Consequences of Filtration for Risk Assessing Metal Contaminated Ground Waters

(1)

KEMI C – SJÄLVSTÄNDIGT ARBETE, 15 HP

HANDLEDARE: STEFAN KARLSSON

EXAMINATOR: MATTIAS BÄCKSTRÖM

ELISABETH ÄNGMYREN | 16 JULI 2016 DIPLOMA WORK

Consequences of filtration for risk

assessing metal contaminated

(2)

Abstract

The purpose of this project was to optimize the filtration process when analyzing ground water for risk assessment. Particulate (>0.45 µm), colloidal (0.45 µm<diam.<0.01 µm) and dissolved metal species were studied using different filtration methods: flow filtration in the field using peristaltic pump and in-line filter; replaceable filters and syringe filter in the field; replaceable filters and syringe filter in a laboratory environment.

Studies where metal concentrations in the filtrate are measured as a function of the amount of material enriched in the filter was carried out. MP-AES were used for analyzing the metals in controlled systems and the reference method was done with ICP-MS. The project was done on the behalf of Structor and evaluation were also made between the sampling methods that Structor has undertaken and the methods that was examined in this report. The results were evaluated by comparison of guideline values set by the World health organization, WHO on drinking water and the regulation and status classification and environmental quality values of ground water recommended by the Swedish Geological Survey, SGU. Produced data was assessed by conventional statistics.

Introduction

In many methods for quality assurance while risk assessing contaminated waters, filtration is a requirement before analysis and quantification of metals can be done. A wide range of error sources may occur during separation of phases and the analysis of dissolved metals can thereby give incorrect results. There is no defined limit for dissolved/unsolved species besides the commonly used 0.45 µm which is an historical value that has been the most practical. In Sweden the groundwater guidelines for dissolved species is almost as strict as WHOs guidelines for drinking water (Table 1).

The United States environmental protection agency (U.S. EPA) has decided that filtration during field sampling of ground water beneath municipal solid waste landfill facilities should be banned because analytical tests will show inaccurate results and the true picture of ground water quality will not be visible [1]. Due to this Ohio EPA did a literature investigation of their own to explain why this was important. They concluded that the sampling method and not filtration was the most important part and particularly what flow rate the sampling pump had. A low rate and an isolated sampling zone are crucial to minimize artifacts such as reduction/oxidation precipitation; dilution or concentration of contaminants; or pH changes to mention a few [1].

When dealing with natural water, metabolic activity of microorganisms will change some of the properties of the water if the samples are not stored at low temperature after sampling. When water comes in contact with oxygen when sampling certain elements, e.g.

(4)

Fe oxidize, and hydrolyzes the excess protons lowers the pH. This can be revealed by increased levels of metals in the filtrate as a function of time. Gimpel et al. [2] showed that separations in situ (on site) without removing the water sample gives the most accurate interpretations of the total amount of elements in the solution. They also concluded that for Fe in particular, in situ dialysis is a better choice than filtration on site.

Horowitz et al. [3] did a study on trace elements in surface water where they tested 0.45/0.40-µm membrane filters from assorted manufacturers and with different diameter. The filters were made of polycarbonate (0.40 µm plate filter), cellulose nitrate (0.45 µm plate filter) and polyether sulfone (0.45 µm capsule filter, surface area of 600 cm2_{). In their}

report they concluded, for what they called the first group: Al, Fe, Ni, Cu and Zn, that the polycarbonate filter produced the lowest concentrations, cellulose came second and the polyether sulfone produced the highest concentrations and this was correlated to the surface area of the filters for Al and Fe but not as obvious for Ni, Cu and Zn. Other elements like Co, Mo, Pb, Sr, Ba and Ca were influenced by the surface area of the membrane filter whereupon they became sorbed to the filter and later, once all sorption sites were filled, released. This allowed the concentration of these substances to increase in the later stage of filtration [3]. In addition to the field study Horowitz et al. also tested the filters in a Class 100 clean room and this time they added a fourth filter: a mixed cellulose acetate/nitrate filter (0.45 µm tortuous path filter) which concentrations came out the highest. In whole their conclusion shows that the common use of 0.45 µm filters simply is not representative for all dissolved particles. When using 0.45 µm pore filters colloidal particles can be retained by adsorption to the filter surface or the filtrate and thus some elements will not be accounted for [3].

STRÖMSHOLM, PRESENTATION OF CASE AND FIELD SITE History

From the early 1940’s until 1979 the small town of Strömsholm on the shore of the lake Mälaren in Sweden, hosted a business company that carried out pressure impregnation of lumber. The work gave emissions of CCA (chromium, copper, arsenic) and zink into the surrounding premises. At the end of the 1960’s cows grazing in the meadow on the adjacent property died from assumed arsenic poisoning. This prompted an investigation of soil and water in the area by SGU in 1979, which showed high levels of CCA and zink. In 2002 Länsstyrelsen, the County Board, categorized the area as a major environmental risk and the sensitivity becomes particularly large when the area next to it is a nature reserve and a Natura 2000 site (Figure 1).

In 2008 the environmental technology company Structor was commissioned to analyze the soil and water and to give recommendations on how the area would be treated. This led to a removal of large amounts of soil from parts of the site and Structor has since then performed regular testing in the area.

(5)

Soil layers and ground water

The top ground layer is mostly a fill layer with a thickness of 0.5-1 m with clay underneath. In some parts of the site the top layer is natural sand-soil. The thickness of the clay is between 1.5 and 6.5 m and becomes thicker to the west. The clay in turn has an under layer of aquifer overburden of gravelly and sandy till. In the installed ground water wells, water level were found at between 0.1 and 3 m below the surface. In the meadow with sampling sites SM 19 and SM20, the ground water level is almost at soil surface. Flow direction is west-southwest from Strömsholmsåsen towards the lake Mälaren (Figure 1).

(6)

Sampling wells

The wells have been installed according to Field guide for environmental investigations 1:2004 by the Swedish

Geotechnical Society and the ground water pipes consists of bottom plug, bottom sump, filters, risers, seals and lids (except B1) [4].

SM13 – Under some trees at the edge of the wetland. Ground level is 2.00 m above mean sea level and sampling depth is 1.11 m below ground level.

SM14 – In a small ditch by a dirt road, right next to a house. Ground level is 2.48 m above mean sea level and sampling depth is 1.63 m below ground level.

SM17 – In deciduous forest. Ground level is 1.54 m above mean sea level and sampling depth is 0.23 m below ground level (Figure 2).

SM18 – In deciduous forest. Height above mean sea level 1.54 m and sampling depth is 0.14 m below ground level.

SM19 – Wetland/field, high grass. Sampling depth is 0.28 m below ground level. SM20 – Wetland/field, high grass. Sampling depth is 0.23 m below ground level. B1 – Iron pipe placed by SGU in 1979, at the edge of the deciduous forest.

Objective

The aim of this project was to find out what size of pores in membrane filters are the most ideal to use when separating colloids from particulate material. The purpose of the controlled systems were to see if the clogging point was detectable using various filters and if it is possible to determine how much fluid is necessary to press through the filter to get the best analytical results with emphasis on inorganic materials. Another part was to find out how much impact pH actually has during filtration, when trying to estimate levels of metals in contaminated water. In the case study part the objective was to determine if the time between sampling and filtration is significant and also if the filter Structor is using is the best alternative.

(7)

Method

A controlled system with clay and four trace metals in a solution of NaClO4 and MΩ-water

will be the first part of this experiment. This is to simulate natural ground water in Sweden where acid woodland water can have a pH of 4.5 and calcareous soil waters have pH around 8. Clays are used in the model because they are good cation exchangers which helps if pH needs to be changed. Soil and clay particles are negatively charged and has a capacity for binding cations that surrounds them via ion exchange. The measurement, cation exchange capacity, CEC is used together with hydrolysis constants to be able to determine when the metals are in solution and when they are bound to the clay particles. When the metal concentration exceeds CEC of the clay, the metal with the highest hydrolysis constant is the one that will remain bound to the clay. If the pH is lowered, CEC will become lower and more metal ions will be in solution. The cation exchange capacity is tabulated and the unit is milliequivalents per gram (meq/g). CEC can be determined in various ways and by various substances, such as copper ethylenediamine [4]; silver–thiourea [5] and the reference method ammonium acetate by Lewis in 1949 [6]. These are just a few methods, there are many more and depending on data inputs and substances the outcome values can be quite different from each other, hence the large range values that can be found in Table

2. To be able to calculate meq/g for the analyzed ion, the charge and concentration of the

specific ion needs to be known. To calculate the CEC for the amount of clay used in an experiment, CEC for that specific clay needs to be known. Since the exact composition of the clay used in this test is unknown, an average value of Illite clay CEC is used when determining if all the metals will be able to adsorb to the clay in the controlled systems. Illite clay is the most common clay in Sweden, as well as elsewhere, and it has a very high CEC. It also has a large specific surface which makes it able to sorb a great amount of ions (Table 2). Which metal ions to choose depends on their ability to sorb to the clay and this is determined by the ions hydrolysis, higher tendency to hydrolyze → higher tendency that the ion sorbs. The metal ions chosen for the controlled sample analysis is copper, cadmium, zink and lead. By looking at their hydrolysis, we can see in which order they will dissociate from the clay, the lower log K the loser the elements will bind to the clay (Table 3): CdOH+

will come off first and then PbOH+_{, CuOH}+_{and lastly ZnOH}+_{. Another variable is pH} zpc,

thepH where the net surface charge of the clay is zero (Table 2). At a pH over zero point of charge all ions in solution will associate with the clay particles while a pH under pHzpc

will make them go into solution again since the clay will change charge and become positive. So in the controlled test part of this study one of the expected outcomes is that there will be more metals in solution at pH 4 than at pH 6 and 8, thereby giving a larger relative concentration when plotted as a function of volume.

(8)

Table 1. Guideline values for different types of water.

Element Guideline value or Provisional guideline value for drinking

water set by the WHO [7]

Guideline values for groundwater on national (Swedish) level set by SGU [8]

Maximum permitted concentration in inland-surface

waters set by Havs- och vattenmyndigheten [9] Aluminum - Antimony 0.02 mg/l (20 µg/l) Arsenic 0.01 mg/l (10 µg/l) 10 µg/l Barium 0.7 mg/l (700 µg/l) Beryllium - Boron 2.4 mg/l (2400 µg/l) Bromide -

Cadmium 0.003 mg/l (3 µg/l) 5 µg/l 0.45-1,5 µg/l (depending on water hardness)

Chlorine 5 mg/l (5000 µg/l)

Chloride 100 mg/l

Chromium Total chromium: 0.05 mg/l (50 µg/l) Conductivity 75 mS/m Copper 2 mg/l (2000 µg/l) Fluoride 1.5 mg/l (1500 µg/l) Inorganic tin - Iodine - Iron - Lead 10 µg/l 10 µg/l - Manganese -

Mercury 6 µg/l for inorganic mercury 1 µg/l 0.07 µg(l/l Molybdenum - Nickel 0.07 mg/l (70 µg/l) - Nitrate 50 mg/l Potassium - Selenium 0.04 mg/l (40 µg/l) Silver - Sodium - Sulfate - 250 mg/l Uranium 0.03 mg/l (30 µg/l) Zinc -

(9)

Table 2. Properties of clays that can be found in Swedish soil.

Kaolin/Kaolinite Montmorillonite Illite

Sorption-property

Soluble in water Water absorbing and swelling

Does not swell Capacity (CEC)1 _{0.001-0.015 meq/g}

Low cation exchange capacity

0.06-0.1 meq/g High cation exchange

capacity

0.1-0.7 meq/g Very high cation exchange capacity Particle-size << 0,002 mm << 0,002 mm < 0,002 mm Specific surface 2 _{10-20 m}2_/g _{50-120 m}2_/g. _{65-100 m}2_/g pHZPC 3 4.6 2.5 2-3 Molecule formula

H2 Al2 O8 Si2 H2O Al2O3.4(SiO2).H2O (KH3O)(AlMgFe)2(SiAl)4

O10[(OH)2H2O])

1_{[10] [11] [12]}

2 _{Particle surface relative particle mass. The smaller the size of a particle the greater the specific surface [13]} 3_{pH at zero point of charge: pH value were the net surface charge is zero [11].}

Table 3. Properties of some of the elements that can be found in soil and water. The ones used in the controlled samples is highlighted.

Hydrolysis Hydrolysis constant (log K) Oxidation number 1:st Ionization energy (kJ/mol)

Li Li+H2O → Li++OH-+½ H2 Weak +1 520.2

Na Na+H2O → Na++OH-+½ H2 +1 495.8 K K+H2O → K++OH-+½ H2 +1 418.8 Zn Zn2+_+H2O→ZnOH+_+H+ Medium-strong -7.69 +2 906.4 Cd Cd2+_+H2O→CdOH+_+H+ or 2Cd2+_+H2O→Cd 2OH3++H+ -10.10 or -6.40 +2 866.0 Cu Cu2+_+H 2O → CuOH++H+ Strong -7.70 +1 or +2 745.5 Pb Pb2+_+H2O→PbOH+_+H+ or 2Pb2+_+H2O→Pb 2OH3++H+ -7.70 or -6.40 +2 or +4 715.6 Mn Mn2+_+H2O→MnOH+_+H+ or Mn3+_+3H2O→MnOH2+_+3H+ -10.95 or 0.40 +2,+3,+4 or +7 717.3

(10)

Hydrolysis Hydrolysis constant (log K) Oxidation number 1:st Ionization energy (kJ/mol) Al Al3+_{+ H} 2O → AlOH2+ + H+ or 2Al3+_{+2H2O →} Al2(OH)24++2H+ Amphoteric -5.02 or -7.69 +3 577.5 Fe Fe3+ _{+H2O→ FeOH}2+_+H+ or Fe3+_{+ 2H} 2O → Fe(OH)2+ + 2H+ or Fe2+_+H2O→FeOH+_+H+ or 2Fe3+_+2H2O→Fe 2(OH)24++2 H+ Amphoteric -2.19; -5.69; -6.74 or -2.90 +2 or +3 762.5 As H3AsO4+H2O→ H++ H2AsO4 -H2AsO4-+H2O→ H++ HAsO3 2-HAsO42-+H2O→ H++AsO3 3-Weak +3 or +5 947.0

PART 1. PREPARATION OF CONTROLLED SYSTEMS

A fresh clay sample was placed in a 1 L beaker and MΩ was added to a clay to water ratio of approx. 1:3. The mixture was stirred with a magnet stirrer for 5 min. Floating organics were skimmed and discarded. The beaker was then placed in an ultrasonic bath and was left to disperse for 20 min. It was then placed on a flat surface for 10 min to allow particles to settle. The supernatant with the dispersed clay was decanted into a clean beaker and allowed to settle. Water was removed using centrifugation for one hour at 6000 RPM. The clay was dried at 105 °C and was then homogenized.

From 10 mM NaClO4, three solutions with different pH were prepared: pH 4

and 6 with the addition of HNO3 and pH 8 with the addition of NaOH. Dry clay was

mixed with the pH-calibrated solutions on end over end shaker for two hours. A metal mix containing Cd, Cu, Pb and Zn were added to a total concentration in the tubes of 0.5 mg/L each. For labeling and what each solution contained see Table 4 and

Table 5. pH was adjusted back to 4, 6 and 8 (for pH after adding the clay and metals but

(11)

Table 6) and the solutions were placed on an end over end shaker for 24 hours.

Table 4. Labeling and ingredients of controlled system batch 1. Metal mix contains Cu, Zn, Pb and Cd 50 mg/L. Batch 1 B1 4:1 B1 4:2 B1 4:3 B1 4:4 B1 6:1 B1 6:2 B1 6:3 B1 6:4 B1 8:1 B1 8:2 B1 8:3 B1 8:4 pH4- solution 45 mL 45 mL 45 mL 45 mL - - - - pH6- solution - - - - 45 mL 45 mL 45 mL 45 mL - - - - pH8-solution - - - 45 mL 45 mL 45 mL 45 mL Illite clay - 0.09 g 0.09 g - - 0.09 g 0.09 g - - 0.09 g 0.09 g - Metal mix - - 0.45 mL 0.45 mL - - 0.45 mL 0.45 mL - - 0.45 mL 0.45 mL

Table 5. Labeling and ingredients of controlled system batch 2. Metal mix contains Cu, Zn, Pb and Cd 50 mg/L. Batch 2 B2 4:1 B2 4:2 B2 4:3 B2 4:4 B2 6:1 B2 6:2 B2 6:3 B2 6:4 B2 8:1 B2 8:2 B2 8:3 B2 8:4 pH4- solution 45 mL 400 mL 400 mL 45 mL - - - - pH6- solution - - - - 45 mL 400 mL 400 mL 45 mL - - - - pH8-solution - - - 45 mL 400 mL 400 mL 45 mL Illite clay - 0.08 g 0.08 g - - 0.08 g 0.08 g - - 0.08 g 0.08 g - Metal mix - - 4 mL 0.45 mL - - 4 mL 0.45 mL - - 4 mL 0.45 mL

(12)

Table 6. Measured pH before recalibration. Illite clay has a pHZPC (pH at zero point of charge) around 3, this means that when the clay is added to the NaClO4-solution it will lower the solutions pHs as shown below.

Sample ID Batch 1 pH Sample ID Batch 2 pH

B1 4:2 4.5 B2 4:2 4.2 B1 4:3 4.1 B2 4:3 3.9 B1 4:4 4.1 B2 4:4 3.9 B1 6:2 5.4 B2 6:2 5.4 B1 6:3 4.8 B2 6:3 4.6 B1 6:4 5.2 B2 6:4 4.3 B1 8:2 ~6 B2 8:2 ~5 B1 8:3 ~5.4 B2 8:3 4.8 B1 8:4 ~5.5 B2 8:4 4.8

PART 2. ANALYSIS OF CONTROLLED SYSTEMS

Polycarbonate filters with a diameter of 47 mm and pore sizes of 1.0, 0.4, 0.2 and 0.05 µm (

Table 10) with detachable filter holder for spray assembly were used while filtrating the

test water. Batch 1 was used when filtrating with 0.4 µm filter and batch 2 was used while filtrating with the other filters. Volumes of at least 10 mL were collected in 15 mL Sarstedt tubes in ~2 mL successive subsets. The clogging point was attempted to be surpassed. Since the subsets all had slightly different volumes an average of 2.2 mL was used when plotting all the diagrams. 10 µL distilled HNO3 per mL solution and 10 µL internal

standard, IS (La, Lu and Y) per mL solution were added to the tubes. Analysis of the controlled systems was done with MP-AES, for settings see Table 7.

Table 7. Settings for the MP-AES during test water analysis. MP-AES Common conditions

Replicates 3

Pump speed (rpm) 10

Sample introduction Autosampler

Uptake time (s) 45 (fast pump)

Switch delay (s) 0

Rinse time (s) 90 (fast pump)

(13)

In this experiment CEC is only used in connection with the controlled experiments to make sure that all metals can bind simultaneously to the clay. For the metals to be able to associate properly with the clay particles in the solution, their concentration needs to be under the CEC value for the clay. For Cu and Zn, the CEC was 0.0006 meq/L; for Cd it was 0.0004 meq/L; and for Pb 0.0002 meq/L. They are all below Illite CEC, which was 0.032 meq/L (batch 1) and 0.036 meq/L (batch 2) in the controlled system.

PART 3. IN THE FIELD

Water was collected at Strömsholm (59.500611, 16.266686 DD) using peristaltic pump without any filtration or water turn over. The pump was also used with an in-line filter to sample water (hereafter called Structor filtered water sample). The in-line filter was a polyethersulphone 0.45 µm filter (see below: Structor method). A maximum of 1 L was collected (Figure 3). The samples were labeled after the sampling site names set by Structor and SGU, for example SM18 1.2 0.2 µm, which indicates sampling site SM18 first day second subset. A or b behind the subset number means that two filters were used to get a sufficient volume and 0.2 µm, 0.4 µm etc. is the pore size of the filter.

Figure 3. Water samples from Strömsholm.

The water collected from sampling site SM17 was filtered on site using the same procedure as in the controlled experiment with the addition of a polypropylene syringe filter with pore diameter of 0.2 µm. The filtration was repeated in the laboratory after one day and again after five days. Using 15 ml Sarstedt tubes, 4 ml in 2-4 successive subsets were produced. pH was measured in all samples from Strömsholm. The filtered samples in the 15 mL tubes were acidified with 10 µL HNO3 per mL sample. Unfiltered water was microwave digested

in duplicates, 5 mL sample and 5 mL HNO3 in Teflon beakers. After digestion they were

diluted 10x with MΩ.

Analysis of trace metals in the filtered samples was done with ICP-MS, settings Table 8. Anions in all Strömsholm samples were analyzed with IEC (settings Table 9) after being filtered with 0.2 µm syringe filter. The Structor filtered water was both analyzed as it was

(14)

and after being filtered with syringe filter to see if the syringe filtration would affect the results. Analysis of the digested unfiltered water samples, was done with ICP-MS and MP-AES.

Table 8. Setting for ICP-MS during trace analysis. ICP-MS analysis

Injections 2

Integration time 0.1 s

Peak pattern Full quant

Rinse 20 s with MQ, rinse 40 s with MQ, rinse 60 s with HNO3, 60 s uptake, 20 s stabilization, measuring.

Table 9. Settings for the IEC during anion analysis IEC analysis

Column Metrosep A SUPP5 250 x 4 m

Mobile phase NaHCO3 1 mM and Na2CO3 3.2 mM

Pump rate 0.7 mL/min

Detector with suppressor

STRUCTOR METHOD

Structor used a peristaltic pump both with and without an in-line filter to gather 125 mL ground water respectively. The in-line filter was a capsule filter of 600 m2

polyethersulphone 0.45 µm filter media (Waterra FHT-45). The water was turned over and then pushed through the filter before sampling. The first capsule filter was used for samples SM14, SM13, SM20 and SM19 (in that order) until it was clogged and SM18, SM17 and B1 were sampled with a new one. At each site the sampling depth was measured. Samples were stored in a cooler until analysis.

Analysis was done by ALS Scandinavia AB, Täby and the analysis package used was

V-3b Bas, Elements in polluted water (after digestion) [14]. EVALUATION

Calibration curves, control solutions, blanks, and/or in-house reference was used in analysis with the different instruments.

RSD was accounted for in the ICP-MS analysis, where a few elements reached very high values but most stayed under 3.

(15)

Scatter plots were used in the controlled samples run to visualize the relationship between pH, pore size and filtrate volume. A relative scale of filtered sample was used to be able to compare elements as the percentage of throughput.

MATERIALS

Quantification both before and after filtration is central in this experiment and a range of instruments were used to achieve the data needed.

AES, Microwave Plasma – Atomic Emission Spectroscopy (Agilent 4200 MP-AES) was used when analyzing Zn, Cd, Cu and Pb in the controlled samples and when analyzing alkali metals, alkaline earth metals, Fe and Mn in the digested samples.

ICP-MS, Inductively Coupled Plasma – Mass Spectrometry (Agilent 7500cx) was used when analyzing trace metals in the filtered and digested Strömsholm samples. In ICP-MS the sample is ionized with Ar-plasma, the ions are separated and measured with MS. Measured isotopes were 27_Al,51_V,53_Cr,55_Mn,56_Fe,59_Co,60_Ni, 63_Cu,66_Zn,69_Ga,75_As,82_Se,85_Rb,88_Sr,95_Mo,107_Ag,111_Cd,125_Te,137_Ba,205_Tl,208_Pb, 209_{Bi and}238_U.

IEC, Ion-exchange chromatography (Metrohm) was used when analyzing anions (fluoride, chloride, nitrite, bromide, nitrate, phosphate and sulfate) in the digested Strömsholm samples. IEC uses the affinity to an ion exchanger to separate ions and polar molecules.

Microwave digestion was made to be able to quantify the total amount of each element in the Strömsholm samples. In microwave digestion, high temperature and pressure of a sample with low pH increases the solubility of metals.

Membrane filters tested was made of hydrophilic polycarbonate with 47 µm in diameter and the pore size of 1.0 µm (Poretics®_{Products, Osmonics Inc.), 0.4 µm (GE Water and}

Process Technologies), 0.2 µm (Nuclepore, GE Water and Process Technologies) and 0.05 µm (Poretics®_{Products, Osmonics Inc.). The pore density was not available at the}

manufacturers’ websites for some filters, but Calvo et al. [15] concluded that track-etched hydrophilic polycarbonate filters from Cyclopore had a pore density according to

Table 10. Area of the filters was 17.35 cm2_{. Syringe filters were also used, 25 mm}

(16)

Table 10. Pore density of membrane filters from Cyclopore examined by Calvo et al. [15] and from GE Water and Process Technologies website [16]

Pore size (µm) Pore density (pores/cm2_{) [15]} Pore density (pores/cm2_{) [16]} Mean density (%) [15] Porosity Void Volume (%) [16] 0.05 - - - - 0.2 4.5 · 109 _{6 x 10}8 1 ₁₄ _- 0.4 1.0 · 109 _{6 x 10}8 ₁₃ _{Min 4} _{Max 20} 1.0 0.2 · 109 _{6 x 10}8 2 ₁₆ _-

1 _{Values for GE Water and Process Tech. Cyclopore 0.2 µm filter, not the Nuclopore 0.2 µm filter that was}

used in the experiment.

2 _{Values for GE Water and Process Tech. 1.0 µm filter, not Osmonics that is the manufacturer of the 1.0 µm}

filter used in the experiment.

Results and discussion

THE CONTROLLED SYSTEMS

When the solution is pressed through the filter, particles that are larger than the pores or have the wrong shape make the filter start to clog and a filter cake is built up during the entire filtration (Figure 4). The amount of solution, the nature of the solution, pore diameter and pH all contributes to how quickly the filter cake builds up. In addition, the chemical composition of the solid material will influence the composition of the solution. The

primary reason for this is of course that it forms a pore system that the sample solution has to pass. This has a large impact on the equilibrium conditions since contact times are prolonged and compressed surfaces have different properties than when they are dispersed.

Pore diameter and load capacity

The membrane filters used in the experiment have spherical holes and spherical particles, pin shaped or very small particles, like water molecules, can pass through. The slight hydrophobic properties of the material will, however, to some extent repel hydrophilic components in the sample solution. As the liquid passes through the filter, the chance for smaller particles to pass through is reduced. During the filtration, chemical reactions and equilibria might be disturbed and this happened while doing the controlled sample experiment. As shown in Figur 5 the composition of the filtered solution is not constant as a function of sample volume that passed through the filter. If all metals in the dissolved phase had the same reaction patterns, the percentage in each set would be the same but this is evidently not the case and several reasons are possible. Syringes were re-used both in the test runs and later in the field, but carefully cleaned between each sample with MΩ

(17)

water. It was discovered, after some time filtrating test samples, that a pressure increase was necessary to maintain a fairly constant flow. This pressure increase is interpreted as the filters clogging point was reached. This in turn could have led to test subsets being heterogeneous with respect to the contact time between the filter cake and the solution phase. An automated approach with a pressure gauge and monitoring of the flow velocity would most likely have given a more accurate subsets to analyze. On the other hand, most pressure filtrations are done by applying a pressure either in a syringe or from a compressed gas.

Figur 5. Comparison between the first samples in each set. The percentage for the metals in each cluster should be the same otherwise chemical reactions are affected while filtrating. Filtered sample indicates how much of the added analyte that can be detectable after filtration.

As shown in Figure 18 to Figure 29 in Appendix 1 the pH 4 series produced the highest concentrations in the filtrates. The same tendency was observed for all metals and all pore diameters. However, the 1.0 µm pore size has a surprisingly high variation in the sub-sets. The fact that in some cases the recovery rate is over 100 % is most likely caused by contamination. A negative development of the trend line between the concentration in the filtered sample and the filter loading (i.e. volume of sample that has passed the filter) would indicate that the filter is starting to clog and dissolved metals are restrained by the clay particles in the filter cake. In a similar way, a positive regression indicates that metals are being released from the filter cake. Such a bias can in theory be caused by several different processes. Upon compression, the clay surface tend to release protons which lowers the pH at the solid/solution interface. This leads to a release of metals that are adsorbed by pH

0 10 20 30 40 50 60 70 80 90 100 F il te re d sa mpl e (% )

First sample in each testset

(18)

dependent mechanisms (examples in Figure 6-Figure 8). pH was, however, not measured after filtering and this theory cannot be confirmed in this experiment. Compression of the particles also favors desorption by ion-exchange since the ion concentrations in the pore system increases with increased compression. The quantitative impact of such processes are difficult to predict but would differ between elements as a function of their overall affinity for the surfaces, irrespective of mechanisms.

Low relative concentrations were found in some of the first fractions of the filtrate (Figure

9). These findings indicate that even without any buildup of clay particles, the elements

interact with the surface of the polycarbonate filter. With increased contact time during the filtration, equilibria is established and a more constant metal concentration in the filtrate is observed. Hence, care should be exercised even when the sample solution is low in particulate matter.

Figure 6. Close-up on the subsets in the start of filtration, filter with pore size 0.2µm and solution with pH 6. A positive trend line indicates that metals are being released from the filter cake. The example shows zinc, but this pattern was exhibited when filtering with some of the other filters, at different pHs and with other elements.

R² = 0,9967 R² = 0,9592 0 10 20 30 40 50 60 70 80 0 2 4 6 8 10 12 14 16 F Il te re d sa mpl e (% ) Volume filtrate (mL) Zn, pH 6, 0.2 µm filter

(19)

Figure 7. The example of Pb in the controlled samples at pH 4 and with 0.2 µm filter, shows that between 13 and 33 mL filtrate, the system was at equilibrium. Average is 70.57, median 70.26 and standard deviation is 0.78, but still only ~70 % of the ingoing amount of Pb is coming through the filter. The small graph shows the whole sequence of the filtration.

Figure 8. The graph shows a negative trend line. The example is Zn, dissolved in a solution with pH 8, which associates to the filter more and more the greater the amount of fluid that passes through. R² = 0,0002 0 10 20 30 40 50 60 70 80 10 15 20 25 30 35 40 F il te re d sa mpl e (% ) Volume filtrate (mL) Pb, pH 4, 0.2 µm filter R² = 0,9375 0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 40 45 50 F il te re d sa mpl e (% ) Volume filtrate (mL) Zn, pH 8, 0.2 µm filter 0 20 40 60 80 0 20 40 60 Pb, pH 4, 0.2 µm, whole sequence

(20)

Figure 9. Low concentrations in the first fractions indicates that the elements interacts with the filter. The impact of pH on Zn while filtrating can also be seen. Some of the measurements for pH 8 were under LOD.

pH and the hydrolysis constant

Each element interacts differently with solid materials such as the filter and the clay particles, and to predict the true outcome is not possible. For instance, pH has a large impact on the solid/solution distribution as noted before. For the practitioner, the general tendency for adsorption of cations to net negatively charged surfaces has a correspondence to the hydrolysis constant of a metal [S. Karlsson comm.]. Hence, it should be possible to extract some information about concentration of metals in the filtrate as a function of treatment in the controlled systems. As mentioned before, CdOH+ _{has the lowest hydrolysis constant}

and will be the first to dissociate into the solution, when the pH is lowered, then the order is PbOH+_{, CuOH}+_{and lastly ZnOH}+_{. Cadmium in natural waters is dominated by its}

divalent ion and is always complexed to other solution species. The most common counter ions are (bi)carbonate, hydroxide and chloride. At high sulfate concentrations, even this ion will be important for the complexation of cadmium. In reducing systems the sulfide

0 20 40 60 80 100 120 0 50 100 150 200 250 F ilte red s am ple (%) Volume filtrate (mL) Zn, 1.0 µm filter 0 20 40 60 80 100 120 140 0 50 100 F ilte red s am ple (%) Volume filtrate (mL) Zn, 0.2 µm filter 0 20 40 60 80 100 120 140 160 0 5 10 15 F ilte red s am ple (%) Volume filtrate (mL) Zn, 0.05 µm filter 0 20 40 60 80 100 0 20 40 F ilte red s am ple (%) Volume filtrate (mL) Zn, 0.4 µm filter

(21)

will dominate the dissolved species. Lead is typically present as lead(II) complexes where (bi)carbonate, hydroxide, sulfate and the halides are dominating, sometimes even as mixed coordination compounds. In addition, lead has a high affinity for many functional groups found in humic substances as well as on cell surfaces. Copper(II) ions combine with almost the same counter ions as lead(II) and both forms very stable sulfides in reducing systems. zinc(II) has a coordination pattern that resembles cadmium(II) why (bi)carbonate, hydroxide and chloride are important counter ions, as well as sulfide in reducing environments.

When comparing the concentrations in the filtrate for the different pore sizes in systems with pH 6 (Figure 10) it seems like cadmium and zinc have a similar behavior although they have hydrolysis constants that are at the far end of the metals in this study. Such a distribution would, however, be possible if the adsorbing species are the same for the two elements. It is commonly accepted that the complex surface of a clay particle contain one set of element specific coordination sites and one set of sites where any cation can adsorb. As long as the pH in the solution is above the pHZPC and the total concentration of metal

ions are below the capacities of the solid phase there will not be any major differences in adsorption efficiency. The differences that occurs with 0.05 µm filter can be caused by the lowering of pH as a response to higher compaction that is caused by the higher pressure that is applied to smaller pore sized filters.

From the example of zinc in Figure 9 (the other three metals are shown in Appendix 1 Controlled samples) the impact of pH and pore size on the metals is clear and as said earlier there is no easy way to measure all elements on the same conditions with one filter. The use of the 0.2 µm pore diameter filter at pH 4 gave the most accurate results compared to the added amount of metals, for all the four metals together (Figure 11). This is however quite an extreme pH for natural waters but is found in surface waters downstream bogs or at acid rock drainages. In this series there is a clear and pronounced difference between the metals. The order of retention is Cd = Zn < Cu < Pb which is frequently observed as the order of adsorption at this pH [17] [18]. At pH 4 the net charge of the clay particle is positive why the adsorption of the metals takes place by coordination to non-protolyzed hydroxyl sites that are specific for each cation in solution [S. Karlsson comm.] Hence, the adsorption mechanisms is by far more specific than in systems with higher pH.

(22)

Figure 10. All the filters compared to each element analyzed at pH 6. Dots with black borders are the filtered metal mix-solution without clay (called 6:4, see Table 4 and Table 5), note that for these dots in Cu-figure, another y-scale is used. Y-scale in all four figures is Filtered sample (%) and x-scale is Volume filtrate (mL).

Figure 11. Relative concentration of the filtered sample 4:3 batch 2, using pore diameter 0.2 µm filter. Dots with black border shows sample 4:4 with no clay.

0 20 40 60 80 100 0 100 200 Zn pH 6 0 20 40 60 80 100 0 100 200 Cd pH 6 0 10 20 30 40 50 60 70 80 90 0 5 10 15 20 25 30 35 0 100 200 Cu pH 6 0 10 20 30 40 50 60 70 80 0 100 200 Pb pH 6 0 20 40 60 80 100 120 140 0 10 20 30 40 50 60 F il te re d sa mpl e (% ) Volume filtrate (mL) 0.2 µm pH 4 Zn Cd Cu Pb

(23)

STRÖMSHOLM SAMPLES

All subsets from sample SM17 filtrated with 0.05 µm filter as well as sample SM17 1.3 filtrated with 0.4 µm filter and SM17 1.2 filtrated with 0.2 µm filter at Strömsholm had precipitated and were excluded from the study. The reasons are unknown, but there was evidently a shift in equilibria as a function of sampling and sample treatment. One possible artifact from systems with high amounts of silicates is the release of silicic acid when the material is exposed to pressure. When the filtrate is acidified before analysis, silica precipitates. Another explanation, that is the most likely in the case of the 0.05 µm filter, is that the filter immediately got blocked and that unfiltered water seeped beneath the sealing o-rings and into the collecting test tube. This may explain the small turbidity the filtrate exhibited directly after filtration. So this was probably a malfunction of the filter holder in combination with a water sample containing too many big particles. The amount of water from sampling site SM17 that could be pressed through the filter before clogging was for the 1.0 µm filter 13.5 mL; for the 0.4 µm filter 13 mL; for the 0.2 µm filter 9 mL; for the 0.2 µm syringe filter 2.5 mL; and for the 0.05 µm hardly anything.

So far it has been shown that depending on the filter pore size, the concentration in the filtrates depends on the loading and that it is crucial to prevent any changes in pH. In addition, the time between sampling, filtration and analysis are important as well. It seems that the best alternative is to filter as soon as possible. This depends, however, on which/what elements you want to analyze. Vanadium, chromium, manganese, cobalt and silver for example, shows the largest relative concentration on the first day of filtration while copper, cadmium and lead had the highest values on the second day of filtration. Molybdenum is the only analyzed element with the highest concentration on the fifth day of filtration.

There are in some cases large differences between the values reported from the commercial laboratory (ALS) and the samples analyzed at Örebro University (Table 11), despite the fact that they are all from the same sampling wells. Both laboratories used the same pre-treatment (digestion) of the samples prior to analysis, but the samples from ALS analysis shows frequently much lower results than the ones from the university, with exception from SM18 and B1. This is of course an important finding since identical samples should produce the same results, irrespective of treatment and analytical technique. Although the reasons are not clear the differences in time of storage, filtration procedures and digestion apparatus should be in focus. There is also the fact that some of the samples analyzed at the university had a higher particle composition due to no water turn over before sampling and, because of that, more bound elements.

(24)

Table 11. Table over the digested samples, analyzed by ALS (Structor) and with ICP-MS at Örebro University. Unit is ppb (µg/L).

Sample V Cr Co Ni Cu Zn As Mo Cd Ba Pb Guideline values - - - 70 2000 - 10 - 3 700 10 SM13 Ö.Uni. 3.19 4.17 0.62 2 4.41 4.92 13.8 2.15 13.5 <0.00 2 56.1 2.15 SM13 Structor 0.97 9 <0.9 0.2 0.68 7 2.55 <4 1.7 13.2 <0.05 36.6 0.71 3 SM14 Ö.Uni. 43.4 38.7 8.23 27.3 79.7 96.9 17.4 14.5 0.189 186 18.6 SM14 Structor 19.8 19.9 4.93 17.2 65.5 77.4 12.9 13.7 0.144 111 14.8 SM17 Ö.Uni. 287 234 59.3 130 102 405 163 7.94 0.895 726 126 SM17 Structor 7.77 9.54 8.68 17.9 12.4 19.9 7.83 0.67 4 0.089 51 5.53 SM18 Ö.Uni. 25.3 45.0 6.79 34.5 26.9 61.1 110 5.82 0.138 123 8.15 SM18 Structor 26.6 50.8 8.74 31 29.5 63 134 0.68 5 0.143 114 10.6 SM19 Ö.Uni. 395 384 78.9 186 333 633 1187 16.4 1.40 913 123 SM19 Structor 35.8 82.9 16.2 41 107 92.4 198 3.36 0.181 90.5 17.1 SM20 Ö.Uni. 4.30 4.71 8.18 25.4 15.0 28.8 1.77 0.55 1 0.123 39.5 2.05 SM20 Structor 3.08 2.63 7.98 21.6 11.6 28.8 1.63 <0.5 0.13 31.5 1.41 B1 Ö.Uni. 21.3 27.9 7.23 17.0 17.9 78.4 6.20 6.56 0.074 120 16.2 B1 Structor 23.2 24.5 11 20.2 21.5 54.4 6.18 5.82 0.128 106 26.8

When the results from the analysis of the Strömsholm samples and the guideline values are compared there are, as expected, differences there too. For example arsenic (Figure 12) are homogenous between the filter types, and that is due to the fact that arsenic is present as anions and will be in solution until the pH is below pHZPC for the clay. The two digested

samples shifts due to sample treatment. Lead (Figure 13) is very inconsistent between filtrations and this probably comes from lead hydrolyzing and forming compounds with organic materials in the water and in the filter cake during filtration. Nickel, on the other hand, is a stabile aqua ion that has weak interactions with other species just like arsenic even though nickel is a cation. Copper(II) ions are present in water below pH 6 but tend to form compounds like CuCO3 and complexes like [Cu(CO3)2]2- if pH is higher [19] thus

lowering the pH of the solution.

These results illustrate that the filtration procedure as well as the digestion are crucial for the interpretation of environmental quality. Depending on which filter is chosen, the

(25)

sample can be either within or above guideline values, i.e. the pore diameter of the filter can be used to change the outcome and thereby affect the objectivity of the analysis.

Figure 12. Comparison of the total amount of As in sample SM17 with the first day filtrations. Digest. is the digested unfiltered sample; Structor unfilt. and filt. are the samples collected by Structor and analyzed by ALS; and Structor 0.45 µm is analyzed with ICP-MS at the university. The red line symbolize WHO drinking water guidelines for As.

The ICP-MS analysis results of sample SM17 shows a contamination of cadmium, zinc, copper and lead in the filtrate from day two, filtered with a 1.0 µm filter. The values for day two and the filter 1.0 µm have thereby been removed from the figures. The probable cause of the contamination, since there were no noticeable problems with the ICP-MS, was IS in the nitric acid used to acidify the samples. Another explanation could be cadmium coming from the filter. Contaminations can be accounted for with blanks prepared at the same time as the samples. As shown in the example figures (Figure 14-Figure 16) the time aspect when filtering needs to be discussed, but it is difficult to make any general conclusions. Cadmium and arsenic, for example, decreases a great deal between the first filtration and the last one, but for copper it is not that obvious. The total content of metals does not change so it has to be the distribution within the samples that changes over time. Alumina and iron were above limit of quantification for all the samples and therefore no conclusions regarding them can be made.

1 10 100 1000 Digest. Structor unfilt. Structor filt. 0.2 µm syr.filt 0.2 µm 0.4 µm 1.0 µm Structor 0.45 µm L og C onc. (p pb ) Sample SM17 As

(26)

Figure 13. Examples of the total amount of some elements in sample SM17compared with the first day filtrations. Digest. is the digested unfiltered sample; Structor unfilt. and filt. are the samples collected by Structor and analyzed by ALS; and Structor 0.45 µm is analyzed with ICP-MS at the university. The red line symbolize WHO drinking water guidelines, the guideline for Cu is 2000 ppb.

1 10 100 1000 L og C onc . (ppb ) Sample SM17 Cu 0,01 0,1 1 10 C onc entrati on (p pb ) Sample SM17 Cd 0,1 1 10 100 1000 L og C onc . (ppb ) Sample SM17 Pb 1 10 100 1000 L og C onc . (ppb ) Sample SM17 Ni

(27)

Figure 14. ICP-MS analysis of trace metals in SM17. Comparison between pore diameter, day and sub-set. Something went wrong with filter 1.0 µm on day two and had a relative concentration of almost 12000 %. Removing those values, sample SM17 still shows a large amount Cd in three out of five filtrations.

Figure 15. ICP-MS analysis of trace metals in SM17. Example comparison between pore size, day and subset. As the previous figure samples 1.0 µm 2.1-2.3 has been removed, since they probably were contaminated.

0 100 200 300 400500 600700 800 900 0 .2 µm 1 .1 0 .2 µm 2 .1 0 .2 µm 2 .2 0 .2 µm 2 .3 0 .2 µ m 3 .1 0 .2 µm 3 .2 0 .2 µm 3 .3 1. 0 µm 1 .1 1. 0 µm 1 .2 1. 0 µm 1 .3 1. 0 µm 3 .1 1. 0 µ m 3 .2 1. 0 µm 3 .3 1. 0 µm 3 .4 0 .4 µ m 1.1 0 .4 µ m 2.1 0 .4 µ m 2.2 0 .4 µ m 2.3 0 .4 µ m 3.1 0 .4 µ m 3.2 0 .4 µ m 3.3 0 .4 µ m 3.4 0 .2 µm s yr.f il t 1 .1 a 0 .2 µm s yr.f il t 1 .1 b 0 .2 µ m s yr.f ilt 2 .1 a 0 .2 µm s yr.f il t 2 .1 b 0 .2 µm s yr.f il t 3. 1a 0 .2 µm s yr.f il t 3. 1b S tr u ct o r 0 .45 µ m 1.1 Re lat iv e c once nt rat ion (% )

Sample (pore size:day:subset)

Cd 0 5 10 15 20 25 30 0 .2 µm 1 .1 0 .2 µm 2. 1 0 .2 µ m 2 .2 0 .2 µm 2. 3 0 .2 µm 3 .1 0 .2 µm 3. 2 0 .2 µm 3 .3 1. 0 µm 1 .1 1. 0 µm 1 .2 1. 0 µm 1 .3 1. 0 µm 3 .1 1. 0 µm 3 .2 1. 0 µm 3 .3 1. 0 µ m 3 .4 0 .4 µ m 1.1 0 .4 µ m 2.1 0 .4 µ m 2.2 0 .4 µ m 2.3 0 .4 µ m 3.1 0 .4 µ m 3.2 0 .4 µ m 3.3 0 .4 µ m 3.4 0 .2 µm s yr.f il t 1 .1 a 0 .2 µm s yr.f il t 1 .1 b 0 .2 µm s yr.f il t 2 .1 a 0 .2 µm s yr.f il t 2 .1 b 0 .2 µm s yr.f il t 3. 1a 0 .2 µm sy r.f il t 3. 1b S tr u ct o r 0 .45 µ m 1.1 R e lat iv e c o nce ntr at io n (% )

(28)

Figure 16. ICP-MS analysis of trace metals in SM17. Example comparison between pore size, day and subset.

Filtering with syringe filter (standard procedure) before analysis of anions (Table 12) does not change the result of the analysis except in the case of nitrites, bromide and nitrates where concentrations of unfiltered and double filtered samples fluctuates. The Structor filtered water is mostly on the same level as the unfiltered water, but fluoride is a little bit overestimated and nitrate is underestimated compared with the acid-soluble content of the analytes in the unfiltered water.

Table 12. IEC analysis of anions. Unfiltered Strömsholm samples filtered with 0.2 µm syringe filter in the laboratory (ofilt.), Strömsholm samples filtered by Structor on site (filt.) and Strömsholm samples filtered by Structor on site and later in the laboratory with 0.2 µm syringe filter (filtx2).

Sample ID Fluoride (F) Chloride (Cl) Nitrite (NO2) Bromide (Br) Nitrate (NO3) Phosphate (PO4) Sulfate (SO4) B1 ofilt. 1.67 167 0.0258 0.415 0.0607 - 0.800 B1 filt. 1.73 169 0.0258 0.474 0.0607 - 0.981 B1 filtx2 1.63 168 0.0258 0.614 0.0607 - 0.984 SM13 ofilt. 1.90 4.85 0.0258 0.136 0.104 0.805 36.2 SM13 filt. 2.05 4.60 0.0388 0.182 0.0607 0.904 18.2 SM13 filtx2 1.99 4.59 0.0267 0.0929 0.0639 0.543 34.8 SM14 ofilt. 1.51 0.520 0.0402 0.0894 3.00 - 9.27 SM14 filt. 1.56 0.508 0.0408 0.0894 2.78 - 9.56 0 10 20 30 40 50 60 70 0 .2 µm 1 .1 0 .2 µm 2 .1 0 .2 µm 2 .2 0 .2 µm 2 .3 0 .2 µm 3. 1 0 .2 µm 3 .2 0 .2 µm 3. 3 1. 0 µm 1 .1 1. 0 µm 1 .2 1. 0 µm 1 .3 1. 0 µm 2 .1 1. 0 µm 2 .2 1. 0 µm 2. 3 1. 0 µm 3 .1 1. 0 µm 3. 2 1. 0 µm 3 .3 1. 0 µm 3 .4 0 .4 µ m 1.1 0 .4 µ m 2.1 0 .4 µ m 2.2 0 .4 µ m 2.3 0 .4 µ m 3.1 0 .4 µ m 3.2 0 .4 µ m 3.3 0 .4 µ m 3 .4 0 .2 µm s yr.f il t 1 .1 a 0 .2 µm s yr.f il t 1 .1 b 0 .2 µm s yr.f il t 2 .1 a 0 .2 µm s yr.f il t 2 .1 b 0 .2 µm s yr.f il t 3. 1a 0 .2 µm s yr.f il t 3. 1b S tr uc to r 0 .45 µ m 1. 1 Re lat iv e c once nt rat ion (% )

(29)

Sample ID Fluoride (F) Chloride (Cl) Nitrite (NO2) Bromide (Br) Nitrate (NO3) Phosphate (PO4) Sulfate (SO4) SM14 filtx2 1.52 0.536 0.0258 0.131 2.79 - 9.39 SM17 ofilt. 0.834 20.4 0.0496 0.275 0.430 - 90.6 SM17 filt. 0.960 18.0 0.0575 0.279 0.0786 - 87.1 SM17 filtx2 0.984 18.0 0.0533 0.248 0.120 - 86.8 SM18 ofilt. 0.460 22.5 0.0613 0.174 0.196 - 136 SM18 filt. 0.309 20.3 0.0471 0.172 0.0721 - 130 SM18 filtx2 0.482 20.2 0.0258 0.185 0.116 - 129 SM19 ofilt. 0.412 21.0 0.0338 0.170 0.0871 - 44.4 SM19 filt. 0.405 17.9 0.0371 0.166 0.0607 - 41.6 SM19 filtx2 0.406 18.0 0.0293 0.177 0.0646 - 41.5 SM20 ofilt. 0.292 17.3 0.0706 0.329 1.05 - 157 SM20 filt. 0.221 16.5 0.0493 0.350 0.0607 - 161 SM20 filtx2 0.236 16.5 0.0489 0.283 0.0662 - 159 Total concentrations

In four out of seven samples arsenic was high above guideline values, for example SM19 was a hundred times higher than drinking- and groundwater guidelines (Table 1). Barium, chromium, nickel, selenium and lead were also above guidelines (see Figure 17). If it depends on the site in itself or the clogged filter is hard to tell by just looking at one filtration. There is nothing in the measurements before or after (with the new filter) that can confirm that the filter was starting to clog. SM17 also had very high values on all the analyzed substances, except molybdenum, compared to the other sampling sites. The results from the anions analysis shows that for B1 values for chlorides is just under 170 mg/L and samples SM13, SM14 and B1 have fluoride values between 1.5 and 2.1 mg/L.

(30)

Figure 17. ICP-MS analysis of total amount of trace metals in the unfiltered sample SM19 from Strömsholm. Al, Mn and Fe is above the calibration range and Te is below LOD.

Conclusion

 The clogging point was hard to detect by just looking at the data produced. Due to the fact that the syringes behaved inconsistently and were handled by a human hand, sufficient volumes may not have been reached. In order to overcome this problem, applied pressure needs to be measured.

 Judging by the data and graphs produced in the controlled experiment (see

Appendix 1 Controlled samples) at least 10 mL solution needed to be filtered before

equilibrium in the membrane filter was reached. In this specific system, between 20-40 mL was optimal for filters 0.2 and 0.4 µm. The results for the filters with pore diameter of 0.05 and 1.0 µm showed to much inconsistency to make a scientific assessment.

 pH has a major impact and should always be taken into account in the analysis of solved and unsolved species. Every little shift in pH increases or decreases the metal ions in solution and since CEC for possible particles binding the ions are unknown hydrolysis constants will be unusable as a guidance for the quantities of metals that will be in solution.

 It is hard to say if the capsule filter that Structor was using is the most suitable one, but underestimation of the trace level metals occurs more often than the other way around when analyzing when using this filter. There is also the aspect of what is faster and more money efficient to be taken under consideration.

A study with more replicates is needed to get the appropriate data, just one set is not enough to be able to determine if the fluctuations are random errors or the scientific truth. Testing of other separation procedures should also be done, like centrifugation, dialysis, ion

0 500 1000 1500 Al V Cr Mn Fe Co Ni Cu Zn Ga As Se Rb Sr Mo Ag Cd Te Ba Tl Pb Bi U C once nt rat ion (p pb ) Element

Total concentration of metals in unfiltered digested sample SM19

(31)

exchange or a hybrid process. A more defined filtration process needs to be addressed, pre-cleaning procedure, pressure applied and which syringes to use for example.

When performing analyzes of groundwater one should consider what the natural occurrences of the metals are in the specific water and also how the metals spread out into the soil and groundwater. One should also reflect over the impact dissolved/colloidal/ particulate metal species have on organisms, are they toxic in measured concentrations, and if they are, can that be observed in the exposed environment? A more comprehensive study with analyzes of plants and animals, as well as water and soil, would show if nature around Strömsholm is (still) harmed by the spill.

Acknowledgments

A special thank you to Viktor Sjöberg for always lending a helping hand and providing his knowledge and to Isabell Berg for kind support with the MP-AES and other problems along the way. Acknowledgments also to Tommy Binbach and Petter Wetterholm at Structor for helping out at Strömsholm and providing all necessary information needed before and during the project.

(32)

References

[1] Ohio Environmental Protection Agency, Solid Waste Policy: Ban on Field

Filtration of Ground Water Samples from Municipal Solid Waste Landfill Facilities [OAC Rule 3745-27-10(C)(1)], Ohio EPA, 1995.

[2] J. Gimpel, H. Zhang, W. Davidson and A. C. Edwards, "In Situ Metal Speciation in Lake Surface Waters Using DGT, Dialysis, and Filtration," Environmental Science

Technology, vol. 37, pp. 138-146, 2003.

[3] A. Horowitz, K. Lum, J. Garbarino, C. Lemieux and C. Demas, "Problems

associated with using filtration to define dissolved trace element concentrations in natural water samples," Environ. Sci. Technol., pp. 954-963, 1996.

[4] F. Bergaya and M. Vayer, "CEC of clays: Measurement by adsorption of a copper ethyenediamine complex," Applied Clay Science, vol. 12, no. 3, pp. 275-280, 1997.

[5] R. Dohrmann, "Cation exchange capacity methodology II: A modified silver-thiourea method," Applied Clay Science, vol. 34, no. 1-4, pp. 38-46, 2006.

[6] F. Aprile and R. Lorandi, "Evaluation of Cation Exchange Capacity (CEC) in Tropical Soils Using Four Different Analytical Methods," Journal of Agricultural

Science, vol. 4, no. 6, pp. 278-289, 2012.

[7] WHO, Guidelines for Drinking-water Quality (Fourth Edition), World Health Organization, 2011.

[8] L. Ljung and H.-G. Jansson, "Sveriges geologiska undersöknings föreskrifter om statusklassificering och miljökvalitetsnormer för grundvatten SGU-FS 2008:2," ISSN 1653-7300 Sveriges geologiska undersökning, 2008.

[9] M. Hellsten and S. Grahn, "Havs- och vattenmyndighetens föreskrifter om

klassificering och miljökvalitetsnormer avseende ytvatten HVMFS 2013:19," Havs- och vattenmyndigheten, 2013.

(33)

[10] J. Srodon, D. Morgan, E. Eslinger, D. Eberl and M. Karlinger, "Chemistry of Illite/Smectite and End-member Illite," Clays and Clay Minerals, vol. 34, no. 4, pp. 368-378, 1986.

[11] M. Hoch, J. Alonso-Azcarate and M. Lischick, "Assessment of adsorption behavior of dibutyltin (DBT) to clay-rich sediments in comparison to the highly toxic

tributyltin (TBT)," Environmental Pollution, vol. 123, no. 2, pp. 217-227, 2003.

[12] B. Baeyens and M. Bradbury, "Cations exchange capacity measurements on illite using the sodium and cesium isotope dilution technique: Effects of the index cation, electrolyte concentration and competition: Modeling," Clay and clay minerals, vol. 52, no. 4, pp. 421-431, 2004.

[13] A.-M. Fällman, O. Holby and K. Lundberg, "Kolloiders betydelse för hållfasthet och föroreningstransport i jord," 2001. [Online]. Available: www.swedgeo.se. [Accessed 07 04 2015].

[14] "ALS Life Sciences," ALS Scandinavia AB, 2014. [Online]. Available: http://www.alsglobal.se. [Accessed 28 05 2015].

[15] J. I. Calvo, G. C. Hernandez and L. Martinez, "Pore Size Distribution in

Microporous Mambranes," Journal of Colloid and Interface Science, vol. 175, no. 1, pp. 138-150, 1995.

[16] "GE Healthcare Life Sciences," General Electric, 2015. [Online]. Available: http://www.gelifesciences.com/. [Accessed 21 05 2015].

[17] X. Gu, L. J. Evans and S. J. Barabash, "Modeling the adsorption of Cd (II), Cu (II), Ni (II), Pb (II) and Zn (II) onto montmorillonite," Geochemica et Cosmochimica

Acta, vol. 74, no. 20, pp. 5718-5728, 2010.

[18] S. G. Lu and Q. F. Xu, "Competitive adsorption of Cd, Cu, Pb and Zn by different soils of Eastern China," International Journal of Geosciences, 6 05 2008.

[19] World health oranization, "World health oranization," 2004. [Online]. Available: http://www.who.int/water_sanitation_health/dwq/chemicals/copper.pdf. [Accessed 06 07 2015].

(34)

[20] P. Suer, "Geokemisk modellering," Statens geotekniska institut, 03 2010. [Online]. Available: www.swedgeo.se. [Accessed 11 04 2015].

[21] KTH, "Visual MINTEQ," Kungliga tekniska högskolan, 2015. [Online]. Available: http://vminteq.lwr.kth.se/. [Accessed 07 04 2015].

[22] N. Plummer, E. Prestemon and D. Parkhurst, "An Interactive Code (NETPATH) For Modeling NET Geochemical Reactions Along a Flow PATH Version 2.0," U.S. Geological Survey Water-Resources Investigations Report 94-4169, 1994.

[23] D. Parkhurst and C. Appelo, "Description of Input and Examples for PHREEQC Version 3—A Computer Program for Speciation, Batch-Reaction,

One-Dimensional Transport, and Inverse Geochemical Calculations," in Modeling

Techniques, Denver, U.S. Geological Survey, 2013, pp. 1-497.

[24] Scientific Software Group , "Balance," 1998. [Online]. Available:

http://www.scisoftware.com/products/balance_details/balance_details.html. [Accessed 07 04 2015].

[25] J. Ball and K. Nordstrom, "User's manual for WATEQ4F, with revised

thermodynamic data base and test cases för calculating speciation af major, trace, and redox elements in natural waters," U.S. Geological Survey, Menlo Park, 1991.

(35)

Appendices

APPENDIX 1 CONTROLLED SAMPLES

Figure 18. The impact of pH on Cd while filtrating with polycarbonate filter with diameter 47 mm and pore size 1.0 µm.

0 20 40 60 80 100 120 0 50 100 150 200 250 F il te re d sa mpl e (% ) Volume filtrate (mL) Cd, 1.0 µm filter pH 8 pH 6 pH 4 0 10 20 30 40 50 60 70 80 0 5 10 15 20 25 30 35 40 F il te re d sa mpl e (% ) Volume filtrate (mL) Cd, 0.4 µm filter pH 8 pH 6 pH 4

(36)

0 20 40 60 80 100 120 0 20 40 60 80 100 F il te re d sa mpl e (% ). Volume filtrate (mL) Cd, 0.2 µm filter pH 8 pH 6 pH 4 0 20 40 60 80 100 120 140 0 2 4 6 8 10 12 14 16 18 F il te re d sa mpl e (% ) Volume filtrate (mL) Cd, 0.05 µm filter pH 8 pH 6 pH 4

(37)

Figure 22. The impact of pH on Pb while filtrating with polycarbonate filter with diameter 47 mm and pore size 1.0 µm. Many of the measurements for pH 6 were below LOD.

Figure 23. The impact of pH on Pb while filtrating with polycarbonate filter with diameter 47 mm and pore size 0.4 µm.

0 10 20 30 40 50 60 70 80 90 100 0 50 100 150 200 250 F il te re d sa mpl e (% ) Volume filtrate (mL) Pb, 1.0 µm filter pH 8 pH 6 pH 4 0 1 2 3 4 5 6 7 8 9 10 0 5 10 15 20 25 30 35 40 F il te re d sa mpl e (% ) Volume filtrate (mL) Pb, 0.4 µm filter pH 8 pH 6 pH 4

(38)

0 10 20 30 40 50 60 70 80 0 20 40 60 80 100 F il te re d sa mpl e (% ) Volume filtrate (mL) Pb, 0.2 µm filter pH 8 pH 6 pH 4 0 10 20 30 40 50 60 70 80 0 2 4 6 8 10 12 14 16 18 F il te re d sa mpl e (% ) Volume filtrate (mL) Pb, 0.05 µm filter pH 8 pH 6 pH 4

(39)

Figure 26. The impact of pH on Cu while filtrating with polycarbonate filter with diameter 47 mm and pore size 1.0 µm.

0 10 20 30 40 50 60 70 80 90 100 0 50 100 150 200 250 F il te re d sa mpl e (% ) Volume filtrate (mL) Cu, 1.0 µm filter pH 8 pH 6 pH 4 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 F il te re d sa mpl e (% ) Volume filtrate (mL) Cu, 0.4 µm filter pH 8 pH 6 pH 4

(40)

0 20 40 60 80 100 120 0 20 40 60 80 100 Fil te re d sa mp le (% ) Volume filtrate (m/L) Cu, 0.2 µm filter pH 8 pH 6 pH 4 0 20 40 60 80 100 120 0 2 4 6 8 10 12 14 16 18 F il te re d sa mpl e (% ) Volume filtrate (mL) Cu, 0.05 µm filter pH 8 pH 6 pH 4

(41)

APPENDIX 2 STRÖMSHOLM SAMPLES

Table 13. Calibration information from the ICP-MS analysis. Limit of detection, LOD and limit of quantification, LOQ.

Mass Element R LOD (ppb) LOQ (ppb)

27 Al 0,9999905 0,0091 0,0137 51 V 0,9999999 0,0013 0,0020 53 Cr 0,9999982 0,0056 0,0084 55 Mn 0,9999980 0,0037 0,0055 56 Fe 0,9999558 0,0314 0,0472 59 Co 1,0000000 0,0015 0,0023 60 Ni 0,9999924 0,0028 0,0043 63 Cu 0,9999998 0,0053 0,0079 66 Zn 0,9999989 0,0000 0,0000 69 Ga 0,9999992 0,0007 0,0011 75 As 0,9999995 0,0075 0,0112 82 Se 0,9999998 1,8119 2,7178 85 Rb 0,9999986 0,0021 0,0031 88 Sr 0,9999996 0,0004 0,0006 95 Mo 0,9999998 0,0058 0,0087 107 Ag 0,9999903 0,0004 0,0006 111 Cd 0,9999976 0,0019 0,0028 125 Te 0,9999975 0,0235 0,0352 137 Ba 0,9999994 0,0042 0,0062 205 Tl 0,9999999 0,0014 0,0021 208 Pb 0,9999996 0,0012 0,0019 209 Bi 0,9999828 0,0024 0,0037 238 U 0,9999983 0,0003 0,0004

(42)

Figure 30. ICP-MS analysis of total amount of trace metals in the unfiltered samples from Strömsholm. Al and Fe is above the calibration range and Se, Cd, and Te is below LOD.

Figure 31. ICP-MS analysis of total amount of trace metals in the unfiltered samples from Strömsholm. Al and Fe is above the calibration range and Te is below LOD.

0 50 100 150 200 250 Al V Cr Mn Fe Co Ni Cu Zn Ga As Se Rb Sr Mo Ag Cd Te Ba Tl Pb Bi U C once nt rat ion (p pb ) Element

Total concentration of metals in unfiltered digested sample SM13

0 50 100 150 200 250 300 Al V Cr Mn Fe Co Ni Cu Zn Ga As Se Rb Sr Mo Ag Cd Te Ba Tl Pb Bi U C once nt rat ion (p pb ) Element

(43)

Figure 32. ICP-MS analysis of total amount of trace metals in the unfiltered samples from Strömsholm. Al, Mn and Fe is above the calibration range and Se, Te is below LOD.

Figure 33. ICP-MS analysis of total amount of trace metals in the unfiltered samples from Strömsholm. Al and Fe is above the calibration range and Se and Te is below LOD.

0 100 200 300 400 500 600 700 800 Al V Cr Mn Fe Co Ni Cu Zn Ga As Se Rb Sr Mo Ag Cd Te Ba Tl Pb Bi U C once nt rat ion (p pb ) Element

Total concentration of metals in unfiltered digested sample SM17