Column Experiments on Arsenic Removal Through Adsorption From Water Using Different Natural and Synthetic Adsorbents

With the aim of exploring a best adsorbent from locally available sands for removing arsenic from water, eight different adsorbents are tested through column experiments using those materials as lter bed. Based on earlier batch experimental results ve locally available sands (Scoria, Skye, Iron ore, NT red and TGS), one commercial sand (GFH) and two synthetic sands (IOCS and IOCS-AOCS) were selected for the column experiments. Target was to treat arsenic from water up to WHO standard level of 10 µg/L for a considerable period. It is found that Skye sand is capable to treat arsenic-contaminated water to the WHO standard for the longest period, followed by TGS, Iron ore and NT red sands. Scoria sand is unable to treat water up to the WHO standard. Although, GFH, IOCS and IOCS-AOCS are capable to remove arsenic to an excellent level, however practically not suitable as they get clogged due to accumulation of ner particles in the lter bed. Also, it is found that articial coatings enhance the arsenic removal capabilities, however susceptible to clogging.


Introduction
Arsenic contaminations in drinking water has become a global issue and there are several natural weathering reactions and anthropogenic activities behind it. According to World Health Organisation (WHO), arsenic contamination above 10 µg/L is not acceptable in the drinking water (Ahmad et al., 2020).
It is evident that long-term exposure and consumption of water having higher concentration of arsenic causes several health hazards (skin diseases, pigmentation and neurological disorders) and even cancer (Maity et al. 2012; Islam and Islam 2010). Although arsenic in the natural water present in both the organic and inorganic forms, its inorganic form is more toxic as compared to the organic form. Most common forms of arsenic in the natural water are As(III) and As(V) depending on the redox potential and pH of water (Mohan and Pittman, 2007). Due to its character of high reactivity with oxygen, arsenic in the groundwater converts to inorganic arsenic such as pentavalent arsenate, As(V) and trivalent arsenite, As(III) (Pizarro et al., 2021). In the surface water, arsenic is mainly present as oxyanions (H 2 AsO 4 − , HAsO 4 2− ), whereas in the groundwater the As(III) is stably present as arsenious acid (H 3 AsO 3 ) (Gu et al., 2005).
Both the commonly found inorganic species, i.e. As(III) and As(V) are potential hazards to the environment and humanity. Being a major threat to the community, over the years various physicochemical techniques have been proposed and developed for the removal of arsenic from drinking water. Among the reported methods, oxidation method, coagulation method, membrane technologies and   (2021) derived adsorbent material from acid mine drainage sludge for the removal of arsenic from water. However, uses of any such recycled material or waste by-products are subject to stringent environmental regulations, which often require thorough investigations for any potential pollutants present in the recycled/waste material. As such, considering strict regulations and cost-effectiveness, lter media made from sand or ceramic is commonly used. Also, it is an objective to secure/make these lters based on locally available material. Consequently, several studies focused on sand/ceramic made lter media. lter. Khan and Imteaz (2021) reported a detailed study involving batch experiments with six different types of natural adsorbents and concluded that Skye sand is having highest arsenic removal capacity. To replicate the Skye sand, Khan and Imteaz (2020) studied the effectiveness of synthetic adsorbents such as iron oxide-coated sand (IOCS), aluminium oxide-coated sand (AOCS) and different mixtures of these synthetic adsorbents. They have reported that IOCS and mix of IOCS-AOCS(50%-50%) are having even higher arsenic removal capacities compared to Skye sand. To study the arsenic removal e ciencies of these (natural and synthetic) adsorbents under practical scenario, this study describes detailed column experiments with the same adsorbents.

Materials And Methods
Five different types of natural materials were collected from different parts of Australia for the current study. Among the studied materials, Skye, TGS and Scoria sands were collected from Melbourne, Victoria. Red sand was collected from Alice Springs, Northern Territory and iron ore sample was collected from Pilbara, Western Australia. Selection of these sands were based on the criteria of the easy availability within Australia, where the experiments were conducted. To assess the locally available materials' removal e ciency with a benchmarked material, a sixth type of sand was selected. This additional sand sample is commercially available and known as granular ferric hydroxide (GFH). GFH sand was obtained from GEH Wasserchemie GmbH & Co., Osnabrueck, Germany. Detailed physical characteristics and arsenic removal mechanisms of these samples were discussed by Khan and Imteaz (2021). The synthetic adsorbents, IOCS and AOCS were prepared in the laboratory using ner fractions (150-300 micron) of quartz sand. Details of the preparation of the synthetic sands and their effectiveness are outlined in Khan and Imteaz (2020).
Column tests were carried out in a glass tube attached to a 500 mL water reservoir. The diameter of the tube was 3 cm and the lter bed was around 15 cm in depth. The column was stabilized with clamps and stands. The lter bed was made with the selected natural adsorbent for each type of selected sample. The volume of each lter bed was 100 mL. However, the mass of the lter bed was different as the different selected materials were having different speci c gravities. The speci c gravity of Skye sand, TGS sand, NT red sand, Scoria and Iron ore were 2.58, 2.57, 2.52, 1.32 and 4.3, respectively. With a 100 mL volume, the corresponding masses of Skye sand column, TGS sand column, NT red sand column, Scoria column and Iron ore column were 258, 257, 252, 132 and 430 g, respectively. For each of the adsorbents, three similar lters were prepared to obtain a representative adsorption capacity of the adsorbents. The initial concentration of arsenic in the feed water of lters was 100 µg/L. However, to assess the removal e ciency of some highly e cient samples, in the later stage for some of the lters, the arsenic concentration in feed water was increased to 500 and 1000 µg/L. The hydraulic capacity of the lters was xed to 500 mL per hour by a control valve. The pH of feed water was adjusted to 7.0. All the column experiments were carried at 22 º C. The ltrate from the column was collected with a beaker of 1L capacity. Special care was taken to prevent air-trapping in the lter bed by maintaining a water head of (5 cm) on lter bed at the intermission of ow. For arsenic analysis, samples were taken at regular intervals from the ltrate in the beaker.

Results
As mentioned earlier for each material, three columns were prepared and tested with the same arsenic contaminated water. To replicate the continuous operation in the real-life, the contaminated water was ltered continuously through these columns having beds of selected adsorbent. Accumulated ltered water was measured as total ltered water in terms of bed volume (100 mL). Arsenic concentrations at the ltered water were measured at regular interval after each 100 bed volume of ltration and each experiment was continued until the arsenic concentration in the ltered water reached to the breakthrough limit, which is 10 µg/L de ned by the WHO as safe limit for human consumption. Figure 1 shows the relationship of accumulated water ows through the columns and corresponding arsenic concentrations in the ltered water for all the three columns using Skye sand. From the gure it is clear that all the three experiments provided identical patterns of diminishing arsenic removal capacity after a threshold treated water volume. It is to be noted that this sort of diminishing arsenic removal capacity after repetitive use of lter media is likely, mainly due to clogging of the lter media. The initial arsenic concentration in the feed water was 100 µg/L (ppb). Considering the high adsorption capacity of Skye sand and as with the low initial concentration of arsenic (100 µg/L), ltered water samples were having no trace of arsenic, the arsenic concentration in the feed water was increased to 500 µg/L after 400 bed volumes and further increased to 1000 µg/L after 1100 bed volumes. It is found that all the three samples started showing trace (1µg/L) of arsenic after almost same bed volumes (1800-1900). The bed volumes of ltered water which could be achieved before the breakthrough concentration (10 µg/L) for columns 1, 2 and 3 were 3200 (320 litres), 3100 (310 litres) and 3300 (330 litres), respectively. The amounts of arsenic adsorbed in column 1, 2 and 3 were 0.96, 1.04 and 0.92 mg of As (per g of sand) respectively at equilibrium concentration of 10 µg/L. Figure 2 shows the performance of TGS sand columns in regard to arsenic removal capacities through the column beds. Like the case of Skye sand, all the three experiments provided identical patterns of diminishing arsenic removal capacities after a threshold treated water volume. The initial arsenic concentration in the feed water was 100 µg/L. The concentration was increased to 500 µg/L after 400 bed volumes and further increased to 1000 µg/L after 1100 bed volumes. The volumes of water ltered through the columns 1, 2 and 3 were 2700 bed volumes (270 litres), 2600 bed volumes (260 litres) and 2500 bed volumes (250 litres) respectively, until breakthrough of the WHO standard of 10 µg/L was reached. Amount of arsenic adsorbed in column 1, 2 and 3 were 0.77, 0.73 and 0.70 mg of As (per g of sand) respectively at equilibrium concentration of 10 µg/L. For all the three samples the threshold bed volume after which the arsenic concentration in the ltered water was traceable (1µg/L) after almost same bed volumes (1400-1500). These threshold bed volumes are lower than the threshold bed volumes for Skye sand. This lower threshold bed volume conforms with the lower bed volumes which can be achieved before the breakthrough level, which were also lower compared to the Skye sand. It is to be noted that in the batch experiments with the same samples as presented by Khan and Imteaz (2021), performance of TGS sand was also lower than the performance of Skye sand. Figure 3 shows performance of Iron ore sand in regard to arsenic removal capacities through the column beds. The initial arsenic concentration in feed water was 100 µg/L. The concentration was increased to 500 µg/L after 800 bed volumes and to 1000 µg/L after 1400 bed volumes. All the three samples showed similar diminishing arsenic removal e ciency patterns. The volumes of water ltered through columns 1, 2 and 3 were 2000 bed volumes (200 litres), 2300 bed volumes (230 litres) and 2200 bed volumes (220 litres), respectively, until breakthrough of the WHO standard of 10 µg/L was reached. Amounts of arsenic adsorbed in columns 1, 2 and 3 were 0.23, 0.29 and 0.27 mg of As (per g of iron ore), respectively. Also, bed volumes at which arsenic concentration of 1µg/L found in the treated water were very close (1100-1300). These amounts are lower compared to the bed volumes at which arsenic presence in the ltered water were traceable for both the Skye and TGS sands. Also, bed volumes achieved before reaching to the breakthrough arsenic concentration were lower compared to both the Skye and TGS sands. This grading of arsenic removal capacities are similar to what were observed in the case of batch experiments presented by Khan and Imteaz (2021). Figure 4 shows the performance of NT red sand columns in regard to arsenic removal e ciency. Again, almost similar diminishing trends of removal e ciencies were observed with the passage increase in ltered water ow (bed volume). The arsenic concentration in the feed water was 100 µg/L. Unlike all the previously mentioned samples, feed water arsenic concentration was not increased for the NT red sand, due to its lower arsenic removal capacity. The bed volumes at which arsenic concentration of 1µg/L found in the treated water were very close (300-310), which is much earlier compared to Skye sand, TGS sand and Iron ore sand. The volumes of water ltered through the columns 1, 2 and 3 were 390 bed volumes, 410 bed volumes and 400 bed volumes respectively, until breakthrough of the WHO standard of mg of As (per g of sand), respectively. These amounts are much lower compared to all the previously mentioned amounts of arsenic adsorptions by Skye sand, TGS sand and Iron ore sand. Figure 5 shows the performance of the scoria sand lter columns in regard to arsenic removal capacity. The arsenic concentration in feed water was 100 µg/L. The volume of water ltered through each column was 80 bed volumes (8 litres). However, none of the lters was able to remove the arsenic to a concentration below the WHO standard of 10 µg/L. The arsenic concentration in the ltrate was above 40 µg/L from the beginning in each lter. This is mainly due to the limited capacity of scoria sand in removing arsenic, which was also established by batch experiments as reported in Khan and Imteaz (2021).
In regard to the testing of the GFH lter columns, contaminated water with the same initial arsenic concentration (100 µg/L) was fed. Being a commercial grade sand considering the higher capacity of GFH, the arsenic concentration in the feed water was increased to 500 µg/L after 300 bed volumes. The performance of all the GFH sand columns were excellent in removing arsenic and residual arsenic concentration in the ltrate was always under the detection limit. However, all the lters clogged due to formation of a layer of ne particles in the lter bed, which occurred after 320 (column 1), 400 (column 2) and 430 (column 3) bed volumes. Similar phenomena were observed with the experiments with the IOCS and IOCS-AOCS columns. With IOCS columns, 100% arsenic removals were observed up to certain ows, after which all the columns were clogged due to formation of a layer of ne particles in the lter bed. Clogging occurred after 80 (column 1), 70 (column 2) and 50 (column 3) bed volumes. With the IOCS-AOCS (50%-50%) same behaviour was observed, i.e. excellent (100%) arsenic removal until the lter was clogged. In this case the clogging occurred earlier; after 25 (column 1), 40 (column 2) and 60 (column 3) bed volumes. Such clogging phenomena is a common nuisance for any lter media. Imteaz et al. (2016Imteaz et al. ( , 2019 have presented detailed modelling studies on such clogging for arsenic ltration.

Discussions
The results presented in the earlier section are summerised in Table 1, where studied adsorbents are graded as per arsenic removal capacity from highest to lowest. From Table 1 it is clear that Skye sand has got the highest arsenic removal capacity. Among the tested samples, the removing capacities from highest to lowest are Skye, TGS, Iron ore, NT red and Scoria sands. Scoria sand was never able to remove arsenic to the WHO standard level. It is to be noted that the batch experimental results with the same samples showed the same grading of arsenic removal e ciencies, except for GFH sand. Also, it is found that the performances of Scoria and NT red sands are much inferior compared to Skye, TGS and Iron ore sands. Similar trend on the differences of performances were also observed with the batch experiments. Although, GFH, IOCS and IOCS-AOCS lters were having very good arsenic removal capacities, they were excluded from the comparison and grading as all the three columns with each of the samples were clogged after repeated operations. GFH, IOCS and IOCS-AOCS lters failed after accumulated ows of 320 ~ 430 bed volumes, 50 ~ 80 bed volumes and 25 ~ 60 bed volumes respectively. It is to be noted that in the batch experiments IOCS and IOCS-AOCS (50-50) showed superior arsenic removal capacities than Skye sand (Khan and Imteaz, 2020). Also, GFH sand showed second highest (slightly below the Skye sand) arsenic removing capacity (Khan and Imteaz, 2021).

Conclusions
Numerous materials were investigated for the assessment of arsenic removal capacity through adsorption. Most of the studies were based on batch experiments, outcomes of which may not be same in the real-life scenario. To overcome this drawback and to ascertain more reliable arsenic removal capacity through certain material, column experiments are performed. This paper presented arsenic removal e ciencies through column experiments using several natural and synthetic adsorbents. Five locally available natural sand samples and one commercial sand sample (GFH) were selected for further assessment following on their earlier batch experimental results with the same samples. Also, two synthetic samples; IOCS and mix of IOCS-AOCS, which were found to be very effective in removing arsenic were selected. These materials were selected for further assessment with the view of their practical application. Each sample was tested with three separate columns having bed material of the same sample. It is found that for all the samples results of three separate columns were similar and were following the same patterns con rming the representativeness of results for each sample.
From the column experimental results, it is found that Scoria sand is not able to treat arsenic to the maximum level in drinking water set by WHO. Among other tested samples, it is found that the Skye sand is having highest arsenic removal capacity, followed by TGS sand, Iron ore sand and NT red sand. At WHO standard, average arsenic removal capacities (per g of sand) of these samples are: 0.97 ± 0.05 mg for Skye sand, 0.73 ± 0.04 mg for TGS sand, 0.26 ± 0.03 mg for Iron ore sand and 0.016 ± 0.010 mg for NT red sand. Among the other tested samples, GFH sand and the two synthetic sands (IOCS and mix of IOCS-AOCS) showed very good arsenic removal e ciencies for some periods (GFH sand: 320 ~ 430 bed volumes, IOCS: 50 ~ 80 bed volumes and IOCS-AOCS: 25 ~ 60 bed volumes). After which all these column lters failed due to clogging of the lter beds through accumulation of ne particles in the lters. It is to be noted that in the real-life such lters are often backwashed to remove such clogging. Nonetheless, usually through such backwashing the lters do not come back to their original state and clogging increases with the passage of time (i.e. ow through the lter). Mentioned ndings warrant testing with such columns and/or real size prototype, as only batch experiments do not reveal these sorts of practical implications. Another inference which can be made through above ndings is that arti cially coated sands may render high effectiveness in removing arsenic, however are susceptible to earlier clogging making those not suitable for real life application.

Declarations
Ethics approval and consent to participate: The study did not involve any human or animal Consent for publication: No data or material used from third person or somewhere else Availability of data and material: All data generated or analysed during this study are included in this published article Competing interests: There is no competing interest to be reported Funding: No funding was received for the study Authors' contributions: Shahnoor Khan was involved in research design and conducting experiments. Monzur Imteaz was involved in managing the research and writing the paper.