Fluoride removal from water using a magnesia-pullulan composite in a 1 continuous fixed-bed column 2

A magnesia-pullulan composite (MgOP) was previously shown to effectively remove fluoride from water. In the present study, a continuous fixed-bed column was used to examine the application of the composite at an industrial scale. The influencing parameters included bed mass (4.0, 6.0 and 8.0 g), influent flow rate (8, 16 and 32 mL/min), inlet fluoride concentration (5, 10 and 20 mg/L), reaction temperature (20, 30 and 40 °C), influent pH (4, 7 and 10) and other existing anions (HCO3-, SO42-, Cl- and NO3-), through which the breakthrough curves could be depicted for the experimental data analysis. The results indicated that MgOP is promising for fluoride removal with a defluoridation capacity of 16.6 mg/g at the bed mass of 6.0 g, influent flow rate of 16 mL/min and inlet fluoride concentration of 10 mg/L. The dynamics of the fluoride adsorption process were modeled using the Thomas and Yan models, in which the Yan model presented better predictions for the breakthrough curves than the Thomas model. Moreover, the concentration of magnesium in the effluent was monitored to determine Mg stability in the MgOP composite. Results indicated the effluent concentration of Mg2+ ions could be kept at a safe level. Calcination of fluoride-loaded MgOP effectively regenerated the material.

especially in developing countries (Mohan et al., 2017). 48 Various technologies for defluoridation based on the principle of adsorption, 49 precipitation-coagulation and the membrane separation process have been developed or are at 50 the development stage. Primarily lime and aluminum salts are utilized to remove fluoride 51 from water in the precipitation-coagulation process; the resulting fluoride-based precipitates 52 achieve satisfactory fluoride removal. However, the method may also produce an excessive 53 quantity of surplus sludge and thus increase the overall operation costs. Moreover, the use of 54 aluminum salts may release some aluminum ions to the water in the defluoridation process. 55 Consequently, human health can be detrimentally affected because the excessive aluminum  where resources or access to technology are limited (Jadhav et al., 2015). A major challenge 62 is fluoride removal from the reactor after treatment and further research is needed to solve 63 this problem. 64 Thus, adsorption is probably the most promising method for fluoride removal in 65 drinking water treatment because of its simple design and operation, high efficiency and low 66 costs compared with other methods (Kameda et al., 2015;Ye et al., 2016). More importantly, 67 adsorption is more likely to be utilized for fluoride removal in less developed countries, 68 where advanced wastewater treatment is unavailable.

87
In the previous study, pullulan (a biodegradable extracellular water-soluble microbial 88 polysaccharide) was found to have highly biocompatible and non-toxic properties; thus, the 89 material potentially could be employed as an adsorbent (Kang et al., 2011). More hydroxyl 90 groups were found in the pullulan saccharide unit than in the chitosan saccharide unit, for 91 which the number of potential sites for adsorption could be increased. In the previous study 92 (Kang et al., 2011), pullulan was spread on MgO to synthesize a magnesia-pullulan 93 composite (MgOP) and fluoride adsorption on the MgOP was explored in a batch system.

94
Compared to other similar adsorbents, the accessibility of the adsorbate-binding sites was 95 increased and the defluoridation capacity of MgOP was hence enhanced to 7.17 mg/g at the   The aim of this study was to evaluate MgOP performance with respect to fluoride 105 removal in a continuous fixed-bed column and provide guidance for design and operation of 106 the reactor. MgOP performance was evaluated under various operating parameters, including 107 bed mass, volumetric flow rate, influent fluoride concentrations, reaction temperature, inlet 108 pH and the presence of coexisting anions. Models developed by Thomas (1944)        Each dynamic adsorption experiment was conducted in duplicate using parallel columns.

174
The data variance derived from each of the duplicate column experiments was determined to 175 be negligible. Experimental data derived from the breakthrough curves were used to optimize 176 the operation and design parameters of the column. Furthermore, mathematical models were 177 employed to describe the experimental data and predict the column performance. The 178 Thomas and Yan models were used to describe the breakthrough curves. was used in the desorption study. In the current desorption study, 100 mL of various solvents 185 (i.e. deionized water, HCl, NaCl, C6H8O7, Na3C6H5O7, NaOH, and Na2CO3 solutions) were Values for parameters in the two models were determined by analysing the fluoride 243 breakthrough curves and are given in Table 1.      Regression coefficients presented in Table 1 show that the Yan model described the  The effects on fluoride adsorption of various volumetric flow rates from 8 to 32 mL/min 295 were conducted at a fixed inlet fluoride concentration of 10 mg/L and bed mass of 6.0 g.

296
Increasing flow rate decreased the volume of effluent treated (Fig. 2). This is because the      Table 1 show that values for the kinetic rate constant in both the Yan model  Correlation coefficients presented in Table 1 show that the Yan model predicted fluoride  The regression coefficients (Table 1)