Rejection of Bromide and Bromate Ions by a Ceramic Membrane

Abstract

Effects of pH and the addition of calcium chloride (CaCl₂) on bromate (BrO₃⁻) and bromide (Br⁻) rejection by a ceramic membrane were investigated. Rejection of both ions increased with pH. At pH 8, the rejection of BrO₃⁻ and Br⁻ was 68% and 63%, respectively. Donnan exclusion appears to play an important role in determining rejection of BrO₃⁻ and Br⁻. In the presence of CaCl₂, rejection of BrO₃⁻ and Br⁻ ions was greatly reduced, confirming the importance of electrostatic interactions in determining rejection of BrO₃⁻ and Br⁻. The effect of Ca²⁺ is so pronounced that in most natural waters, rejection of both BrO₃⁻ and Br⁻ by the membrane would be extremely small.

Introduction

A major concern in the treatment of drinking waters using ozone is the formation of bromate (BrO₃⁻). The bromide ion (Br⁻) is found naturally in many water supplies, and the ozonation of water containing Br⁻ leads to the formation of BrO₃⁻, a potential human carcinogen. BrO₃⁻ is regulated by the U.S. Environmental Protection Agency (U.S. EPA, 1998) and the Ministry of the Environment, Canada (Ministry of the Environment, Canada, 2002). The typical means of controlling BrO₃⁻ formation are to adjust the pH, add ammonia, or reduce the ozone dosage (Pinkernell and von Gunten, 2001). A reduction in the ozone dosage can have a negative impact on pathogen inactivation (Gerba et al., 2003) or the control of disinfection by-product formation during chlorination (Singer, 1994). Adjusting the pH or adding ammonia can be expensive as well as add to the total dissolved solids of the finished water. The use of activated carbon (Sanchez-Polo et al., 2006) and other techniques for BrO₃⁻ removal, such as UV photocatalysis (Zhang et al., 2009) and biological reduction (Matos et al., 2008), has been investigated. Listiarini et al. (2010) discussed the limitations of each of these treatment methods, and they concluded that existing BrO₃⁻ removal techniques are not cost effective and there is a need to develop new treatment technologies.

Reverse osmosis (RO) can remove Br⁻ and BrO₃⁻ (Gyparakis and Diamadopoulous, 2007), but it is an expensive process, as membrane fluxes are low and high operating pressures are needed. Listiarini and coworkers (2010) reported that polyamide nanofiltration (NF) membranes are not very effective at Br⁻ or BrO₃⁻ removal. The removal of Br⁻ and BrO₃⁻ ions by the NF membranes tested ranged from 1% to 7% and from 6% to 19%, respectively.

Ceramic membrane filtration has the potential to reduce BrO₃⁻ levels in drinking water. The surface of ceramic membranes is typically charged, as the surface of metal oxides from which they are made are usually charged, due to the amphoteric nature of the surface hydroxyl groups on the metal oxide surface (Tsuru, 2001). Previous studies have shown that due to their surface charge, ceramic ultrafiltration (UF) or NF membranes can reject ions, even though the pore size of the membrane is much larger than the size of the ions (Combe et al., 1997; Fievet et al., 2002; Labbez et al., 2003; Skluzacek et al., 2006). Charge effects are largely responsible for the rejection of the ions. If a charged membrane is in contact with an electrolyte solution, electrostatic repulsion will result in a lower concentration of ions with the same charge as the membrane (referred to as co-ions) near the membrane surface and within the membrane pores. On the other hand, counterions, which have an opposite charge to that of the membrane surface, are attracted to the membrane, and the concentration of these ions within the membrane and at its surface is higher than that in bulk solution. Models have been developed that describe the rejection of ions by charged membranes based on the Donnan, steric, and dielectric exclusions at the interfaces (Escoda et al., 2011). In these models, the extended Nernst–Planck equation is used to describe transport within the pores of the membrane (Escoda et al., 2011).

Due to their ability to reject ions, ceramic UF membranes could potentially be used to reduce BrO₃⁻ levels in drinking water and because of their greater permeability, UF membranes could be operated at higher membrane fluxes and/or lower transmembrane pressures than NF or RO membranes (U.S. EPA, 2005). This would result in considerable savings in operating and/or capital costs. In this study, we have measured the rejection of Br⁻ and BrO₃⁻ by a ceramic UF membrane. Rejection by the membrane was studied at various pH. The influence of the addition of calcium chloride (CaCl₂) was also studied, as the presence of Ca²⁺ is likely to affect the rejection of Br⁻ or BrO₃⁻, as it will interact more strongly with the membrane than monovalent ions, such as Na⁺. In addition, since it is a divalent ion, Ca²⁺ will have a greater shielding effect than monovalent ions (Labbez et al., 2003).

Materials and Methods

Experimental setup

The apparatus employed is shown in Fig. 1 and is described in greater detail in Moslemi et al. (2011). The system was operated at a transmembrane pressure of 138 kPa (20 psi). The average permeate flux was 6.5 mL/min. The cross-flow was 200 mL/min. The bleed flow rate was held at 5.0 mL/min. In order to duplicate the hydrodynamic conditions used in the hybrid ozonation-membrane filtration experiments (Moslemi et al., 2011), oxygen was injected into the water stream flowing into the retentate loop through a Tee fitting at a flow rate of 50 mL/min. The temperature of the water inside the system was kept constant at 20°C±0.1°C. Tubular ceramic ultrafiltration membranes (TAMI North America) with molecular weight cutoffs of 5 kDa were used in these experiments. These membranes have a titania support and a filtration layer. The membrane had three channels and a total filtering surface area of 41.2 cm².

FIG. 1.

Schematic of the experimental setup.

Analytical methods and reagents

Br⁻ and BrO₃⁻ concentrations in the retentate and permeate streams were measured by high-performance liquid chromatography (Moslemi et al., 2011). pH was measured using a pH meter (pHTestr 30, Eutech Instruments). Milli-Q water with a resistivity of greater than 18 MΩ was used to prepare all solutions and reagents. In order to obtain the desired concentrations of Br⁻ and BrO₃⁻ ions in the feed water, Milli-Q water was spiked with appropriate amounts of sodium bromate (NaBrO₃⁻, 99.7+%, Fisher Scientific) or sodium bromide (NaBr, 99.995% metals basis, Fluka). Phosphoric acid (H₃PO₄) (99.99%, Sigma-Aldrich) and/or sodium hydroxide (NaOH) (98.0+%, Fluka) was used to adjust the water pH.

The data reported are averages of the results obtained in triplicate experiments, and the error bars represent the standard deviations of the results for the triplicate experiments.

Accumulation of ions in the retentate loop

Since there is no reaction, the accumulation in the retentate loop can be calculated using the mass balance equation:

(1)

Our experimental results indicated that the adsorption of BrO₃⁻ and Br⁻ by the system at the various pHs examined (3.0, 6.0, and 8.0) was negligible (less than 1%). Thus, the mass balance equation can be written as follows:

(2)

• where

• V=total volume of water in the system (1500 mL)

• C_in=concentration of the ion in the inflow into the system (μM)

• Q_in=inlet flow rate (mL/min)

• Q_P=permeate flow rate (mL/min)

• Q_B=bleed flow rate (mL/min)

• t=time (min)

After rearrangement, Equation (2) can be written as follows:

(3)

The amount of BrO₃⁻ and Br⁻ that accumulated in the system after 180 min can be calculated by integrating both sides of Equation (2) over the course of the experiment (180 min). In order to obtain the accumulation from Equation (2), the discretization method was used to calculate the areas underneath the concentration versus the time graphs derived from the experimental data.

(4)

where Q_in is the flow rate, and C_in is the concentration entering the membrane; Q_p and C_p pertain to the permeate; and Q_R and C_R, pertain to the reject stream.

Results and Discussion

Figures 2–4 present the results for BrO₃⁻ and Br⁻ concentrations in the permeate and retentate streams, obtained at pH values of 3.0, 6.0, and 8.0, respectively. As shown in Fig. 2, at pH 3.0, the Br⁻ and BrO₃⁻ concentrations are constant, and the concentrations of these ions in the permeate and retentate are nearly identical, indicating that the rejection of this ions is negligible. However, at pH 6.0 and 8.0 (Figs. 3 and 4), BrO₃⁻ and Br⁻ ions were rejected by the ceramic membrane. The ions cannot be rejected due to sieving. The pore size of the membrane was estimated to be 4 nm using the following relationship:

(5)

FIG. 2.

Bromate (BrO₃⁻) and bromide (Br⁻) concentration vs. time. Membrane: 7 channel–5 kDa, [BrO₃⁻]_o: 7.8 μM (1 mg/L), [Br⁻]_o: 12.5 μM (1 mg/L), [CaCl₂]_o: 0 μM, pH: 3.0.

FIG. 4.

BrO₃⁻ and Br⁻ concentration vs. time. Membrane: 7 channel–5 kDa, [BrO₃⁻]_o: 7.8 μM, [Br⁻]_o: 12.5 μM, [CaCl₂]_o: 0 μM, pH: 8.0.

FIG. 3.

BrO₃⁻ and Br⁻ concentration vs. time. Membrane: 7 channel–5 kDa, [BrO₃⁻]_o: 7.8 μM, [Br⁻]_o: 12.5 μM, [CaCl₂]_o: 0 μM, pH: 6.0.

where d is the pore size in nm, and MWCO is the nominal molecular weight cutoff of the membrane in Daltons (Howe and Clark, 2002). The radii of hydrated BrO₃⁻ and Br⁻ ions are 0.35 and 0.33 nm, respectively (Nightingale, 1959); so, they are much smaller than the pore size of the membrane. Donnan exclusion appears to play an important role in determining the rejection of BrO₃⁻ and Br⁻. The filtration layer of the membrane consists of TiO₂, which has a pH of zero point of charge (pH_zpc) in the range of 4.1–7.2 (with a median 6.0) (Kosmulski, 2009). At pH 8, the surface charge of the membrane would be negative. The rejection of ions by this membrane at higher pH can be explained by the repulsive interactions between the negatively charged membrane and the BrO₃⁻ and Br⁻ ions (Labbez et al., 2003). In order to maintain electroneutrality in the retentate and permeate streams, the counter ion in this feedwater solution, Na⁺, would also be rejected by the membrane; however, no measurements were made of Na⁺ levels in the retentate or permeate to confirm that this occurred. H₂PO₄⁻, the predominant phosphate species at pH 3, will sorb onto the titania surface (Kang et al., 2011). Previous studies have shown that phosphate is strongly sorbed on TiO₂ at pH 3 (Kang at al., 2011), and spectroscopic evidence indicates that phosphate sorbed onto TiO₂ forms inner-sphere complexes with the Ti atom (Connor and McQuillan, 1999; Kang et al., 2011). In pure water, the surface charge of the TiO₂ membrane would be positive at pH 3; however, the minimal rejection of the Br⁻ and BrO₃⁻ at pH 3 suggests that at this pH, Donnan exclusion is minimal, as the sorption of phosphate significantly reduces the surface charge on the TiO₂ surface.

The concentration of BrO₃⁻ and Br⁻ ions in the retentate increases with time, as these ions are retained within the recycle loop. The rejection, R, can be calculated using the expression:

(6)

• where

• C_P=the concentration of the ion in the permeate (μM)

• C_R=the concentration of the ion in the system (retentate) (μM)

For all the experiments conducted, R is essentially independent of time. Table 1 shows the values of R for Br⁻ and BrO₃⁻ ions at the end of each experiment. The rejection of BrO₃⁻ is slightly greater than that for Br⁻, and the rejection of both ions increases with pH.

Table 1.

Rejection Coefficients at 180 Min

		Rejection coefficient (%)
pH	[CaCl2] μM	BrO3−	Br−
3.0	0	1.4	0.6
6.0	0	41	33
8.0	0	68	63
6.0	10	13.3	7.6
6.0	250	4.2	1.2
6.0	500	3.4	1.1
6.0	1000	2.5	1.1

As shown in Table 1, the presence of CaCl₂ greatly reduces the rejection of both Br⁻ and BrO₃⁻. At pH 6 and a Ca²⁺ concentration of 10 μM (0.4 mg/L), there is a 70%–80% decrease in the retention of Br⁻ and BrO₃⁻. Typical Ca levels in freshwaters range from 0.5 to 100 mg/L (12.5–2,500 μM) (Manahan, 1991); so, in most natural waters, the Ca²⁺ levels would be sufficiently high that the rejection of both BrO₃⁻ and Br⁻ by the membrane would be extremely small. Zhao et al. (2005) showed that the addition of CaCl₂ shifts the isoelectric point of a ceramic membrane to a higher pH value. In the presence of CaCl₂, the surface charge of the membrane is more positive. The lower rejection of BrO₃⁻ and Br⁻ seen in CaCl₂ solutions can also be explained by the shielding of the surface charge in the presence of CaCl₂ (Labbez et al. 2003).

Accumulation

Due to their rejection, BrO₃⁻ and Br⁻ accumulated in the retentate recirculation loop over the course of the experiment (180 min). The accumulation of these ions in the retentate loop is presented in Figs. 5 and 6. The insets in these figures depict the accumulation normalized with regard to the initial concentration of the ions in the system. As can be seen, regardless of pH, BrO₃⁻ is accumulated in the reactor to a slightly greater degree than Br⁻. This is because the rejection of BrO₃⁻ was slightly greater than the rejection of Br⁻.

FIG. 5.

BrO₃⁻ and Br⁻ accumulation in the system after 180 min at various pHs. Membrane: 7 channel–5 kDa, [BrO₃⁻]_o: 7.8 μM, [Br⁻]_o: 12.5 μM, [CaCl₂]_o: 0 μM.

FIG. 6.

BrO₃⁻ and Br⁻ accumulation in the system after 180 min at various CaCl₂ concentrations. Membrane: 7 channel–5 kDa, [BrO₃⁻]_o: 7.8 μM, [Br⁻]_o: 12.5 μM, pH: 6.0.

Conclusions

The rejection of Br⁻ and BrO₃⁻ increased with pH. At pH 8, the rejection of BrO₃⁻ and Br⁻ was 68% and 63%, respectively. Donnan exclusion appears to play an important role in determining the rejection of BrO₃⁻ and Br⁻. In the presence of CaCl₂, the rejection of BrO₃⁻ and Br⁻ ions was greatly reduced, confirming the importance of electrostatic interactions in determining the rejection of BrO₃⁻ and Br⁻. As such, membrane filtration can be used to reduce bromate levels in drinking water. If ozonation is used, the BrO₃⁻ levels in the treated water can be reduced by using membrane filtration to remove Br⁻ before ozonation or by removing BrO₃⁻ after ozonation. When ozonation is used in combination with membrane filtration, as was accomplished in this hybrid system, fouling can be controlled (Karnik et al., 2005b) and the formation of disinfection by-products after chlorination can be reduced (Karnik et al., 2005a). In very soft waters, the combined use of ozonation and ceramic membrane may lead to significant decreases in BrO₃⁻ levels in the product water; however, in most natural waters, the Ca²⁺ levels would be high enough such that the rejection of both BrO₃⁻ and Br⁻ by the membrane would be extremely small.

Acknowledgments

The authors acknowledge financial support of this work from the National Science Foundation Research (Grant No. CBET-0506828) and from the Natural Sciences and Engineering Research Council of Canada Discovery Grant Program.

Disclosure Statement

No competing financial interests exist.