P from waters is of prime importance, especially for aqua sources near urban and agricultural areas. Various P removal methods from contaminated water resources have been reported in the literature, including physical such as membrane technologies, biological such as P bacterial removal or plant and algae uptake (Drizo et al. 1999) and chemical methods, such as precipitation and adsorption. Each of the above methods has advantages and disadvantages; however, chemical adsorption techniques, due to their simplicity and low cost, are attracting more attention. Different materials can be used for removing P from waters, including fly ash (Agyei et al. 2002; Drizo et al. 1999; Grubb et al. 2000), slag (Agyei et al. 2002), gas concrete (Οguz et al. 2003), Fe and Al oxides, gibbsite, dolomite (Hesterberg and Beauchemin 2000), aluminium oxide-hydroxide (Tanada et al. 2002), calcined layered double hydroxides (Hosni and Srasra 2010), hematite and bentonite (Dimirkou et al. 2002), calcium-modified bentonite (Khadhraoui et al. 2002), bauxite, shale, limestone and clay aggregates (Drizo et al. 1999). However, many of these materials exhibited insufficient adsorption and regeneration capacities as well as poor selectivity and limited surface area (Saad et al. 2008). Therefore, it is worthy studying other materials, with high surface area, open pore structure and high adsorption capacity, as absorbents for P.

Zeolite is a porous alumino-silicate mineral, with high adsorption capacity for cations because of its high negative charge (Bowman et al. 2000). Zeolites have been widely used to improve the quality of water systems through removing inorganic cations, trace and radioactive elements (Faghihian et al. 1999; Mercer and Ames 1978; Misaelides et al. 1995; Rajec et al. 1998). Surfactant-modified zeolite (SMZ), i.e. zeolite with positive surface charge, because of its surface modification with a cationic surfactant substance, may be useful in removing P anionic species from water through adsorption. The fact that zeolite is highly available from natural sources at low cost, in certain countries, combined with the simple techniques that can be used for its surface modification, suggests that this material could be an alternative to other adsorbents for remediating P enriched waters. Several studies have indicated that SMZ is an effective adsorbent for many anions, such as ΝΟ3 −, SO4 2− (Li et al. 1998), CrO4 2− (Haggerty and Bowman 1994; Li 2004), Η2AsO4 − and HAsO4 2− (Li et al. 2007), ClO4 − (Zhang et al. 2007), HSeO3 − and SeO4 2− (Frenay et al. 2000). There are few studies, however, that have investigated P adsorption on SMZ (Allred 2010; Dao 2003; Vujakovic et al. 2000, 2003; Widiastuti et al. 2008).

The objectives of this study were to further evaluate certain factors that could affect P adsorption on a SMZ, in the perspective of using the SMZ for P removal from waters enriched with P. Specifically, the effect of: (a) temperature (exothermic or endothermic character of adsorption reaction, (b) pH (as a factor controlling P speciation) and (c) NO3 − (as competing anion, present at two concentrations typical for agricultural subsurface drainage systems), on P adsorption on the SMZ were studied. In addition, the desorbability of externally added P from the SMZ was investigated.

2 Materials and Methods

2.1 Surface Modification of the Zeolite

The natural zeolite used was clinoptilolite with the chemical formula (Ca1.5K1.4Mg0.6Na0.5) (Al6.2Si29.8O72)·20H2O and its chemical composition, as it was determined by the Department of Mineralogy-Petrology-Economic Geology of the Aristotle University of Thessaloniki using TXRD analysis (Denaix et al. 1999), is presented in Table 1. The zeolite was sieved through a 2 mm stainless steel screen sieve, and this material was used for all analyses and P adsorption-desorption experiments, after the modification of its surface charge. The total and the external cation exchange capacity of the zeolite determined using the method of Ming and Dixon (1987) and found to be 226 and 10 cmolc kg−1, respectively (Table 1).

Table 1 Certain chemical properties of the natural zeolite
Full size table

Modification of the zeolite’s surface charge was accomplished using hexadecyltrimethylammonium bromide (C19H42BrN or HDTA-Br), according to Bowman et al. (2000). Specifically, a quantity of zeolite sufficient to conduct all the experiments was equilibrated with 0.05 M HDTMA-Br, for 24 h, at 25 ± 1 °C, with intermittent shaking. The solution of 0.05 M HDTMA-Br was added to the zeolite at a ratio 4/1 (v/w) and contained an amount of HDTMA-Br equivalent to the double external cation exchange capacity of the zeolite. At the end of the equilibration period, the suspension of the SMZ and HDTMA-Br was centrifuged (at 7,000 rpm, for 5 min) washed three times with distilled water and air-dried. This material was used to all subsequent experiments.

2.2 Phosphorus Adsorption Experiments

To establish the optimum equilibration solution/adsorbent ratio and time for P adsorption on the SMZ, two preliminary batch adsorption experiments were conducted (data not shown). For the first parameter, a tentative equilibration time of 24 h was used, the solution/adsorbent ratios tested were 5/1, 10/1, 20/1 and 50/1 (v/w). From the four ratios, the highest percentages of P adsorbed on the SMZ were obtained for the ratios 5/1 and 10/1 (v/w) and finally the ratio 10/1 (v/w) was selected for practical reasons, in respect to laboratory work mainly. Using the solution/adsorbent ratio of 10/1 (v/w), the equilibration time was selected by testing different periods of equilibration time (1, 2, 4, 6, 8, 10 and 24 h) and found to be 1 h, as it was found also by Vujakovic et al. (2000, 2003). Consequently, the equilibration solution/adsorbent ratio and time used in all subsequent P adsorption experiments on the SMZ were 10/1 (v/w) and 1 h, respectively.

Phosphorus adsorption data were obtained by a slight modification of the technique used by Li et al. (2007). Four grammes sub-samples of the SMZ, in three replicates, were placed in 50 mL polypropylene centrifuge tubes with 40 mL of deionized water, containing varying concentrations of P (0, 0.5, 1, 2, 5, 8, 10, 12, 15, 20, 30, 40 and 50 mg L−1) as KH2PO4. The suspensions were equilibrated for 1 h at 25 ± 1 °C with intermittent shaking. After equilibration the tubes were centrifuged and the supernatant liquid was filtered and analysed for P, employing the ascorbic acid method (Murphy and Riley 1962), using a UV–vis Spectrophotometer (Lambda 5, Perkin-Elmer). The amount of adsorbed P was calculated by the difference between the amount added and that found in solution at equilibrium. Phosphorus adsorption data were obtained at: (a) three temperatures, namely, 25 ± 1 °C, 35 ± 1 °C (data not shown) and 65 ± 1 °C, (b) two pH values of the external solution, namely, 4.2 ± 0.1 (adjusted by 0.5 M HCl) and 10.4 ± 0.1 (adjusted by 0.5 M NaOH), at 25 ± 1 °C and (c) two NO3 − concentrations in the external solution, namely 25 and 50 mg NO3 − L−1, as ΚΝΟ3, at 25 ± 1 °C. In all cases, pH in the equilibrium solution was determined.

The following nonlinear equations were fitted to P adsorption data by nonlinear regression, using the Levenberg–Marquardt algorithm and the statistical package SPSS, version 19:

$$ \mathrm{Langmuir}\ \mathrm{equation}:x = {KMC \left/ {{\left( {1 + KC} \right)}} \right.} $$
(1)
$$ \mathrm{Freundlich}\ \mathrm{equation}:x=k{C^n} $$
(2)

where x is the amount of P adsorbed (in milligrammes per kilogramme), C is Ρ concentration in the equilibrium solution (in milligrammes per litre), K is a parameter related to the affinity of the adsorbent to adsorbate (in litres per milligramme), Μ is the maximum P adsorption capacity (in milligrammes per kilogramme), n is a parameter that can be considered as a measure of adsorption intensity and k is a parameter corresponding to the amount of P adsorbed when C is equal to 1.

To evaluate the goodness of fit of each equation, the standard errors (SE) of the estimate were used along with F tests. In addition, the standard errors of the estimated parameters along with t tests were used. Appropriate comparisons among the estimated parameters were performed using t test.

2.3 Phosphorus Desorption Experiments

Sub-samples of the SMZ were spiked with P in the form of a solution of K2HPO4, at rates equal to 200 and 50 mg P kg−1 of the substance; these concentrations were equal to the maximum and a quarter of the maximum P adsorption capacity, respectively, according to the Langmuir equation (see next section). After P addition, the samples were left to equilibrate for 10 days with periodical wetting and drying, then air-dried and used to the two following (one short-term and the other long-term) batch desorption experiments. In both experiments, a slight modification of the technique reported by Li (2003) was used.

For the short-term desorption experiment of P from the SMZ, 4 g of the P-spiked sub-samples of the SMZ, in three replications, were placed in 50 mL polypropylene centrifuge tubes with 40 mL of deionised water and equilibrated for different periods, namely, 0.25, 0.5, 1, 2, 4, 6, 8, 10 and 12 h, at 25 ± 1 °C, by intermittent shaking. At the end of each equilibration period, three tubes were centrifuged and the supernatant liquid was filtered and analysed for P, employing the ascorbic acid method. The same desorption experiment was conducted with 0.01 M KCl as external solution instead of deionised water (data not shown). For the long-term desorption experiment of P from the SMZ, 4 g of the P-spiked sub-samples of the SMZ, in three replications, were placed in 50 mL polypropylene centrifuge tubes with 40 mL of deionised water and equilibrated for 24 h, at 25 ± 1 °C, by intermittent shaking. After equilibration the tubes were centrifuged, 25-mL aliquot of the supernatant solution was removed and analysed for P, employing the ascorbic acid method. The 25-mL aliquot removed was replaced with 25 mL of deionised water; the mixtures were re-suspended and equilibrated for 24 h, at 25 ± 1 °C. This procedure was repeated until P concentration in the supernatant liquid reached undetectable levels (< 0.1 mg L−1), resulting in a total of four desorption steps, i.e. a maximum of 96 h desorption time. The amount of P desorbed at each step was corrected for the amount of P transferred from the previous step.

Multiway factorial analysis of variance was performed to the desorption data to evaluate main effects and interactions and Duncan’s multiple range test was used for mean comparisons, using the statistical package SPSS, version 19.

3 Results and Discussion

3.1 Phosphorus Adsorption on the SMZ

Phosphorus adsorption isotherm obtained at 25 ± 1 °C was L-shaped (Fig. 1) and according to Sparks (1984), such adsorption behaviour, could be explained by the strong affinity of the SMZ for P, at low P concentrations, which then decreases as P concentration increases. Adsorption data conformed satisfactorily to the nonlinear forms of both the Langmuir and Freundlich equations (Fig. 1). However, the Langmuir equation was slightly superior as judged by data fit (presented in Fig. 1), but also by the lower SE of the estimate and the higher value of the F test which were used as the criteria of goodness of fit (Table 2). Consequently, the Langmuir equation was selected to evaluate adsorption data obtained from all adsorption experiments. In both cases, the estimated values of the parameters, K and M of the Langmuir equation but also of the parameters k and n of the Freundlich equation, were significant, as evidenced by their low standard errors (Table 2) but also by the p values of the t tests, which were < 0.001. According to literature, the nonlinear Langmuir equation satisfactorily described the adsorption of different anions onto SMZs, such as CrO4 2− (Bowman et al. 2000; Li et al. 1998), NO3 − (Li 2003), H2AsO4 − , HAsO4 2− (Li et al. 2007) and ClO4 − (Zhang et al. 2007). In all cases of this and also in the rest of the adsorption experiments where un-buffered external solutions were used, pH of the equilibrium solution was 6.9 ± 0.3.

Fig. 1
figure 1

Isotherms (experimental, Langmuir and Freundlich) of P adsorption on the SMZ, at 25 ± 1 °C

Full size image
Table 2 The standard errors of the estimate (SE), the values of the F test and the respective p values of the equations Langmuir and Freundlich and their estimated parameters, at 25 ± 1 °C
Full size table

Comparing the SMZs P adsorption Langmuir M to other adsorbents, certain other materials exhibited comparable adsorbing capacity, such as kaolinite–goethite systems (Ioannou et al. 2001) and some kaolinite clays (Sei et al. 2002). However, other showed higher adsorbing capacity, such as fly ash, synthetic zeolite by-product (Dao 2003), ammonium-functionalised mesoporous silica materials (Saad et al. 2008), natural and synthetic palygorskite (Hengpeng et al. 2006) and calcined-layered double hydroxides (Hosni and Srasra 2010).

Conducting the adsorption experiments at 25 ± 1 and 35 ± 1 °C, it was revealed that no safe conclusions concerning the exothermic or endothermic character of the adsorption reaction could be drawn. For this reason, data obtained at 35 ± 1 °C are not shown and P adsorption on the SMZ at 65 ± 1 °C was also studied. Phosphorus adsorption data at the two temperatures (25 ± 1 and 65 ± 1 °C) showed that the SMZ removed more than 80 % of the added P, when P concentration in the external solution was less than 20 mg L−1. The Langmuir equation was adequately fitted to the adsorption data obtained at 65 ± 1 °C (SE = 17, F = 42, p < 0.001; p values of the t tests were <0.001 for M and < 0.01 for K). Plotting P adsorption isotherms at 25 ± 1 and 65 ± 1 °C revealed the exothermic character of the reaction since the isotherm at the lower temperature was located above the isotherm at the higher temperature; this trend was more pronounced at higher P concentrations in the equilibrium solution (Fig. 2). The exothermic character of the reaction was reflected in the estimated Langmuir adsorption M which was significantly higher at 25 ± 1 than 65 ± 1 °C (Table 3). Reported studies relevant to the endothermic or exothermic nature of adsorption of different anions on SMZs are rare. However, Sullivan et al. (2003) reported that As adsorption on a SMZ was an endothermic reaction since maximum As adsorption declined when temperature decreased from 25 to 15 °C.

Fig. 2
figure 2

Isotherms of P adsorption on the SMZ, at the two temperatures

Full size image
Table 3 Parameters of the Langmuir equation, at the two temperatures
Full size table

The SMZ removed P effectively regardless of the pH of the equilibrium solution (4.2 ± 0.1 and 10.4 ± 0.1) (Fig. 3). More than 70 and 40 % of the added P was removed when P concentrations in the external solution exceeded 15 and 20 mg L−1, respectively, and more than 80 % of the added P was removed when P concentrations in the external solution was 1–5 mg L−1 Adsorption data obtained at both pH values conformed satisfactorily to the Langmuir equation (for pH = 4.2 ± 0.1 SE = 9.5, F = 677, p < 0.001 and for pH = 10.4 ± 0.1 SE = 25, F = 123, p < 0.001; p values of the t tests were <0.001 for M in both cases and for K at pH = 4.2 ± 0.1 and <0.025 for K at pH = 10.4 ± 0.1). As pH of the equilibrium solution rose from the acidic to alkaline range, P adsorption on the SMZ also increased. This increase was reflected in the estimated Langmuir parameters, the capacity parameter M and the affinity parameter K, which were significantly increased with the pH rise (Table 4).

Fig. 3
figure 3

Isotherms of P adsorption on the SMZ, at the two pH values of the equilibrium solution

Full size image
Table 4 Parameters of the Langmuir equation, at the two pH values of the equilibrium solution
Full size table

Soluble inorganic P may be present in four different chemical forms, depending on pH. When solution pH is less than 2, P exists as Η3ΡΟ4, between pH 3 and 6.5 the dominant chemical form is Η2ΡΟ4 −, HPO4 2− dominates between 7.5 and 12 and when pH exceeds 12, P exists as ΡΟ4 3− (Tisdale et al. 1990). Consequently, when the pH of the equilibrium solution was acidic (4.2 ± 0.1), the univalent form of Η2ΡΟ4 − was dominant, whereas at the alkaline pH (10.4 ± 0.1) P was mostly in the divalent form of ΗΡΟ4 2−. According to Li and Bowman (1997), the anion adsorption mechanism on the SMZ (with HDTMA-Br) is anion exchange. Specifically, the anions are exchanged for Br− originating from the HDTMA-Br on the surface of the zeolite and held by electrostatic bond. Furthermore, the electrostatic bond is stronger for the polyvalent anions than the univalent and for anions with a smaller hydrated ionic radius (Schroeder 1984). Based on this and taking into account that the univalent and divalent P anions have similar hydrated ionic radii estimated at 0.4 nm (Dean 1979), the increase in P adsorption in response to a rise in pH could be due to a stronger electrostatic bond between the SMZ and the divalent P anions than the univalent P anions.

Phosphorus adsorption on the SMZ in the presence of NO3 − was conformed satisfactorily to the Langmuir equation (for NO3 − = 25 mg L−1 SE = 18, F = 77, p < 0.001 and for NO3 − = 50 mg L−1 SE = 9.6, F = 247, p < 0.001; p values of the t tests were < 0.001 for M in both cases, <0.05 for K at NO3 − = 25 mg L−1 and <0.005 for K at NO3 − = 50 mg L−1) and reduced (Fig. 4). However, the reduction was significant only when there was 50 mg L−1 of NO3 − in the external solution, i.e. the estimated Langmuir parameters, M and K were significantly deceased with the NO3 − concentration increase from 0 to 50 mg L−1 (Table 5). In all cases, the amount of P removed by the SMZ was substantial. More than 50 and 80 % of the added P was removed when the P concentration in the external solution was less than 30 and 5 mg L−1, respectively.

Fig. 4
figure 4

Isotherms of P adsorption on the SMZ, at the two ΝΟ3 − concentrations in the external solution adsorption isotherm at 0 mg L−1 NO3 − is the same with that presented in Fig. 1, for 25 ± 1 °C

Full size image
Table 5 Parameters of the Langmuir equation, at the two ΝΟ3 − concentrations in the external solution
Full size table

According to Schroeder (1984), adsorbent’s affinity for anions depends mainly on the anion’s valence and size. The higher the valence and the smaller the hydrated ionic radius, the stronger is the adsorption. Phosphorus anions commonly found in a pH, like the pH of the equilibrium solution, 6.9 ± 0.3 (see first paragraph of this section), are Η2ΡΟ4 −, HPO4 2−. Although HPO4 2− have valence higher than NO3 −, both P anionic species have hydrated ionic radius (0.4 nm) larger than NO3 − (0.3 nm) (Dean 1979). Consequently, the reduction of P adsorption on the SMZ in the presence of NO3 − may be due to the size difference of the specific anions.

Contradictory results are reported in the literature with respect to the adsorption of competing anions on SMZs. Specifically, Li and Bowman (1997) reported that CrO4 2− adsorption was decreased in the presence of ΝΟ3 −, ΗCO3 −, SO4 2− and HPO4 2−, whereas Zhang et al. (2007), reported that, from the anions studied, Cl−, ΟΗ−, NO3 −, CO3 2− and SO4 2−, only NO3 − reduced ClO4 − adsorption on the SMZ, and only when the concentration of ClO4 − in the external solution was low.

3.2 Phosphorus Desorption from the SMZ

The data obtained from the short-term desorption experiments, performed using deionised water or 0.01 M KCl as external solution, showed that P desorption from the SMZ within 12 h, was not affected by the kind of the external solution. For this reason, only the data obtained using deionised water are presented.

The results of both (short-term and long-term) desorption experiments showed that P desorption from the SMZ was affected only by time and not by the initial P concentration added to the SMZ (Table 6; Fig. 5). As it can be seen from the desorption data, especially from Table 6, P desorption from the SMZ was a slow reaction and this is in agreement with the findings of Bansiwal et al. (2006) who studied P release from a SMZ by leaching experiments. At both rates of P addition, the greatest part of the total desorbed P was released within 24 h and found to be 37 and 51 % of the total desorbed P, for P addition at 50 and 200 mg kg−1, respectively. After the 24 h, the cumulative amounts of desorbed P continued to increase and at the end of the last desorption step (96 h) reached the 7.5 mg kg−1, in both cases of P addition. Consequently, the SMZ retained a large amount of added P.

Table 6 Amounts of desorbed P from the SMZ, after P addition at 50 and 200 mg kg−1
Full size table
Fig. 5
figure 5

Cumulative amounts (mean and standard deviation) of desorbed P from the SMZ, after addition of 50 and 200 mg P kg−1 of SMZ. Each step represents a 24-h desorption interval. Values followed by different letter are significantly different, at p ≤ 0.05, using the Duncan’s multiple range test

Full size image

At both rates of P addition, P concentration in the equilibrium solution ranged between 0.1 and 0.4 mg L−1. The above outcome is an indication that in the case of using SMZ for reducing P levels from P enriched waters, after P adsorption on the SMZ, the concentration in waters will be less than 2 mg L−1, above which environmental risks are likely (Council Directive 91/271/EC 1991). Furthermore, according to Tisdale et al. (1990) who reported that P concentrations in the soil solution between 0.2 and 0.3 mg L−1 are adequate for a variety of crops, it seems that if P-laden SMZ is applied to soils after its use for removing P from waters could supply soil solution and plants with probably sufficient P. This is also supported by the findings of Bansiwal et al. (2006) who reported that SMZ has a great potential as a fertiliser carrier for slow release of P.

4 Conclusions

In conclusion, the SMZ exhibited high adsorption capacity for P, which was comparable to certain other materials, used as adsorbents for P. The P adsorption capacity of the SMZ increased with pH increase and although decreased with temperature increase and in the presence of a relative high NO3 − concentration, remained appreciable in all cases. In addition, the SMZ retained the greatest amount of added P and the P desorbed concentrations ranged at levels below than those above which environmental risks are likely. Taking into account zeolite’s high availability, its low cost and the simplicity of the techniques that can be used for its surface modification, it was concluded that SMZ can be a promising adsorbent for cleaning up waters enriched with P. However, further research, including additional laboratory experiments, economical analysis and especially large-scale experiments, is needed.