Chemical Stabilization of Arsenic in Contaminated Soil under Low Redox Conditions

(1)

Chemical Stabilization of Arsenic in Contaminated Soil under Low Redox

Conditions

Eva Smedborn Paulsson

Geosciences, master's level (120 credits) 2019

Luleå University of Technology

Department of Civil, Environmental and Natural Resources Engineering

(2)

Chemical Stabilization of Arsenic in Contaminated Soil under Low Redox Conditions

Abstract

Remediation techniques for arsenic contaminated soil have previously focused mostly on the surface soil layers where aerobic conditions are prevalent. In this master thesis, chemical stabilization by adding zero-valent iron (Fe(0)) and calcium oxide (CaO) to an arsenic- contaminated soil under low redox conditions have been studied through up-flow percolation tests. Over a time of 2 months, pH, conductivity, redox potential and concentrations of major and minor elements including As, Zn, Cu, Cd, Fe and Ca under both fluctuating and continuous flow were measured.

Results indicate that CaO is a very promising amendment to use under conditions of low redox, immobilizing 98% of the As and 50-65% of Zn, Cd and Cr over the 8 weeks of testing, compared to the untreated soil. Copper on the other hand was mobilized by the treatment; but as the concentrations of Cu in the soil was low the increased leaching would in this case not be problematic. The leached water also had a very alkaline pH at 13, while conductivity was relatively high at 6.5 mS cm^-1 and the redox potential remained negative throughout the two months. Geochemical modelling indicate that the immobilization of As is most likely controlled by precipitation of Ca-As-complexes as well as ettringite and portlandite. The precipitation of these minerals are controlled by pH, while redox conditions were shown to not be a controlling factor. The alkalinity of the CaO-amended soil was very high and it is expected that the alkaline conditions controlling the immobilization of As will prevail for hundreds of years.

Fe(0) that previously has shown good results in stabilizing As under oxidized conditions did not effectively immobilize As under the low redox conditions in this study. Rather, the leaching of As and metals like Cr, Cd, Cu and Zn showed very similar results as the untreated soil. The results signifies the importance of activating the Fe(0) beforehand to allow amorphous Fe-oxides to form to which the metal(loids) can sorb to.

In the study it was also observed that a fluctuating groundwater table could have an effect on both untreated and treated soil as wet and dry cycles influenced the leached concentrations of As and many other elements. More As was leached during the dry days in the untreated soil and the soil treated with Fe(0), while the opposite was true for the CaO- treated soil.

The results suggests that using CaO as an amendment to treat As-contaminated soil is a method that potentially can be used for soils that have low redox conditions, for example soils below the groundwater table and landfills, although caution should be used in situations with more complex contamination, e.g. where both As and Cu are present.

Keywords: arsenic, contaminated soil, chemical stabilization, low redox, calcium oxide.

(3)

Sammanfattning

Efterbehandlingsmetoder för arsenikförorenad jord har tidigare fokuserat på ytliga jordlager där syre finns tillgängligt och förhållandena är oxiska. I denna masteruppsats har kemisk stabilisering av arsenik under anoxiska förhållanden i en förorenad jord studerats. I perkolationsförsök över två månader har effekten av två olika kemiska tillsatser, nollvärt järn (Fe(0)) och kalciumoxid (CaO), på utlakningen av halvmetaller och metaller som As, Zn, Cd, Cu och Cr samt påverkan på pH, konduktivitet och redoxpotential undersökts under både fluktuerande och konstanta flöden.

Resultaten visar att CaO är en lovande tillsats att använda för kemisk stabilisering av As i jordar som har anoxiska förhållanden. 98% av As immobiliserades jämfört med den obehandlade jorden. Behandlingen med CaO resulterade även i ett stabilt basiskt pH på 13.

Konduktiviteten hölls också på en stabil nivå på 6,5 mS cm^-1 medans redoxpotentialen varierade en del men var negativ under hela experimentet. Den utlakade koncentrationen av As låg mellan 0.059 och 0.021 mg L^-1. CaO immobiliserade även Zn, Cd och Cr med en effektivitet som låg mellan 50-65%, jämfört med den obehandlade jorden. Koppar däremot mobiliserades och lakades ut i större utsträckning; men då Cu-koncentrationen i jorden var låg från början utgör detta inget större problem. Geokemisk modellering indikerade att immobiliseringen av As kontrolleras av pH, men förblev opåverkad av redoxpotentialen. Då alkaliniteten i den CaO-behandlade jorden var hög förväntas de basiska förhållandena som kontrollerar den kemiska stabiliseringen av As att kvarstå under flera hundra år.

Tidigare studier har visat att Fe(0) till stor del stabiliserar As under oxiska förhållanden, men under de anoxiska förhållandena i denna studie hade den Fe(0)-behandlade jorden liknande resultat som den obehandlade jorden, både vad gäller utlakning av metaller, pH, konduktivitet och redoxpotential. Detta markerar vikten av att aktivera Fe(0) genom att låta den reagera med syre och vatten i jorden för att bilda amorfa Fe-oxider som (halv)metallerna kan adsorbera till. I detta fall tillsattes inget vatten i förväg, vilket ledde till att det mesta av järnet lakades ut direkt, utan att bilda fler ytor som speciellt As kan binda till. För båda dessa jordar var utlakningen av As två storleksordnar större jämfört med den CaO-behandlade jorden.

Studien visade också att en fluktuerande grundvattennivå kan ha en effekt på utlakningen av grundämnen både behandlad och obehandlad jord, då cykler av anoxiska och oxiska förhållanden påverkade koncentrationerna i utlakningsvattnet. Mer As lakades ut under de dagar som hade mer anoxiska förhållanden jämfört med oxiska dagar för den Fe(0)- behandlade jorden och den obehandlade jorden, medans det motsatta var sant för den CaO- behandlade jorden.

Resultaten indikerar att CaO har en potential till att användas för att kemiskt stabilisera As i jordar som har låga redox-förhållanden, exempelvis i jordar som är under grundvattenytan och i deponier, men att försiktighet ska användas i jordar som har en mer komplex föroreningssituation, speciellt i jordar som är förorenat av både As och Cu.

(4)

Acknowledgements

I would like to express gratitude to my supervisor Prof. J. Kumpiene at LTU for all her encouragement and thoughtful comments throughout this project. Your knowledge is an inspiration. Also, for introducing me to this project and for providing me with soil samples and data from Kagghamra, I would also like to thank Golder Associates AB and my supervisor at the company, Mikael Lundström. Furthermore, for all the help and inspiring conversations in the lab, I would like to thank the employees at the Department of Waste Science at LTU.

Through darkness, northern lights and midnight sun, my classmates Anna, Anna-Lena, Anton and Pär made my time here in Luleå a memorable one; thanks for all the “fika”- moments and game nights that brightened the time.

Last, but not least, I would like to thank my family and friends who have encouraged and supported me through all of my studies. Whenever I lost motivation or things were tough, you were always there with supportive words, hugs and kisses. Thank you!

(5)

1. Introduction

Arsenic is a hazardous metalloid considered to be carcinogenic and genotoxic for humans and other organisms (IARC, 2012; SEPA, 2017a). It is an element that is and has been used industrially for its properties, including in agriculture as a pesticide, as a wood impregnation agent to prevent insect attacks and moulding as well as a raw material in glass manufacturing.

Historical use of arsenic has led to many places containing elevated concentrations of the element and due to the toxicity of the element, these arsenic contaminated sites are prioritized for remediation.

Remediation of metal(loid) contaminated sites is most often done by excavating and landfilling as this is a well-established method that can be done relatively fast (SGF, 2015).

As the method is generally associated with the generation of large quantities of solid waste and large emission of green-house gases due to long transportations, it is not considered to be a sustainable method in a long-term perspective (Hou and Al-Tabbaa, 2014). Considering that one of the strategies for Sweden to lessen the environmental impact of landfills is to reduce the amount of waste deposited at the landfills (SEPA, 2017b), there is a need for considering other remediation methods for metal(loid) contaminated sites that are both more sustainable in a long-term perspective and cost-effective.

One other method for remediation of metal(loid) contaminated sites that have been developed during the last decades is solidification/stabilization (S/S). S/S is a method that aims at reducing the mobility and/or toxicity of the contaminants. The method can either consist of only solidification or stabilization, or a combination of both, and can be used both in-situ and ex-situ. Solidification involves the encapsulation of the contaminated soil by a material with a low permeability (e.g. cement), thereby reducing the amount of water that can percolate the soil and leach out the contaminants (SGF, 2018). Stabilization encompasses the use of chemicals/compounds (hereby called amendments) that adsorbs or co-precipitate the contaminants in the soil, leading to an immobilization of the contaminants and/or a reduced toxicity (SGF, 2018). Studies of different amendments used for chemical stabilization of arsenic include metal-oxides (Fe, Al and Mn), Ca-oxides and various industrial residues containing these compounds (García-Sanchez et al., 2002; Kumpiene et al., 2009, 2006;

Masscheleyn et al., 1991; Michálková et al., 2016; Moon et al., 2011, 2004; Nielsen et al., 2011; Raven et al., 1998; Travar et al., 2015a, 2015b; Warren et al., 2003; Yoon et al., 2010).

Most of these amendments have been proven to be effective in sorbing arsenic under oxidizing conditions, but in more reducing conditions reductive dissolution of the metal- oxides leads to a remobilization of the previously immobilized arsenic (Dixit and Hering, 2003; Kumpiene et al., 2009; Zhang et al., 2018). Ca-oxide has shown promising results of stabilizing arsenic under more reducing conditions, by forming stable Ca-As-complexes (Travar et al., 2015b).

In long-term stabilization of arsenic, it is important to consider that the conditions in the soil or waste may change. Landfills can after closure develop reducing conditions due to the limited access to oxygen which can greatly affect the stabilization of the metalloid. In soil, reducing conditions are often prevalent deeper down in the ground, below the groundwater

(7)

2

table. A fluctuating groundwater table can thus lead to changing redox conditions, which can have consequences for the chemical stabilization of arsenic when the method is used in-situ.

Amendments that are not sensitive to changing redox conditions may thus be imperative for a successful stabilization in a long-term perspective.

1.1 Scope of thesis

In this Master Thesis, chemical stabilization of arsenic in low redox conditions in a contaminated soil from Kagghamra, Sweden, has been studied. The aim of the project is to find a method for stabilizing arsenic in soil in a long-term perspective, where low-redox conditions are prevalent. The method could thus be possible to use for both treatment of arsenic-contaminated soil in-situ and for arsenic-rich waste in landfills. The main focus has been on the effectiveness of calcium oxide (CaO) as an amendment, as it has previously shown promising results (Travar et al., 2015b), but iron oxides (iron grit and an iron industrial by-product from steel surface processing) have also been studied. The amendments were selected based on a literature study that was conducted within the course Senior Design Project at Luleå University of Technology (LTU) during the autumn of 2017 and experience from researchers at the Division of Waste Science and Technology at LTU. The literature study can be found in Appendix A.

To cover the scope of the thesis, the following research questions have been considered:

1. How effective are the different amendments in stabilizing arsenic and metals at depths affected by a fluctuating groundwater table and below the groundwater table?

2. How long does it take for iron grit (Fe(0)) to start stabilizing arsenic if not being oxidised beforehand?

3. How is the stabilization of arsenic by CaO and Fe(0) dependent on pH and redox potential?

4. Can chemical stabilization by the amendments be used as a remediation method for arsenic contaminated deeper soil layers at Kagghamra?

1.2 Delimitations

This project was carried out during 5 months in spring 2018. Due to time and equipment constraints, percolation tests of only two different amendments as well as one control, were able to be carried out over a period of 2 months.

Amendments are usually injected into the groundwater when used in-situ. In this bench study, up-flow percolation tests was performed. The amendments were mixed directly with the soil, as the used equipment would be clogged if injecting a slurry of the amendments, especially when using iron. Also, as the mixing was performed a week before the experiments started, the amendments were allowed to react with contaminants in the soil beforehand. For iron grit, it is a common practise to mix it with the soil and wetting the mixture to field capacity. In this study, the iron grit was only mixed with the air-dried soil, to evaluate if and how long it would take for the iron to oxidize under the low redox conditions of the experiment. It is important to note that the percolation tests here provide an indication of the stabilizing effect one could expect, but as it is a bench experiment some caution should be used when scaling up for field conditions.

(8)

3

2. Site description

The study area is located close to the village Kagghamra, in Botkyrka municipality, approximately 25 km SW of Stockholm, Sweden (Figure 1). Historically, the site has been used as an experimental wood impregnation facility during 1930-1940, where impregnation agents consisting mainly of arsenic compounds (for example CCA – chromium copper arsenate) were tested and produced. Before that, the area was used as a ship yard during 1880- 1890. The area of the old facility is approximately 0.9 ha and is located next to Kaggfjärden, a narrow coastal inlet of the Baltic Sea. The surrounding land consists of forests and fields.

Figure 1. Map indicating the location of the contaminated site. Red rectangle shows the study area where the wood impregnation facility in Kagghamra was located. Map of Sweden: GSD-Map of Sweden scale 1:10 million © Lantmäteriet 2015; background map: GSD-General map scale interval 1:100 000 – 1:500 000 © Lantmäteriet 2018.

(9)

4

According to Golder (2011), the soil in the area consists mainly of silt and sand, with some inclusions of gravel and stones and covered by a thin layer of humus. At approximately 1-2 m depth, clay is present in some parts of the area, while the groundwater table is approximately 2-3 m below ground surface and flowing in a southern direction with a flow in the magnitude of 5×10^-8 m s^-1 (Golder, 2011). The yearly average precipitation in the area is around 600 mm year^-1 (SMHI, 2017), and the yearly groundwater recharge in the area is 150- 225 mm year^-1 (Rodhe et al., 2006).

An environmental site investigation according to MICS (Method for Investigating Contaminated Soil) was carried out by SGI in 2009, concluding that the area should be classified as risk class 1 (very high risk) (Golder, 2011). Golder Associates AB (Golder) carried out investigations of the area on behalf of Botkyrka municipality during 2010-2011, determining that the area is mainly contaminated by arsenic, but high levels of chromium, zinc and lead have also been detected. For arsenic, levels exceeding the limit for hazardous waste (1000 mg kg^-1) has been found in the area as well as levels exceeding the guideline values for less sensitive and sensitive land use (25 mg kg^-1and 10 mg kg^-1respectively) extending to at least 6 m depth in some places (Golder, 2011). A more detailed investigation was carried out during 2016-2017 by Golder, where soil samples over the whole area were taken and analysed, to determine a plan for remediation.

The area will in the future be used as a peri-urban recreational area (Botkyrka kommun, 2014).

(10)

5

3. Materials and methods

3.1 Materials 3.1.1 Soil

Soil from Kagghamra had been sampled by Golder during 2016 and 2017 over the whole area and at different depths. The soil samples were analysed for the concentrations of different major, minor and trace elements in situ by XRF and was also sent to an accredited lab (ALS Scandinavia) and analysed by ICP. Most of the soil samples had been visually classified at the time of sampling. The soil samples had been stored in plastic bags sealed with a zip-tie.

Due to quite small quantities of soil from each depth, samples that were determined to be of similar soil type, from the same or from adjacent sample points and contained high concentrations of arsenic (as determined by the element analysis from ALS Scandinavia) were mixed into two composite samples (C1 and C2) after the soil samples had been air dried for 7 days; a detailed description of which soil samples that were mixed can be seen in Appendix B.

C1 consisted mainly of soil samples that had been visually determined to be dominated by silt or clay, while C2 consisted mainly of samples determined to be dominated by sand. Each composite sample weighed about 15 kg.

3.1.2 Amendments

Three different amendments were used in the experiments; iron grit (Fe(0)), iron by-product (Feres) from steel surface processing and calcium oxide (CaO). Fe(0) was obtained from SSAB Merox AB, Sweden, and contains approximately 97% iron. The Feres is an industrial residue obtained from Ferruform AB, Sweden, which contains approximately 15% iron. The CaO was in a fine powder form from Riedel-de Haën AG of an analytical grade of 96- 100.5%.

3.2 Methods

3.2.1 Soil characterization

Before any soil characterization was made, the two composite samples C1 and C2 were homogenized thoroughly by hand to reduce dusting and the loss of smaller particles. The soil characterization included determination of the grain size distribution, total solids, organic content, total metal(loid)s concentration, porosity, soil pH and electrical conductivity.

Grain size distribution

The grain size distribution of the two composite samples were determined according to the European standard EN-ISO 17892-4:2016, with some modifications:

• the test sieves consisted of 25 mm, 20 mm, 10 mm, 6.3 mm, 5 mm, 4 mm, 2.5 mm, 1 mm, 0.63 mm, 0.315 mm, 0.2 mm, 0.16 mm and 0.063 mm.

• No wet sieving to remove particles less than 0.063 mm was performed before the dry sieving.

• No determination of the grain size distribution of particles < 0.063 mm was done.

(11)

6 Total solids and Loss on Ignition

Total solids (TS) were determined according to the Swedish standard SS-EN 14346:2007 followed by a determination of the loss on ignition (LOI) according to SS-EN 15169:2007 for both composite soils. Both TS and LOI of both soils were determined in four replicates and mean and standard deviation calculated.

Total metal(loid) concentrations

The total metal(loid) concentration of the two composite soil samples were determined using aqua regia. 1 g of d.w. soil was digested in 15 mL of aqua regia (mixture of concentrated HNO3 and HCl with the proportions v/v 3:1) and heated to 195°C in a microwave oven for 10 minutes and then filtered and diluted to 100 mL with deionized water. The samples were then filtered through a 0.45 µm sterile filter before the element concentrations of the metal(loid)s were determined with ICP-OES.

Porosity

The porosity of the soil samples was determined by saturating the soil samples from below. A 100 mL glass beaker was filled with 20 mL of soil. A burette was filled with water and a plastic tip was positioned at the tip of the burette and lowered into the soil until it was at the bottom of the beaker. Water was slowly allowed to saturate the soil, so that the pore air was pushed up until all pores are filled with water. The volume of water thus corresponds to the pore volume (Vw=Vp). The porosity could then be calculated with Equation 1:

𝑛 =^𝑉^𝑝

𝑉 (Eq. 1)

where,

n = porosity Vp = pore volume V = total volume

The experiment was performed three times per soil (C1 and C2) and the mean and standard deviation were determined.

Soil pH and electrical conductivity

Determination of the soil pH and electrical conductivity (EC) was done by mixing 15 g of soil with 30 mL of distilled water, so that the water-to-soil ratio was 2:1. The mixture was stirred and left for 90 minutes to allow particles to settle before the EC and pH of the mixture were measured.

3.2.2 Screening test

To determine which amendments and what proportions of the amendments to use for the percolation tests, a pre-study in the form of a one-step batch leaching test was done according to the Swedish standard SS-EN 12457-2:2003, with some modifications; the amount of soil and water were reduced to approximately 5% of the recommended amounts, the soil was not divided by a riffle splitter and the samples were homogenized by hand due to the small mass of the sample and to avoid loss of the smallest soil particles through dust.

(12)

7

Seven sets of samples (all in triplicates) with soil from C1 were prepared by mixing the air-dried and sieved (to <4 mm) soil with the different amendments according to the proportions in Table 1, so that the total weight of each sample was approximately 5 g. For the two iron products (Fe(0) and Feres a proportion of 1 wt% iron was used, as a compilation by Komárek et al (2013) showed that many previous studies have reported effective stabilization with 1 wt% of iron. For CaO, four different proportions were tested; 20%, 30%, 40% and 50%. One of the samples contained soil without any amendment and constituted a control sample. The samples were put in 100 mL acid-washed polyethylene bottles and left undisturbed for 4.5 days, after which deionized water was added so that a L/S of 10 l/kg was achieved in each bottle. Thereafter the samples were placed in a rotating device for 24 h at 5 rpm. Samples were left to settle for 15 min and then filtered through a sterile 0.45 µm filter.

4.9 mL of the filtered leachate was mixed with 0.1 mL concentrated HNO3 and stored in 4 °C for later analyses with ICP-OES. Electrical conductivity and pH were measured in the rest of the filtered leachates.

Table 1. Amendments and their proportions used in the screening test.

Sample Amendment Proportion

1 Fe(0) 1% iron

2 Feres 1% iron

3 CaO 20% calcium oxide

7 None 100% soil

3.2.3 Up-flow percolation test

To evaluate the long-term performance of the different amendments, up-flow percolation tests over a total of 8 weeks were performed. Below, the materials used in the percolations test as well as the design of the test are described.

Materials used in the percolation test

The results from the soil basic characterization and the screening test was used to determine which soil and what amendments to use in the percolation test. Because contaminants are often concentrated to finer particles in soils of mixed compositions (SEPA, 2006), C1 was used in both the screening test and the percolation test the leaching tests as it contained a higher proportion of small (<0.063 mm) particles compared to C2.

Of the two iron products (Fe(0) and Feres) evaluated in the screening test, the sample mixed with Fe(0) leached less As, as well as most trace metals (e.g. Cd and Cu) and was thus chosen for the percolation test. For CaO, only the sample with 20 wt% CaO leached As, while the higher proportions had As-concentrations below the detection limit. Thus, 30 wt% CaO was chosen for the percolation test. This is a similar proportion as used in a study by Moon et al (2011), where it was shown that treatment with 25 wt% of calcined oyster shells (which chemically is defined as CaO) efficiently stabilized As in mine tailings.

(13)

8 Design of the percolation test

A schematic figure of the column set-up can be seen in Error! Reference source not found..

The columns (20 cm in height and 5 cm in diameter) were filled with soil and fitted with an inert membrane (hereby called geomembrane) on the bottom and the top to prevent larger particles from being washed off the column. A rubber seal was also used in the top and bottom parts of the column to prevent leakage. The bottom of the columns was connected to a peristaltic pump, and the top of the columns was connected to an acid-washed sample bottle to collect the leachate that was pumped through the columns. The flow rate through the filled columns was set to 7 mL h^-1 (the linear velocity through the empty column was 7.33 cm day^-1 or 6 mL h^-1). An inert atmosphere (N2) was used in the sample bottles to prevent oxidation and carboxylation of the leachates; the sample bottles were sealed with a rubber cork and sparged with N2 for 1.5-2.0 minutes depending on the size of the sample bottles (300 mL and 500 mL respectively) and an air trap was used to regulate overpressure in the sample bottles.

A three-way valve was placed on either side of the column (top and bottom), to be able to turn off the flow completely when changing sample bottles.

Three sets of columns (each in triplicates) were tested, one was filled with only soil and used as a control, and the other two were filled with soil treated with either 1wt% Fe(0) or 30 wt% CaO. The amendments were mixed directly in the soil and homogenized by hand and left undisturbed for 7 days before the column leaching tests were started. The experiment was carried out in a room temperature of 16 ± 2 °C.

The test period was divided into two four-week blocks. During the first four weeks, conditions of a fluctuating groundwater table were simulated by using wet and dry cycles.

Each cycle consisted of five wet days, where deionized water was pumped through the columns, and two dry days where air was pumped through. Sampling of the leachates was done after 1 day (24 hours), 4 days, 5 days and 7 days of each cycle. During each sampling the volume of the leachate was measured, and the leachates were then filtered through a 0.45 µm membrane filter (Filtropur S 0.45, Sarstedt). A sample for elemental analysis was taken

Figure 2. Schematic design of the up-flow percolation tests. Not to scale.

(14)

9

on day 1, 5 and 7. Due to the limited size of the sample bottles, the sample of day 5 was a composite sample collected during the day 4 and day 5. The samples for element analyses were prepared by mixing 4.9 mL of the filtered leachate with 0.1 mL of concentrated HNO3

and were stored in 4 °C for later analysis with ICP-OES. Measurements of redox potential (EH), electrical conductivity (EC) and pH were done on all sampling days directly after the sampling on filtered samples.

In the second four-week block, conditions corresponding to depths below the groundwater table were simulated by the column being water saturated throughout the whole period. This was achieved by continuously pumping water through the columns. Samples were taken four times a week. EH, EC and pH were measured at every sampling, while samples for element analyses were taken two times per week, one of which was a composite sample. The sampling procedure during the second block was identical to the procedure during the first block.

The leached amount of metal(loid)s for each sampled fraction were determined with Equation 2:

𝑈_𝑖 =^𝑉^𝑖^×𝑐^𝑖

𝑚₀ (Eq. 2)

where,

Ui = the released amount of a component in the leachate at each sampling time per amount of soil in the column (mg kg^-1dry matter).

i = the index of the leachate sample

Vi = the volume of leachate of each sample, i, in L

ci = the concentration of the specific component in the leachate fraction, i, in mg L^-1 m0 = the dry mass of the soil in the column, in kg

For each column, the cumulative leached amount of metal(loid)s during the first and second four-week block was calculated, as well as the total cumulative amount. The mean leached amount and the standard deviation for each soil (Fe(0)-amended soil, CaO-amended soil and control) were determined (n=3). Furthermore, to determine how efficient the amendments were, the cumulative leached amount from the soils mixed with the amendments was compared to the control samples. Table 2 defines how the efficiency was classified.

After the percolation test was concluded, the soil from the columns were used to determine the alkalinity. The international standard EN ISO 9963-1:1995 was used, with the modification of using a solution consisting of 1 g of soil from the columns mixed with 100 mL deionized water (L/S ratio of 100 L kg^-1). For the columns with the control soils and the soil treated with Fe(0), the used titrant was 0.02 M HCl, while for the columns with the

Table 2. Classification of efficiency of the immobilization.

% immobilized compared to the control Efficiency

90 – 100 Very efficient

60 – 90 Efficient

30 – 60 Somewhat efficient

0 – 30 Not efficient

(15)

10

CaO-treated soil 0.1 M HCl was used. Furthermore, for the CaO-amended soil an equilibrium period of 15 minutes was used at the end of the test because of the high buffering capacity of the CaO-amended soil, while for the Fe(0)-amended soil and the control only 30-60 seconds were used. Titration was done to pH 8.3 (phenolphthalein alkalinity) to account for all the hydroxide and half of the carbonate content of the solution and to pH 4.5 (total alkalinity) to account for the concentrations of hydrogen carbonate, carbonate and hydroxide in the solution. Equations 3 and 4 were used to calculate the different alkalinities:

𝐴_𝑃 = ^𝑐^𝐻𝐶𝑙^×𝑉^{𝐻𝐶𝑙 8.3}^×1000

𝑉𝑠𝑎𝑚𝑝𝑙𝑒 (Eq. 3)

𝐴_𝑡𝑜𝑡 =^𝑐^𝐻𝐶𝑙^×𝑉^{𝐻𝐶𝑙 4.5}^×1000

𝑉_{𝑠𝑎𝑚𝑝𝑙𝑒} (Eq. 4)

where,

AP = phenolphtalein alkalinity (mmol L^-1) Atot = total alkalinity (mmol L^-1)

cHCl = concentration of the used HCl (mol L^-1)

VHCl 8.3 = the volume of HCl used to reach pH 8.3 (mL) VHCl 4.5 = the volume of HCl used to reach pH 4.5 (mL) Vsample = sample volume (mL)

An estimation was done of how many years it would take for the buffering capacity of the soil to be depleted by the percolating rainwater under the conditions present at the study site. As was described in section 2. Site description, the ground water recharge at Kagghamra is about 150-225 mm year^-1. For these calculations it was assumed that the percolation rate of rainwater into the ground is 200 L mm year^-1and that the pH of the rainwater is 4.6 (HaV, 2018). The calculations can be seen in Appendix E – Calculations.

3.2.4 Geochemical Modelling

Visual Minteq ver. 3.1 was used to model the speciation of As and to identify possible solid phases that can precipitate in the treated soils. For each set of columns (Fe(0)-treated, CaO- treated and control) two models were made with data from beginning of the percolation test (day 2) and the end of the test (day 56) respectively, to enable evaluating changes over the leaching period. As input parameters, the mean concentrations of elements, pH and EH from respective day were used. To identify possible solid phases, the saturation index (SI) of different minerals in the system given the input parameters in each model were evaluated; an SI-value close to 0 indicate that the mineral is in/or close to equilibrium in the system, while a negative SI-value indicate that the mineral is undersaturated in the system and prone to dissolve and a positive SI-value indicate that the mineral is oversaturated in the system and thus prone to precipitate. Minerals with an SI-value close to 0 or with a positive SI-value is thus indicative of possible solid phases in the system.

Furthermore, to get an indication if the pH or EH is a controlling factor for the possible solid phases, a pH-sweep was performed for the input files of day 56 for the CaO-treated soil where SI-values for each mineral at different pH-values or EH-values were produced.

(16)

11

4. Results

4.1 Soil characterization

The characterizations of the two composite soils can be seen in Table 3. C1 had a higher proportion of small particles compared to C2, with about 30% of the grain sizes being less than 0.063 mm compared to the 5% of C2. The full grain size distribution curves can be seen in Appendix C. C1 also had a lower proportion of organic matter. The porosity of the two soils also differ, with C1 having a larger porosity of 44%.

Based on the grain size distributions and the porosities of the two soils, C1 could be classified as grsaSi (silt with gravel and sand) and C2 as grSa (sand with gravel). This corresponds well to the visual classification performed during sampling by Golder in 2016 and 2017.

Table 3. Basic soil characteristics of the two composite soil samples C1 and C2 (mean ± standard deviation, n=3, except for TS and LOI where n=4).

Parameter Unit C1 C2

Grain size Gravel % 20 45

Sand % 50 50

Silt/Clay % 30 5

TS % 99.24 ± 0.04 98.76 ± 0.19

LOI % of TS 2.84 ± 0.20 3.51 ± 0.39

Porosity 0.44 ± 0.01 0.35 ± 0.01

Bulk density g cm^-3 1.33 ± 0.03 1.55 ± 0.07

pH 5.61 ± 0.11 7.66 ± 0.24

EC µS cm^-1 155.6 ± 27.8 78.2 ± 14.5

Total concentrations of some major, minor and trace elements in the two composite soils are given in Table 4. Some variations between the two soils could be seen, with C1 having larger concentrations of most metal(loid)s such as As, Cd, Cr and Zn, while C2 typically exhibited somewhat larger concentrations of the major elements (e.g. Fe, Al, Ca and Mg).

Table 4. Total concentrations of some major and trace elements in the two composite soils. Mean concentrations ± standard deviation (n=3). For trace elements also the guideline value for contaminated soil for less sensitive land use (MKM) and sensitive land use (KM) (SEPA, 2016). All concentrations in mg kg^-1.

Major elements Minor/trace elements Guideline values

C1 C2 C1 C2 MKM KM

Al 18200 ± 1070 18600 ± 1250 As 715 ± 47.9 266 ± 3.91 25 10 Ca 6040 ± 183 8250 ± 911 Cd 2.92 ± 0.47 0.82 ± 0.03 15 0.8 Fe 16600 ± 1290 22500 ± 874 Cr 220 ± 32.2 71.9 ± 8.98 150 80 K 3980 ± 264 3710 ± 166 Cu 20.1 ± 1.54 38.8 ± 1.88 200 80 Mg 4970 ± 353 5390 ± 429 Ni 12.0 ± 1.01 17.6 ± 0.96 120 40 Mn 258 ± 15.4 313 ± 11.7 Pb 20.8 ± 0.81 141 ± 8.05 400 50 Na 398 ± 26.2 411 ± 50.1 V 41.2 ± 2.31 43.9 ± 1.93 200 100 P 470 ± 43.4 540 ± 12.4 Zn 514 ± 45.9 183 ± 3.71 500 250 S 579 ± 65.3 377 ± 45.2

(17)

12 4.2 Screening test

The leached concentrations of some chosen metal(loid)s in the leachates from the batch leaching test can be seen in Figure 3. Arsenic was leached in the samples treated with Fe(0), Feres and 20 wt% CaO, with concentrations of 3.01, 6.31 and 0.21 mg kg^-1 dw soil respectively. For samples treated with higher proportions of CaO, As-concentrations were below the detection limit (0.01 mg L^-1). The control sample showed a higher concentration of As, 18.2 mg kg^-1 dw soil. All three amendments thus show a very efficient or efficient immobilization of As by 99% (20 wt% CaO), 86% (Fe(0)) and 65% (Feres), compared to the control.

Looking at some other metals, all amendments were efficient in immobilizing Cr, as detectable concentrations could only be seen in the control sample. For Cu, only Fe(0) efficiently immobilized the metal (67%). CaO mobilized Cu compared to the control sample, but with higher proportions of CaO the mobilizing effect was less and with 50 wt% CaO, the concentration of Cu was below the detection limit (0.3 µg L^-1). The soils treated with Feres and CaO were also efficient in immobilizing Zn; for the Feres the Zn-concentration was below the detection limit (2 µg L^-1), while for CaO the efficiency increased with increasing proportion of CaO from 82% to 95%.

The pH and electrical conductivity (EC) in the leachates from the batch leaching test can be seen in Table 5. The pH was slightly alkaline in the samples treated with the iron products Fe(0) and Feres (around pH 9) and in the control sample (pH 8), and very alkaline in the samples treated with CaO (pH 13). The soils treated with CaO exhibited the highest EC (84.23-94.97 mS/cm). The sample treated with Fe(0) had a conductivity similar to the control sample (around 1.0 mS/cm), while Feres displayed slightly higher values (1.53 mS/cm).

0.10 1.00 10.00 100.00

Concentration As (mg/kg d.w. soil)

0.00 0.05 0.10 0.15 0.20

Concentration Cu (mg/kg d.w. soil)

0.10 1.00 10.00 100.00

Concentration Zn (mg/kg d.w. soil)

0.00 0.01 0.02 0.03 0.04 0.05

Concentration Cr (mg/kg d.w. soil)

Figure 3. Mean leached concentrations (mg kg^-1dw soil) of As (a), Cu (b), Zn (c) and Cr (d) after the batch leaching test at L/S 10. Only detectable concentrations are shown. Error bars represent standard deviation (n=3).

Fere s

a)

c)

b)

d)

(18)

13

Table 5. pH and conductivity of the samples from the screening test (mean and standard deviation, n=3).

Fe(0) Feres CaO 20% CaO 30% CaO 40% CaO 50% Control

pH 8.87 ±

0.12

9.35 ± 0.12

12.99 ± 0.03

13.04 ± 0.02

13.03 ± 0.03

13.04 ± 0.07

7.97 ± 0.36 Conductivity

(mS/cm)

1.03 ± 0.05

1.53 ± 0.46

84.23 ± 14.41

94.40 ± 0.35

94.97 ± 0.58

88.50 ± 9.46

1.07 ± 0.07

4.3 Up-flow percolation test

During the two months that the percolation experiment took place, the mean total L/S reached for the three soils (Fe(0)-amended, CaO-amended and the control) were 12.60 L/kg, 16.67 L/kg and 13.37 L/kg respectively. For all nine columns the mean L/S reached was 14.21 L/kg (± 1.91), which corresponds to ca 80-130 years of leaching in field assuming 150-225 mm year^-1of precipitation that infiltrates into the soil.

4.3.1 Changes in redox potential, electrical conductivity and pH

The change in EH, EC and pH over time in the leachates from the percolation tests can be seen in Error! Reference source not found.a-c. The soil amended with CaO exhibited noteworthy differences in all the measured parameters in the leachates compared to the soil amended with Fe(0) and the control sample by having a lower EH, a higher EC and a higher pH throughout the leaching test.

Redox potential

For the CaO-amended soil, the EH was negative, ranging between -4.33 to -119.30 mV during the entire two months period. For the soil amended with Fe(0) and the control sample, the redox potentials were instead positive, ranging between 109.70 and 255.60 mV for Fe(0) and 194.03 and 276.07 mV for the control sample. During the first four-week block, each wet and dry cycle showed a pattern of decreasing EH during the wet days, followed by an increase in EH during the dry days, although the changes were relatively small.

The second four-week block, with continuous water saturated conditions showed a stabilisation at approximately -75 mV after 36 days of the leaching test for the soil amended with CaO. The soil treated with Fe(0) and the control sample exhibited a larger variation in EH

between sampling days compared to the soil treated with CaO.

Electrical Conductivity

The conductivity in the soil treated with CaO was stable during the whole leaching experiment, with values around 6.5 mS/cm. For the control sample and the soil treated with Fe(0) the conductivity decreased within the first week of the experiment from approximately 5.2 and 2.6 mS/cm to less than 0.1 mS/cm.

pH

pH also showed noticeable differences between the samples, with a pH of approximately 13 for the soil treated with CaO and a pH around 6.5 for the soil treated with Fe(0) and the

(19)

14

-150.00 0.00 150.00 300.00

0 7 14 21 28 35 42 49 56 EH(mV)

Days

10 100 1000 10000

0 7 14 21 28 35 42 49 56

EC (µS/cm)

Days

4.00 6.00 8.00 10.00 12.00 14.00

0 7 14 21 28 35 42 49 56

pH

Days

Fe⁰ CaO Control

Figure 4. Mean changes (n=3) in EH, EC and pH of the leachates during the percolation test. The dotted vertical lines illustrate the change from the wet days and dry days of each cycle during the first four-week block.

a) b)

c)

control sample. No clear trends related to the wet and dry cycles could be discerned for either EC and pH for all three soil.

4.3.2 Leached concentrations of elements

Changes in the concentrations of some metal(loid)s in the leachates during the percolation tests can be seen in Figure 5a-d. Individual plots for the mean change in concentration in the leachates from the two amended soils and the control soil can be seen in Appendix D – Individual figures of leached concentrations of As.

Arsenic

For As, a distinct difference in concentrations could be seen between the soil amended with CaO and the two other soils (control and the soil amended with Fe(0); with a difference of 2 orders of magnitude (Figure 5a). The soil amended with Fe(0) and the control sample both showed an increase in the leached concentrations during the first 2 weeks, culminating at day 14 with As-concentrations of 4.91 mg L^-1 for the Fe(0)-amended soil and 4.48 mg L^-1 for the control sample. In the following weeks, the concentrations in both these soils decreased slowly to 2.38 mg L^-1 and 2.04 mg L^-1 respectively after 56 days. The highest As- concentrations in the leachate from the CaO-amended soil could be seen at the second day (0.059 mg L^-1).

During the first four-week block, both the amended soil (Fe(0) and CaO) appeared to be affected by the wet and dry cycles, but in opposite ways. The leachate of the CaO-amended soil showed trends of decreasing concentrations during the wet days followed by a rapid increase in the concentrations during the dry days or in the immediate days following the dry days. The leachate from the Fe(0)-amended soil showed trends of increasing As- concentrations during the wet days and decreasing trends during the dry days or in the

(20)

15

immediate days following the wet days. The control soil could possibly have been affected by the wet and dry cycles in a similar pattern as the Fe(0)-treated soil, although the changes in concentrations are small and barely noticeable (Figure 5c).

In the second four-week block, with simulated water saturated conditions, the concentrations of As decreases slowly for the Fe(0)-amended soil and the control sample. The soil treated with CaO appeared to be a bit more varied, compared to the soil treated with Fe(0) and the control, with concentrations ranging between 0.023 mg L^-1 and 0.045 mg L^-1.

Chromium, Copper and Zinc

In the leachates of the Fe(0)-amended soil and the control sample, concentrations of Cr, Cu and Zn are very similar in the two samples and follow the same patterns, while the concentrations in the leachate from the CaO-amended soil differed.

For Cr, the Fe(0)-amended soil and the control exhibited the highest concentrations in the leachates during the first week of the leaching test, after which the concentrations steadily decreased (Figure 5b). The control sample was slightly affected by the wet and dry cycles, with a small increase in the concentrations of Cr during the dry days and decreasing

concentrations during the wet days. The CaO-amended soil showed a similar, but more pronounced trend during the wet and dry cycles as the control sample. But, in the last four weeks, the concentrations of Cr in the leachates from the CaO-treated soil increased and at the end of the leaching test they were slightly higher compared to the concentrations in the leachate from the Fe(0)-amended soil and the control.

Also the Cu-concentrations in the CaO-amended soil was largely affected by the wet and dry cycles (Figure 5c); with decreasing trends during the wet days and a large and rapid

0.00 0.10 0.20 0.30 0.40

0 7 14 21 28 35 42 49 56 [Cu] (mg L-1)

Leaching time (days)

Fe⁰ CaO Control

0.001 0.01 0.1 1

0 7 14 21 28 35 42 49 56 [Cr] (mg L-1)

Leaching time (days) 0.01

0.1 1 10

0 7 14 21 28 35 42 49 56

[As] (mg L-1)

0.1 1 10 100 1000

0 7 14 21 28 35 42 49 56 [Zn] (mg L-1)

b)

c) d)

a)

Figure 5. Mean leached concentrations (n=3) of As (a), Cr (b), Cu (c) and Zn (d) over time in the percolation test.

(21)

16

increase in the immediate day following the dry days. In the second four week-block, the concentrations decreased and the stabilized after about 44 days with an approximate concentration of 0.06 mg L^-1. The Fe(0)-amended soil and the control sample did not exhibit the same pattern, but instead showed a large decrease during the first week followed by a smaller decrease until the concentrations stabilized after approximately four weeks at approximately 0.045 mg L^-1.

For Zn (Figure 5d), the highest concentrations could be seen during the first day for all three samples, but the difference in the concentrations were 2 orders of magnitude between the CaO-amended soil and the two other samples (Fe(0)-amended soil and the control). After the initial flush out, the concentrations decreased until it stabilized for all three samples after the first four-week block. All three samples were affected by the wet and dry cycles during the first four week-block, with increasing concentrations during the dry days, but for each cycle the decrease during the wet days were larger than the increase during the dry days, leading to an overall decreasing trend during these four weeks. Furthermore, the decrease in the concentrations were larger for the Fe-amended soil and the control, compared to the CaO- treated soil and at the end of the leaching experiment, the concentrations for all three samples were within the same order of magnitude.

Iron and Calcium

Figure 6a-b show the leached concentrations of the major elements Fe and Ca in the treated soils and the untreated soil (control). The Fe-treated soil had a high initial concentration of leached Fe on day 1, but the concentration then decreased rapidly and on day 2 it was about half the initial concentration (from 0.66 mg L^-1 to 0.32 mg L^-1). The Fe-concentrations in the Fe-amended soil was affected by the wet and dry cycles, and displayed increasing concentrations during the wet days and decreasing concentrations during the dry days or in the day following the dry days. Although, the overall trend of the Fe-concentrations during the first four-week block was decreasing concentrations. During the second four-week block, the Fe-concentrations in the leachates from the Fe-amended soil were somewhat stable. The CaO-amended soil showed very low leached concentrations of Fe throughout the whole leaching test, and was below the detection limit (0.002 mg L^-1) after three weeks of leaching.

The untreated soil (control) showed increasing concentrations during the first week of the

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

0 7 14 21 28 35 42 49 56

[Fe] (mg L-1)

Fe⁰ CaO Control

1 10 100 1000

0 7 14 21 28 35 42 49 56

[Ca] (mg L-1)

b)

Figure 6. Mean leached concentrations (n=3) of Fe (a) and Ca (b) over time in the percolation test of the treated soils (Fe(0) and CaO) and the untreated soil (control).

a)

(22)

17 percolation test, followed by a decreasing trend.

The Ca-concentrations also displayed differences between the CaO-amended soil and the two other samples (Fe-amended soil and the untreated soil). For the CaO-treated soil the leached Ca-concentrations were quite stable throughout the leaching test, although a small overall decrease could be discerned from approximately 790 mg L^-1 on the 2^nd day to 680 mg L^-1 on day 56. The Fe-treated soil and the control instead showed high initial Ca- concentrations that decreased rapidly during the first week and then more slowly. After four weeks, the Ca-concentrations stabilized at approximately 2 mg L^-1 for the untreated soil and 1.2 mg L^-1 for the Fe-treated soil.

4.3.3 Leached amount of elements

Of the total concentration of metal(loid)s in the soil (see Table 4), only a few percentages leached out during the percolation test (Table 6). Most noticeable is that 5% of the As leached out from the control and Fe(0)-treated soil, while for the CaO-treated soil only 0.1% leached out. Almost 12% of the Cu leached out from the CaO-treated soil.

Table 6. Percentage of the total metal(loid) concentrations that leached out from the soil in the treated and untreated soils during the percolation test.

Element Fe(0) (% leached) CaO (% leached) Control (% leached)

As 5.12 0.09 5.51

Cd 2.94 1.05 3.01

Cr 0.17 0.11 0.22

Cu 3.08 11.90 3.63

Zn 4.79 1.07 5.14

Figure 7a-d shows the leached amount of a As, Cr, Cu and Zn during the percolation test, highlighting the amount leached during the first four-week block, the second four-week block and the total leached amount. The amount of As that leached out from the Fe-amended soil and the control sample during the first and second block are approximately the same. The soil amended with CaO on the other hand, had a larger amount of As leached out during the second four-week block compared to the first. For Cr, more was leached out during the first block in the soil amended with Fe(0) and the control compared to the second block, while the opposite was true for the CaO-amended soil. The leached amount of Cu showed no large differences between the first and second block for any of the soils. Zinc exhibited large differences between the two blocks for the soil treated with Fe(0) and the control; with more leached out during the first block compared to the second block. The CaO-amended soil on the other hand, showed similar amounts of Zn leached out during the two blocks.

By comparing the amount of leached metal(loid)s between the two treated soils (Fe(0) and CaO) and the control sample, a percentage of immobilization or mobilization could be determined (Table 7). The CaO-amended soil demonstrated a very efficient immobilization of As (98%) and efficient immobilization of most metals, although for Cr after 8 weeks the CaO- amendment was only somewhat efficient. The only exception being Cu, for which the CaO- treatment instead caused a mobilisation, leading to a higher release of Cu compared to the

(23)

18

control; although important to note is that the concentrations of Cu in the soil was rather low from the beginning. The Fe(0)-amended soil only showed a slight immobilization effect of the metal(loid)s.

Table 7. Percentage decrease in total leached amount of a few metal(loid)s in the amended soils compared to the control. Mean values (n=3). Negative values indicate an increase in the amount metal(loid)leached.

Decrease in leached amount from soil treated with Fe(0) (%)

Decrease in leached amount from soil treated with CaO (%)

Element After 4 weeks After 8 weeks After 4 weeks After 8 weeks

As 10 7 99 98

Cd -3 2 77 65

Cr 29 23 78 50

Cu 21 15 -258 -228

Zn -6 6 85 61

4.3.4 Alkalinity

The alkalinities of the different amended soils and the untreated soil can be seen in Table 8.

The soil treated with CaO had a high phenolphthalein alkalinity (39 mmol L^-1), accounting for the content of hydroxide and half the carbonates within the solution, while the total alkalinity (accounting for the hydroxide, carbonate and hydrogen carbonate content in the solution) was not much higher (41 mmol L^-1). The soil treated with Fe(0) and the untreated soil both had a very low alkalinity (0.11 mmol/l).

It was estimated that it would take approximately 800 years for the buffering capacity to be depleted in the soil (for calculations see Error! Not a valid bookmark self-reference.

0 0.2 0.4 0.6

Fe⁰ CaO Control

Amount Cr leached (mg kg-1dw soil)

0 1 2 3 4

Fe⁰ CaO Control

Amount Cu leached (mg kg-1dw soil)

Week 1-4 Week 5-8 Total 1 10 100

Fe⁰ CaO Control

Amount Zn leached (mg kg-1dw soil) 0.1

1 10 100

Fe⁰ CaO Control

Amount As leached (mg kg-1dw soil)

Figure 7. Mean leached amount of As (a), Cr (b), Cu (c) and Zn (d) in the Fe(0)-amended soil, CaO- amended soil and the control during the first 4 week block (week 1-4), the second four-week block (week 5-8) and in total of the percolation test.. Error bars indicate the standard deviation (n=3).

d)

a) b)

c)

Chemical Stabilization of Arsenic in Contaminated Soil under Low Redox Conditions