Removal of Arsenic in Ground Water from Northern Burkina Faso through Adsorption with Granular Ferric Hydroxide

(1)

Removal of Arsenic in Ground Water from Northern Burkina Faso through Adsorption with Granular Ferric Hydroxide

-‐ A SIDA Minor Field Study at the Department

of Chemistry, University of Ouagadougou

Emma Lundin and Hannes Öckerman

June – August 2013

_______________________________________________

Master Programme in Environmental and Water Engineering Uppsala University Swedish University of Agricultural Sciences

(2)

Title:

Removal of Arsenic in Ground Water from Northern Burkina Faso through Adsorption with Granular Ferric Hydroxide

Authors:

Emma Lundin and Hannes Öckerman

Supervisors:

Prof. Dr. Samuel Paré, environmental chemist, Department of Chemistry, University of Ouagadougou, Burkina Faso

Prof. Dr. Ingmar Persson, Department of Chemistry, Swedish University of Agricultural Sciences, Uppsala, Sweden

Dr. Johan Mähler, Department of Chemistry, Swedish University of Agricultural Sciences, Uppsala, Sweden

Granting institution:

International Science Programme, Uppsala University

Course title: Project work in environmental and water engineering

Course code: 1TV009

Credits: 15

Key words: Burkina Faso, arsenic, Granular Ferric Hydroxide, adsorption, column test, drinking water, ground water, self-‐regeneration

(3)

Acknowledgements

We would like to send a great thanks to Samuel Paré for welcoming us with such warmth to the equally warm Burkina Faso and Ouagadougou. You gave great guidance and expertise in the laboratory work and also gave us the possibility to explore Ouagadougou outside of the university.

Abdoul Karim Sakira, your patience with us in the laboratory was magnificent.

Joel, thank you for your great support in the practical work every day. Bakouan, that you kept an eye on us and checked in with us every day made it easier to keep up the work. Both of you have a great part in this project and we hope to see that you continue the work also in the future.

We would never have experienced this without the support from our wonderful supervisor Ingmar Persson. Thank you so much for wanting us to get this opportunity and for all the support in the practicalities around the project. Johan Mähler, thank you for your dedication to the experiments and the valuable never-‐ending inputs of scientific knowledge and advice in the laboratory work.

Knowing the importance of companies investing in clean water for everyone we are very thankful and happy to acknowledge GEH Wasserchemie GmbH & Co. KG for providing the adsorbent free of charge.

Finally, the financial support through SIDA has made this possible. Thank you Peter Roth and the International Science Programme at Uppsala University for believing in the necessity and importance of the study and granting us the MFS scholarship.

(4)

Abstract

The need of making arsenic contaminated ground water potable is urgent in parts of Burkina Faso.

An implementation of a treatment design using Granular Ferric Hydroxide (GFH) is under development. Water from a tube-‐well in Lilgomdé, Yatenga province, Burkina Faso, has been treated with the adsorbent GFH through column experiments. The water had an arsenic concentration varying between 99 and 215 μg/L and an average pH of 7.9. The study has shown that arsenic, predominantly in the form of arsenate, can be adsorbed to the material in significant amounts despite a high natural pH and the presence of ions competing with arsenic for adsorption sites on the GFH.

When run through the column, the pH of the effluent water drastically decreased in the beginning.

However, the low pH was soon followed by a slower readjustment towards the pH of the influent water. The adsorption of phosphates and fluorides was also studied. Both competitors exist in higher molar quantities than arsenic in the ground water. Even though arsenic displays a higher affinity for the GFH, an average 44 % of total phosphate and 64 % of the fluoride were adsorbed, making them a factor affecting the results of the study. Hydrogen carbonate is also believed to affect the adsorption process but this could not be confirmed. The empty bed contact time (EBCT), describing the average time of contact between the adsorbent and the water, has shown to be of importance. Increasing the EBCT resulted in notably more arsenic being adsorbed per volume GFH. When increasing the contact time, the study showed that reducing the speed of the flow was more effective than increasing the volume of the adsorbent.

The GFH was also found to have a self-‐regenerating ability to a certain extent. When interrupting the experiment and leaving the column material in the aqueous solution for several days, the arsenic adsorption capacity after the break was shown to be higher than just before it. A 13 % increased capacity was shown in one experiment. Conclusively, the results of this study suggest no hindrances towards developing larger scale columns and prototypes to be applied at tube-‐well pump stations.

Further investigations on the treatment method with GFH, on arsenic contaminated water, are recommended.

(5)

1 Aim

This rapport aims to present a complimentary study to Competition for Adsorption Sites on Iron Oxyhydroxide Based Column Adsorbents for the Removal of Arsenic Oxyacid Species (Mähler et al., 2013) where Granular Ferric Hydroxide (GFH) has been proven to be an efficient material to adsorb arsenic from deionized water through column experiments. This study aims to determine whether the GFH is an effective adsorbent on natural ground water from the Yatenga Province in northern Burkina Faso and to find the physical and chemical parameters affecting the efficiency of the treatment method.

(8)

2 Introduction

Out of the occurring trace elements in the ground water of northern Burkina Faso, arsenic has the greatest impact on human health. Medical problems such as skin lesions and cancer are widely known to occur as a consequence of chronic arsenic intake. Skin lesions such as melanosis and keratosis on hands and feet are among the first seen effects, usually occurring after 5-‐15 years of arsenic exposure (Agusa et al., 2009). Observed skin lesions among a study population in the northern regions of Burkina Faso, West Africa, coincide with high arsenic concentrations in the drinking water being pumped up from deep boreholes (Somé et al., 2012). In these regions geological sources are the contaminating factor of arsenic in the ground water (Smedley et al., 2007). The drilled boreholes are specifically exposed to arsenic contamination as this intrusion increases the mobility of arsenic through the natural process of leaching from the bedrock (Somé et al., 2012).

Both forms of arsenic found in natural waters, penta-‐ and trivalent-‐forms, are considered toxic. The difficulty to find suitable drinking water in the arid northern parts of Burkina Faso has led to the initiative of both local and international scientific studies to find a water treatment solution for the arsenic contamination (Somé et al., 2012; Mähler and Persson, 2013b).

The problem with arsenic contamination is feared to exist throughout West Africa due to the spread of the same type of bedrock (P. Genthon 2013, pers. comm., 15 July). The boreholes have become essential in supplying the northern villages of Burkina Faso with water. Tube-‐well water is used by approximately 87 % of the villagers in the Yatenga province (Somé et al., 2012). Among the first recognized consequences from chronic exposure to arsenic is melanosis, a skin disorder of hyperpigmentation or keratosis where the skin goes rough and dry with skin papules. The effect arsenic has on human health depends on dose and duration of exposure, but arsenic-‐related diseases also include internal types of cancer (Somé et al., 2012).

2.1 Problem background

In 2012, more than one half (52 %) of the tube-‐wells studied in the Yatenga Province, had an arsenic concentration exceeding the guideline limit on drinking water by the World Health Organization of 10 μg/L. Out of the examined wells, 6 % showed a concentration above 100 μg/L. High concentrations are thought to reflect the oxidation/weathering of the mineral arsenopyrite (FeAsS) (Somé et al., 2012). Arsenic concentration in the urine samples of the residents was correlated to the arsenic concentration of the water. 3 % of the population showed an arsenic concentration above

(9)

one half (46 %), respectively, of the population (Somé et al., 2012). A positive but weak correlation between the prevalence of skin lesions and the arsenic concentration of the tube-‐well water was established. It is considered that nutritional intake and genetics also play a big role in the expression of arsenic toxicity (Somé et al., 2012).

2.2 Geology of the region

The basement rock of Burkina Faso consists mainly of Birimian (Lower Proterozoic) meta-‐

sedimentary metaigneous rocks. Arsenic is found mainly in the gold rich granite veins of the Birimian volcano-‐sedimentary sequences (Smedley et al., 2007). While the top mineral soil layer can be very porous but the pores not well connected, fractures in the basement rock create well-‐connected pores for ground water and trace element movement (P. Genthon 2013, pers. comm., 15 July).

Therefore, the fracture zones create good conditions to find drinking water but there is also a severe risk of the water being contaminated with toxic trace elements. One of these is arsenic whose presence in boreholes is linked to the zones of gold, primary sulfide minerals and/or iron oxides within the Birimian basement sequences.

The natural flow velocity of the ground water aquifer is approximately 1 m/year where the fracture zones contribute to the majority of the movement. All kind of activity in the basement rock, such as drilling of boreholes and mining, can possibly affect and change the movement of trace elements, such as arsenic.¹ The mobilization of trace elements in the ground water depends on pH, redox conditions, mineral solubility, kinetics, surface reactions (Smedley et al., 2007) as well as the seasons (P. Genthon 2013, pers. comm., 15 July).

2.3 Earlier studies

Over the past ten years, research on the ground water quality has been carried out in the region around the town of Ouahigouya in northern Burkina Faso by study teams such as Direction de l’Approvisionnement en Eau Potable et de L’Assainissement (DAEPA) in collaboration with P. L.

1 Arsenic has also been found in surface water. Through mining, the minerals are taken to the surface and can through precipitation be spread to near-‐by surface waters. In Nakambe River, 30 km from Ouahigouya, Yatenga Province, arsenic has been detected at concentrations up to 18 µg/l. The only way for the arsenic to reach this surface water is through mining exploitation (P. Genthon 2013, pers. comm., 15 July).

(10)

Smedley and the British Geological Survey (2007). Issa T. Somé and Abdoul K. Sakira, at the University of Ouagadougou have also analyzed water samples from the country’s different boreholes (2012).

As a consequence of the results from the research, boreholes have been closed due to extremely high arsenic levels. In these areas the dug wells, more traditionally used for domestic water supply, return to play an important roll. As the contact time between the ground water and trace elements in a dug well is shorter and the pH lower, arsenic, if present, is more likely to be adsorbed to metal oxides surfaces in the ground. However, when yields are low such as during dry seasons the supply is not always sufficient from the dug wells alone. The quality of the water may also be poor due to higher levels of bacteria growth (Smedley et al., 2007). A treatment method to make the water from the boreholes drinkable according to WHO’s guidelines is therefore desired.

(11)

3 Theory

In order to fully understand the mechanisms in the column experiments, the properties of arsenic, GFH and possible competitors in the adsorption process have been studied.

3.1 Arsenic

The commonly occurring arsenic species in natural ground water are arsenate, As(V), and arsenite, As(III). In oxidized environments the dominating form is arsenate, while arsenite on the contrary is generally found in anaerobic ground water (Saha et al., 2005). Studies done by Smedley (2002) suggest that the highest concentrations of arsenic in ground water are typically found together with a dissolved oxygen (DO) concentration of 2 mg/l or less. Also, the deeper the boreholes are, the higher the arsenic concentration often is, since the ground water potentially has a longer residence time giving a higher possibility for reaction with minerals and organic matter to occur (Smedley and Kinniburgh, 2002).

In the focus area of this study, the ground water seems to be mainly oxic and the dominating arsenic form found is arsenic(V) (Smedley et al., 2007). Since arsenic(V) is an ionic species within the range of pH often found in natural groundwater, it is more readily removed than the neutral arsenic(III) form (Saha et al., 2005). The arsenic speciation is mainly a result of the variation of pH (Table 1) and redox potential (Eh) (Smedley and Kinniburgh, 2002).

Due to the variation of pH within water the dominating species of the dissolved arsenate will either be H2AsO4-‐ or HAsO42-‐, where an equilibrium between the two species is expected at pH 7.10 = pKa2 (Hall, 2003). The species H3AsO4 and AsO43-‐ are practically nonexistent within the natural pH range. In ground water, arsenic(III) is likely to be found in its neutral form As(OH)3 (Figure 1).

(12)

Figure 1. In natural water the pH effect on the forms of inorganic arsenic in more clearly with arsenic(V) than arsenic(III) as the pH usually differs between 6 and 8. In arsenic enriched ground water, taken from seven boreholes in Northern Burkina Faso, the pH ranges from 6.91 to 7.72 (Smedley et al., 2007). Within this range inorganic arsenic in the arsenic(III) species is dominantly found as As(OH)3 and arsenic(V) as H2AsO4-‐ (dominates below pH 7.10) or HAsO42-‐(dominates above pH 7.10).

The figure is kindly borrowed from Issa et al. (2010).

3.2 Granular Ferric Hydroxide

Research on Granular Ferric Hydroxide (GFH) has been carried out at the Technical University of Berlin (Saha et al., 2005). These studies showed that GFH is an efficient arsenic adsorption material, utilizing a small residual mass and found to be better than other commonly used setups for similar water treatment. The following summarized description of GFH, unless stated otherwise, is given by Saha et al., (2005). GFH is a poorly crystallized β-‐FeOOH containing chloride, with the active components being Fe(OH)3 and am-‐FeOOH. The density of GFH is high due to the many water filled pores, which increases the number of available adsorption sites. GFH has a point of zero charge at pH 7.5 (pHpzc) and the specific surface area of the material varies between 250 – 350 m²/g with an average particle size of 500-‐650 µm in diameter. The material consists of a varied mix between fine particles on the surface of larger grains.

The optimal adsorption of arsenic, as arsenate, onto GFH is stated by Saha et al. (2005) to be at pH 4.

However, this is also the lowest solution pH at which the uptake capacity was tested in the cited study. The surface charge of the GFH is dependent on the pH and the degree of positive surface charge on the GFH increases as the pH decreases, which will affect the adsorption capacity of negatively charged ions as H2AsO4-‐ and HAsO42-‐.

(13)

3.3 Surface adsorption on mineral surfaces

This chapter is based on the theory in of Chemistry of the Solid-‐water Interface by Stumm (1992) unless stated otherwise.

When put in an aqueous solution a mineral generally forms a surface of neutral and stable hydroxide groups. The pH of the solution determines the net electrical charge of the surface. If the pH is below pHpzc of the mineral, then the surface will have a positive net charge, while the opposite is true when pH is above pHpzc. Thus, at pH below pHpzc the mineral can electrostatically bind anions to form outer sphere complexes. For instance, if H2AsO4-‐ were to form an outer sphere complex with the GFH in a solution with pH below 7.5 (pHpzc of the GFH), an OH^-‐ would be released and make the solution more alkaline. On the other hand, at pH above pHpzc cations, such as Pb²⁺, could form outer sphere complexes with the mineral and acidify the solution since an H⁺ is released in the binding process.

A mineral in an aqueous solution may also form inner sphere complexes with ions. The extent to which these are formed depend on “geometrical availability” (I. Persson 2013, pers. comm.). In our case, arsenate is pH-‐dependent. Thus, it will form inner sphere complexes on the surface of the GFH particles independently of the pH, but only form outer sphere complexes when the pH is below 7.5.

Arsenic(III), on the other hand, is pH-‐independent and will only form inner sphere complexes at which an OH^-‐ molecule is released from the surface and an H⁺ from the molecule, As(OH)3. This results in no change of the pH.

3.4 Factors affecting adsorption

The chemical water properties are expected to affect the adsorption capacity of solved arsenic onto GFH. In a former study on GFH’s arsenic adsorption capacity, done by Mähler and Persson (2013b), deionized water contaminated by arsenic was used. It is feasible to believe that in natural waters the adsorption would be interfered by other ions likely to be adsorbed onto the GFH as well. The species of the interfering ions is shown to influence the likeliness of adsorption occurring and the ions affinity to the adsorption material (Mähler et al., 2013).

3.4.1 pH

Arsenate is adsorbed more easily onto the positively charged surface of GFH than arsenite since arsenate exists as an ion within the pH range of typical water environments, and arsenite as a neutral molecule. Arsenate can be removed from the solution both through ion exchange and surface adsorption through bonds of partly covalent character. The affinity to adsorb differs on the material’s

(14)

surface, with the strong binding sites firstly occupied followed by the weaker binding sites. The internal surface charge also varies. It is, however, more negatively charged than the external surface (Saha et al., 2005).

A batch test reported by Saha et al. (2005), showed that the uptake of arsenate decreased from its maxima of 170 µmol/g to about 140 µmol/g as pH increased from 4 to 9, when the initial arsenate concentration was 400 µg/L. The attractive force towards the anion is reduced with increasing pH.

However, arsenate is still adsorbed despite the negatively charged GFH surfaces showing that strong inner-‐sphere complexes are formed. The surface hydroxyl-‐groups on the GFH pick up a proton from the un-‐dissociated acid (i.e. H2AsO4-‐) to form water and leaves a site for the created anion to attach (Saha et al., 2007). The affinity to form inner-‐sphere complexes depends on the chemical binding procedure rather than on the charge of the anion.

3.4.2 Competition from other anions

The similarity in pKa-‐values between arsenic acid and phosphorous acid leads one to believe similar behavior in adsorption processes can occur for arsenate and phosphate (Table 1). Phosphate has in fact been proven to be an adsorbent competitor with arsenic, through both batch and column testing (Mähler and Persson, 2013a; Driehaus et al., 1998). Phosphate is adsorbed to a comparable extent, within the same pH range, as arsenate (Saha et al., 2005). The affinity for adsorption onto the surface of the GFH depends on the ion charge. Phosphates should thus be considered a competing ion.

However, in the study presented by Saha et al. (2005), the GFH has been found to prefer to adsorb arsenate over phosphate. Mähler and Persson (2013a) have also shown that fluoride and hydrogen carbonate, if present in high concentration, are possible competing ions.

Table 1. The form of arsenate found as arsenic acid depending on pH and the correlating phosphoric acid. pKa values taken from Hall (2003).

pKa1 pKa2 pKa3 Arsenic acid H3AsO4 2.30 7.10 11.53 Phosphoric acid H3PO4 2.12 7.21 12.32

(15)

4 Experimental

GFH was evaluated regarding its ability to adsorb arsenic from ground water. The water used in the research came from a tube-‐well in Lilgomdé, in the Yatenga Province of northern Burkina Faso, which according to the Department of Health held an arsenic concentration of 85 μg/L (S. Paré 2013, pers.

comm.). The study was conducted during an eight-‐week period in a laboratory at the Department of Chemistry at the University of Ouagadougou, Burkina Faso.

Six sets of column experiments were conducted on the water in order to find the treated bed volume² and the amount of arsenic added per volume GFH, before the residual arsenic concentration exceeded 10 µg/l in the effluent; this will be referred to as the breakthrough. The concentrations of probable competing ions were measured in the influent and effluent. A glass ion exchange tube with inner diameter of 15 mm was primarily used to test the treatment efficiency on arsenic contaminated water with GFH.

4.1 Preparation of Column

The GFH was washed with water to remove small particles prior to use to prevent clogging of the column. A hydrogravimetrical method was used where GFH in an aqueous solution was stirred and the smaller particles floating up to the top discarded. This was repeated until a relatively clear solution was formed. It was then packed in the ion exchange tube to a height of five times the diameter as in the original experiment carried out by Mähler and Persson (2013b). Later in the study the height of the column bed material was changed in order to evaluate the impact of the empty bed contact time (EBCT)³. The material rested on an approximately 10 mm high bed of glass wool and 4 mm of glass beads. The packing was made through adding GFH with a spatula, a spoonful at a time to the column. In order to prevent interfering air bubbles and to make sure that the packing was made with an even density, the tube was filled up with distilled water, which was let to drain out half way between every other spoonful.

2 Bed volume = volume of water added / volume of adsorbent

3 EBCT = volume of adsorbent / flow of water

(16)

An electric pump was installed to steadily pump water from an inlet tank to the top of the column (Figure 2). Stroke length and rate were adjusted to reach a steady state flow through the column. Nonetheless, variations occurred over time due to the human imperfection and small fluctuations of flow in the column due to the adsorbent becoming more and more saturated. The flow rate through the column was measured by placing a 40 mL beaker below the tap of the column and filling up with effluent water while timing with a stopwatch.

Outflow samples for analysis of concentrations of arsenic and competing ions were taken periodically, more frequently when an arsenic break through was expected. In order to evaluate the treatment efficiency, samples of the incoming water prior treatment were taken for analysis. All samples were collected in approximately 30 mL plastic beakers in the beginning of the study and replaced by 50 mL and 100 mL glass flasks by the end. All water samples were stored in a fridge until analysis.

Figure 2. The basic setup of the column experiments. The inlet water of the plastic tank was pumped with an electric pump to the ion exchange column where the water passed through the GFH material resting on a bed of glass wool and glass pearls. Photo by Emma Lundin.

4.2 Summary of experimental setup

In experiment 1 the GFH was washed with tap water while stirring it in a high-‐sided beaker and pouring off the colored water. This was done repeatedly until the water became fairly clear. In experiments 2, 3, 4 and 5 the washing procedure was made with distilled water and using a magnetic stirrer. In experiment 6 the procedure was made with tap water and a magnetic stirrer.

(17)

Each experiment setup differed from one another (Table 2). In experiment 1, samples of influent and effluent water were taken for analysis of arsenic concentration. In experiments 2 and 3, arsenic as well as the concentration of phosphate was analyzed in order to evaluate a potential adsorption competition. For experiments 4 and 5 the analyses also included fluoride and hydrogen carbonate.

Experiment 2 was performed to closer specify the column capacity before breakthrough, considering results from experiment 1. In experiments 3 and 4, the height of the GFH column was increased to 185 mm while keeping approximately the same flow velocity as in experiment 2, in order to study the impact of the EBCT on the treatment efficiency. Experiment 4 was a replicate of experiment 3, which ran dry during an electrical blackout. In experiment 5 the EBCT was kept the same as in experiment 4.

However, this time the column height was the same as in experiment 1 and 2 while the flow velocity was kept low. In an attempt to up-‐scale the method a 1.5 L plastic bottle was used as a column in experiment 6. After a while it was unintentionally aborted when the GFH fell out of the bottle during an attempt to pause the experiment.

Table 2. Each experiment setup aimed to give more information about when the breakthrough occurs and what factors affect the adsorption capacity. Note that experiment 1-‐6 all used GFH as the adsorbing material. experiment 3 and 6 were canceled due to unforeseen circumstances.

Experiment 1 2 3 4 5 6

Height of glass wool (mm) 9 11 14 12 12 20

Height of glass pearls (mm) 4 5 8 7 4 6

Column height (mm) 74 75 185 185 75 39

Column diameter (mm) 15 15 15 15 15 Upper: 84

Lower: 22

Average flow (L/h) 2.68 2.38 2.44 2.31 0.94 5.72

Conducted analyses [As] [As], [PO43-‐]

[As], [PO43-‐]

[As], [PO43-‐], [F^-‐], [HCO3-‐]

[As], [PO43-‐],

[F^-‐], [HCO3-‐] -‐

(18)

4.3 Analysis methods

Temperature and pH was measured with a pH-‐meter, Martini Instruments MI806, which was calibrated with Martini Instruments 4.01 and 7.01 buffer solutions.

One influent and four effluent samples from experiment 1 were analyzed for arsenic concentration with a fast sequential atomic absorption spectrophotometer (AA240FS) with a lamp with the wavelength 193.7 nm. Samples from experiments 2-‐5 were analyzed for arsenic externally at Le Laboratoire National d’Analyse des Eaux (The National Laboratory for Water Analyses, LNAE). The equipment used was a Wagtech Arsenator. The expected breakthrough could be estimated through knowledge from earlier test rounds and this was taken into account when choosing the samples to be analyzed.

Analysis of the phosphate concentration was conducted with a Wagtech WTD Automatic Wavelength Selection Photometer 7100, in the range 0-‐4 mg/L. Two tablets of Wagtech WTD Phosphate LR were added to a 5 mL sample in a glass cuvette. Phosphate first reacts with ammonium molybdate to form phosphor-‐molybdic acid under acidic conditions. The acid is reduced by ascorbic acid to form the colored “molybdenum blue” complex (Lenoble et al., 2005). The samples rested 10 minutes to let the complexes form. The intensity of the color is proportional to the phosphate concentration.

Phosphate was measured with a colorimetric method using a photometer. Note that arsenic may interfere with this method as arsenic may also form a blue complex under these conditions (J.

Mähler 2013, pers. comm.).

The fluoride analysis was also carried out using the Wagtech Photometer 7100, this time within the range 0-‐1.5 mg/L and one tablet was added. Zirconyl chloride and Eriochrome Cyanine R in acid solution form a red colored complex. The color is destroyed by fluoride, which gives a pale yellow color of Eriochrome Cyanine. Different amounts of fluoride produce a color range from red to yellow.

Potentially interference by calcium and phosphates could affect the results but should not be significant at levels found in drinking water.

The hydrogen carbonate concentrations were calculated from measurements of the alkalinity in the samples, which similarly were measured with the Wagtech Photometer 7100, in the range 0-‐500 mg/L CaCO3. One tablet was added to the water, which produced a range from yellow, through green, to blue. The color indicated the alkalinity and was measured using the photometer. The CaCO3 concentration was multiplied with 1.22 to get the HCO3-‐ concentration.

(19)

5 Results and discussion

During the six experiments, several conditions were kept constant when studying the arsenic adsorption efficiency onto GFH. The preparation of the column was carried out in the same way every time in the same laboratory. Air conditioning in the laboratory kept the temperature of the water relatively stable at around 25-‐26 °C. During the overnight breaks the air condition was turned off causing the temperature to have increased slightly in the beginning of the day. However, no large effects in the results could be seen from these variations in temperature.

The average inlet concentration on the water differed quite noticeably from the official value in the well given by the Department of health (85 μg/L). The average inlet concentration during the six experiments was 168 μg/L (N = 14), with single values ranging from 99 to 215 μg/L. The variations might be explained by imprecise analyses method, performed with an Arsenator where samples containing more than 100 μg As/L had to be diluted. The large interval in which the arsenic concentration fall into might also be due to that outtake volumes from the well were collected at two different times, with approximately one month in-‐between.

5.1 Initial Arsenic Adsorption Testing

Experiment 1 and 2 aimed to have the same flow and volume of adsorbent. The EBCT for the experiments were 17 and 20 seconds respectively. 788 (experiment 1) and 785 (experiment 2) bed volumes were treated before the limit of 10 μg As/L was exceeded. In experiment 1 the breakthrough occurred after 106 μg Asadded/cm³adsorbent, and in experiment 2 the figure was 121 μg Asadded/cm³adsorbent. However, samples were taken seldom within the time span of the occurring breakthrough since the breakthrough was initially thought to occur later.

Previous experiments by Mähler and Persson (2013b) have shown that the arsenic concentration in the effluent exhibits a sharp increase when the breakthrough is approaching. The extrapolations between data points have been done with straight lines in order to not overestimate the adsorption capacities. Experiment 2, where the EBCT is short and the pH naturally high, also suggests that the curve might actually be more linear under these conditions (Figure 3). In conclusion, the first two experiments showed that the GFH has the capacity to adsorb arsenic from natural ground water but not as efficiently as desired for a practical tube well pump device. A longer contact time with the adsorbent was thus investigated.

(20)

Figure 3. The concentrations of arsenic in the effluent water of experiments 1 and 2, which were run under similar conditions. Red dotted line is the WHO guideline value of 10 μg As/L. In experiment 1, four of the water samples were analyzed using an Arsenator, and another four with Atomic Absorption Spectroscopy. One of the results from the AAS was considered an anomaly and has not been included in the graph. 1) A break over the weekend was done in the experiment. 2) The concentration of arsenic in these two samples were measured from two water samples which were diluted 1:8 and 1:9 respectively.

5.2 Impact of Empty Bed Contact Time

The following experiments focused on the effects from changing the EBCT. Experiment 3 intended to answer this question but was abruptly ruined by an electricity blackout. Experiment 4 and 5 however showed different results from the initial experiments. In experiment 4 the EBCT was increased, roughly keeping the same flow as in experiments 1 and 2, but increasing the volume of the GFH by a factor of 2.5. In experiment 5 the same volume of GFH was used as in experiments 1 and 2 but instead the flow was decreased by a factor of 2.5. In both cases the EBCT ended up being 51 seconds, therefore generating comparable results.

In both experiment 4 and 5 numerous bed volumes, 2160 and 2369 respectively, were successfully treated before the breakthrough was reached.⁴ A longer EBCT also resulted in an increased efficiency in the amount of arsenic being treated per volume unit of GFH before breakthrough was reached;

356 μg Asadded/cm³adsorbent in experiment 4 and 463 μg Asadded/cm³adsorbent in experiment 5⁵. Previous literature (Mähler and Persson, 2013) suggests that the height of the adsorbent should be around five times the diameter of the column. In experiment 4, this height/diameter ratio was 12.3.

(21)

Comparing increased volume of adsorbent with decreased flow; the latter one seems to be more efficient, as around 30 % more arsenic was adsorbed per volume GFH in experiment 5 compared with experiment 4 (Figure 4).

Since the exact speciation of the arsenic in the tube-‐well water from Lilgomdé is unknown⁶, it is difficult to compare the results with previous studies (Mähler and Persson, 2013b; Driehaus et al., 1998) where distilled water was polluted with arsenous acid and arsenate separately. However, it can be concluded that the adsorption efficiency in natural water is significantly lower than that achieved under laboratory conditions by Mähler and Persson (2013b).

The EBCT strongly influences the arsenic adsorption capacity. Even though the EBCT in experiments 4 and 5 were 2.9 times longer than in experiment 1 and 2.6 times longer than in experiment 2, the amounts of arsenic adsorbed per volume GFH were 3.4 and 2.9 times higher for experiment 4 compared to experiment 1 and 2, respectively. The same figures for experiment 5 were 4.4 and 3.8, respectively.

Figure 4. Concentration of arsenic in the effluent water of experiments 4 and 5, which were run with the same EBCT. The EBCT was higher than in experiments 1 and 2. In experiment 4, a larger volume GFH was used than in experiments 1 and 2 while the flow was decreased in experiment 5 compared to experiments 1 and 2. Red dotted line is the WHO guideline value of 10 μg As/L. 3), 4) Breaks were done in both experiments over a weekend.

6 Though it is assumed that the dominant form is As(V) (Smedley et al., 2007).

(22)

5.3 Self-regeneration

In experiment 1 the effluent arsenic concentration suddenly goes from 32 to < 2 μg/L within a short time span after nearly 200 μg Asadded/cm³adsorbent (Figure 3). Samples were taken 3 minutes before and 10 minutes after pausing the experiment over a weekend, suggesting that the 64-‐hour break affects the adsorption capacity in the column. The adsorbent has potentially undergone self-‐regeneration.

Having the water in the column stationary during a certain time period increases the adsorption capacity of the GFH.

Experiments 4 and 5 were also paused over weekends and showed proof of the self-‐regeneration property. After 30.5 hours the breakthrough concentration of 10 μg/L was reached in experiment 4.

The column was then let to rest during 70 hours before once again being operated for another hour (Figure 4). Two following water samples of the effluent, taken 12 minutes and 55 minutes after the breakthrough, showed an arsenic concentration below 1 μg/L. In experiment 5, the concentration in the effluent was 7 μg As/L after letting 27.5 liters of water pass through the column. It was then let to rest for 66 hours. When restarting the experiment it could be calculated, using linear extrapolation, that another 3.5 liters of water passed through the column before the same concentration was reached once again. This corresponds to an increased efficiency of 13 % and suggests that self-‐regeneration plays a significant role in GFH’s arsenic adsorption capacity (Figure 5).

Figure 5. The effect of self-‐regeneration in experiment 5. Black line is the total amount of arsenic added per volume GFH.

Blue line is the effluent concentration of arsenic. Both are plotted against the total time of the experiment (including breaks). Red dotted line is the WHO guideline value of 10 μg As/L. (With kind permission from Johan Mähler)

0 100 200 300 400 500 600 700

0 5 10 15 20 25 30 35 40

0 24 48 72 96 120 144 168 192

µg Asadded/cm3 adsorbent

[As] (µg/L)

Total time (hrs)

(23)

It is hypothesized that when GFH is left in an aqueous solution to rest for several days the adsorbed anions, arsenic compounds as well as its competitors, free previously occupied sites and make them available for adsorption once again. This could happen in two ways (Figure 6). During this resting period the anions could (1) move to more unavailable adsorptions sites that are located further into the pores of the GFH since the adsorbent has a documented large surface area. The other possibility is that (2) the material absorbs the anions, actually incorporating them into the GHF particles. This

“migration” of anions occurs continuously throughout the experiment. It is not until the column is stationary and adsorption no longer occurs, however, that the effect is noticeable.

Figure 6. When the material self-‐regenerate, it is hypothesized that the adsorbed particle (mostly anions; arsenic acid, phosphoric acid, fluorides, hydrogen carbonates) can either (1) move further into the pores of the Granular Ferric Hydroxide or (2) be absorbed by the material and thus incorporated into the structure of the GFH.

5.4 Adsorption of competitors

Since arsenate occurs as an anion in the range of natural pH, competition from other anions can play an important role in the arsenic adsorption efficiency of the GFH. Phosphates are known to significantly affect the adsorption of arsenic (Driehaus et al., 1998; Saha et al., 2005; Mähler and Persson, 2013b), as well as fluorides and hydrogen carbonates (Mähler et al., 2013).

5.4.1 Phosphates

The initial molar ratios (IMR) of phosphate to arsenic in the influent water ranged between 2.1 to 1 and 3.5 to 1 in experiments 2-‐5. The pattern of how phosphates are adsorbed is more complex than that of arsenic (Figure 7). Studying experiment, 4 two peaks can be identified. The first peak, after 259 μg Asadded/cm³adsorbent, corresponding to 21 hours and 40 minutes of running the experiment,

Removal of Arsenic in Ground Water from Northern Burkina Faso through Adsorption with Granular Ferric Hydroxide