Effect of silica fouling on fluoride rejection in RO membrane system
(1) General introduction
Silica (SiO2)n is one of the foulants present in feed waters in membrane filtration processes. Silica can be found in many different forms and is influenced by a variety of variables including temperature, pH, metal ions, and ionic strength (S.D.N. Freeman and R.J. Majerle, 1995).
R.Y. Ning (2002) proposed that silica is the final and most stable form of silico-oxygen acid polymerization at normal conditions. First, monosilicic acid (Si(OH)4), which is a weak tetravalent acid with pKa values of 9.9, 11.8, and 12 will polymerize by dehydration to form Si-O-Si anhydride bonds: n Si(OH)4 → (OH)3Si-O-Si(OH) dimer. Then, dimer (two similar subunits linked) and oligomers (a finite number of subunits linked) are formed, respectively. Finally, colloidal polymers and particulate silica will be formed. The silica (SiO2)n formula is shown in Figure 1. Thus, the molecules of silica represented by the formula SiO2 is polymeric form, and is more accurately represented by the formula of (SiO2)n where n is infinited in number, allowing for extensions in term of amorphous and crystalline forms of silica.
(Source: K.D. Demadis et al., 2005)
Figure 1 Silica (SiO2)n formula
In crystalline silica the tetrahedral structure is continued over a large range, forming a well-ordered lattice (crystal). Whereas, amorphous silica is the non-crystalline form of silica. In amorphous silica, this long range order does not present due to the disordered nature of the material, that some atoms are missing a neighbor to which it would be able to bind (
P. Sahachaiyunta et al. (2002) reported that silica is generally found in natural waters in three different forms (L.F. Comb, 1996), namely, reactive silica (dissolved silica), colloidal silica and particulate silica. Reactive silica in water can be found in many forms i.e. monosilicic acid, disilicic acid and polysilicic acid with molecules smaller than the colloidal range (G.B. Alexander, 1953). Colloidal silica is widely thought to be both silica that has polymerized with infinited units of silica dioxide and silica that have formed with organic compounds or with other complex inorganic compounds. Particulate silica is larger in size and mostly comprised of sand or suspended solids in water.
The dissolution of silica involves a chemical reaction can be expressed by:
The dissolution rate mainly depends on pH and temperature. The soluble form of silica is initial monomer, as it contains only one silicon atom. In this monomeric form, or often called monosilicic acid. Monosilicic acid is generally deionized at natural pH. At a pH of 8.5 only 10% of the monosilicic acid is ionized and when the pH reaches at 10, 50% is ionized (R. Sheikholeslami and S. Tan, 1999). Besides, high water temperature can accelerate hydrolysis of SiO2.
In contrast, Silanols (Si-OH) can react with each other to form a siloxane bond (Si-O-Si), which is the beginning of the polymerized silica as discussed previously (
(2) Silica fouling and hyperfiltration (NF and RO filtration)
Silica fouling mechanisms can be separated into three categories. These phenomena will reduce water flux in membrane filtration processes.
(1) Polymerization of dissolved silica when concentration exceeds its solubility. According to R. Sheikholeslami and S. Tan (1999), the solubility of silica was observed of 6 ppm and 100-104 ppm for crystalline silica and amorphous silica, respectively (K.B. Krauskoph (1956); G.B. Alexander, W.M. Heston and R.K. Iler (1954). When silica is dissolved in water, it forms monosilicic acid, which will remain in the monomeric state as long as its concentration remains less than about 2 mM (R. Sheikholeslami and S. Tan, 1999).
(2) Deposition of colloidal and particulate silica.
(3) A deposition of silica as copolymerization of silicic acid with hydroxideof metal elements or reaction of silicic acid with organic compound to form anhydrides (R.Y. Ning, 2002). These formations of silica scale in hyperfiltration systems leads to limited recovery in many installations and unnecessary increase in costs.
When Fe(OH)3, Al(OH)3, Ca(OH)2, and Mg(OH), are presented in water at suitably high pH, these hydroxides can react in the copolymerization resulting in the incorporation of Fe, Al, Ca and Mg into the complex anhydrous silicate structures (R.Y. Ning, 2002).
(3) Mechanisms of solutes through silica fouling layer
Solutes transport through the silica layer could be either convective or diffusive depending on the structure of the silica layer. If the silica layer is made of a cake layer of colloidal or particulate silica, then the solutes transport through the layer dominated by convection flow, whereas if the silica layer is made of polimerized silica layer, then the solutes can move through the layer only by diffusion.
(4) Effect of physicochemical on silica fouling
R. Sheikholeslami and S. Tan (1999) reported laboratory tests with simulated waters in the range of those in a prospective desalination plant. The objective of this study was to determine the water quality effects (i.e. cations) on silica precipitation both in batch and dynamic tests using RO membranes. Operating pH was set at 6.5 for all experiments, which prevented polymerization as well as precipitated of silicates. Results indicated that magnesium hardness was more effective than Ca hardness in silica polymerization rate. At a given hardness, decreasing the ratio of calcium to magnesium was found to increase the polymerization rate. This was apparent in batch tests and during the initial period of dynamic tests. Moreover, batch tests showed that polymerization increases with the degree of supersaturation. However, in dynamic tests the flux decline was higher at low feed silica concentrations and higher Ca: Mg ratio after a period of operation. This inflection on the effect of feed silica concentration and Ca: Mg ratio in dynamic tests is might be due to precipitation of monomeric silica and deposition at the surface resulting in a lower feed silica concentration. While at high silica concentrations or when polymerization is rapid (at high Ca: Mg), bulk polymerization results and deposits are of more porous colloidal silica.
R. Sheikholeslami et al. (2001) reported that the pretreatment for reducing silica fouling potential was operating pH range in absence of cations was above 9.5 and below 5.5; while in presence of cations, it was for pH<5.5. It was also found that the polymerization of silica was increased with incrasing of intial silica concentration. The increasing in polymerization rate was also observed when calcium and magnesium increase, and magnesium hardness seemed to be more effective than calcium hardness. Effect of increasing in pH of sodium aluminate, lime and soda ash solutions on silica reduction was experimented and the relationships for them were carried out. It was observed that the increasing in pH in lime softening was much more effective in silica reduction than addition of sodium aluminate, that was up to 70% reduction in silica was obtained at pH of 10.2 as opposed to 50% reduction with sodium aluminate (at about dosing rate of 22 ppm). However, it was not cleared for relationship between silica reduction and lime/soda-ash.
T. Koo, et al. (2001) studied silica fouling and membrane cleaning of reverse osmosis membranes. In batch test experiments, it was found that increasing the concentration of calcium and magnesium also enhanced the polymerization of silica. This also means that increasing the total hardness of the solution enhances silica polymerization. The presence of calcium and magnesium in low concentrations did not help catalyze silica polymerization. But the presence of calcium and magnesium in higher concentrations did help catalyze silica polymerisation. At silica concentrations greater than 300 ppm, polymerization will take place even in the absence of calcium and magnesium. Moreover, carbonate was found to have greater effect on silica polymerization than chloride. It was found that carbonate ions promote silica polymerisation more compared to chloride ions. The cleaning of membranes was using distilled water. However, distilled water did not restore the membrane fluxes properly.
P. Sahachaiyunta et al. (2002) investigated the effect of silica fouling in presence of minute amounts of inorganic salts in reverse osmosis membranes. The various cations including iron, manganese, nickel, and barium, which existed in industrial and mineral processing wastewaters, on silica fouling was studied. Structure of fouled membrane is analysed using scanning electron microscope (SEM). The concentration of soluble silica and total silica (soluble plus colloidal) in the feed solution monitored during the runs using UV spectrophotometer and ICP/AES, respectively. It was found that additions of manganese, nickel and barium in silica feed solution accelerated flux decline of the feed silica solution. Since the structure of the deposit and absence of these cations was the same, thus it might be concluded that manganese, nickel and barium do not react with dissolved silica, and the reduction of flux was probably due to these cations accelerating polymerization of silica. However, when iron of 3 ppm was added, the foulant structure was changed drastically. It was observed monolayer structure with cracks developing. This layer probably developed due to association of iron (III) with dissolved silica to form the complex iron silicate, Fe(OH)3 · SiO2.
(5) Chemical cleaning of silica fouling
Chemicals used for silica fouling control can be divided into three categories: (1) chemicals for silica inhibition, (2) chemicals for silica dispersion (E. Neofotistou and K.D. Demadis, 2004), and (3) chemicals for silica scale removal from membrane surfaces. Figure 2 shows silica inhibition and silica dispersion processes.
(Source: E. Neofotistou and K.D. Demadis, 2004)
Figure 2 Silica inhibition and silica dispersion
Silica inhibitors can retard the polymerization of silica, which exceeded its solubility i.e phosphonate-based chemicals. Dispersants can work by placing a surface charge on the polymerized silica and cause repulsion and dispersion of these polymerized structures into water phase (L. Y. Dudley and J. S. Baker, 1999).
Polymerization of dissolved silica (i.e. supersaturation and catalyzed by hardness) can be very difficult to remove. It should be noted that this type of silica fouling is different from colloidal and particulate silica, which may be associated with either metal hydroxides or organic matter (Toray, 2003). Thus, it might difficult to remove polymerized silica by typical chemical cleaning process. However, polymerized silica scales removed with a high pH cleaning solution (pH of 10-11) were reported. Caustic (sodium hydroxide) at the maximum pH allowed by the membrane manufacturer will remove polymerized silica scales but it will take many hours to remove a silica scale (David H. Paul, 1996). While colloidal silica and particulated silica coating, which not associated with either metal hydroxides or organic matter can be removed from membrane surface by physical flushing (Nitto Denko Corporation, 2000).
Sayed Siavash Madaeni et al. (2001) studied the fouling problem in the application of membranes for water treatment. Chemical cleaning processes were carried out for membrane regeneration. The chemical cleaning of fouled membrane using acid, alkaline, chelating agent, and surfactant and detergent solutions has been investigated. It was found that cleaning efficiency depended on the type of cleaning agent and its concentration. The results showed that the efficiency increases with increasing of chemical cleaning concentration while the mix cleaning solution of EDTA, SDS, and NaOH (0.05%ww) was the most effective solution for foulants removal. Operating conditions such as cross flow velocity, turbulence in the vicinity of membrane surface, temperature, pH, and cleaning time were also affected the cleaning process. The increasing of these conditions to their optimum values resulted in the highest efficiency of cleaning processes.
E. Neofotistou and K.D. Demadis (2004) investigated the use of antiscalants for mitigation of silica fouling and deposition in desalination systems. Chemical for silica inhibition was examined. The inhibition performance of Starburst® polyaminoamide (PAMAM) dendrimers of generations 0.5, 1, 1.5, 2, and 2.5 were investigated. The results showed that the NH2-terminated dendrimers, PAMAM 1 and 2 were much more effective for SiO2 inhibitors with the optimum dosage of 40 ppm. However, PAMAM 1 and 2 dendrimers also formed SiO2-PAMAM composite precipitates after long period. Thus, formation of such materials should be considered. Moreover, polyethyloxazolines were also effective for SiO2 inhibitors, but their performance is less dependent on structural features.
L. Y. Dudley and J. S. Baker (1999) proposed that phosphonate-based chemicals were widely used in water treatment as scale and corrosion inhibitors. In membrane systems, they act as scale inhibitors. These products have the ability to hold highly supersaturated solutions in a stable condition during the finite time it takes the water to exit the membrane system. They claimed that the phosphonate-based chemicals offered the better bicarbonate rejection and less corrosive to membranes.
(6) Preliminary experiment
(a) Introduction
The preliminary experiment was conducted on December 2006. The objectives of this experiment were to investigate the optimal contact time of cleaning chemical solutions on silica fouling membrane and to evaluate the performance of silica fouling membrane before/after chemical cleaning process on fluoride and chloride removal.
(b) Methodology
Batch test was experimented. The operating pressure of 0.8 MPa and temperature of 25 °C were controlled. Permeate flux was monitored by using balance and connected to PC. Surface structure of membrane was observed by using SEM while fluoride and chloride were analyzed by using IC. Cleaning efficiency was evaluated by comparing permeate flux recovery and fluoride and chloride removal (feed concentration of 10 mg/L) both before and after chemical cleaning processes.
The examined cleaning chemical was mixed of NaOH, SDS, and EDTA. The concentrations of all cleaning chemicals were 0.05 % ww. The cleaning period for each batch experiment will be varied of 30, 60, and 120 min, respectively. The procedure for cleaning experiments was set up as followed:
(1) The membranes were previously fouled with silica solution at pilot plant of Tamakawa water treatment plant.
(2) The silica-fouled membrane was washed for 20 minutes to remove weak-bonded particles from membrane surface and cut into small pieces and preserved in milli-Q water.
(3) The chemical cleaning solutions were prepared and followed by washing the fouled membrane from (2) with prepared solutions without applying pressure. The cleaning period for each batch experiment will be varied of 30, 60, and 120 min, respectively.
(4) Fouled membrane after chemical cleaning process was washed again with milli-Q water.
(5) The permeate flux after chemical cleaning process was measured using milli-Q water.
(6) The permeate flux recovery is determined.
(4) Results
Figure 3 shows the SEM images of membrane surfaces of each condition: (a) silica fouled membrane, (b) silica fouled membrane with 30 min chemical cleaning, (c) silica fouled membrane with 60 min chemical cleaning, and (d) silica fouled membrane with 120 min chemical cleaning. It was found that increase in chemical cleaning time, increasing of clearance on membrane surface was observed. And it leaded to increasing in permeate flux of membrane significantly.
(a) (b)
(c) (d)
Figure 3 SEM images of membrane surfaces (Yasuhiro Matsui, D3)
Figure 4 shows concentrations of fluoride and chloride in permeate water of each experimental condition. It was found that increasing of contact time of 30 and 60 min, fluoride and chloride concentration in permeate water was reduced. It means that chemical cleaning with these two contact time values were successfully to clean the silica fouled membranes and efficiently to recover the membrane performance. However, fluoride and chloride concentration in permeate water were increased in the case of contact time of 120 min. This phenomenon might be due to the rupture of membrane surface when contacted with cleaning chemical for prolonged periods.
Figure 4 Fluoride and chloride concentration in permeate versus contact time
(7) Preliminary research plan
(7.1) Chemical cleaning experiment
(a) Objectives
(1) To study the effects of physicochemical conditions on chemical cleaning of silica fouled membranes including chemical compositions, contact time, and pH.
(2) To study the efficiency of a virgin membrane, a fouled membrane, and a membrane after chemical cleaning on permeate flux recovery and rejection of fluoride.
(3) To observe the effect of chemical cleaning on zeta potential and surface charge density of membrane surface before and after cleaning process.
(b) Methedology
Batch test will be conducted. Silica fouled membranes at pilot plant of Tamakawa water treatment plant, which tested in preliminary experiment will be experimented. Experiments will be conducted with 64 (4 x 4 x 4) combinations. Chemical solutions of NaOH, NH4OH, NaOH+EDTA+SDS, and NH4OH+EDTA+SDS, will be examined at controlled temperature of 25 °C. Contact time of 30, 60, 90, and 120 min, will be investigated. And pH values of chemical solutions in range of 8 – 11 will be evaluated. Flow diagram of this preliminary research plan is illustrated in Figure 5.
In the literatures, the procedure for cleaning experiments was set up as followed (M.J. Munoz-Aguado et al., 1996).
- The permeate flux of virgin membrane was measured by using DI water (jo).
- The virgin membrane was fouled by observed foulant solution; the monitoring of water flux during this process was also done. After fouling was finished, the permeate flux of fouled membrane was measured by using DI water (ji).
- The fouled membrane was washed for 20 minutes to remove weak-bonded particles from membrane surface and the permeate flux was measured again by using DI water (jk).
- The chemical cleaning solutions were prepared and followed by washing the fouled membrane from (3) with prepared solutions without applying pressure.
- The permeate flux after chemical cleaning was measured using DI water (jc).
- The permeate flux recovery is determined as jc/jo.