Nitrate, phosphate and fluoride removal from water using adsorption process
- Publication Type:
- Issue Date:
The wastewater treatment industry has identified the discharge of inorganic anions such as nitrate, phosphate and fluoride into waterways as a risk to the natural environment and human health. Of the various methods for removing these anions, adsorption/ion exchange methods are promising because they are simple, efficient, economical and result in less sludge production and therefore experience minimal disposal problems. Of the four ion exchange resins tested (Purolite A520E, Purolite A500PS, Purolite FerrIX A33E and Dowex 21k XLT), Purolite A520E emerged as the most efficient adsorbent, having a high adsorption capacity for the removal of nitrate from water. Purolite A520E proved to be the most efficient at removing nitrate (84%) followed by Dowex 21k (81%), and Purolite A500PS (75%) within 120 min. The lowest removal efficiency was found for Purolite FerrIX A33E (48%).The Langmuir adsorption capacity was 33 mg N/g for this resin and the highest column adsorption capacity was 21.3 mg N/g at an inlet concentration of 20 mg N/L, 12 cm bed height and 2.5 m/h filtration velocity. The kinetics of nitrate adsorption by Purolite A520E in the batch study was satisfactorily described using pseudo-first-order, pseudo-second-order and HSDM models. The experimental and Thomas models predicted breakthrough adsorption capacities (12.0-13.5 mg N/g and 8.2-9.7 mg N/g, respectively) agreed fairly well. The presence of other co-ions such as SO₄²⁻, F⁻, C1⁻ and PO₄³⁻ did not compete much with nitrate at equal concentrations for adsorption on Purolite A520E; only high concentrations reduced the effectiveness of this resin’s ability to adsorb nitrate. Moreover, at all nitrate to phosphate ratios in solution, the ratio of nitrate to phosphate adsorbed was higher for Purolite A520E which suggested that the nitrate selectivity for adsorption was higher than phosphate. It was found that solution pH had little effect on nitrate adsorption in the pH range 5-8. Moreover, Purolite A520E was regenerated and used at least three times without significantly reducing the adsorption capacity. Of the six adsorbents tested in a batch study (Purolite A520E, Purolite A500PS, Purolite FerrIX A33E, Dowex 21k XLT, HFO (iron (iii) oxide HFeO₂) and Zirconium (IV) hydroxide (H₄O₄Zr)), Purolite FerrIX A33E had the highest phosphate removal efficiency (98%) followed by Dowex 21k (91%), Zirconium (IV) hydroxide (89%), Purolite A500PS (75%) and Purolite A520E (69%) within 240 min. HFO was found to have the least removal efficiency (36%). The batch adsorption isotherm data for Purolite FerrIX A33E was satisfactorily explained using the Langmuir, Freundlich and Tempkin isotherm models. Meanwhile the kinetic adsorption data fitted reasonably well to the pseudo-second-order, Elovich and intraparticle diffusion models. The Langmuir maximum adsorption capacity of Purolite FerrIX A33E was 48 mg P/g which constituted one of the highest phosphorus adsorption capacities reported in the literature. Three empirical models - Bohart-Adams, Thomas and Yoon-Nelson - and a numerical model based on the advection-dispersion equation satisfactorily described phosphate adsorption behaviour in a fixed-bed column containing Purolite FerrIX A33E. The phosphate adsorption capacity of the column was estimated by: firstly, the Thomas model to be 22.7 mg P/g; and secondly, the breakthrough curve calculation to be 16.3 mg/g at the inlet concentration of 30 mg P/L, 12 cm bed height and 10 m/h filtration velocity. pH had little effect on phosphate adsorption by Purolite FerrIX in the pH range 4 – 10. The decreasing order of the anions’ competition with phosphate was as follows: SO₄²⁻ > C1⁻ > NO₃⁻ > F⁻. The Purolite FerrIX A33E resin was regenerated by leaching the adsorbed phosphate with 1 M NaOH solution and reused at least four times without significantly reducing the adsorption capacity. This phosphorus desorbed was recovered as struvite by adding magnesium chloride (MgC1₂.6H₂O) and ammonium sulphate (NH₄)₂SO₄ at the molar ratio of phosphate, ammonium and magnesium of 1:1:1. Calcium hydroxide (Ca (OH) ₂) was added to the desorbed solution to recover phosphorus as hydroxyapatite at the molar ratio of phosphate and calcium of 1:0.5 and 1:2. The XRD and FTIR analyses confirmed the recovered compounds were struvite and hydroxyapatite. These compounds’ phosphorous content was 12-14% which was similar to the phosphorus content of struvite and hydroxyapatite. Hydrous ferric oxide (HFO) had the highest fluoride adsorption capacity of seven adsorbents tested (Purolite A520E, Purolite A500PS, Purolite FerrIX A33E, Dowex 21k, HFO (iron (iii) oxide HFeO₂), Zirconium (IV) hydroxide (H₄O₄Zr) and α-Alumina (A1₂O₃)). Among the seven adsorbents, HFO had the highest fluoride removal efficiency (56%) followed by Dowex 21k (52%), Zirconium (IV) hydroxide (38%), Purolite A502PS (35%), Purolite FerrIX A33E (29%) and Purolite A520E (25%) within 120 min. The lowest removal efficiency was found for α-Alumina (4%). The batch adsorption of fluoride on HFO was satisfactorily explained using both the Langmuir and Freundlich isotherms while the column adsorption data fitted reasonably well to the Thomas model. However, by using an artificial neural network approach the model’s capability did improve. The Langmuir maximum adsorption capacity at pH 6.5 was 6.71 mg F/g and the highest column breakthrough adsorption capacity was 7.06 mg F/g at the inlet concentration of 30 mg F/L, 12 cm bed height, pH 5 and 2.5 m/h filtration velocity. The adsorption capacity predicted by the Thomas model was also highest (12.7 mg F/g) for these experimental conditions. The kinetic data concerning the fluoride adsorption on HFO was better described with the pseudo-second-order model compared to the pseudo-first-order model. The HFO was regenerated by leaching the adsorbed fluoride with 0.1 M NaOH solution and reused for at least three times. However, the fluoride adsorption capacity declined with repeated use.
Please use this identifier to cite or link to this item: