Fluorides are found in the waste discharges from process streams in a number of industries. Significant amounts of fluoride come from the following: glass manufacturers, electroplating operations, steel and aluminum, pesticides and fertilizer, groundwater and the semiconductor industry. The original fluoride effluent levels can vary over a large range, and restrictions on final effluent level depend on the place of disposal. When there is any risk of fluoride seeping back to water supplies, a limitation of about one ppm fluoride is normal. Apart from the treatment of industrial waste streams, the other main application of fluoride removal is the treatment of municipal water supplies to reduce the fluoride content to 1 ppm or less.
High levels of fluoride are generally reduced by precipitation of CaF2 with lime. However, the solubility of CaF2 is such that ~8 ppm fluoride remains in the distilled water, and in industrial water, residual fluoride can be considerably higher. Since pollution control boards are requiring effluent limits of 1 ppm fluoride in many cases, these saturated CaF2 solutions must undergo further treatment.
Reports in the literature suggest that activated alumina is the best way to reduce fluoride levels down to below 1 ppm. If the initial fluoride content exceeds 15 – 20 ppm, however, a prior treatment with lime to reduce the fluoride and prevent rapid saturation of the alumina will be economically advantageous.
Several laboratory studies have been reported. They are all in agreement that fluoride can be removed below one ppm by adsorption on alumina. Some results are not quantitative, and others are not in agreement over the amount of fluoride that can be removed, or the best method of regeneration, etc. A comprehensive literature review of fluoride removal has been published in German. A more recent article summarizes fluoride removal technology; this includes cost estimate data for the use of alumina, but it is based on laboratory scale work.
A pilot scale operation using alumina is described by Zolotva and is reported as operating successfully. While a number of plant operations are referred to in the literature, most are not described in detail. An exception is a water purification plant in Kansas. The size of the plant, mode of operation and overall cost of treatment are discussed in considerable detail.
Another benefit of using activated alumina in water treatment is its arsenic removal capability. Current federal limitation on arsenic in drinking water is 10 ppb. Such levels are reported to have been easily achieved using alumina. This paper will focus on the regenerable alumina, although, Tramfloc, Inc. has offered the disposable grade since 1998.
It is interesting to note the type of physical properties of the activated alumina are never discussed in these articles, although they may have a significant effect upon the fluoride or arsenic removal performance. Other factors likely to have an effect upon alumina performance are flow rate, other ions in the water to be treated, pH of the water, and the method and conditions of regeneration.
This bulletin summarizes the performance of Tramfloc’s Activated Alumina in removing fluoride from aqueous streams. The variables mentioned above are considered to some degree, but obviously, not all aspects of fluoride from different streams have been considered. The information is intended to give a general idea of the fluoride removal capabilities of alumina. For all but the simplest systems, it is recommended that a small scale test is carried out with the particular stream to be treated.
The granular activated alumina used in the evaluations discussed in this report is 14 X 28 grind activated alumina (AA). This is a transition alumina with a high surface area (>300 m2/g), which makes it especially suited for adsorption of certain species. It is a fairly high purity alumina with a pore volume of ~0.5 cc/gm and a bulk density of 46 lbs./ft3. The 14 X 28 S product is a similarly activated alumina but in a spherical form. The granular alumina has the advantage of being available in smaller sizes, making the internal active surface of the alumina more readily available. However, the spherical alumina has the advantage of a lower pressure drop in packed bed (i.e., downflow) systems. The fluoride removal data described were obtained with laboratory scale experiments using 14 x 28 mesh size granular alumina in a packed column.
The efficiency of the activated alumina for adsorbing fluoride is generally poor on the first adsorption cycle unless the alumina is pretreated. A pretreatment which involves allowing a dilute aluminum sulfate solution (~29 g Al2 (SO4)3 •18 H2O per liter) to remain in contrast with the alumina for 1 hour is found to be particularly satisfactory. The dramatic improvement of treated over untreated alumina is illustrated in Figures 1 and 2. This pretreatment is very important if the alumina is being used on a once-through basis or where good performance is necessary on the first cycle. In a cyclic system, the regeneration procedures described later will activate the surface for subsequent adsorption cycles.
The effectiveness of alumina in removing fluoride from aqueous NaF is shown in Figure 1. Curves are given for initial fluoride concentrations of 10 and 20 ppm. The fluoride level is readily reduced to ~0.2 ppm in both cases. The fluoride capacity of the alumina is slightly greater for the 10 ppm level, but in both cases, it is around 1.5%. These data are for alumina pretreated with aluminum sulfate; note the poor performance of the untreated alumina.
Fluoride removal from a saturated calcium fluoride solution is shown in Figure 2. This calcium fluoride solution is of special interest because it is typical of the residual after removing high levels of fluoride by precipitation with lime. This feed solution was made up by dissolving excess calcium fluoride in deionized water; the fluoride level was ~8 ppm. The fluoride in the effluent after passing through the alumina column was 0.2 ppm or less, and the capacity of the alumina was about 1.5%. Again note the poor performance of the alumina which was not pretreated with aluminum sulfate.
Two solutions containing hydrofluoric acid in deionized water were used to evaluate fluoride removal at lower pH. One contained 9 ppm fluoride (pH 3.63) and the other 25 ppm fluoride (pH 3.35). The fluoride removal curves are shown in Figure 3. The fluoride in the effluent was less than 0.2 ppm, and the capacity of the alumina at 2.0% was higher than for the neutral solutions. For these tests, the alumina was not pretreated, hence, the higher fluoride in the effluent (up to 1.8 ppm) during the initial period of adsorption, Figure 3. However, this rapidly changed and the fluoride level dropped. If this initial small amount of fluoride passing through cannot be tolerated, then the alumina should be pretreated even for acidic systems
For bed design purposes, the most important relationship is that between the efficiency and the flow rate. The data presented so far were measured at a flow rate of 6-bed volumes per hour, which is slow enough to enable the full bed capacity to be utilized. The efficiency of fluoride removal from a 20 ppm neutral solution at several higher flow rates is shown in Figure 4. At 12 bed volumes per hour, fluoride removal to ~0.2 ppm is still achieved and the capacity is similar to that at 6-bed volumes per hour. At 16.4 bed volumes per hour, some efficiency is lost; the fluoride in the effluent ranges from 0.5 to 0.8 ppm and the alumina capacity at 1.0 ppm fluoride breakthrough is 1.3%. At 24 bed volumes per hour, the fluoride level falls in the range of 1.0 to 1.5 ppm, and the alumina capacity is about 1.0% at the 1.5 ppm fluoride level.
These flow rate data are for 14 x 28 mesh alumina. It was noted that if the flow was stopped for several hours, subsequent to breakthrough, then restarted again, an improvement in fluoride removal occurred. This phenomenon has also been reported in the literature and suggests diffusion rate limitations.
Therefore, it follows that flow rate efficiency is affected by particle size. The smaller the particle size, the higher the flow rate that can be used, but this must be balanced against the higher pressure drop which results from smaller size material.
The data shown up to this point represent single component systems. In practice, many aqueous streams to be treated will contain other components. These other components could have an effect on fluoride removal efficiency. Therefore, any particular stream should actually be tested with alumina. Some data are shown here to indicate the effect some frequently-encountered ions can have on the adsorption efficiency.
In Figure 5 the fluoride removal efficiency is shown for a 10 ppm fluoride solution containing much larger amounts of sodium and one of the following three anions: chloride, sulfate, bicarbonate. In all cases, the fluoride effluent level is reduced to 0.2 ppm or less, but there are differences in alumina capacity for fluoride removal. The effects of chloride and sulfate are very small, but the bicarbonate causes a major decline in capacity. For the feed solution containing 522 ppm bicarbonate, the capacity declines to 0.30%, compared to 1.2 to 1.45% for the other streams. Clearly the bicarbonate ion has a larger inhibiting effect; presumably, competitive adsorption is occurring.
The presence of bicarbonate at the 500 ppm level reduces the fluoride adsorption capacity of the alumina by 75 to 80%. In practice, bicarbonate exists in raw water streams at a variety of levels. The relative effect of different bicarbonate levels is shown in Figure 6. The curves show fluoride removal for 10 ppm feed solutions made from sodium fluoride plus 50, 100, 200, and 522 ppm bicarbonate. In all cases, the fluoride is always reduced to less than 0.2 ppm, but the total adsorption capacity of the alumina is considerably reduced. Even for 50 ppm bicarbonate, the capacity is only 0.75%, which is about half that for the same solution without any bicarbonate.
Higher levels of bicarbonate continue to depress the adsorption capacity, but the incremental effect is less.
The above data indicate that special consideration has to be given to designing a system for fluoride removal in the presence of bicarbonate. Either the bicarbonate has to be removed first, or the system has to be designed for much lower fluoride capacities. This subject is also discussed further in the section on regeneration.
Most fluoride removal applications are long-term and necessitate regeneration of the alumina. There are three known methods for regeneration: (1) NaOH/H2SO4, (2) Al2 (SO4)3 and (3) H2SO4. The conditions for some laboratory tests on the regeneration of A-2 are summarized in Table I. The effectiveness of the different regenerations is judged by their subsequent adsorption performance as shown in Figures 7-10. The letters A-H shown on the curves identify the regeneration conditions given in Table I. Note that the intention here is to present a range of regeneration conditions, and none of them should be taken as necessarily being the optimum.
Table I. Regeneration Data for Activated Alumina
Type F Solution Regenerant Total Volume Time
(ml)/100g A12 O3
________________________________________________________________________
A NaF (1) 1% NaOH 1000 100 minutes
(2) H2O 760 80 minutes
(3) 0.05N H2SO4 1000 100 minutes
________________________________________________________________________
B NaF (1) 1%NaOH 1400 180 minutes
(2) H2O 2000 80 minutes
(3) 0.05N H2SO4 1000 90 minutes
________________________________________________________________________
C NaF 2%Al2(SO4)3
•18 H2O 1290 6.5 hours
________________________________________________________________________
D NaF 2%Al2(SO4)3
•18 H2O 1090 5.25 hours
________________________________________________________________________
E NaF 2%Al2(SO4)3
•18 H2O 2650 24 hours
________________________________________________________________________
F NaF 2% Al2(SO4)3
•18H2O 8000 5 hours
________________________________________________________________________
G CaF2 2% Al2(SO4)3
•18 H2O 1260 5.75 hours
________________________________________________________________________
H NaF 2% H2SO4 8000 5 hours
________________________________________________________________________
The NaOH/H2SO4 method has been well proven in large-scale operation at potable municipal water treatment plants across the country. Also, the laboratory data shown here indicate that it is the most effective method.
The fluoride removal performance of A-2 after regeneration with 1% NaOH/H2O rinse/0.05N H2SO4 is shown is figure 7. The curves are for 10 ppm and 20 ppm fluoride streams made up fromNaF. Note that the regeneration procedure for B takes longer time and uses more NaOH than A. It results in a slightly higher adsorption capacity for the alumina. The 1% NaOH strips the fluoride off of the alumina, then the 0.05 H2SO4 neutralizes residual caustic left after the rinse step and also reactivates the alumina.
The use of aluminum sulfate as a regenerant is described in the literature for laboratory scale testing. The fluoride adsorption performance of alumina after regeneration with 2% Al2(SO4)3 •18 H2O is shown in Figures 8, 9, and 10. In Figure 8, note that after startup the fluoride effluent levels are slow in approaching ~1 ppm and do not fall below this 1 ppm level. The fluoride capacity of the alumina is also lower in the initial cycle for C and D, but in E, where more aluminum sulfate solution and longer times are used, the original capacity of the alumina is restored. Figure 10 (Curve G) shows a case where considerably more aluminum sulfate (8.0L/100 g alumina) is used. The regeneration is much more effective, with fluoride effluent levels of 0.2 ppm being achieved. Note that, even so, the effluent level is slowly coming down to this 0.2 ppm level, and also the alumina capacity is slightly less than for the original adsorption step.
A case with 2% H2SO4 is also shown in figure 10. The use of 2% H2SO4 is mentioned in the literature. For the example in Figure 10, 8L 2% H2SO4/100 g alumina is used more than 5 hours i.e., the same conditions as for the aluminum sulfate regeneration in this same figure. The regeneration efficiencies are essentially the same. However, the use of 2% H2SO4 for regeneration is not recommended because it can have a harmful effect upon the physical strength of the alumina over a period of time.
A water source containing 10 ppm fluoride required treatment to reduce the fluoride to <1 ppm. This particular example illustrates the treatment of water containing significant amounts of bicarbonate. The chemical analysis of the water is given in Table II.
Raw Water Analysis |
|
---|---|
Name | PPM |
F | 10.0 |
F | 10.0 |
Cl | 182 |
SO4 | 21 |
PO4 | <0.02 |
HCO3 | 170 |
Fe | 0.06 |
As | 0.12 |
Na | 150 |
Ca | 135 |
Mg | 0.35 |
The results of treating this water with alumina are shown in Figure 11. First, note that very little fluoride removal is achieved with untreated alumina (Curve 1). The alumina pretreated with aluminum sulfate reduces the fluoride to <0.2 ppm and has a capacity of 0.46 g F/100 g alumina (Curve 2). This is about the capacity for a bicarbonate concentration of 170 ppm based upon the data in Figure 6. The remaining curves are all obtained after successful regenerations of this same alumina column. Curve 3 shows the adsorption after regenerating with 1.4L 1% NaOH/100 g alumina more than 90 minutes, rinse, then 1L 0.05N H2SO4/100 g alumina. There is a dramatic reduction in total capacity for this cycle. (0.22 g fluoride/100 g alumina) although fluoride is reduced to 0.2 ppm.
Visit Our Home Page