Targeted and untargeted analysis of organic contaminants from

on-site sewage treatment facilities

Removal, fate and environmental impact

Kristin Blum

Department of Chemistry

Umeå 2018

This work is protected by the Swedish Copyright Legislation (Act 1960:729)

Dissertation for PhD

ISBN: 978-91-7601-836-1

The front cover photo was taken by Patrik Andersson and Kristin Blum during a sampling

trip.

Electronic version available at: http://umu.diva-portal.org/

Printed by: Lars Åberg, KBC Service center

Umeå, Sweden 2018

When everything feels like an uphill struggle, just think of the view

from the top. – Unknown

i

Table of Contents

Abstract .......................................................................................................... ii

Enkel sammanfattning på svenska ............................................................... iv

Abbreviations ................................................................................................ vi

List of Publications ........................................................................................ x

1. Introduction ................................................................................................ 1

2. Background ................................................................................................ 3 2.1. On-site sewage treatment facilities .............................................................................. 3 2.2. Environmental fate ....................................................................................................... 9 2.3. Environmental risk assessment ...................................................................................13 2.4. Contaminant analysis using mass spectrometry .......................................................13

3. Material and Methods ............................................................................... 19 3.1. Sampling ....................................................................................................................... 19 3.2. Sample preparation ..................................................................................................... 23 3.3. Instrumental analysis .................................................................................................. 24 3.4. Data processing ........................................................................................................... 30 3.5. Environmental risk assessment .................................................................................. 32 3.6. Mass fluxes per capita ................................................................................................. 32 3.7. Statistics ....................................................................................................................... 32 3.8. Molecular descriptors ................................................................................................. 34

4. Results and Discussion ............................................................................. 35 4.1. Prioritization of environmentally-relevant contaminants in wastewater .............. 35 4.2. Method development for urban water contaminants ............................................ 40 4.3. Removal patterns in soil beds compared to large-scale STPs ................................. 42 4.4. Wastewater-impacted surface water and sediment ................................................ 43 4.5. Environmental fate of wastewater contaminants .................................................... 45 4.6. Mass fluxes per capita at wastewater-impacted sites ............................................. 48 4.7. Removal in char-fortified soil beds ........................................................................... 50

5. Conclusions and future perspectives ........................................................ 53

6. Acknowledgements ................................................................................... 57

7. References ................................................................................................. 59

ii

Abstract

On-site sewage treatment facilities (OSSFs) are widely used all over the world to

treat wastewater when large-scale sewage treatment plants (STPs) are not

economically feasible. Although there is great awareness that the release of

untreated wastewater into the environment can lead to water-related diseases and

eutrophication, little is known about organic contaminants and their removal by

OSSFs, environmental load and fate. Thus, this PhD thesis aims to improve the

knowledge about treatment efficiencies in current OSSFs, the environmental

impact and fate of contaminants released from OSSFs, as well as how biochar

fortification in sand filter (soil beds) OSSFs might increase removal of these

contaminants. State-of-the-art analytical techniques for untargeted and targeted

analyses were used and the results evaluated with univariate and multivariate

statistics.

Environmentally-relevant contaminants discharged from OSSFs were identified

using untargeted analysis with two-dimensional gas chromatography mass

spectrometry (GC×GC-MS) and a MS (NIST) library search in combination with a

prioritization strategy based on environmental relevance. A method was

successfully developed for the prioritized contaminants using solid phase

extraction and GC×GC-MS, and the method was also applicable to untargeted

analysis. This method was applied to several studies. The first study compared

treatment efficiencies between STP and soil beds and showed that treatment

efficiencies are similar or better in soil beds, but the removal among the same type

of treatment facilities and contaminants varied considerably. Hydrophilic

contaminants were generally inadequately removed in both types of treatment

facilities and resulted in effluent levels in the nanogram per liter range.

Additionally, several prioritized and sometimes badly removed compounds were

found to be persistent, mobile, and bioavailable and two additional, untargeted

contaminants identified by the NIST library search were potentially mobile. These

contaminants were also found far from the main source, a large-scale STP, at Lake

Ekoln, which is part of the drinking water reservoir Lake Mälaren, Sweden. The

study also showed that two persistent, mobile and bioavailable contaminants were

additionally bioaccumulating in perch. Sampling for this study was carried out over

several seasons in the catchment of the River Fyris. Parts of this catchment were

affected by OSSFs, other parts by STPs. Potential ecotoxicological risks at these

iii

sites were similar or higher at those affected by STPs compared to those affected

by OSSFs. Mass fluxes per capita were calculated from these levels, which were

higher at STP-affected than at OSSF-affected sites in summer and autumn, but not

in winter. Possibly, the diffuse OSSF emissions occur at greater average distances

from the sampling sites than the STP point emissions, and OSSF-affected sites may

consequently be more influenced by fate processes.

The studies carried out suggested that there is a need to improve current treatment

technologies for the removal of hydrophilic contaminants. Thus, the final study of

this thesis investigated char-fortified sand filters (soil beds) as potential upgrades

for OSSFs using a combination of advanced chemical analysis and quantitative

structure-property relationship modeling. Removal efficiencies were calculated

from a large variety of contaminants that were identified by untargeted analysis

using GC×GC-MS and liquid chromatography ion mobility mass spectrometry as

well as library searches (NIST and Agilent libraries). On average, char-fortified

sand filters removed contaminants better than sand, partly due to an enhanced

removal of several hydrophilic contaminants with heteroatoms. After a two-year

runtime, sorption and particularly biodegradation must have contributed to the

removal of these compounds.

Generally, the combination of targeted and untargeted analysis has proven

valuable in detecting a large variety of organic contaminants, as well as unexpected

ones. The results imply that OSSFs have similar or better removal efficiencies,

similar or lower environmental risks and similar or lower mass fluxes per capita,

compared to STPs. Biochar fortification can improve the removal of organic

contaminants in soil beds, but further research is needed to find technologies that

reduce the discharge of all types of organic contaminants.

iv

Enkel sammanfattning på svenska

Enskilda avlopp och andra små avloppsreningsanläggningar används över hela

världen för att behandla avloppsvatten när storskaliga reningsverk inte är

ekonomiskt försvarbara. Det är välkänt att utsläpp av obehandlat avloppsvatten i

miljön kan leda till vattenburna sjukdomar och igenväxning. Effekter av utsläpp av

organiska föroreningar från små avlopp är däremot relativt okänt. Denna

avhandling syftar till att förbättra kunskapsläget kring hur organiska föroreningar

avskiljs i små avlopp, om dessa leder till miljöpåverkan samt om befintliga små

avlopp kan förses med biokolfilter för att förbättra reningsgraden. I avhandlingen

har avancerad kemisk analys använts för riktade analyser av välkända miljögifter

och förutsättningslös analys på jakt efter nya, okända föroreningar.

I den förutsättningslösa analysen isolerades föroreningar i avloppsvatten med en

ny fast-fas extraktionsmetod, identifierades med tvådimensionell gaskromatografi

kopplat till masspektrometri följt av ranking och prioritering med avseende på

miljörelevans. Samma metodik användes för riktad analys men då med selektiv

datautvärdering. Resultaten har utvärderats med både traditionell och multivariat

statistik. Avhandlingen är baserad på fyra studier där den första jämför rening i

stora verk med markbäddar, som är en mycket vanlig teknik för små avlopp.

Avskiljningen av föroreningar visade sig vara lika bra eller bättre i markbäddar men

också att den varierar kraftigt mellan anläggningar och mellan kemikalier.

Vattenlösliga föroreningar avlägsnades generellt sämre i båda markbäddar och

konventionella anläggningar. Andra och tredje studien genomfördes i Fyrisåns

avrinningsområde nära Uppsala där vatten och andra matriser provtogs i områden

påverkade av små avlopp eller nedströms stora kommunala reningsverk. Flera av

de föroreningar som den första studien identifierat som svåra att avskilja, visade

sig också vara långlivade, lättrörliga och biotillgängliga (påträffades i fisk). Dessa

kemikalier återfanns också långt från källan och når till Mälaren som är en av

Stockholms och Sveriges viktigaste ytvattentäkter. Utsläppt mängd föroreningar

per person och år beräknades och visade sig vara högre under sommaren och

hösten för vatten påverkade av stora reningsverk jämfört med vatten påverkade av

små avlopp, vilket inte kunde ses under vintern. Utsläppen från små avlopp sker

längre bort från provtagningsplatserna än utsläppen från stora reningsverk vilket

ökar betydelsen av biologisk nedbrytning, UV nedbrytning samt inbindning till

partiklar. En uppskattning av studerade föroreningarnas ekotoxikologiska risker

indikerade lika eller högre risker i områden påverkade av stora reningsverk jämfört

v

med områden påverkade av små avlopp. De tre första studierna visade på ett behov

av förbättrad reningsteknik för vattenlösliga föroreningar vilket var i fokus i den

sista studien. Utsläpp från markbäddar med biokolfilter studerades med avancerad

kemisk analys och multivariat statistik och resultaten pekar på att en biokol-sand

blandning i markbädden ökade avskiljningen av föroreningar jämfört med enbart

sand. Markbäddarna hade använts under två år och skillnaden i avskiljning var

relativt liten men pekade på betydelsen av framförallt biologisk nedbrytning men

också inbindning till kol.

vi

Abbreviations

3-C12-LAB 3-Phenyldodecane

4OP 4-Octyl phenol

6-C12-LAB 6-Phenyldodecane

AHTN Tonalide

ANTC Anthracene

APCI Atmospheric pressure chemical ionization

ATS Aerobic treatment system

BANT Benz(a)anthracene

BCF Bioconcentration factor

BFLA Benzo(b+k)fluoranthene

BOD Biological oxygen demand

BP-1 Benzophenone-1

BPA Bisphenol-A

BPYR Benzo(a)pyrene

CCS Collision cross section

CHR Chrysene

ChV Chronic value

EC50 Half maximal effective concentration

EDC Endocrine disrupting chemical

EHDPP 2-Ethylhexyldiphenyl phosphate

EI Electron ionization

ERA Environmental risk assessment

ESI Electrospray ionization

FLA Fluoranthene

GC×GC Two-dimensional gas chromatography

GC×GC-MS Two-dimensional gas chromatography mass

spectrometry

GC×GC-ToF-MS Two-dimensional gas chromatography time-of-flight

mass spectrometry

GC×GC-HRToF-MS Two-dimensional gas chromatography high resolution time-of-flight mass spectrometer

GC-EI Gas chromatography electron ionization

GC-MS Gas chromatography mass spectrometry

GC-sectorMS Gas chromatography sector field mass spectrometry

GPC Gel permeation chromatography

vii

GW Grey water separation/ Source separation system

H0 Null hypothesis

HCB Hexachlorobenzene

HHCB 1,3,4,6,7,8-Hexahydro-4,6,6,7,8,8,-hexamethyl-

cyclopenta[g]benzopyran (Galaxolide)

HRT Hydraulic retention time

IM Ion mobility

KOC Organic carbon-water partition coefficient

LC50 Median lethal dose

LC-ESI-IM-QToF-MS Liquid chromatography electrospray ionization ion mobility quadrupole time-of-flight mass spectrometer

LC-HRMS Liquid chromatography high resolution mass

spectrometry

LC-IM-MS Liquid chromatography ion mobility mass spectrometry LC-MS Liquid chromatography mass spectrometry

LOD Limit of detection

LOEC Lowest observed effect concentration

Log KOC Logarithm of the organic carbon-water partition

coefficient

Log KOW Logarithm of the octanol-water partition coefficient

LOQ Limit of quantification

MEC Measured environmental concentration

MKK Musk ketone

MKX Musk xylene

MLQ Method limit of quantification

MRM Multiple reaction monitoring

MS Mass spectrometry

MS/MS Tandem mass spectrometry

MSTP Medium-scale sewage treatment plant

MTBT 2-(Methylthio)benzothiazole

nBBSA N-Butylbenzenesulfonamide

NIST National Institute of Standards and Technology

NOEC No observed effect concentration

N-tot Total nitrogen

OC Organic carbon

OCR Octocrylene

OECD Organization for Economic Co-operation and

Development

OSSF On-site sewage treatment facility

viii

PAH Polycyclic aromatic hydrocarbon

PBT Persistent, bioaccumulative, and toxic

PC Principle component

PCA Principle component analysis

PEC Predicted environmental concentration

PLS Partial least squares

PMT Persistent, mobile and toxic

PNEC Predicted no effect concentration

POCIS Polar organic chemical integrative sampler

P-tot Total phosphorous

PYR Pyrene

QqQ Triple quadrupole

QSPR Quantitative structure-property relationship

QToF Quadrupole time-of-flight mass spectrometer

REACH Registration, evaluation, authorization and restriction

of chemicals

SB Soil bed

SPE Solid phase extraction

SPME Solid-phase micro extraction

SRM Selected reaction monitoring

STP Sewage treatment plant

SVHC Substances of very high concern

SW Water solubility

TBEP Tris-(2-butoxyethyl) phosphate

TBP Tributyl phosphate

TCEP Tris-(2-chloroethyl) phosphate

TCIPP Tris-(1-chloro-2-propyl) phosphate

TCP Tricresyl phosphate

TCS Triclosan

TDCPP Tris-(1,3-dichloro-2-propyl) phosphate

TEHP Tris-(2-ethylhexyl) phosphate

TMDD 2,4,7,9-Tetramethyl-5-decyn-4,7-diol

TMF Trophic magnification factor

ToF Time-of-flight mass spectrometer

TPP Triphenyl phosphate

U.S. EPA United States Environmental Protection Agency

USE Ultrasonic extraction

UV Ultraviolet

VIP Variable influence on projection

ix

vPvB Very persistent and very bioaccumulative

αTPA α-Tocopheryl acetate

x

List of Publications

I Blum KM, Anderson PL, Renman G, Ahrens L, Gros M, Wiberg K and

Haglund P, Non-target screening and prioritization of potentially

persistent, bioaccumulating and toxic domestic wastewater

contaminants and their removal in on-site and large-scale sewage

treatment plants,

Science of Total Environment (2017), 575, 265-275

II Blum KM, Anderson PL, Ahrens L, Wiberg K and Haglund P, Persistence,

mobility and bioavailability of emerging organic contaminants

discharged from sewage treatment plants,

Science of the Total Environment (2018), 612, 1532-1542

III Blum KM, Haglund P, Gao Q, Ahrens L, Gros M, Wiberg K and Anderson,

PL, Mass fluxes per capita of organic contaminants from on-site sewage

treatment facilities,

Submitted manuscript (Chemosphere, September 2017)

IV Blum KM, Gallampois C, Anderson PL, Renman G, Renman A, and

Haglund P, Comprehensive assessment of organic contaminant removal

from on-site sewage treatment facility effluent by char-fortified filter

beds,

Manuscript

xi

Author’s contributions

I The author was involved in planning the study, carrying out parts of the

sampling, and was responsible for sample preparation,

GC×GC-(HR)ToF-MS analysis, method development, target and

untargeted data processing, as well as identification, data evaluation,

interpretation and representation. The prioritization strategy was

developed together with the co-authors. The author also wrote,

submitted and revised the paper for publication.

II The author took part in the planning of the study, partly participated in

sampling and was responsible for sample preparation,

GC×GC-HRToF-MS analysis and data processing, partial GC-sectorMS

analysis and data processing, as well as identification, data evaluation,

interpretation and representation. The author also wrote, submitted and

revised the paper for publication.

III The author took part in the planning of the study, partly participated in

sampling and was responsible for sample preparation,

GC×GC-HRToF-MS analysis and data processing, partial GC-sectorMS

analysis and data processing, mass flux per capita estimations, data

interpretation and representation, as well as writing and submitting of

the manuscript.

IV The author carried out sample preparation, untargeted and targeted

GC×GC-HRToF-MS and LC-ESI-IM-QToF-MS analysis and data

treatment of additional data generated using GC-ToF-MS. The author

was also responsible for identification, data evaluation, interpretation

and representation, and wrote the manuscript.

xii

Own publications not included in this thesis

Lindberg RH, Fedorova G, Blum KM, Pulit-Prociak J, Gillman A, Järhult J,

Appelblad P and Söderström H, Online solid phase extraction liquid

chromatography using bonded zwitterionic stationary phases and tandem mass

spectrometry for rapid environmental trace analysis of highly polar hydrophilic

compounds - Application for the antiviral drug Zanamivir,

Talanta (2015), 164-169

Blum KM, Norström SH, Golovko O, Grabic R, Järhult JD, Koba O and Lindström

Söderström H, Removal of 30 active pharmaceutical ingredients in surface water

under long-term artificial UV irradiation,

Chemosphere (2017), 176, 175-182

Gros M, Blum KM, Jernstedt H, Renman G, Rodríguez-Mozaz S, Haglund P,

Anderson PL, Wiberg K, and Ahrens L, Screening and prioritization of

micropollutants in wastewaters from on-site sewage treatment facilities,

Journal of Hazardous Materials (2017), 328, 37-45

1

1. Introduction

Domestic wastewater contains a variety of organic contaminants. However, since

wastewater treatment plants were built to remove pathogens, phosphor and

nitrogen species but not to remove organic contaminants [1], their discharge into

the environment has raised concerns for ecosystem and human health [2].

Whereas centralized sewage treatment plants (STPs) provide the largest economic

benefit if the population is dense and large enough, smaller decentralized on-site

sewage treatment facilities (OSSFs) provide a larger economic benefit for smaller

communities and single households in rural or suburban areas [3]. Thus, they can

be a solution for the 90% of wastewater in developing countries that is not treated

before it is discharged into the environment and the 2.6 billion people worldwide

without proper sanitation [3]. In the industrialized world, 13% (2014) of Sweden’s

population [4], 20% (2011) of the population of the United States [5] and 3.9%

(2007) of Germany’s population [6] do not have access to a central STP. This

corresponds to 26.1 million (2007) housing units in the United States alone [7].

It is already well known that inadequately treated wastewater results in water-

related diseases and eutrophication. However, the removal of emerging organic

contaminants in OSSFs and their environmental impact has rarely been studied.

Therefore, this PhD thesis aims to improve the knowledge about treatment

efficiencies and contaminant discharge of current OSSFs, the environmental load

of these OSSFs, and the fate of released contaminants. Finally, possible

improvements in technology were examined to reduce the discharge.

Throughout this thesis, state-of-the-art analytical techniques including

comprehensive two-dimensional gas chromatography mass spectrometry

(GC×GC-MS) and liquid chromatography ion mobility mass spectrometry

(LC-IM-MS) were used for targeted and untargeted contaminant identification.

Contaminants were identified through library searches and results evaluated using

univariate and multivariate statistics.

2

Objectives

o To identify environmentally-relevant compounds that are discharged from

OSSFs using a GC×GC-MS based screening approach, to develop an

analysis method for these compounds in order to quantify the discharge

and to evaluate treatment efficiencies of OSSFs (Paper I).

o To investigate the fate of wastewater-discharged contaminants in the

aquatic environment in terms of persistence, mobility and bioavailability

(Paper II).

o To determine the environmental load of OSSFs compared to STPs using

mass flux per capita estimations, to evaluate temporal variations and

associated environmental risks (Paper III).

o To screen for contaminants in char-fortified sand filter (soil beds) effluents

using GC×GC-MS and LC-IM-MS, evaluate these filters for the removal of

organic contaminants, and to understand the relationship between

molecular properties and removal (Paper IV).

3

2. Background

2.1. On-site sewage treatment facilities

2.1.1. Technologies

According to data by Statistics Sweden, 28% of housing units (including holiday

houses) [4] and 13% of inhabitants [8] had OSSFs in Sweden in 2014, corresponding

to around 690 000 housing units. Other sources, such as the SMED report in 2015,

report 750 000 housing units with OSSFs in Sweden of which 83% treat sewage and

17% treat greywater only [9]. As source separation systems (greywater systems,

GWs) are a type of OSSF, the statistics for both categories were combined and

proportions for different treatment types recalculated (Figure 1). The most

common treatment technologies are infiltration systems, followed by septic tanks,

GWs and soil beds (or filter beds, SBs). Aerobic treatment systems (ATSs) were the

least common, present in only 2% of the households connected to OSSFs.

Figure 1. Proportions of the different OSSF types in Sweden, calculated and adapted from [9].

A septic tank consists of a container that retains wastewater and allows anaerobic

digestion and sedimentation [10] (Figure 2). Solids and digested organic matter

settle to the bottom, floatable solids rise to the top and are discharged with the

effluent from the tank [11]. In Sweden, septic tanks often have three consecutive

17%

25%

21%

16%

2%

9%

10%

source separation system

infiltration system

septic tanks (without add-on)

soil beds

aerobic treatment system

closed septic tanks

unknown

4

sedimentation compartments (“trekammarbrunn”) from which the sedimented

sludge needs to be removed periodically. The degree of macropollutant removal is

usually assessed with the biochemical oxygen demand (BOD), total nitrogen

(N-tot) that includes organic nitrogen, NH4+, NO3

-, and NO2- species, and the total

phosphorous (P-tot) consisting mainly of polyphosphates, orthophosphate (PO43-),

and organic phosphorous [12]. In septic tanks, the percentages of removed BOD,

N-tot and P-tot are approximately 20 ± 10%, 10 ± 5% and 15 ± 10%, respectively [9].

If the septic tank effluent is drained through leach fields into the ground to remove

macropollutants further, the technology is referred to as an infiltration system

(Figure 2). Other common names are drain field, soil absorption system, leach field

or soil treatment unit [10]. Degradation conditions in the uppermost soil are

usually aerobic and become more anaerobic further downgradient [13]. BOD,

N-tot, and P-tot removal in infiltration systems are around 90 ± 5%, 30 ± 10%, and

50 ± 30%, respectively [9]. Infiltration systems do not have a defined outlet, which

often makes effluent sampling impossible. Therefore, more and more enclosed SBs

with defined outlets are installed. Septic tank effluent is drained through a bed that

consist of layers of soil, gravel and sand and is surrounded by an impermeable

material [14]. The system prevents uncontrolled infiltration and the outlets make

sampling easier. Removal percentages for this type of system are 90 ± 5% for BOD,

25 ± 10% for N-tot and 40 ± 20% for P-tot [9].

ATSs, also called package treatment plants or sequential batch reactors, exist as

continuous or batch flow systems. By actively aerating the wastewater, they

promote biological activity and enhance aerobic degradation processes [15,16].

Some plants additionally use precipitation agents (e.g. ferric chloride) to remove

P-tot by precipitation of Al, Fe and Ca phosphates and sorption to Fe and Al

(hydr)oxide phases [13]. Aerobic treatment systems remove 90 ± 10% of BOD,

40 ± 10% of P-tot and 80 ± 10% of N-tot [9].

Source separation systems (greywater separation, GWs) separate blackwater (from

toilet flushing) and greywater (e.g. water from sinks, showers, bath, washing

machines and dishwashers). Greywater is treated and sometimes recycled on-site,

whereas blackwater is collected and treated separately or transported to a

municipal treatment plant. Other less common treatment technologies exist such

as constructed wetlands, urine separators and membrane bioreactors. Sometimes,

OSSFs are upgraded by filtering effluent through an inorganic sorbent to reduce

the P-tot discharge [17].

5

2.1.2. Removal processes of organic contaminants

For ease of understanding, the word “removal” is used in this thesis to address

losses of the parent compound, thus not only for mineralization but also for e.g.

transformation or sorption. To quantify the removal, removal efficiencies

(treatment efficiencies) were calculated.

Previous studies have found organic wastewater contaminants from various

chemical categories in OSSF effluent (Table 1), with concentrations ranging from

the low nanogram per liter up to several micrograms per liter [10]. Organic

contaminants originate from sources such as food, pharmaceutical intake and

excretion, cleaning and personal care products, furniture, building materials, floor

treatments, and dust. The contaminants reach the treatment facilities from toilets,

washing machines, sinks and bath tubs. Once at OSSFs, the major removal

processes are microbial transformation/biodegradation, as well as sorption to

solids and sedimentation [18,19] (Figure 2). Volatilization can be relevant for some

compounds [20]. The removal processes of on-site sewage treatment plants are

dependent on treatment technology, the wastewater composition and on the

structure and properties of a contaminant [19].

Figure 2. Schematic of an on-site sewage treatment plant and fate processes during treatment and

transport in the vadose zone. Adapted from [21].

A main driving force for retention of lipophilic compounds by soil or sludge

depends on the rejection of the solute by the hydrophilic water (solvophobic effect

[22]), due to the strong hydrogen bonding between the water molecules. Sorption

of these lipophilic compounds is caused by hydrophobic interactions of aromatic

septic tank

leach lineswell

ground water zone

infiltrationvadose zone

aerobic &

anaerobic

biodegradation

sorption

to drinking water to surface water

sources

anaerobic biodegradation

sorption & sedimentation

volatilization

to surface water

6

and aliphatic groups with lipophilic constituents of solids and the cell membranes

of microorganisms [19]. Alternatively, electrostatic interactions between

compounds’ positively charged functional groups and negatively charged surfaces

lead to adsorption [19]. Sorption to soil, as found in soil beds, and infiltration

systems and sorption to sludge, as found in septic tanks, is influenced by the soil

type and organic matter content, pH [23], hydraulic retention time (HRT) [24] and

temperature [25].

Organic carbon-water partition coefficients (KOC) that reflect partitioning into

organic carbon, increase with increasing molecular size and hydrophobicity [23] of

a contaminant. However, biodegradation decreases with a compound’s increasing

hydrophobicity, when sorption becomes more dominant [25,26] due to a reduction

in bioavailability (see section 2.2.3).

Biodegradation largely depends on the susceptibility to microbial attack, and thus

on a compound’s stability and the general microbial activity in the facility due to

nutrient availability and chemical and physical conditions (e.g. temperature).

Thus, it can be separated into ultimate biodegradation (mineralization) and

primary biodegradation (transformation of the parent compound). Furthermore,

biodegradation can be aerobic (e.g. package treatment plants [27], soil beds [28]),

mixed anaerobic-aerobic (e.g. infiltration systems [29],Figure 2) and anaerobic

(e.g. septic tanks [30]). Aerobic bacteria oxidize compounds by using oxygen as an

electron acceptor, whereas anaerobic bacteria use other terminal electron

acceptors such as nitrate, Fe(III), Mn(IV), SO42-, and CO2, thus reducing them

under anoxic conditions [31]. Anaerobic processes are slower than aerobic

processes, since these reactants supply less energy than oxygen. Although

oxidizing conditions have shown improved degradation of organic contaminants

in OSSFs [29,30], reducing conditions can, for instance, favor the degradation of

halogenated substances through reductive dehalogenation [32]. Reductive

dehalogenation belongs to the group of co-metabolism degradation processes [33],

which are defined as being when energy and carbon that are gained from the

compound (secondary substrate) are not used for microbial growth and the

degradation happens at the expense of growth-supporting substrates (primary

substrates) [33].

2.1.3. Removal efficiencies in OSSFs

Schaider et al. (2017) summarized studies of OSSFs and calculated removal

efficiencies where they were not calculated in the original studies (Table 1). In this

7

review and this thesis, removal efficiencies were calculated with the following

equation:

𝑅𝑒𝑚𝑜𝑣𝑎𝑙 𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑐𝑦 = 100% − 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝑖𝑛 𝑒𝑓𝑓𝑙𝑢𝑒𝑛𝑡

𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝑖𝑛 𝑖𝑛𝑓𝑙𝑢𝑒𝑛𝑡× 100%

Large variations were found for each compound in different facilities. For instance,

0 to 99.8% of triclosan and 27 to 76% bisphenol A were removed. Other studies

reported that the removal efficiencies were similar in OSSFs (ATSs, SBs) and STPs

(activated sludge treatment and sedimentation tanks) [5,27]. However, anaerobic

septic tanks [5,27] often performed worse than large-scale STPs. This is likely

because aerobic biodegradation and sorption to organic matter are important

removal processes in both STPs and OSSFs, whereas anaerobic septic tanks mainly

remove by sorption and anaerobic biodegradation (Figure 2)

Table 1. Chemical categories and examples of organic wastewater contaminants found in OSSF effluent

and their removal efficiencies in OSSFs summarized by [10].

Category Compound example Removal in OSSFs [10]

Fragrances Citronellol [34], tonalide [35] 29 to 99.5% (galaxolide)

Biocides Triclosan [21,34] 0 to 99.8%

UV stabilizers 2-Phenyl-5-benzimidazolesulfonic acid

[35], oxybenzone [36,37]

41 to 99% (oxybenzone)

Pharmaceuticals Ibuprofen [36,38], carbamazepine

[36,37,39,40]

0 to 85% (carbamazepine)

Hormones Estrogens [35,41] 0 to 94%

(17β-Estradiol)

Sweeteners Acesulfame [42,43] -

Polymer impurities Bisphenol A [36,38] 27 to 76%

Alkyl phenols and

ethoxylates

Nonyl-phenols and ethoxylates [21,41,44] 0 to 70% (octyl phenol)

Removal efficiencies vary considerably for the same contaminants between

different facilities and studies. These variations can be explained by several factors:

o Influent variability: The population discharging to each OSSF is very small

and thus represents only a small fraction of the total population [10]. This

also leads to a high daily and weekly variability in the inflow [45].

o Facility technology variability: Removal processes are dependent on

treatment technology e.g. soil infiltration versus aerobic treatment

systems.

8

o Site-specific variability: Each site varies in HRT, maintenance state and soil

composition. A higher HRT leads to more time for sorption and

biodegradation, limited aeration results in anaerobic conditions, and soil

infiltration depends on the site-specific soil characteristics [46], bedrock

and the groundwater table [12].

o Sampling representability: Grab sampling is often carried out without

adjustment to HRTs. Thus, the influent heterogeneity is problematic due

to a lack of mixing and lack of correspondence between influent and

effluent sample [10]. Some treatment technologies have sampling-related

issues e.g. the very common soil infiltration system rarely has a defined

outlet where effluent sampling could take place.

o Temporal and regional variability: Precipitation can affect removal

efficiencies due to dilution, and temperature differences can affect

removal processes [25]. Water consumption, product use and national

regulations differ regionally [10].

Another phenomenon often observed in studies of OSSFs and STPs are negative

removal efficiencies and negative mass balances due to higher concentrations in

effluent than in influent [47]. Possible explanations are desorption from solids [48],

lack of adjustment for HRTs and flows [49], release from broken-down feces [50],

and deconjugation (retransformation of the metabolite to its parent

compound) [19]. Additionally, an analytical bias could translate to negative

removal efficiencies due to a higher matrix content in the influent than effluent

impacting the extraction efficiency and instrumental sensitivity.

2.1.4. Biochar fortification to improve removal in OSSFs

To reduce the discharge of organic contaminants into the environment, simple and

cost-efficient upgrades for existing OSSFs need to be found. One cost-effective

option is the additional filtration of OSSF effluent though a low-cost sorbent or

adding the sorbent to the soil bed during construction of new SBs. Biochar is such

a cost-effective sorbent and is preferably produced from waste biomass [51].

Biochar is a carbon-rich material that is produced by thermochemical treatment of

biomass using, for instance, pyrolysis, torrefaction or hydrothermal carbonization

[52]. Torrefaction and pyrolysis typically convert biomass to biochar in the absence

of oxygen at 200 to 300˚C [53] and at approximately 350 to 800˚C [54], respectively,

while in hydrothermal carbonization, water-surrounded biomass is converted to

hydrochar under pressures up to 20 bar and at temperatures between 180 and

9

250˚C [54,55]. These approximate temperature-based definitions differ between

studies. Feedstock type, heating rates, and temperature conditions in these

processes determine the biochar characteristics and product fractions [52,55,56].

High carbonization temperatures (400 – 700˚C) have been shown to produce

biochars suitable for the removal of more hydrophobic contaminants due to an

increase in surface area, microporosity and hydrophobicity, while low

temperatures (200 – 300˚C) create biochars that remove polar organic

contaminants better due to more oxygen-containing functional groups and

electrostatic attraction [52,57–59].

Activated carbon is char that has been additionally processed to increase

microporosity and surface area [52]. Granulated activated carbon is widely used in

drinking water treatment to remove organic contaminants [60] and has also been

shown to remove personal care products and pharmaceuticals from wastewater

[61–63]. In addition, biochars have previously been evaluated for the removal of

specific organic water contaminants, such as pesticides [64,65], pharmaceuticals

[66–68], and color and dyes [59,69].

2.2. Environmental fate

2.2.1. Processes and pathways

Environmental fate describes the environmental “behavior” of a compound when

it is released into the environment and is determined by interconnected physical,

chemical and biological processes [70]. OSSF effluent is either discharged directly

to rivers and lakes or it is infiltrated into the soil through the vadose zone, from

where it can reach the subjacent groundwater zone or seep into surface water [12]

(Figure 2). Thus, OSSF-discharged organic contaminants have not only been found

in coastal [71,72] and surface water [73,74], but also in ground and drinking water

wells [29,30,40,75–81]. Amongst other organic contaminants, the insecticide

diethyltoluolamide (DEET), pharmaceuticals such as ibuprofen and

sulfamethoxazole, and the organophosphate tris(2-butoxyethyl) phosphate have

all been found in ground or drinking water wells.

Organic contaminants and their transformation products that enter the aquatic

environment from OSSF effluent can undergo intramedia and intermedia

transportation, dilution, volatilization, and sorption to suspended solids,

sediments and organic matter. Additionally, organic contaminants can be

transformed through biodegradation, direct and indirect photolysis, hydrolysis,

10

oxidation, and reduction [82,83]. To facilitate direct photolysis, the natural

sunlight’s radiation spectrum has to overlap with the contaminant’s absorption

spectrum. In this way, high-energy photons are absorbed, which initiate the

molecule’s transformation [84]. Indirect photolysis is the reaction with

photogenerated species such as photoexcited natural organic matter or reactive

oxygen species [85]. These fate processes are influenced by the contaminant’s

intrinsic properties such as conjugated π-systems, functional groups,

physicochemical properties [84], and by factors such as water flow, pH,

precipitation, temperature, irradiation, and biomass levels [83].

2.2.2. Persistence

Persistence describes the degradability of a compound and is often quantified in a

regulatory context, based on transformation half-lives in different medias that are

measured in laboratory tests such as biodegradability tests in water [86–88].

Current biodegradability tests from the Organization for Economic Co-operation

and Development (OECD) guidelines include three groups: ready biodegradability

tests (e.g. OECD 301), inherent biodegradability tests (e.g. OECD 302), and

simulation tests (e.g. OECD 303) [89]. Annex XIII of the Registration, Evaluation,

Authorization and restriction of Chemicals (REACH) regulation defines a

substance as persistent if its half-life exceeds 60 days, 40 days, 180 days or 120 days

in marine water, fresh water, marine sediment, or fresh water sediment and soil,

respectively. The regulation defines a substance as very persistent if the half-life in

water or sediment and soil is > 60 days, or > 180 days, respectively [90].

The overall persistence of a compound depends on the persistence in the media

into which most of the compound partitions [86,91]. Therefore, it is important to

distinguish between half-lives in an environmental media, environmental

compartment and overall half-lives in the environment [91,92]. As discussed in

earlier chapters, environmental fate, as a result also persistence, is impacted by

environmental conditions such as temperature, pH, microbial activity and

irradiation [86], which control, for instance, photolysis, volatilization and

biodegradation (sections 2.1.2. and 2.2.1). Consequently, large spatial and temporal

variability has been found when investigating the persistence of contaminants in

the field [93,94].

Field studies to determine persistence are rare and generally either utilize mass

fluxes [95] or benchmarking [93,96–98]. In this context, benchmarking is the

normalization of concentrations or rather degradation rates to that of a persistent

11

tracer [88]. If half-lives cannot be calculated, persistence is, for instance, assessed

based on attenuation between study sites [94,99].

2.2.3. Bioavailability

Bioavailability depends on interactions between contaminants, matrix and

organism and is very specific for their individual physical, chemical and biological

properties [100,101]. There have been many definitions of bioavailability in

environmental science; however, one currently often used definition distinguishes

between bioavailability and bioaccessibility. Semple et al. (2004) [102] defined the

bioavailable compound as “that which is freely available to cross an organism’s

cellular membrane from the medium the organism inhabits at a given time” and

the bioaccessible compound as what is additionally potentially available over time

(e.g. through desorption) [102]. In Reichenberg and Mayer (2006)’s [103]

bioavailability definition, they distinguished chemical activity from

bioaccessibility and defined chemical activity as the potential of a contaminant for

physicochemical processes (e.g. diffusion, sorption, partitioning), which does not

include the biological and ecological aspects [103].

Chemical methods to measure bioavailability are, for instance, equilibrium

sampling devices that, by this definition, measure the chemical activity when in

equilibrium with the freely dissolved and reversibly bound fractions [103]. For

instance, the polar organic chemical integrative sampler (POCIS) was

developed [104] to mimic respiratory exposure of aquatic organisms (biomimetric

method [100]). Thus, it can assess the cumulative aqueous exposure to bioavailable

compounds [105]. Another semi-permeable membrane sampling device that

retains the dissolved fraction are solid-phase micro extraction (SPME) fibers that

are coated with a thin polymer used for sampling and subsequent introduction into

a gas chromatograph coupled to a mass spectrometer (GC-MS) [106]. Apart from

these chemical methods that identify and quantify contaminants, there are

biological methods to measure bioavailability that measure the effect of the

bioavailable concentration on an organism such as in earthworm reproduction

tests [107].

2.2.4. Bioaccumulation

Bioaccumulation leads to higher contaminant concentrations in an organism than

derived from its direct environment [31] and happens by all routes of exposure. For

aquatic organisms, these routes are: bioconcentration (direct uptake from water

via respiratory and dermal surfaces), biomagnification (intake of contaminated

food), and ingestion of suspended particles [108,109]. Even without acute or

12

chronic effects, bioaccumulation can impact ecosystems by multi-generation

effects or effects on higher trophic levels [108].

The bioconcentration factor (BCF) can be calculated as the ratio of concentration

in the organism and the freely dissolved concentration in the water at

steady-state [109]. The bioaccumulation factor, which is usually determined from

field data, can be calculated as the ratio of concentration in the organism and the

total concentration in water [109]. Trophic magnification factors (TMFs), that are

calculated from the slope of the function between concentration and trophic level

of the organism, were proposed for assessing bioaccumulation [110]. However, the

European Union’s REACH and the United States Environmental Protection Agency

(U.S. EPA) both define a compound as bioaccumulative if the BCF in aquatic

species is ≥ 2 000 (REACH) and ≥ 1 000 (U.S. EPA), and very bioaccumulative if the

BCF is ≥ 5 000 [87,111].

2.2.5. Mobility

Mobility has been discussed in the past in the context of soil and was defined by

REACH as: “Mobility in soil is the potential of the substance (…), if released to the

environment, to move under natural forces to the groundwater or to a distance

from the site of release” [112]. Although mobility in water is currently the subject

of intense discussion and persistent, mobile and toxic (PMT) criteria have been

proposed to be included for the identification of substances of very high concern

(SVHC), a definition for mobility in water is still lacking. Hence, “the tendency of

a chemical to move in the environment”[113] describes mobility more generally and

is used in this thesis. Additionally, quantitative measures for mobility are lacking

[113], thus several options, usually used for describing water solubility and sorption

tendency, have been proposed [114]. For instance, the water solubility

(SW > 0.15 mg L−1) together with the organic carbon normalized adsorption

coefficient (log KOC < 4.5) for non-ionic compounds [113] or the pH-adjusted

octanol-water partition coefficients (DOW < -1) for ionizable compounds and

log KOW for neutral compounds [114] have all been suggested. Rogers et al. (1996)

[32] suggested a lower sorption potential for log KOW < 4 [32], which is therefore

used in this thesis. These measures, however, do not account for ionic interactions

and, since most natural surfaces are negatively charged, compounds, especially

cationic ones, could adsorb and might not be mobile [114]. Evidently, more research

on the development of better mobility quantifiers is necessary.

13

2.3. Environmental risk assessment

To identify chemicals of environmental concern, assessments need to cover various

aspects of risks to the ecosystem. For instance, chemicals can be very persistent

and very bioaccumulative (vPvB, sections 2.2.2 and 2.2.4), persistent,

bioaccumulative and toxic (PBT), or/ and endocrine disrupting chemicals (EDCs).

Effects on the endocrine system are subtle and can occur at low exposure levels

that may lead to long-term adverse effects, which are not captured with traditional

toxicity tests [31]. vPvB chemicals may as well reach critical levels in biota over time

that could induce adverse effects [31]. However, these highly lipophilic compounds

would also not be covered by standard ecotoxicity tests. Typically, environmental

risk assessments (ERAs) are completed where data on the predicted environmental

concentration (PEC) are compared with a predicted no effect concentration

(PNEC). Here, the measured environmental concentration (MEC) was used instead

of PEC. The effects are assessed by calculating predicted no effect concentrations

(PNECs), defined as “a concentration below which an unacceptable effect will most

likely not occur” [115] and relies on the assumption that the sensitivity of the most

sensitive species can be extrapolated to the entire ecosystem [115]. The European

commission recommends obtaining the PNEC by dividing the lowest short-term

(acute) or long-term (chronic) ecotoxicological endpoint from the most sensitive

species by an assessment factor to accommodate for uncertainties [115].

Uncertainties include laboratory tests to field extrapolation, intra and inter

laboratory variation, intra and inter species variation, short-term to long-term

extrapolation, tests on limited species and mixture effects in the field [115]. For

instance, if only short-term data are available, an assessment factor of 1 000 is

advised and if long-term data for three species representing three trophic levels are

available, the assessment factor can be 10 [116]. Finally, risks are characterized by

calculating the PEC/PNEC or MEC/PNEC ratios (risk quotients) for each

contaminant.

2.4. Contaminant analysis using mass spectrometry

2.4.1. Targeted and untargeted analysis strategies

Instrumental analysis of emerging contaminants can be carried out using targeted

and untargeted data acquisition (Figure 3). Targeted data acquisition aims to

identify and quantify target contaminants using mass spectrometers, such as the

triple quadrupole (QqQ), and carrying out tandem mass spectrometry (MS/MS)

experiments [117]. Specific precursor ions are selected, fragmented and selected

product ions detected, the so-called selected reaction monitoring (SRM)

transitions. If SRM is applied to multiple product ions from one or more

14

precursors, it is called multiple reaction monitoring (MRM). Experiments carried

out using targeted data acquisition techniques cannot be processed in the same

way as untargeted data acquisition [117]. However, experiments carried out using

untargeted data acquisition with MS/MS can be used for target analysis.

Mass spectrometers operating in full-spectrum mode are used in untargeted data

acquisition and allow retrospective identification without selection of analytes in

advance [118]. Data processing workflows differ between data acquired by electron

ionization (EI) in GC-MS and for soft ionization techniques in liquid

chromatography mass spectrometry (LC-MS) and GC-MS (Figure 3). Whereas

workflows for untargeted data acquired by high resolution LC-MS (LC-HRMS) are

commonly separated into target analysis, suspect and non-target screening [119],

workflows in GC-MS are less well defined (Figure 3).

Figure 3. Differences between untargeted and targeted data acquisition and data processing.

2.4.2. Untargeted analysis workflow

A simplified overview of the workflow is shown in Figure 4. Samples are prepared

for untargeted analysis by the extraction of compounds from their original

matrices and by removing potential interferences. It is important to have a generic

method that captures a broad range of analytes [120,121], but still removes matrix

interference sufficiently. The next step is to carry out instrumental analysis.

Instruments that provide high peak capacity, high resolution and accurate mass

information [122] are preferred. To date, most research facilities use Quadrupole

time-of-flight mass spectrometers (QToFs) or ToFs in GC-MS [123–129] and ToFs,

QToFs or Orbitraps in LC-MS [119,120,129–140]. The use of the soft ionization

techniques, for instance atmospheric pressure chemical ionization (APCI) or

Untargeted data acquisition

Suspect

screening

Target

analysis

Targeted

data

acquisition

Target

analysis

GC-EI LC-HRMS

Spectral

library

search

Non-

target

screening

LC-MS and

GC-MS

In-silico

tools

GC-HRMS

Soft-EI/ CI

Unknown

analysis

Target

analysis

15

electrospray ionization (ESI) in LC-MS, enables the detection of molecular ions.

Using a QToF or an Orbitrap, MS/MS spectra are acquired by either data-

independent acquisition (alternating full-spectrum acquisitions at low and high

collision energies) or data-dependent acquisition (switching between

full-spectrum acquisitions and products ion scan in scan cycles based on a product

ion list or ion abundance) [122].

LC-HRMS

Target analysis is carried out using reference standards and produce quantitative

data. Targeted data processing utilizes reference standards or in-house libraries,

built up from reference standards, to match data based on retention time, HRMS

and possibly MS/MS information [119,141]. It can be quantitative or semi-

quantitative [129]. In suspect screening, prior information indicates that certain

molecules could be present in a sample and a suspect list is

created [129]. The molecular formulae of these suspects are used

to calculate exact masses, isotope patterns, and expected

adducts, which are used for identification [122,129]. If MS/MS

spectra are available in vendor libraries, the identification

confidence is increased by matching recorded MS/MS spectra

against them. Suspect screening is initially carried out without

reference standards [119], but may produce semi-quantitative

data if a wide-range of internal and external standards has been

used. Non-target screening tries to identify the remaining

unknown components without prior information available [119].

A component is a group of accurate masses associated with one

compound [129]. The data evaluation includes processing

(feature extraction), filtering (e.g. detection frequency, blank

subtraction), identification including elemental composition

determination from the accurate mass of the molecular ion and

isotope abundance analysis, structural elucidation by search for

the molecular formula in MS libraries and substance databases,

library spectrum matching and ranking of the structure

candidates [122,133,142].

Schymanski et al. (2014) [141] developed identification confidence levels for target,

suspect and non-target screening (Figure 5). The levels run from highest

confidence level 1 to the lowest level 5. Level 5 components only have an accurate

mass, level 4 components have an unequivocal molecular formula assigned using

isotope pattern information, level 3 components have possible structure

candidates utilizing MS/MS fragmentation and/or other experimental data, level 2

Sampling

Sample

preparation

Instrumental

analysis

Data processing

Identification

Prioritization

Filtering

Confirmation

Figure 4. Simplified

screening workflow.

16

components have a probable structure assigned by additional library MS/MS

spectrum match (2a) or in silico fragmentation match (2b), and finally, level 1

components have a confirmed structure by reference standard matching based on

mass, isotope abundance and spacing, fragmentation, and retention time [141].

Target screening starts at level 1, suspect screening at level 3 and non-target

screening at level 5 [129].

Figure 5. Schymanski identification confidence levels for target analysis, suspect screening and non-target

screening in LC-HRMS. Adapted from [129,141].

GC-MS

In GC-MS based screening, the fragmentation reproducibility using EI is taken

advantage of and recorded spectra can be compared to mass spectral libraries such

as the NIST database (> 200 000 substances) [143] (Figure 3). Features with high

spectral similarity and probability count as tentatively identified. In the scope of

this thesis, screening based on GC-EI spectral library searches is referred to as non-

target screening. Since the molecular ion is often lost through hard ionization, the

identification of unknowns remains more complicated and in silico fragmentation

tools such as MetFrag [144] can be used for identification by assuming the largest

fragment is the molecular ion. Alternatively, data acquired using low energy “soft”

EI or chemical ionization (CI) can be used for unknown analysis (Figure 3). First,

the molecular formula is determined from the molecular ion and subsequently

structural candidates are found with the help of databases such as PubChem

(www.ncbi.nlm.nih.gov/pubmed) or ChemSpider (www.chemspider.com/). The

data processing workflow resembles the workflow for LC-HRMS.

The screening for thousands of chemicals makes it necessary to develop

prioritization strategies to focus on the most relevant ones. Especially in

Start Level 1 Confirmed structure

MS, MS/MS, Rt, reference standard

Level 2 Probable structure

MS, MS/MS, (2a) Library MS/MS or (2b) diagnostic evidence

Start Level 3 Tentative candidates

MS, MS/MS

Level 4 Molecular formula

MS, isotopic pattern

Start Level 5 Mass of interest

MS

Suspect

screening

Non-target

screening

Target

analysis

Identification

confidence Data requirements

17

untargeted analysis, the identification of many features without environmental

relevance information makes it sometimes necessary to filter out the most relevant

ones. Different strategies have been used: environmental risk assessment [121],

prioritization scoring based on abundance, detection frequency, exposure and

bioactivity [145] or exposure modeling [146]. The final and optional step would be

to confirm the tentatively identified compounds with reference standards (Figure

4).

2.4.3. Challenges

Untargeted analysis ideally detects all analytes accessible by the method used [133];

hence, unknown and unexpected compounds such as transformation products can

be found [133]. However, untargeted analysis presents challenges. The mass spectra

produced are not necessarily unique as, for example, isomers can result in the same

fragmentation spectrum [147]. In addition, mass abundances can vary when

measured with different instruments which affects the spectral similarity match in

e.g. NIST database searches [147]. Untargeted analysis is also limited by the

exclusion of compounds during sample preparation, chromatography and

ionization [139]. Moreover, special software or programming skills are often

needed to deal with the extensive data mining process [148] and the identification

of unknowns. In general, the handling of the large volume of data generated is

time-consuming [128,148]. Finally, it can be a disadvantage that compounds can

only be semi-quantified and not quantified [122].

2.4.4. Future considerations

Untargeted analysis by LC-HR-MS has already adopted well-accepted

nomenclature for the identification of unknowns [129], whereas nomenclature for

untargeted analysis by GC-MS remains to be defined. For instance, it is common

to describe different types of discovery workflows in GC-MS as untargeted analysis,

non-target analysis, non-target screening or screening. Additionally, LC-HRMS has

well-defined confidence levels [141] but a similar definition of confidence levels for

GC-based techniques is missing. Until consensus has been reached regarding the

proper concepts and terms in GC-MS, similar confidence levels to those in

LC-HRMS could be used (Figure 6). For instance, GC-EI most often yields highly

reproducible fragmentation patterns, which can be associated with compounds

present in the NIST library and give a candidate that is similar to a level 2a

candidate in LC-HRMS (“probable structure” through an unambiguous MS/MS

library match [141]). If the fragmentation does not match well with candidates in

experimental spectral libraries, but a computer-aided “in silico” spectra

interpretation results in a “likely candidate”, it gives an identification confidence

18

similar to level 2b in LC-HRMS. The corresponding workflow could be called

“spectral library search”. If the spectral library search or interpretation produce

several candidates of sufficient quality it gives a level 3 identification confidence,

“tentative candidates” using the LC-HRMS nomenclature [141]). The workflow to

raise a level 3 tentative candidate to a level 2 likely candidate could be called

“unknown analysis”. If the mass spectrum only has low molecular weight

fragments, and larger fragments and molecular ions are missing, the EI spectra

could be complemented with, for instance, low energy “soft” EI or CI spectra. In

addition, consensus structure elucidation could be used to rank the candidates and

find the most likely candidate [149].

Figure 6. Proposed nomenclature and identification confidence levels for untargeted analysis in GC-MS.

Start Level 1 Confirmed structure

Reference standard spectrum + retention time

Start

Level 2 Probable structure a) Library spectrum b) Other evidence

EI-fragmentation spectrum + unambiguous match (2a) library spectrum or (2b) in silico spectrum

Start

Level 3 Tentative candidate(s)

EI-fragmentation spectrum + several library or in silico tool matches

Untargeted analysis

Spectral

library

search

Unknown

analysis

Target

analysisIdentification

confidence Data requirements

19

3. Material and Methods

3.1. Sampling

The work for this thesis involved four sampling campaigns from 2013 to 2016.

Sampling campaigns I and II were carried out for Paper I. The first sampling

campaign aimed to screen a variety of OSSF types in Sweden including SBs, ATSs

and GWs to increase the knowledge of contaminants discharged from OSSFs. The

second sampling campaign, with improved sampling techniques compared to

campaign I, aimed specifically to study SBs and their treatment efficiency for the

removal of contaminants and compare them to STPs. For Papers II and III, a third

sampling campaign was carried out in the catchment of the River Fyris, close to

Uppsala, Sweden. The catchment is affected by a high proportion of the

households (35%) that are connected to OSSFs [150]. Thus, the work for Paper III

aimed to estimate the environmental load per capita from levels detected in the

river water. The work for Paper II concentrated on several environmental matrices

and followed the fate of the contaminants discharged from the large-scale STP

Kungsängsverket downriver in the catchment. To evaluate the removal of

contaminants by biochar in a long-term field setting for Paper IV, samples were

taken from an OSSF field study site that was upgraded with a char-fortified soil

bed.

3.1.1. On-site and municipal sewage treatment facilities

SBs, ATSs and GWs were selected for sampling in Paper I stage I (Table 2) to cover

some of the most widely-used OSSF types in Sweden (section 2.1.1, Figure 1). Since

infiltration systems do not have a defined outlet, these OSSF types were not

covered. Stand-alone septic tanks were also not included as sampling before a

septic tank creates heterogeneity issues.

The campaign was carried out in October and November 2013; at each facility,

influent and effluent was grab sampled. Influent samples were taken after the last

chamber of the septic tank to improve homogeneity. Additionally, samples from

conventional medium-scale STPs (MSTPs) and one large-scale STP were taken for

comparison. Influent/ effluent samples from similar OSSF types were pooled to

avoid poor representativity due to small facility sizes (1 to 40 users connected).

20

Table 2. Pooled samples (SB, ATS, GW), individual samples MSTP1, MSTP2, MSTP3, and STP Paper I

stage I.

Type Sample

name

Total persons connected/

Population equivalents

Number of

facilities

Soil beds SB 60 to 100/- 6

Aerobic treatment systems ATS ~50/- 4

Source separation systems GW ~20/- 3

Medium-scale sewage treatment plant MSTP1 -/125 1

Medium-scale sewage treatment plant MSTP2 -/2 500 1

Medium-scale sewage treatment plant MSTP3 -/1 000 1

Large-scale sewage treatment plant STP -/440 000 1

A second sampling campaign, from September 2015 to November 2015, was carried

out for Paper I stage II. Samples were taken at five SBs as representative of OSSFs

and five STPs (Table 3). SBs represent 16% of facilities in Sweden and function like

soil infiltration systems (25%) (section 2.1.1, Figure 1). To improve OSSF

representativity and removal efficiency reliability compared to stage I, SB influents

and effluents were sampled in a time-integrated manner by collecting an aliquot

every hour for one day. Influent samples were taken from the last stage of the septic

tank, and effluent samples were taken after the SB. Additionally, larger SBs that

were shared by multiple households were selected. The large and medium-scale

STPs were sampled flow-proportionally over one week and one day, respectively.

Table 3. Soil bed and sewage treatment plant samples for removal efficiency evaluation from Paper I

stage II.

Type Name Persons/Population

equivalents

Treatment steps

Soil bed SB1 30 to 40/- S, SB

SB2 60/- S, SB

SB3 14/- S, SB

SB4 80 to 90/- S, SB

SB5 39/- S, SB

Sewage

treatment

plant

STP1 -/ 1200 M, C, S, A, S, (C), (D)

STP2 -/ 110 000 1) M, C, S, A, S, C, (D) 2) M, C, S, BB, S, C, (D)

STP3 -/100 000 M, C, S, A, S, (D)

STP4 -/150 000 M, C, S, A, S, C, (D)

STP5 -/780 000 M, C, S, A, S, C, SF

S = sedimentation, M = mechanical treatment, C = chemical treatment, A = active sludge, BB = bio

bed, D = disinfection, SF = polishing sand filter, optional treatment steps are in brackets, 1) = line 1,

2) = line 2

21

3.1.2. The catchment of the River Fyris

For Paper II and Paper III, the catchment of the River Fyris in Uppsala

municipality (2 200 km2, ~210 000 inhabitants) in Sweden was studied (Figure 7).

Figure 7. Sampling sites from Paper II (A, B, C, S) shown as blue stars and sampling sites from Paper III

(OSSF A, OSSF B, OSSF C, STP A, STP B) shown as red stars. Site A equals STP B and Site S equals OSSF

C. Small brown triangles indicate medium-scale municipal sewage treatment plants (< 6500 persons,

< 4 000 population equivalences) that were in use during 2014/2015. The large triangle indicates

Kungsängsverket sewage treatment plant serving 171 000 people (160 000 population equivalences). The

shaded areas area shows the catchment of River Fyris. Maps were retrieved from

https://vattenwebb.smhi.se/hydronu/, 2017-01-16).

22

Uppsala municipality has a high proportion of households (35%) connected to

OSSFs [150]. The effluent from OSSFs and municipal STPs around Uppsala drains

into the River Fyris, which has a total catchment area of around 2 000 km2 and is

80 km long. The large-scale STP Kungsängsverket also drains into the River Fyris,

which is one of the major contributors of contaminants to Lake Ekoln, a sub-basin

of Lake Mälaren that supplies drinking water to around 2 million inhabitants in

and around Stockholm, Sweden.

Three sites primarily received OSSF effluent, whilst two sites primarily received

STP effluent. One further site was located before the River Fyris flows into Lake

Ekoln and one site was located at Lake Ekoln. The studies for Paper II and

Paper III shared two sampling sites (Site A = STP B and Site S = OSSF C), whereas

OSSF A, OSSF B, and STP A were unique to Paper III and Site B and Site C were

unique to Paper II. However, Paper III only describes grab water samples,

whereas Paper II deals with sediment, fish and effluent samples in addition to grab

samples. Grab and POCIS sampling were carried out over four months covering

four seasons, in December 2014 (winter), March 2015 (spring), June 2015 (summer)

and September 2015 (autumn). POCIS was deployed for two weeks at ~1 m under

the water’s surface. Grab water samples were collected twice in glass bottles, when

deploying and when collecting the POCIS samples, and were pooled before

extraction. Additionally, sediment core samples were taken with the top 4 cm

layers being sliced off (September 2015), a six-day composite effluent sample was

collected from Kungsängsverket STP (November 2015) and perch (Perca fluviatilis)

were caught in the River Fyris after Kungsängsverket STP (June 2014).

3.1.3. Experimental on-site sewage treatment facility site

Paper IV’s study site is described in detail in [151]. Briefly, the wastewater

originated from a total of four households and was distributed on three types of

filter beds (Table 4). The first sorbent bed was sand, the second bed was sand and

hardwood-derived biochar, and the third bed was gas concrete (Sorbulite®) with

biochar.

Table 4. Filter materials and their physical parameters. Adapted from [151].

Parameter Concrete Sand Char

Particle size (mm) 10 – 80 0.3 – 25 0.5 – 20

Porosity (%) (unsaturated) 60 45 50

Char-Concrete Sand Char-sand

Hydraulic retention time (hours) 2 – 2.5 2.1 – 3.0 2.1 – 3.0

23

Outlets enabled effluent sampling and beds were covered to avoid any leakage

caused by precipitation. The wastewater was distributed on the sorbent beds at

timed intervals. After a one-year run period, samples were taken from the influent

and from the three types of effluent (sand, char-sand, char-concrete).

3.2. Sample preparation

Samples were generally prepared with methods suitable for targeted and

untargeted analysis. Thus, preparation was focused on the extraction of a variety

of organic contaminants with different molecular properties and a limited clean-

up to reduce the loss of contaminants during an extensive sample preparation

workflow. The following sections explain concepts and give short summaries of the

methods used, while details including materials are described in their publications.

3.2.1. Water samples

Water samples were filtered before the separate extraction of suspended solids and

water phase in order to assure optimal extraction from both fractions. In Paper I

stage I, filtered water samples were extracted by liquid-liquid extraction with

dichloromethane. Thus, analytes are extracted based on partitioning between the

water and the dichloromethane phase in which the analytes are diverse soluble.

Dichloromethane was selected because it covers extractable compounds in an

extended polarity range. Briefly, water samples were added to a separatory funnel

and extracted three times with dichloromethane. Next, the extracts were combined

and evaporated. The suspended solids on filters were extracted using Soxhlet

extraction with toluene. Soxhlet extraction is a continuous extraction using a

heated solvent in a round bottom flask, a Soxhlet extractor and a reflux condenser.

Both extracts were combined for analysis.

A solid phase extraction (SPE) method for filtered surface water was developed for

Paper II and Paper III and extended to wastewater in Paper I stage II and applied

again to wastewater in Paper IV. The separation in SPE happens due to

physicochemical properties such as polarity or ionic charge. SPE cartridges are

made of different stationary phases for the separation desired. Normal phase

retains polar analytes, reversed phase retains non-polar analytes, anion-exchange

retains anions and cation-exchange retains cations. Reverse-phase Oasis HLB

cartridges were chosen, since the polymer sorbent is made from two monomers,

the hydrophilic n-vinylpyrrolidone and the lipophilic divinylbenzene, and thus has

a wide selectivity and can sorb both hydrophilic and hydrophobic compounds.

Briefly, the cartridges were conditioned with dichloromethane, acetonitrile and

water, and the filtered water samples were loaded. The analytes interacted with the

24

stationary phase and were retrained on the column. Next, the cartridges were

washed, dried under vacuum and the analytes eluted with acetonitrile and

dichloromethane. Analytes retained on Oasis HLB in POCIS samples were eluted

using the same method (Paper II).

In parallel, suspended solids on filters were extracted using ultrasonic extraction

with acetonitrile and dichloromethane (Paper I stage II, Paper II, Paper III, and

Paper IV). The filters were stepwise sonicated with acetonitrile and

dichloromethane and the analytes were dissolved by ultrasound-induced

cavitation. Cavitation is generated when sound waves travel through matter and

cause compression and expansion cycles [152]. The latter produces a negative

pressure so that bubbles build up, expand and collapse. This process facilitates the

leaching of compounds from particles [152]. Water phase extracts and suspended

solid extracts were combined and filtered through sodium sulfate to remove any

water residues before analysis.

3.2.2. Sediment and fish samples

In Paper II, sediment samples were extracted using sequential ultrasonic

extraction (USE) and the fish samples were blended with sodium sulfate and

extracted using column extraction first with acetone/n-hexane (71:29, v:v) and then

with n-hexane/diethyl ether (90:10, v:v) [153]. Since lipids were co-extracted with

this method, and since USE is an exhaustive extraction method and sediment is

rich in matrix, the co-extracted matrix had to be removed prior to analysis. Hence,

to remove large molecules without losing any analytes, gel permeation

chromatography (size exclusion chromatography, GPC) was used after USE or

column extraction. Compounds in GPC are separated according to their

hydrodynamic diameter. Thus, a gel works as a molecular sieve. Small molecules

penetrate into the pores of the gel and are retained on the column, whereas large

molecules pass through and elute first [154].

3.3. Instrumental analysis

Samples for Papers I, II, III and IV were analyzed using GC-MS and samples for

Paper IV were additionally analyzed using LC-MS. The use of GC-MS enabled the

detection of semi-volatile, volatile and lipophilic compounds that are stable at high

temperatures [119], whereas thermolabile and more hydrophilic compounds were

detected using LC-MS [119]. While instrumental and operational details are

explained in the respective papers, the methods used are summarized and

explained in the following sections.

25

3.3.1. Two-dimensional gas chromatography time-of-flight mass

spectrometry

GC×GC-MS was used throughout this work since the coupling of two columns with

different separation mechanisms increases the separation space and thus allows

better separation of analytes of interest from interference in complex samples,

without requiring extensive sample preparation. Hence, GC×GC gives higher

resolving power, peak capacity and provides cleaner MS spectra. In complex

matrices such as wastewater, the probability of a correct assignment and (semi-)

quantification is therefore enhanced.

In Paper I stage I, a low-resolution two-dimensional gas chromatograph coupled

to a time-of-flight mass spectrometer (GC×GC-ToF-MS) was used, with a

(midpolar) 50% phenyl polysilphenylene-siloxane phase in the primary column

and a (nonpolar) 100% dimethylpolysiloxane phase in the secondary column. The

instrument was replaced with a high-resolution GC×GC-ToF-MS for Paper I stage

II, as well as Papers II, III and IV. For this setup, the primary column was coated

with a (nonpolar) 5% phenyl phase and the secondary column was coated with a

(midpolar) 50% phenyl phase. Both instruments were made by Leco (St. Joseph,

MI, USA). Additionally, for improved sensitivity of some target analytes, a

magnetic sector field GC-HRMS (GC-sectorMS) from Waters (Milford, MA, USA)

was used for the targeted analysis described in Papers II and III. Helium was used

as carrier phase in all three instruments. Since GC×GC-ToF-MS was used most

often in this thesis, it is explained in more detail in the following section.

Modulator

1 D column

2D

column

GC oven

Secondary

GC oven

MS

Injector

Figure 8. Schematic of a GC×GC system adapted from [183] to the GC×GC-ToF-MS by Leco used for this

thesis.

26

Primarily, separation in GC happens because of the different volatilities of the

analytes. More volatile compounds partition more into the gas phase and thus

move faster and elute earlier. As a secondary process, analytes are separated by

interaction with the stationary phase. More strongly interacting compounds are

retained longer and elute at higher temperatures. Interactions with the stationary

phase are especially important for the separation of compounds with similar

volatility.

Figure 9. Example of a two-dimensional chromatogram (A) and a one-dimensional chromatogram

showing the zoomed in slices of squalene (B).

2nd dimension

separation

1st dimension

separation

Rt 2

Rt 1

B

Slices

A

Abundance

27

Samples were injected by pulsed splitless injections and introduced onto the

column at a temperature below the boiling point of the solvent (i.e. toluene, boiling

point 111˚C) to ensure condensation of the solvent as a film on the column. The

solvent slowly evaporates, and analytes are focused in a narrow band. For the

comprehensive two-dimensional GC set-up (Figure 8), a modulator was set

between the primary column and the secondary column. The system used had a

cryogenic modulator, which repetitively trapped and released the analytes that

were eluted from the first-dimension column using cold nitrogen and hot air jets.

The modulation ensured that the separation achieved in the first dimension was

not lost in the second dimension. Instead of a classical peak as in one-dimensional

chromatography, each first-dimension peak was cut into several second-dimension

chromatogram slices (Figure 9B) and could be visualized a two-dimensional

chromatogram (Figure 9A).

Analytes are introduced from the capillary column through a transfer line into the

EI source. The EI source consists of a filament that releases electrons that are

accelerated towards an anode, thereby interacting with analytes. An electron is

expelled from the molecule and results in a radical cation (molecular ion) [155].

However, due to the high energy supplied, bonds in molecules can dissociate and

molecules fragment [155]. The most common acceleration potential is 70 eV, due

to the maximum number of ions that can be produced and since small changes in

potential do not significantly affect the fragmentation [155]. The ions are

accelerated towards a mass analyzer that measures the time an ion takes to travel

through a flight tube. The time is then translated into its mass-to-charge ratio

(m/z), so the faster an ion reaches the detector, the lower the m/z. To reach high

resolution, an electrostatic mirror (a reflectron) was used to deflect ions back

through the flight-tube towards the detector. Faster ions penetrate the reflectron

deeper and thus spend a longer time in the device than slower ions. Isobaric ions

(with the same m/z) are thus focused, which increases the resolving power. The

GC×GC-HRToF-MS by Leco used in this thesis had a “folded” flight path design

(multi-reflectron analyzer). The flight path is guided by gridless mirrors and

periodic ion lenses, which compress ion beam dispersion and result in high

resolution [156].

3.3.2. Liquid chromatography ion mobility quadrupole time-of-flight mass

spectrometry

The LC-MS system used for Paper IV consisted of a liquid chromatograph, with

electrospray ionization, coupled to an ion mobility spectrometer that was

connected to a quadrupole time-of-flight mass spectrometer

28

(LC-ESI-IM-QToF-MS). Using ion mobility (IM) had the advantage, that in

addition to the separation achieved by LC-MS, an additional dimension of

separation according to the shape-to-charge ratio was introduced. IM can resolve

isomers and increases the peak capacity [157]. Also, it allows use of the collision

cross section (CCS), which describes rotationally average size and shape [158]

(Figure 10), as a distinct qualifier for analytes in addition to m/z and retention time

[157]. The CCS is specific for each ion and thus increases the certainty of

identification [157].

Figure 10. Movement of a charged molecule through an electric field in an ion mobility drift cell filled with

Nitrogen (N2) gas resulting in a rotationally averaged collision cross section (CCS). Adapted from

Christine Gallampois (unpublished) and from presentation by Frederic Abts (Agilent Technologies,

2017-03-01).

The chromatographic separation was achieved with a reversed phase octadecyl

carbon chain bonded silica column (C18 column) as the stationary phase and

MilliQ water (0.1% formic acid)/methanol (0.1% formic acid) as the mobile phase.

The column was tempered at 40 °C throughout the run. The hydrophobic

molecules retained on the stationary phase were eluted by decreasing the

hydrophilicity of the mobile phase which was achieved by gradually increasing the

methanol content (from 10% to 95% methanol, total run time 60 min). Thus,

interactions decrease between aqueous mobile phase molecules that tend to

promote self-association over interactions with dissimilar solutes [156] (the

solvophobic effect [22] decreases) and hydrophobic compounds partition more

from the hydrophobic stationary phase into the mobile phase. Thus, a

hydrophilicity separation is obtained, and more hydrophilic molecules are retained

less and elute first.

Before introduction of the separated molecules into the ion mobility spectrometer,

the eluted analytes were ionized by ESI, an ionization method where, under

29

atmospheric pressure, an electric field is applied to liquid that passes through a

capillary [155]. Charges accumulate on the surface of the liquid located at the end

of the capillary, which breaks and forms highly charged droplets [155]. The droplets

are dispersed, and heated N2 evaporates solvent molecules from the charged

droplets [155]. With evaporation of the solvent, electrostatic repulsion between

charged species becomes larger than the surface tension, and lead to cycles of

explosions of droplets (Coulomb fission) from which gas-phase charged analyte

molecules are formed [159]. Charged molecules are introduced into the ion funnel

through the transfer line, where they are focused and the remaining gas is removed

[155] (Figure 11). Next, the ions move to the trapping funnel where they are

accumulated and introduced as packages into the drift tube. In the drift tube, the

gaseous ions move through a voltage gradient and energy gained is partially lost

through collisions with N2 molecules [158].

Figure 11. Ion mobility schematic in a LC-ESI-IM-QToF-MS from Agilent Technologies.

The drift time depends on an analyte’s mass, charge and CCS [158]. Since small

molecules have less resistance, they have a shorter drift time. From the drift tube,

the ions reach the rear funnel, where the ions are refocused and introduced into

the QToF mass analyzer. Using a quadrupole consisting of four parallel cylindrical

rods as a mass filter, ions may be filtered by the oscillating voltage applied between

one rod pair and the other pair (a fixed direct current and alternating radio

frequency are applied) [155] and only selected precursors pass through the

oscillating electric field. Between the first quadrupole and the time-of-flight mass

analyzer, a hexapole serves as a collision cell (only radio frequency is applied) [155].

The collision cell is filled with N2, so if voltage is applied, ions undergo collisions

with N2 and fragment. Filtering by m/z and fragmentation in the collision cell are

both optional and their usage depends on the goal of the experiment. Ions passing

through the quadrupole and hexapole are orthogonally accelerated towards a flight

tube, the ToF, which is explained in section 3.3.1.

For Paper IV, different experiments were carried out: i) IM plus full spectrum

acquisition without m/z filtering and fragmentation, ii) IM plus All Ions MS/MS

without m/z filtering but alternating no fragmentation and fragmentation in cycles

using 0 V in the first frame and ramped collision energies from 5 V to 90 V in the

second frame, iii) Targeted MS/MS by filtering for selected precursors that were

Ion

source

Front ion

funnel

Trapping

funnelDrift tube

Rear

funnel

Mass

analyzer

30

then fragmented at 10 V, 20 V and 40 V, and iv) Auto MS/MS by fragmenting the

most abundant precursors at 10 V, 20 V and 40 V.

3.4. Data processing

3.4.1. Two-dimensional gas chromatography time-of-flight mass

spectrometry

Targeted data processing involved searches of mass calibrated GC×GC-HRToF-MS

data files for characteristic ions of target analytes within given retention times and

mass windows. Untargeted data processing of GC×GC-HRToF-MS data included

mass calibration, peak picking, retention index calculation, and NIST library search

(Figure 12). Accurate mass spectra were converted to nominal mass spectra and

were aligned with the Java application GUINEU [160] (Paper II and Paper IV).

Extracted features were filtered based on detection frequency and blank levels and

preliminary identified (NIST) features were manually investigated for the accurate

masses of the expected molecular ion and abundant fragments. For semi-

quantification purposes, the remaining features were reprocessed for characteristic

ions within given mass and retention time windows. Since the analysis for Paper I

utilized a low-resolution GC×GC-ToF-MS, data processing excluded mass

calibration and conversion to nominal mass spectra, but otherwise resembled the

described workflow for Paper II and Paper IV.

3.4.2. Liquid chromatography ion mobility quadrupole time-of-flight mass

spectrometry

Data files were recalibrated using references masses, features were extracted, and

calibrated reference ions were used to calculate CCS values. Features were aligned,

blank subtracted, and compared to both in-house and vendor libraries (Agilent

Technologies). The identification was based on molecular formula, isotope

abundance and spacing for MS data files. In All Ions MS/MS data files, fragments

were additionally used for identification (Figure 12).

Targeted MS/MS runs were carried out for the remaining compounds that were

extracted from the MS data using Workflow I, and the results were compared to

the in-house and Agilent Technologies libraries. If the in-house or Agilent libraries

did not contain any MS/MS spectra, an attempt was made to find the information

in Massbank (https://massbank.eu/MassBank/, Norman Association, retrieved

2017-10-16). For compounds not present in Massbank, the in silico fragmentation

tool MetFrag (https://msbi.ipb-halle.de/MetFrag/) [144] was used for confirmation

or rejection. For the compounds confirmed by Targeted MS/MS (Workflow I) or

31

All Ions MS/MS (Workflow II), linear regression analysis of retention times and the

log KOW values of a set of contaminants were used to check retention time

plausibility. Finally, reference standards were purchased for most of the tentatively

compounds and compounds were considered confirmed if the retention time and

major fragments were in agreement.

Figure 12. Identification workflow for GC×GC-HRToF-MS and LC-ESI-IM-QToF-MS in Paper IV.

GC×GC-HRToF-MS LC-ESI-IM-QToF-MS

Targeted data

processing

Peak picking and

alignment

Blank subtraction +

NIST library search

Detection frequency

filtration

+

manual

investigation

including

soft EI re-run

Tentatively

identified

compounds

Identified

compounds

Feature extraction and

alignment

Targeted MS/MS mode

Library search or Metfrag

fragment confirmation

Feature extraction in influent

sample

MS mode All Ions MS/MS mode

Workflow IIWorkflow I

Blank subtraction +

library search

Identification based on

molecular formula, isotope

abundance and spacing

Library search

Identification based on

molecular formula, isotope

abundance and spacing +

MS/MS fragmentation

Targeted search in MS mode

samples

Tentatively identified compounds

Targeted

reprocessing

Confirmation with reference standards

32

3.5. Environmental risk assessment

The ecotoxicological risks of contaminants found in effluent and surface water for

Papers I and II were assessed by carrying out ERAs. First, the exposure was

evaluated by calculating the PEC for a contaminant found in effluent, using a

dilution factor of 1 000 as advised for local surface water scenarios [161] or using

the MEC for contaminants found in surface water. Next, the effects were assessed

by calculating PNECs. For Papers I and II, the chronic value (ChV) was used,

which is a long-term endpoint and is defined as “the geometric mean of the no

observed effect concentration (NOEC) and the lowest observed effect

concentration (LOEC)” [162] calculated using U.S EPA’s ECOSAR model [163]. The

most sensitive (lowest ChV) was divided by an assessment factor of 10 as advised,

if long-term data for three species representing three trophic levels were available

[116,162]. Finally, the risks were characterized for each contaminant by calculating

the PEC/PNEC or MEC/PNEC ratios (risk quotients).

3.6. Mass fluxes per capita

In Paper III, environmental loads from OSSFs were compared to STPs by

calculating mass fluxes per capita from contaminant levels detected in the river

water. Mass fluxes per capita were obtained by multiplying the measured

environmental contaminant concentrations by the river flow at each sampling site

and dividing the product by the number of people estimated to be connected to

OSSFs or STPs in the corresponding catchment area. Estimated river flows, flow

uncertainties, and catchment sizes, were obtained from the Swedish

Meteorological and Hydrological Institute’s HYPE model [164]. Uncertainties in

mass flux per capita estimations were accounted for by calculating the overall

uncertainty, using error propagation including flow, analytical and capita estimate

uncertainties.

3.7. Statistics

3.7.1. Univariate statistics

The Wilcoxon Rank Sum test is a non-parametric statistical test to assess whether

two independent populations differ and can be used when the data are not

normally distributed [165]. Numerical data are ranked from 1 to n, the ranks are

substituted for the actual data, and the sum of the ranks is calculated for each

group (A,B) [165]. If the null hypothesis is true (H0: A = B) i.e. there is no difference

between the medians of each group, then the sum of ranks of one group should be

around half of the sum of all ranks. The Wilcoxon Rank Sum test was used to test

33

for compound-specific removal efficiency differences between SBs and STPs for

Paper I. The Mann-Whitney test [166] uses similar statistics as the Wilcoxon Rank

Sum test [167] and was used for Paper III to test for differences between mass

fluxes per capita at OSSF- and STP-affected river sites. The work for Paper IV

utilized the Wilcoxon signed-rank test and the Friedman test for dependent

populations. Wilcoxon signed-rank test is also a non-parametric test based on

ranks, but can be used when comparisons are paired, which leads to positive or

negative differences between the pairs in the two sample groups [168]. Ignoring the

sign, ranks are assigned to differences between paired comparisons. Then,

differences that are negative are ascribed a negative sign [168]. Next, the test

statistics are calculated, and it is determined whether the two groups are from

populations with the same distribution (H0: A = B) [169]. The Friedman test [170–

172] is also a rank-based non-parametric test for dependent populations, but is

used to compare three or more sample groups. Ranks are assigned within each row

and the sum of ranks for each group is used to calculate the chi-squared statistics

[167]. For Paper IV, paired contaminant dependent removal efficiencies of two

different sorbents and three different sorbents were compared with the Wilcoxon

signed-rank test and with the Friedman test, respectively. The tests for Paper III

and IV were carried out using OriginLab (Version 9.1.0, Northampton, MA, USA,

www.OriginLab.com).

3.7.2. Multivariate statistics

In principle component analysis (PCA), systematic variation in data consisting of

multiple variables and observations is extracted and projected into a low-

dimensional plane (principle components) [173]. The first principle component

(PC) describes the largest variation in the data, the second principle component is

orthogonal to the first and explains the second largest variation and so on. The

projected observations are represented with score plots and the variables in

loading plots [173]. With PCAs, relationships between observations and variables

can be found and patterns, outliers, and groupings elucidated [173]. The work in

Paper I utilized PCA to study variations in contaminant removal efficiencies for

the different treatment plant types and, for Paper III, PCA was used to study

differences in contaminant mass fluxes per capita between OSSF- and STP-affected

surface water. For both papers, samples were treated as observations and

contaminant concentrations as variables.

Projections to latent structures by means of partial least squares (PLS) relate X and

Y variables to each other and model their association [173]. PLS was used to develop

quantitative structure-property relationship (QSPR) models for Paper IV.

34

Computed molecular descriptors (X variables) were used to characterize the

structural variation of compounds and to predict the measured response, here the

removal efficiencies (Y variables). Variable influence on projection (VIP) values

that are the weighted sum of squares of the PLS weights, summarizing all

components into one VIP vector [173], were used to determine significant

molecular descriptors. Internal cross validation was used to estimate the predictive

power of the principle components by dividing the data into groups and then

developing parallel models from reduced data sets. Models were developed using

the deleted data as a test set and difference between observed and predicted Y

values were calculated into a Q2 value (cross-validated R2) that gives information

about the predictive power of a model [173]. In addition, response permutation

tests were carried out to estimate the significance of the Q2 value. In this test,

randomly ordered Y-response data were modeled and the real Q2 was compared to

Q2 values of models with randomly ordered Y data. Overfitted models give high R2

and Q2 for the permutated response models [173]. Multivariate statistics were

carried out using the software package SIMCA (Version 13.0.3, Umetrics, Sartorius

Stedim Biotech, Umeå, Sweden, https://umetrics.com/).

3.8. Molecular descriptors

Molecular descriptors were calculated for the QSPR modelling in Paper IV. For

charge-dependent descriptors, the contaminant structure’s ionic state was

selected, which was most abundant at the present pH. Two-dimensional molecular

descriptors were calculated for each structure, including partial charge,

connectivity and constitutional descriptors, as well as descriptors based on

Abraham solvation parameters, physicochemical properties, surface area, and

stereochemistry. Descriptors were calculated in ACD/Percepta (Version 2017.1.3,

www.acdlabs.com) and MOE (Version, 2015.1001, Chemical Computing Group Inc.

Montreal Canada, https://www.chemcomp.com/).

35

4. Results and Discussion

4.1. Prioritization of environmentally-relevant contaminants in

wastewater

The work for Paper I aimed to increase the knowledge of contaminants discharged

from OSSFs and to evaluate treatment efficiencies of OSSFs. Thus, it consisted of

a two-staged design (Figure 13) including a screening for environmentally-relevant

compounds (stage I) and an evaluation of removal patterns and environmental

loads of SBs compared to STPs (stage II).

Figure 13. Two-staged design of the work for Paper I. The first stage consisted of sampling, contaminant

screening, prioritization and analyte selection. The second stage consisted of method development,

sampling, removal pattern and load analysis.

In Paper I stage I, the samples were analyzed by GC×GC-ToF-MS using a non-

target screening approach. Peak picking and alignment of peaks resulted in

≥ 200 000 features that were filtered based on blank detection, detection

frequency, and manual inspection, resulting in approximately 300 tentatively

identified compounds. The data set was further reduced by filtering based on

environmental relevance: i) Half-life in water ≥ 60 days (EPI Suite’s BIOWIN 3,

[163]), or ii) BCF ≥ 1 000 (EPI Suite’s BAFBCF, [163]), or iii) PEC/PNEC ≥ 0.01 (see

sections 2.3 and 3.5), or iv) European or Swedish industrial chemical with an acute

(LC50/EC50) or chronic (ChV) ecotoxicity endpoint ≤ 1.0 mg L−1 or ≤ 0.1 mg L−1,

respectively. Industrial chemicals were found from European low and high

production volume chemicals as given in [174], the EINECS database [175], and a

database of chemicals used in large amounts in Sweden compiled by the Swedish

Chemicals Agency [176]. Finally, the list was checked for anthropogenic origin and

obviously biogenic compounds were excluded.

Stage I

Sampling I

GCxGC-MS based non-target

screening

Compound prioritization

Target analyte selection

Stage II

Method development for target

analytes

Sampling II

Removal pattern analysis

Environmental load analysis

36

Environmental relevance filtering, the removal of biogenic compounds from the

data set and targeted reprocessing, resulted in 46 anthropogenic contaminants of

potential environmental concern. These compounds included the pharmaceuticals

acetyl salicylic acid, carbamazepine, ethosuximide, and mirtazapine; the pesticide

2,3-dichlorobenzonitrile, and chlorinated, alkyl and aryl organophosphates; the

UV stabilizers octyl salicylate, oxybenzone and octocrylene; the detergent

impurities 4-phenylundecane, 5-phenylundecane, 4-phenyldodecane and 6-

phenyldodecane, and the surfactants 2,4,7,9-tetramethyl-5-decyn-4,7-diol

(TMDD) and N,N,N′,N′-tetraacetylethylenediamine. These 46 contaminants were

further ranked with a scoring system from 1 to 5 for i) removal efficiency, ii) half-

life, iii) BCF, iv) PEC/PNEC, and v) maximum concentration. The lowest total score

represented the most environmentally-relevant compound. The top ranked

compounds were a fragrance (galaxolide) that scored low in all five categories, a

food additive and cosmetic ingredient (α-tocopheryl acetate) that scored low in

persistence and toxicity, the UV stabilizers (octocrylene and octyl salicylate) that

scored low in bioconcentration and removal efficiency, followed by the surfactant

additives and impurities (TMDD, 4-phenyldodecane, 6-phenyldodecane,

5-phenylundecane, 4-phenylundecane), organophosphorus plasticizers and flame

retardants (Figure 14).

The top ranked chemicals (Figure 14) were studied for Paper I stage II, Paper II,

Paper III and Paper IV. These compounds were complemented with structurally-

related compounds including several alkyl, aryl and chlorinated

organophosphates, nitro and polycyclic fragrances and the UV stabilizer

2

2

2

2

3

2

1

4

1

2

2

2

2

2

1

5

3

5

2

1

5

2

5

5

5

5

2

4

1

3

2

5

4

2

5

2

5

2

2

1

1

2

5

1

3

5

1

3

3

2

1

1

1

4

3

1

3

2

4

3

4

3

1

5

5

0 5 10 15

Galaxolide

a-Tocopheryl acetate

Octocrylene

2,4,7,9-Tetramethyl-5-decyn-4,7-diol

4-Phenyldodecane

Tris(3-chloropropyl) phosphate

Tris(1,3-dichloroisopropyl) phosphate

6-Phenyldodecane

Tri(2-chloroethyl) phosphate

Octyl salicylate

Caffeine

5-Phenylundecane

4-Phenylundecane

Total Score

Removal efficiency Half-life BCF PEC/PNEC Concentration

Figure 14. The top ranked compounds identified by GC×GC-ToF-MS, with the lowest total score and their

scoring in removal efficiency, half-life for aquatic biodegradation, bioconcentration factor, PEC/PNEC and

maximum concentration found in samples.

37

benzophenone-1. Furthermore, the set was complemented with a few reference

compounds to expand the physicochemical domain and facilitate the

interpretation of removal processes such as the classical biocides, triclosan and

hexachlorobenzene, and the polymer impurity bisphenol A. Structures of the

selected compounds and their water solubility (SW) and octanol-water partition

coefficients (log KOW) from the EPI SuiteTM toolbox (www.epa.gov, 2008) are

presented in the following pages. For Papers II, Paper III, and Paper IV, several

polycyclic aromatic hydrocarbons (PAHs) were added to the selection of target

compounds as indicators for road runoff.

Biocides

Hexachlorobenzene (HCB) CAS 118-74-1

SW = 0.0062 mg L-1

log KOW = 5.9

Triclosan (TCS) CAS 3380-34-5 SW = 29 mg L-1

log KOW = 4.7

Thiabendazolea (TBZ) CAS 148-79-8

SW = 340 mg L-1

log KOW = 2.5

Food additive

α-Tocopheryl acetate (αTPA)

CAS 58-95-7 SW = 1.3 10-7 mg L-1

log KOW = 12

Fragrances

Galaxolide (HHCB) CAS 1222-05-5 SW = 1.8 mg L-1

log KOW = 6.3

Musk ketone (MKK) CAS 81-14-1

SW = 1.9 mg L-1

log KOW = 4.3

Musk xylene (MKX) CAS 81-15-2

SW = 0.47 mg L-1

log KOW = 4.5

a Only for Paper I and Paper IV.

38

Tonalide (AHTN) CAS 1506-02-1 SW = 1.3 mg L-1

log KOW = 6.3

Linear alkyl benzenes

3-Phenyldodecanea (3-C12-LAB) CAS 2400-00-2

SW = 0.0030 mg L-1

log KOW = 7.9

6-Phenyldoceanea (6-C12-LAB) CAS 2719-62-2

SW = 0.0032 mg L-1

log KOW = 7.9

Organophosphates

Tributyl phosphate (TBP) CAS 126-73-8

SW = 280 mg L-1

log KOW = 3.8

Triphenyl phosphate (TPP) CAS 115-86-6 SW = 1.9 mg L-1

log KOW = 4.7

Tricresyl phosphateb (TCP) CAS 1330-78-5 SW = 0.36 mg L-1

log KOW = 6.3

Tris(2-chloroethyl) phosphate

(TCEP) CAS 115-96-8

SW = 7 000 mg L-1

log KOW = 1.6

Tris(1,3-dichloro-2-propyl) phosphate (TDCPP)

CAS 13674-87-8 SW = 28 mg L-1

log KOW = 3.6

Tris(1-chloro-2-propyl) phosphate (TCIPP) CAS 13674-84-5 SW = 1 200 mg L-1

log KOW = 2.9

aOnly for Paper I, Paper III, and Paper IV. bApart from tri-p-cresyl phosphate (p,p,p) shown here, there are tri-m-cresyl phosphate (m,m,m) (CAS 563-04-2), tri-o-cresyl phosphate (o,o,o) (CAS 78-30-8), and isomers consisting of combined o, p and m substituents. The CAS 1330-78-5 refers to the isomer mixture and SW and log KOW to the p,p,p isomer.

39

2-Ethylhexyldiphenyl phosphatea (EHDPP)

CAS 1241-94-7 SW = 0.19 mg L-1

log KOW = 5.7

Tris(2-butoxyethyl) phosphate (TBEP)

CAS 78-51-3 SW = 1 100 mg L-1

log KOW = 3.0

Tris(2-ethylhexyl) phosphate (TEHP)

CAS 78-42-2 SW = 0.60 mg L-1

log KOW = 9.5

Plastic/rubber additives

n-Butylbenzenesulfonamide (nBBSA)

CAS 3622-84-2 SW = 400 mg L-1

log KOW = 2.3

Bisphenol A (BPA) CAS 80-05-7

SW = 42 mg L-1

log KOW = 3.6

2-(Methylthio) benzothiazole (MTBT)

CAS 615-22-5 SW = 130 mg L-1

log KOW = 3.2

Surfactants

2,4,7,9-Tetramethyl-5-decyn-4,7-diol (TMDD)

CAS 126-86-3 SW = 26 mg L-1

log KOW = 3.6

4-Octyl phenol (4OP) CAS 1806-26-4 SW = 3.1 mg L-1

log KOW = 5.5

UV stabilizers

Benzophenone-1 (BP-1) CAS 119-61-9

SW = 140 mg L-1

log KOW = 3.1

Octocrylene (OCR) CAS 6197-30-4

SW = 0.0038 mg L-1

log KOW = 6.9

aFor Paper III, 2-Ethylhexyldiphenyl phosphate was wrongly presented as CAS 109925-03-3, SW = 0.12 mg L-1 and log KOW = 6.3.

40

Polycyclic aromatic hydrocarbonsa

Anthracene (ANTC)

CAS 120-12-7 SW = 0.043 mg L-1

log KOW = 4.5

Fluoranthene (FLA) CAS 206-44-0

SW = 0.26 mg L-1

log KOW = 4.9

Pyrene (PYR) CAS 129-00-0

SW = 0.14 mg L-1

log KOW = 4.9

Benz(a)anthracene (BANT) CAS 56-55-3

SW = 0.0094 mg L-1

log KOW = 5.5

Chrysene (CHR) CAS 218-01-9

SW = 0.0020 mg L-1

log KOW = 5.5

Benzo(k)fluoranthene (BFLA) CAS 207-08-9

SW = 0.0008 mg L-1

log KOW = 6.1

Benzo(b)fluoranthene (BFLA) CAS 205-99-2

SW = 0.0015 mg L-1

log KOW = 5.8

Benzo(a)pyrene (BPYR) CAS 50-32-8

SW = 0.0016 mg L-1

log KOW = 6.1

4.2. Method development for urban water contaminants

A SPE method for filtered surface water was developed for Paper II and Paper III

and extended to wastewater for Paper I stage II. Although the method was

developed for the selected target analytes, clean-up was minimized to enable

untargeted analysis of the extracts. The final solid phase extraction method was

tested with triplicate recovery experiments of native and labeled analytes (Figure

15).

Native analytes were recovered with median relative recoveries of 95%, 97%b, and

96% in effluent, influent, and surface water, respectively. Recoveries relative to

structurally identical or carefully matched deuterated/13C-labeled standards were

generally > 50%, except for tris(2-ethylhexyl) phosphate in influent (30 ± 1%) and

surface water (43 ± 11%), and TMDD in effluent (44 ± 3%). Only a few compounds

had recoveries > 150%: 6-Phenyldodecaneb, 3-phenyldodecanec, musk ketone,

a Only for Paper II and Paper III. b The median relative recovery was wrongly described in Paper I as 95% due to an error in the recovery of TCIPP which was, in fact, 159% instead of 15%. c 6-Phenyldodecane and 3-phenyldodecane were not validated in effluent and surface water due to a late delivery of the standards.

41

tris(2-ethylhexyl) phosphate in influent, 4-octyl phenol and tris(1-chloro-2-propyl)

phosphate (TCIPP) in effluent, and hexachlorobenzene in surface water.

Figure 15. Boxplot representing recoveries in influent, effluent, and surface water. The boxes represent the

25th, 50th and 75th percentiles, the whiskers show the 10th and 90th percentiles, the cross represents the mean

and the dots represent data points that lay outside the 10th and 90th percentiles. Analytes included in

wastewater recovery tests (n = 34) do not completely overlap with the analytes included in surface water

tests (n = 28). Analytes are listed in the Supplementary Material of Paper I and Paper III

Recoveries > 100% are caused by matrix enhancement in the liner, column and ion

source [177]. During calibration, analytes bind to active sites and are not

completely released. In turn, when analyzing samples containing matrix, the

matrix competes with analytes to bind to active sites and higher analyte responses

are observed [177]. This effect could have been avoided by extracting a non-spiked

sample and using it for preparation of the calibration standards (“matrix-matched

calibration”) [177] or the standard addition method. By spiking standard to sample

aliquots at different concentrations to obtain a calibration curve and through

extrapolation to the signal measured for the pure sample, the unknown original

sample concentration could have been found. Alternatively, adding analyte

protectants, that exhibit the same effect as matrix, to calibration standards has

been proposed [177]. Recoveries comprised the narrowest interquartile range for

surface water followed by effluent due to the lowest matrix content in this type of

sample.

In conclusion, the developed method had high extraction efficiencies for a large

variety of compounds. However, the method worked best for water with less

influent effluent surface water

25

50

75

100

125

150

175

200R

eco

ve

rie

s (

%)

42

matrix, so best for surface water followed by effluent and finally influent. The

method could be improved for several very hydrophobic compounds and the

missing structurally matched labeled standard could be complemented.

4.3. Removal patterns in soil beds compared to large-scale STPs

For Paper I stage II, the treatment efficiencies of SBs (n = 5) were compared to

STPs (n = 5) for the target contaminants selected using the prioritization strategy

from Paper I stage I; removal efficiencies were correlated to the hydrophobicity of

these compounds to draw conclusions about removal processes.

The removal efficiencies varied considerably between the different facilities, which

can be seen from the whiskers representing the 25th and 75th percentiles (Figure

16). Nevertheless, total and treatment type specific median removal efficiencies

were significantly correlated to the log KOW of each contaminant (Spearman rank

correlation, α = 0.01), due to sorption playing a significant role in the removal of

organic contaminants in both STPs and SBs. However, some of the more

hydrophilic compounds – TMDD, TCIPP, tris(1,3-dichloro-2-propyl) phosphate

(TDCPP), tributyl phosphate (TBP), and tris(2-butoxyethyl) phosphate (TBEP) –

were removed better in SBs, although only significantly for TMDD (33% vs. 0%)

and TBEP (80% vs. 68%) (Wilcoxon Rank Sum test, α = 0.01). This difference could

be due to a higher soil-to-water ratio and an increased sorption of compounds with

polarizable functional groups in SBs.

SBs, which are a type of OSSF, removed organic contaminants as effectively as STPs

with the exceptions of TBEP and TMDD, for which SBs worked significantly better.

However, these compounds were moderately hydrophobic. The removal of more

hydrophilic contaminants was inadequate by both STPs and SBs. For instance, the

median removal efficiencies of the chlorinated organophosphates

tris(2-chloro-ethyl) phosphate (TCEP), TCIPP and TDCPP were < 56% in SBs and

STPs.

For several of the contaminants, removal efficiencies by OSSFs were reported for

the first time. Amongst them, the lowest median removal efficiencies were

2-(methylthio)benzothiazole (MTBT) (64%), TDCPP (56%), TMDD (33%), TCIPP

(34%), and TCEP (19%). The low removal efficiency resulted in median effluent

concentrations of these analytes between 67 (MTBT) and 1 100 ng L-1 (TMDD).

The in-depth investigation of removal patterns was limited to SBs and well-

functioning plants only. Thus, future studies should include a variety of commonly

43

used OSSF types and investigate poor-functioning plants. The results also imply

that upgrades of both STPs and OSSFs are necessary to reduce contamination of

surface water with hydrophilic contaminants. Additionally, OSSFs could

contaminate groundwater, if the water table and OSSF density are high.

Groundwater contamination could especially represent a risk for human exposure,

if drinking water wells are installed too close to OSSFs.

Figure 16. Median removal efficiencies in sewage treatment plants (STPs) and soil beds (SBs) versus the

logarithm of the octanol–water partition coefficient (log KOW) for compounds with detection frequency =

100 % (detected in influent or effluent from the same treatment plant). Whiskers show the 25th and 75th

percentiles and log KOW values were retrieved from EPI Suite’s KOWWIN v.1.68 [163].

4.4. Wastewater-impacted surface water and sediment

For Paper II and Paper III, concentrations of the selected target analytes from

Paper I in OSSF- and STP-impacted surface water were examined, and

contaminants were tentatively identified in surface water affected by the

large-scale STP Kungsängsverket.

Concentrations were strongly impacted by water flow, thus, a low flow resulted in

higher contaminant concentrations. The flow was highest during the spring flood

in March [178,179], followed by December, June and September. Hence, highest

EHDPP

α-TPATCS

AHTNBP TPP

TBP

MTBT

TDCPP

TMDD

TCIPP

TCEP

OC

TEHPHHCBTBP

TBEP

TDCPP

TMDD

0

20

40

60

80

100

1 3 5 7 9 11 13

Media

n r

em

oval

eff

icie

ncy

(%)

log KOW

STP

SB

44

concentrations were detected in September directly after a large-scale STP (STP B

= Site A) for galaxolide (1 200 ng L−1), TMDD (240 ng L−1), TBEP (270 ng L−1),

n-butylbenzenesulfonamide (nBBSA) (130 ng L−1), TCEP (110 ng L−1), TDCPP

(62 ng L−1) and MTBT (44 ng L−1). The lowest concentrations were detected in

March, except TCIPP after the large-scale STP Kungsängsverket (1 100 ng L-1). The

high concentration could be due to increased runoff from roadside snow

contaminated with emissions from polyurethane foam from car interiors [180]. In

addition to the target contaminants, ten contaminants were tentatively identified

in surface water after the STP using GC×GC-MS and a NIST library search.

Amongst them were pharmaceuticals, an insecticide, a fragrance and plastic

impurities: profolol, tolycaine, carbamazepine, diethyltoluamide, rose acetate,

7,9-di-tert-butyl-1-oxaspiro(4,5) deca-6,9-diene-2,8-dion, 4-acetyl-morpholine,

dipropylene glycol, 1-butyl-2-pyrrolidinone, and diphenyl sulfone.

Several contaminants were also found in Lake Ekoln (Site C). They were the target

analytes benzophenone-1, fluoranthene, galaxolide, triclosan, nBBSA, TMDD,

TCEP, TCIPP, TDCPP and TBEP, and the tentatively identified compounds

dipropylene glycol and rose acetate.

Most of the contaminants found were also found in the effluent of the large-scale

STP, except a few were not detected due to higher detection limits (Paper II).

Contaminants found with high detection frequencies (≥ 50%) at OSSF-impacted

sites and in OSSF effluent (Paper I stage II) were galaxolide, TMDD, TCEP, TCIPP

and TDCPP. Contaminants found in OSSF effluent with high detection

frequencies, but rarely in OSSF impacted surface water, were MTBT, TBEP,

6-phenyldodecane, tonalide, octocrylene, triphenyl phosphate and

α-tocopheryl acetate. These compounds were predominantly found in sediment

(Paper II).

Comparing sediment and water composition profiles (Paper II), the more

hydrophilic compounds dominated the water phase, whilst the more hydrophobic

contaminants dominated the sediment phase. For instance, sediment levels

normalized to the organic carbon content (OC) at sites after the STP reached up

to 9 800 ng g−1 OC−1 for individual PAHs, 3 100 ng g−1 OC−1 for galaxolide,

1 400 ng g−1 OC−1 for 6-phenyldodecane, 730 ng g−1 OC−1 for tris(2-ethylhexyl)

phosphate, 640 ng g−1 OC−1 for tonalide, and 210 ng g−1 OC−1 for MTBT.

High environmental levels of galaxolide in surface water after the large-scale STP

presented a high potential ecotoxicological risk (MEC/PNEC = 2.4). Medium

potential risks (MEC/PNEC > 0.1) were also found for the organophosphates TBEP

45

and TCIPP at both STP sites in summer and for the UV stabilizer octocrylene, at

both OSSF- and STP-affected sites.

In conclusion, wastewater contaminants that likely originated from OSSFs and

STPs occurred in water and sediment affected by OSSFs and STPs. Seasonal

changes in flow impacted the water concentration, and several contaminants were

transported far from the source to Lake Ekoln. Lake Ekoln is part of the drinking

water reservoir, Lake Mälaren, that supplies drinking water to around 2 million

inhabitants in and around, Stockholm, Sweden. Thus, both the ecosystem and

human health could be at risk.

4.5. Environmental fate of wastewater contaminants

The aim of the work for Paper II was to assess the fate (persistence, mobility,

bioavailability) of emerging contaminants in the catchment of the River Fyris.

Persistence was assessed by calculating mass flux attenuations, mobility was

assessed by sorption and partition potentials along with water solubility in

addition to detection far from the source in Lake Ekoln, and bioavailability was

assessed by sampling of the bioavailable fraction of contaminants with POCIS

samplers. Additionally, fish were analyzed to determine bioaccumulative

contaminants.

To assess persistence, the mass flux attenuation from Site A to Site B in surface

water during summer and autumn was calculated for the top ten detected

compounds at sites A, B, and C in June and September (detection frequency > 50 %,

n = 6). In addition to an attenuation < 50%, contaminants had to be found far from

the source (Lake Ekoln, Site C) to be considered persistent (Table 5). Compounds

fulfilling these requirements were TCIPP, TDCPP, TCEP, TMDD, nBBSA,

galaxolide and fluoranthene. These compounds have relatively stable molecular

structures e.g. chlorinated aliphatic moieties and stabilizing conjugated π-systems.

Furthermore, attenuations were higher in June and September, possibly due to an

increase in microbial activity at higher temperatures (15 to 17°C in June and

September compared to 3 to 6°C in December) and an increase of indirect

photolysis through excited dissolved organic substances by the higher irradiation

intensity in summer and autumn.

Mobility, which is a compound’s potential to be transported from the source [114]

without removal through for instance sorption to particulate matter, was

estimated for the studied target compounds for Paper II. Regulated quantifiers for

mobility are currently unavailable [113], thus the analytes were evaluated by their

46

log KOW/ log DOW plus SW and experimentally determined log KOC and SW

(section 2.2.5). Log KOC values could only be calculated, if the compound was found

in both water and sediment phase. According to their log KOW (< 4) [32] and

SW (> 0.15 mg L-1) [113], resulting in low sorption potentials, possible mobile target

analytes are TBP, TDCPP, TCIPP, TCEP, TBEP, nBBSA, MTBT, TMDD and

benzophenone-1 (Table 5). The Log DOW evaluation did not yield any additional

mobile contaminants. Additionally, MTBT, TBEP, galaxolide and tonalide are likely

mobile according to SW (> 0.15 mg L-1) and their log KOC (< 4.5) calculated from the

water-sediment distribution [113].

Table 5. Compounds with detection frequency > 50% (n = 6, sites A, B, and C in June and September) plus

mobile compounds), their mobility indicators (log KOW, log KOC, WS) and the mass flux attenuation from

Site A to B, the occurrence in Lake Ekoln (Paper II), SB effluent (Paper I) and OSSF-impacted surface

water (Paper III). Compounds fulfilling mobility criteria (section 2.2.5) [32,113,114] and the persistence

criteria set by the author are marked in bold.

Chemical Log KOW

Log KOC WS in mg L-1

Attenuation A to Ba)

Paper II

Lake Ekoln Paper II

SB effluent Paper I

OSSF surface water Paper III

BP-1 3.1 - 140 b)

galaxolide 6.3 3.4, 4.1 1.8 43 ± 26%

Fluoranthene 4.9 6.5, 4.7 0.26 0 ± 121%

c)

MTBT 3.2 4.0, 3.5 130 23 ± 20%

nBBSA 2.3 - 400 43 ± 12%

TBEP 3.0 3.4 1 100 71 ± 20%

TBP 3.8 - 280 19 ± 35%

TCEP 1.6 - 7 000 47 ± 16%

TCIPP 2.9 - 1 200 24 ± 36%

TDCPP 3.6 - 28 35 ± 23%

TMDD 3.6 - 26 38 ± 14%

a) Mean of June and September with standard error (n = 2) b) Detection frequency (17 %, n = 6) too

low to determine attenuation c) Fluoranthene was not analyzed for Paper I.

Apart from TBP, MTBT, benzophenone-1 and tonalide, these compounds and the

tentatively identified compounds dipropylene glycol (log KOW = -0.64,

SW = 10-6mg L-1) and rose acetate (log KOW = 3.85, SW = 19 mg L-1) were transported

far from the source to Lake Ekoln (Site C), which indicates, along with their

log KOW and SW, high mobility (Table 5). Although the main source for these

contaminants was a large-scale STP, the mobile target contaminants were also

found in SB effluent (Paper I) and in OSSF-impacted surface water (Paper III),

excluding nBBSA and TBEP, respectively. Thus, mobile contaminants are also

discharged from OSSFs and could be transported far from the source.

47

POCIS samplers can show the bioavailability of the target compounds by

mimicking the uptake of the bioavailable fraction in the water. Consequently, only

13 instead of 17 target analytes were found in POCIS samples compared to grab

samples. However, spatiotemporal trends were similar in POCIS and grab samples,

thus, sampled amounts were highest in summer when the flow was lowest. This

resulted in a higher amount available for uptake into biota when biological

productivity was at its highest.

Emerging organic contaminants that were identified for Paper II as being

persistent, mobile and bioavailable were galaxolide, TMDD, TCEP, TDCPP, and

TCIPP. Galaxolide and TMDD were additionally found in fish and thus found to be

bioaccumulative (Figure 17). As mentioned in section 4.4, galaxolide and TCIPP are

a potential risk to the aquatic ecosystem (Paper III). Thus, galaxolide and TCIPP

are not only potentially persistent, mobile and bioavailable, but also an

ecotoxicological threat.

The reoccurrence of these contaminants in the work in Paper I, Paper II and

Paper III, as well as the indication of possible ecosystem and human health risks

demand an improvement in treatment technologies to remove organic

contaminants during wastewater treatment, before they can enter the

environment. Moreover, the field-based studies to investigate persistence could be

complemented with laboratory studies of photolysis and biodegradation to

confirm that these fate processes play a key role in the removal of the studied

Figure 17. Bioavailable, persistent and mobile compounds identified for Paper II.

48

contaminants in surface water. Furthermore, the method to analyze contaminants

in fish did not cover all target analytes due to lipid interference in the GC×GC-MS

analysis (e.g. TCEP and TCIPP). Therefore, the method needs to be improved for

the removal of lipids and extended to the analytes measured in water, POCIS and

sediment. Consequently, further studies of the bioaccumulation of targeted and

untargeted analytes in fish would increase the knowledge of discharge from OSSFs

and STP and the subsequent uptake in biota. Accumulated levels in biota are not

only a threat for the ecosystem, but also humans who could consume these

organisms and could thus be exposed to toxic levels.

4.6. Mass fluxes per capita at wastewater-impacted sites

The environmental loads of on-site sewage treatment plants were compared to

large-scale sewage treatment facilities by calculating mass fluxes per capita at sites

impacted by either of them (Paper III). Mass fluxes per capita were calculated by

multiplication of the measured water concentration at the sampling site with the

river flow at the site and by division with the number of people estimated as having

their housing unit connected to OSSFs and STPs in the catchment area (Figure 18A,

section 3.6).

Mass fluxes per capita are presented in a PCA biplot, a combined loading and score

plot (Figure 18B). Correlation in the positive direction of variables in one PC means

that these variables have a high positive impact on the observations located in the

same PC direction. In Figure 18B, March samples and most STP samples are located

in the positive direction along PC1, because of higher contaminant fluxes than the

samples located in PC2 i.e. June and September samples. Thus, OSSF-affected sites

deviated more from STP sites in June and September than in December and March.

March samples were impacted by an ongoing spring flood [178,179] and levels could

therefore be affected by both dilution and increased runoff (section 4.4).

Of the compounds with the highest detection frequency (> 33%), median mass

fluxes per capita were similar at OSSF- and STP-impacted sites for TDCPP, TCIPP,

TCEP and nBBSA. However, mass fluxes per capita of TBEP, MTBT and galaxolide

were significantly lower (~2 to 3-fold) at OSSF sites than at STP sites

(Mann-Whitney, α = 0.05).

Calculating mass fluxes downstream of OSSFs is challenging, since OSSFs are

diffuse sources and the distance between sampling site and source can influence

the measured levels. Hence, enhanced biodegradation and photodegradation in

June and September as compared to December could affect contaminant levels

49

more at OSSF sites. Therefore, mass fluxes could have been lower at OSSF sites not

due to a lower environmental load, but due to “treatment” along the way to the

recipient and sampling site. Thus, contaminants identified as persistent and

mobile in Paper II – Galaxolide, nBBSA, TCEP, TCIPP, TDCPP and TMDD – should

be less affected by these fate processes than compounds which were less mobile

and persistent – triclosan, fluoranthene, TBEP and MTBT. Thus, the significantly

lower mass fluxes per capita of MTBT and TBEP at OSSF-affected sites could

originate from a higher impact of fate processes on OSSF sites. Alternatively, these

compounds could have been removed by hydrophobicity-driven adsorption

(log KOW = 3.0 to 6.3) in OSSFs due to a higher solid-to-liquid ratio and long

retention time in, for instance SBs, whereas the relatively hydrophilic

contaminants were removed in OSSFs and STPs to a similar degree. In addition,

differences could have originated from a higher use and emission of these

compounds in urbanized areas with a higher percentage of non-residential

buildings, which could have affected the differences: MTBT due to tire abrasion,

TBEP from floor polish and galaxolide from cleaning products.

Figure 18. (A) Schematic overview of sampling sites and (B) Principal Component Analysis biplot of mass

fluxes per capita at OSSF-affected and STP-affected sites (observations) of contaminants (variables) with

detection frequency ≥ 33%. Ellipses represent 50%, 75%, and 100% correlation. Samples were collected at

five sites (OSSF A, B, C and STP A and B) and collected in four seasons (December (D), March (M), June (J)

and September (S), n = 20). Mass flows for < limit of detection (LOD), < limit of quantification (LOQ),

and < method limit of quantification (MQL) were calculated based on LOD/2, LOQ/2, and MQL/2,

respectively.

TMDD

MTBT

TCEP

TCIPP

nBBSA

HHCB

FLA

TCSTDCPP

TBEP

OSSF A.D

OSSF A.M

OSSF A.J

OSSF A.S

STP A.D

STP A.M

STP A.J

STP A.SOSSF B.D

OSSF B.M

OSSF B.J

OSSF B.S

OSSF C.D

OSSF C.M

OSSF C.J

OSSF C.S

STP B.D

STP B.MSTP B.J

STP B.S

PC

2 (

16

%)

-1

-0.5

0

0.5

-1 -0.5 0 0.5

PC 1 (45%)

Contaminant

OSSF

STP

Kungsängsverket

STP 171 000 pers

River Sävja

River Fyris

Björklinge STP

3700 pers

OSSF A660 pers

STP A5500 pers

OSSF B560 pers

STP B198 000 pers OSSF C

10400 pers

Lake Ekoln

River Björklinge Storvreta STP

6400 pers

Lake Mälaren

Vattholma STP

1500 pers

Skyttorp STP600 pers

Gåvsta STP

690 pers

Länna STP1300 pers

A B

50

Although the calculation of mass fluxes per capita is affected by uncertainties from

the analysis, water flow and estimated numbers of people using OSSFs in the

catchment, the estimations suggest that the environmental loads of OSSFs are

similar or weaker compared to STPs. There are several factors that could explain

such differences, including differing patterns of product and material usage

between urban and rural areas, as well as differing treatment technologies and

removal mechanisms between STPs and OSSFs. Further studies of the impact of

OSSFs on the environment are necessary by studying contaminants in recipients

with an exclusive inflow of OSSF effluent, so that results could be solely attributed

to OSSFs. In these studies, different distances between OSSFs and recipient could

also be investigated. Water levels, bioaccumulation in biota and the impact of

seasonal factors such as temperature, irradiation and nutrient availability could be

further investigated.

4.7. Removal in char-fortified soil beds

For Paper IV, the removal of organic contaminants from OSSF effluent using a

stand-alone sand filter, commonly used in SBs, was investigated in a long-term

(two-year) field setting and compared to removal using a char-fortified sand filter

and a char-fortified gas concrete (Sorbulite®) filter. Biochar was investigated as it

may enhance the removal of (semi-)polar contaminants, whilst gas concrete was

included since it is currently used to upgrade OSSFs for the removal of phosphates

[181] and had already been installed at the study site. The removal of compounds

identified by targeted and untargeted analysis with GC×GC-ToF-MS, GC-ToF-MS

and LC-ESI-IM-QToF-MS was thoroughly investigated with QSPR modeling, to

better understand the relationship between molecular properties and compound

removal by biochar.

The comprehensive screening of OSSF samples with three different instruments

resulted in 24 compounds confirmed by reference standards and 47 tentatively

identified compounds. Amongst these were 12 LC compounds on level 1

identification confidence (”confirmed structure”) and five LC compounds on

level 2 identification confidence (“probable structure” confirmed with MS/MS

library) [141] (see section 2.4.2). Twelve target GC-compounds were found that

overlapped with four GC compounds found by library search. A further

47 GC compounds remained tentatively identified with eight GC compounds

confirmed with low energy EI GC-ToF-MS. Two compounds were found using

GC-MS and LC-MS. Amongst the 74 compounds found in total were plasticizers,

UV stabilizers, fragrances, pesticides, surfactant and polymer impurities,

pharmaceuticals and their metabolites and many biogenic compounds.

51

Char-fortified sand filters removed these contaminants the best, followed by sand,

while gas concrete fortified with biochar performed worst (Figure 19). The

difference between the three treatments was significant according to the Friedman

test (α = 0.05) and the treatment efficiencies were significantly higher in

char-fortified sand compared to sand according to the Wilcoxon signed-rank test

(α = 0.05).

Figure 19. Boxplot showing the removal efficiencies of organic contaminants in char-concrete, sand, and

char-sand beds. The boxes represent the 25th and 75th percentiles, the whiskers indicate the 10th and 90th

percentiles, the horizontal lines indicate the median and the crosses represents the mean. Data points

representing the individual removal efficiencies are indicated as dots next to each boxplot (Paper IV).

QSPR modeling showed that sand filtration commonly used in soil bed/ infiltration

OSSFs can remove contaminants that are hydrophobic, have few heteroatoms and

few hydrogen bond donors and acceptors. Char fortification enhances the

treatment efficiencies of such contaminants (Figure 20) and a few hydrophilic

compounds. However, most of the very polar and hydrophilic were also not

completely removed by char-fortified sand, thus additional treatment methods are

needed to remove these compounds.

The difference in removal of individual compounds spanned from -7% to 49%, and

the small (4%), but significant increase in average removal efficiencies in char-

fortified sand, compared to sand, indicates that the difference is attributed to

sorption to biochar. However, microorganisms that colonized the filter beds after

a two-year run in a field setting and formed biofilm, are likely responsible for the

overall removal through biodegradation or sorption in the two filter beds. The

char-concrete sand char-sand

0

10

20

30

40

50

60

70

80

90

100

Rem

ova

l e

ffic

iencie

s (

%)

52

worst removal efficiency was by char-enhanced gas concrete, showing that it was

a poor medium for biofilm formation or that the combination of two coarse

materials reduced the hydraulic retention time to an extent, resulting in reduced

biodegradation and sorption (section 3.1.3,Table 4). Most studies on wastewater

treatment with biochar were carried out in short-term laboratory settings and

therefore discuss sorption as the major removal pathway [182] and ignore

biodegradation. In our study, we show that organic contaminants are still

significantly better removed with char-fortified sand, as compared with sand only,

after a two-year run period. In summary, this study suggests that biodegradation

plays a key role in the overall removal of organic contaminants and biofilm

formation and long hydraulic retention times should be promoted as well as

further investigated. The reduction of the discharge of very polar and hydrophilic

contaminants requires further research on alternative treatment methods, such as

ozonation.

Figure 20. Properties negatively and positively correlated to removal in char-fortified filter beds.

Although a large variety of contaminants with different polarities were covered by

the comprehensive analysis with GC-MS and LC-MS, differences in removal could

be further examined by extending the contaminant range with experiments in

ESI(-) mode to cover, for instance, carboxylic acids and alcohols that preferentially

form negative ions. Additionally, many unidentified features that might be

environmentally-relevant, were not further investigated due to the sole

identification via library searches. Features remained unidentified due to a lack of

library entries (NIST and vendor libraries) and the challenging workflows to

identify unknowns through non-target screening in LC-HRMS and unknown

analysis in GC-MS.

Hydrophobicity

van der Waals surface area (partial positive charge)

Negatively charged

Positively charged

van der Waals surface area (partial negative charge)

Heteroatoms

Polarizability

Polarity

Positive

correlation

Negative

correlation

53

5. Conclusions and future perspectives

Environmentally-relevant contaminants were found (Paper I) by untargeted

contaminant analysis with GC×GC-MS in combination with filtering and

prioritization based on estimated ecotoxicological risks, bioconcentration and

persistence, as well as the determination of removal efficiencies and semi-

quantified concentrations in OSSFs. A method was successfully developed for

these target contaminants and was also applicable to untargeted analysis. The

method was based on solid phase extraction and GC×GC-MS analysis in

wastewater and surface water and was used for subsequent studies (Paper II,

Paper III and Paper IV).

The prioritization strategy for untargeted GC×GC-MS data was successful in

identifying environmentally-relevant compounds discharged from OSSFs, since

several of the prioritized untargeted contaminants were found in high

concentrations in the effluent samples from the second OSSF sampling campaign

(Paper I). Some of these contaminants were also identified as being persistent,

mobile and bioavailable in a field study (Paper II). These contaminants were

galaxolide, TMDD, TCIPP, TCEP and TDCPP. Galaxolide and TMDD were also

found to be bioaccumulative.

Mobility, persistence, bioavailability and especially bioaccumulation could be

studied in other recipients or on a laboratory-scale to confirm the results of this

assessment. For instance, the field-based studies to investigate persistence could

be complemented with laboratory-scale studies of photolysis and biodegradation

to investigate whether these fate processes play a key role in the removal of the

studied contaminants from surface water in brighter and warmer months. Thus,

the impact of seasonal factors such as temperature, irradiation and nutrient

availability could be further investigated. Additionally, it would be simpler to

compare studies, if as a first step, harmonization discussions would lead to a

general agreement on persistence and mobility quantifiers in the field and as a

second step, better quantifiers would be found through further research.

The analysis method for fish did not cover all target analytes due to lipid

interferences and needs to be improved to cover a wider range of analytes. An

improved and extended method to remove lipids from fish extracts could be

valuable to further investigate bioaccumulation of untargeted wastewater

54

contaminants in biota. Accumulated levels in biota are not only a threat to the

ecosystem, but also to the health of fish consumers.

SBs, which are a type of OSSF, removed organic contaminants with a similar

effectiveness to STPs, with the exceptions of TBEP and TMDD for which SBs

worked significantly better. Yet, the removal of in particular more hydrophilic

contaminants was inadequate by both STPs and SBs (Paper I). For instance, the

median removal efficiencies of the chlorinated organophosphates TCEP, TCIPP

and TDCPP were < 56% in both SBs and STPs and their effluent concentrations

were in the nanogram per liter range. The poorer removal of hydrophilic

contaminants by both OSSF and STPs (Paper I), that were later shown to be

mobile, persistent and bioavailable (Paper II), implies that an upgrade of the

treatment facilities is necessary to reduce discharge of organic contaminants into

the environment and to avoid adverse ecological and human health risks.

Despite the similar concentrations and removal efficiencies noted in the study of

paired OSSF and STP influent and effluent samples (Paper I), the surface water

concentrations measured in river water seem to be somewhat lower at OSSF- than

STP-affected sites (Paper II and Paper III). The associated potential

ecotoxicological risks (MEC/PNEC) were higher at sites receiving primarily STP

effluents as compared with sites receiving primarily OSSF effluents (Galaxolide,

TBEP and TCIPP) and similar for octocrylene at OSSF and STP sites. No elevated

risks at the studied sites were found for the other studied contaminants. The

differences between OSSFs and STPs in terms of mass fluxes per capita could be

associated with deviations in user patterns between urban and rural areas as well

as deviations in treatment technologies and removal mechanisms. However, the

pronounced differences in June and September compared to December indicate

that fate processes, such as biodegradation and photolysis, may have influenced

the measured environmental concentrations. OSSFs are diffuse sources and there

were relatively long average distances between the OSSFs and the sampling sites,

whereas STPs are point sources with release points closer to the sampling sites.

Large uncertainties in water flow (up to 36%), estimated people (22%) using OSSFs

in each catchment, and the 95% confidence intervals from the recovery tests used

for the analytical uncertainty that become rather large due to the small sample size

(n = 3), all result in large overall uncertainties of the mass flux per capita

calculations and make interpretation more difficult. Additionally, none of these

sampling sites had direct and sole inflow of OSSF effluents or STPs, thus risks and

loads cannot solely be attributed to either OSSFs or STPs. Hence, further studies

on the impact of OSSFs on the environment are necessary by studying

55

contaminants in recipients mainly receiving OSSF effluent, allowing results to be

more certainly attributed to OSSFs.

OSSF soil beds can be fortified with biochar to enhance the removal of

hydrophobic contaminants and partly to enhance the removal of several

hydrophilic contaminants with heteroatoms (Paper IV). Removal in sand and

char-fortified sand was not solely associated to sorption, but also to

biodegradation. In fact, biodegradation seems to play a significant role after two

years of running in a field-based system (Paper IV). However, further research on

treatment technologies to remove all types of organic contaminants is necessary as

some of these contaminants were neither removed by sand filters nor with a

biochar-fortified sand filter. For this purpose, other sorbent types, such as

granulated activated carbon filtration or technical solutions e.g. ozonation, should

be tested in an OSSF field-scale setting. Such a study, investigating different

sorbents in a field study, is currently being carried out and is part of the same

project as this thesis (“RedMic”).

Untargeted analysis by LC-HRMS has already well accepted nomenclature [129]

and well-defined confidence levels [141] for the identification of unknowns. In

comparison, a nomenclature for untargeted analysis by GC-MS remains to be

defined and a similar definition for confidence levels for GC-based techniques is

missing. Until consensus has been reached regarding the proper concepts and

terms in GC-MS, similar confidence levels as for LC-HRMS could be used. For

instance, GC-EI peaks with good similarity and probability to NIST library entries

or computer-aided “in silico” spectra, which give a “likely candidate”, are similar to

level 2a or 2b candidates in LC-HRMS (“probable structure” [141]), respectively. If

the spectral library search or interpretation produce several candidates of

sufficient quality, it gives a level 3 identification confidence, “tentative candidates”

using the LC-HRMS nomenclature [141]. The additional use of low energy “soft” EI

or CI could increase the identification confidence for some candidates and

consensus structure elucidation could be used for candidate ranking to find the

most likely candidate [149].

Generally, the combination of targeted and untargeted analysis throughout

Paper I, Paper II and Paper IV has been proven valuable to investigate a large

variety of organic contaminants. In addition, the PCA used for Paper I and

Paper III helped to elucidate patterns in removal and mass fluxes per capita,

respectively, and the PLS used for Paper IV visualized the strength of latent

variables instead of single molecular properties to correlate properties and removal

efficiencies. The results from Paper I, Paper II, and Paper III implied that OSSFs

56

have similar or better removal efficiencies and have similar or lower environmental

impact regarding environmental risks and mass fluxes per capita. Biochar

fortification can improve the removal of organic contaminants in SBs, however,

further research on alternative technologies to remove all types of organic

contaminants is needed.

57

6. Acknowledgements

Thanks to my supervisor Peter Haglund for giving me the opportunity to do my

PhD in his lab. Thanks for all the interesting discussions, great scientific advice

and access to a unique selection of instruments. Thanks to my co-supervisor

Patrik Andersson for great discussions, for your support along the way, and

always keeping me on track. Thanks, that I could be part of the project RedMic and

had the opportunity to collaborate with Karin Wiberg, Lutz Ahrens, Gunno

Renman, Meri Gros, Wen Zhang, Qiuju Gao, Pablo Gago Ferrero, Berndt

Björlenius, Jerker Fick, Misse Wester, Erik Kärrman, Henrik Jernstedt,

Rikard Tröger, and Linnéa Tjernström. I also want to thank Stina Jansson and

Erik Björn for advice during my annual evaluations.

Thanks to all my current and former colleagues from Miljökemi for your support,

fruitful discussions, and company during fikas. Thanks to my fellow PhD council

members for interesting discussions and having a lot of fun during our organized

events. I would like to kindly thank Mátyás Ripszám for his technical support,

teaching me a lot about the HRT, and for a lot of fun at the climbing wall. Special

thanks goes to João Figueira for writing data processing scripts for me and for

being the best climbing partner and fellow dwarven companion. I want to

especially thank Christine Gallampois for many scientific discussions,

proofreading my thesis, and having you as a friend. To Mirva, who was one of my

first friends in Umeå and has ever since been a great support along the way. Thanks

also for reading parts of my thesis. Thanks to Sandra for helping me out as a

neighbor, proofreading my thesis, and for being an awesome friend. Thanks to Mar

for all the good times, for proofreading parts of my thesis, and for all the

dissertation tips and tricks. Qiuju, you are awesome, and it was great to work with

you in RedMic. Thanks Darya for your openness and friendship. Ivan, continue to

have a nice life ;). Thanks to Elin Näsström for introducing me to Guineu and

thanks to Ziye Zheng for help with MOE.

Thanks to Anna for our long-lasting friendship, our refreshing weekly lunches, and

being a patient Swedish teacher. Thanks to Dagmar for hanging out with me at

IKSU, but most of all for always smiling and passing on your good mood. Thanks

to Henke for a lot of fun during parties, dinners and pathfinder sessions.

58

Thanks to all the friends who have been a large part of the Umeå experience but

have unfortunately moved away, foremost Palle, Lina, Casper, Jon, Tobi, Moritz,

Hanna, Herja, and Warin. Thanks to Steffi, you have been a great support during

our Bachelor and thanks for all the great hiking adventures we could experience

together. Emmy, Nina, Bela, and Lisa, I am glad to have you as friends. You have

always been there for me, even with thousands of kilometers between us.

To Mama, Papa, Anne, Norman, Gisela, and the rest of my family for your

support and love.

Finally, thanks to Artur for going on this Sweden journey together with me,

enduring me during all this time, being a great support, and giving me your love.

59

7. References

[1] M. Petrovic, Analysis and removal of emerging contaminants in wastewater and drinking water, TrAC Trends Anal. Chem. 22 (2003) 685–696. doi:10.1016/S0165-9936(03)01105-1.

[2] R.P. Schwarzenbach, B.I. Escher, K. Fenner, H. B, C.A. Johnson, U. von Gunten, et al., The Challenge of Micropollutants in Aquatic Systems, Science (80-. ). 313 (2006) 1072–1077. doi:doi: 10.1126/science.1127291.

[3] E. Corcoran, C. Nellemann, R.B. E. Baker, D. Osborn, H. Savelli, Sick Water? The Central Role of Wastewater Management in Sustainable Development, 2010. doi:10.1007/s10230-011-0140-x.

[4] SCB Statistics Sweden, Befolkningen efter region, avloppsanslutning och vart 5:e år, (2014). http://www.statistikdatabasen.scb.se/pxweb/sv/ssd/START__MI__MI0902__MI0902C/MI0902T01/table/tableViewLayout1/?rxid=b2ef512e-30ba-41d4-86b0-8e2a64a95ec0 (accessed February 13, 2017).

[5] B. Du, A.E. Price, W.C. Scott, L.A. Kristofco, A.J. Ramirez, C.K. Chambliss, et al., Comparison of contaminants of emerging concern removal, discharge, and water quality hazards among centralized and on-site wastewater treatment system effluents receiving common wastewater influent, Sci. Total Environ. 466–467 (2014) 976–984. doi:10.1016/j.scitotenv.2013.07.126.

[6] Bundesverband der Energie- und Wasserwirtschaft e. V, Abwasserdaten Deutschland Zahlen und Fakten der Abwasserentsorgung, (2010) 15.

[7] U.S. EPA, Septic Systems Fact Sheet, 2008.

[8] SCB Statistics Sweden, Befolkningens anslutning till kommunalt vatten och avlopp. Vart femte år 1960 - 2014, (2014). http://www.statistikdatabasen.scb.se/pxweb/sv/ssd/START__MI__MI0902__MI0902C/MI0902T05/table/tableViewLayout1/?rxid=b2ef512e-30ba-41d4-86b0-8e2a64a95ec0 (accessed January 16, 2017).

[9] M. Olshammar, A. Sidvall, S. Svanström, M. Ek, L. Rosenquist, H. Ejhed, SMED rapport 2015: Uppdatering av kunskapsläget och statistik för små avloppsanläggningar, 2015.

60

[10] L.A. Schaider, K.M. Rodgers, R.A. Rudel, Review of Organic Wastewater Compound Concentrations and Removal in Onsite Wastewater Treatment Systems, Environ. Sci. Technol. 51 (2017) 7304–7317. doi:10.1021/acs.est.6b04778.

[11] U.S. EPA, Decentralized Systems Technology Fact Sheet Septic System Tank, 2000.

[12] J.E. McCray, S.L. Kirkland, R.L. Siegrist, G.D. Thyne, Model parameters for simulating fate and transport of on-site wastewater nutrients., Ground Water. 43 (2005) 628–39. doi:10.1111/j.1745-6584.2005.0077.x.

[13] W. Robertson, S. Schiff, C. Ptacek, Review of phosphate mobility and persistence in 10 septic system plumes, Ground Water. 36 (1998).

[14] U.S. EPA, Decentralized Systems Technology Fact Sheet Septic Tank - Soil Absorption Systems, 2000.

[15] U.S. EPA, Wastewater Technology Fact Sheet Package Plants, 2000. doi:EPA 832-F-99-062.

[16] U.S. EPA, Decentralized Systems Technology Fact Sheet Aerobic Treatment, 2000.

[17] A. Renman, On-site wastewater treatment - Polonite and other filter materials for removal of metals nitrogen and phosphorus, 2008.

[18] R.L. Oulton, T. Kohn, D.M. Cwiertny, Pharmaceuticals and personal care products in effluent matrices: A survey of transformation and removal during wastewater treatment and implications for wastewater management, J. Environ. Monit. 12 (2010) 1956. doi:10.1039/c0em00068j.

[19] P. Verlicchi, M. Al Aukidy, E. Zambello, Occurrence of pharmaceutical compounds in urban wastewater: removal, mass load and environmental risk after a secondary treatment--a review, Sci. Total Environ. 429 (2012) 123–55. doi:10.1016/j.scitotenv.2012.04.028.

[20] A. Joss, S. Zabczynski, A. Göbel, B. Hoffmann, D. Löffler, C.S. McArdell, et al., Biological degradation of pharmaceuticals in municipal wastewater treatment: Proposing a classification scheme, Water Res. 40 (2006) 1686–1696. doi:10.1016/j.watres.2006.02.014.

[21] K.E. Conn, R.O.L.S. Siegrist, L.B. Barber, M.T. Meyer, Fate of trace organic compounds during vadose zone soil treatment in an onsite wastewater system, Environ. Toxicol. Chem. 29 (2010) 285–293. doi:10.1002/etc.40.

61

[22] A. Vailaya, C. Horváth, Retention in reversed-phase chromatography: Partition or adsorption?, J. Chromatogr. A. 829 (1998) 1–27. doi:10.1016/S0021-9673(98)00727-4.

[23] H.C. Tülp, K. Fenner, R.P. Schwarzenbach, K.U. Goss, pH-dependent sorption of acidic organic chemicals to soil organic matter, Environ. Sci. Technol. 43 (2009) 9189–9195. doi:10.1021/es902272j.

[24] Y.-H. Li, H.-B. Li, T.-H. Sun, X. Wang, Effects of hydraulic loading rate on pollutants removal by a deep subsurface wastewater infiltration system, Ecol. Eng. 37 (2011) 1425–1429. doi:10.1016/j.ecoleng.2011.03.029.

[25] K. Thompson, J. Zhang, C. Zhang, Use of fugacity model to analyze temperature-dependent removal of micro-contaminants in sewage treatment plants, Chemosphere. 84 (2011) 1066–1071. doi:10.1016/j.chemosphere.2011.04.063.

[26] G. Byrns, The fate of xenobiotic organic compounds in wastewater treatment plants, Water Res. 35 (2001) 2523–2533. doi:10.1016/S0043-1354(00)00529-7.

[27] S.N. Garcia, R.L. Clubbs, J.K. Stanley, B. Scheffe, J.C. Yelderman, B.W. Brooks, Comparative analysis of effluent water quality from a municipal treatment plant and two on-site wastewater treatment systems., Chemosphere. 92 (2013) 38–44. doi:10.1016/j.chemosphere.2013.03.007.

[28] K.E. Conn, R.L. Siegrist, Occurrence and Fate of Trace Organic Contaminants in Onsite Wastewater Treatment Systems and Implications for Water Quality Management, Fort Collins, 2009.

[29] C. Carrara, C.J. Ptacek, W.D. Robertson, D.W. Blowes, M.C. Moncur, E. Sverko, et al., Fate of pharmaceutical and trace organic compounds in three septic system plumes, Ontario, Canada., Environ. Sci. Technol. 42 (2008) 2805–11. http://www.ncbi.nlm.nih.gov/pubmed/18497127.

[30] C.H. Swartz, S. Reddy, M.J. Benotti, H. Yin, L.B. Barber, B.J. Brownawell, et al., Steroid estrogens, nonylphenol ethoxylate metabolites, and other wastewater contaminants in groundwater affected by a residential septic system on Cape Cod, MA, Environ. Sci. Technol. 40 (2006) 4894–4902. doi:10.1021/es052595+.

[31] C.J. van Leeuwen, T.G. Vermeire, Risk Assessment of Chemicals: An Introduction, 2nd ed., Springer, Dordrecht, The Netherlands, 2007.

[32] H.R. Rogers, Sources, behaviour and fate of organic contaminants during

62

sewage treatment and in sewage sludges., Sci. Total Environ. 185 (1996) 3–26. doi:10.1016/0048-9697(96)05039-5.

[33] R.S. Horvath, Microbial co-metabolism and the degradation of organic compounds in nature., Bacteriol. Rev. 36 (1972) 146–155.

[34] E. Eriksson, K. Auffarth, A. Eilersen, M. Henze, A. Ledin, Household chemicals and personal care products as sources for xenobiotic organic compounds in grey wastewater, Water SA. 29 (2003) 135–146.

[35] L.H. Leal, N. Vieno, H. Temmink, G. Zeeman, C.J.N. Buisman, Occurrence of xenobiotics in gray water and removal in three biological treatment systems, Environ. Sci. Technol. 44 (2010) 6835–42. doi:10.1021/es101509e.

[36] B. Subedi, N. Codru, D.M. Dziewulski, L.R. Wilson, J. Xue, S. Yun, et al., A pilot study on the assessment of trace organic contaminants including pharmaceuticals and personal care products from on-site wastewater treatment systems along Skaneateles Lake in New York State, USA., Water Res. 72 (2014) 1–12. doi:10.1016/j.watres.2014.10.049.

[37] V. Matamoros, C. Arias, H. Brix, J.M. Bayona, Preliminary screening of small-scale domestic wastewater treatment systems for removal of pharmaceutical and personal care products, Water Res. 43 (2009) 55–62. doi:10.1016/j.watres.2008.10.005.

[38] K.E. Conn, K.S. Lowe, E. Drewes, C. Hoppe-jones, M.B. Tucholke, Occurrence of Pharmaceuticals and Consumer Product Chemicals in Raw Wastewater and Septic Tank Effluent from Single-Family Homes, Environ. Eng. Sci. 27 (2010). doi:10.1089/ees.2009.0364.

[39] M. Clara, B. Strenn, N. Kreuzinger, Carbamazepine as a possible anthropogenic marker in the aquatic environment: Investigations on the behaviour of Carbamazepine in wastewater treatment and during groundwater infiltration, Water Res. 38 (2004) 947–954. doi:10.1016/j.watres.2003.10.058.

[40] E. Godfrey, W.W. Woessner, M.J. Benotti, Pharmaceuticals in on-site sewage effluent and ground water, Western Montana, Ground Water. 45 (2007) 263–271. doi:10.1111/j.1745-6584.2006.00288.x.

[41] B.D. Stanford, H.S. Weinberg, Evaluation of on-site wastewater treatment technology to remove estrogens, nonylphenols, and estrogenic activity from wastewater, Environ. Sci. Technol. 44 (2010) 2994–3001. doi:10.1021/es903422b.

63

[42] W.D. Robertson, D.R. Van Stempvoort, D.K. Solomon, J. Homewood, S.J. Brown, J. Spoelstra, et al., Persistence of artificial sweeteners in a 15-year-old septic system plume, J. Hydrol. 477 (2013) 43–54. doi:10.1016/j.jhydrol.2012.10.048.

[43] D.R. Van Stempvoort, W.D. Robertson, S.J. Brown, Water level monitoring pressure transducers : a need for industry-wide standards, Ground Water Monit. Remediat. 31 (2011) 82–94. doi:10.1111/j1745.

[44] B.E. Huntsman, C.A. Staples, C.G. Naylor, J.-B. Williams, Treatability of Nonylphenol Ethoxylate Surfactants in On-Site Wastewater Disposal Systems, Water Environ. Res. 78 (2006). doi:https://doi.org/10.2175/106143005X72966.

[45] J. Teerlink, A.S. Hering, C.P. Higgins, J.E. Drewes, Variability of trace organic chemical concentrations in raw wastewater at three distinct sewershed scales, Water Res. 46 (2012) 3261–3271. doi:10.1016/j.watres.2012.03.018.

[46] S. Roberts, C. Higgins, J. McCray, Sorption of emerging organic wastewater contaminants to four soils, Water (Switzerland). 6 (2014) 1028–1042. doi:10.3390/w6041028.

[47] B. Blair, A. Nikolaus, C. Hedman, R. Klaper, T. Grundl, Evaluating the degradation, sorption, and negative mass balances of pharmaceuticals and personal care products during wastewater treatment, Chemosphere. 134 (2015) 395–401. doi:10.1016/j.chemosphere.2015.04.078.

[48] R. Salgado, R. Marques, J.P. Noronha, G. Carvalho, A. Oehmen, M.A.M. Reis, Assessing the removal of pharmaceuticals and personal care products in a full-scale activated sludge plant, Environ. Sci. Pollut. Res. 19 (2012) 1818–1827. doi:10.1007/s11356-011-0693-z.

[49] A. Rodayan, M. Majewsky, V. Yargeau, Impact of approach used to determine removal levels of drugs of abuse during wastewater treatment, Sci. Total Environ. 487 (2014) 731–739. doi:10.1016/j.scitotenv.2014.03.080.

[50] A. Göbel, C.S. McArdell, A. Joss, H. Siegrist, W. Giger, Fate of sulfonamides, macrolides, and trimethoprim in different wastewater treatment technologies, Sci. Total Environ. 372 (2007) 361–371. doi:10.1016/j.scitotenv.2006.07.039.

[51] S. Brick, Biochar: Assessing the Promise and Risks To Guide US Policy, Madision, Wisconsin, USA, 2010. http://arxiv.org/abs/1112.1077%5Cnhttp://devstaging.win.nrdcdev.org/ener

64

gy/files/biochar_paper.pdf.

[52] M. Ahmad, A.U. Rajapaksha, J.E. Lim, M. Zhang, N. Bolan, D. Mohan, et al., Biochar as a sorbent for contaminant management in soil and water: A review, Chemosphere. 99 (2014) 19–23. doi:10.1016/j.chemosphere.2013.10.071.

[53] M.J.C. van der Stelt, H. Gerhauser, J.H.A. Kiel, K.J. Ptasinski, Biomass upgrading by torrefaction for the production of biofuels: A review, Biomass and Bioenergy. 35 (2011) 3748–3762. doi:10.1016/j.biombioe.2011.06.023.

[54] D. Mohan, A. Sarswat, Y.S. Ok, C.U. Pittman, Organic and inorganic contaminants removal from water with biochar, a renewable, low cost and sustainable adsorbent - A critical review, Bioresour. Technol. 160 (2014) 191–202. doi:10.1016/j.biortech.2014.01.120.

[55] J.A. Libra, K.S. Ro, C. Kammann, A. Funke, N.D. Berge, Y. Neubauer, et al., Hydrothermal carbonization of biomass residuals: a comparative review of the chemistry, processes and applications of wet and dry pyrolysis, Biofuels. 2 (2011) 71–106. doi:10.4155/bfs.10.81.

[56] A. Demirbaş, G. Arin, An overview of biomass pyrolysis, Energy Sources. 24 (2002) 471–482. doi:10.1080/00908310252889979.

[57] B. Chen, D. Zhou, L. Zhu, Transitional adsorption and partition of nonpolar and polar aromatic contaminants by biochars of pine needles with different pyrolytic temperatures, Environ. Sci. Technol. 42 (2008) 5137–5143. doi:10.1021/es8002684.

[58] M. Uchimiya, L.H. Wartelle, I.M. Lima, K.T. Klasson, Sorption of deisopropylatrazine on broiler litter biochars, J. Agric. Food Chem. 58 (2010) 12350–12356. doi:10.1021/jf102152q.

[59] Y. Qiu, Z. Zheng, Z. Zhou, G.D. Sheng, Effectiveness and mechanisms of dye adsorption on a straw-based biochar, Bioresour. Technol. 100 (2009) 5348–5351. doi:10.1016/j.biortech.2009.05.054.

[60] P.E. Stackelberg, E.T. Furlong, M.T. Meyer, S.D. Zaugg, A.K. Henderson, D.B. Reissman, Persistence of pharmaceutical compounds and other organic wastewater contaminants in a conventional drinking-water-treatment plant., Sci. Total Environ. 329 (2004) 99–113. doi:10.1016/j.scitotenv.2004.03.015.

[61] T.A. Ternes, M. Meisenheimer, D. McDowell, F. Sacher, H.J. Brauch, B.

65

Haist-Gulde, et al., Removal of Pharmaceuticals during Drinking Water Treatment, Environ. Sci. Technol. 36 (2002) 3855–3863. doi:10.1021/es015757k.

[62] M. Ek, C. Baresel, J. Magnér, R. Bergström, M. Harding, Activated carbon for the removal of pharmaceutical residues from treated wastewater, Water Sci. Technol. 69 (2014) 2372–2380. doi:https://doi.org/10.2166/wst.2014.172.

[63] D.P. Grover, J.L. Zhou, P.E. Frickers, J.W. Readman, Improved removal of estrogenic and pharmaceutical compounds in sewage effluent by full scale granular activated carbon: Impact on receiving river water, J. Hazard. Mater. 185 (2011) 1005–1011. doi:10.1016/j.jhazmat.2010.10.005.

[64] K.T. Klasson, C.A. Ledbetter, M. Uchimiya, I.M. Lima, Activated biochar removes 100 % dibromochloropropane from field well water, Environ. Chem. Lett. 11 (2013) 271–275. doi:10.1007/s10311-012-0398-7.

[65] M. Essandoh, D. Wolgemuth, C.U. Pittman, D. Mohan, T. Mlsna, Phenoxy herbicide removal from aqueous solutions using fast pyrolysis switchgrass biochar, Chemosphere. 174 (2017) 49–57. doi:10.1016/j.chemosphere.2017.01.105.

[66] L. Lin, W. Jiang, P. Xu, Comparative study on pharmaceuticals adsorption in reclaimed water desalination concentrate using biochar: Impact of salts and organic matter, Sci. Total Environ. 601–602 (2017) 857–864. doi:10.1016/j.scitotenv.2017.05.203.

[67] M.B. Ahmed, J.L. Zhou, H.H. Ngo, W. Guo, M.A.H. Johir, D. Belhaj, Competitive sorption affinity of sulfonamides and chloramphenicol antibiotics toward functionalized biochar for water and wastewater treatment, Bioresour. Technol. 238 (2017) 306–312. doi:10.1016/j.biortech.2017.04.042.

[68] C. Jung, J. Park, K.H. Lim, S. Park, J. Heo, N. Her, et al., Adsorption of selected endocrine disrupting compounds and pharmaceuticals on activated biochars, J. Hazard. Mater. 263 (2013) 702–710. doi:10.1016/j.jhazmat.2013.10.033.

[69] B.H. Hameed, M.I. El-Khaiary, Kinetics and equilibrium studies of malachite green adsorption on rice straw-derived char, J. Hazard. Mater. 153 (2008) 701–708. doi:10.1016/j.jhazmat.2007.09.019.

[70] R.P.R.P. Schwarzenbach, D.M. Imboden, P.M. Gschwend, D.M. Imboden, Environmental organic chemistry, 2nd ed., Wiley, Hoboken, 2003.

66

[71] S. Gaw, K. V Thomas, T.H. Hutchinson, Sources, impacts and trends of pharmaceuticals in the marine and coastal environment., Philos. Trans. R. Soc. Lond. B. Biol. Sci. 369 (2014) 20130572-. doi:10.1098/rstb.2013.0572.

[72] S.P. Singh, A. Azua, A. Chaudhary, S. Khan, K.L. Willett, P.R. Gardinali, Occurrence and distribution of steroids, hormones and selected pharmaceuticals in South Florida coastal environments, Ecotoxicology. 19 (2010) 338–350. doi:10.1007/s10646-009-0416-0.

[73] J.H. Writer, L.B. Barber, G.K. Brown, H.E. Taylor, R.L. Kiesling, M.L. Ferrey, et al., Anthropogenic tracers, endocrine disrupting chemicals, and endocrine disruption in Minnesota lakes, Sci. Total Environ. 409 (2010) 100–111. doi:10.1016/j.scitotenv.2010.07.018.

[74] B.H. Baker, D. Martinovic-Weigelt, M. Ferrey, L.B. Barber, J.H. Writer, D.O. Rosenberry, et al., Identifying Non-point Sources of Endocrine Active Compounds and Their Biological Impacts in Freshwater Lakes, Arch. Environ. Contam. Toxicol. 67 (2014) 374–388. doi:10.1007/s00244-014-0052-4.

[75] J.A. Dougherty, P.W. Swarzenski, R.S. Dinicola, M. Reinhard, Occurrence of Herbicides and Pharmaceutical and Personal Care Products in Surface Water and Groundwater around Liberty Bay, Puget Sound, Washington, J. Environ. Qual. 39 (2010) 1173. doi:10.2134/jeq2009.0189.

[76] L. Standley, R. Rudel, Wastewater-contaminated groundwater as a source of endogenous hormones and pharmaceuticals to surface water ecosystems, Environ. Toxicol. Chem. 27 (2008) 2457–2468. doi:10.1897/07-604.1.

[77] B.G. Katz, D.W. Griffin, P.B. McMahon, H.S. Harden, E. Wade, R.W. Hicks, et al., Fate of Effluent-Borne Contaminants beneath Septic Tank Drainfields Overlying a Karst Aquifer, J. Environ. Qual. 39 (2010) 1181. doi:10.2134/jeq2009.0244.

[78] S.R. Hinkle, R.J. Weick, J.M. Johnson, J.D. Cahill, S.G. Smith, B.J. Rich, Organic wastewater compounds, pharmaceuticals, and coliphage in ground water receiving discharge from onsite wastewater treatment systems near La Pine, Oregon: Occurrence and implications for transport, Usgs Sir 2005-5055. (2009) 98 p. http://pubs.usgs.gov/sir/2005/5055/.

[79] K.L. Del Rosario, S. Mitra, C.P. Humphrey, M.A. O’Driscoll, Detection of pharmaceuticals and other personal care products in groundwater beneath and adjacent to onsite wastewater treatment systems in a coastal plain shallow aquifer, Sci. Total Environ. 487 (2014) 216–223.

67

doi:10.1016/j.scitotenv.2014.03.135.

[80] P.J. Phillips, C. Schubert, D. Argue, I. Fisher, E.T. Furlong, W. Foreman, et al., Concentrations of hormones, pharmaceuticals and other micropollutants in groundwater affected by septic systems in New England and New York, Sci. Total Environ. 512–513 (2015) 43–54. doi:10.1016/j.scitotenv.2014.12.067.

[81] L.A. Schaider, R.A. Rudel, J.M. Ackerman, S.C. Dunagan, J.G. Brody, Pharmaceuticals, perfluorosurfactants, and other organic wastewater compounds in public drinking water wells in a shallow sand and gravel aquifer, Sci. Total Environ. 468–469 (2014) 384–393. doi:10.1016/j.scitotenv.2013.08.067.

[82] Y. Luo, W. Guo, H.H. Ngo, L.D. Nghiem, F.I. Hai, J. Zhang, et al., A review on the occurrence of micropollutants in the aquatic environment and their fate and removal during wastewater treatment, Sci. Total Environ. 473–474 (2014) 619–641. doi:10.1016/j.scitotenv.2013.12.065.

[83] S. Mompelat, B. Le Bot, O. Thomas, Occurrence and fate of pharmaceutical products and by-products, from resource to drinking water, Environ. Int. 35 (2009) 803–814. doi:10.1016/j.envint.2008.10.008.

[84] J.K. Challis, M.L. Hanson, K.J. Friesen, C.S. Wong, A critical assessment of the photodegradation of pharmaceuticals in aquatic environments: defining our current understanding and identifying knowledge gaps, Environ. Sci. Process. Impacts. 16 (2014) 672. doi:10.1039/c3em00615h.

[85] A.L. Boreen, W.A. Arnold, K. McNeill, Photodegradation of pharmaceuticals in the aquatic environment: A review, Aquat. Sci. 65 (2003) 320–341. doi:10.1007/s00027-003-0672-7.

[86] T. Gouin, D. Mackay, E. Webster, F. Wania, Screening chemicals for persistence in the environment, Environ. Sci. Technol. 34 (2000) 881–884. doi:10.1021/es991011z.

[87] European Commission, COMMISSION REGULATION (EU) No 253/2011 of of 15 March 2011 amending Regulation (EC) No 1907/2006 of the European Parliament and of the Council on the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) as regards Annex XIII, Off. J. Eur. Union. 69 (2011).

[88] M.S. McLachlan, H. Zou, T. Gouin, Using Benchmarking to Strengthen the Assessment of Persistence, Environ. Sci. Technol. 51 (2017) 4–11. doi:10.1021/acs.est.6b03786.

68

[89] A. Kowalczyk, T.J. Martin, O.R. Price, J.R. Snape, R.A. van Egmond, C.J. Finnegan, et al., Refinement of biodegradation tests methodologies and the proposed utility of new microbial ecology techniques, Ecotoxicol. Environ. Saf. 111 (2015) 9–22. doi:10.1016/j.ecoenv.2014.09.021.

[90] ECHA, Guidance on information requirements and chemical safety assessment Chapter R.11: PBT/vPvB assessment Version 3.0, 2017. doi:10.2823/128621.

[91] D. Mackay, E. Webster, Environmental persistence of chemicals., Environ. Sci. Pollut. Res. Int. 13 (2006) 43–49. doi:http://dx.doi.org/10.1065/espr2006.01.008.

[92] E. Webster, D. Mackay, F. Wania, Evaluating environmental persistence, Environ. Toxicol. Chem. 17 (1998) 2148–2158. doi:10.1002/etc.5620171104.

[93] H. Zou, M. Radke, A. Kierkegaard, M.S. McLachlan, Temporal Variation of Chemical Persistence in a Swedish Lake Assessed by Benchmarking, Environ. Sci. Technol. 49 (2015) 9881–9888. doi:10.1021/acs.est.5b01720.

[94] A. Daneshvar, J. Svanfelt, L. Kronberg, M. Prévost, G.A. Weyhenmeyer, Seasonal variations in the occurrence and fate of basic and neutral pharmaceuticals in a Swedish river-lake system, Chemosphere. 80 (2010) 301–9. doi:10.1016/j.chemosphere.2010.03.060.

[95] C. Tixier, H.P. Singer, S. Oellers, S.R. Müller, Occurrence and fate of carbamazepine, clofibric acid, diclofenac, ibuprofen, ketoprofen, and naproxen in surface waters, Environ. Sci. Technol. 37 (2003) 1061–1068. doi:10.1021/es025834r.

[96] H. Zou, M. Radke, A. Kierkegaard, M. Macleod, M.S. McLachlan, Using chemical benchmarking to determine the persistence of chemicals in a swedish lake, Environ. Sci. Technol. 49 (2015) 1646–1653. doi:10.1021/es505548k.

[97] U. Kunkel, M. Radke, Fate of pharmaceuticals in rivers: Deriving a benchmark dataset at favorable attenuation conditions, Water Res. 46 (2012) 5551–5565. doi:10.1016/j.watres.2012.07.033.

[98] U. Kunkel, M. Radke, Reactive tracer test to evaluate the fate of pharmaceuticals in rivers, Environ. Sci. Technol. 45 (2011) 6296–6302. doi:10.1021/es104320n.

[99] Z. Li, A. Sobek, M. Radke, Fate of Pharmaceuticals and Their Transformation Products in Four Small European Rivers Receiving

69

Treated Wastewater, Environ. Sci. Technol. 50 (2016) 5614–5621. doi:10.1021/acs.est.5b06327.

[100] A. Cachada, R. Pereira, E.F. da Silva, A.C. Duarte, The prediction of PAHs bioavailability in soils using chemical methods: State of the art and future challenges, Sci. Total Environ. 472 (2014) 463–480. doi:10.1016/j.scitotenv.2013.11.038.

[101] M.J. Riding, K.J. Doick, F.L. Martin, K.C. Jones, K.T. Semple, Chemical measures of bioavailability/bioaccessibility of PAHs in soil: Fundamentals to application, J. Hazard. Mater. 261 (2013) 687–700. doi:10.1016/j.jhazmat.2013.03.033.

[102] K.T. Semple, K.J. Doick, K.C. Jones, P. Burauel, A. Craven, H. Harms, Peer Reviewed: Defining Bioavailability and Bioaccessibility of Contaminated Soil and Sediment is Complicated, Environ. Sci. Technol. 38 (2004) 228A–231A. doi:10.1021/es040548w.

[103] F. Reichenberg, P. Mayer, Two Complementary Sides of Bioavailability: Accessibility and Chemical Activity of Organic Contaminants in Sediments and Soils, Environ. Toxicol. Chem. 25 (2006) 1239. doi:10.1897/05-458R.1.

[104] J.D. Petty, J.N. Huckins, D.A. Alvarez, Device for sequestration and concentration of polar organic chemicals from water., US Pat. 6,478,961 (2002) November 12, 2002.

[105] D.A. Alvarez, J.D. Petty, J.N. Huckins, T.L. Jones-Lepp, D.T. Getting, J.P. Goddard, et al., Development of a Passive, in Situ, Integrative Sampler for Hydrophilic Organic Contaminants in Aquatic Environments, 2004. doi:10.1897/03-603.

[106] C.L. Arthur, J. Pawliszyn, Solid Phase Microextraction with Thermal Desorption Using Fused Silica Optical Fibers, Anal. Chem. 62 (1990) 2145–2148. doi:10.1021/ac00218a019.

[107] J.J. Ortega-Calvo, J. Harmsen, J.R. Parsons, K.T. Semple, M.D. Aitken, C. Ajao, et al., From Bioavailability Science to Regulation of Organic Chemicals, Environ. Sci. Technol. 49 (2015) 10255–10264. doi:10.1021/acs.est.5b02412.

[108] R. van der Oost, J. Beyer, N.P.E. Vermeulen, Fish bioaccumulation and biomarkers in environmental risk assessment: a review, Enviironmental Toxicol. Pharmacol. 13 (2003) 57–149. doi:10.1016/S1382-6689(02)00126-6.

70

[109] J.A. Arnot, F.A. Gobas, A review of bioconcentration factor (BCF) and bioaccumulation factor (BAF) assessments for organic chemicals in aquatic organisms, Environ. Rev. 14 (2006) 257–297. doi:10.1139/a06-005.

[110] K. Borgå, K.A. Kidd, D.C.G. Muir, O. Berglund, J.M. Conder, F.A.P.C. Gobas, et al., Trophic magnification factors: Considerations of ecology, ecosystems, and study design, Integr. Environ. Assess. Manag. 8 (2012) 64–84. doi:10.1002/ieam.244.

[111] US Environmental Protection Agency, USEPA. 1976. Toxic substances control act, (1976).

[112] European Union. Regulation (EC) No 1907/2006 of the European Parliament and of the Council of 18 December 2006 concerning the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH)., Off. J. Eur. Communities. (n.d.) 3. doi:2004R0726 - v.7 of 05.06.2013.

[113] F. Kalberlah, J. Oltmanns, M. Schwarz, Guidance for the precautionary protection of raw water destined for drinking water extraction from contaminants regulated under REACH, 2014.

[114] T. Reemtsma, U. Berger, H.P.H. Arp, H. Gallard, T.P. Knepper, M. Neumann, et al., Mind the Gap: Persistent and Mobile Organic Compounds - Water Contaminants That Slip Through, Environ. Sci. Technol. 50 (2016) 10308–10315. doi:10.1021/acs.est.6b03338.

[115] European Commission Joint Research Centre, Technical guidance document on risk assessment in support of Commission Directive 93/67/ EEC on risk assessment for new notified substances and Commission Regulation (EC) No. 1488/94 on risk assessment for existing substances. Part II. EUR 20418 EN/2, Eur. Chem. Bur. Part II (2003) 7–179.

[116] Echa, Guidance on information requirements and chemical safety assessment. Chapter R.10: Characterisation of dose [concentration]-response for environment, Eur. Chem. Agency. (2008) 1–65.

[117] R. Díaz, M. Ibáñez, J. V. Sancho, F. Hernández, Target and non-target screening strategies for organic contaminants, residues and illicit substances in food, environmental and human biological samples by UHPLC-QTOF-MS, Anal. Methods. 4 (2012) 196. doi:10.1039/c1ay05385j.

[118] A. Agüera, M.J. Martínez Bueno, A.R. Fernández-Alba, New trends in the analytical determination of emerging contaminants and their transformation products in environmental waters., Environ. Sci. Pollut.

71

Res. Int. 20 (2013) 3496–515. doi:10.1007/s11356-013-1586-0.

[119] M. Krauss, H. Singer, J. Hollender, LC-high resolution MS in environmental analysis: from target screening to the identification of unknowns., Anal. Bioanal. Chem. 397 (2010) 943–51. doi:10.1007/s00216-010-3608-9.

[120] J.F. García-Reyes, M.D. Hernando, A. Molina-Díaz, A.R. Fernández-Alba, Comprehensive screening of target, non-target and unknown pesticides in food by LC-TOF-MS, TrAC Trends Anal. Chem. 26 (2007) 828–841. doi:10.1016/j.trac.2007.06.006.

[121] P.M. Bastos, P. Haglund, The use of comprehensive two-dimensional gas chromatography and structure–activity modeling for screening and preliminary risk assessment of organic contaminants in soil, sediment, and surface water, J. Soils Sediments. 12 (2012) 1079–1088. doi:10.1007/s11368-012-0533-x.

[122] J. Aceña, S. Stampachiacchiere, S. Pérez, D. Barceló, Advances in liquid chromatography–high-resolution mass spectrometry for quantitative and qualitative environmental analysis, Anal. Bioanal. Chem. 407 (2015) 6289–6299. doi:10.1007/s00216-015-8852-6.

[123] T. Portoles, E. Pitarch, F.J. Lopez, J. V Sancho, F. Hernández, Methodical approach for the use of GC-TOF MS for screening and confirmation of organic pollutants in, J. Mass Spectrom. 42 (2007) 1175–1185. doi:10.1002/jms.

[124] M.I. Cervera, T. Portolés, E. Pitarch, J. Beltrán, F. Hernández, Application of gas chromatography time-of-flight mass spectrometry for target and non-target analysis of pesticide residues in fruits and vegetables., J. Chromatogr. A. 1244 (2012) 168–77. doi:10.1016/j.chroma.2012.04.063.

[125] F. Hernandez, T. Portole, E. Pitarch, F.J. Lopez, Target and Nontarget Screening of Organic Micropollutants in Water by Solid-Phase Microextraction Combined with Gas Mass Spectrometry, Anal. Chem. 79 (2007) 9494–9504.

[126] E. Pitarch, T. Portolés, J.M. Marín, M. Ibáñez, F. Albarrán, F. Hernández, Analytical strategy based on the use of liquid chromatography and gas chromatography with triple-quadrupole and time-of-flight MS analyzers for investigating organic contaminants in wastewater., Anal. Bioanal. Chem. 397 (2010) 2763–76. doi:10.1007/s00216-010-3692-x.

[127] R. Serrano, J. Nácher-Mestre, T. Portolés, F. Amat, F. Hernández, Non-

72

target screening of organic contaminants in marine salts by gas chromatography coupled to high-resolution time-of-flight mass spectrometry., Talanta. 85 (2011) 877–84. doi:10.1016/j.talanta.2011.04.055.

[128] S. Hashimoto, Y. Zushi, A. Fushimi, Y. Takazawa, K. Tanabe, Y. Shibata, Selective extraction of halogenated compounds from data measured by comprehensive multidimensional gas chromatography/high resolution time-of-flight mass spectrometry for non-target analysis of environmental and biological samples., J. Chromatogr. A. 1282 (2013) 183–9. doi:10.1016/j.chroma.2013.01.052.

[129] E.L. Schymanski, H.P. Singer, J. Slobodnik, I.M. Ipolyi, P. Oswald, M. Krauss, et al., Non-target screening with high-resolution mass spectrometry: Critical review using a collaborative trial on water analysis, Anal. Bioanal. Chem. 407 (2015) 6237–6255. doi:10.1007/s00216-015-8681-7.

[130] M.D.M. Gómez-Ramos, A. Pérez-Parada, J.F. García-Reyes, A.R. Fernández-Alba, A. Agüera, Use of an accurate-mass database for the systematic identification of transformation products of organic contaminants in wastewater effluents., J. Chromatogr. A. 1218 (2011) 8002–12. doi:10.1016/j.chroma.2011.09.003.

[131] M. Petrovic, D. Barceló, Special feature : Application of liquid chromatography / quadrupole time-of-flight mass spectrometry ( LC-QqTOF-MS ), J. Mass Spectrom. 41 (2006) 1259–1267. doi:10.1002/jms.

[132] M.J. Gómez, M.M. Gómez-Ramos, O. Malato, M. Mezcua, a R. Férnandez-Alba, Rapid automated screening, identification and quantification of organic micro-contaminants and their main transformation products in wastewater and river waters using liquid chromatography-quadrupole-time-of-flight mass spectrometry with an accurate-mass , J. Chromatogr. A. 1217 (2010) 7038–54. doi:10.1016/j.chroma.2010.08.070.

[133] A. Müller, W. Schulz, W.K.L. Ruck, W.H. Weber, A new approach to data evaluation in the non-target screening of organic trace substances in water analysis., Chemosphere. 85 (2011) 1211–9. doi:10.1016/j.chemosphere.2011.07.009.

[134] S. Terzic, M. Ahel, Nontarget analysis of polar contaminants in freshwater sediments influenced by pharmaceutical industry using ultra-high-pressure liquid chromatography-quadrupole time-of-flight mass spectrometry., Environ. Pollut. 159 (2011) 557–66. doi:10.1016/j.envpol.2010.10.009.

[135] A. Hogenboom, J. van Leerdam, P. de Voogt, Accurate mass screening and

73

identification of emerging contaminants in environmental samples by liquid chromatography-hybrid linear ion trap Orbitrap mass spectrometry., J. Chromatogr. A. 1216 (2009) 510–9. doi:10.1016/j.chroma.2008.08.053.

[136] D.E. Helbling, J. Hollender, H.-P.E. Kohler, H. Singer, K. Fenner, High-Throughput Identification of Microbial Transformation Products of Organic Micropollutants, Environ. Sci. Technol. 44 (2010) 6621–6627.

[137] M. Bourgin, E. Bichon, J. Antignac, F. Monteau, G. Leroy, L. Barritaud, et al., Chemosphere Chlorination of bisphenol A : Non-targeted screening for the identification of transformation products and assessment of estrogenicity in generated water, Chemosphere. 93 (2013) 2814–2822. doi:10.1016/j.chemosphere.2013.09.080.

[138] C. Hug, N. Ulrich, T. Schulze, W. Brack, M. Krauss, Identification of novel micropollutants in wastewater by a combination of suspect and nontarget screening., Environ. Pollut. 184 (2014) 25–32. doi:10.1016/j.envpol.2013.07.048.

[139] M. Ibáñez, J. V. Sancho, F. Hernández, D. McMillan, R. Rao, Rapid non-target screening of organic pollutants in water by ultraperformance liquid chromatography coupled to time-of-light mass spectrometry, TrAC Trends Anal. Chem. 27 (2008) 481–489. doi:10.1016/j.trac.2008.03.007.

[140] M. Mezcua, O. Malato, J.F. García-Reyes, A. Molina-Díaz, A.R. Fernández-Alba, Accurate-mass databases for comprehensive screening of pesticide residues in food by fast liquid chromatography time-of-flight mass spectrometry., Anal. Chem. 81 (2009) 913–29. doi:10.1021/ac801411t.

[141] E. Schymanski, J. Jeon, R. Gulde, K. Fenner, M. Ruff, H.P. Singer, et al., Identifying small molecules via high resolution mass spectrometry: communicating confidence, Environ. Sci. Technol. 48 (2014) 2097–2098. http://pubs.acs.org/doi/abs/10.1021/es5002105 (accessed December 10, 2014).

[142] M. Zedda, C. Zwiener, Is nontarget screening of emerging contaminants by LC-HRMS successful? A plea for compound libraries and computer tools., Anal. Bioanal. Chem. 403 (2012) 2493–502. doi:10.1007/s00216-012-5893-y.

[143] F. Hernández, M. Ibáñez, T. Portolés, M.I. Cervera, J. V. Sancho, F.J. López, Advancing towards universal screening for organic pollutants in waters, J. Hazard. Mater. 282 (2015) 86–95. doi:10.1016/j.jhazmat.2014.08.006.

74

[144] S. Wolf, S. Schmidt, M. Müller-Hannemann, S. Neumann, In silico fragmentation for computer assisted identification of metabolite mass spectra, BMC Bioinformatics. 11 (2010) 148. doi:10.1186/1471-2105-11-148.

[145] J.E. Rager, M.J. Strynar, S. Liang, R.L. McMahen, A.M. Richard, C.M. Grulke, et al., Linking high resolution mass spectrometry data with exposure and toxicity forecasts to advance high-throughput environmental monitoring, Environ. Int. 88 (2016) 269–280. doi:10.1016/j.envint.2015.12.008.

[146] H.P. Singer, A.E. Wössner, C.S. McArdell, K. Fenner, Rapid Screening for Exposure to “Non-Target” Pharmaceuticals from Wastewater Effluents by Combining HRMS-Based Suspect Screening and Exposure Modeling, Environ. Sci. Technol. 50 (2016) acs.est.5b03332. doi:10.1021/acs.est.5b03332.

[147] E. Schymanski, Integrated Analytical and Computer Tools for Toxicant Identification in Effect-Directed Analysis, Fakultät für Chemie und Physik der Technischen Universität Bergakademie Freiberg, 2011.

[148] F. Hernández, J. V Sancho, M. Ibáñez, E. Abad, T. Portolés, L. Mattioli, Current use of high-resolution mass spectrometry in the environmental sciences., Anal. Bioanal. Chem. 403 (2012) 1251–64. doi:10.1007/s00216-012-5844-7.

[149] E.L. Schymanski, C.M.J. Gallampois, M. Krauss, M. Meringer, S. Neumann, T. Schulze, et al., Consensus structure elucidation combining GC/EI-MS, structure generation, and calculated properties, Anal. Chem. 84 (2012) 3287–3295. doi:10.1021/ac203471y.

[150] SCB Statistics Sweden, Antal småhusfastigheter efter region, fastighetstyp och avloppsanslutning. Vart femte år 2000 - 2014, (2014). http://www.statistikdatabasen.scb.se/pxweb/sv/ssd/START__MI__MI0902__MI0902C/MI0902T02/table/tableViewLayout1/?rxid=772e850f-82f8-4e16-a23b-c199b6c2d0ee (accessed January 16, 2017).

[151] E. Kholoma, G. Renman, A. Renman, Phosphorus removal from wastewater by field-scale fortified filter beds during a one-year study, Environ. Technol. 37 (2016) 2953–2963. doi:10.1080/09593330.2016.1170888.

[152] J. Luque-Garcı́a, M. Luque de Castro, Ultrasound: a powerful tool for leaching, TrAC Trends Anal. Chem. 22 (2003) 41–47. doi:10.1016/S0165-9936(03)00102-X.

[153] K. Wiberg, M. Oehme, P. Haglund, H. Karlsson, M. Olsson, C. Rappe,

75

Enantioselective analysis of organochlorine pesticides in herring and seal from the Swedish marine environment, Mar. Pollut. Bull. 36 (1998) 345–353. doi:10.1016/S0025-326X(97)00187-2.

[154] G. Gellerstedt, 8.1 Gel Permation Chromatography, in: Methods Lignin Chem., Springer-Verlag, Berlin Heidelberg, 1992: p. 487.

[155] E. de Hoffmann, V. Stroobant, Mass Spectrometry Principles and Appiliactions, Third Edit, Wiley-VCH Verlag, 2007.

[156] LECO Corporation, Pegasus GC-HRT+ Product Brochure (209-212), (2017). https://www.leco.com/component/edocman/pegasus-gc-hrt-brochure-209-212 (accessed January 17, 2018).

[157] S. Stephan, C. Jakob, J. Hippler, O.J. Schmitz, A novel four-dimensional analytical approach for analysis of complex samples, Anal. Bioanal. Chem. 408 (2016) 3751–3759. doi:10.1007/s00216-016-9460-9.

[158] H. Borsdorf, G.A. Eiceman, Ion mobility spectrometry: Principles and applications, Appl. Spectrosc. Rev. 41 (2006) 323–375. doi:10.1080/05704920600663469.

[159] S. Banerjee, S. Mazumdar, Electrospray Ionization Mass Spectrometry: A Technique to Access the Information beyond the Molecular Weight of the Analyte, Int. J. Anal. Chem. 2012 (2012) 1–40. doi:10.1155/2012/282574.

[160] S. Castillo, I. Mattila, J. Miettinen, M. Orešič, T. Hyötyläinen, Data analysis tool for comprehensive two-dimensional gas chromatography/time-of-flight mass spectrometry, Anal Chem. 83 (2011) 3058–3067. doi:10.1021/ac103308x.

[161] J.P. a Lijzen, M.G.J. Rikken, RIVM Report no. 601900005/2004 EUSES 2.0, 2004.

[162] U.S. EPA, Interpretive Assistance Document for Assessment of Discrete Organic Chemicals, Sustainable Futures Summary Assessment, 2013. https://www.epa.gov/sites/production/files/2015-05/documents/05-iad_discretes_june2013.pdf.

[163] www.epa.gov, EPIWEB 4.1, (2008).

[164] Swedish Meteorological and Hydrological Institute (SMHI), Modelldata per område, (2017). http://vattenweb.smhi.se/modelarea/ (accessed January 16, 2017).

76

[165] F. Wilcoxon, Individual Comparisons by Ranking Methods, Biometrics Bull. 1 (1945) 80–83.

[166] H. Mann, D. Whitney, On a test of wheter one of two random variables is stochastically larger than the other, Ann. Math. Stat. 18 (1947) 50–60. doi:10.1214/aoms/1177728422.

[167] J.C. Miller, J.N. Miller, Statistics for analytical chemistry, Second, Ellis Horwood Limited, Chichester, England, 1988.

[168] F. Wilcoxon, Individual comparisons of grouped data by ranking methods., J. Econ. Entomol. 39 (1946) 269. doi:10.2307/3001968.

[169] D. Hawkins, Biomeasurement. A student’s guide to biological statistics, Third, Oxford University Press, New York, 2014.

[170] M. Friedman, A Correction : The Use of Ranks to Avoid the Assumption of Normality Implicit in the Analysis of Variance Author ( s ): Milton Friedman Source : Journal of the American Statistical Association , Vol . 34 , No . 205 ( Mar ., 1939 ), p . Published by : Tayl, J. Am. Stat. Assoc. 34 (1939) 1–2.

[171] M. Friedman, A Comparision of alternative tests of significance for the problem of m rankings, Ann. Math. Stat. 11 (1940) 86–92.

[172] M. Friedman, The Use of Ranks to Avoid the Assumption of Normality Implicit in the Analysis of Variance, J. Am. Stat. Assoc. 32 (1937) 675–701.

[173] L. Eriksson, E. Johansson, N. Kettaneh-Wold, J. Trygg, C. Wikström, S. Wold, Multi-and Megavariate Data Analysis Part I Basic and Applications, Second edi, Umetrics AB, Umea, Sweden, 2006.

[174] S. Rännar, P.L. Andersson, A Novel Approach Using Hierarchical Clustering To Select Industrial Chemicals for Environmental Impact Assessment, J. Chem. Inf. Model. 50 (2010) 30–36. http://www.ncbi.nlm.nih.gov/pubmed/20050708.

[175] M. Stenberg, A. Linusson, M. Tysklind, P.L. Andersson, A multivariate chemical map of industrial chemicals--assessment of various protocols for identification of chemicals of potential concern., Chemosphere. 76 (2009) 878–84. doi:10.1016/j.chemosphere.2009.05.011.

[176] S. Fischer, Exposure Index Table on Product Register data for year 2011, (2011).

77

[177] M.M. Rahman, a. M. Abd El-Aty, J.H. Shim, Matrix enhancement effect: A blessing or a curse for gas chromatography?-A review, Anal. Chim. Acta. 801 (2013) 14–21. doi:10.1016/j.aca.2013.09.005.

[178] Swedish Meteorological and Hydrological Institute (SMHI), Februari 2015 - Lugn vårflod i söder, (2015). http://www.smhi.se/klimat/manadens-vader-och-vatten/sverige/laget-i-sveriges-sjoar-och-vattendrag/februari-2015-lugn-varflod-i-soder-1.84785 (accessed March 9, 2017).

[179] Swedish Meteorological and Hydrological Institute (SMHI), Mars 2015 – Normala vattenflöden och höga grundvattennivåer, (2015). http://www.smhi.se/klimat/manadens-vader-och-vatten/sverige/laget-i-sveriges-sjoar-och-vattendrag/mars-2015-normala-vattenfloden-och-hoga-grundvattennivaer-1.85605 (accessed March 9, 2017).

[180] A. Marklund, B. Andersson, P. Haglund, Traffic as a source of organophosphorus flame retardants and plasticizers in snow, Environ. Sci. Technol. 39 (2005) 3555–3562. doi:10.1021/es0482177.

[181] Ecofiltration Nordic AB, (n.d.). http://www.ecofiltration.se/en/reactive-filter-media/ (accessed December 18, 2017).

[182] M. Kah, G. Sigmund, F. Xiao, T. Hofmann, Sorption of ionizable and ionic organic compounds to biochar, activated carbon and other carbonaceous materials, Water Res. 124 (2017) 673–692. doi:10.1016/j.watres.2017.07.070.

[183] J. Dallüge, J. Beens, U. a T. Brinkman, Comprehensive two-dimensional gas chromatography: a powerful and versatile analytical tool., J. Chromatogr. A. 1000 (2003) 69–108. http://www.ncbi.nlm.nih.gov/pubmed/12877167.