Arsenic removal from drinking water using low-pressure nanofiltration under various operating conditions

Arsenic (As) is toxic and fatal. It is almost tasteless and has no odor. Consumers cannot, therefore, recognize it in water, which could lead to a slow and painful death.

Arsenic is a natural element in the earth's crust. As water flows over arsenic-containing rocks, they are dissolved and transported to ground- and/or surface- waters. They can also mix with drinking water as a result of industrial and agricultural activities (Nriagu 1994; Meng et al. 2000; Younos & Grady 2014). Arsenic is found in groundwater in two forms (valencies or oxidation states): 1) pentavalent arsenic (also known as As(V), As⁵⁺, or arsenate) and 2) trivalent arsenic (As(III), As³⁺, or arsenite). Their relative abundance depends essentially upon pH and redox conditions. Sometimes, a combination of them can be found in groundwater (Gomez-Caminero et al. 2001; Lescano et al. 2015). At typical pH values for drinking water (between 6 and 9), arsenic is often found in the inorganic forms, arsenite or arsenate. As(III) species are neutral (H₃AsO₃) while As(V) are found in the anionic forms HAsO₄⁻² and H₂AsO⁻⁴ (Chen et al. 1985; Karim 2000; Morales et al. 2000; Jiang 2001). Cancers, including those of the bladder, kidney, skin, lung, liver and prostate, are the main chronic diseases caused by arsenic, and peripheral vascular diseases are widely reported (Harisha et al. 2010; Younos & Grady 2014). Long-term ingestion of 0.21 mg-As/L is reported to be detrimental (Crittenden et al. 2012). Another study showed that the average lifetime risk of death from liver, lung, bladder, and kidney cancer combined is 13.4 in 1,000 from consuming only 1 L/d of water containing 50 μg-As/L (Smith et al. 1992). It has also been demonstrated that an arsenic maximum contaminant level (MCL) of 2 μg-As/L would decrease risk cancer levels to below 1 in 10,000 (Hering & Elimelech 1996).

It is important to determine which arsenic species is/are present in any water before selecting a treatment method (Wang et al. 2010). Statistics show that arsenic toxicity is widespread across the world, particularly in developing countries, and some 10 to 20% of the population are prone to arsenic toxicity (Jiang 2001). In Argentina, Bangladesh, Chile, China, India, Mexico, and the United States, native inorganic arsenic in groundwater presents a public health threat (Younos & Grady 2014). In various Iranian provinces, including East Azerbaijan, Kurdstan, and the city of Bijar, arsenic has also been found. The concentration of arsenic in groundwater is higher than that in surface water, e.g., rivers and lakes. Thus, controlling arsenic concentration within acceptable limits is important where groundwater is used for drinking (Davis et al. 1996).

Inorganic arsenic groups are more active than organic ones. It is noted in this context, too, that arsenite groups are more toxic, soluble and active than those of arsenate, and are also 10 to 60 times more poisonous than the latter (Gomez-Caminero et al. 2001; Lescano et al. 2015).

The United States Environmental Protection Agency (USEPA) declared on January 22, 2001, that arsenic's MCL would be lowered ‘from 50 to 10 ppb’ [μg/L] and systems must comply by January 23, 2006 (USEPA 2001). More than 130 million people are thought to be potentially exposed to arsenate in drinking water at concentrations above the guideline (Meher et al. 2015). USEPA calculated that reaching this arsenic standard will involve 3,000 water systems in the US, serving 11 million people, to take action (Han et al. 2002).

The most common arsenic removal techniques are 1) precipitation or coagulation using metal salts; 2) precipitation or adsorption on activated alumina, and oxides or hydroxides of mainly iron; 3) ion-exchange; 4) desalination – e.g., reverse osmosis (RO), nanofiltration (NF), electrodialysis (ED) and distillation (Meng et al. 2000; Korngold et al. 2001; Cakmakci et al. 2009; Elcik et al. 2013; Jasrotia et al. 2013; Chang et al. 2014). Nowadays, distillation, NF, and RO (with pretreatment) are the most common arsenic removal methods (Harisha et al. 2010). Lescano et al. (2015) studied As(III) and As(V) removal from aqueous media, using three commercial adsorbents (titanium dioxide, granular ferric hydroxide, and activated alumina). Many studies have shown that titanium-based materials seem to be the most appropriate adsorbent, followed by granular ferric hydroxide (Lescano et al. 2015). Shorney et al. (2001) found good arsenic removal using ferric sulfate (Shorney et al. 2001; AWWA 2005). Other studies have shown that alumina could achieve 85 to 99% arsenic removal (Hammad Khan et al. 2013). In a more recent study almost 100% arsenic removal from drinking water was reported using magnetic iron oxide nanoparticles coated with sand (MIONCS) from an initial concentration of 87.0 μg/L at pH 7, within 25 minutes (Afzali et al. 2016). Korngold et al. (2001) showed that in water containing relatively low concentration of anions, use of a strong-base anion-exchange resin makes 99% arsenic removal possible.

On the other hand, some studies have related to arsenic removal by NF membranes (Thompson & Chowdhury 1993; Chang et al. 1994; Waypa et al. 1997; Brandhuber & Amy 1998; Urase et al. 1998; Vrijenhoek & Waypa 2000; Seidel et al. 2001; Saitúa et al. 2005). In this field, the efficiency of arsenic removal is a function of organic matter concentration (Yu et al. 2013; Jiang et al. 2016). It was shown that As(V) rejection was reduced by increasing the ionic strength and concentration of ions such as sulfate, phosphate, and calcium (Fang & Deng 2014). Also more than 90% arsenic removal was by adding a small amount of nanoscale zero valent iron (0.2 g nZVI/L) to the water (Nguyen et al. 2009). Harisha et al. (2010) achieved 99.80% arsenate removal using high pressure-NF (50 bar). All these studies reported that arsenate removal was proportionately higher than arsenite (Sato et al. 2002; Xia et al. 2007).

In this study, the feed was groundwater, and arsenite and arsenate were added to the samples. To vary the lab conditions, four solutions (arsenite and arsenate) at concentrations of 100, 300, 500 and 1,000 μg/L were prepared. Sodium arsenate dibasic heptahydrate (Na₂HAsO₄.7H₂O, manufactured by Merck, Germany) was used for the arsenate solution, and sodium arsenite dibasic (Na₂AsO₂ manufactured by Fluk, Germany) for the arsenite. Two solutions, of arsenate and arsenite respectively, were prepared at 1,000 μg/L. The more dilute solutions – i.e., 100, 300, 500 and 1,000 μg/L – were prepared by diluting the 1,000 μg/L solution with distilled water. A pilot arsenic removal membrane was made – see Figure 1. The membrane module was spiral-wound, thin-film composite and aromatic polyamide, model NF90-4040, manufactured by Dow Filmtec. The membrane specification is given in Table 1.

The experiments were based on the standard methods (APHA 2005), and the permeate and reject streams were collected in the same tank. Some preliminary studies were carried out using water without added arsenic, to try to find system problems, etc. The tank was also rinsed several times with nitric acid and then distilled water, to remove all pollutants.

Water was passed through a spiral copper tube to control and maintain the temperature in the tank. To determine the membrane's removal efficiency for arsenite and arsenate, the different parameters could all be varied or kept constant. Thus, a constant pressure of 5 bar, at constant 28 °C and pH 8, were used with input concentrations of 100, 300, 500 and 1,000 μg/L for both arsenate and arsenite. Pressure effects were tested – 4, 5, 6 and 7 bar – using a constant input concentration of 500 μg/L, at 28 °C and pH 8. The effect of temperature was also studied with constant arsenate and arsenite input concentrations of 500 μg/L, at constant 5 bar and pH 8. The temperatures tested were 28, 31, 34, and 37 °C.

At the start, the tank was filled with groundwater, which was transferred to the cartridges and then to the membrane. From the membrane, it was returned to the tank. Samples of filtered water were taken from the tank and the arsenic concentration measured. Analysis was done with a Unican 919-AA spectrophotometer, which can determine arsenic at a precision of 1 μg/L. All samples were analyzed three times.

The basic quality characteristics of the groundwater used in the tests are given in Table 2.

The results of varying the initial feed-water arsenite and arsenate concentrations are shown in Figure 2. Under constant lab conditions (P = 5 bar; T = 28 °C; pH = 8), increasing the initial feed-water concentration from 100 to 1,000 μg-As/L, led to declines in both arsenite and arsenate removal efficiency. The highest arsenite removal level was 90% at an input concentration of 100 μg/L, while for arsenate it was 93%, also at a feed concentration of 100 μg/L.

The arsenite and arsenate removal efficiencies decreased at higher concentrations because, as the input concentration increases, the arsenic concentration increases on the membrane surface so that the concentration of accumulated arsenic might also increase. This is likely to enable more atoms to pass through the membrane, reducing removal efficiency. Arsenate removal is thought to have exceeded arsenite removal because of the impact of its negative charge on the membrane surface. Nguyen et al. (2009) showed that about 57% of As(III) and 81% of As(V) were removed from 500 mg-As/L solutions by NF. Other studies have also shown decreasing removal efficiency for arsenic with increasing of initial concentration (Cakmakci et al. 2009; Figoli et al. 2010; Harisha et al. 2010).

The results of varying the trans-membrane pressure are shown in Figure 3. The removal efficiencies for both arsenite and arsenate rose with increasing pressure, from 4 to 7 bar. The highest arsenite and arsenate removal efficiencies occurred at 7 bar, and were 90 and 93%, respectively. (At 4 bar, the respective removal efficiencies were 82 and 86%.) As the removal efficiency at 6 bar pressure is not significantly different from that at 7 bar, and taking account of the cost of electricity, it would probably be more appropriate to use the lower pressure.

The same results as are shown in Figure 3 have been achieved in other, similar studies (Uddin et al. 2007; Harisha et al. 2010; Elcik et al. 2013).

The results of applying various temperatures are compared in Figure 4. As the temperature rises, the arsenate and arsenite removal efficiencies decrease. With the temperature increased from 28 to 37 °C, the arsenate and arsenite removal efficiencies dropped from 87 to 82% and from 86 to 81%, respectively. Similar studies indicated that reducing the operating temperature leads to higher arsenic removal efficiency (Uddin et al. 2007; Figoli et al. 2010).

Increasing the feed arsenic concentration from 100 to 1,000 μg/L, led to reduced arsenate and arsenite removal efficiencies, because of the concentration accumulation on the membrane surface leading to the plugging of filtration passages through the membrane.

As pressure was increased from 4 to 7 bar, arsenate and arsenite removal efficiency increased, probably because of the relationship between trans-membrane pressure and water flux. Thus, as the pressure rises the flux increases. With temperature increases, however, arsenate and arsenite removal efficiency decreased. This is likely to be because arsenic's diffusivity increases with temperature, so that the diffusive transport of arsenic across the membrane is increased.

Under the lab conditions used – pressure 5 bar, temperature 28 °C, pH 8 – and with initial feed concentrations of 100 and 300 μg-As/L, the concentrations of arsenite and arsenate in the NF permeate were below 50 μg/L. Only with a feed concentration of 100 μg/L, however, was the permeate concentration below 10 μg/L. Increasing the pressure had a positive effect on the NF system's efficiency in arsenite and arsenate removal, but increasing the temperature had a negative impact.

The authors are grateful to the city of Bijar Water Authority for providing data and documents. We also wish to thank Dr. Rabbani for his technical and logistical assistance in modeling and analysis works.