Research on water defluoridation with composite filters of bauxite, gypsum and magnesite was carried out at Ngurdoto Research Station to determine capacity, nature of fluoride-sorbent interaction and changes in water quality. The objective was to identify an optimum composite filter that would not alter the water quality beyond World Health Organisation recommendations. The three materials were characterized through X-Ray Fluorescence. The major components were Al 2 O 3 (30.33%) for bauxite, CaO (28.09%) for gypsum and MgO (34.57%) for magnesite. The materials were calcined at temperatures of 150, 200, 250 and 300 o C. Six composite filters were prepared for each calcine temperature of bauxite:gypsum:magnesite in the ratios 1:2:3, 1:3:2, 2:1:3, 2:3:1, 3:1:2 and 3:2:1. The highest capacity, 11.77 mg F/g, was obtained with the 3:1:2-200 o C calcine. The optimum filter was however the 1:2:3-200 o C calcine with capacity of 10.19 mg F/g. This filter did not alter the water quality beyond the considered WHO limits. The fluoride-sorbent interaction could best be described by SO , however a composite filter of these materials is plausible.

Research was carried out at Ngurdoto research station in Tanzania were within the prescribed limits. Sorption behavior followed strongly to Langmuir isotherm, except for the 450°C calcined samples for which the Temkin isotherm behavior was pronounced. Despite the limitations of high residual sulphates and iron, a composite filter of bauxite, gypsum and magnesite was shown to be workable.

One of the parameters that impact on fluoride sorption capacities of materials in water defluoridation is initial concentration. Water from various sources will have different fluoride concentrations; as such the employment of a particular media in defluoridation requires information on how the media will perform in different concentrations. Consequently, research was carried out to determine how initial fluoride concentrations in raw water affect capacity in defluoridation with bauxite, gypsum, magnesite and their composite, and to investigate the possibility of predicting loading capacity of their composite filter through initial fluoride concentration to optimize their application in defluoridation. The results showed that sorption capacities increased with increase in initial concentrations. However, higher starting concentrations resulted in larger residual fluoride concentrations; as such the benefit of large sorption capacities obtained was overshadowed. Polynomial relations of capacity (Cs, mg/g) and initial concentration (Co, mg/l), were obtained for bauxite and gypsum. Magnesite obtained a logarithmic relation for Cs and Co. A power relation was obtained between Cs and Co for their composite, Cs = 0.0328C1.20160. Approximation of capacities of this composite from initial fluoride concentrations was feasible. Water defluoridation has become important as a result of dental and skeletal fluorosis.

Water defluoridation in a fixed bed column may be optimized through choice of flow rate. This, however, affects breakthrough characteristics. In any fixed bed system an understanding of flow rate and breakthrough interactions is therefore important to guide choice of design velocity. Consequently interactions between flow rate and break through characteristics were determined in a fixed bed column in defluoridation with a composite filter of bauxite, gypsum and magnesite. This was an attempt to optimize defluoridation with the three materials. Lower flow rates obtained longer service times but higher residual concentrations of sulphates and chlorides. Higher flow rates registered lower residual sulphate and chloride concentrations but obtained lower service times. There were no significant differences in residual concentrations of aluminium, iron, calcium and magnesium among the different flow rate regimes. Adsorbent exhaustion rates and critical bed depths increased with increase in flow rates, indicating that lower flow rates offered better technical system performance. Critical bed depths ranged from 5.23 to 10.89 Cm for flow rate range of 0.68 to 1.25 ml/s. operating lines were polynomic with generic form 3 2 2 1 C t C t C     , where  is adsorbent exhaustion rate and t is empty bed residence time. Choice of flow rates, in this system requires a balance between service-time optimisation and water quality.