Page 8: Guidelines for Canadian Drinking Water Quality: Guideline Technical Document – Selenium

Part II. Science and Technical Considerations (continued)

7.0 Treatment technology and distribution system considerations

Selenium has several oxidation states, but only two are predominant in drinking water: Se(IV) and Se(VI). The chemical speciation and behaviour of selenium are highly dependent on the pH and redox potentials of the environment. In natural water (pH 6.0-9.0) under oxidizing conditions, Se(VI) predominates as the divalent ionic form SeO42−. Se(IV) will predominate under reducing conditions; at pH below 8.15, the monovalent biselenite anion HSeO3− will be the dominant form; at pH greater than 8.15, the divalent anion SeO32− will dominate. Reduced selenium species such as elemental selenium (Se0) and selenides (HSe−) are insoluble and likely to be released as colloidal suspensions in the surface water. Organic selenium species occur in the natural water by means of microbiological assimilation and degradation (McKeown and Marinas, 1985; Clifford, 1999).

Control options for addressing selenium in drinking water include blending of waters, selection of alternative low-selenium sources and the removal of excess selenium by treatment processes at the public water supply or household level. Blending can reduce the equipment capacity requirements and costs by blending or mixing a portion of the feed water with the treated water. However, the initial concentration of selenium in the source water and the efficiency of the treatment process will determine if the blending process is advantageous (U.S. EPA, 1985).

7.1 Municipal scale

The removal of excess selenium from drinking water has not been studied on a full-scale treatment plant basis, and limited data exist on laboratory and pilot plant tests. The speciation of selenium in the raw water plays a critical role in the effectiveness of treatment methods used for the removal of selenium. As the species of selenium will determine the effectiveness of the treatment technologies, the removal of Se(IV) and Se(VI) will be considered separately when applicable.

The U.S. EPA has identified the following technologies as the best available technologies (BATs) for selenium removal from drinking water: coagulation/filtration (for Se(IV) only), lime softening, ion exchange (for Se(VI) only), RO, activated alumina and electrodialysis reversal (for Se(IV) only). The removal efficiency of most of the BATs ranged from 75% to 99%; however, lime softening and electrodialysis reversal achieve lower removal rates of approximately 50% and 71%, respectively (U.S. EPA, 1991c). Although conventional coagulation and lime softening processes demonstrated a limited capacity for removing Se(VI), these two technologies may be used when the removal of Se(IV) is sufficient to meet the guideline value for selenium in drinking water (U.S. EPA, 1989).

Several studies demonstrated that adsorptive materials containing various iron oxides were capable of removing selenium species in the water. Se(IV) has been found to adsorb more readily than Se(VI) (Lo and Chen, 1997; Zingaro et al., 1997; Li and Viraraghavan, 1998; Rovira et al., 2008).

The selection and effectiveness of the treatment process are driven by several factors, including source water chemistry, selenium oxidation state, selenium concentration, pre-existing treatment processes and facilities, treatment goals, residual handling concerns and costs.

7.1.1 Conventional coagulation/filtration and lime softening processes

The removal efficiency of selenium by conventional coagulation/filtration treatment depends on the oxidation state of selenium, coagulant type and dose, the selenium concentration in the raw water and the pH of the treated water (Sorg and Logsdon, 1978; Sorg, 1985). The removal of selenium from drinking water by conventional coagulation processes has not been investigated in a full-scale treatment plant. Jar tests and pilot-scale tests demonstrated that conventional coagulation/filtration techniques are moderately successful in removing Se(IV) (80% removal efficiency) and ineffective in removing Se(VI) from the drinking water supply. Conventional coagulation/filtration and lime softening processes are not defined as BATs for small systems unless these treatment processes are currently in place (U.S. EPA, 1991c). Removal of Se(IV)

Jar tests evaluated the impact of different parameters, such as coagulant type, pH of the raw water and the influent selenium concentration, on the effectiveness of the coagulation process. The optimum treatment conditions have been confirmed by a pilot plant study. Both the bench and pilot scale tests showed that Se(IV) was effectively removed (70-80%) by ferric coagulation whereas alum coagulation was relatively ineffective (10-20% removal).

Jar test experiments demonstrated that the application of a ferric sulphate dose of 25 mg/L achieved an approximately 80% reduction of the spiked Se(IV) concentration (0.03 mg/L) in surface water in the pH range of 6-7. Parallel experiments conducted with groundwater achieved up to 70% Se(IV) reduction. The removal efficiency decreased to approximately 10% when the pH increased to 9 for both types of water (Sorg, 1985).

Alum was found to be independent of pH and less effective (10-25%) than the ferric sulphate coagulant for Se(IV) removal. An alum dose of 25 mg/L achieved up to 25% removal of the 0.03 mg/L spiked Se(IV) in surface water at pH 6-7, whereas the tests conducted with groundwater reported up to 15% removal. Both ferric and alum coagulants achieved greater removal efficiency for Se(IV) in surface water than in groundwater (Sorg and Logsdon, 1978; Sorg, 1985).

Sorg and Logsdon (1978) reported that an increase in selenium removal was achieved by increasing the dose of either coagulant. Jar test experiments demonstrated that increasing the ferric doses at a pH in the range of 8.5-8.6 achieved better removal improvement than at a pH in the range of 5.5-7.0. Conversely, experiments conducted with increased alum doses achieved better improvements in the lower pH range.

To determine the effect of the influent Se(IV) concentrations on the effectiveness of the coagulation process, jar tests were conducted with feed Se(IV) concentrations up to 10 mg/L. Se(IV) removal capacity was decreased from 58% to approximately 30% when the influent Se(IV) concentrations increased from 0.1 mg/L to 10 mg/L using a ferric sulphate dose of 25 mg/L and pH of 7.2 (Sorg and Logsdon, 1978; Sorg, 1985).

In a pilot-scale study, the coagulation/dual-media filtration techniques demonstrated that an average influent Se(IV) concentration of 0.027 mg/L in surface water was reduced to 0.005 mg/L (81% removal) using a ferric sulphate dose of 23 mg/L and pH of 6.9. The removal rate decreased to approximately 30% as the pH increased to 8.3. A ferric sulphate dose of 30 mg/L achieved 79% reduction of a Se(IV) concentration of 0.047 mg/L in groundwater at pH 6.4. Parallel experiments conducted with alum doses in the range of 30-34 mg/L achieved up to 20% removal of the influent Se(IV) concentrations in the range of 0.019-0.03 mg/L in surface water and pH in the range of 6.6-8.3. An alum dose of 28 mg/L reduced an influent concentration from 0.056 mg/L to 0.04 mg/L (29% removal) in groundwater at pH 7.8 (Sorg and Logsdon, 1978; Sorg, 1985).

Co-precipitation involves sorption/inclusion of contaminants to an actively precipitating substrate, resulting in the formation of mixed solid-phase. Surface adsorption is one of the principal mechanisms of co-precipitation. Laboratory experiments investigated the factors affecting the iron/selenium precipitation system, such as the type of precipitant (FeCl3 and FeSO4), pH, mixing time, turbidity and temperature. Because SO42- anions showed ability to compete for the adsorption sites, experiments were only conducted and reported for FeCl3. The initial Se(IV) concentration of 0.05 mg/L in synthetic water was reduced by between 83.8 and 93.4% using FeCl3 doses ranging from 5 to 20 mg/L and pH levels between 6 and 8. The optimum mixing time was reported as being between 5 and 10 minutes. The turbidity and temperature of the water ranged widely. A water supply with design output of 2000 m3/day used the process of iron/selenium co-precipitation to remove selenium from spring water. An influent selenium concentration (the form of selenium was not specified) in the range of 0.03-0.04 mg/L was reduced to below 0.01 mg/L using a ferric chloride dose of 5 mg/L, a mixing time of 10 minutes and a pH of 7 (Shi et al., 2009).

Studies reported that the prechlorination process may affect the removal of Se(IV) in drinking water, as the chlorine will tend to oxidize Se(IV) to Se(VI) (Sorg, 1985; Boegel and Clifford, 1986). Coagulation tests, following a prechlorination process with a chlorine dose of 2 mg/L and pH of 6.4, showed a reduction in Se(IV) removal efficiency from 56% to 21% when the chlorine contact time increased from 0 to 60 minutes. The coagulation process was conducted on spiked well water with a Se(IV) concentration of 0.1 mg/L and a ferric sulphate dose of 25 mg/L. The same trend of decrease was found at pH in the range of 7.9-8.1, but not in the range of 6.8-7.8 (Sorg, 1985). However, Boegel and Clifford (1986) found that the optimum pH for the oxidation of Se(IV) using free chlorine at a concentration of 2 mg/L ranged from 6.5 to 7.5. Removal of Se(VI)

Jar tests and pilot-scale tests demonstrated that conventional coagulation/filtration techniques are ineffective in removing Se(VI) from the drinking water supply. Jar tests conducted with alum and ferric sulphate doses as high as 200 mg/L and at pHs in the range of 6-8 reported less than a 10% removal of Se(VI) from both surface water and groundwater (Sorg and Logsdon, 1978; Sorg, 1985; U.S. EPA, 1985).

Pilot-scale studies, using ferric sulphate and alum coagulants, achieved 11% and 18% reduction of Se(VI) concentration, respectively, in the settled water. The tests were conducted with a ferric sulphate dose of 32 mg/L, an initial Se(VI) concentration of 0.097 mg/L and a pH of 6.5. Tests with the alum coagulant were conducted with an alum dose of 25 mg/L, an initial Se(VI) concentration of 0.028 mg/L and a pH of 6.8 (Sorg and Logsdon, 1978; Sorg, 1985). Lime softening

Jar test and pilot plant lime softening investigations demonstrated that the technology achieved approximately 50% Se(IV) removal and was ineffective for Se(VI). Lime softening is a pH-dependent process.

The maximum Se(IV) removal (range of 45-50%) was observed at pH 11.5, whereas the removal rate decreased to 30% when the pH was decreased to 9.5. An increase of the influent Se(IV) concentration in the range of 0.05-10 mg/L showed that the removal efficiency remained constant at approximately 50% (Sorg, 1985).

The results for Se(VI) removals were similar to those reported for conventional coagulation treatment, achieving a maximum 10% removal at a pH in the range of 9-11.5 with initial Se(VI) concentrations ranging from 0.03 to 10 mg/L (Sorg and Logsdon, 1978; Sorg, 1985; U.S. EPA, 1985).

Pilot plant studies confirmed the jar tests results. Pilot-scale experiments demonstrated approximately 50% and 10% reductions of Se(IV) (0.028 mg/L) and Se(VI) (0.038 mg/L), respectively, at a pH of 11.3. These percentages correspond to calculated reductions of finished water concentrations of 0.013 mg/L for Se(IV) and 0.034 mg/L for Se(VI) in groundwater (Sorg and Logsdon, 1978; Sorg, 1985).

7.1.2 Ion exchange

Ion exchange is a physicochemical process in which there is an exchange of ions in the raw water with ions within the solid phase of a resin. As raw water ions displace ions on the resin, the capacity of the resin is gradually exhausted, resulting in finished water concentrations that increase (i.e., contaminant breakthrough). Once the resin has reached its capacity (i.e., when all the resin sites are occupied by the contaminant ion), the resin must be regenerated to reverse the process. The presence of organic contaminants, suspended solids, calcium or iron can cause fouling of the ion exchange resins.

Removal efficiency of greater than 80% is considered to be achievable by strong base anion (SBA) exchange resins for selenium in drinking water (U.S. EPA, 1990b, c). Factors affecting selenium removal by ion exchange include the oxidation state of selenium, the concentration of competing anions and the type of the selected resin. Selenium is usually present at trace concentrations in the drinking water, and the efficiency of its removal is controlled by the concentrations of common drinking water anions, such as sulphate, nitrate, chloride and bicarbonate. The ion exchange behaviour of Se(IV) was found to be similar to that of nitrate, whereas the behaviour of Se(VI) was identical to that of sulphate (Maneval et al., 1985). However, Boegel and Clifford (1986) found that nitrate was preferred over Se(IV) and that Se(VI) was clearly preferred over sulphate. Pilot studies should be conducted in order to verify the effectiveness of ion exchange technology for site-specific water quality. Removal of Se(IV)

The low position of the Se(IV) ion in the ion selectivity sequence suggests that ion exchange is not favourable for Se(IV). The strong-base anion exchange resins have less preference for Se(IV) anions in comparison with Se(VI), nitrate and sulphate (Maneval et al., 1985; Boegel and Clifford, 1986; Li and Viraraghavan, 1998). Another factor affecting the ion exchange removal of Se(IV) is the pH of the treated water. The ion exchange behaviour of the monovalent biselenite form (HSeO3−) and divalent form (SeO32−) should be considered, as the monovalent anion is less preferred than the divalent anion (Maneval et al., 1985).

Clifford (1999) indicated that "chromatographic peaking" of Se(IV) may occur. Chromatographic peaking is a process in which less preferred ions will be concentrated in the column and will, at some time, exit the column in concentrations exceeding their influent concentrations. A laboratory ion exchange column treated mineralized synthetic groundwater containing total dissolved solids at a concentration of 712 mg/L, sulphate at a concentration of 192 mg/L and Se(IV) at a concentration of 0.1 mg/L. The strong-base anion exchange unit achieved a Se(IV) concentration of 0.01 mg/L, with run lengths of 152 bed volumes and an empty bed contact time of 5 minutes. Chromatographic peaking of 0.54 mg Se(IV)/L (5.4 times the feed concentration) occurred at 237 bed volumes (Boegel and Clifford, 1986).

A laboratory experiment using strong-base type I quaternary anion exchange resin was capable of reducing the Se(IV) concentration from 0.1 mg/L to 0.01 mg/L with run lengths of 238 bed volumes, and an adsorption capacity of 23.3 mg Se(IV) per litre of media. Although empty bed contact time (EBCT) was not provided in the published article, this article was based on the author's thesis (Li, 1998), which stated an EBCT of approximately 2.87 minutes.  The resin was exhausted after 18 hours of operation, achieving 371 bed volumes and adsorption capacity of 31.9 mg per litre of media. The experiment was conducted with spiked tap water and in the presence of 366 mg/L sulphate concentration (Li and Viraraghavan, 1998; Li, 1998). Although the authors have not observed chromatographic peaking, it is a major operational consideration when using anion exchange for Se(IV) treatment. Chromatographic peaking causes the effluent Se(IV) concentration to be greater than the influent concentration due to the presence of sulphate and nitrate ions, which displace Se(IV) ions on the resin (Clifford, 1999).

Modified cation exchange resins and chelating resins have been used for Se(IV) removal. Maneval et al. (1985) showed that a weak acid cation resin loaded with ferric ions could be used to remove selenite from water containing sulphate and chloride ions. In a later in-depth study, laboratory tests using chelating polymer resin with immobilized copper(II) ions in the solid phase showed a high affinity towards Se(IV) anions, over Se(VI), sulphate and chloride anions. A fix-bed ion exchange column reduced a Se(IV) concentration to below 0.01 mg/L from an influent concentration of 2 mg/L in the presence of Se(VI) at a concentration of 2 mg/L, sulphate at a concentration of 100 mg/L and chloride at a concentration of 200 mg/L, at pH 9.5 and with an empty bed contact time of 0.21 hour (12.6 minutes). The experiments demonstrated that the ion exchange resin selectively reduced the Se(IV) concentration, as it achieved greater than 1000 bed volumes of treated water for Se(IV) and 200 bed volumes for Se(VI) (Ramana and Sengupta, 1992). Although the authors indicated that regeneration of the resin is possible using brine or sodium carbonate, regeneration and reuse of copper loaded chelating resin may be challenging. Removal of Se(VI)

Strong-base anion exchange resins demonstrated greater removal efficiency for Se(VI) than for Se(IV) (Boegel and Clifford, 1986; Li and Viraraghavan, 1998). However, the effectiveness of the Se(VI) removal is limited by the presence of sulphate ions, as the affinity of sulphate is nearly as great at that of Se(VI), and sulphate, typically present in a much higher concentration, compete strongly with Se(VI),  for ion exchange sites (Maneval et al., 1985; Boegel and Clifford, 1986). Boegel and Clifford (1986) predicted that the number of bed volumes that can be treated before Se(VI) breakthrough occurs may increase when the sulphate concentration is decreased. An increase of greater than 700 bed volumes was expected when a strong-base anion resin treated source water containing sulphate at a concentration of 50 mg/L in comparison with a sulphate concentration of 192 mg/L.

Pilot-scale experiments using a SBA exchange resin, primarily developed for selective nitrate removal in drinking water, showed a high affinity towards Se(VI) anions (Cousin et al., 2011). The experiments reported that the removal of selenium was concurrent to the removal of the competing sulphate and chloride anions. The authors also reported results for Se(VI) removal with both fresh and regenerated resin. Data indicated that after the second regeneration cycle, the  SBA exchange resin was capable of reducing an influent Se(VI) concentration in groundwater from 30.1µg/L to 0.5 µg/L with run lengths of 920 bed volumes, while the nitrate concentration was reduced from approximately 20 mg/L to 3.44 mg/L.. The Se(VI) concentrations in the water samples treated by SBA exchange resin increased gradually over time and ranged from below 0.5 µg/L after 64 hours (920 bed volumes)  to 8.3 µg/L after 84 hours of treatment (1240 bed volumes). The nitrate concentration in the samples increased from 3.44 mg/L to 16.8 mg/l after 84 hours (1240 bed volumes) (Cousin et al., 2011).

Laboratory experiments evaluated a SBA resin in chloride form for the removal of Se(VI) from mineralized synthetic water containing total dissolved solids at a concentration of 712 mg/L and sulphate at a concentration of 192 mg/L. The ion exchange column treated influent Se(VI) concentrations ranging from 0.1 mg/L to 0.01 mg/L, achieving run lengths of 235 bed volumes. The ion exchange column operated at a pH of 8.3 and an empty bed contact time of 5 minutes. Se(VI) was eluted after the sulphate ions and was not subject to chromatographic peaking (Boegel and Clifford, 1986). Another laboratory experiment found that a strong-base type I anion exchange resin was capable of reducing the Se(VI) concentration from 0.1 mg/L to 0.01 mg/L with run lengths of 361 bed volumes, achieving an adsorption capacity of 36.3 mg Se(VI) per litre of media. Although empty bed contact time (EBCT) was not provided in the published article, this article was based on the author's thesis (Li, 1998), which stated an EBCT of approximately 2.87 minutes.The resin was fully exhausted after 28 hours of operation, achieving 595 bed volumes and adsorption capacity of 52.3 mg Se(VI) per litre of media. The experiment was conducted with spiked tap water and in the presence of 366 mg/L sulphate concentration (Li and Viraraghavan, 1998).

Based on laboratory experiments, Clifford (1999) suggested that an ion exchange process in combination with an oxidation pretreatment step of Se(IV) to Se(VI) may be considered as a technically feasible process. A free chlorine concentration of 2 mg/L achieved a 60% oxidation of Se(IV) within 5 minutes at a pH in the range of 6.5-8.0. However, at pH 9, only 15% of Se(IV) could be oxidized in 5 minutes. In this study, hydrogen peroxide and potassium permanganate were found to be less effective, whereas oxygen was found to be ineffective (Clifford, 1999).

A consideration when using strong-base anion exchange resins is the potential for the release of nitrosamines from the resin. Kemper et al. (2009) found that new resin or resin that is exposed to disinfectants (chlorine and chloramines) may release nitrosamines due to shedding of manufacturing impurities. To minimize nitrosamine formation, attention should be paid when selecting strong-base ion exchange resins (Kimoto et al., 1980; Najm and Trussell, 2001).

7.1.3 Membrane processes: reverse osmosis

RO technology is based on forcing water across a membrane under pressure while the ionic species, such as Se(IV) and Se(VI), are retained in the waste stream. The performance of the RO membrane systems depends on a variety of factors, including the quality of the raw water, the type of the membrane, molecular weight cut-off and recovery of the system (Jacangelo et al., 1997). The presence of iron, manganese, silica, scale-producing compounds and turbidity could negatively affect the system performance. A pretreatment of the feed water is required to prevent scaling and fouling of the RO membranes. The product water typically requires post-treatment, consisting of pH and alkalinity adjustments.

RO technology has been shown to be an effective method for the removal of selenium from drinking water. Pilot-scale studies demonstrated that RO may achieve removal efficiency in the range of 75-99% (Sorg et al., 1980; U.S. EPA, 1985, 1989; Huxstep and Sorg, 1988) for selenium in drinking water, and the technology is typically used when high concentrations of other dissolved solids need to be removed (U.S. EPA, 1989). As RO systems generally produce high-quality water, the blending of treated water and raw water to produce finished water of acceptable quality may be a factor in selecting an RO system (U.S. EPA, 1985).

Eight RO systems with varied design capacities in the range from 800 to 1 million gallons per day (Mgd) (0.003-3.8 ML/day) have effectively reduced selenium concentrations in groundwater. The systems used hollow fibre and/or spiral wound cellulose acetate membranes, supplied by six different manufacturers. Data from RO systems with a capacity ranging from 0.115 to 3.8 ML/day indicated that the selenium concentrations in the range of 0.014-0.025 mg/L were lowered to below the detectable level (0.005 mg/L). These systems operated with design water recovery in the range of 50-75% and a feed pressure of 2800-2900 kPa (400-425 psi). Smaller systems (0.003-0.019 ML/day) having a design water recovery of 35-50% and a feed pressure of 1400-2800 kPa (200-400 psi) achieved a finished water concentration below the detectable level (0.005 mg/L) from a feed concentration in the range of 0.015-0.025 mg/L. All treatment configurations consisted of pretreatment of the raw water, RO unit and post-treatment. Pretreatment included filtration, pH adjustment to 6.0-6.2 as well as calcium and magnesium sequestration. Post-treatment consisted of pH adjustment, degassing and disinfection (Sorg et al., 1980).

Pilot-scale testing evaluated the effectiveness of five different RO membranes for the rejection of inorganic contaminants. Each membrane has been tested according to the manufacturer's specifications. The feed water pressure, water recoveries and product water flow rates differed between membrane elements. The investigations demonstrated a high rejection rate in the range from 95% to 99% and from 98% to greater than 99% for Se(IV) and Se(VI), respectively, under a variety of operating conditions. The reported influent concentrations ranged from 0.33 to 1.5 mg/L for Se(IV) and from 0.61 to 2.7 mg/L for Se(VI) (Huxstep and Sorg, 1988).

Considerations when using RO treatment include disposal of the reject water and possible increased corrosivity of the treated water (Schock and Lytle, 2011). RO rejects a significant portion of the influent water as contaminant-rich brine (Taylor and Wiesner, 1999), and the concentrate discharge must be disposed of appropriately. The removal of contaminants can cause mineral imbalances that could increase the corrosive nature of the treated water (Schock and Lytle, 2011). In most cases, post-treatment corrosion control measures need to be undertaken.

7.1.4 Adsorption Activated alumina

The removal of selenium from drinking water by activated alumina has been investigated on a laboratory basis. Activated alumina treatment demonstrated a selenium removal in the range of 85-95%. However, the feasibility of the treatment process depends on the selenium species in the raw water, as the activated alumina preferentially adsorbs Se(IV) (U.S. EPA, 1985, 1989).

Activated alumina is a physicochemical process by which ions in the feed water are sorbed to the oxidized activated alumina surface. Activated alumina is used in packed beds, which may operate in series or parallel. Feed water is continuously passed through the packed bed. The contaminant ions in the water are exchanged with the surface hydroxides on the alumina. When adsorption sites on the activated alumina surface become filled, the bed must be regenerated. Regeneration of activated alumina is accomplished through a sequence of rinsing with regenerant (sodium hydroxide), flushing with water and neutralizing with acid, such as sulphuric acid for Se(IV) recovery and hydrochloric acid for Se(VI) recovery (U.S. EPA, 1998).

Studies have shown that activated alumina is an effective treatment technique for the removal of inorganic contaminants, including arsenic, selenium, fluoride and silica. As a result of the amphoteric nature of activated alumina, the adsorption process is influenced by pH. Below pH 8.2 (a typical zero point charge for activated alumina), the activated alumina surface has a net positive charge, and it will adsorb anions found in the water (Clifford, 1999). Factors such as pH, contaminant oxidation state, regenerant dose and flow rate, competing ions and empty bed contact time can influence the inorganic contaminant removal by activated alumina. When employing activated alumina technology, operational issues that must be considered include the degradation of activated alumina through the regeneration process and the fouling of the activated alumina bed, resulting in an increase in headloss across the media bed. The activated alumina process also has the potential for "chromatographic peaking" where the effluent selenite concentration would exceed its influent concentration due to the presence of more preferred ion such as fluoride or phosphate in the influent water.

Activated alumina may not be suited for small systems because of the special operational requirements. The technology requires adequate surveillance and maintenance, including the use of concentrated acids (sulphuric acid or hydrochloric acid) and base (sodium hydroxide) for regeneration of activated alumina. These can be hazardous, particularly if the operator's knowledge and skills are insufficient for handling hazardous materials. Utilities need to consider the chemical handling and disposal requirements prior to selecting this treatment technology (U.S. EPA, 1989, 1998).

Adsorption of Se(IV)

A laboratory-scale continuous-flow column of activated alumina studied the removal efficiency of Se(IV) from synthetic well water (Trussell et al., 1980). The study developed activated alumina breakthrough capacities for three different influent Se(IV) concentrations, pH in the range of 5-7 and a surface loading rate of 3 gpm/ft2 (7.3 m/h) and an EBCT of 1.87 min which can be calculated for the 9 inch deep column and the flow rate used, in the study. The breakthrough capacity has been defined as the amount of selenium adsorbed per litre of activated alumina before the effluent selenium concentration exceeded the treatment goal of 0.01 mg/L. Trussell et al. (1980) reported the following breakthrough capacity for activated alumina:

Table 2. Breakthrough capacities for activated alumina
pH Influent Se (IV) concentrations Bed Volumes
0.05 mg/L 0.1 mg/L 0.2 mg/L
Breakthrough capacity (mg Se(IV)/L activated alumina)
5 60 120 235 1200
6 45 90 175 900
7 25 50 100 500

Activated alumina showed optimum adsorption capacity for Se(IV) at pH in the range of 5-6, and the capacity of the media was proportional to the initial selenium concentration in the raw water. The study investigated the impact of various ions on the adsorption efficiency of Se(IV) by activated alumina. Whereas bicarbonate ions had the more pronounced effect on Se(IV) adsorption, chloride, nitrate and sulphate ions showed only marginal interference. Based on the graphical representation of the experimental data, an increase in bicarbonate concentration from 50 to 200 mg/L reduced the removal efficiency of Se(IV) by approximately 10% at a pH in the range of 6.0-6.5. Calcium, magnesium and sodium cations, at concentrations as high as 200 mg/L, did not have a negative impact on the adsorption of Se(IV).

Activated alumina was regenerated with 0.5% sodium hydroxide at a flow rate of 0.5 gpm/ft2 (1.2 m/h). The regenerant's flow rate demonstrated a great effect on Se(IV) recovery and its subsequent removal. Experiments showed that the percent recovery of Se(IV) during the regeneration was increased twice when the regenerant flow rate was decreased from 1.0 gpm/ft2 to 0.5 gpm/ft2 (2.4 to 1.2 m/h) EBCT (calculated from the information provided n the study), increased from 5.6 minutes to 11.2 minutes for 9 inch deep column.

Adsorption of Se(VI)

As Se(VI) has a low position in the selectivity series of activated alumina, it is more susceptible to interference with adsorption. The experiments found that the adsorption capacity of activated alumina with respect to Se(VI) was approximately 1/13th of the capacity for Se(IV) under similar conditions (Trussell et al., 1980; Kreft, 1985).

Sulphate ions strongly interfered with Se(VI) removal by activated alumina. Trussell et al. (1980) reported that the number of bed volumes that can be treated before Se(VI) breakthrough occurs may decrease significantly when sulphate is present. An increase of the sulphate concentration from 5 to 500 mg/L would decrease the bed volumes of the treated water from 450 to 15 at pH 6. These capacities have been reported for an influent Se(VI) concentration of 0.05 mg/L in the presence of a bicarbonate concentration of 100 mg/L. At a low ratio of sulphate to Se(VI), the adsorption capacity of the activated alumina for Se(VI) was increased. Similar tests showed that the effect of bicarbonate concentration was not as great as that of sulphate. An increase of the bicarbonate concentration from 5 to 500 mg/L decreased the bed volumes of the treated water from 125 to 33 at pH 6. Alkalinity tests have been conducted for an influent Se(VI) concentration of 0.05 mg/L and sulphate concentration of 100 mg/L. Chloride and nitrate ions had no pronounced effect on Se(VI) adsorption. Sodium, magnesium and calcium, at concentrations as high as 200 mg/L for each, did not negatively affect the adsorption of Se(VI). The study found that the calcium and magnesium concentrations may slightly enhance the adsorption of Se(VI), due to the "secondary adsorption" phenomenon. Secondary adsorption occurs as a joint adsorption of anions with multivalent cations or as a joint adsorption of cations with multivalent anions (Trussell et al., 1980).

A laboratory-scale study was conducted using a continuous-flow column of activated alumina, treating synthetic well water to Se(VI) levels of 0.01 mg/L with run lengths of 100, 70 and 35 bed volumes (adsorption capacities of 4.5, 3.2 and 1.6 mg/L alumina, respectively), at pHs of 5, 6 and 7, respectively. The treatment goal of 0.01 mg/L was achieved by treating an influent Se(VI) concentration of 0.05 mg/L in the presence of sulphate anions at a concentration of 100 mg/L. The study used 0.5% sodium hydroxide for the regeneration of activated alumina at a flow rate of 2 gpm/ft2 (4.8 m/h) and hydrochloric acid for the neutralization of the bed (Trussell et al., 1980).

7.1.5 Electrodialysis/electrodialysis reversal

Electrodialysis is an electrochemical separation process in which charged species from water are transported through semipermeable membranes under the influence of an electric potential. The membranes are configured in "stacks" parallel to one another, and each successive membrane carries direct electric current. Cations and anions migrate through the cation and anion membranes, respectively. In electrodialysis reversal, the polarity of the electrodes is changed periodically across the ion exchange membranes, causing a reversal in ion movement. This step minimizes the scale build-up on the membranes, and thus electrodialysis reversal can operate for a longer period of time between cleaning. Electrodialysis is generally automated and allows for part-time operation, and it may be an appropriate technology for small systems (U.S. EPA, 1998).

A field test reported that a selenium concentration of 0.05 mg/L in the groundwater was reduced to 0.002 mg/L. No information was provided on the operational conditions of the electrodialysis systems (U.S. EPA, 1985). Data from two mobile units treating public water supply demonstrated that electrodialysis reversal was capable of reducing selenium concentrations in source water. This study demonstrated an average selenium reduction of 71% from an influent concentration in the range of 0.005-0.0075 mg/L in well water. The pretreatment of the raw water included carbon filtration (Folster et al., 1980; U.S. EPA, 1991c).

Utilities planning to utilize electrodialysis for the reduction of selenium, total dissolved solids and other trace metals would require pilot plant testing of the feed water in order to verify the effectiveness of selenium removal.

7.1.6 Manganese greensand filtration

The active component in "greensand" is glauconite, a green, iron-rich mineral that has ion exchange properties. In manganese greensand filtration, the soluble metals in the water, such as iron and manganese, are oxidized and precipitated when they come in contact with oxides of manganese on the greensand granules. When the manganese greensand bed is exhausted, the bed is regenerated to restore its oxidizing capacity.

Laboratory column tests (Li and Viraraghavan, 1998) studied the efficiency of manganese greensand filtration for the removal of Se(IV) and Se(VI) from spiked tap water. The column was capable of reducing the feed water Se(IV) concentration of 0.1 mg/L to 0.01 mg/L (90% removal), achieving an adsorption capacity of 1.73 mg/L media. It was found that the addition of ferric chloride to the raw water enhanced the removal of Se(IV). Selenium(IV) adsorption capacities of 2.37 and 3.2 mg/L of media have been reported when the iron to Se(IV) ratios were 10:1 and 20:1, respectively. The process has been reported to be ineffective for Se(VI) (Li and Viraraghavan, 1998).

7.1.7 Emerging treatment technologies

Emerging treatment technologies include the following:

  • Iron based materials: Granular ferric hydroxide (GFH) is an iron based adsorptive media currently used for treatment of arsenic in drinking water. Full-scales GFH media-based systems for arsenic removal reported that the media was capable of removing selenium in drinking water. However, no operational data for selenium have been reported (Cumming et al., 2007).
    Ferric oxide media is an iron based media that adsorbs arsenic and other ions including selenium (U.S. EPA, 2005). Effective adsorption occurs at pH values ranging between 6.0 and 8.0. At pH values greater than 8.0 to 8.5, pH adjustment is recommended to ensure adsorption capacity is maintained. Silica and phosphate are the competing ions that can reduce the adsorption capacity. No operational data for selenium have been reported for selenium removal.
    Several researchers have studied iron oxide-coated sand (Lo and Chen, 1997; Li and Viraraghavan, 1998), aluminum oxide-coated sand (Kuan et al., 1998) and natural iron oxides (Zhang and Sparks, 1990; Rovira et al., 2008) for the removal of selenium from water. The studies reported that both selenium species had a different degree of adsorption on the iron and aluminum oxides and that Se(IV) was found to adsorb more readily than Se(VI).
    A laboratory column study found that selenium removal by iron oxide-coated sand was possible for Se(IV). The study demonstrated a 90% reduction of a Se(IV) concentration of 0.1 mg/L in spiked tap water, achieving an adsorption capacity of 2.7 mg Se(IV)/L of  media. However, the process was reported as being ineffective for Se(VI) (Li and Viraraghavan, 1998). Lo and Chen (1997) reported practically complete removal of Se(IV) with an initial concentration of 10 mg/L after 10 minutes in contact with 100 g/L iron oxide-coated sand. Se(VI) removal required approximately 90 minutes for a similar reduction at the same initial concentration. The authors reported a Se(IV) adsorption capacity in the range of 0.014-0.017 mmol/g and 0.013-0.014 mmol/g for iron oxide-coated sand (Lo and Chen, 1997). In laboratory experiments, iron(II) hydroxide was capable of reducing an influent selenium concentration of 0.004 mg/L to below 0.001 mg/L at pH 8.8 (Zingaro et al., 1997). Another experiment, using aluminum oxide-coated sand, was conducted for the removal of selenium from water. The study reported an adsorption capacity of approximately 0.5 mg/g media for Se(IV) and 0.23 mg/g media for Se(VI) at pH in the range of 7.18-8.3 (Kuan et al., 1998).
  • Adsorption with polymer-clay composite:Several polymer-clay composites have been designed and tested for selenium removal from water. An adsorption capacity of 18.4 mg Se(VI) per gram media was obtained for a chitosan-clay composite. The composite showed selectivity towards selenium in the presence of sulphate and was able to reduce a Se(VI) concentration of 0.64 mg/L in groundwater to below 0.01 mg/L (Bleiman and Michael, 2010).
  • Dissimilatory reduction: Biological treatment uses microorganisms to reduce, oxidize or eliminate groundwater contaminants, either as the sole treatment technique or combined with other conventional physicochemical processes, such as sorption and filtration. The basic principle of the biological treatment is that remediation takes place as the result of oxidation-reduction potential changes (Zouboulis and Katsoyiannis, 2005). The metabolism of dissimilatory metal-reducing microorganisms may reduce uranium, selenium, chromium and possibly other metals to insoluble forms that can be removed from contaminated waters (Lovley, 1995).

7.1.8 Distribution system materials: non-leaded brass alloys

Typically, non-leaded brass alloys contain lead in the range of 0.1-0.25% by weight as an incidental impurity from the recycled materials or ores used as the source metals. Metals such as bismuth and selenium are added to the alloys to replace the lead and improve the mechanical characteristics. Data on the potential of non-leaded brass to leach metals such as selenium are limited, and additional studies need to be undertaken in order to gain a better understanding of the leaching propensity of these alloys.

Sandvig et al. (2007) synthesized the current state of knowledge related to the use of non-leaded brass components in drinking water, identified and prioritized recommended research needs related to non-leaded brass and provided a preliminary structure for the highest-priority projects to meet these needs. Based on this report, the Water Research Foundation considered the potential of the non-leaded brass alloys to leach metals, including selenium, as one of the highest-priority research needs.

Sandvig et al. (2012) compared the leaching characteristics of typical utility service connections and premise plumbing devices under the relevant NSF/ANSI 61 Standard test conditions. The objective of the study was to determine whether the current testing conditions set out in the standard for use with chlorinated water can also be used to accurately reflect the release of metals (including lead, copper and selenium) in chloraminated water with sufficient accuracy to meet the public health and regulatory needs of utilities. The project also evaluated the leaching of other metals, including selenium, from leaded- and non-leaded brass devices (e.g., ball valve, corporation stop, water meter, kitchen faucet). Six different types of the devices (triplicate samples), were tested under Section 8 of the NSF/ANSI 61 Standard with 11 types of source water. The normalized metal data showed that selenium was detected in only one sample (0.8 µg/L) of the 198 samples analyzed (reported detection level was 0.5 µg/L). Four different types devices (triplicate samples) were tested under Section 9 of the NSF/ANSI 61 Standard with 8 types of source water. The normalized metal data indicated that selenium was not detected in any of the 96 samples analyzed (reported detection level in the range 0.2 - 2 µg/L).

A Water Research Foundation project (No. 4191) identified and prioritized key water quality characteristics and changes that might adversely impact non-leaded brasses and increase their leaching in drinking water distribution systems. Five different water qualities, including both surface water and groundwater of varying alkalinity, hardness, total organic carbon, pH and secondary disinfectant, were used to study the suitability of non-leaded brass materials for use in drinking water. Preliminary results showed no leaching of selenium in the four non-leaded brass materials tested, indicating that they are suitable for use in drinking water of various water qualities (Turković et al., 2014).

Another experiment was conducted with bismuth/selenium-modified cast red brass alloy containing 0.016% lead, 2.23% bismuth and 1.02% selenium. The tested specimens were large faucet body castings. According to the report, the 19th-day mean normalized leached selenium concentrations were below the NSF's maximum allowable level, at that time, of 12.5 µg/L. The test protocol used in the study was not specified (Peters, 1995).

Recent laboratory experiments have been conducted to quantify the levels of the metal leached from two commercially available non-leaded brass alloys: Envirobrass® I and Envirobrass® II. Envirobrass® I alloy may contain selenium in the range of 0.35-0.75%, bismuth from 0.5% to 1.15% and lead at 0.25%, and Envirobrass® II may contain selenium in the range of 0.8-1.1%, bismuth from 1.6% to 2.2% and lead at 0.25%. The tests were conducted with two different leaching solutions: a synthetic test water (NSF/ANSI Standard 61 section 9 methodology) and a solution identified as aggressive for lead leaching from brass alloys. Selenium concentrations leached from these alloys were below the detection limit of 2.98 µg/L for both tested leaching solutions (Triantafyllidou and Edwards, 2006).

7.1.9 Selenium in the distribution system

Contaminants may enter the distribution system through a variety of mechanisms: 1) in a dissolved state from the source water; 2) attached to turbidity particles; 3) added to the source water from the treatment techniques; and/or 4) as by-products of corrosion of piping and plumbing material. The fate and transport of the contaminants and their subsequent accumulation and release within the distribution system are complex processes controlled by a number of chemical, physical and microbial mechanisms. Inorganic contaminants may accumulate on or within materials commonly found in the water distribution system, such as scales, biofilms or sediments, and may be released back to the distribution system water (Schock, 2005; U.S. EPA, 2006).

Case studies, based on available published literature, characterized the potential for the accumulation and release of inorganic contaminants in the distribution system. Schock (2005) found that the scale material from a lead service line contained selenium at concentrations of 0.3, 0.5, 1 and 7.8 mg/kg. However, the stability of the metals accumulated in the scales is unpredictable, and their concentration in the bulk water is not quantifiable.

7.2 Residential scale

Generally, it is not recommended that drinking water treatment devices be used to provide additional treatment to municipally treated drinking water. In cases where an individual household obtains its drinking water from a private well, a residential drinking water treatment device may be an option for reducing selenium concentrations in drinking water. The treatment processes that are capable and able to be certified for selenium removal at the residential scale include adsorption, reverse osmosis (RO) and distillation.

Before a treatment device is installed, the well water should be tested to determine general water chemistry and to verify the concentration of selenium. The testing should also include assessing the presence and concentration of competing ions (e.g., sulphate, nitrate, chloride) and organic matter in the water, which could interfere with selenium removal. Devices can lose removal capacity through usage and time and need to be maintained and/or replaced. Consumers should verify the expected longevity of the components in their treatment device as per the manufacturer's recommendations.

Health Canada does not recommend specific brands of drinking water treatment devices, but it strongly recommends that consumers use devices that have been certified by an accredited certification body as meeting the appropriate NSF/ANSI drinking water treatment unit standards. These standards have been designed to safeguard drinking water by helping to ensure the material safety and performance of products that come into contact with drinking water. Currently, there are certified devices for the reduction of selenium from drinking water that rely on RO and distillation treatment processes.

Certification organizations provide assurance that a product conforms to applicable standards and must be accredited by the Standards Council of Canada (SCC). In Canada, the following organizations have been accredited by the SCC to certify drinking water devices and materials as meeting NSF/ANSI standards (SCC, 2014):

An up-to-date list of accredited certification organizations can be obtained from the SCC (

Water treatment technologies able to be certified to NSF standards for reduction of selenium include adsorption, RO and distillation. Applicable standards are NSF/ANSI Standards 53, 58 and 62.

Treatment devices to remove selenium from untreated water (e.g., a private well) can be certified for either the removal of selenium alone or the removal by surrogate testing.

Drinking water treatment devices certified to NSF/ANSI Standards 53, 58 and 62 specifically for selenium removal, must be capable of reducing the concentration of selenium in water from an influent (challenge) concentration of 0.1 mg/L (added as 0.05 mg/L for Se(IV) and 0.05 mg/L for Se(VI)) to a maximum final (effluent) concentration of 0.05 mg/L (NSF/ANSI, 2009a,b, 2011b). However, treatment devices certified to NSF/ANSI Standard 62 to remove selenium can be certified either specifically for selenium as noted above or for the removal of total dissolved solids (TDS) which is used as a surrogate for selenium in this standard. Treatment devices certified to NSF/ANSI Standard 62 using TDS as a surrogate, must achieve a minimum TDS reduction of 99.0% from an influent (challenge) concentration of 1000 mg/L(NSF/ANSI, 2009b).

RO systems certified to NSF/ANSI Standard 58 (Reverse Osmosis Drinking Water Treatment Systems) are intended for point-of-use installation only. RO requires larger quantities of influent water to obtain the required volume of drinking water, because these systems reject part of the influent water. A consumer may need to pretreat the influent water to reduce fouling and extend the service life of the membrane.

Distillation systems certified to NSF/ANSI Standard 62 (Drinking Water Distillation Systems) are also intended for point-of-use installation only. The distillation process is effective for the reduction of inorganic contaminants, but requires an electrical energy input.

NSF/ANSI Standard 61 (Drinking Water System Components--Health Effects) limits the leaching of selenium into drinking water. The standard ensures that materials meet health-based leaching requirements and are safe for use in potable water applications. When materials are certified to the standard, the concentration of selenium must not exceed the single product allowable concentration of 0.005 mg/L (NSF/ANSI, 2011a).

Report a problem or mistake on this page
Please select all that apply:

Privacy statement

Thank you for your help!

You will not receive a reply. For enquiries, contact us.

Date modified: