Pharmaceuticals – improved removal

from municipal wastewater and their

occurrence in the Baltic Sea

Berndt Björlenius

Doctoral Thesis

KTH Royal Institute of Technology

School of Engineering Sciences in Chemistry, Biotechnology and Health

Stockholm 2018

© Berndt Björlenius 2018 KTH Royal Institute of Technology School of Engineering Sciences in Chemistry, Biotechnology and Health AlbaNova University Center SE-106 91 Stockholm Sweden Printed by Universitetsservice US-AB 2018 ISBN 978-91-7873-047-6 TRITA-CBH-FOU-2018-62 Front cover: The mobile pilot plant visiting Västerås WWTP. Photo: Berndt Björlenius

Till min älskade familj

i

Abstract

Pharmaceutical residues are found in the environment due to extensive use in

human and veterinary medicine. The active pharmaceutical ingredients

(APIs) have a potential impact in non-target organisms. Municipal wastewater

treatment plants (WWTPs) are not designed to remove APIs.

In this thesis, two related matters are addressed 1) evaluation of advanced

treatment to remove APIs from municipal wastewater and 2) the prevalence

and degradation of APIs in the Baltic Sea.

A stationary pilot plant with nanofiltration (NF) and a mobile pilot plant with

activated carbon and ozonation were designed to study the removal of APIs at

four WWTPs. By NF, removal reached 90%, but the retentate needed further

treatment. A predictive model of the rejection of APIs by NF was developed

based on the variables: polarizability, globularity, ratio hydrophobic to polar

water accessible surface and charge. The pilot plants with granular and

powdered activated carbon (GAC) and (PAC) removed more than 95% of the

APIs. Screening of activated carbon products was essential, because of a broad

variation in adsorption capacity. Recirculation of PAC or longer contact time,

increased the removal of APIs. Ozonation with 5-7 g/m3 ozone resulted in 87-

95% removal of APIs. Elevated activity and transcription of biomarkers

indicated presence of xenobiotics in regular effluent. Chemical analysis of

APIs, together with analysis of biomarkers, were valuable and showed that

GAC-filtration and ozonation can be implemented to remove APIs in WWTPs,

with decreased biomarker responses.

Sampling of the Baltic Sea showed presence of APIs in 41 out of 43 locations.

A developed grey box model predicted concentration and half-life of

carbamazepine in the Baltic Sea to be 1.8 ng/L and 1300 d respectively.

In conclusion, APIs were removed to 95% by GAC or PAC treatment. The

additional treatment resulted in lower biomarker responses than today and

some APIs were shown to be widespread in the aquatic environment.

Keywords

Advanced wastewater treatment, WWTP, pilot plant, pharmaceutical

residues, removal of pharmaceuticals, activated carbon, ozonation,

nanofiltration, biomarker, Baltic Sea

ii

Sammanfattning

Den omfattande användningen av human- och veterinärmediciner gör att

läkemedelsrester kan återfinnas i miljön. Dessa substanser kan ha en

påverkan på andra organismer än människan. Kommunala reningsverk är

inte byggda för att ta rena bort läkemedelsrester.

I avhandlingen diskuteras två forskningsområden 1) utvärdering av avancerad

rening för att ta bort läkemedelsrester från kommunalt avloppsvatten och 2)

förekomst och nedbrytning av läkemedelsrester i Östersjön.

En stationär och en mobil pilotanläggning med nanofiltrering (NF), aktiverat

kol och ozonering designades för att studera avskiljning av läkemedelsrester

vid fyra reningsverk. NF avskilde 90% av dessa, men retentatet måste

behandlas ytterligare. En väl predikterande modell för avskiljningen med NF

utvecklades med variablerna: polariserbarhet, sfäriskhet, kvoten mellan

hydrofob och polär tillgänglig yta och ämnets laddning. Linjerna med

granulerat (GAC) och pulveriserat (PAC) aktiverat kol tog bort minst 95% av

läkemedelsresterna. Urvalstest av aktiverat kol är viktiga pga stor variation i

adsorptionskapacitet mellan olika produkter. Recirkulation av doserad PAC

eller längre kontakttid ökade avskiljningsgraden. Ozonering med dosen 5–7

g/m3 gav 87–95% avskiljning av läkemedelsrester. Biomarkörer i regnbågslax

indikerade förekomst av xenobiotika i dagens utgående vatten. Den

avancerade rening som utvecklats, minskade signifikant biomarkörernas

respons. Kemisk analys i kombination med analys av biomarkörer visade att

ozonering och aktiverat kol kan användas i reningsverk för att ta bort

läkemedelsrester till 90%-95%.

En provtagning av Östersjön visade på förekomst av läkemedelsrester på 41

av 43 platser. En utvecklad ”grey-box” modell predikterade koncentration och

halveringstid av karbamazepin i Östersjön till 1,8 ng/l respektive 1300 d.

Sammanfattningsvis visade studien att läkemedelsrester kunde avskiljas till

95% med aktiverat kol, med lägre respons i biomarkörer än idag och att de är

vitt spridda i miljön.

Nyckelord

Avancerad avloppsvattenrening, reningsverk, pilotanläggning, läkemedel,

avskiljning av läkemedelsrester, aktiverat kol, ozonering, nanofiltrering,

biomarkör, Östersjön

iii

List of publications

This thesis is based on the following publications:

Paper I.

Björlenius, B., Ripszám, M., Haglund, P., Lindberg, R.H., Tysklind, M., and

Fick, J. (2018). Pharmaceutical residues are widespread in Baltic Sea coastal

and offshore waters – Screening for pharmaceuticals and modelling of

environmental concentrations of carbamazepine. Sci. Total Environ. 633,

1496–1509.

Paper II.

Flyborg, L., Björlenius, B., Ullner, M., and Persson, K.M. (2017). A PLS

model for predicting rejection of trace organic compounds by nanofiltration

using treated wastewater as feed. Sep. Purif. Technol. 174, 212–221.

Paper III.

Kårelid, V., Larsson, G., and Björlenius, B. (2017). Pilot-scale removal of

pharmaceuticals in municipal wastewater: Comparison of granular and

powdered activated carbon treatment at three wastewater treatment plants. J.

Environ. Manage. 193, 491–502.

Paper IV.

Kårelid, V., Larsson, G., and Björlenius, B. (2017). Effects of recirculation

in a three-tank pilot-scale system for pharmaceutical removal with powdered

activated carbon. J. Environ. Manage. 193, 163–171.

Paper V.

Beijer1 K., Björlenius1 B., Shaik, S., Lindberg, R.H., Brunström, B., and

Brandt, I. (2017). Removal of pharmaceuticals and unspecified contaminants

in sewage treatment effluents by activated carbon filtration and ozonation:

Evaluation using biomarker responses and chemical analysis. Chemosphere

176, 342–351. 1Equal contribution.

iv

Contribution to appended publications

Paper I: BB initiated the study and planned the major sampling campaign,

collected additional data, developed and evaluated the model for predicting

environmental concentrations. He wrote major parts of the manuscript and

was the corresponding author.

Paper II: BB initiated the study, proposed the strategy of using MVA, designed

the pilot setup and took part in the design of the experiments, operation of the

pilot plant and in the evaluation of data. BB derived the final equation for

prediction of rejection of substances by nanofiltration. He wrote equal parts

of the manuscript.

Paper III. BB initiated the study, designed the pilot plants, wrote parts of the

manuscript and planned and performed the experiments together with Victor

Kårelid (VK). BB planned, evaluated and made parts of the bench scale

screening of PAC products. BB was the principal supervisor of the PhD student

VK.

Paper IV. BB initiated the study, design the flexible PAC lines, wrote parts of

the manuscript and was responsible for the original treatment concept. He

was together with VK responsible for the planning and execution of the

experiments. BB was the principal supervisor of the PhD student VK.

Paper V. BB initiated the study, designed the pilot plants and fish tank system,

operated the pilot plant during the Käppala biotests and evaluated the

chemical analysis. He wrote parts of the manuscript and was the

corresponding author.

v

Related publications not included in this thesis

Östman M., Björlenius B., Fick J., Tysklind M. (2018) Effect of full-scale

ozonation and pilot-scale granular activated carbon on the removal of

biocides, antimycotics and antibiotics in a sewage treatment plant. Sci. Total

Environ, 649, 1117–1123.

Zhang W., Blum K., Gros M., Ahrens L., Jernstedt H., Wiberg K., Andersson

P.L. Björlenius B., Renman G.W. (2018) Removal of micropollutants and

nutrients in household wastewater using organic and inorganic sorbents.

Desalination Water Treat, 120, 88–108.

Pohl, J., Björlenius, B., Brodin, T., Carlsson, G., Fick, J., Larsson, D.G.J.,

Norrgren, L., and Örn, S. (2018). Effects of ozonated sewage effluent on

reproduction and behavioral endpoints in zebrafish (Danio rerio). Aquat.

Toxicol. 200, 93–101.

Wang, H., Sikora, P., Rutgersson, C., Lindh, M., Brodin, T., Björlenius, B.,

Larsson, D.G.J., and Norder, H. (2018). Differential removal of human

pathogenic viruses from sewage by conventional and ozone treatments. Int. J.

Hyg. Environ. Health 221, 479–488.

Bengtsson-Palme, J., Hammarén, R., Pal, C., Östman, M., Björlenius, B.,

Flach, C.-F., Fick, J., Kristiansson, E., Tysklind, M., Larsson, D.G.J., (2016).

Elucidating selection processes for antibiotic resistance in sewage treatment

plants using metagenomics. Sci. Total Environ. 572, 697–712.

Ågerstrand M., Berg C., Björlenius B., Breitholtz M., Brunström B., Fick J.,

Gunnarsson L., Larsson D. G. J., Sumpter P. J., Tysklind M., and Rudén C.

(2015). Improving Environmental Risk Assessment of Human

Pharmaceuticals. Environ. Sci. Technol., 49, 5336–5345.

Cuklev, F., Gunnarsson, L., Cvijovic, M., Kristiansson, E., Rutgersson, C.,

Björlenius, B., Larsson, D.G.J., (2012). Global hepatic gene expression in

rainbow trout exposed to sewage effluents: A comparison of different sewage

treatment technologies. Sci. Total Environ. 427–428.

Minten, J., Adolfsson-Erici, M., Björlenius, B., Alsberg, T., (2011). A method

for the analysis of sucralose with electrospray LC/MS in recipient waters and

in sewage effluent subjected to tertiary treatment technologies. International

Journal of Environmental Analytical Chemistry 91, 357–366.

vi

Samuelsson M. L., Björlenius B., Förlin L., Larsson D. G. J., (2011).

Reproducible 1H NMR-Based Metabolomic Responses in Fish Exposed to

Different Sewage Effluents in Two Separate Studies. Environ. Sci. Technol.,

45, 1703–1710

Wahlberg, C., Björlenius, B., Paxéus, N., (2011). Fluxes of 13 selected

pharmaceuticals in the water cycle of Stockholm, Sweden. Water Sci. Technol.

63, 1772–1780.

Lundström, E., Adolfsson-Erici, M., Alsberg, T., Björlenius, B., Eklund, B.,

Lavén, M., Breitholtz, M., (2010). Characterization of additional sewage

treatment technologies: Ecotoxicological effects and levels of selected

pharmaceuticals, hormones and endocrine disruptors. Ecotoxicol. Environ.

Saf. Safety 73.

Lundström, E., Björlenius, B., Brinkmann, M., Hollert, H., Persson, J.-O.,

Breitholtz, M., (2010). Comparison of six sewage effluents treated with

different treatment technologies—Population level responses in the

harpacticoid copepod Nitocra spinipes. Aquat. Toxicol. 96, 298–307.

Flyborg, L., Björlenius, B., Persson, K.M., (2010). Can treated municipal

wastewater be reused after ozonation and nanofiltration? Results from a pilot

study of pharmaceutical removal in Henriksdal WWTP, Sweden. Water Sci.

Technol., 61, 1113-1120.

Gunnarsson, L., Adolfsson-Erici, M., Björlenius, B., Rutgersson, C., Förlin,

L., Larsson, D.G.J., (2009). Comparison of six different sewage treatment

processes—Reduction of estrogenic substances and effects on gene expression

in exposed male fish. Sci. Total Environ. 407, 5235–5242.

Reports

Wahlberg, C., Björlenius, B., Ek, M., Paxéus, N., Gårdstam, L., 2008.

Avloppsreningsverkens förmåga att ta hand om läkemedelsrester och andra

farliga ämnen redovisning av regeringsuppdrag: 512-386-06 Rm.

Naturvårdsverket, Stockholm.

Wahlberg, C., Björlenius, B., Paxéus, N., 2010. Läkemedelsrester i

Stockholms vattenmiljö. Stockholm Vatten, Stockholm

vii

Table of contents

Abstract i

Sammanfattning ii

List of publications iii

List of abbreviations ix

Introduction

1 Introduction to pharmaceuticals in wastewater 1

1.1 The history of water and wastewater treatment and management 2

1.2 The development of treatment technologies 6

1.3 Development of legislation on water pollution in Europe 13

1.4 Market of pharmaceuticals 14

1.5 Routes/pathways of pharmaceuticals in the environment 14

1.6 Reported effects of pharmaceuticals in the environment 17

1.7 Environmental risk assessment (ERA) 19

1.8 Evaluation of chemical data -Deconjugation and accuracy 22

1.9 Pharmaceuticals in WWTPs and the aquatic environment 23

1.10 Potential technologies for removal of pharmaceuticals 29

Potential of biological treatment for removal of pharmaceuticals 30

Potential of separation processes for removal of pharmaceuticals 39

Potential of chemical oxidation for removal of pharmaceuticals 45

Technologies in development 52

Comparison of treatment technologies 52

viii

Present investigation

2 Aim and strategy 56

3 Methodology 58

Sampling 58

Measurements and analysis 58

Data collection 59

Calculations and Modelling 60

Wastewater treatment plants selected for pilot tests 60

Design, construction and description of pilot plants 62

4 Result and discussion 69

4.1 Use of pharmaceuticals (Paper I-V) 69

4.2 Pharmaceuticals in the environment (Paper I) 74

4.3 Pharmaceutials in wastewater effluents (Paper II-V) 80

4.4 Removal of pharmaceutical residues by nanofiltration (Paper II) 84

4.5 Treatment with GAC and PAC (Paper III and IV) 87

4.6 Examination of process parameters in the PAC lines (Paper IV) 98

4.7 Treatment with ozone (Paper V) 103

4.8 Evaluation of treatment technologies by biomarkers (Paper V) 108

4.9 Comparison of treatment technologies 110

4.10 Spin-off of the ozonation pilot tests 113

4.11 Summary of work done and main results 114

5 Conclusion, future perspective and recommendations 118

6 Acknowledgements 121

7 References 123

ix

List of abbreviations

AC Activated carbon

AGS Aerated granular sludge

AMR Antimicrobial resistance

AOP Advanced oxidation processes

API Active pharmaceutical ingredient

BNR Biological nitrogen removal

BOD Biochemical oxygen demand

BV Bed volume

CAS Conventional activated sludge

CEC Critical environmental concentration

COD Chemical oxygen demand

CUR Carbon usage rate

CYP Cytochrome P450

DDD Daily defined dose

DOC Dissolved organic carbon

EBCT Empty bed contact time

EQS Environmental Quality Standard

ERA Environmental risk assessment

EROD Ethoxyresorufin-O-deethylase

GAC Granular activated carbon

HRT Hydraulic retention time

LOQ Limit of quantification

MEC Measured environmental concentration

MLSS Mixed liquor suspended solids

MWCO Molecular weight cut off

NF Nanofiltration

NSAIDS Non-steroidal anti-inflammatory drugs

OSSF On-Site Sewage Facilities

x

PAA Peracetic acid

PAC Powdered activated carbon

PCA Principle Components Analysis

PE Population equivalents

PEC Predicted environmental concentration

PLS Partial Least Squares Projection of Latent Structures Analysis

RO Reverse osmosis

SS Suspended solids

STP Sewage treatment plant

TOC Total organic carbon

VRF Volume reduction factor

WFD Water Framework Directive

WHO World Health Organization

WWTP Wastewater treatment plant

xi

Ideas are like rabbits. You get a couple and learn how to handle them, and pretty soon you have a dozen.

John Steinbeck

Introduction | 1

1 Introduction to pharmaceuticals in wastewater

Residues of active pharmaceutical ingredients (APIs) are found in the

environment due to extensive use in human and veterinary medicine1–5. The

APIs are designed to pass the stomach intact and fit in molecular receptors in

humans which are often evolutionary conserved in a variety of organisms

leading to a potential impact of APIs in non-target organisms in the

environment6–8. The human body conjugates many of the APIs to facilitate the

excretion of APIs via urine9. Most of the APIs consumed by humans are

excreted in urine and feces, which to a large extent are collected by sewer

systems. Some APIs are used externally on skin and will also mainly end up in

the sewer system after shower and bath. The existing municipal wastewater

treatment plants (WWTPs) are not designed to remove pharmaceutical

residues and the most advanced WWTPs remove up to a maximum average of

50-60% of the pharmaceuticals in the influent1,10.

The present work focused on treatment of effluent from municipal WWTPs

since 95% of the APIs will appear in the water phase and only a small minority

of API will accumulate in sludge. Traditional upstream control measures are

valuable but will only influence 5-10 percent of the number of pharmaceuticals

in the wastewater – in many cases we still must take our medicines and the

residuals will, after passage in our bodies, enter the sewer network or on-site

sewage facilities for single households. Examples of upstream control is to

lower the doctors´ prescription of antibiotics or prescription of physical

activity instead of medication. A potential upstream measure is the

development of “green” medicines that decompose into non-persistent, non-

toxic or not bio accumulative substances, but they will take decades to bring

to the market, if ever feasible, to replace existing pharmaceuticals. Hospitals,

though not contributing with more than a few percent of API to the total

amount in a city10, are point sources of some APIs e.g. from clinics where

antibiotics are used and are potential target for implementation of

pretreatment before discharge to WWTPs. Several technologies have been

proposed to remove APIs in municipal wastewater and the most promising

can also be implemented in pretreatment facilities at hospitals, but only after

a pretreatment which also must be installed to degrade bulk organic

substances in hospital wastewater.

In summary, 90-95% of the mass of APIs in municipal wastewater has passed

human bodies, which suggests that end of pipe treatment at municipal

2 | Introduction

WWTPs must be implemented, if we want to remove APIs from the water

cycle. The end of pipe installations will be the main solution for decades, but

upstream control is important until biologically degradable APIs are brought

to the market.

1.1 The history of water and wastewater treatment and

management

Water and wastewater management has a long history. Irrigation of

agricultural land was the earliest application of water management. The long-

term increasing population in cities has demanded solutions for water supply

and wastewater disposal and the policy of all times has been development and

improvement of the water and wastewater systems.

1.1.1 Ancient history

Irrigation of farmland was the first organized water management measure

taken starting before urban development of water distribution and sewer

networks. In Egypt, dams were constructed to level out flooding and facilitate

storage of water for irrigation. Examples of early urban water and wastewater

application come from Habuba Kebira, a Sumerian settlement, where

terracotta pipes were used for urban drainage and water distribution in 3500

BC. The layout of the city drainage system in Habuba Kebira was however not

a first-time design, but incorporated knowledge of already existing systems in

the Sumerian metropolis of Uruk, situated 900 km from the settlement11,12.

Toilet facilities from different eras in Mesopotamia were found in a few houses

of the Akkadian (2335–2155 BC) and at Tell Asmar, but also toilet structures

from earlier eras have been found13. In 3500 to 3000 BC, many cities in

Mesopotamia had networks of wastewater and stormwater drainage. The

Indus civilization had bathrooms in houses and sewers in streets 3000 BC14.

During early civilizations, the water distribution in urban locations included

canals transferring water from rivers, rainwater harvesting systems, wells,

aqueducts, and underground storage tanks15.

The Minoans and Mycenaeans built aqueducts from springs connected with

tunnels and self-cleaning pipes. The water distribution system included

sedimentation basins for removal of silt. During the later Hellenistic period,

pressurized pipes and inverted siphons were incorporated in water

distribution systems. The cisterns for drinking water, built in many places,

were also a safety in wartime. In the Minoan era, lavatories were flushed with

reused grey water or stormwater and this principle remained in use in later

eras14.

Introduction | 3

1.1.2 Water and wastewater management in Rome and the

Roman empire

The history of water management in Rome starts 600 BC with the first stretch

of Cloaca Maxima. It was first built during the eras of Etruscan kings as

drainage from an area in Rome, that later would be Forum Roman, to the river

Tiber. The channel had stone walls with wooden bridges that also prevented

the walls from collapsing 16. Side channels from public lavatories “latrina” and

street run-off were later connected to Cloaca Maxima, which then was

converted to be a combined sewer tunnel by a construction of a roof on top of

the stone walls. No channels or pipes from private buildings were connected.

The houses were connected to cesspits instead17. At least twelve other major

combined sewers (cloacae) were built in Rome before 1 BC18.

The first centuries after the foundation of Rome in 753 BC, water from wells

and river Tiber was used as water supply. Rome had later its major water

supply by the aqueducts of which the oldest Aqua Appia, built 312 BC, was 16

km long, stretching mainly underground and had a discharge capacity of

73 000 m3 per day. The last constructed aqueduct suppling Rome with water,

Aqua Alexandrina was built in 226 AD. At the end of the Republic, the one

million inhabitants in Rome benefited from a daily supply of some

453 000 m3 water19. Frontinus, senator and responsible water commissioner

or director for the advanced water supply, named as “curator aquarum” for

the city of Rome, was in charge 97-103 AD. He described in a booklet, later

translated into English, the water quality, volume, supply and distribution via

the nine active aqueducts, tunnels, cisterns and pipes to Rome, but also

relevant laws20,21. The detailed descriptions in Frontinus booklet portray e.g.

the individual capacity of the aqueducts, the 25 design flows of the bronze

nozzles used to deliver fixed amount of water and a list of the 16 previous

curator aquarum, from 13 AD, until Frontinus arrival in 97 AD. Frontinus also

described the maintenance of the aqueducts e.g. how calcium carbonate

deposits should be removed, which otherwise lined the conduit with 1 mm per

year.

The technical skills of the engineering and construction are shown in the

example of the aqueduct of Nemausus, built around 20 BC. This aqueduct

conveyed water approximately 50 km from Uzes to the castellum in the

Roman city of Nemausus (Nimes in France), with an elevation difference of

17 m, corresponding to an average slope of 0.34 mm/m along the distance.

Settling tanks (piscinae) were located along the aqueducts to remove

sediments. In the cities, a fraction of the water was stored in cisterns. The

4 | Introduction

cistern Piscina Mirabilis, near Naples, Italy, is one of the largest Roman

cisterns, with a capacity of 12 600 m3 of water19. The largest ancient cistern is

probably the Yerebatan Saryi or Basilica Cistern in Istanbul, built during 6th

century AD, with a storage capacity of 100 000 m3 water22.

In the distribution systems of water from the aqueducts, lead pipes were used

in the junction boxes19. There are however two reasons why lead poisoning of

the romans from water pipes did not occur. Firstly, calcium carbonate scaling

formed an insulation layer that prevented the water to come in contact with

the lead. Secondly, the water was in constant flow through the pipe, giving a

short contact time in the last short pipe stretch, outlined in lead.

In the Roman Empire, over 1 600 aqueducts have been described in the

Mediterranean basin23, e.g. Figure 1, but over 100 aqueducts were constructed

in other parts of the Roman Empire, in today´s Austria, Belgium, Bulgaria,

Germany, Hungary, Luxemburg, the Netherlands, Portugal, Romania,

Switzerland, Ukraine and United Kingdom24.

Figure 1: Section of the Fontveille aqueduct and a distribution lead pipe

built and installed in the 2nd century in Antibes, France. Photo: B. Björlenius

Introduction | 5

1.1.3 Post roman time in Europe

After the fall of the Roman empire, there was a recession in hygienic

conditions and although the aqueducts were still in operation, they were not

maintained, resulting in malfunctions. Efforts to restore aqueducts were made

in Paris during the 15th century. In the same city, the first sewer was built in

1370, beneath rue Montmartre.

The great conduit in London was an underground channel, constructed in mid

13th century, which distributed spring water more than 4 km into central

London. Later, 6 inch lead pipes were installed to transport drinking water in

12 similar conduits systems. A major leap in development was taken in 1580,

when Peter Morice applied to city officials for permission to construct a water-

wheel and pump under one of the arches of London Bridge. The purpose was

to supply the city with cooking water. It was in operation for many years, but

it was destroyed in the great fire in 1666, as the great conduits consisted of

lead pipes and junction boxes, which melted by the heat. It was however

rebuilt and was in operation until early nineteenth century25.

In London, public latrines, provided with running water for flushing, were

available in the 13th century26. The wastewater was directly or finally

discharged into river Thames, still serving as the major source of potable

water.

On the European continent, the first water conduit to be built after the

aqueduct era was in Hamburg in 1370, where drilled hollow logs led fresh

spring water, from Altona to several wells in Hamburg. The water conduit was

named Katharinen-Brunnen and was in operation for centuries, as can be

understood from the reported chemical analysis of the water, dating from

180127,28.

Some early examples of wastewater treatment come from the 16th and 17th

century, with the concept of sewage farms, where crops were irrigated with

wastewater, collected in sewer systems. Larger cities like Berlin and Paris

implemented large sewage farms. In Paris, irrigation on farmland was

established along the river Seine, eventually treating all sewage from the

city29,30. In 1948 the sewage farms covered an area of 4487 ha and they

produced more than 10% of the vegetables sold at the central market Les

Halles31. The yields were higher in the sewage farm fields, than in fields at

regular farms48. Still in the beginning of the 21th century, a fraction of partly

treated sewage in Paris was irrigated onto 600 ha of the sewage fields. Later,

6 | Introduction

the cultivation of only non-food crops on the irrigated fields was

considered30,33.

In 1858, the so-called Great stink in London, forced measures to clean the

river Thames from the waste from over 2 million Londoners and many

industries. It was a great opportunity to implement treatment facilities for the

wastewater, e.g. sewage farms, but at this time, only measures for collection

and transport of the wastewater in a sewer network was undertaken. The

collected wastewater was sophisticatedly discharged and diluted in river

Thames, at a few spots downstream the city: In the ends of the two major

sewers on the north and south side of river Thames, Abbey Mills and

Crossness pumping station were built respectively. The wastewater was stored

in reservoirs and pumped at high tide to river Thames. The large-scale project

was the solution to the great stink and it saved many humans from new

outbreaks of cholera in the 19th century. In 1891 the first treatment step with

sedimentation tanks was taken in operation at the Crossness Sewage

Treatment Works, which now is one of largest WWTPs in Europe34–36.

1.2 The development of treatment technologies

The driving force for the development of municipal wastewater treatment

plants (WWTPs) has often been political decisions or successively sharpened

legislation, based on technological and scientific conclusions on the influence

on human health or the receiving water. The development of the WWTPs has

proceeded during more than 150 years, starting from the water works

development in 19th century. The wastewater treatment plants were preceded

of a period when wastewater was collected and transported in pipes and

tunnels away from the cites, to less populated areas, to avoid spreading of

diseases and the general pollution in the waters, in the vicinity of the cities.

During the 20th century, WWTPs have been established in many cities. The

WWTPs have been extended successively to remove more and more groups of

pollutants, starting with coarse material and particles, later bulk organic

compound, phosphorus and nitrogen. The individual extensions have often

been made during different decades.

In the line of continually improved wastewater treatment, technologies for

removal of micropollutants, i.e. inorganic and organic substances with

negative effects on the environment, are developed. The negative effects come

from the persistent, bioaccumulative and toxic properties of the substances

and the name micropollutants comes from their prevailing concentrations in

the low micrograms per liter or lower. Currently, focus is mainly on the large

Introduction | 7

group of pharmaceutical residues, that is mapped and evaluated and, in a few

cases, treatment technologies for their removal from wastewater are

implemented.

1.2.1 Early development of treatment technologies – removal of

organic compounds

Today, most removal technologies for wastewater treatment operate in

continuous mode and are designed to have a sufficient average removal

efficiency, independently of the hydraulic load. The first developed

technologies for biological wastewater treatment were however operated in

batch mode.

Biofilm processes

An early introduced method for wastewater treatment, applying irrigation

over filter materials, like crushed rock or sand, was tested in Paris, Berlin and

Manchester and it was first operated in batch mode. In the attached growth of

microorganisms, a biofilm containing different bacteria species will developed

on the surface of a filter material and the bacteria will break down organic

material in the wastewater. When the easily degradable organic material has

been degraded, nitrification of the wastewater will take place potentially. The

succeeding development of biofilm processes, included trickling filters on

initially coarse stone material and later on corrugated plastic sheets37. Starting

in the late 1980’s, the development of moving carrier biofilm processes had

led to many applications for removal of organic matter and nitrogen38.

SBR Sequencing Batch Reactor

The SBR process was developed in the period 1914-1920 when also full-scale

SBR-plants were in operation39. The batch process provided great flexibility

with fill, reaction, settling, decanting and idle sequences for treating

wastewater. Problems with the equipment, high demand for operators

attention and the limited possibilities for automation at the time being,

limited the use until the end of the 1950’s, when development of SBRs started

again40,41. Today SBR technology is used for removal of organic matter,

nitrogen and phosphorous in the main stream in some municipal WWTPs, but

also in treating side streams, like supernatant from digested sludge41.

8 | Introduction

Activated sludge

The concept of the activated sludge process was presented in 1914, based on

research and development that was ongoing from 1882. The essential lab test

with activated sludge were performed as batch tests, where a fraction of the

accumulated solids was kept in the system between the batches. Oxidation of

organic matter and nitrification were observed after five weeks of batch

operation. The researchers behind the experiments, Ardern and Lockett

named the accumulated solids in the wastewater activated sludge. Further

development was intensive in UK and USA and in 1914 the process was

operated at two full-scale plants in UK: one plant with fill and draw (SBR-

type) and one plant with continues flow. The first plant in USA was taken in

operation in 1916. Plants with continuous flow were dominating from the

start, 18 of the first 21 plant built during 1914-1927, used continues

operation42. The activated sludge process is still central in biological treatment

all over the world and can contain biological nitrogen and phosphorous

removal with different process configurations.

Activated carbon

Hindu documents dating from 450 BC refer to the use of sand and charcoal

filters for the purification of drinking water. The specific adsorptive properties

of charcoal (the forerunner of activated carbon) were first described by

Scheele in 1773 in the treatment of gases. Later, in 1786, Lowitz performed

experiments on the decolorizing of solutions. In 1862, Lipscombe prepared a

carbon material to purify potable water. Activated carbon has gained

importance, especially since the mid 1960s, as an adsorptive material in the

treatment of municipal and industrial wastewaters. The first full-scale

advanced (tertiary) wastewater treatment plant incorporating GAC was put

into operation in 1965 in South Lake Tahoe, California43.

1.2.2 Development and implementation of wastewater

treatment plants in Sweden

In Sweden, the first centralized treatment of wastewater was taken in

operation in 1904 and consisted of a septic tank for 250 inhabitants, which

was constructed in Storängen, Nacka east of Stockholm. The first WWTP with

biological treatment with biofilters was taken in operation in Skara in 1911 and

the first oxidation ponds were built in 1933-39 in Lund, followed of the first

activated sludge process built in Kristianstad 1941. An early and cost effective

Introduction | 9

design of WWTPs was to construct sedimentation basins for removal of

particles and apply anaerobic digestion of the removed sludge, which was

installed in Stockholm at Bromma and Henriksdal WWTPs in 1936 and 1941

respectively44,45. The WWTPs in Sweden have successively been extended to

meet the effluent standards given to reduce the impact of increasing pollution

loads on receiving waters. On average, a new fundamental treatment step has

been introduced every 20 years in Sweden44, Table 1. The last two posts have

been selected by the author, as they are the most probable process

representatives of the last decades.

Table 1. Cycles of implementation of treatment steps in Swedish WWTPs.

Decade Treatment

1910’s Septic tanks

1930’s Mechanical treatment

1950’s Biological treatment

1970’s Removal of phosphorous

1990’s Removal of nitrogen

2010’s Removal of pharmaceutical residues /

Micropollutants

The implementation of the main treatment steps in Swedish has continued

from 1930’s until today, Figure 2. Four major causes of the extensions of the

WWTPs can be identified44: 1) Primary treatment to remove course material

and sludge that polluted lakes and caused odour and esthetic problems 2)

Biological treatment to remove organic matter and decrease the number of

bacteria – this extension was accelerated by the outbreak of Salmonella in

Sweden 195346,47. 3) Chemical precipitation of phosphorous to reduce the

eutrophication in Swedish lakes cause by detergents and increasing

population in the late 1960’s 4) Biological nitrogen removal to reduce nitrogen

in mainly marine environment which otherwise causes eutrophication and

oxygen shortage at sea bottoms along the Swedish coastline in the 1980’s.

10 | Introduction

An example from activities in the latest cycle is Sweden’s first full-scale

treatment step with ozonation for removal of pharmaceutical residues at

Knivsta WWTP. This treatment step was designed by the author, based on

findings partly presented in this thesis. The ozonation step was designed for

12 000 population equivalents (PE) and was installed and operated at Knivsta

WWTP in 2015-2016. The second full-scale plant with removal of

pharmaceutical residues in Sweden is currently under operation trials at

Nykvarn WWTP in Linköping (Robert Sehlén, personal communication,

October 18, 2018). However, the latter plant is the first permanent and largest

full-scale ozonation step in Sweden, with a connected population of 145 200

persons, but with a load of organic matter corresponding to 235 000 PE48.

Figure 2: Development of Swedish WWTPs. Type of treatment and share of total volume treated49.

1.3.4 Number of Swedish WWTPs and their removal efficiencies

The official statistics of water emissions from WWTPs cover WWTPs which

have at least 2 000 people connected or a corresponding load of organic

material, measured as biochemical oxygen demand (BOD7), of at least 2 000

population equivalents (PE)49. The sizes of WWTPs are divided into five

classes, depending of the number of people connected, Table 2.

The removal efficiencies of phosphorous (P) are independent of the size of the

WWTP, due to the applied chemical precipitation, that can be controlled to

achieve 90-95% removal of phosphorous, depending of the effluent standard.

Removal efficiency for nitrogen (N) is lower for smaller WWTPs, since their

biological treatment is designed to fulfil the effluent standards of either 50%

or 80% removal of nitrogen. BOD7 have a generally high removal efficiency in

Introduction | 11

the WWTPs. The removal efficiency increases slightly with increasing size of

the WWTP, mainly as a result of measure taken to fulfil higher effluent

standards for phosphorous and nitrogen at larger WWTPs49. In addition,

there are smaller WWTPs, which are divided into two major groups: those

designed for 25-200 PE and those designed for 200 to 2 000 PE respectively.

They are however not included in the statistics. These smaller WWTPs are

considered to achieve lower removal efficiencies than the larger plants.

Furthermore, Sweden has 700 000 on-site sewage facilities (OSSFs) for

summer houses and rural areas. The OSSFs have very diverse removal

efficiencies depending of type of applied treatment technology and type of

substance.

Table 2: Number of treatment plants in Sweden and removal efficiencies [%] in 201449.

Number of

people

connected [PE]

Number

of

WWTP

Total

number of

people

connected

[PE]

Removal

efficiency of

P [%]

Removal

efficiency

of N [%]

Removal

efficiency

of BOD7 [%]

2 000 – 10 000 246 678 682 95 38 93

10 001 – 20 000 71 602 021 95 56 96

20 001 – 50 000 64 1 190 827 94 55 96

50 001– 100 000 31 1 351 440 95 60 97

>100 000 19 4 226 783 95 70 97

Total / Average 431 8 049 753 95 56 96

1.2.3 Centralized water distribution and sewer systems in

Stockholm – a case study

In the end of 17th century, Stockholm had 300 wells supplying the

Stockholmers with drinking water. The number of wells increased in parallel

with increasing population. In general, the wells had good water quality, due

to the continuous water outflow along the Brunkeberg esker, which had a high

hydraulic capacity to provide ground water. Some wells situated close to

tanneries and central shorelines had worse water quality than wells along the

esker. In the 17th century a water conduit of hollow logs was constructed

leading water to a fountain in the Royal Garden, north of the Royal Castle50.

The pandemics of cholera in the 19th century became the driving force for a

centralized water distribution system. The 2nd pandemic of cholera reach

Stockholm in 1834 and the 3rd in 1850, with yearly cholera outbreaks

summertime until 1859. Despite the fact that both the cholera causing toxin,

12 | Introduction

and the bacteria Vibrio cholerae producing it, was unknown, the spreading of

cholera in London was located to well water, by the English physician John

Snow51. Filippo Pacini wrote a paper on the cholera causing bacteria in 1854,

but the discovery seems to have been ignored, or not spread, until Koch

published and claimed his identification and discovery of Vibrio cholerae in

188052.

In 1861, the distribution of drinking water in a newly built water network

began in Stockholm. The distributing of larger and larger volumes of water,

led to a plan for a sewer network, presented in 1866, to manage the larger

volumes of wastewater. In Stockholm, limitations in the sewer system, did not

allow installations of WCs until 1909, where after the installation rate of WCs

exploded, in 1915 and 1936, 50 000 and 200 000 WCs had been installed

respectively53. In 1930, a plan for intercepting sewers, to collect the

wastewater from many smaller pipes, were launched and 11 years later, the

first wastewater treatment plant for central Stockholm, Henriksdal WWTP

was taken in operation. Seven years earlier, Bromma WWTP was taken in

operation to serve the western parts of Stockholm44.

Today, the Stockholm region has several WWTPs of different sizes, but the

trend has been that smaller plants are rebuilt to pumping stations, for feeding

larger plants. The reasons were mainly sharpened effluent standards, claiming

costly reconstructions in the plants or precautions for securing raw water

quality for water works in the region. The specific cost per treated water

volume is also lower in larger plants, which promotes the centralization. The

treatment plants in the Stockholm region have been extended stepwise during

the 20th century, Table 3,

Table 3: Implementation of treatment step in larger WWTP in the Stockholm region44,54,55.

WWTP Mechanical Biological Chemical

phosphorus

removal

Biological

Nitrogen

Removal

Sand

filters

Connected

population

Peak value

Bromma 1934 1966 1970 1986/97 1993 351 100

Henriksdal 1941 1970 1973 1997 1997 833 500

Loudden 1950 1969 1970 - - 50 000

Eolshäll 1961 1961 1970 - - 100 000?

Käppala 1969 1969 1969 2000 2000 516 700

Himmerfjärds-

verket

1974 1974 1974 1997 1993 322 000

Introduction | 13

The year of implementation of mechanical treatment is also the year of

inauguration of the WWTP. Noticeable is the late implementation of biological

treatment in Stockholm. Today, wastewaters from Loudden and Eolshäll

WWTP’s former catchment areas are pumped to Henriksdal WWTP and

Himmerfjärdsverket WWTP respectively.

1.3 Development of legislation on water pollution in Europe

In Sweden, the regulation from 1880 of water rights prohibited in principle

the discharge of harmful substances into water. In 1918, the law of water and

wastewater was introduced with several amendments; in 1941 concerning

judicial control, leading to the obligation of judicial review of point sources.

Further strengthening of the legislation by amendments was undertaken in

1955, after the salmonella outbreak in 1953, and in 1970 due to eutrophication.

However, the legislation was not sufficient so the Environmental Protection

Act of 1969 was launched56. Many of these laws were replaced by the Swedish

Environmental Code at 1st of January 1999: Miljöbalken 1998:80857.

In the European union, the Water Framework Directive (WFD) was adopted

in 2000 and it requires that all inland and coastal waters, within certain river

basins, must reach at least good ecological and chemical status by 2015 and it

defines how this should be achieved through environmental objectives and

ecological targets for surface waters. The goal is a healthy water environment

with environmental, economic and social considerations taken into account.

A list of priority substances, which could threaten human health or

ecosystems, was presented in 2000, with the goal to decrease naturally

occurring pollutants back to the background values and man-made synthetic

pollutants to values close to zero. This first list was replaced by Annex II of

the Directive on Environmental Quality Standards (Directive 2008/105/EC),

also known as the Priority Substances Directive, which set environmental

quality standards (EQS) for the substances in surface waters. In 2012, the

European commission proposed to include estradiol, ethinyl-estradiol and

diclofenac on the Watch list, which consist of pollutants to be monitored in

the environment for eventual inclusion on the list of prioritized substances. In

2015, the antibiotics azithromycin, clarithromycin and erythromycin were

added to the watch list58–61

In Switzerland, an extensive research on micropollutant prevalence, effects in

the environment and removal in WWTPs has been undertaken during the last

15 years. The Swiss government decided in 2014 that technical measures must

14 | Introduction

be implemented over the next 20 years. A selection of 100 WWTPs, out of the

existing 700 WWTPs in Switzerland, will be upgraded to remove

micropollutants, resulting in more than 80% removal in 50% of the total

volume of municipal wastewater in Switzerland62.

Five indicator substances are proposed in Switzerland to represent the

micropollutants in wastewater, three APIs: sulfamethoxazole, diclofenac and

carbamazepine, one herbicide: mecoprop and one corrosion inhibitor:

benzotriazole. All five can be analyzed with the same analytical method62.

The discussion above concerns mainly the concentration levels and the

pinpointed, and thereby critical, substances for removal according to the

authorities. This has obvious consequences of the selection of the treatment

technology, since different methods remove APIs with radically different

efficiency10,63–65.

1.4 Market of pharmaceuticals

About 4 000 active pharmaceutical ingredients (APIs) are used in human

and veterinary medicine worldwide and they are produced by pharmaceutical

companies to a combined mass of 100 000 tons per year66. The global market

for APIs was valued at USD 134.2 billion in the year 2015 and is estimated to

reach a value of USD 239.8 billion by 2025, growing with an average annual

growth rate of 6.0%, specifically calculated as Compound Annual Growth Rate

(CAGR)67.

In the European Union, about 3000 different APIs are used in human

medicine68 and in Sweden, approximately 1 200 APIs are authorized in more

than 13 700 human and animal pharmaceutical products69. In 2014, human

pharmaceuticals in Sweden had a total sales value of 37 829 Million SEK,

excluding VAT, corresponding to 1 728 defined daily doses (DDD) per

thousand inhabitants and day. For veterinary medicines, the total sales value

of 780 SEK millions, excluding VAT, and a total sales volume 3.21 Million

packaging. In 2014, the sales value for veterinary medicines corresponded to

2.0% of the total sales value of human and veterinary medicines, where pets

represents over 50% of the veterinary consumption70.

1.5 Routes/pathways of pharmaceuticals in the environment

APIs in the aquatic environment originate from several sources, mainly from

human use of medicines, Figure 3. The APIs or conjugated forms of the API

are subsequently excreted into urine and feces, which are transported to

WWTPs (mWWTPs in Figure 3) or OSSFs, where some APIs are removed, but

Introduction | 15

most substances remain, to different extents, in the treated wastewater, which

is discharged to surface or ground water.

In Sweden, 97% of the human medicines (reported as number of DDDs) are

consumed in non-institutional care, only 3% of the human pharmaceuticals

are used in hospital care71. The hospitals are nevertheless discussing on-site

treatment of wastewater from some clinics to remove e.g. antibiotic residues

from wastewater with high concentrations, but in Sweden today, the hospital

wastewaters are treated in municipal WWTPs.

Figure 3: Pathways of pharmaceuticals from production to the environment. Inspired by 10,72,73.

At API production sites outside Europe, discharges of untreated or partly

untreated industrial wastewater have been reported74. In a specific case, the

concentrations of ciprofloxacin in wastewater exceeded the maximum

therapeutic dose in human plasma. The concentrations of e.g. losartan,

Manufacturing of pharmaceuticals

Excretion

Distribution

Fishdebris

Humans

Pharmacies

Animals

Unused medicine

Municipal Wastewater

treatment plants (mWWTP)

Wash off

Surface water

Onsite Sewage Facilities (OSSF)

Manure

OSSF Sludge

Soil

Excretion

Manure storage /

processing

Sediment

Ground water

Landfill

Ash

Householdgarbage

Incineration700-850°C

Oceans

Incineration1100-1200°C

Incineration 750-1000°C

Drinking water

Industrial Wastewater

treatment plants (iWWTP)

mWWTP Sludge

Ash

iWWTP Sludge

Excretion Excretion

Stormwater

Farmanimals

PetsFish in

fishfarms

Excrements

16 | Introduction

cetirizine, metoprolol and citalopram were typically 1 000 higher in the

industrial wastewater, than in municipal wastewater in general74. In Europe,

industrial WWTPs (iWWTPs) generally treat the wastewater from the API

production sites.

The relative importance of veterinary medicines, as sources of APIs in the

aquatic environment, are lower in Sweden, than in some other countries.

Within the EU, it is forbidden to use antibiotics for growth-promoting

purposes in all animal farming, which also includes fish farming75. However,

in Czech Republic, Denmark, Finland and the Netherlands the usage of

antibiotic substances increased as sales per kg of animals in 2009, compared

to 2005. They are often categorized as “therapeutic” antibiotics, e.g. in

Denmark the antibiotic dosage per pig increased by 24% between 2001 and

2008 and antibiotics are used systematically to control diseases in intensive

farms and involved in mass medication76. Use of antibiotics in production of

poultry, pigs and cattle are still widespread in the US and elsewhere outside

the EU76. An estimation shows that the global consumption of antimicrobials

increased by 67% between 2010 and 203077.

The APIs of veterinary origin are excreted in pet excrements, manure or fish

debris. The pet excrements and fish debris will, to a large extent, rapidly come

in contact with the aquatic environment. The manure from farms will be used

as fertilizer on soil, where some APIs can be adsorbed and removed, mainly

by microbial activity 78. Potentially, some low adsorptive APIs will migrate to

surface and groundwater.

The collection of unused medicines is an important upstream measure to

avoid discharge of APIs into the aquatic environment. In Sweden, the

estimated volume of unused medicines is 5%. All pharmacies in Sweden take

the unused medicines in return, free of charge. The collected medicines are

incinerated at higher temperatures, than conventional household garbage,

which is incinerated at 700-850°C. The waste incineration EU directive

2000/76/EC states that the temperature by incineration of hazardous waste

containing 1% Cl must be 1100 °C during 2 seconds79. Pharmaceutical waste

must be incinerated at >1000 °C or for halogenated content >0.5% at

1200°C80. The waste of used medicines in household garbage is thus not

recommendable, but it is still better than flushing them into the sewer.

The sludge from the WWTPs is incinerated at 750-1000°C or is used as

fertilizer in agriculture. Theoretically the incineration temperatures are too

low to destruct all APIs. Like in the case of manure, APIs in sludge from

WWTPs can be adsorbed and removed, but some APIs will migrate to surface

Introduction | 17

and groundwater. APIs in treated wastewater, ending up in surface water,

might be degraded by natural UV light, but some APIs will find their way to

drinking water, since traditional treatment of surface water in waterworks, do

not include ozonation or activated carbon treatment, which otherwise would

remove them to a large extent. However, in more densely populated areas with

water scarcity or raw water with low quality, waterworks have been extended

with treatment of micropollutants, including APIs.

1.6 Reported effects of pharmaceuticals in the environment

Pharmaceutical residues have been reported to cause optically observable

adverse effects in wild organisms. Feminization of male fish, due to natural

and synthesized estrogens, was the first observations from downstream

locations of municipal WWTPs in England and was reported in the 1970s, in

a few scientific paper or reports, which increased the interest and awareness

of hormones in the environment among researchers and the public81.

Additional scientific studies were undertaken in the 1970s, e.g. of the fate of

some veterinary APIs (phenothiazine, sulfamethazine, clopidol, and

diethylstilbestrol) in aquatic model ecosystems82 and on APIs and their

metabolites as environmental contaminants83. During the 1980s, mapping of

APIs in the effluent of WWTPs continued and since the late 1990s interests

into the fate of pharmaceuticals in the WWTPs and in the environment has

accelerated81. The most important reason for the interest are the reports of

observed effects of pharmaceuticals on water-living organisms;

hermaphrodite fish in ponds, downstream of a WWTP, initiated a study with

caged fish in the effluent from 15 WWTP in England, showing dramatically

increased levels of vitellogenin, a precursor protein of egg yolk, also in male

fish84. Vitellogenin has since then been an important biomarker for estrogenic

contamination of the aquatic environment85. Concentrations of natural

steroidal estrogens in British rivers were sufficient to increase vitellogenin

synthesis observed in male fish86.

In 2004, two studies showed on effects of diclofenac on organisms at

environmental relevant concentrations. In the first study, rainbow trout

(Oncorhynchus mykiss) was observed to bioconcentrate diclofenac in liver,

kidney, gills and muscle. Based on the observed effects on kidney and gills at

a water concentration of 5 µg/L, the no observed effect concentration (NOEC)

of diclofenac was proposed to be 1 µg/L87. Further evaluation showed on

effects also in tests with 1 µg/L, proposing the NOEC to be lower than 1 µg/L88

18 | Introduction

i.e. corresponding to typical concentration in effluent at Swedish WWTPs. In

a recent paper, the low NOEC values from 2004 were disputed as the

moderately reduced observed growth-rates were interpreted as artefacts in the

new study, suggesting a much higher NOEC value of 320 µg/L89.

Another route for diclofenac in the environment was discovered on the Indian

subcontinent where the population of three spices of vultures (Gyps

bengalensis; G. indicus and G. tenuirostris) declined by 34-95% during the

period 1990s-2003 and the death of the vultures was associated with renal

failure resulting in visceral gout. Residues of veterinary diclofenac in animal

carcasses were proposed to be responsible for the decline in vulture

populations. In Pakistan, veterinary diclofenac is sold without prescription for

treatment of cattle, which after death, are left for scavengers like vultures to

remove90.

Further examples demonstrating the effects of APIs on aquatic organisms, at

environmentally relevant concentrations, show that the lowest effect

concentration of fluoxetine and ibuprofen on the activity of Gammarus pulex,

an amphipod crustacean, was 100 ng/L and 10 ng/L respectively91. The lipid

regulator Gemfibrozil was bioconcentrated in goldfish (Carassius auratus)

and the plasma concentration of testosterone was reduced by over 50% in the

fish92.

Antibacterial drugs, as environmental contaminants, were discussed in the

early 1970s93. Antibiotic substances are essential to treat bacterial infections,

but they are also used in the less requisite application on livestock in

agriculture. The extended use and misuse of antibiotics has increased the

prevalence of antimicrobial resistance (AMR) e.g. among clinically relevant

bacteria94. The spreading of AMR is a substantial threat to human health and

many initiatives and measures are in progress to decrease the spreading of

AMR and to extend the usability and life-time of the available antibiotics. No

novel classes of antibiotics have been presented to the treatment of diseases

since 1995. One reason for the lack of new classes is the decrease in budgets

for antibiotic R&D in the major pharmaceutical companies. However, creative

design of new molecules is possible within the existing antibiotic classes to

improve their therapeutic properties94,95.

Antibiotic residues in wastewater from production sites are considered to be

one source for induction of AMR74,96. In municipal WWTPs, the continuous

input of AMR bacteria from connected humans has been considered to be

much more important than the inflow of antibiotic residues. The

concentrations of antibiotics in municipal wastewater are very low compared

Introduction | 19

to therapeutic concentration, but also low compared to the concentration in

wastewater from hospitals97. In a recent study, the concentrations of

ciprofloxacin and tetracycline in the influent to three Swedish WWTPs,

exceeded the predictive threshold concentrations for resistance selection.

However, no enrichment of any particular class of antibiotic resistant genes in

the WWTPs were seen98.

Synthesized progestins, a class of contraceptive pharmaceuticals have recently

moved into focus in the field of ecotoxicology. Natural progestins are involved

as reproductive hormones in all vertebrates. In total 20 synthesized progestins

are in use in human and veterinary medicine99,100. The synthetic progestins,

used for contraception so far, are structurally related either to testosterone or

to progesterone. Several new progestins have been designed to minimize side-

effects. The most potent progestins can be used at very low doses99,101. This

indicates that low environmental concentrations could have an effect on water

living organisms. Two progestins, levonorgestrel (LNG) and norethindrone

(NET) have been identified as highly potent androgenic pollutants in the

aquatic environment, at low ng/L level100,102

A recent study showed that oxazepam altered the behavior and feeding rate of

wild European perch (Perca fluviatilis), at environmental relevant

concentrations, in effluent-influenced surface waters. The change in behavior,

in form of increased activity and reduced sociality, will probably have

ecological and evolutionary consequences, as well as that the increased

feeding rate can influence the structure of the aquatic community103.

The substances now discussed are hitherto the most important examples of

pharmaceuticals giving adverse effects at environmentally relevant

concentrations. Together with several hundred other APIs, they can be found

in the Wikipharma database containing publicly available ecotoxicity data for

pharmaceutical substances104,105.

1.7 Environmental risk assessment (ERA)

ERA is an important tool in the work chain, from research on environmental

effects, to action taken to prevent the environment from harmful substances;

Research → Risk assessment → Risk management104.

In the ERA, identification and characterization of environmental risks form

the basis for a decision of the risk connected to a specific chemical, to prevent

20 | Introduction

unacceptable harm to the environment, taking into account economic,

engineering, political and social information106.

Since 2006, environmental risk assessments are required for all new

marketing authorization applications for APIs. It includes recommendations

of using OECD test protocols for physical-chemical, fate and effects studies in

the first phase. If potential risks have been identified in the first phase, then

additional OECD tests should be performed. However, a market introduction

is not prohibited, despite detected negative impacts in the tests107.

In 2016, ten years after the release of the guidance for environmental risk

assessment of human pharmaceutical products, ten recommendations to

improvements of the assessment were published by the MistraPharma

research project team108.

1. Include substances introduced to the market before 2006

2. Requirements to assess the risk for development of antibiotic

resistance

3. Jointly performed assessments by several companies

4. Refinement of the test proposal

5. Mixture toxicity assessments on active pharmaceutical ingredients

with similar modes of action

6. Use of all available ecotoxicity studies

7. Mandatory reviews at regular intervals

8. Increased transparency

9. Inclusion of emission data from production

10. Inclusion of environmental risks in the risk-benefit analysis

The published guidelines have been presented for several stakeholders,

including the European Parliament and national water and wastewater

organizations.

1.7.1 Ecotoxicology tests of wastewater

Treated wastewater contains complex mixtures of micropollutants, raising

concerns about effects on aquatic organisms. The addition of advanced

treatment steps could for some processes contain, or potentially produce,

effluents affecting exposed organisms by known or unknown modes of

action109.

Ecotoxicological assays for studies on the effect of micropollutants have for

many years focused on organism´s individual level, with endpoints like

Introduction | 21

individual growth, reproduction and mortality. Today, the biomarker

responses on biochemical and cellular systems are frequently used, being

more sensitive and with faster response to changes in the test environment.

Many ecotoxicological biomarkers originate from biomedical sciences and

many of the mechanisms are conserved in mammals and different organisms,

allowing the biomarkers to indicate the same or similar processes on

biochemical or cellular level110,111. However, the conserved number and

similarities to mechanisms in human differ in different species e.g. zebrafish

Danio rerio has orthologs to 88% of the human drug targets, while 63% are

conserved in Daphnia pulex , 36% in green algae Chlamydomonas reinhardtii

and 19% in E. coli112.

The combination of ecotoxicological assays and studies of biomarker

responses are still valuable for development of biomarker assays and the

modelling of mechanisms at different levels in the organisms, but also in the

evaluation of the effects of individual and complex mixtures of chemicals like

the situation in municipal wastewater110.

1.7.2 Biomarkers

The term biomarker is generally almost any measurement reflecting an

interaction between a biological system and a potential chemical, physical or

biological hazard. The measured response can be functional, physiological,

biochemical at the cellular level or a molecular interaction113.

One of the most commonly used biomarkers in studies of the aquatic

environment is the induction of vitellogenin in male and juvenile fish, as a

result of exposure to estrogenic compounds. As a complement, the induction

of the glue protein spiggin is the only known quantitative, molecular

biomarker for androgenic compounds in fish. Only the male fish produces the

spiggin for nest building and a production of spiggin in female fish is induced

by exposure to exogenous androgenic substances114.

The Cytochrome P450 (CYP) monooxygenases are members of the

hemoprotein superfamily and are involved in metabolism of endogenous

compounds such as steroids, fatty acids, and prostaglandins and exogenous

compounds such as chemical pollutants including pharmaceuticals. Fish has

been reported to have 18 families of CYP genes (CYP1, CYP2, CYP3, CYP4,

CYP5, CYP7, CYP8, CYP11, CYP17, CYP19, CYP20, CYP21, CYP24, CYP26,

CYP27, CYP39, CYP46 and CYP51) and numerous subfamilies. CYPs catalyze

the conversion of lipophilic substances to more water-soluble substances

22 | Introduction

primarily by oxidation115. The mRNA expression of CYP1A is known to be

highly inducible by a number of aryl hydrocarbon receptor (AhR) agonists like

PAHs and other planar aromatic (aryl) hydrocarbons in various fish species116.

One non-target ecotoxicity test, the 1H NMR (proton nuclear magnetic

resonance spectroscopy) metabolomics of fish blood plasma can be used to

explore responses not identified by more targeted (chemical or biological)

assays109.

1.8 Evaluation of chemical data -Deconjugation and accuracy

Pharmaceutical residues appear in concentrations of ng/L to µg/L in

wastewater and they can be in form of parent (orginal) substances or

conjugated substances. For some APIs, higher concentrations are reported in

the WWTP effluent, than in the influent, in corresponding samples, at the

same WWTP. This negative removal can be a result of either deconjugation,

analytical uncertainty or improper sampling. Deconjugation means cleavage

of a conjugate of an API, with a molecule like glycine, glucuronic acid or

glutathione. The conjugates are produced in humans to facilitate excretion of

APIs with urine or bile63,117. In the WWTPs, bacterial enzymes are active in the

deconjugation, which increases the concentrations of APIs in the effluent,

compared with the influent, for some slowly degradable substances.

The analysis of pharmaceutical residues demands advanced methods, but no

standardized analytical method is available, although several methods have

been reported. Most of the APIs are present at ng/L level and considerable

analytical uncertainties follow with the low concentrations. To evaluate the

accuracy of the different in-house analytical methods, which all use SPE

sorbents, chromatographs and mass spectrometers, an intercalibration was

performed at five Swedish laboratories in 2008118. The intercalibration

showed that the distribution in results between the laboratories, for the same

API and water, is higher than the spread between replicates within the same

laboratory, which means that the laboratories perform reproducible results,

but the results are not the same, i.e. different systematic errors are built into

the individual laboratory method. The variation of the results was

significantly greater for APIs that are poorly removed, or not removed at all,

in the WWTPs. Interestingly, for APIs with poor removal (<30%), an

analytical error of 20% can lead to a calculation result of negative removal and

be mistakenly interpreted as deconjugation of API metabolites. The cause of

systematic errors lies primarily in complexity in composition of the

Introduction | 23

wastewater. Interfering particles and other substances cause a matrix effect,

particularly in samples of influent wastewater118.

Ion suppression is one form of matrix effect that negatively affects detection

capability, precision, and accuracy. Ion suppression is caused by irrelevant

substances from wastewater, often reducing the signal for the APIs, resulting

in improperly lower values. However, depending upon the type of sample, it

also can be observed as an increase in the response of the desired analyte.

Strategies have been developed to validate the presence and quantitatively

calculate the extent of ion suppression119.

Another factor that can contribute to different results from different

laboratories is the distribution of APIs between water and particles in the

samples. Some APIs tend to bind to particles and the pretreatment of the

samples, by filtration (0.45-1.6 µm) of influent wastewater, which almost all

laboratories performed prior to concentration and analysis, may have resulted

in the determination of the waterborne API component only. The filtration

also means that reported levels in WWTPs’ influent samples often are lower

than the actual levels. This in turn leads to the fact that too low removal

efficiencies over the WWTPs are calculated for certain APIs 118.

Improper sampling of a non-constant flow of wastewater results in bad raw

data. Normally a diurnal variation in substance concentration and wastewater

hydraulic flow occurs, but also mass flow of pollutants has a diurnal variation

in the WWTPs. This is also the case for APIs120–122. Grab samples from influent

and effluent wastewater, taken without regard to the hydraulic retention time

in the WWTP, will probably cause errors in calculations of removal efficiencies

due to non-correlating samples. Flow proportional 24h composite samples

normally taken at many WWTPs are more likely to be a good base for

calculations, although they are not corresponding regarding the same water

portion, but they cover the daily, often repeated, diurnal variation. Sampling

points in the WWTP must be selected so internal streams like supernatant

from the dewatering of digested sludge will not be discharged to the influent

wastewater, prior to sampling123.

1.9 Pharmaceuticals in WWTPs and the aquatic environment

The municipal wastewater treatment plants are designed to treat household

wastewater regarding suspended material (particles), easily biodegradable

organic matter, phosphorus and nitrogen and the WWTPs are not designed to

specifically remove micropollutants like pharmaceutical residues. However,

some APIs are partly or fully removed by sorption and biological degradation

24 | Introduction

or transformation in the biological treatment in the existing WWTPs64. For

this introduction, the median concentration of APIs in the influent, effluent

and the resulting removal efficiency in Swedish WWTPs were determined,

based on published data of pharmaceutical residues from different Swedish

authorities63,124–127. The Swedish Environment Protection Agency, County

Councils and municipalities have sampled the influent and effluent of many

WWTPs, to get an idea of the situation of pharmaceutical residues in

wastewater. However, only a few compilations of the raw data on

pharmaceutical residues have been done previously and then on subsets of the

data. In the following paragraphs, the results of the compilation made for this

thesis is presented and discussed.

1.9.1 Occurrence of APIs in the influent in Swedish WWTPs

Samples from the inlet to WWTPs have been taken by several authorities, e.g.

Swedish Environment Protection Agency and County Councils, to study the

concentrations and load of pharmaceutical residues on the WWTPs, but also

to enable calculation of removal efficiencies in the regular WWTPs. The

available data used in the compilation come from in total 91 Swedish WWTPs

including 85 samples from influent wastewaters and 150 samples from

WWTPs effluents63,124–127. One explanation to the lower number of inlet

samples can be that the authorities want to get an idea of the remaining APIs

in the effluent to estimate the pollution load from the WWTPs on the

environment. The reported samples were taken as grab samples or composite

samples and they were analyzed by one of a few external labs that can offer

analysis of pharmaceutical residues in Sweden.

The concentrations of API in influent wastewater differed a lot and they are

therefore presented in a logarithmic diagram to make a visual comparison

feasible, Figure 4. In the influent to the WWTPs, seven APIs had a median

concentration over 1 µg/L, whereof the painkiller paracetamol reached the

highest concentration 69 µg/l, followed by two other painkillers ibuprofen and

naproxen.

Introduction | 25

Figure 4. Median concentration in influent to Swedish WWTPs. Notation #W corresponds to the

number of WWTPs sampled and #S corresponds to the number of samples with quantified

concentrations.

Next in concentration order was furosemide, a diuretic API, followed by

another painkiller, ketoprofen. The last two substances, with concentrations

exceeding 1 µg/L, were the beta-blockers atenolol and metoprolol. The four

painkillers in “top 7” are all non-steroidal anti-inflammatory drugs (NSAIDS)

with the exception of paracetamol. The remaining concentration level groups

of APIs represent several therapeutic classes in each group. Data showed that

22 APIs had a median concentration between 100 and 1000 ng/L. In the range

10-100 ng/L, 19 APIs were recorded, and 11 APIs had concentrations in the

range of 1 to 10 ng/L. Ethinyl estradiol was the only API quantified below 1

ng/L, through the specific analytical methodology. The concentration levels

reflected the consumption and excretion of parent substance after

consumption: The DDD is for paracetamol 3000 mg and for ethinyl estradiol

0.035 mg128. The DDDs for the two APIs differs by a factor of 100 000 and the

concentration in the influent differs by a factor of 200 000, thus being in the

same range. The excretion from humans of paracetamol and ethinyl estradiol

is reported as 2-3% and 2-40% respectively129,130. In conclusion, many of the

APIs analyzed for, were found in the influent to the WWTPs, with varying

concentrations, reflecting the API’s DDD and the total use.

26 | Introduction

1.9.2 Occurrence of APIs in the effluent from Swedish WWTPs

The national data described above were compiled to evaluate the

concentrations of APIs in the effluent from Swedish WWTPs63,124–127. In

parallel with the case of API concentrations in the influent, the concentrations

of APIs in the effluent are presented in a logarithmic diagram to facilitate

comparisons, Figure 5. The effluent concentrations are related to the inlet

concentrations and the different processes in the WWTPs, including

adsorption and biological transformation as oxidation and deconjugation. The

removal of different APIs varied a lot in the existing WWTPs and combined

with the large range of influent concentrations, the effluent concentrations

showed a large range as well, however not as broad as the inlet concentrations.

The order of substances sorted in concentration magnitude has changed in

comparison with the order of substances in the influent, which indicates the

substances are removed to different degrees.

Figure 5. Median concentration in effluent from Swedish WWTPs. Notation #W corresponds to the

number of WWTPs sampled and #S corresponds to the number of samples with quantified

concentrations.

From the top 7-list of high concentrations in the influent, furosemide,

metoprolol and atenolol remained at approximately the same concentration

in the effluent, showing the low removal efficiencies in existing WWTP. Data

provided concentrations of four of the six APIs and hormones on the WFD

Introduction | 27

present watch list; diclofenac, erythromycin, 17-beta-estradiol (E2), and 17-

alpha-ethinylestradiol (EE2)58. The median concentration in the effluents for

these substances were 251, 220, 0.5 and 0.3 ng/L respectively. Proposed

annual average environmental concentrations in freshwater (marine water)

for diclofenac, erythromycin are set to 100 (10) ng/L and 200 (20) ng/L and

for 17-beta-estradiol and 17-alpha-ethinylestradiol 0.4 (0.08) ng/L and 0.035

(0.0070) ng/L. respectively, leading to the conclusion that as long that the

effluents of the Swedish WWTPs is diluted by a factor of 10 in the receiving

water, the environment concentrations will be lower than the proposed

environmental quality standards EQSs131–134.

In conclusion, many of the APIs are still present in the effluent from the

WWTPs, which shows that the removal capacity is limited in regular WWTPs

and that pharmaceutical residues are discharged into the aquatic environment

via WWTPs. The authorities use the data to map the magnitude of the

discharged pharmaceutical residues and, in a few cases, to model the

environment concentrations of pharmaceutical residues in watersheds.

1.9.3 Removal efficiencies of APIs in Swedish WWTPs and the

presence of APIs in the aquatic environment and in

drinking water

The removal efficiency describes to what extent the APIs have been removed

in a regular municipal WWTP. The statistical processing of national raw data

of influent and effluent concentrations presented above63,124–127, showed the

removal efficiencies for a selection of APIs in Swedish wastewater treatment

plants. Raw data were of different origin, from different years and analyzed at

different laboratories. WWTPs represent both small and large plants with

different process design and the removal efficiencies are calculated in the

corresponding samples. In the case of a lower concentrations in the effluent

than the limit of quantification, LOQ, the removal efficiency was calculated on

the concentration LOQ/2 in the effluent. A big difference in removal

efficiencies was observed whether the sewage treatment plant has biological

nitrogen removal or not Figure 6.

28 | Introduction

Figure 6. Median removal efficiencies of pharmaceuticals in Swedish WWTPs. Notation #W

corresponds to the number of WWTPs sampled and #S corresponds to the number of data pairs.

Nine APIs had a negative removal in the studied WWTPs due to

deconjugation, improper sampling and analytical limitations, which can

explain the results as discussed earlier. Sixteen APIs were removed more than

80%, including three of the four painkillers with high concentrations in the

influent. Paracetamol and ibuprofen were removed more than 99%. In

conclusion, the removal efficiencies in the regular WWTPs differed much

between the APIs. The average removal for all APIs was 44% and the median

removal efficiency was 52%. The compilation showed that the removal

efficiency of APIs was limited in regular municipal WWTPs and measures

must be taken if improved removal efficiencies of APIs are required.

This evaluation showed that APIs were present in regular municipal

wastewater, both in the influent and effluent waters. Thereby, there is a risk

for accumulation of APIs in the receiving water bodies and ultimately a

potential effect on the living organisms in the receiving water.

Introduction | 29

Pharmaceutical residues in the aquatic environment

More than 600 active pharmaceutical ingredients, their metabolites or

transformation products have been found in aquatic systems globally, due to

a combination of worldwide usage and low removal efficiency in wastewater

treatment plants (WWTPs), or a complete lack of WWTPs. In surface waters,

concentrations of pharmaceuticals usually range from low ng/L to low µg/L.

The concentration range shows that APIs are true micropollutants3–5,66,135.

Discharges from pharmaceutical production sites, have been shown to result

in concentrations as high as mg/L in receiving surface waters136–138.

Pharmaceutical residues in drinking water

The concentrations of pharmaceutical residues in drinking water are more

than 1000-fold below the minimum therapeutic dose, which is the lowest

clinically active dosage. The measured trace concentrations of

pharmaceuticals in drinking water are deemed unlikely to pose risks to human

health, due to the safety margin between the concentrations quantified and

the concentrations of lowest clinically active dose. Based on the current low

concentrations of carbamazepine and other pharmaceutical residue in

drinking water, WHO has concluded, that impact on human health is very

unlikely from drinking water and consequently, no WHO guideline values for

pharmaceuticals in drinking water will be formulated for the time being139.

1.10 Potential technologies for removal of pharmaceuticals

To increase the removal efficiencies of pharmaceutical residues in regular

municipal WWTPs, additional treatment measures must be applied.

Biological treatment, separation and chemical oxidation have been suggested

as main process categories for removal of pharmaceuticals from

wastewater63,64. The specific treatment technologies or unit operations were

divided into these categories in Table 4 and are further described in detail

below, to show the principles, process conditions and removal efficiencies of

pharmaceutical residues.

30 | Introduction

Table 4: Regular and advanced treatment technologies

Category Unit operation

Biological treatment

Conventional Activated Sludge (CAS)

Aerated granular sludge (AGS)

Membrane Bio Reactor (MBR)

Carrier Biofilm Systems (MBBR)

Enzyme transformation (ENZ)

Transformations by microorganisms (Algae or Fungi)

Separation

processes

Sorption:

Activated Carbon (AC)

Adsorption on Sludge

Membrane filtration:

Nano Filtration (NF)

Reverse Osmosis (RO)

Chemical oxidation

Ozonation (O3)

Ultra Violet Light + Hydrogen Peroxide (UV/H2O2) or TiO2

Ozonation + Hydrogen Peroxide (O3/H2O2)

Chlorine dioxide (ClO2)

Peracetic acid (PAA)

Potential of biological treatment for removal of pharmaceuticals

Enhanced biological degradation of pharmaceutical residues can be achieved

by increasing the sludge age in existing biological treatment or by treating the

effluent by biofilm processes (biofilm systems, trickling filters, slow sand

biofilter). Generally, the average removal of pharmaceutical residues by

biological treatment is not sufficient, but for some APIs, biological treatment

gives high removal efficiencies10,109,140,141. Processes involving enzymes, algae

and fungi are out of scope of this thesis.

Introduction | 31

1.10.1 Conventional Activated Sludge (CAS)

The activated sludge process is a system where suspended growth of bacteria

form settling flocs with incorporated organic and inorganic material. The

flocs are kept in suspension by aeration, which also supplies the bacteria with

oxygen for degradation of organic material. The aeration also strips the

produced carbon dioxide and other gases from the wastewater. The

continuous flow of wastewater transports the flocs, from the activated sludge

basin, to a subsequent sedimentation basin, where the flocs settle to the basin

bottom, from which approximately 98% is recirculated back to the inlet of the

activated sludge basin. The remaining 2% of the flocs are removed from the

activated sludge step in form of an excess sludge for further processing. The

removal efficiency of organic matter is generally very high, between 90 and

95 percent measured as BOD7 (biochemical oxygen demand for 7 days).

Adsorption of some APIs to the activated sludge contributes to the removal of

pharmaceutical residues over the biological treatment step. Furthermore,

activated sludge degrades some pharmaceutical residues and sludge age is an

essential parameter for the removal efficiency10. Aerobic sludge ages

exceeding seven days resulted in higher efficiencies for some APIs, than

aerobic sludge ages of less than three days, but still the average removal

efficiency of pharmaceutical residues was limited to approximately 50%10,142.

1.10.2 Aerated granular sludge (AGS)

The performance of the conventional activated sludge process is dependent

of sludge with good settling properties. Aerated granular sludge (AGS)

consists of biologically active granules with high settling velocities, 30-40

m/h143, in comparison with typical settling velocities of activated sludge, 0.5-

8 m/h, the latter range covers different sludge concentrations and sludge

morphologies144,145. The aerobic granules are most commonly formed in a

sequencing batch reactor (SBR) with process conditions that retain the

aerobic granules, but the setup let activated sludge flocs to be washed out of

the system146. The AGS technology enables the design of compact WWTPs

with simultaneous removal on organic material, nitrogen and

phosphorous147. In the few published papers on removal of pharmaceuticals

by AGS, the reported removal efficiencies are for some APIs inconsistent with

the removal efficiencies in conventional activated sludge148–150. The removal

efficiencies were however not evaluated in parallel at the same site and the

number of studies is limited due to the recently developed AGS technology.

32 | Introduction

The sorption of APIs to the AGS was substantial in one study, 50-85% of the

APIs in the feed ended as adsorbed substances148.

1.10.3 Membrane Bio Reactor (MBR)

The separation of the active sludge in an activated sludge processes usually

takes place in a sedimentation basin and is driven by gravity. During the past

20 years, membrane technology in form of micro- or ultrafiltration has been

used more and more for the separation of activated sludge, due to better

separation of suspended solids, also at higher mixed liquor suspended solids

(MLSS) concentration, than in conventional activated sludge. The high MLSS

concentration makes the treatment step, entitled membrane bioreactor

(MBR), compact with a smaller footprint. Increasing the MLSS in

combination of an increased sludge age has been shown to improve the

removal of pharmaceutical residues by 10-20%10,109.

The activated sludge is separated over a membrane by cross flow filtration,

which means that a small flow of particulate-free water is taken out through

the membrane surface, while a much higher flow passes the membrane

tangentially. The large flow is created by air, blown from beneath the

membrane modules. The large amount of air requires input of more electrical

energy, but low energy systems are developed. Furthermore, chemical

cleaning is needed regularly to avoid lower hydraulic capacity due to fouling

and scaling. The cleaning frequencies vary for different membrane brands

and strategies are still developed to decrease the fouling. There are two main

types of membranes: flat sheet and hollow-fibre membranes. The membrane

bioreactors are implemented in full-scale worldwide10.

1.10.4 Biofilm Systems (MBBR)

In biofilm processes, different bacteria species are attached on surfaces of a

support material like sand, crushed rock or plastic. In the biofilm, the bacteria

will decay organic material in the wastewater. When the easily degradable

organic material has been degraded, potentially nitrification of the

wastewater will take place. The attached growth makes the bacteria more

protected from variation in flow and wastewater quality than the bacteria in

many suspended growth applications. An important process parameter to

monitor in biofilm applications is the mass transport of organic substances,

nutrients and gases, to and from the biofilm. Limitation in diffusion due to

Introduction | 33

e.g. the biofilm thickness will slow down the reaction rates. Layers with

different redox potential are evolved in a biofilm, which facilitates e.g. both

nitrification and denitrification to take place. Today, the availability of

moving plastic carriers for biofilm has resulted in many applications for

removal of organic matter and nitrogen, as alternatives to activated sludge,

which for long was the dominating process for biological treatment37,38,151.

One hypothesis for biofilm systems is that the long sludge age for the attached

bacteria will facilitate the establishment of a diverse microbial community

that can degrade also slowly degradable substances, like some pharmaceutical

residues10. Recent research has shown that biofilm systems degrade some

pharmaceutical residues more efficiently than activated sludge152,153.

1.10.5 On-site sewage facilities - Prevalence and removal of

pharmaceuticals

In a Swedish study, two on-site sewage facilities (OSSFs) with soil filter bed

systems, showed higher removal efficiencies of 15 pharmaceutical residues,

than four large Swedish municipal WWTPs. On the contrary, diclofenac and

ketoprofen had significantly lower removal efficiencies in the OSSFs, than in

the WWTPs154. Another Swedish study, with pooled samples from eight soil

filter beds, showed that the removal of diclofenac was on average 87% in the

soil beds, compared to the achieved 34% removal efficiency in the reference

WWTP155.

The removal of organic substances in soil filter bed involves a biofilm process

where bacteria are attached to soil particles. A comparison of removal rates

of diclofenac and ketoprofen in activated sludge and biofilm processes,

performed in parallel in lab scale batch tests, showed that the biofilm system

removed diclofenac and ketoprofen faster than the activated sludge156.

The residence time for the bacteria in a soil filter bed is longer than in a

regular WWTP with activated sludge154. In the soil filter bed, the bacteria are

not removed, except with occasional losses of biofilm to the effluent water, in

contrary to WWTP, where the sludge age is generally controlled in an interval

of 3-25 days, depending on applied process and conditions such as

wastewater temperature. The longer hydraulic and bacteria retention time

seems to give the higher removal efficiency in the soil filter beds154. One

conclusion from the study is that OSSFs both can remove some

pharmaceutical residues better than a WWTP, but also give a lower removal

34 | Introduction

of other pharmaceutical residues. In the latter case, this leads to a discharge

of diffuse pollution of pharmaceuticals over large land areas154.

Prevalence and removal of micropollutants in OSSFs were extensively studied

in a recently ended Swedish research project, “Redmic”, where screening and

evaluation of add on methods to remove pharmaceuticals, personal care

products, pesticides, phosphorus-containing flame retardants and PFAS were

performed. In total 79 micropollutants were successfully identified whereof a

prioritized set of 20-26 substances was further evaluated in fate studies,

monitoring in watersheds and used as model substances in the development

of treatment technologies.

Concentrations of micropollutants were similar in influents and effluents of

OSSFs and WWTPs, respectively. Overall, the removal rates of

micropollutants in OSSFs and WWTPs were rather low. Removal of PFASs

and PFRs (phosphorus-containing flame retardants) were higher in the

OSSF´s package treatment and the WWTPs, while the selected

pharmaceuticals and personal care products (PPCPs) were more efficiently

removed in the soil beds in the OSSFs155,157,158.

1.10.6 Factors/processes contributing to removal of APIs in

regular WWTP

There are three main conceivable mechanisms to reduce pharmaceutical

residues in today's sewage treatment plant: Stripping to air, adsorption to

particles (sludge) and biochemical transformation (biodegradation)64.

Stripping

The loss of APIs to the air in the WWTPs depends on the distribution of

substance between air and water at equilibrium and how much air that comes

into contact with the wastewater. The distribution is described according to

Henry's law159, where the concentration in air is in equilibrium with the

concentration of the substance in the aqueous phase. The distribution of the

substance between water and air depends on how volatile the substance is and

is determined or calculated as Henrys constant, which varies widely between

different substances. Usually, APIs have very low Henry's constants (<10-5

atm, m3/mol) because the substances must remain in the human blood and

not be lost with the expired breath64.

Introduction | 35

Henry's constants for the selected substances in this introduction vary

between 5.4·10-29 (Erythromycin) and Ibuprofen 1.5·10-7 (ibuprofen) (atm,

m3/mol)160. The low values show that the studied APIs are non-volatile and

normally do not evaporate into the atmosphere in the WWTP. In summary,

no or very little stripping of APIs occurs in wastewater treatment plants64.

Adsorption to sludge

Adsorption of APIs to different types of sludge in the WWTPs, followed by

separation of the sludges from the wastewater, is for some non-biodegradable

APIs an important factor for removal.

Adsorption of pharmaceutical residues onto particles occurs partly to primary

sludge, in the primary sedimentation, and partly to the activated sludge, in

the biological treatment. When adsorption occurs, it is relative fast and

supposed to reach equilibrium during the retention time in the respective

treatment steps in a sewage treatment plant. Adsorption can normally be

described by one of three models: linear (Henry´s), Freundlich or Langmuir

isotherms with the corresponding distribution coefficients for linear (Kd),

Freundlich (Kf) and Langmuir (KL) isotherms. In a study of sorption of 75

APIs in municipal wastewater onto primary sludge and secondary sludge,

sorption was predominantly best described with linear isotherms, followed by

the Freundlich isotherms161. This was valid for secondary sludge with either

short or long sludge age.

Adsorption with a linear isotherm means that the mass of a substance

adsorbed per unit of sludge is linearly proportional to the equilibrium

concentration in the solution. The sorption coefficients Kd for a selection of

APIs to different types of sludge shows a broad range, Table 5. A high value

for the sorption coefficient Kd means high adsorption. For example,

ciprofloxacin has ten times higher adsorption to excess sludge, than to

primary sludge. Diclofenac has opposite properties, that is, lower adsorption

to excess sludge, than to primary sludge.

The most lipophilic substances, with high log Kow (Octanol water distribution

constant) values, are hypothesized to adsorb more both onto primary and

excess sludge, like e.g. clotrimazole does, Table 5.

36 | Introduction

Table 5: Sorption coefficients, Kd, in primary and excess sludge for a selection of APIs

and their octanol water distribution constants, log Kow161,162

API Kd primary sludge

[L/gSS]

Kd excess sludge

[L/gSS]

Log Kow

Ciprofloxacin 2.6±1.6 26±7.3 0.2

Clotrimazol 32 34 6.1

Cyclophosphamide 0.055 0.0024 0.8

Diazepam 0.044 0.02 2.9

Diclofenac 0.46 0.016 4.3

Naproxen 0.013 3.2

Norfloxacin 2.5±1.5 37±13 1.0

Paracetamol <0.001 0.5

Sulfametoxazol 0.32 0.30 0.5

However, no simple correlation between adsorption and lipophilicity,

measured as log Kow, can be seen among the selected substances in Table 5,

where e.g. ciprofloxacin is readily adsorbed in sediments and onto primary

and excess sludge10,163. However, replacing log Kow with log Dow (commonly

log D, the distribution coefficient) the correlation between hydrophobicity i.e.

lipophilicity and removal efficiency is improved, since log Dow takes the

distribution of ionized and un-ionized molecule into account164.

Lower pH in the wastewater provides higher adsorption onto sludge for many

substances since acidic condition is preferable for the removal of the acidic

pharmaceuticals165. However, in the relevant pH range, 6-8 for municipal

wastewater, no significant effect on the removal via sorption was observed,

even though the sorption coefficients, Kds, were significantly different in the

pH range 6-8161.

Biochemical transformation – biodegradation

The biological treatment is the most important treatment step in the existing

WWTPs for removal of pharmaceutical residues. For biologically readily

degradable APIs, the removal is primarily depending on the sludge age. An

important conclusion from the comprehensive EU project Poseidon was that

the biological degradation of APIs normally follows a pseudo first order

kinetics and depends both on the concentration of the substance and the

microbial composition of the sludge, as in turn depends on the sludge age.

Introduction | 37

Another conclusion was that sludge age, the number of biological reactors in

series and the dilution of effluent wastewater with stormwater, are important

parameters that affect the removal efficiency in the biological treatment162.

Impact of biological nitrogen removal (BNR)

Data from the Swedish WWTPs were also evaluated on the dependence of

biological nitrogen removal (BNR) on the removal efficiencies of

pharmaceutical residues. One key hypothesis is that high sludge ages in the

activated sludge, like the situation in BNR systems, are beneficial for the

removal of some APIs. BNR also implies longer hydraulic retention time for

the wastewater in the biological treatment, which is beneficial for slowly

degradable substances. A comparison of removal efficiencies for a selection

of APIs in plants with and without BNR showed that the median removal

efficiency was 24% higher in BNR plants, than in WWTPs without biological

nitrogen removal63, Table 6.

Table 6. Comparison of removal efficiency of APIs in WWTPs with and without biological

nitrogen removal (BNR)63.

Pharmaceutical Removal

efficiency in

WWTPs with

BNR [%]

Number of data

pairs (N) used

in calculations

Removal

efficiency in

WWTPs with-

out BNR [%]

Number of data

pairs (N) used

in calculations

Atenolol 27 12 -45 4

Citalopram -48 9 -104 3

Dextropropoxyphene -124 11 -149 3

Diclofenac 11 24 12 20

Ethinyl estradiol 36 13 74 3

Furosemide 11 12 7 4

Hydrochlorothiazide -13 12 -64 4

Ibuprofen 91 25 77 19

Ketoprofen 58 26 42 22

Metoprolol -17 17 -41 7

Naproxen 78 26 54 19

Oxazepam -18 12 -16 3

Sertraline 27 8 -20 2

Sulfamethoxazole 48 18 64 2

Median 19 13 -5 4

38 | Introduction

The negative removal efficiencies reported in Table 6, reflect higher

concentrations in the effluent, than in the influent to the WWTPs, which

probably is an effect of deconjugation of parent substances in the WWTPs.

The amount of data on pharmaceutical residues is limited for Swedish non-

BNR plants, in some cases only two to three samples, so the compilation just

gives an indication of differences between WWTPs with or without biological

nitrogen removal. However, also other studies have shown on the benefit of a

higher sludge age of removal of pharmaceuticals. Although with some

exceptions, one study showed that for WWTPs operated at SRTs higher than

10 days, low effluent concentrations of many micropollutants was achieved140.

The removal efficiency of APIs increased by 40%, from 50% to 70%, at an

extremely high sludge age of 75 days in a membrane bio reactor MBR,

compared to a conventional activated sludge process, with the normal sludge

ages of 5-15 days. A higher sludge age than 75 days did not improved the

removal further10. The increased removal efficiency by an increased sludge

age is probably caused by a change in the heterotrophic microbial community,

rather than an increase in the amount of nitrifying bacteria in the BNR

plant153. The relative contribution of the adsorption onto sludge and the

biochemical degradation to the total removal differs between APIs, which can

be estimated by mass flow analysis over the WWTP10,161.

Biofilm processes

Biofilm processes like suspended carriers are hypothesized to achieve higher

removal of APIs compared to activate sludge in BNR plants due to a longer

retention time for the attached bacteria in biofilm systems10,153. In parallel lab

scale batch test, carrier biofilms showed considerably higher removal rates

per unit biomass for several pharmaceuticals, such as ketoprofen, and

diclofenac, than activated sludges. However, the causes of this are not

known156. A comparison in lab and full-scale of carrier biofilm and activated

sludge originated from the same WWTP showed that the removal capacity

differed for several compounds, with higher removal efficiencies for many

micropollutants including diclofenac, mefenamic acid and trimethoprim,

what showed significantly higher removal efficiencies with the carrier biofilm

compared to different types of activated sludge systems in the examined

WWTP152.

Introduction | 39

Potential of separation processes for removal of

pharmaceuticals

Several types of sorbents (activated carbon, minerals and molecular

imprinted polymers) have characteristics that make them suitable for

removal of APIs10,166,167. While there are relevant sorbents with affinity for

most APIs, a common problem with their use is that the full sorption capacity

cannot be used, due to fouling of the materials and loss of sorption capacity

to irrelevant organic molecules present in the treated wastewater168.

Membrane filtration is a process with separation of particles or solutes over a

semi-permeable membrane. In descending order of membrane pore size, the

treatment is either called microfiltration (MF), ultrafiltration (UF),

nanofiltration (NF) or reverse osmosis (RO). Increasing hydraulic pressure

must be applied with the decreasing pore size to separate the substances over

the membranes, which demands more electric energy for pumps, but also

influence the choice of different types of pumps. MF and UF are applied in

membrane bioreactors (MBRs) to separate activated sludge from treated

wastewater. The pore sizes of micro- and ultrafilters are much larger than the

APIs physical molecular sizes. Tight nanofilters (150 Da) and especially

reverse osmosis membranes, have smaller pore sizes than the physical

molecular sizes of all APIs available. NF and RO are widely applied in

drinking water preparation, to remove salts and small molecules, including

trace pollutants from contaminated raw waters and seawater. MF and RO are

also used in water reclamation plants, treating effluents from WWTPs though

after months in reservoirs10,81,169,170.

1.10.7 Activated Carbon (AC)

Adsorption with activated carbon (AC) is a widespread and general

technology to remove organic substances from fluids and gases. The

adsorption capacity of activated carbon is high for many organic substances,

due to the large porous surface, which varies in the range of 500-1500 m2/g.

AC is produced from a variety of raw materials such as coal, anthracite,

lignite, wood and coconut shells. The production of commercial AC products

involves carbonization and activation. The carbonization process includes

drying, heating, pyrolysis and carbonization, within a temperature range of

400–600° in an oxygen-deficient atmosphere. The activation is normally

achieved thermally by the use of oxidation gases, such as steam at above

40 | Introduction

800°C or carbon dioxide at higher temperatures43. During the activation,

pores of different sizes are formed, and these are divided into three groups,

depending on the sizes of the pore openings: micropores < 2 (3.5) nm,

mesopores in the range of 2 (3.5)- 50 nm and macropores >50 nm. Most of

the adsorption takes place in the micropores, due to Wan der Waals forces.

The values within parentheses relates to the largest distance between two

crystalline planes in the microcrystallites, the main constituent in activated

carbon43,171 and should be a more correct dividing limit value between

micropores and mesopores, than the often stated 2 nm. Two main forms of

activated carbon are used in water and wastewater treatment: powdered

activated carbon (PAC) and granular activated carbon (GAC)43. GAC has a

larger particle diameter, ranging from 0.2 to 5 mm compared to PAC with

particle diameters < 0.2 mm, typically 0.015–0.025 mm. A large specific

surface area of an activated carbon comes from a high distribution of

micropores and small mesopores, which was shown to correlate well with a

high average removal of organic micropollutants172.

A high solute hydrophobicity (log D), was well correlated with higher

adsorption of solutes to AC, which has a predominantly hydrophobic

surface. In addition, polarizability, aromaticity and the presence of H-

bond donor/ acceptor groups will influence the adsorption onto activated

carbon. These adsorption mechanisms will occur in parallel to different

extent, depending of solute and AC characteristics168.

Like in membrane processes, remaining bulk organic substances in the

wastewater will adsorb onto the AC surface, giving it a negative charge,

which favours the adsorption of positively charged substances. In a study

of adsorption of PFOS and PFOA, the adsorbed organic bulk material

responsible for the negative charge, had a MW < 1000 Da173. Like for the

application of oxidation methods for removal of APIs, it is important that

the feed of wastewater to the activated carbon contains low concentrations

of particles and bulk organic substances. Otherwise, the consumption of

activated carbon becomes unnecessarily high and the GAC filters will clog

with accumulated particles.

Introduction | 41

1.10.8 GAC granular activated carbon

Treatment with GAC is normally performed in fixed filter bed units, in smaller

applications often with two filters in series, and in larger plants with several

filter lines in parallel, to meet the hydraulic capacity demand in the plant. In

larger plants, normally only one filter will be installed as a single step, with

no succeeding filter. The filter bed is normally 1-3 m deep and must be

backwashed regularly to remove accumulated substances that otherwise clog

the filter. The filters are filled with a proven GAC product i.e. a product that

has shown to adsorb an adequate set of substances listed for removal. GAC

products are available in different average particle sizes, in the range of 0.6 to

3.0 mm.

To supervise GAC filter performance, the position of the unsaturated

adsorption zone or (average) mass transfer zone (MTZ) is followed to

determine when the GAC bed must be exchanged. Organic substances in the

municipal wastewater have different concentrations and adsorption

characteristics, meaning that they will travel through the GAC bed with

different retention times.

Depending on the operational target level of removal efficiency, the exchange

of a saturated GAC bed will be done after a varying number of thousand

passed bed volumes. The bed volumes (BVs) are normally calculated based on

the empty filter volume of filled GAC, meaning the volume that the GAC

product later will occupy. The hydraulic retention time in the GAC bed is also

calculated based on the BV, giving the useful parameter empty bed contact

time (EBCT). At the time of GAC exchange, it is only the first of the two filters

in series that will be exchanged. The second filter will be put first in series to

be fully saturated it before exchange, all for economic reasons.

The spent GAC will preferable be removed and transported to a regeneration

plant, where thermal treatment for 30 minutes will volatilize and degrade the

adsorbed substances by oxidation at 700-1000°C, followed by reactivation in

steam and carbon dioxide. In the regeneration process, 2-10% of the GAC is

lost and must be replace by virgin product. The GAC product can be

regenerated five times before it is finally wasted and incinerated. The

possibility to regenerate used GAC, gives an economic advantage compared

to the handling of used PAC, which must be incinerated without regeneration

and reuse171,174,175.

42 | Introduction

1.10.9 Powdered activated carbon (PAC)

PAC has a smaller particle size than GAC and the average particle sizes of PAC

products are in the range of 0.01 to 0.03 mm. Treatment with PAC is either

made in existing or additional treatment steps in the WWTPs. The main parts

in PAC treatment is a mixing tank for wastewater and PAC slurry, contact

tanks, a sedimentation tank and in some application an additional separation

unit: a sand filter or an ultrafilter. The contact tanks are completely mixed to

ensure effective contact between PAC and organic solutes in the water. The

conditions for adsorption of organic solutes onto PAC is closer to steady-state

than in GAC applications, due to smaller particle sizes and shorter diffusion

pathways. In the sedimentation tank, and in the sand filter, the main fraction

of the added PAC is accumulated. To better utilise the dosed PAC, and come

closer to adsorption steady-state, the separated PAC in the sedimentation and

in the sand filter is recycled back to the contact tanks. This internal

recirculation will prolong the solid retention time for PAC in the system and

increase the PAC concentration in the contact tanks176–178. The spent,

separated PAC is discharged to sludge handling or to a separate handling of

PAC, for later incineration. PAC is not recovered or reactivated, due to the

handling difficulties and economic considerations. The implementation of

PAC in existing treatment trains is easy and has low capital cost43,174. The cost-

effective implementation of PAC in existing WWTPs involves the utilisation

of the activated sludge stage as contact tank with relatively long retention time

for PAC and organic solutes. The spent PAC will mix with the activated sludge

and follow the excess sludge to the sludge treatment. In central Europe, with

sludge incineration, the adsorbed solutes and PAC itself will be incinerated,

together with other organic substances. In Sweden and some other countries,

where recycling of anaerobically digested sludge to farmland still is in

practice, the PAC dosing to activated sludge would ruin the sludge quality and

rule out recirculation of sludge to farmland10.

1.10.10 Nanofiltration and reverse osmosis

The moderate to high pressure-driven tight nanofilters (150 Da) and

especially reverse osmosis membranes have been evaluated in pilot scale

studies for their potential for removal of APIs in effluents from municipal

WWTPs. In a South Korean study, fourteen APIs, six hormones, two

antibiotics, three hygiene products and a flame retardant were removed more

than 95% by an RO system179. The overall removal efficiencies for

Introduction | 43

micropollutants, including 11 APIs in a full-scale water recycling plant with

MF and RO in Australia, were above 97%, with the exception of 90% removal

of bisphenol, with a residual concentration of 0.5 g/L in the final recycled

water, which in itself is a serious challenge for process operators180.

The importance of an appropriate choice of NF membrane is demonstrated

by a study of the removal of APIs with MWs in the range of 151-791 Da, over

an NF membrane with a MWCO of 600 (±200) Da. The removal efficiency

varied between 30–90% except, for naproxen (MW 230 Da), which had lower

than 10% removal efficiency181. The membrane MWCO must be in the same

order as the lowest MW of the APIs expected to be removed from the

wastewater. In another study with an NF membrane with MWCO of 150 Da,

the removal efficiency of six APIs, all within a relative narrow range of MW

(214.6 to 296.1 Da), varied between 68% and 95%169. In a test with two NF

membranes in parallel, one tight NF90 (considered to have a MWCO of 200

Da) and one loose HL (considered to have a MWCO of 150-300 Da), the

average removal efficiency of eight antibiotics in the MW range of 204-460

Da was 99 and 88% respectively. The loose NF membrane retained smaller

antibiotics molecules less effective. The rejection of the examined tight NF

membrane was remarkably high182.

By all these examples, it can be concluded that RO, and in some cases NF, can

yield very high removal efficiency, 95-99%, of APIs in municipal wastewater.

Theoretically an NF with low MWCO should have a higher removal efficiency

than an NF with higher MWCO, but in the range 150-300 Da. However, the

comparisons above shows the uncertainties for the proportional dependency

of MWCO and observed removal efficiency.

The rejection of organic pollutants in NF and RO depends on three major

interactions of the organic solute and the membrane: 1) steric hindrance, 2)

hydrophobic interactions, and 3) electrostatic interactions. The grade of steric

hindrance depends on the molecule size, in comparison to the pore size of the

membrane. Hydrophobic interactions mean adsorption of hydrophobic

substances onto and into the membrane matrix, which also can facilitate

migration of some hydrophobic substance to the permeate, but the influence

of hydrophobicity decreases with increasing molecular size in relation to

molecular weight cut off (MWCO). Electrostatic interactions are important

for rejection of charged organic solutes, due to the negatively charged

membrane surface, that repulse negatively charged organic solutes from the

membrane surface and prohibit passing to the permeate. Furthermore,

44 | Introduction

positively charged solutes are enriched close to the membrane surface,

facilitating passing of negatively charged molecules, which will lower the

rejection183–185.

Nanofiltration uses less tight membranes than reverse osmosis, but as shown

above, NF can have a comparable removal efficiencies of APIs, and the use of

the less tight NF membranes can reduce the energy costs for pumping by 25–

50%, compared to RO186,187. Still, the energy consumption compared to

ozonation is estimated to be 2 and 3 times higher for NF and RO respectively,

at high VRFs of 20-60187,188.

Retentate handling

Appling NF and RO for removal of pharmaceuticals in wastewater lead to the

question of retentate handling. The membrane technologies concentrate APIs

in the retentate, but the APIs will not be eliminated. NF and RO separate

much more compounds than pharmaceutical residues, especially inorganic

salts when applying RO. NF retains only multivalent inorganic ions, not

monovalent inorganic salts, which makes NF a good alternative to RO, as long

as the removal of APIs is sufficient189. The salt enriched retentates demand

treatment so the effluent standards of APIs for a WWTP can be fulfilled. In

water reclamation plants, normally no effluent standards for APIs are set for

the wastewater, which lead to a disposal of retentate with enriched API

concentrations to a river or the sea. However, the retentate from RO units in

wastewater reuse plants, can be treated with advanced oxidation processes

(AOPs), in combination with a biologically active sand filters190.

In a WWTP, extended with NF or RO for removal of pharmaceuticals, 90-95%

of the APIs might be concentrated in the retentate, which must be treated to

fulfil the effluents standards, therefore a reasonable small volume of retentate

is required for an affordable further processing. VRF 20 is an accessible

volume reduction factor for NF and RO applied for wastewater treatment,

corresponding to 5% of the inflow ending up in retentate volume. At VRF 60,

the retentate volume would correspond to 1.7% of the inflow to a WWTP. This

latter volume of retentate is in the same order of magnitude as the volume

of supernatant from digested and dewatered sludge, which constitutes

0.5% – 1%, of the incoming flow to the WWTP123,191.

Introduction | 45

Potential of chemical oxidation for removal of pharmaceuticals

Advanced oxidation processes (e.g. UV/H2O2, H2O2/O3 and UV/O3) and

selective oxidation reagents (O3) have been used to oxidize APIs remaining in

traditionally treated wastewater10,192–196. The alternative oxidants have

different power of oxidation. A comparison, where chlorine gas is used as the

reference, shows that ozone has 1.5 times the oxidation power compared with

chlorine, Table 7. Hydroxyl radicals have the strongest oxidation power and

they have been tested to remove pharmaceutical residues in a system with UV

and hydrogen peroxide where they are produced, to a larger extent than by

ozonation. Furthermore, ClO2, H2O2, PAA have been tested to remove APIs,

see below.

Table 7: Relative oxidation power related to chlorine197.

Oxidizing species Relative oxidation power

Chlorine dioxide (ClO2) 0.94

Chlorine (Cl2) 1.00

Hydrogen peroxide (H2O2) 1.31

Peracetic acid (PAA) 1.34

Ozone (O3) 1.52

Atomic oxygen (∙O) 1.78

Hydroxyl radical (∙OH) 2.05

1.10.11 Ozonation - oxidation of substances in wastewater with

ozone

Traditionally ozonation, i.e. injection/introduction of ozone (triplet oxygen,

O3) into water, has been used in drinking water production for disinfection,

taste and odour control and to oxidize iron, magnesium, cyanide, phenol,

benzene, chlorophenol, atrazine and other pollutants. Applications on

wastewater are also know from water reclamation plants, fed with treated

wastewater170,198,199.

Ozone is a highly toxic and reactive, with blue colour and intense odour.

Ozone is a relatively heavy gas with a density of 2.14 kg/m3, compared to air

with a density of 1.29 kg/m3 at T = 0 °C, p = 1 atm. The odour threshold is

46 | Introduction

about 0.02 ppm and the maximum daily exposure level during 8 h is set to

120 µg/m3, approximately 0.05 ppm in the European union (European

Commission, 2016). In Sweden, the maximum daily 8 h exposure level is set

to 0.1 ppm and the 15 minute exposure level is set to 0.3 ppm200. Acute effects

of ozone exposure are headache, dry throat and mucous membranes, and

irritation of the nose. Higher concentrations of ozone can also cause delayed

lung oedema, cough, choking sensation and other illness. Chronic exposure

symptoms are similar to acute exposures, with lung function decrements,

asthma, allergies. Tumorigenic, direct and indirect genetic damage have also

been found in animal and human tissue studies. However, ozone is also used

in medicine by ozone-therapists, despite difficulties in publishing studies in

peer-review journals, which have limited the knowledge in official

medicine201.

Ozone is unstable and must be produced on site. The leading production

method is to lead pure oxygen (95-99%), through an ozone generator with a

chamber with a high-voltage electrical field, where up to 20% percentage of

the oxygen is converted into ozone. In the electrical field, electrical energy

excitation of oxygen results in the formation of oxygen radicals, which in turn

react with oxygen, forming two ozone molecules from three oxygen

molecules. A portion of the formed ozone gas is converted into oxygen again,

releasing energy in form of heat, which demands water cooling of the ozone

generator. UV or ultraviolet ozone generators are usually not used in

demanding ozone applications, due to low production of ozone with relatively

low concentrations. The life span of the UV lamps is limited and they must be

changed yearly10.

Treatment with ozone requires electric energy for on-site oxygen generation,

ozone production (including cooling pumps) and for destruction units of

residual ozone after treatment.

Ozone reacts with substances in two different ways, by indirect and direct

oxidation, leading to substance specific, but mostly unknown oxidation

products and the reactions are controlled by different kinetics. The indirect

reaction pathway involves radicals to form secondary oxidants such as

hydroxyl radicals (•OH) which are formed in the decay of ozone, which is

accelerated by initiators, e.g. OH- ions. The •OH radicals react non-selectively

with a number of organic compounds in water and are essential in

degradation of persistent substances 192,202. In the direct oxidation, ozone

reacts and splits an unsaturated bond, due to its dipolar structure. Ozone

reacts also with aromatic compounds, faster with aromates with electron

Introduction | 47

supplying substituents, such as the hydroxyl group and amino group, and

slower with un-activated aromatic compounds 203,204.

Ozone in sufficient quantity and under the right process conditions can

oxidize many organic compounds. In WWTPs, with typical water temperature

of 10-20°C and pH 7, it is necessary to apply very high doses of ozone to

oxidize all substances to carbon dioxide and water (and nitrate, phosphate

and sulphate). Often the oxidation stops at the carboxylic acid level. The

ozone dose must be selected so especially micropollutants are eliminate or

transformed, while preferable the oxidation of bulk organic substances is

limited. The ozone treatment step should be installed after the regular

biological treatment, where after the concentrations of organic substances are

low, to avoid excessive use of ozone. Oxidation of the APIs clofibrinic acid,

ibuprofen and diclofenac occurs within the first few seconds after addition of

ozone205. In the diclofenac molecule, ozone attacks the amino group and, in

amoxicillin, in addition to the attack on the amino group, benzene rings and

sulphur atoms may be possible targets for ozone. The conditions in the water,

for example the pH, can influence which functional group that reacts with

ozone203. By-products formed after selective reaction with ozone, react

further with OH radicals, which causes a chain of reactions. Therefore, only

small doses of ozone are usually required for oxidation. The oxidation with

ozone, and its reaction with functional groups in the APIs, usually results in

the elimination of the pharmacological effect e.g. the estrogenic effect of 17α-

ethinyl estradiol is reduced proportionally to the added ozone dose193.

1.10.12 Application of ozonation in water and wastewater

treatment

Ozone as a disinfectant of water was documented by de Meritens in 1886 and

after pilot plants tests in Germany and a series of discoveries and

development of equipment in Germany and France the first full-scale plant

for drinking water treatment was taken in operation in 1893 at Oudshoorn in

the Netherlands. Many full-scale ozone installations were undertaken until

1914, thereafter the development of inexpensive production of chlorine for

chemical warfare opened up for use of chlorine in disinfection of water206.

During the first half of the 20th century, ozonation was used both for

disinfection and taste and odour control. In the 1960s, ozonation plants for

oxidation of iron and manganese were built, but also ozonation plants for

48 | Introduction

oxidation of phenolic compounds and pesticides in drinking water. In the

1970s, with the discovery of trihalomethanes (THM) formation by

disinfection with chlorine, the market for ozonation plants for disinfection of

drinking water increased207. In 2003, water scarcity in Singapore forced

recycling of wastewater to produce drinking water for the public. The

multibarrier concept for preparing drinking water is named NEWater, where

ozonation is used mainly for disinfection of wastewater170.

In 2014, the first municipal full-scale WWTP with ozonation for removal of

pharmaceutical residues was taken in operation in Switzerland. The Neugut

WWTP in Dübendorf has a design capacity of 150 000 population equivalents

(PE) at the present load of 105 000 PE. The overall removal efficiency of five

indicator substances was over 80%, which was also level of the effluent

standard208. This first full-scale plant constitutes an important step towards

a broader implementation of treatment steps for removal of pharmaceutical

residues and it will also serve as a reference plant for other installations,

probably worldwide. In 2015, Sweden’s first full-scale plant for removal of

pharmaceutical residues was taken in operation in Knivsta (Björlenius,

unpublished results). In the latter case, the knowledge that a full-scale

implementation of ozonation has been done in Switzerland, confirmed and

supported the ongoing work with the ozonation plant in Knivsta.

1.10.13 Ultraviolet Light and Hydrogen Peroxide (UV/H2O2)

The UV/ H2O2-process is an advanced oxidation process (AOP) and therefore

involves production of reactive radicals, particularly the OH radical,

favourable for oxidizing organic contaminants in water. Hydroxyl radicals,

one of the most powerful oxidants, Table 5, are theoretically easiest produced

by cleavage of hydrogen peroxide by ultraviolet radiation. Unfortunately,

hydrogen peroxide adsorbs UV-light poorly and therefore the concentration

of hydrogen peroxide must be high in the wastewater to generate sufficient

levels of hydroxyl radicals. In addition, the UV/H2O2 may involve direct or

photosensitized photolysis of the organic contaminants. The UV/H2O2

process is affected by radical scavengers and competitive UV absorbers in the

wastewater, limiting the rate of oxidation, e.g. bicarbonate ions are

scavenging the hydroxyl radicals and nitrate ions are shielding the UV-

radiation209,210.

Introduction | 49

UV light is produced by UV lamps mounted in pipes, where wastewater

passes. If the wavelength of the produced UV light is within the range of 200

to 280 nm it is called UVC, and it is in this wavelength range, disinfection of

bacteria occurs, and hydroxyl radicals are formed, which in turn, degrade

organic compounds. In the water and wastewater applications, the produced

UV light is either monochromatic with a single wavelength, often 254 nm, or

it has a wider spectrum with several wavelengths. UV light alone can cause

disinfection, but UV light gives generally a limited removal of APIs in

wastewater. However, photolysis by means of UV-light in sunlight can be

significance for degradation of some pharmaceuticals in the aquatic

environment, e.g. carbamazepine, diclofenac and ketoprofen211–216.

The removal efficiencies for several APIs are high in UV/H2O2 systems.

Diclofenac was removed to 95% by UV/H2O2 treatment after 90 minutes

treatment. In a parallel setup, ozone removed 95% after 10 minutes retention

time217. Paracetamol was mineralized to carbon dioxide, up to 40 % in a

UV/H2O2 system in comparing study with an ozonation system, which

achieved 30% mineralization to carbon dioxide. Many intermediates and

products were identified for both systems218.

The removal efficiency of naproxen in wastewater exceeded 90% after 3

minutes in a UV/H2O2 process219. The dosage of hydrogen peroxide is

normally in the range of 3-30 g/m3 of wastewater and the demanded contact

time ranges from 3 to 90 minutes. The pathway of UV light in wastewater is

short, just a few centimetres, depending on the turbidity in wastewater. The

design of the UV reactor is therefore important to enable good contact

between UV-lamps and the wastewater to get high removal efficiencies of

APIs. The UV-radiation doses range from 70 to 400 Wh/m3 10.

In summary, UV/H2O2 systems remove many APIs, but the hydraulic

retention time is longer, and the energy consumption is higher, than in

ozonation systems. The addition of H2O2 and the periodical exchange of UV-

lamps increase the cost further. The scaling reduces the UV transfer and too

high turbidity is critical for the application of UV/H2O2 in wastewater

treatment.

50 | Introduction

1.10.14 Combination of UV/TiO2

Under UV exposure of a solid phase catalyst, generally titanium dioxide

(TiO2). The catalyst adsorbs UV light and hydroxyl radicals are formed from

water molecules or hydroxyl groups, adsorbed onto the catalyst surface. The

formed radicals react with organic substances by redox reactions220–223. One

benefit of using TiO2, instead of a continuous dosing of hydrogen peroxide, is

the lower running cost for the treatment. UV in combination with titanium

dioxide (TiO2), has shown promising results for e.g. for diclofenac, with 95%

removal efficiency, also at high diclofenac concentrations224. Photocatalysis

with TiO2 was more efficient than photolysis alone, in the degradation of 17β-

estradiol, estriol and 17α-ethinyl estradiol in water225.

1.10.15 Combination of hydrogen peroxide and ozone (H2O2/O3)

The addition of hydrogen peroxide (H2O2) to an ozonation step can enhance

the production of OH radicals and consequently increase the potential for a

higher removal of APIs. In a lab scale study, the addition of H2O2, up to 1.3

g/m3 wastewater, had a limited impact on the removal efficiency of APIs, but

the addition reduced the necessary hydraulic retention time, which is

advantageous in full-scale applications. In another study, the elimination rate

for cyclophosphamide and the APIs building block quinoxaline-2-carboxylic

acid, increased by 26% by an addition of H2O2, corresponding to 2.3 g/m3

wastewater226–228.

1.10.16 Chlorine dioxide (ClO2)

Chlorine dioxide is used in drinking water production for disinfection and for

taste- and odour control. Typical doses of ClO2 for taste and odour control, or

for disinfection, may be in the range of 0.07–2 g/m3 229. The disinfection

efficiency in water and wastewater is comparable to use of chlorine, but

chlorine dioxide is a better alternative, since it limits the formation of organic

disinfection by-products e.g. chloramines, and trihalomethanes230. However,

chlorine dioxide applied in water treatment is converted to chlorite and

chlorate, which is inorganic disinfection by-products formed from chlorine

dioxide decay229. Toxicity of chlorine dioxide, chlorite and chlorate ions to

some species e.g. rainbow trout and brown algae has been reported231,232.

Chlorine dioxide is a strong oxidant over a broad pH range230. Thus, oxidation

of APIs with chlorine dioxide has been proposed as an alternative to

Introduction | 51

ozonation, due to high removal efficiencies at low doses of ClO2 for some APIs,

e.g. 17α-ethinyl estradiol, diclofenac, roxithromycin, sulfamethoxazole and

mefenamic acid233,234. However, clofibrinic acid and ibuprofen were not

removed when wastewater was treated with ClO2 doses up to 20 g/m3.

Extended studies showed on a low degree of reactivity using ClO2, with an

average removal efficiency of 32% for 56 APIs tested, even when the ClO2 dose

was increased to 20 g/m3. APIs with electron-withdrawing functional groups

appear to be more resistant to the ClO2 oxidation.

The high removal efficiency for 17α-ethinyl estradiol, at a low dose of chlorine

dioxide, was explained by ClO2’s high reactivity to phenolic moieties. Higher

removal efficiency of APIs was achieved treating effluent wastewater from a

BNR process, holding a relatively low COD concentration, 30 g/m3, than

treating effluent from a WWTP without BNR, which produced a wastewater

with a COD concentration of 55 g/m3. This shows the importance of a low

content of bulk organic substances in the wastewater to be treated, to reduce

the consumption of oxidation agents202,235. The residual concentrations of

ClO2 in wastewater, after different doses of chlorine dioxide, were low, up to

an added ClO2 dose of 10 g/m3 235. In summary, ClO2 will only be effective to

oxidize certain APIs. Reactions with ozone are generally faster and ozone

reacts with a larger number of APIs than ClO2 does202,235.

1.10.17 Peracetic acid (PAA)

Peracetic acid (PAA) is a strong disinfectant, with bactericidal, viricidal,

fungicidal, and sporicidal effects shown in industrial applications, and PAA

has been discussed and used in disinfection of wastewater effluents in full-

scale236,237. PAA for wastewater disinfection is relatively simple to implement,

due to low investment costs. Additional advantages are the absence of toxic

or mutagenic residuals or by-products, a small dependence on pH and the

need for only a short contact time. PAA can be used for disinfection in both

primary and secondary effluents and no quenching of PAA residues is

required. Disinfectant doses range from 2 to 7 g/m3 PAA in secondary and

tertiary effluents, and from 5 to 15 g/m3 PAA in primary effluents238. Major

disadvantages associated with peracetic acid disinfection are the high running

cost and the increase of organic material in the effluent, due to acetic acid,

which also facilitates microbial regrowth236,239.

52 | Introduction

The relatively strong oxidation potential of PAA, slightly higher than

hydrogen peroxide, makes it a potential candidate for the removal of

pharmaceuticals residues233. Due to a minor risk of explosions in the

preparation of PAA on site at WWTPs, the number of reported tests on

removal efficiencies are limited. Batch lab tests, on three different biologically

treated municipal wastewaters, showed that PAA was not sufficiently efficient

to remove the evaluated pharmaceutical residues233. However, diclofenac and

mefenamic acid, which were removed most of the tested substances, achieved

a removal efficiency of approximately 90%, at a PAA dose of 25 g/m3. The

removal efficiencies for the other substances varied between 10-50%, at the

same PAA dose. The removal was negatively influenced by an increasing COD

concentration233, indicating the consumption of PAA to degrade bulk organic

substances present in the wastewater. In summary, PAA cannot be

considered as a candidate oxidant of pharmaceuticals, but it remains as an

interesting disinfectant chemical for wastewater233.

Technologies in development

Removal of pharmaceuticals in water and wastewater has met an increased

attention since two decades. Many alternative technologies have been

proposed. In addition to the processes described in this introduction, e.g.

Fenton reactions, membrane distillation and use of enzymes, algae and fungi

have been discussed as candidate process or measures for removal of

pharmaceuticals. These candidates are either resource demanding, still

under development or adaptation to municipal wastewater applications and

they are not further discussed in this thesis.

Comparison of treatment technologies

The different treatment technologies described previously have their pros

and cons. Removal efficiency, total cost, practical applicability and

complexity in process control are essential parameters to be evaluated before

a treatment technology can be selected. A comparison of the treatment

methods discussed above shows that, among the methods with high and

broad removal efficiency of pharmaceutical residues, ozonation are the most

favorable method. This is a result of the ozonation’s need for a relatively short

hydraulic retention time, ozonation provides at least a partial disinfection of

the wastewater and it can be implemented at a relatively low cost, Table 8.

Introduction | 53

Due to the risk for by products by ozonation, activated carbon is added to the

group of main candidates of most suitable methods for removal of APIs, in

spite the moderately higher costs than for ozonation. Today, the membrane

technologies are too costly to be an alternative for implementation in larger

scale. The selected main candidates ozonation and treatment with activated

carbon have previously been pointed out as potential methods for the

removal of APIs, at least for implementation in full-scale WWTPs10,122.

Ozonation and activated carbon can reach high removal efficiencies for many

different APIs, at a relatively low total cost, and they can be implemented in

existing WWTPs. Still questions of by-products formulation by ozonation

and high consumption of activated carbon, due to interferences and site

competition on adsorptive surfaces, must be more investigated, but that can

be studied in full-scale plants as well. The by-product formation by ozonation

can presumably be minimized by dose control and subsequent biofilm

processes, treating the ozonated wastewater. The other treatment

technologies described above, have either limitation in overall removal

efficiencies, high total costs or must be further studied before

implementation.

54 | Introduction

Table 8: Comparison of treatment methods for removal of pharmaceutical residues.

Treatment

method

Hy

dra

uli

c

rete

nti

on

tim

e

(HR

T)

Cle

an

ing

/

reg

ener

ati

on

By

-pro

du

cts

Dis

infe

ctio

n

Rem

ov

al

effi

cie

ncy

Co

st

Activated sludge

(CAS)

Long No Excess

sludge

No Low to high

and narrow

Low

Granular sludge

(AGS)

Moderate No Excess

sludge

No Low to high

and narrow

Moderate

Membrane bio

reactor (MBR)

Moderate Chemical

cleaning

Excess

sludge

No Low to high

and narrow

High

Biofilm systems Moderate No Excess

sludge

No Low to high

and narrow

Moderate

Activated Carbon

(GAC)

Short Backwash

of GAC

filters

Spent GAC

must be

regenerated

No High and

broad

Moderate

Activated Carbon

(PAC)

Short Backwash

of filter

units

Spent PAC

must be

incinerated

No High and

broad

Moderate

Nanofiltration

(NF) & reverse

osmosis (RO)

Short Chemical

cleaning

Retentate

must be

treated

Partly/

Yes

High and

broad

Moderate to

High

Ozonation (O3) Short Backwash

of sand

filters

Partly

degraded

substances

Partly/

Yes

High and

broad

Low to

moderate

UV/H2O2 Short Chemical &

mechanical

cleaning

Partly

degraded

substances

Yes Low to high

and narrow

High

H2O2/ O3 Short Backwash

of sand

filters

Partly

degraded

substances

Yes High and

broad

Moderate to

high

Chlorine dioxide

(ClO2)

Short No Partly

degraded

substances/

chlorite &

chlorate

Yes Moderate Moderate

Peracetic acid

(PAA)

Short No No toxic

by-

products

Yes Low to high

and narrow

Moderate to

high

Introduction | 55

Conclusion

Medicines are produced and consumed to cure or alleviate the symptoms of

diseases. The active pharmaceutical ingredients in medicines are

administrated at high doses, so the effect concentration of the active substance

still can be reached in the body. The pharmaceutical residues are excreted

mainly in urine, but also in faeces, which are transported in sewer networks to

municipal wastewater treatment plants (WWTPs) or discharged directly into

watersheds, if no treatment has been implemented in the area. Many

pharmaceuticals are resistant to applied removal technologies1,10 and passes

therefore through the WWTPs, to be discharged into the receiving waters.

Recent investigations240,241 suggest that pharmaceutical residues are

accumulating in the aquatic environment, with hitherto unknown half-life.

The effects of the discharged pharmaceutical residues from municipal WWTPs

are mainly unknown, but it can be assumed that the risk of adverse effects in

the environment is relatively high, due to the persistence, toxicity and

bioaccumulation of many substances. Receptors in animals, living in the

environment and subjected to discharged wastewater, are supposed to act

either similar to humans or dissimilar, to rise completely different, but still

severe, biological effects242. Since this might be an urgent existing and future

problem, there is a need of development and implementation of appropriate

technical solutions for WWTPs to remove pharmaceuticals.

56 | Present investigation

2 Aim and strategy

The overall aim of the present work is to provide the appropriate knowledge

enabling the WWTPs to remove pharmaceutical residues (APIs), according to

the demand of legislation and with a methodology and efficiency according to

the present scientific frontline.

The study was disposed into two main parts, formulated as four research

questions. In the first part, prevalence and modelling of APIs in the

environment was examined to confirm the magnitude of the problem with

persistence and spreading of APIs in the environment. In the second part,

different advanced treatment technologies to remove APIs from municipal

wastewater were evaluated in field studies.

The applied strategies were divided into four areas and they can be

summarized as follows:

Area 1) Mapping of sources and transfer of APIs into the aquatic environment. (Paper I, II and III)

The consumption patterns of APIs and the concentrations of APIs present in

the effluent from major Swedish WWTPs were studied and modelled to

describe the present situation with transfer of APIs to the aquatic

environment. The prevalence of APIs in the Baltic Sea was mapped by

sampling in the sub-basins of the Baltic.

Area 2) Do APIs accumulate in the aquatic environment? (Paper I)

The Baltic Sea was chosen as a model environment to study and determine the

prevalence of APIs in the aquatic environment, since the Baltic Sea is the

receiving waterbody for treated wastewater from many Swedish WWTPs.

The prevalence of APIs was mapped by sampling the Baltic Sea sub-basins and

the data were used to model accumulation, half-life and predict future

concentrations of a selected target substance i.e. carbamazepine. The model

was thereafter used to predict environmental concentrations (PEC) of this

substance and later compared them with the measured environmental

concentrations (MEC) of carbamazepine.

Aim and strategy | 57

Area 3) Among conventional and advanced treatment systems - is there a technology that can remove APIs in a majority of the WWTPs? (Paper II-V)

The work on removal of APIs was based on finding treatment technologies

that work broadly, on as many substances as possible. From a list of potential

technologies, the most promising were activated carbon, nanofiltration and

ozonation, which were selected based on potential high performance and low

to moderate cost. Initial lab tests supported the design and construction of

pilot plants, with the most promising technologies. To evaluate the validity of

the results, pilot tests were performed at four WWTPs, with different

characteristics in wastewater composition and process configuration. The goal

for the average removal efficiency of the APIs by the selected technologies was

set to 95%. The pharmaceuticals were chosen as representatives for APIs that

are persistent, bioaccumulating and toxic.

Area 4) Does the present discharge of APIs lead to biological consequences

in the aquatic environment? (Paper V)

Ecotoxicity tests were planned to estimate the potential effects of regular and

advanced treated wastewater on the aquatic environment. For the selected

technologies ozonation and activated carbon, field experiments were

conducted to study the effects on rainbow trout of untreated and treated

effluents. Selected biomarkers in the fish were used to establish the

bioeffects in both rapid and slow responding metabolism of xenobiotics such

as APIs.

58 | Present investigation

3 Methodology

This work was based on sampling, results from analysis, data collection and

modelling, as well as the designing and construction of two pilot plants,

whereof one was equipped with fish tanks for ecotoxicity tests.

Sampling

Grab sampling of surface water from 43 locations in the Baltic Sea and

Skagerrak was carried out by the crews on three ships, including the tall ship

Briggen Tre Kronor. Samples in the Barents Sea and the Greenland Sea were

taken from drilled holes in the sea ice. Samples of the WWTPs effluents and

treated wastewater in the mobile pilot plant were continuously collected, using

a 16-channel peristaltic pump (Ismatec IP16, Cole-Parmer GmbH, Germany).

Grab samples were taken in the nanofiltration/reverse osmosis (NF/RO) plant

and some of the permeate samples were pooled to form composite samples of

mixed permeate. The samples were stored at 6°C during sampling and frozen

at -20°C, before shipping to the analytical laboratory.

Measurements and analysis

3.1.1 Process parameters

Conductivity and pH were measured using a portable multiparameter meter

(Orion Star A329, Thermo Scientific). Turbidity was measured using a

spectrophotometer with a light detector placed 90° in relation to main light

path (Nanocolor VIS, Macherey-Nagel) and oxygen concentration in

wastewater was measured with an oxygen meter (HQ30D Portable Multi

Meter, HACH). In the mobile pilot plant, the incoming wastewater flow to the

individual treatment lines was recorded by water meters, supplied with

internal registers and pulse outputs for external reading (Z-System, Qn 1.5,

Systeme-sh). The transformed flow values were stored in a logger 2020 system

(WebIQAB) together with wastewater temperatures, measured every minute

by digital sensors (1-wire PRO, Dallas).

Methodology | 59

3.1.2 Lab analysis of traditional WWTP parameters

Total organic carbon (TOC) and dissolved organic carbon (DOC) were

determined using cuvette tests (LCK 385, Hach Lange) and measured with a

Hach Lange DR 3900 spectrophotometer. DOC was analyzed like TOC, but

after filtration through 0.45 mm membrane filters (Puradisc AQUA,

Whatman). Absorbance measurements at 254 nm (UVA254) were done with

a Hitachi U-2900 UV spectrophotometer, using quartz cuvettes with a 50 mm

path length (VWR). Suspended solids analysis was performed according to

wastewater standards (ESS Method 340.2), using glass-microfiber filters of

grade GF/A (Munktell).

3.1.3 Analysis of pharmaceutical residues

The pharmaceutical residues in the wastewater samples were analysed by

Umeå University, as a collaboration within the MistraPharma project. The

concentrations of 100 target pharmaceuticals were determined using liquid

chromatography coupled with mass spectrometry (LC-MS/MS). A triple-stage

quadrupole MS/MS TSQ Quantum Ultra EMR (Thermo Fisher Scientific, San

Jose, CA, USA) coupled with an Accela LC pump (Thermo Fisher Scientific,

San Jose, CA, USA) and a PAL HTC autosampler (CTC Analytics AG, Zwingen,

Switzerland) operated using Xcalibur software (Thermo Fisher Scientific, San

Jose, CA, USA) was used for the analysis of the water samples. A detailed

description of the method is given in a published paper243.

Data collection

Data on consumption of APIs in different countries, counties in Sweden and

concentrations in influent and effluent in Swedish WWTPs were collected

from on-line data bases, official reports and compilations of available

literature data64,65,244–254,254–270.

60 | Present investigation

Calculations and Modelling

3.1.4 Removal efficiency

The removal efficiency for an API i, REi, was calculated as

REi (%) = 100*(ci-cj)/ci (1)

Where ci and cj were the compound concentrations in the influent and effluent

of a tank, watershed system etc. In the case where influent concentration was

above LOQ and the effluent concentration was below LOQ, half the LOQ value

was used as cj in the calculation of RE.

3.1.5 Modelling of environmental concentrations of APIs

A grey box model for calculation of concentration of an API in the Baltic Sea

was developed based on a series or system of mass balances for substances in

individual sub basins, where each sub basin was approximated as a completely

mixed tank reactor. The mass balances were set up based on general

expression of the conservation of mass, equation (2).

Input + Produced = Output + Accumulated (2)

Data used in the model were: 1) sub basins characteristics; geographical

position, volume, surface area, air and water temperature, precipitation,

evaporation, river discharge and removal efficiency of pharmaceuticals from

smaller watersheds and 2) statistics on population size and geographical

distribution, specific consumer use of API, excretion rate in humans and

average removal efficiency in wastewater treatment plants.

The solutions of the equations for different sub basins were obtained by

iteration. A set of yearly mass balance equations were set up for the system of

interconnected sub basins and solved by iterations (N=20) in MS Excel. The

output of the model was time series of annual concentration of an API, in all

sub basins.

Wastewater treatment plants selected for pilot tests

Four WWTPs were selected in this study, due to significant differences in

connected population, different consumption pattern of pharmaceuticals in

the WWTPs catchment area, discrepancies between tertiary treatment

processes in the WWTPs, as well as differences regarding process conditions

such as hydraulic retention time (HRT), sludge loading rate and sludge age.

Methodology | 61

The four selected plants were Henriksdal WWTP, Käppala WWTP,

Kungsängsverket in Uppsala (Uppsala WWTP) and Kungsängsverket in

Västerås (Västerås WWTP). Data on WWTP characteristics and effluent

wastewater quality are given in Table 12 in the present investigation.

Henriksdal WWTP is the second largest WWTP in Sweden and the largest

in the Stockholm region and it treats on average 250 000 m3 wastewater per

day, corresponding to 750 000 population equivalents. The wastewater

treatment consists of pre-treatment (screening and grit removal), primary

sedimentation, biological treatment and sand filtration for removal of

particles. The biological treatment is an activated sludge process with pre-

denitrification, typical for Swedish WWTPs, with biological nitrogen removal

(BNR). Chemical precipitation of phosphorous is done by addition of ferrous

sulphate to the primary sedimentation tanks and to the sand filters. The

treated wastewater is discharged at 30 m depth, in the inner part of the

Stockholm archipelago.

In this study, only nanofiltration was piloted at Henriksdal WWTP.

Käppala WWTP is the second largest WWTP in the Stockholm region and

the plant treats 149 000 m3 wastewater per day, corresponding to 425 000

population equivalents. The treatment consists of pre-treatment (screening

and grit removal), primary sedimentation, biological treatment and sand

filtration. Two thirds of the wastewater are treated in a conventional activated

sludge pre-denitrification, with chemical pre and post precipitation of

phosphorous by addition of ferrous sulfate. One third of the wastewater is

treated biologically with the UCT (University of Cape Town) process271, which

involves enhanced biological phosphorous and nitrogen removal in activated

sludge systems. The biologically treated wastewater is collected and

distributed to the final sand filters. The treated wastewater is discharged at 45

m depth, in the inner part of the Stockholm archipelago.

Kungsängsverket (Uppsala WWTP) treats 50 000 m3 wastewater per day,

corresponding to 148 000 population equivalents. The treatment consists of

pre-treatment (screening and grit removal), primary sedimentation,

biological treatment and final chemical precipitation of phosphorous by

flocculation and lamella separation. Approximately 40% of the wastewater is

62 | Present investigation

treated in a conventional activated sludge pre-denitrification setup, while the

remaining 60% is treated using activated sludge in a step-feed pre-

denitrification setup. Both pre- and post-precipitation are used to remove

phosphorous through addition of ferric chloride. The treated wastewater is

discharged into river Fyris.

In Uppsala, the mobile pilot plant was extended with a pretreatment step in

the form of two shallow sand filter lines, after the early observations of a fast

clogging in the upper surfaces of the GAC filters. The sand filters were installed

upstream of the leveling tank in the pilot plant, from which all treatment lines

were fed.

Kungsängens WWTP (Västerås WWTP) treats 48 000 m3 wastewater per

day, corresponding to 102 000 population equivalents. The treatment consists

of pre-treatment (screening and grit removal), primary sedimentation and

biological treatment, the latter in the form of a conventional activated sludge,

with pre-denitrification setup. Methanol and ethylene glycol are used as

carbon sources to improve the nitrogen removal process. Pre-precipitation is

used to achieve phosphorous removal through addition of ferrous sulfate.

Polymeric coagulants are added to the secondary sedimentation tanks to

improve particle separation, before discharge to the effluent channels. The

treated wastewater is discharged into lake Mälaren.

The shallow sand filters installed in Uppsala were in operation in Västerås as

well.

Design, construction and description of pilot plants

Two different pilot plants were used in the study. The nanofiltration plant was

designed and constructed within the Stockholm Water project “Prevalence

and removal of pharmaceuticals in wastewater”10, where the author was

responsible for the pilot testing and the evaluation of the tests performed

within the project. The nanofiltration plant was only operated at Henriksdal

WWTP. The mobile pilot plant, with activated carbon, ozonation and biofilm

processes, was designed, constructed and operated within the KTH part of the

Swedish MistraPharma project. The latter pilot plant was operated at the

WWTPs in Käppala, Uppsala and Västerås.

Methodology | 63

3.1.6 Nanofiltration (NF) pilot plant

The membrane pilot plant unit was designed to work in a semi-continuous

mode, with an average hydraulic capacity of 800 L/h. Two NF membranes

were arranged in a two-staged array system, Figure 7. The total water volume

in the membrane unit was about 200 L, including the work tank volume of 180

L. After the final sand filter treatment in the WWTP, wastewater was

conducted to the pilot plant work tank and the flow was controlled with a ball

cock. From the work tank, the feed was pumped through a cartridge filter (10

µm mesh), to the high pressure pump. After passing the two membranes, the

retentate flow was divided, one part was continuously recycled back to the

work tank, through a heat exchanger to keep the temperature in the

wastewater at 21°C. The other part was bypassed to the inlet of the high-

pressure pump. The applied hydraulic pressure was regulated by a needle

valve placed on the bypassing retentate pipe. In order to allow for backflow

diffusion, the membrane process was stopped for 20 s every 20 min. The

membranes used in this study were two commercial spiral wound thin film

nanofilter membranes, with aromatic composite polyamide sheets (ESNA1-

LF-4040, Hydranautics). The nominal area of each membrane was about 7.9

m2 and the molecular weight cut-off (MWCO) of the membranes was 150 Da.

The membranes have a low fouling tendency and were neutral to slightly

negatively charged and were chosen due to the relatively low MWCO, which

theoretically gives a high retention of nearly all APIs.

Figure 7. Outline of the nanofiltration pilot plant.

64 | Present investigation

3.1.7 Activated carbon, ozonation and biofilm mobile pilot plant

A mobile pilot plant was designed and constructed in-house for application at

Käppala, Uppsala and Västerås WWTPs. Treatment tanks, and equipment for

sampling and process control, were connected and installed into an insulated,

20-foot shipping container, allowing for operation at outdoor temperatures

down to freezing, Figure 8.

Figure 8. The mobile pilot plant with technology for additional treatment Photo: Berndt Björlenius.

The pilot plant contained eleven treatment lines; three designed for GAC

application, another three for PAC, two ozonation lines, two lines using

biofilm (MBBR) and finally one line with sand filtration after ozonation. The

process tanks were all made of stainless steel, except the contact columns for

the ozonation, which were made from PVC pipes. An applied control system

served the lines to control inlet pumps, water levels in GAC and sand filters,

the ozonation dose, dosing of PAC and of flocculation agent. The control

system also stored process data for later evaluation.

Methodology | 65

An overview of the treatment lines and the hydraulic pathways in the mobile

pilot plant is presented in Figure 9.

Figure 9. Overview of the treatment lines in the mobile pilot plant.

During operation, effluent wastewater was pumped by a submerged impeller

pump, via coarse particle filters (2 mm perforation), to a leveling tank located

outside the container. The treatment lines were fed directly or indirectly from

the leveling tank by separate screw pumps. GAC filtration was performed in

three identical treatment lines, but with filter lines filled with different GAC

products. Each one of the GAC lines consisted of two filters (columns) in

series. The filters were 2 m height and had an inner diameter of 0.15 m each.

The operational volume for each GAC line was approximately 60 L. The filters

were filled with 1 m of GAC and operated with down-flow configuration. The

water level in the GAC filters was kept constant by a control valve in the

bottom of the filter, which was connected to the pilot plant control system.

Backwashing of the filters was performed with tap water regularly.

PAC treatment was equally performed using three identical treatment lines.

Each line consisted of an initial mixing tank for effluent wastewater and dosed

PAC, followed by three sequential aerated contact tanks, a sedimentation tank

and a final dual media sand filter. The contact tanks were aerated for mixing

purposes. The contact tanks, sedimentation tank and sand filter had an

operating volume of 100 L, 180 L and 75 L respectively, making a total

operational volume for each PAC line of 360 L, six times the volume in of a

GAC line. The sand filter was filled with a 0.33 m bottom layer of sand, 1.2-2

mm particle size (Rådasand AB) and a 0.77 m top layer of crushed, expanded

66 | Present investigation

clay with 2-4 mm particle size (Filtralite MC, Saint Gobain Byggevarer AS).

Recirculation of PAC, from the bottom of the sedimentation tank, back to the

first, second or third contact tank was accomplished by an airlift pump.

The design of the two ozonation lines in the pilot plant was based on tests with

a lab scale ozonation plant, with a 5 m high contact tank equipped with two

alternative devices for dosing of ozone: a) bubble diffusor or b) venturi

injector. The lab tests showed that higher removal efficiency was achieved

with a venturi injector, than with bubble diffusion for the ozone dosing. Based

on this, two ozonation lines were designed, constructed and fitted into a

mobile pilot plant, Figure 10, where ozone was produced from oxygen, in an

ozone generator and the produced ozone, and the remaining oxygen, was

added to the wastewater by a venturi injector.

The ozone was produced by an ozone generator (ICT-10, Ozone Tech Systems,

Sweden) supplied with 92-99.5% oxygen and was supplied with water cooling.

A venturi injector (Mazzei Injector Company LLC, Bakersfield, CA, USA) was

used for the addition of the ozone-containing oxygen into the wastewater. The

wastewater was thereafter distributed between ozonation line 1 and line 2.

Ozonation line 1 consisted of three, 5 m high, contact columns (CC1, CC2 and

CC3), with an operational volume of 83, 167 and 267 L. The contact columns

had a subsequent, 200 L aerated stripping tank, for the removal of potentially

remaining residues of free ozone, and finally a sand filter, allowing growth of

biofilm, for the potential removal of biodegradable products, released in the

ozonation. One of the contact columns was used at a time, or they were

coupled in series, to extend the retention time for ozone-wastewater contact.

Alternatively, in line 2, the wastewater passed a pressurized contact tank (CT1)

for wastewater containing 300 L, with a height 1.7 m and a diameter of 0.54

m. The wastewater was pressurized by an inlet pump to 1.4 bar absolute

pressure. A final 33 L aeration tank was installed for the removal of potential

residues of free ozone, from the ozonated wastewater. Oxygen was supplied in

gas cylinders (99.5% O2) or alternatively by an on-site oxygen concentrator

(92-96% O2) (Onyx+, AirSep Corporation, Buffalo, NY, USA). The flow of

wastewater through the plant was typically 0.4-1.2 m3/h. An ozone dose

between 1 and 20 g/m3 could be added to the wastewater and the dosing was

managed from the control system in the mobile pilot plant.

Methodology | 67

Figure 10: Layout of the ozonation pilot plants. Line 1 has three 5 m high contact column (CC1-CC3)

and line 2 has one 1.7 m high contact tank (CT1).

Ten 50 L fish tanks were installed in a rack inside the mobile pilot plant for

intermittent ecotoxicity tests using rainbow trout. In the setup, five

wastewaters with duplicates could be tested in parallel. Effluent wastewater

from the WWTP served as the positive control and tap water as the negative

control. The retention time for the wastewater was one hour in each fish tank.

Aeration via ceramic air stones supplied the water with sufficient, >95%

oxygen saturation.

68 | Present investigation

3.1.8 Operation of the mobile pilot plant at the specific WWTPs

The mobile pilot plant was operated for 57 weeks in total, at three different

WWTPs. After commissioning of the mobile pilot plant, evaluation tests with

flow proportional water sampling started in October 2013 at Käppala WWTP.

Weekly samples were collected from 14 sampling point. In addition to the

mainly continuous operation of the lines in the pilot plant, factorial

experiments with e.g. dose, temperature and pH were performed.

After the first four months of operation, sampling and preliminary evaluation

at Käppala WWTP, the pilot plant was moved to Kungsängsverket WWTP in

Uppsala. The disassembling and assembling of the plant were relatively

efficient, in less than two weeks, the plant was up and running in Uppsala. The

most obvious observation, showing just after a few days, was that without a

final sand filter like at Käppala WWTP, the up time for granular activated

carbon filters (GAC) was limited. To prolong the uptime for the GAC-filters, a

pretreatment in form of sand filters was built upstream of the pilot plant. In

Uppsala additional factorial experiments were undertaken in parallel with

continuous operation of the eleven lines.

In the end of September 2014, the pilot plant was disassembled and moved to

Kungsängsverket WWTP in Västerås, where continuous operation and

factorial experiments were commenced and performed to mid December

2014, when the pilot plant was moved back to Käppala WWTP, where

complementary tests were done on pretreatment, control of ozone addition,

performance of GAC and powdered activated carbon (PAC), during spring

2015.

Results and discussion | 69

4 Result and discussion

To reach the proposed aim, the work has been divided into four strategic areas,

stated under chapter 2, “Aim and strategy”. The areas are:

Area 1) Mapping of sources and transfer of APIs into the aquatic environment. (Paper I, II and III) Area 2) Do APIs accumulate in the aquatic environment? (Paper I) Area 3) Among conventional and advanced treatment systems - is there a technology that can remove APIs in a majority of the WWTPs? (Paper II-V) Area 4) Does the present discharge of APIs lead to biological

consequences in the aquatic environment? (Paper V)

Each strategic area is treated below and relates mainly to the results presented

in the attached papers, but results are also related to other publications.

Area 1) Mapping of sources and transfer of APIs into the aquatic environment.

4.1 Use of pharmaceuticals (Paper I-V)

The specific use of pharmaceuticals was shown to vary between country

regions and countries. The use of specific pharmaceuticals also changes with

time. The overall tendency is 3% yearly increase of total consumption of

medicines, especially of pharmaceuticals related to ageing and chronic

diseases, but also an increase due to changes in clinical practice244,272. The

studied APIs were selected as preferable being persistent, bioaccumulation

and persistent, but also sold in considerable amounts.

4.1.1 Use of carbamazepine in countries in the Baltic Sea

catchment area

The specific consumption of carbamazepine [DDD/1000 inhabitants] in the

countries in the Baltic Sea catchment area were extracted from different,

mainly national, official sources but compilation of available literature data

was also performed. In the case of unavailable official data, interpolation and

70 | Present investigation

estimations of similar trends in consumption patterns, as in countries with

available data sets, were used to estimate the load of carbamazepine to the

Baltic Sea, Table 9.

Table 9. Specific consumption in 2013 of carbamazepine in the countries in the Baltic Sea catchment

area. The specific consumption is given as daily defined dose per 1000 inhabitants. *Estimated values

based on few data and general consumption pattern in Sweden (Paper I).

Country DDD/1000 inhabitants

in 2013

Country DDD/1000 inhabitants

in 2013

Belarus 0.78* Lithuania 0.98

Czech Republic 1.47* Norway 1.37

Denmark 1.00 Poland 2.4*

Estonia 2.15 Russia 0.34*

Finland 1.54 Slovakia 1.23*

Germany (including

former GDR)

1.47 Sweden 1.74

Latvia 0.9 Ukraine 0.21*

Table 9 shows that the specific consumption of carbamazepine is not uniform

in the Baltic Sea catchment area. One explanation can be that alternative APIs

to treat epilepsy are in use to different extents in different countries, with the

most obvious difference for Ukraine, Russia and Belarus. This is most likely a

legacy from the former Soviet Union, where methindione was used to treat

epilepsy247.

4.1.2 Use of pharmaceuticals in regions – counties in Sweden

The consumption of pharmaceuticals varies also between regions in a country.

In 2013, selecting carbamazepine as an example, consumption in the 22

counties in Sweden, showed a variation from the lowest consumption in the

Stockholm County, 1.45 DDD/1000 inhabitants, to the highest consumption

in the Jönköping county, where 2.59 DDD/1000 inhabitants were used245.

A comparison of statistics from 2014, for the three counties visited during the

tests with the mobile pilot plant, showed discrepancies in consumption of ten

APIs in this study, which were quantified in almost all samples from regular

effluent from the three WWTPs, Table 10.

Results and discussion | 71

Table 10. Specific consumption of ten APIs, quantified in the pilot studies in this study245.

Specific consumption of

API [DDD/1000

inhabitants, day]

Käppala

WWTP

Uppsala

WWTP

Västerås

WWTP

Average

County council

Stockholm

(Paper III-V)

Uppsala

(Paper III-V)

Västmanland

(Paper III-IV)

All

Atenolol 4.24 10.5 12.4 8.93

Carbamazepine 1.39 1.59 1.97 1.65

Citalopram 16.9 25.0 23.1 20.9

Codeine 0.13 0.13 0.24 0.15

Diclofenac 3.88 4.26 5.39 5.80

Metoprolol 24.0 23.4 22.4 23.3

Mirtazapine 6.51 10.4 9.26 7.85

Oxazepam 2.71 2.86 4.37 3.21

Tramadol 2.49 4.15 5.47 3.96

Trimethoprim 0.13 0.12 0.14 0.19

Σ of 10 APIs

[DDD/1000 inhabitants]

62.3 82.4 84.8 75.9

The consumption of APIs can be imagined as varying within and between

cities, due do the demographic situation and maybe also due to the doctors’

customs of prescribing different APIs, especially when alternative APIs are

available.

The most quotable differences in consumption between counties, were for

atenolol and citalopram. In general, the prescriptions of the selected APIs

were highest in Västerås, followed by Uppsala and lastly Käppala. This order

corresponds also to the concentrations of APIs in the effluent from the three

WWTPs (Paper III, Table 2). In relative numbers, the specific consumption of

the sum of the ten selected APIs was 32% higher in Uppsala WWTP catchment

area and 36% higher in Västerås WWTP catchment area, than in the Käppala

WWTP catchment area. The different consumption patterns influenced the

72 | Present investigation

loads of APIs on the WWTPs and thereby were the concentration of APIs in

influent and effluent wastewater influenced. The concentrations of APIs in

wastewaters can also have been influenced by the amount of stormwater in

the influent to the WWTP, wastewater temperature, process configuration and

conditions at the WWTPs, including hydraulic retention time and sludge age

in the biological treatment.

4.1.3 Long-term variations in consumption of pharmaceuticals

(Paper I)

After the market introduction of a pharmaceutical, the consumption of the

pharmaceutical will increase. The long-term development in consumption

pattern depends on many factors such as marketing, medical evaluation of

effects, new medical treatment areas for the API etc. In the modelling of the

concentrations of carbamazepine in the Baltic Sea, the access to long-term

consumption data were essential. Carbamazepine was introduced to the

market in Switzerland in 1962 and approved as an anticonvulsant in UK in

1965246. The consumption of carbamazepine in four countries in the Baltic Sea

catchment area developed differently during the years, Figure 11.

Figure 11: Specific consumption of carbamazepine in 1975-2013 in four countries in the catchment

area of the Baltic Sea (Paper I).

Results and discussion | 73

The consumption patterns were relatively similar in Sweden and Denmark

during the first decade after 1975, which was the starting year of the

simulations in Paper I. Then, the specific consumption increased and was

maintained at higher levels in Sweden. During the last decade, a similar

annual decrease in consumption of carbamazepine was observed in Sweden

and Denmark. In Germany, the initially low levels of consumption, might be

an effect on the low consumption in former GDR. A similar explanation can

probably be found for Estonia, being a former Soviet Union republic, where

methindione was used to treat epilepsy247. In summary, the tendency is a

general decrease in specific consumption of carbamazepine. Alternative APIs

have partly and successively substituted carbamazepine in the treatment of

epilepsy. The awareness of non-steady state conditions for the specific

consumption of pharmaceuticals, is crucial in the long-term modelling of

environmental concentrations of pharmaceutical residues.

However, the total specific consumption of all pharmaceuticals, calculated as

DDD/1 000 inhabitants, is steadily increasing in many countries. During the

period 2000-2015, the average specific consumption of pharmaceuticals in

nine ATC groups increased linearly 2.9% per year (R2=0.99) in 14 countries in

the OECD (Australia, Belgium, Czech Republic, Denmark, Estonia, Finland,

Germany, Hungary, Iceland, Italy, Norway, Portugal, Slovak Republic and

Sweden). The ATC groups in the evaluation were: A-Alimentary tract and

metabolism, B-Blood and blood forming organs, C-Cardiovascular system, G-

Genito urinary system and sex hormones, H-Systemic hormonal preparations,

excluding sex hormones and insulins, J-Anti-infectives for systemic use, M-

Musculo-skeletal system, N-Nervous system and R-Respiratory system272.

The increased consumption will most likely lead to higher concentrations of

pharmaceuticals in the environment, above all in aquatic systems, where they

have been found globally, often in combination with low removal efficiency in

wastewater treatment plants (WWTPs) or a complete absence of

WWTPs2–5,273.

74 | Present investigation

Area 2) Do APIs accumulate in the aquatic environment?

4.2 Pharmaceuticals in the environment (Paper I)

A sampling campaign of the Baltic Sea water was initiated and planned to

study the prevalence and mitigation of pharmaceutical residues in the aquatic

environment. Understanding the fate of APIs in the effluent from existing

wastewater treatment plants, later transported or processed in the receiving

water, can help us to select the APIs that have to be removed in the wastewater

treatment plants, instead of being accumulated in the environment.

4.2.1 Measured environmental concentrations (MECs)

Since the Baltic Sea is vulnerable to anthropogenic activities, due to a long

turnover time and a sensitive ecosystem in the brackish water274,275, it is a

suitable location for environmental studies. Furthermore, it is the receiving

water body for many Swedish WWTPs.

A sampling campaign, all over the Baltic Sea, was initiated to study the

prevalence and persistence of pharmaceutical residues in larger water bodies,

Figure 12, and the sampling was designed to cover as much as possible of both

coastal and offshore locations. The potential accumulation of pharmaceutical

residues can lead to successively higher concentration, which finally reach

effect concentrations in organisms in the water. The sampling points were

chosen to cover all sub-basins in the Baltic Sea, to provide environmental

concentrations, also for modelling purposes later on.

The results from the chemical analysis displayed the range of quantified

environmental concentrations of a selection of 93 pharmaceuticals, in 43

locations in the Baltic Sea and Skagerrak area, Figure 13.

Thirty-nine of the 93 pharmaceuticals were detected in at least one sample,

with concentrations ranging between 0.01 and 80 ng/L. In general, coastal

locations had higher concentrations of APIs, than offshore locations and the

highest concentrations were found outside major cities. The anti-epileptic

drug carbamazepine was widespread, both in coastal and offshore seawaters

(present in 37 of 43 samples) and is thus a good indicator substance for

anthropogenic impact in water. The median concentration of carbamazepine

was 2.6 ng/L in the Baltic Sea, which was lower than previous reported

median, 22 ng/L, but the latter was based on samples from a limited area of

the Baltic Sea, namely the coastal areas of Northern Germany240.

Results and discussion | 75

Figure 12: Baltic Sea with sampling points and main sub basins: BB=Bothnian Bay, BS=Bothnian

Sea, GF=Gulf of Finland, GR=Gulf of Riga, BP=Baltic Proper, DS=Danish Straits, KT=Kattegat and

SK=Skagerrak. Two samples, 44 and 45, were taken at Svalbard according to the pasted map in the

upper right corner. Values of the environmental concentrations are presented in the supplementary

material to paper I.

Figure 13: Minimum, median and maximum concentrations (denoted as vertical bars) of quantified

APIs, sorted in declining frequency [%].

76 | Present investigation

4.2.2 Prediction of environmental concentration of pharmaceutical

residues – Example carbamazepine

Prediction of environmental concentrations of chemicals is valuable in the risk

assessment of the use of chemicals and the discharge of chemical residues to

the environment. It is also valuable in the work to identify parameters

essential for the spreading of chemicals. Furthermore, prediction of

environmental concentrations can be used in the prioritizing work of regions

relevant for implementation of advanced treatment at WWTPs or at

industries.

In this study, a mathematical model was set up to predict concentrations of

pharmaceuticals in the sub basins of the Baltic Sea and the persistent and

widely used carbamazepine was selected as model substance in the

simulations. The modelling included input data from all countries, not only

Sweden, to predict environmental concentrations, in particular

concentrations of carbamazepine.

The grey box model consisted of a system of mass balances for a single

substance, distributed to sub-basins, where each sub-basin was approximated

as a completely mixed tank reactor. The mass balances were set up based on

the expression for the conservation of mass, equation (2), also stated in the

methodology chapter, but is repeated below.

Input + Produced = Output + Accumulated (2)

APIs were not produced in the sub-basins, except in the case if the parent

substance was reformed from a conjugated API. Many APIs were removed to

different extents in the sub-basins, giving a negative value to the production

term. A graphical representation of a mass balance for sub-basin i, is shown

in Figure 14.

Results and discussion | 77

Figure 14: Graphical representation of the components of a mass balance for the sub-basin I, where Vi was the volume of the basin, QLi , QPi ,QEi , QXij and Qxji were the water flow from land, precipitation, evaporation, water exchange from connected sub-basin j to sub-basin i and water exchange from connected sub-basin i to sub-basin j respectively ; cLi, cj and ci were the concentrations of the substance in the water streaming from land (L), sub-basin j and sub-basin i.

The system of mass balances enclosed the Baltic Sea and was divided into

seven main sub-basins, plus, the to the Baltic Sea adjacent Atlantic sub-basin,

Skagerrak, Table 11. Flows of water, originated from precipitation,

evaporation and rivers, and mass flows of substances, including discharges of

APIs from sewers and WWTP effluents were calculated.

A set of equations were set up for the system of sub-basins and solved

iteratively. The input data into the model were sub-basin characteristics and

statistics on national population size and geographical distribution in each

sub-basin, use of API, excretion rate of humans and average removal in

wastewater treatment plants. The output from the model was a collection of

time series of annual concentrations of a substance, in all separate sub-basins.

The predicted concentrations of carbamazepine ranged from 0.51 to 2.5 ng/L

in the sub-basins, and the corresponding measured concentrations, amounted

to 0.57-3.2 ng/L, depending on sub basin location, Table 11. An analysis of the

mean absolute errors (MAEs) for the predictive values showed that the data

series, applied for the parameters in the model, seemed to be acceptable, the

mean absolute error for the predictive values was 0.43 ng/L, which

corresponded to 23% of the average measured concentration of

carbamazepine in the Baltic Sea waters.

Q P i

c i c j

Q E i

Q Li c L

Q xij Q xji

Sub-basin i

c i V i

78 | Present investigation

Table 11. Predicted concentrations of carbamazepine in Baltic Sea with sub-basins, and the

adjacent Atlantic sub-basins Kattegat and Skagerrak (Paper I).

Basin PEC

[ng/L]

MEC ± standard deviation

(min – max) [ng/L]

Bothnian Bay 1.2 0.57 ± 0.64 (0.05 – 1.5)

Bothnian Sea 1.8 2.3 ± 0.90 (1.1 – 3.7)

Gulf of Finland 2.1 3.2 ± 0.37 (2.7 – 3.6)

Gulf of Riga 2.2 2.3

Baltic Proper 2.5 2.5 ± 0.46 (1.6 – 3.3)

Danish Straits 1.9 1.9 ± 0.51 (1.2 – 2.4)

Kattegat 0.95 1.3

Skagerrak 0.51 0.75 ± 1.22 (0.05 – 3.1)

Baltic Sea 1.8 1.9 ± 1.16 (0.05 – 3.7)

The average concentration of carbamazepine in the Baltic Sea was predicted

to be 1.8 ng/L in 2013, compared to the average measured environmental

concentration of 1.9 ng/L, which corresponded to an accumulated mass of

carbamazepine in the Baltic Sea of 55.6 metric tons. A simulation, with help

of the model, was performed to predict future annual concentrations of

carbamazepine in the sub basins of the Baltic Sea. A scenario with a complete

stop in consumption of carbamazepine, showed the long response time, a

matter of decades, for the total removal of accumulated carbamazepine in the

Baltic Sea, Figure 15.

The predicted concentrations of carbamazepine ranged from 0.51 to 2.5 ng/L

in the sub-basins, and the corresponding measured concentrations, amounted

to 0.57-3.2 ng/L, depending on sub basin location, Table 11. An analysis of the

mean absolute errors (MAEs) for the predictive values showed that the data

series, applied for the parameters in the model, seemed to be acceptable, the

mean absolute error for the predictive values was 0.43 ng/L which

corresponded to 23% of the average measured concentration of

carbamazepine in the Baltic Sea waters.

Results and discussion | 79

Figure 15. Simulated concentration of carbamazepine in the Baltic Sea and sub basins in the past

and in the future after a hypothetical case with a complete stop of consumption in 2014, indicated

with a vertical dashed line. Highest concentration in year 2013 was predicted to be present in Baltic

Proper (Ӿ) followed by Gulf of Riga (+), Gulf of Finland (— —), Danish Straits ( ● ),

Bothnian Sea (‐ ‐), Bothnian Bay (-), Kattegat (▲) and Skagerrak (♦) in descending order. The

average annual concentration in the Baltic Sea (■) is based on sub basin volumes times predicted

concentration divided by total volume in Baltic Sea and Skagerrak.

The simulation clearly showed that carbamazepine will be present in the Baltic

Sea decades after a stop in use. The average flushing and degradation half-

time for carbamazepine was ten years. It is important to keep in mind the long

response time in the environment, when measures or regulations are

discussed to phase out persistent chemical substances – a complete stop in

use will not immediately result in zero concentrations in the environment.

80 | Present investigation

4.2.3 Half-life of carbamazepine (Paper I)

Based on removal efficiencies of carbamazepine in surface water, which were

calculated from literature data and samples taken in this study, an estimation

of the average removal efficiency for one year, and the corresponding removal

rate were calculated. An overall removal efficiency of 17.3% per year was

achieved from measured and calculated data, corresponding to a half-life, t1/2,

of 3.54 years, or nearly 1 300 days, at 10°C. Assuming a 1th order kinetics for

the degradation of carbamazepine, the average removal rate, r, was 6.2-09 s-1,

calculated according to r= ln2/ t1/2. The persistence of carbamazepine makes

it thus a good indicator of wastewater intrusion in natural waters.

The half-life for carbamazepine in water, previously reported in two separate

studies, was much shorter: 63 days276 and 38 days in constant sun light211,

although they incorporated flushing rate and had a constant sun light in a

laboratory environment respectively. In the second study, the water depth was

2.1 cm, compared with the average depth of 55 m in the Baltic Sea. The

intensity of solar UV-light decreases with water depth, e.g. at 2 m, the

remaining UV intensity at λ=325 nm is <1% to 55%, depending of the amount

of particles, algae etc. in the water. At 10 m depth, the remaining UV intensity

is <2%277–279. In a recent study that used biodegradation models, the half-live

time of carbamazepine in water was estimated to be 201 days280. This

demonstrates clearly, independent of used model, that slowly degradable

substances can generate at least long-term esthetic, or worse, problems in the

aquatic environment.

4.3 Pharmaceutials in wastewater effluents (Paper II-V)

The WWTPs, where the pilot tests were done in this study, were selected to

represent different geographical regions, process layouts and treatment

performance, which might influence the wastewater matrix and the

composition of the treated wastewater. Evaluation of the characteristics of the

WWTPs showed discrepancies in all evaluated parameters (Table 12).

Results and discussion | 81

Table 12. Characterization of the WWTPs during the studies in this thesis, paper II-V.

Parameter

Henriksdal

WWTP

(Paper II)

Käppala

WWTP

(Paper III-V)

Uppsala

WWTP

(Paper III-V)

Västerås

WWTP

(Paper III-IV)

Connected population [pe] 750000 425 000 148 000 102 000

Specific wastewater flow [L/pe,d] 333 390 305 471

Temperature [°C] 15 12 18 16

Sludge age [d] 16 16 23 10

Hydraulic retention time (HRT) [h] 29 36 38 14

Suspended solids in effluent [mg/L] 2.8 0.6 4.6 3.7

DOC in effluent [mg/L] 6.8 9.3 9.0 9.4

Dilution factor to recipient 68x 116x 20x 12x

The mapping of the concentrations of APIs in effluent wastewater, from the

examined WWTPs, was performed to describe and understand how much

pharmaceutical residues, that regular WWTPs release to the aquatic

environment after treatment. The results of the mapping were also used as

part of the characterization of the selected WWTPs, but mainly as reference

values in the evaluation of additional treatment technologies for removal of

pharmaceutical residues. Effluent concentrations of a selection of APIs, that

were frequently quantified in the effluent from the four WWTPs, were

compared with the critical environmental concentration (CEC) for the APIs,

Table 13. The CECs are the levels of environment concentrations of different

APIs, which will bioconcentrate in fish, so the plasma concentration of APIs

in exposed fish will be equal to the human therapeutic plasma

concentration281. Data on concentrations of other APIs are available in paper

II-V.

82 | Present investigation

Table 13. Effluent concentrations, critical environmental concentrations and proposed environmental

quality standards of chronic toxicity for freshwater in EU134 and Switzerland (CH)282 of ten APIs from

the four WWTPs with pilot tests.

Concentration

[ng/L]

Henriksdal

WWTP

(Paper II)

Käppala

WWTP

(Paper III-

V)

Uppsala

WWTP

(Paper III-

V)

Västerås

WWTP

(Paper III-

IV)

Critical

environmental

Concentration

(CEC)281

EQS

fresh

water

CH (EU)

Atenolol 487 250 220 729 792 000 150 000

Carbamazepine 387 221 524 202 346 000 2 000

Citalopram 220 92 357 206 141 n.d.

Codeine 57 85 74 498 26 620 n.d.

Diclofenac 183 287 200 690 4 560 50 (100)

Metoprolol 1 330 1 203 1 095 684 15 400 8 600

Mirtazapine 61 119 189 120 14 000 n.d.

Oxazepam 325 178 330 741 30 700 n.d.

Tramadol 553 258 563 824 4 800 n.d.

Trimethoprim 84 22 77 62 3 300 000 120 000

The concentrations in the effluent varied a lot for the different WWTPs.

Several explanations are plausible: 1) At relatively short sludge age, like in

Västerås, removal efficiencies of slowly degradable APIs are lower140. 2)

Higher specific flow of wastewater, mainly due to stormwater intrusion, will

dilute the influent and thereby lower the concentrations of pollutants. 3)

Variation in the inhabitants use of API will influence the concentrations. 4) A

higher wastewater temperature will increase the removal rate, which will

lower the concentrations of degradable APIs283. 5) A long HRT will promote

the removal efficiency, due longer time for degradation to APIs, especially if

the biological treatment has a long HRT.

The concentrations of APIs in the effluent from the WWTPs were diluted by

natural inflows, e.g. surface run-off, in the receiving waters. The average

dilution factors in the receiving waters were for Henriksdal WWTP: 68x,

Käppala WWTP: 116x, Uppsala WWTP: 20x and Västerås WWTP: 12x, during

the year when the pilot tests were done, Table 13.

Results and discussion | 83

The dilution factors were sufficiently high to dilute all effluents, even the

concentrations of the most critical API in this list – citalopram – were lower

in the receiving water, than the predicted critical environmental concentration

(CEC), based on bioaccumulation281. The safety factor between critical

environmental concentration and calculated environmental concentration

was less than 10 for citalopram in the receiving water in Uppsala and Västerås.

The concentrations of diclofenac in the effluents were higher than the

proposed environmental quality standards (EQS). However, thanks to the

dilution, diclofenac concentrations in the receiving waters were lower than the

EQS for fresh water.

In the WWTPs, the concentrations in the effluents indicated different

prescription patterns between cities and municipalities, where Västerås

probably had the highest prescription, and the municipalities connected to

Käppala WWTP, had the lowest prescription. The dilution in the receiving

water lowered the concentrations of all APIs, below both the critical

environmental concentrations and the proposed environmental quality

standards (EQS) for the APIs, which indicates that the discharged

pharmaceutical residues will not influence fish in the receiving waters by

potential bioaccumulation.

To summarize the observations so far in this study, prescription pattern and

the configuration of WWTPs, influenced the concentrations and mass flows of

APIs into the aquatic environment. Based on the measured environmental

concentrations and performed calculations for the Baltic Sea, the half-life of

the chemical substance differed between APIs and was as long as several years,

as for carbamazepine. A consequence of the slow degradation of hydrophilic,

or dissolved APIs, is that they can pass the WWTPs and they can accumulate

in water bodies. Carbamazepine was found in offshore waters all over the

Baltic Sea. However, the concentrations of the examined APIs were lower in

the specific aquatic environment, than the corresponding critical

concentration for effects of bioaccumulation in fish. In receiving waters with

lower dilution and higher concentrations of APIs in the effluent from WWTPs,

critical environmental concentrations can likely be reach for some APIs.

84 | Present investigation

Area 3) Among conventional and advanced treatment

systems - is there a technology that can remove APIs in a

majority of the WWTPs?

4.4 Removal of pharmaceutical residues by nanofiltration (Paper

II)

The pilot plant for nanofiltration (NF) was configured to treat effluent

wastewater from Henriksdal WWTP in Stockholm. The NF removed solutes

and ions of a certain size from wastewater by an applied pressure over a

semipermeable membrane. The treated wastewater was discharged as a

permeate from the nanofiltration, while the separated substances were

accumulated in the retentate, which therefore must be further treated or

disposed. Most APIs have relatively low molecule weights, typically 150-900

g/mol. Initially, pilot tests were designed to evaluate the removal efficiencies

of a selection of APIs, abounded in the effluent of Henriksdal WWTP. A

multivariate model was built to identify the most descriptive parameters for

the removal of APIs.

During the pilot tests, 32 of the 95 APIs analyzed were quantified in the regular

effluent from Henriksdal WWTP, which is equipped with sand filters as the

last treatment step. The results from the chemical analysis of the nanofilter

permeate were used to calculate the removal efficiencies of different APIs,

which varied between 38-99%, with a median removal of 90% of the 32 APIs.

The total volume reduction factor (VRF) was 20 during the tests, which

implies that the treated effluent from Henriksdal WWTP was split into 95%

permeate and 5% retentate, whereof the latter must be further treated. The

project aim of 95% removal of APIs was not fulfilled with the selected

nanofilter membranes, but still the removal efficiencies were high. The NF

membranes were selected based on an offered membrane molecular weight

cut-off (MWCO), specified to be 150 g/mol, which was lower than the lowest

molecular weight (MW) of the selected APIs in the present study. A

comparison of the removal efficiencies achieved versus MW, showed that the

cut-off for 90% removal for an individual API was approximately 300 g/mol,

instead of the offered 150 g/mol, Figure 16. To reach 95% removal of any

individual API, the MW of the API must exceed 420 g/mol in this study. The

observed lower removal of the smallest APIs, with MWs in the interval 150—

300 g/mol, is thus a membrane problem.

Results and discussion | 85

Figure 16: Removal of 32 pharmaceuticals vs. molecular weight (MW) in the nanofiltration studies.

To better understand and predict the removal of APIs in nanofiltration, a

model was developed with multivariate analysis methodology, to predict the

rejection for a nanofiltration plant, treating effluent from full-scale WWTPs.

Many of the studies previously reported on the removal of micropollutants by

nanofiltration, were performed in a laboratory environment. By these lab

tests, virgin membranes and spiked pure water were used, to avoid

interference of the wastewater matrix in the evaluation183,189,284–291.

To model the rejection of APIs by nanofiltration, Principle Components

Analysis (PCA) and Partial Least Squares Projection of Latent Structures

Analysis (PLS) were used. Initially, PCA was used to give an overview of the

data, check for outliers and discover correlations between the variables. In this

process, four APIs were excluded from the development of the model, being

significantly different due to large size or small size in combination with large

polar water accessible surface areas compared to the other APIs. However, the

four APIs were included in the latter testing of the models.

In the modelling, 59 physiochemical properties of the pharmaceutical

residuals present in the treated wastewater were initially used as variables

(Paper II – Table 3 in supplementary material). Examples of the molecular

properties were molecular size and shape, electrostatic properties, polarity,

86 | Present investigation

hydrophobicity and presence of different functional groups. To distinguish

between different physical characteristics, ratios between different variables

were also included. Data for the chemical and physical parameters were

collected from open databases such as ChemAxon292 and Chemspider293.

In the PLS modelling, the initial 59 variables were reduced to four by

successively selecting the best representing variable for similar properties in

combination with the relative importance and relationship in the PCA

modelling, equation (3). Models were set up and developed for each test and

for every VRF in the range from 2 to 20, using all data, as well as a grouping

of data into a training set and a test set, Figure 17.

Figure 17: Regression lines for developed PLS models.

PLS models based on the same VRF, showed very good similarity. However,

the accuracy of the models differed for different VRFs. Models based on VRF

10 (ModA:10 and ModB:10 in Figure 17) showed the best conformity to the

observed rejections. The rejection was found to be best described by

equation (3):

Rejection=α+β∗polarizability+γ∗globarity+δ∗phob/polar+ε∗charge (3)

Where α, β, γ, δ and ε were plant specific constants and phob/polar was the

ratio between hydrophobic and polar accessible surface area of the molecule.

The coefficients in the model were dependent on wastewater composition,

Results and discussion | 87

wastewater treatment, water recovery, membrane type and brand, etc. The

model was tested on previously reported data from pilot tests with spiked

tapwater294, since no data were found from tests with regular wastewater. The

training sets and validation sets were used to calibrate and simulate rejection

of different substances. The simulations showed that the proposed modelling

approach can be used, also for other studies, to simulate rejection of

substances by nanofiltration. The rejection model was valid in a range of the

molecular weights MW, around the molecular weight cut-off value (MWCO).

For higher MW, preliminary 2-4 times the MWCO, the clearly determining

factor for rejection was the molecule size. The model is recommended as being

a practical tool for operators of WWTPs and wastewater reuse facilities, to

estimate the rejection of emerging organic compounds, based on their

physiochemical characteristics. The influence of the variables varied between

the VRF’s and the model coefficients must be developed for each specific

WWTP, or water reuse facility, and should be checked up seasonally or yearly.

At low VRFs, removal efficiencies for negatively charged APIs were higher

than at high VRFs. The repulsion of negatively charge APIs comes most

probably from the typically negatively charged membranes.

In summary, the nanofiltration pilot plant gave high removal efficiencies of

APIs, but the achieved 90% average removal efficiency could most probably

have been higher if the delivered nanomembrane had met the manufacturer

specified value of 150 g/mol for molecular weight cut-off (MWCO). The

membrane was shown to have an actual MWCO value of 300 g/mol. The

higher MWCO value of the membrane resulted in poor removal of the smallest

APIs, having MWs in the range of 150-300 g/mol. The developed model for

predicting the rejection of APIs can be a useful tool for operators of a

treatment plant, but also in the estimation of removal efficiencies of not

analyzed substances. However, the model coefficients must be determined for

the individual WWTP, since they are site specific.

4.5 Treatment with GAC and PAC (Paper III and IV)

The aim of the pilot tests was to achieve 95% removal of a selected set of 22

APIs, frequently occurring in Swedish municipal wastewater. This degree of

removal was chosen to have a safety margin to critical environmental

concentrations (CECs) and future effluent quality standards (EQSs). The 22

APIs were selected after mapping effluent wastewater, among 100 preselected

bioactive APIs. The selection criterium was presence of the substance in more

than 50% of the samples.

88 | Present investigation

Two GAC units and two PAC units were operated in parallel in the mobile pilot

plant, consecutively at three WWTPs. The plants were selected to facilitate

evaluation of site-specific conditions, especially wastewater characteristics

and plant configuration. Ozonation was operated in parallel with the GAC and

PAC units (partly described in paper V). Each treatment line was designed for

a hydraulic flow of 100 L/h. The GAC lines consisted of two filter filters in

series, with a valve arrangement to allow backwashing and change of filter

order, after saturation of the first filter. The HRT, or more specifically the

EBCT, was 20 minutes per GAC line. Including the water column above and

below each filter, the total HRT in each GAC line was 36 minutes during

operation. The corresponding HRTs in the PAC lines, including sedimentation

tank and sand filter were 3.5 h, with 60 minutes contact time in all three

contact tanks together, Figure 18.

Figure 18. Schematics of the GAC and PAC treatment systems. The GAC system is composed of

two filter columns in series using down-flow operation. The PAC system is composed of an initial

mixing tank (denoted “M”), three sequential contact tanks, a sedimentation tank and a sand filter.

Sampling points are symbolized “S”. R1, R2 and R3 indicates the different recirculation pipes in the

PAC system, while RX symbolizes an operation without recirculation of settled PAC from the

sedimentation tank.

4.5.1 Selection of GAC and PAC products

A set of 14 activated carbon products was initially selected, based on a broad

range of adsorption specific properties, e.g. specific surface area, iodine

number and particle size. Among the 14 products, in total eight GAC and PAC

products were selected for tests in pilot scale.

Five GAC products were selected for evaluation in pilot scale, based on a

previously performed in-house tests10, but also based on carbon properties

and prequalifying recommendations from manufacturers. Initially, eight

prequalified PAC products were tested in bench scale and evaluated on the

Results and discussion | 89

removal efficiencies of all quantified APIs, among the list of 100 preselected

bioactive APIs10,243,281. Three PAC products were selected for testing in pilot

scale, together with the five GAC products. The screening of PAC products in

lab scale showed different, but typical dose response curves, Figure 19.

Figure 19: Dose response curves from PAC screening in bench-scale with 60 minutes contact time.

To achieve 95% removal of the 22 APIs, the bench scale tests indicated that

the best performing PAC products must be applied with a dose of 30 mg/L,

which consequently was set as the initial dose for the pilot tests. Figure 19

shows the importance of selecting a proper PAC product for treatment of a

specific wastewater. PAC A, B and C in the figure legend above were the

selected products, later used in the pilot test.

4.5.2 Experiences from the pilot plant operation of GAC and PAC

A comparison of the pilot plant performance, with the design values, showed

that the hydraulic capacity of the GAC filters was influenced by the effluent

quality, mainly regarding particles, measured as suspended solids (SS).

During the first tests, all performed at Käppala WWTP, the SS concentration

was below 1 mg/L in the regular effluent and the design hydraulic capacity in

the GAC filters was maintained. This was also the case in filters with small

GAC particle sizes (MESH 12*40), and with the applied backwashing, for

removal of accumulated particles, two times a week. Käppala WWTP has full-

90 | Present investigation

scale sand filters as a final treatment step and these lowered the SS

concentrations below 1 mg/L. After relocation from Käppala to Uppsala

WWTP, the GAC filters clogged after only one day of operation. The backwash

frequency was therefore increased to five days a week. The immediate

hypothesis was that presence of suspended solids in the effluent, partly in

form of flocs from the final chemical treatment for removal of phosphorous,

clogged the GAC filters. Installation of shallow sand filters, upstream the GAC-

lines, partly reduced the clogging, but the hydraulic capacity of the GAC filters

was still 50-80% of the capacity achieved at Käppala. The limited hydraulic

capacity of the GAC filters in the pilot plant remained at Västerås WWTP. By

shifting to a coarser GAC product (MESH 8*30), the hydraulic capacity of the

GAC filters was restored to 90%, but unfortunately the removal efficiency of

APIs was reduced to 75-85%, in-stead of the typically 95%. From this study, it

is unclear if it is the coarse GAC grains themselves, or just the selected product,

that gave the lower removal efficiency. On contrary to the GAC lines, PAC lines

were not influenced by the particles in the regular effluent. The final sand filter

in each PAC line was backwashed once a week and maintained a high

hydraulic capacity at all WWTPs visited.

The risk of clogging of the GAC filters was an important finding during the

pilot tests and must be taken into account before GAC filters are implemented

in full-scale. Typical design values for full-scale filters are hydraulic loads of

6-10 m/h295 (the pilot filters had a design value of 5.7 m/h). Measures to avoid

clogging can be an installation of preceding sand filters, or to find GAC

products that have both coarse grain fractions, and high adsorption capacity.

The PAC lines were in continuous operation and worked well during the tests,

after solving problems with the dosing of PAC-slurry. Initially, clogging of the

narrow hoses (2 mm inner diameter) for PAC slurry dosing, was solved by

installing a sieve on the hose inlet, to prevent coarse grains of PAC to block

hose couplings. The PAC lines were operated without limitations in the

hydraulic capacity, throughout the basin trains. Minor difficulties to control

the feed pumps occurred, due to wear and tear of the helical rotor pumps.

Backwashing of the sand filter was performed once a week, but long-term tests

showed that even after 14 days without backwashing, the capacity of the sand

filter was sufficient for the applied feed flow, corresponding to linear filter

velocity of 2-7 m/h in the sand filters. The applied linear filter velocity was

lower than the value discussed in recent publications295, but was chosen from

experiences from long-term operation of full-scale sand filters in regular

Swedish WWTPs296,297.

Results and discussion | 91

4.5.3 Adsorption of pharmaceutical residues - GAC

Two GAC lines were operated in the pilot plant to evaluate two GAC products

in parallel. In total five GAC products were tested during the pilot test. GAC

A—E correspond to product/supplier; GAC A: Aquacarb 207C/Chemviron;

GAC B: Aquasorb 5000/Jacobi; GAC C: Filtrasorb 400/Chemviron; GAC D:

GPP-20/Chemviron; GAC E: Carbsorb 30/Chemviron. GAC A and D were

used in Käppala, GAC B was used in Käppala and Uppsala, and GAC C and E

were used in Uppsala and Västerås.

Treatment of wastewater with GAC filters is a long-term process, with

continuous adsorption and desorption of organic substances onto the internal

surfaces of the activated carbon, trying to reach equilibrium after changes in

wastewater composition.

The adsorption in the GAC filters was followed by breakthrough curves for

each API, where the ratio of concentration in the advanced treated and the

regularly treated wastewater, (C/C0), was plotted versus the number of passed

bed volumes (BV). The patterns of the breakthrough curves differed for the

APIs but also among GAC products. To demonstrate this, breakthrough curves

for three indicator APIs, stated in EU and Switzerland, plus a fourth API,

irbesartan were compared, Figure 20.

The comparison showed that GAC A and E had the lowest removal capacity,

i.e. breakthrough occurred earlier, at a lower number of treated BV, than for

the other products, GAC B, C and D. These products had more than five times

as high capacity to adsorb clarithromycin and irbesartan than GAC A and E.

The ratio C/C0 scattered mainly due to variations in the API concentration in

the WWTP effluent. GAC E was selected and used to improve the hydraulic

capacity of the GAC-filters, as described above, and unfortunately of the

expenses of adsorption capacity.

Breakthrough curves for all 22 substances were plotted to determine the

specific adsorption capacity of the GAC products. Generally, GAC A and

GAC E showed lower specific adsorption capacities, compared with the other

products, except for high adsorption capacities for carbamazepine and

trimethoprim.

92 | Present investigation

Figure 20: Breakthrough curves from the GAC pilot experiments. Breakthrough, C/C0, the ratio of

concentration in advanced treated wastewater, C and the concentration in the regularly treated

wastewater C0, is plotted against treated bed volumes (BV) and displayed for carbamazepine,

clarithromycin, diclofenac and irbesartan. Dashed lines between the discrete samples are added to

improve visual interpretation.

The different GAC products adsorbed individual API very differently, even

though most APIs contain one or more aromatic groups, which lead to the

conclusion that not only aromaticity of the molecules contributes to

adsorption onto activated carbon. For the individual APIs, the specific

adsorption capacities, varied between 0.03 μg fexofenadine per g dry GAC, to

23 μg metoprolol per g dry GAC, at a breakthrough level (C/C0) of 0.05. The

adsorbed specific amount of APIs was approximately linear versus the number

of treated bed volumes, Figure 21, which indicates that the for the sum of the

selected 22 APIs has not reach its maximum value. The values of specific

adsorption capacities achieved were much lower than previously reported

values in distilled water and surface waters298. This indicates the matrix effect,

when wastewater bulk organic materials compete with APIs in the adsorption

onto the available adsorptive surfaces of the activated carbon. This is a

reasonable hypothesis, looking at the average removal of total organic carbon

(TOC) from the regular effluent, which was 43% in the GAC-lines.

Results and discussion | 93

Figure 21: – Accumulative load (solid line) and uptake (dashed line) for the sum of the 22

pharmaceuticals during the GAC pilot-scale experiments. Vertical dashed lines indicate change of

location, from one WWTP to another WWTP.

The average carbon usage rate (CUR) of GAC, to remove 95% of the selected

APIs, was 110 g/m3 wastewater, for all five GAC products and at all three

WWTPs studied. The corresponding carbon usage rate for the three, best

adsorbing GACs were: GAC B<43 g/m3 (at 6440 BV), GAC C<41 g/m3 (at

10210 BV) and GAC D<28 g/m3 (at 16130 BV). Final values could not be

determined, since the test periods were too short to reach 5% breakthrough,

corresponding to 95% removal efficiencies of the selected APIs. The remaining

two products A and E showed much higher carbon usage rates: GAC A 230

g/m3 (at 2290 BV) and GAC E 170 g/m3 (at 2360 BV). No general explanation

for the differences in carbon usage rates and adsorption capacity was found

from the available GAC product characteristics. However, the author observed

that a higher methylene blue number (MBN) corresponded well to a higher

94 | Present investigation

adsorption of APIs, i.e. products GAC A and E both have an MBN of 230 and

showed lower removal than GAC B and C with an MBN of 260. Consequently,

MBN should be further evaluated, as a guiding parameter in the selection

process of GAC products for removal of APIs.

4.5.4 Adsorption of pharmaceutical residues - PAC

Three different PAC products were used in this work. PAC A—C correspond

to product/supplier; PAC A: Pulsorb C/Chemviron; PAC B: Aquasorb

MP20/Jacobi; PAC C: Aquasorb 5000P/Jacobi. PAC A was used in the tests

at Käppala WWTP and Uppsala WWTP, PAC B was used in tests at all three

plants and PAC C was used in tests at Uppsala WWTP and Västerås WWTP.

The applied fresh PAC dose was initially set to 30 g/m3, based on lab tests, and

then stepwise lowered during the treatment campaigns. The dose of 30 g/m3

was more than sufficient to reach 95% overall removal, in both Käppala and

Uppsala, while 94% overall removal was achieved in Västerås, by applying a

reduced fresh dose of 15 g/m3. The accumulation of PAC product in the

treatment tanks due to recirculation, contributed to an increase of the removal

efficiencies of APIs. The average specific adsorption capacity at a

breakthrough of 0.05, corresponding to 95% removal efficiency for the three

PAC products, varied between 1.0 µg/g (clarithromycin) and 23 µg/g

(metoprolol), the latter had the same high value as for the best GAC product.

4.5.5 Removal efficiencies of individual pharmaceutical residues

The removal efficiencies of individual APIs were related to the carbon usage

rate (CUR) for both GAC and PAC. CUR for GAC was calculated relating the

loaded amount of GAC to the treated volume of wastewater. In the PAC

systems, the accumulated amount of PAC was related to the treated volume of

wastewater, Figure 23.

Removal efficiencies of individual for GAC products

In order to evaluate the removal efficiency for the individual substances

related to CUR, results from all discrete calculations of removal efficiencies

were divided into dose intervals. The individual removal efficiencies of APIs

showed that 8 out of 22 substances were removed more than 95% at a carbon

usage rate of 30-100 mg/L i.e. 30-100 g/m3, for all experiments and GAC

products, Figure 22.

Results and discussion | 95

Figure 22: Average removal efficiencies of APIs for the five GAC products in the pilot experiments

divided into carbon usage rate intervals.

The indicator APIs, carbamazepine and clarithromycin were removed more

than 95% and diclofenac to 93% in this carbon usage rate interval. A carbon

usage rate, which exceeded 100 mg/L, did not result in a considerable

improvement of removal, mainly due to poor removal efficiencies for GAC A

and E. In the present test, the removal efficiencies were in the same range or

higher than in other reported evaluations of GAC performance in regular

effluent wastewater10,299,300. Comparisons of removal efficiencies can be

difficult to make, since the number of treated bed volume, at the sample

occasions, at pilot or full-scale plants, are sometimes missing in published

papers.

Removal efficiencies of individual for PAC products

Analogous to the evaluation of removal efficiencies in the GAC pilot tests, all

the PAC experiments were sorted in PAC dose intervals. The individual

removal of APIs showed that 16 out of 22 substances were removed more than

95% by a carbon usage rate of <30-100 mg/L for all experiments and the three

PAC products. Concerning the indicator APIs, carbamazepine, clarithromycin

and diclofenac, all were removed more than 95% in the carbon usage rate

interval 30-100 mg/L, Figure 23.

96 | Present investigation

Figure 23: Average removal efficiencies of APIs for the three PAC products in the pilot experiments

divided into carbon usage rate intervals.

Seven of the 22 APIs were removed less than 95% at all CURs, although

fexofenadine, bupropion and venlafaxine were removed by 94%, close to the

target value. Fluconazole, mirtazapine, diltiazem and memantine were clearly

removed less than 95%, but PAC removed these substances better than GAC

did, by approximately 10 %-units. However, the added PAC dose in most

experiments with was 26—30 mg/L, which was in accordance with the lab test,

performed to select proper PAC products and doses. By the last pilot tests,

which were performed in Västerås, the dose was lowered from 26 mg/L to 15

mg/L. By this decrease in PAC dose, the average removal efficiency dropped

from 99% to 95%. The removal efficiencies were higher than, or similar to,

previously reported results from PAC applications301,302.

4.5.6 Mechanisms for adsorption of APIs on GAC and PAC

In general, the specific adsorption of APIs was higher on PAC than GAC, most

probably due to the smaller particle size of PAC, which makes the pores more

accessibly for diffusion and adsorption of APIs. The comparison of GAC and

PAC in this work was based on all products tested. Still, if the two low-

performing GAC products were excluded, PAC showed a higher specific

adsorption, although less pronounced.

A ranking of the removability of the 22 APIs, where the results from GAC and

PAC tests were combined, Table 14, showed that the first nine APIs, i.e. with

the highest removability, are all positively charged at normal wastewater pH.

This indicates that the net surface charge of the activated carbon was negative

when it was in operation, likely as a result of co-adsorption of organic matter

Results and discussion | 97

present in the effluent wastewater. The higher removability of positively

charged APIs by activated carbon has previously been reported176,302.

Table 14. Ranking of removability, equally based on removal efficiency and % below LOQ. Molecular

structures are show in declining removability, ordered left-to-right then top-to-bottom.

The four best ranked substances, bisoprolol, atenolol, sotalol and

trimethoprim, were all positively charged at the actual wastewater pH 6.7-7.7

and they all contain at least one aromatic group and hydrogen-bond

donor/acceptor groups, Table 14.

The four poorest ranked substances were memantine, fluconazole, irbesartan

and clindamycin. Memantine is positively charged at actual wastewater pH

6.7-7.7, has a low molecular weight and a compact molecular structure and is

not aromatic, the latter has proven to be a disadvantage for adsorption onto

activated carbon303. Fluconazole is neutrally charged at wastewater pH and is

therefore more dependent on hydrophobic interactions with the activated

carbon, than positively charged substances. The low log D value of fluconazole

indicates predominantly hydrophilic properties, which are unfavorable for the

adsorption. Irbesartan is negatively charged, which contributes to a low

Pharmaceutical GAC PAC

Bisoprolol 4 1

Atenolol 1 6

Sotalol 2 4

Trimethoprim 4 3

Clarithromycin 8 2

Metoprolol 5 6

Citalopram 6 5

Codeine 7 8

Tramadol 9 11

Diclofenac 15 9

Bupropion 12 13

Mirtazapine 12 14

Flecainide 11 15

Diltiazem 16 10

Oxazepam 9 18

Fexofenadine 18 11

Carbamazepine 14 17

Venlafaxine 17 16

Clindamycin 21 18

Irbesartan 21 20

Fluconazole 20 21

Memantine 18 22

98 | Present investigation

adsorption. Clindamycin is positively charged at wastewater pH, is not

aromatic and it has a moderate to low log D value, that can explain the low

adsorption.

In summary, activated carbon, in form of treatment lines with GAC and PAC,

was shown to be an efficient technology for the removal of APIs in the effluents

of municipal wastewater treatment plants. In the designed pilot treatment

lines, 95% removal efficiency was reached for almost all tested substances.

Large variations in specific carbon usage rates were observed between GAC

products, but also between different APIs.

The APIs to be regulated in effluent standards will influence the selection of

GAC and PAC products to a specific WWTP and a variation of both GAC and

PAC qualities should be screened before acquisition, to find products with

lower carbon usage rate and thereby lower cost. Concerning GAC, screening

with respect to hydraulic capacity is also important before purchase.

PAC seemed to have a generally higher removal efficiency at a lower carbon

usage rate than GAC, but well-performing GAC products approached the

performance of PAC, as indicated by the results.

4.6 Examination of process parameters in the PAC lines (Paper

IV)

The PAC lines were designed based on literature data and bench-scale tests of

the dependency of contact time and removal efficiency. The comparable

testing in pilot scale described above showed that a fresh PAC dose of 15-16

mg/L, a contact time of 60 minutes and recirculation of used PAC to the first

contact tank was sufficient to remove 95% of the 22 APIs in the study. In the

pilot tests, the PAC dose was successively lowered from 30 mg/L and finally

levelled out at 15 mg/L to maintain 95% removal. The research question below

was raised to examine the dependency of contact time and recirculation on the

removal efficiency of APIs in a PAC system.

4.6.1 Contact time in contact tanks for wastewater and PAC in

pilot scale

Three contact times in the contact tanks were evaluated: 30, 60 and 120

minutes, at a fresh PAC dose of 15 mg/L and under recirculation of used PAC.

The corresponding contact time in one contact tank was 10, 20 and 30 minutes

respectively. The overall removal efficiencies of the APIs were marginally

Results and discussion | 99

improved by a longer total contact time than 30 minutes. The removal

efficiency increased from 95% at 30 minutes, to 97% at 60 minutes and 98%

at 120 minutes. A long contact time was however beneficial for the removal

efficiency of clindamycin, fluconazole, irbesartan, oxazepam and tramadol.

They all approved the 95% removal level, when increasing the contact time

from 60 to 120 minutes. Memantine and venlafaxine did not reach more than

80% removal efficiency, even at the longest contact time.

Similar test with 60- and 120-minute contact times were performed, but

without recirculation of used PAC. The use of recirculation increased the

overall removal efficiency of APIs from 92 to 97%, at 60 minutes contact time,

and from 96 to 98%, at 120 minutes contact time. A comparison of removal

efficiencies for individual APIs in systems, with and without recirculation of

used PAC, showed that recirculation was beneficial for the removal of

carbamazepine, fluconazole, Irbesartan, sotalol, tramadol and trimethoprim,

Figure 24. Experiences from other studies showed that 50-67% lower doses of

PAC can be applied through recirculation of PAC from different separation

units, following the contact tanks301,304.

Figure 24: Effects on the removal comparing operation with recirculation to the first contact tank (R1) and without (RX) at nominal contact times of 60 min (top) and 120 min (bottom). Values below 95% removal and below the LOQ are marked with an asterisk (*). Error bars indicate the standard deviation of duplicate experiments.

100 | Present investigation

Recirculation had a relative larger positive effect of removal efficiency of APIs

at the shorter contact time of 60 minutes. In summary, the aim of 95% overall

removal of the APIs was reach by a fresh dose of 15 mg PAC/L, at a contact

time of 30 minutes and with recirculation of used PAC. However, the removal

efficiency of some less readily adsorbed APIs, increased with longer contact

times and applied recirculation of used PAC. Again, the selected set of APIs in

effluent standards will influence the design of the PAC process at a specific

WWTP.

4.6.2 Recirculation release point of used PAC

The influence on removal efficiency of different recirculation release points,

i.e. discharge of the recirculation stream to any of the three contact tanks

(configuration R1-R3), was evaluated at 30 min contact time and at fresh PAC

dose of 15 mg/L. The removal efficiency was similar for all three recirculation

points, with a slightly higher removal (95-96%) for recirculation to the first

(R1) or second tank (R2) in series, compared to recirculation to the last tank

(R3), which resulted in 94% overall removal efficiency, Figure 25.

Figure 25: Removal using the different recirculation configurations at 15 mg/L fresh PAC dose and 30 min nominal contact time. Values below 95% removal but decreasing the effluent concentrations below LOQ are marked with an asterisk (*). Diclofenac was below the LOQ in the WWTP effluent used in test R2. Error bars indicate the standard deviation of duplicate experiments.

Generally, a significant difference for the removal efficiency of individual

APIs, due to recirculation point, could not be observed, but some APIs,

clindamycin, irbesartan and tramadol, were mostly influenced by the position

of the recirculation point, R1 was favorable over R2 and R3, Figure 25. Despite

the expectations of a better performance in configuration R1, the removal in

R2 was generally similar or slightly better for five out of nine APIs, while three

of nine APIs performed better in R1. In summary, comparison of different

Results and discussion | 101

discharge points for the recirculation showed that recirculation to R2 resulted

in the best performance with respect to removal efficiencies of individual

substances.

With regards to 95% overall removal efficiency, the classical setup, i.e.

recirculation of used PAC to R1 and R2, must be used. Regarding the indicator

APIs, the only example of insufficient removal was for carbamazepine in

configuration R3, where the removal efficiency reached 92%.

4.6.3 Retention of PAC in a pilot plant – coagulation and

flocculation

To achieve a high removal efficiency of APIs, the used PAC must be retained

in the treatment unit. A discharge of PAC residues will carry adsorbed APIs to

the next treatment step or to the receiving water. Additionally, an internal

recirculation of PAC can be valuable to increase the retention of PAC, mainly

through flocculation and sedimentation. Coagulants and flocculation aids

have been applied in contacts tank to increase the retention of PAC176,305.

Initial, separately performed pilot tests in this work showed that with an

addition of both Al and Fe-based coagulants to the third contact tank, a faster

blocking of the sand filter occurred. In connection with the evaluation of

recirculation of PAC, a few experiments were performed with continuous

addition of poly-aluminum chloride (PAX-XL350, Kemira) to the third

contact tank, at 60 min contact time (configuration R1). Coagulant was added

as 4 mg/L Al2O3 and the fresh PAC dose was 12 mg/L. The overall removal

efficiency of APIs was somewhat higher, 98% compared to 97%. The hydraulic

capacity in the final sand filter was however significantly reduced. In

opposition to the hypothesis, the PAC-sludge was less well retained in the

treatment tanks, due to a bulkier sludge appearance. The major parts of the

pilot tests were therefore run without coagulation and flocculation aids.

4.6.4 Treatment of secondary or tertiary effluent with PAC

Käppala WWTP has a sand filter step as a tertiary treatment to remove

suspended solids and particle bound phosphorous. Uppsala WWTP and

Västerås WWTP have not installed sand filters, which otherwise had produced

tertiary effluent, so the latter WWTPs discharge different types of secondary

effluent. Many WWTPs lack a tertial treatment, so the pilot tests are valuable

also from this perspective. During the pilot tests, it was obvious that the sand

102 | Present investigation

filter removed particles, that otherwise had interfered with the GAC filters,

while the PAC lines were not apparently affected.

During the evaluation of PAC process parameters, the average concentration

of suspended solids in the secondary or tertiary effluent were 10 and 2 mg

SS/L respectively. The hydraulic capacity in the PAC-lines was not influenced

by the higher load of suspended substances, so the frequency of backwash in

the final sand filter once every week was maintained.

In addition, the removal efficiencies of APIs seemed not to have been

influenced by the higher concentrations of organic material in the secondary,

than the tertiary effluent. Reconsider that the tests on tertiary effluent were

performed with substantially higher fresh PAC doses, 25-37 mg/L, than the

tests on secondary effluent, where the average dose was 15 mg/L, tests with

lower PAC doses on tertiary effluents must be performed as well.

4.6.5 Influence of mixing with air on the pH in the contact tanks

During the experiments, a rapid increase in pH was observed in samples taken

from the pumped effluent wastewater (pH 6.9), to the first (pH 7.3), second

(pH 7.7) and third (pH 7.8) contact tank, which was hypothesized to be caused

by stripping of carbon dioxide, which in turn is a result from the use of

aeration as mixing aid. Bench-scale experiments showed a similar trend, when

aeration was used for mixing, while the change in pH with mechanical stirring

was negligible. pH influences the log D value differently for different APIs,

which might influence the adsorption. No considerable difference in

adsorption is however expected, due to the probability that PAC surfaces have

a predominantly negative charge at pH 7, mainly because of co-adsorption of

organic matter, present in the wastewater.

4.6.6 Conclusions from the evaluation of PAC parameters

The PAC-based systems, with recirculation, in a series of three tanks, achieved

high removal of APIs, both from secondary and tertiary effluents.

Recirculation to improve the removal of APIs or lower the fresh PAC dose to

maintain a proper removal efficiency, should be either established to the first

or the second tank. Physical retention of PAC, e.g. by use of sand filtration is

required to maintain a high recirculation degree of PAC, but also to prevent

PAC with adsorbed APIs to reach the effluent, which will lower the removal

efficiency. Depending on the effluent standards and PAC product selected, the

Results and discussion | 103

contact time and PAC dose must be varied to achieve different degrees of

removal efficiency, both with respect to the overall removal of

pharmaceuticals, and with respect to the removal of individual APIs.

4.7 Treatment with ozone (Paper V)

Ozonation was the third technology to be evaluated for removal

pharmaceutical residues from wastewater, in addition to nanofiltration and

activated carbon. The aim for the tested removal technologies was to achieve

95% removal of a selected set of 22 APIs, initially detected as frequently

occurring in Swedish municipal wastewater.

During the field tests, two ozonation units were operated in parallel in the

mobile pilot plant when it was in operation at Käppala WWTP, Uppsala

WWTP and Västerås WWTPs. The ozonation units were operated also in

parallel with the GAC and PAC units (described in paper III-IV), to facilitated

comparison of different treatment technologies for removal of pharmaceutical

residues.

4.7.1 Removal efficiencies of pharmaceutical residues by

ozonation

Paper V describes the results from a relative short period, two weeks out of 57

weeks of the pilot tests, when ozonation for removal of APIs was evaluated at

Käppala and Uppsala WWTPs, in connection with ecotoxicity tests. The first

week, at Käppala WWTP, is more representative for the ozonation tests during

the remaining 55 unreported weeks, than the presented second week, at

Uppsala WWTP, when problems with one feed pump, disturbed the ozone

dosing and prolonged the hydraulic retention time.

During the described periods in paper 5, ozonation line 1 was operated with

the 5 m high contact column 1 (CC1), which had an operational volume of 83

L, where the reaction between ozone and organic substances occurred during

on average 36 and 47 minutes (Käppala WWTP and Uppsala WWTP,

respectively). The longer hydraulic retention time in Uppsala was an effect of

lost capacity of the feed pump. The ozonated wastewater passed the aerated

stripping tank for the removal of potential residues of free ozone, that

otherwise might interfere with the biotests. During the tests at Käppala

WWTP, parts of the ozonated wastewater also passed a sand filter with a

surface load of 2.7 m/h, for a potential removal of biodegradable products,

104 | Present investigation

released in the ozonation during the tests. In Uppsala the same pilot sand filter

was used to study the effects of biomarkers in rainbow trout, only adding a

sand filter to treat the regular effluent wastewater.

During the tests, the ozone doses were 7 g/m3 (specific ozone dose 0.92 g O3

per g TOC) in Käppala WWTP and 5.4 g/m3 (specific ozone dose 0.82 g O3 per

g TOC) in Uppsala WWTP. The average removal efficiencies of the 22 APIs

were 89% and 87% in Käppala WWTP and Uppsala WWTP, respectively.

These results were mainly in accordance with the results from a published lab

study on effluents from six Swedish WWTPs, spiked with APIs306. The average

removal of 13 APIs present in regular effluent from three of the six WWTPs in

the lab study were 81% and 71%, at the corresponding ozone doses of 7.0 and

5.4 g/m3, respectively. The corresponding removal efficiencies in Käppala and

Uppsala were 81% and 79%, respectively. The three selected WWTPs had

similar average DOC concentrations in their effluents, 9.1 g/m3 in both lab and

pilot groups of the WWTPs. A direct comparison of the 13 selected APIs, shows

that the removal efficiencies of the individual APIs were scattered between the

two groups of WWTPs, Figure 26.

Figure 26: Correlation of removal efficiencies in lab scale test (literature data) and performed pilot

tests at Käppala and Uppsala WWTPs. The ozone doses are marked with orange for 5.4 g/m3 and

blue for 7 g/m3 respectively.

Results and discussion | 105

The results indicate that the removal efficiency was higher in pilot scale for

many APIs, than in lab scale at an ozone dose of 5.4 g/m3, but similar results

were achieved in both scales at an ozone dose of 7 g/m3. One reason for this

disagreement might be the relatively longer retention time in the contact tank

at 5.4 g/m3 in the pilot tests in Uppsala.

At Käppala WWTP, the ozonation resulted in 89% overall removal efficiency

of APIs by the addition of 7 g/m3, while the number of quantified APIs was

reduced from 24 to 3 by the ozonation. At Uppsala WWTP, the ozonation of

effluent wastewater resulted 87% overall removal efficiency of APIs by the

addition of 5.4 g/m3 and the number of quantified APIs was reduced from 25

to 10 by the ozonation. It seems that the level of ozone dose is more important,

than the retention time in the contact tank, to remove as many APIs as

possible. The removal efficiencies of individual APIs were not uniform at a

specific ozone dose, but differed from the average, Figure 27.

Figure 27. Removal efficiencies of the APIs quantified in effluent in Käppala WWTP and Uppsala

WWTP during the pilot test during the evaluation with biomarkers.

Seven of the 26 APIs quantified were removed to more than 95%. Half the

number of the 26 APIs had a removal efficiency between 80-95%. In the

interval 50-80% removal efficiency, five APIs were observed, among them

flecanide and fluconazole, which are reported to be only moderately removed

by ozonation307.

106 | Present investigation

Mechanisms for removal of APIs by ozonation

The reason for the relatively low removal of flecanide and fluconazole can be

that the molecules have electron-withdrawing fluoro groups, that also shield

from ozone attack. Bupropion was the only API removed less than 50%. The

poor removal of bupropion is not apparent, simply by looking on the chemical

structure, since bupropion contains an aromatic functional group and should

be vulnerable for an attack of ozone. The limited removal seems to be a

question of ozone dose in relation to DOC concentration, which in the two

pilot tests was 0.68 g ozone per g DOC on average, but more likely one

disregarded shielding effect or simply deconjugation of some of the in human

formed metabolites, back to the parent substance. In the controlled lab study

on real wastewater effluent, but spiked with APIs to individual concentrations

of 1 µg/L, a similar specific ozone dose of 0.74 g per g DOC resulted in 45%

removal of bupropion, which is moderately higher than the achieved results

in pilot scale306. The low ng/L concentration of bupropion in un-spiked

effluent might also have influenced the lower removal efficiency, due to lower

reaction rates at lower concentrations according to first order kinetics.

4.7.2 Selection of ozone dose and hydraulic retention time in

contact column

Dose-response tests were done in the pilot plant during the tests at Käppala

WWTP to find a proper ozone dose to remove 95% of the APIs. By previous in-

house ozonation tests it was shown that at high ozone doses, e.g. 15 g/m3,

caused adverse effects in test organisms, so as low dose of ozone as possible to

remove 95% of the APIs was favorable109, but also beneficial from the

perspective of resource consumption. Based on these reflections, an ozone

dose of 7 g/m3 (0.77 g per g DOC) was selected as target value for the long-

term operation in Käppala WWTP and Uppsala WWTP, although the average

removal of APIs was 90% at 7 g/m3 in the dose-responds tests and thus below

the goals of 95% removal efficiency.

As a comparison, the lab tests described in connection with Figure 26,

demanded 0.93 g O3 per g DOC to achieve 90% removal efficiency of 13

overlapping APIs. The applied specific ozone dose 0.77 g O3 per g DOC, during

the pilot test, resulted in 80% removal efficiency of 13 overlapping APIs.

The ration between these doses and corresponding removal efficiencies

confirms earlier observations that the relation of ozone dose to removal

Results and discussion | 107

efficiency is not linear, but logarithmic, more apparent at higher removal

efficiencies, >80%.

During the dose-response tests, the hydraulic retention time (HRT) in the

contact column was 36 minutes, equal to the total HRT in one GAC line. The

applied HRT in the ozonation contact column was based mainly on preceding

tests in the mobile pilot plant, where the necessary HRT showed to depend on

API concentrations i.e. a greater sum of concentrations of APIs in the effluent

demanded a longer HRT (unpublished data). Strictly from this evaluation, the

HRT could have been set to 30 minutes, but to add a margin of safety for

occasional high concentrations of APIs during the biotests, 36 minutes was

chosen as HRT in the ozonation contact tank.

By ozonation, the pilot tests showed that the kinetics of degradation of

individual APIs varied substantially. Again, the selection of targeted

substances, and their concentrations, will influence the applied treatment,

here in form of the length of the HRT in the ozonation contact tanks

(unpublished data). During the full period of pilot tests, the dependence of pH,

temperature and wastewater pressure on removal efficiency, were tested and

evaluated in addition to the effect of ozone dose and HRT (data not shown).

In summary, the aim of 95% overall removal of APIs was not consistently

reached by ozonation, but with an appropriate ozone dose of 5-7 g/m3,

ozonation reach 85-90% overall removal efficiency of the selected APIs. The

ozone dose must be adopted in relation to the concentration of bulk organic

substances in the wastewater, to amount at least a ratio of 0.77 g O3 per g DOC,

to reach 89% removal efficiency of 24 selected APIs. The HRT in the ozonation

contact tanks had to be 30 minutes for the concentration level of APIs during

the biotests. The sand filter, following the ozonation, did not improve the

removal of pharmaceuticals, but the TOC concentrations decreased by 5%.

Fluconazole and bupropion were the most resistant APIs to ozonation.

108 | Present investigation

Area 4) Does the present discharge of APIs lead to biological

consequences in the aquatic environment?

4.8 Evaluation of treatment technologies by biomarkers

(Paper V)

To validate the removal efficiency of advanced treatment technologies,

chemical analysis of pharmaceutical concentrations is crucial. It is however a

risk that substances, not being analyzed or potentially formulated as

byproducts in ozonation, will be harmful in the aquatic environment. Use of

biomarkers, complementary to the chemical analysis, offers a good indication

of the presence or production of harmful substances in untreated or treated

wastewater.

Biomarkers reflect interaction between a biological system and a potential

chemical, physical or biological hazard and were used in parallel with chemical

analysis, to evaluate the effluents from ozonation and activated carbon.

Biomarkers used in the evaluation of polluted water are often linked to the

metabolism of hazardous substances.

The set of biomarkers applied in this study, integrates responses both of

known and unknown chemical substances. The biomarkers were measure

both on mRNA and protein level and were chosen to span the effects of a large

variety of possible substances, i.e. aryl hydrocarbons estrogens (EROD,

CYP’s), metals (MT), oxidative stress inducers (HSP 70, SOD) and pregnane

X receptor agonists (CYP3A45). Rainbow trout (Oncorhynchus mykiss) was

used as biological system and samples were taken from both gills and liver to

allow measurement from both a more rapid and slower reacting system.

In total three exposures tests on rainbow trout were performed, one on lab

scale treatment in Käppala WWTP, one on pilot scale treatment in Käppala

WWTP and one on pilot scale treatment in Uppsala WWTP. The initial

exposure test in Käppala was performed on effluents from lab scale setups of

ozonation and activated carbon, which was constructed to supply supporting

data for the design of the treatment lines in pilot scale. The initial exposure

test was also used to select relevant biomarkers for the subsequent exposure

tests in Käppala and Uppsala.

In the lab scale ozonation, the added ozone dose was 7 g O3/m3 and it resulted

in 65% average removal efficiency of APIs, which indicated a limited transfer

of ozone to the wastewater, likely due to the low column (0.9 m). Later

performed pilot scale tests, with addition of 7 g O3/m3 in a 5 m high column,

Results and discussion | 109

gave 87-90% removal efficiency, showing the shortcomings of a low contact

column. The activated carbon filtration (GAC) resulted in 99.8% removal of

APIs.

The relatively poor removal of APIs in the lab scale ozonation plant was also

reflected in some biomarker responses308. The liver induction of CYP1A1 and

CYP1A3 transcription in fish exposed to ozonated water was probably due to

remaining persistent aryl hydrocarbon receptor agonists. This has been shown

earlier, when the ozone dose was insufficient to remove the target

substances309.

In pilot scale, the first exposure of rainbow trout to untreated and treated

effluent was performed at Käppala WWTP for one week, when ozonated,

ozonated plus sand filtered and granular activated carbon treated effluents,

were compared with the negative control; tap water and the positive control:

effluent from the wastewater treatment plant.

Chemical analysis showed that both activated carbon filtration and ozonation,

with the addition of 7 g/m3, resulted in 89% average removal of APIs in the

effluent wastewater. These treatments significantly reduced EROD activity to

levels below that recorded in rainbow trout held in regular effluent and even

below that in fish held in tap water.

The induction of CYP1A1 and CYP1A3 transcription in the liver in fish exposed

to effluent from the lab scale ozonation described above, was significantly

reduced in the pilot scale ozonation, likely due to a taller contact column,

which facilitated a higher removal of AhR agonists. Furthermore, the

indicated increase in transcription of heat shock protein 70 (HSP70) and

metallothionein (MT) in gills of fish, exposed to the ozonated effluent in the

lab study, could not be confirmed in the pilot scale, indicating that the

ozonation did not induce oxidative stress in the studied biological system

The second major exposure on rainbow trout of effluent from pilot scale

treatment lines with sand filter, ozonation and granular activated carbon was

performed for one week at Uppsala WWTP. Tap water was used as negative

control and regular effluent was used as positive control. During the exposure,

the ozonation, with the addition of 5.4 g/m3, resulted an 87% removal

efficiency and activated carbon filtration removed 95% of the APIs.

Fish exposed to regular effluent, with or without sand filter, showed

significantly induced gill EROD activity and increased the transcription of

cytochrome P450 (CYP1As and CYP1C3) in liver, compared to activity and

transcription in fish held in tap water. This has also been observed in other

110 | Present investigation

studies309,310. EROD activity in fish exposed to GAC-treated effluent was lower

and EROD activity in fish exposed to ozonated water was slightly, but

significantly higher than in those exposed to tap water.

Ozonation means addition of very oxidative ozone into the wastewater, which

led to the application of two relevant biomarkers, the heat shock protein 70

(HSP70) and the superoxide dismutase (SOD), which response to oxidative

stress in fish. The tests did not show any increased oxidative stress response

in the fish, due to ozonation in pilot scale, on contrary to the results in lab

scale, where the ozone transfer was limited. Previous studies309 have shown

increases in HSP70 responses in ozonated effluent, at ozone doses of 5 and 15

g/m3, but it was not observed in the pilot tests in Käppala and Uppsala. The

ozonated effluents were however stripped by aeration to remove free ozone,

thus no effect could be expected from the O3 molecule itself. The conclusion

from the ozonation tests were that xenobiotic byproducts and/or oxygen

radicals were not present in concentration high enough to evoke response

from the chosen biomarker in gill and liver.

In summary, biomarker responses indicated that the effluents from WWTPs,

using regular treatment, contained substances to a concentration that elevated

some of the biomarkers (EROD, CYP1’s), specifically responding to aryl

hydrocarbon receptor (AhR) agonists. However, these responses disappeared

by the applied advanced treatment technologies. GAC treatment and

ozonation significantly decreased the concentrations of APIs present in

regular effluent water. Furthermore, ozonation did not increase the oxidative

stress responses in the fish. Consequently, GAC treatment and ozonation are

considered to be suitable and implementable treatment technologies to

remove biologically active contaminants in wastewater. To increase the

validity of this conclusion, repeated and additional tests with biomarker

responses, especially concerning chronic toxicity, are recommended.

4.9 Comparison of treatment technologies

The tested treatment technologies have shown good overall removal

efficiencies for APIs, but for nearly all technologies, limitations in the removal

efficiencies of some individual APIs were observed. A comparison of the

removal efficiency for the total set of APIs, shows that the overall removal

efficiency was higher than 80% for all tested technologies, Table 15.

Results and discussion | 111

Table 15. Ranking of APIs in descending order of average removal efficiency (RE). O3= Ozonation;

GAC=Granular Activated Carbon; PAC=Powdered Activated Carbon and NF=Nanofiltration.

Pharmaceutical O3

RE [%]

GAC

RE [%]

PAC

RE [%]

NF

RE [%]

Average

RE [%]

Alfuzosin 92 - 99.6 - 96

Trimethoprim 97 89 99.9 90 94

Clarithromycine 92 90 99 - 94

Carbamazepine 98 85 98 - 94

Atenolol 87 94 98 95 94

Tramadol 98 88 95 92 93

Codeine 90 84 99.6 99 93

Sotalol 91 88 99 - 93

Citalopram 97 82 94 96 92

Venlafaxine - 84 99 - 92

Clindamycine 96 79 99 - 91

Metoprolol 91 92 99.7 77 90

Bisoprolol 86 82 99.7 - 89

Fexofenadine 90 82 95 - 89

Irbesartan 83 86 97 - 88

Diclofenac 96 72 98 88 88

Flecainide 77 88 99.5 - 88

Mirtazapine 82 81 98 92 88

Ketoprofen 86 - - 89 88

Oxazepam 71 85 99 87 85

Diltiazem 80 72 93 - 82

Fluconazole 58 74 91 - 74

Memantine 68 66 88 - 74

Bupropion 28 50 98 - 58

Average 84 82 97 90 88

The results represent the outcome of typical operation of pilot plants and in

the case of GAC, the average for the five tested GAC products is reported,

including removal data for two low performing GAC products. The most

striking result is the outstanding high removal efficiency, for most APIs,

achieved by the PAC lines. The substances in Table 15 are sorted by average

112 | Present investigation

removal efficiencies, in descending order, with the purpose to show the easiest

removable substance at the first row.

In the list above, the APIs reported to have adverse effects on aquatic

organisms: diclofenac, oxazepam and the antibiotics (trimethoprim,

clarithromycine and clindamycine), were removed more than 80% by any of

the tested technologies, with the exception of ozazepam, which was removed

only by 71% applying ozonation. However, the other technologies tested

removed ozazepam in the range of 85%-99%. The LOQs of the applied

analytical procedure were too high, to give values of the hormones reported to

have adverse effects in aquatic organisms.

Three substances were removed less than 80% in average, due to relatively

poor removal by ozonation and GAC. Fluconazole was, as discussed earlier,

less effectively removed by ozonation, due to flouro groups in the molecule.

The relatively poor average removal by GAC compared to PAC, comes from

the two less well performing GAC products, of the five products tested, just

comparing the three well-performing GACs, the same removal efficiency as

PAC was reach, 91%, again showing the necessity to screen for an appropriate

product before purchase. Memantine is, as described above positively charged

at wastewater pH, has a low molecular weight and a compact molecular

structure and is not aromatic, the latter has proven to be a disadvantage for

adsorption on activated carbon, but also less reactive with ozone. Bupropion

demands a higher ozone dose than 7 g/m3 to be removed – higher ozone doses

can however be problematic due to the production of substances with adverse

effects on water living organisms. The very high removal efficiency of

bupropion by PAC was not observed for any of the GAC products, after

treatment of 4000 bed volumes or more, which corresponds to a higher

specific carbon usage rate of GAC, than PAC. Of the remaining APIs in treated

effluent, fluconazole, used as an antifungal pharmaceutical, has the potential

to inhibit the catalytic activity of CYP1A, but the concentrations of fluconazole

in effluent after advanced treatment, seems to be so low, that inhibition will

most likely not occur.

Investment and running costs for the studied treatment technologies are not

explicitly evaluated in this work, but the author calculated these costs in two

national and regional reports in 2008 and 201010,115. The total cost

relationship was 1 : 1.7 : 5.3 for ozonation : GAC : nanofiltration and the cost

constitutes an important part in the selection of treatment technology to be

implemented in full scale.

Results and discussion | 113

In summary, all examined APIs, in this work could be removed to at least 88%.

In most cases, PAC was the broadest working removal process to achieve very

high removal efficiencies and PAC was the only process that fulfilled the

study´s goal of 95% average removal of APIs. Aiming at more realistic removal

efficiencies as 80%, all tested technologies are good candidates, but the total

cost must be taken into account. For nanofiltration, the waste problem with

retentate must be solved before nanofiltration can be implemented in the large

scale.

4.10 Spin-off of the ozonation pilot tests

In late 2014, and not as a part in this thesis, the author designed Sweden´s

first full-scale treatment for removal of micropollutants, including

pharmaceutical residues, based on the mobile pilot tests within

MistraPharma. Due to the limited budget and tense time schedule, ozonation

was selected as treatment technology. The ozonation step has a capacity to

treat wastewater from all 12 000 PE in the municipality of Knivsta, situated in

the Stockholm Region. Knivsta WWTP was selected due to its reasonable size,

12 000 inhabitants, and the small receiving watercourse, that is heavily

influenced by the treated wastewater.

After the construction phase, the ozonation in Knivsta WWTP constituted the

final treatment step after the existing mechanical, biological and chemical

wastewater treatment. By the full-scale operation, and thereby treatment of

all effluent wastewater from Knivsta WWTP, the average removal efficiency,

of the previously selected set of APIs (Table 15), was 80% at an ozone dose of

7 g/m3. This was in the same range as the results achieved in long term

operation of the ozonation pilot plant, when comparing periods with the same

concentration of organic material (TOC) in the effluent, typically 11 g/m3

(unpublished data). The Swedish Agency for Marine and Water Management,

SwAM, funded a major part of the project in Knivsta.

114 | Present investigation

4.11 Summary of work done and main results

In this work, the prevalence of pharmaceutical residues in municipal

wastewater and in the aquatic environment has been studied, in parallel with

the design, construction, operation and evaluation of the most promising

treatment technologies in pilot scale, for removal of pharmaceuticals in

municipal wastewater. The reason for carrying out this study comes from that

pharmaceutical residues have been reported to have adverse effects in the

environment and observations show that they are generally poor removed in

municipal WWTPs.

Data collection showed that the specific consumption of pharmaceuticals

varied between country regions and countries, but also altered with time.

Differences in demography and prescription habits probably caused the

variance in consumption of APIs between counties and the countries. One

example from the Baltic Sea catchment area showed a large variation between

the highest and lowest country specific consumption of carbamazepine in

2013, corresponding to a factor of 11. However, the availability and the details

of the statistics on the consumption of pharmaceuticals varied a lot between

countries, which complicated the modelling of the load of pharmaceuticals to

the WWTPs in some countries and to the aquatic environment.

In the performed sampling campaign in the Baltic Sea, residues of

pharmaceuticals were found in many locations, particularly outside major

cities. The Baltic Sea is directly or indirectly receiving effluents from the

majority of Swedish WWTPs, but also effluents from many WWTPs in the 13

other countries in the Baltic Sea catchment area. Carbamazepine was found

all over the Baltic Sea, including in offshore locations, but the average

concentration was low, 1.9 ng/L. In comparison to the estimated critical

environmental concentration, carbamazepine itself, cannot be considered as

a threat to the aquatic environment, not even in lakes with 100 times higher

concentrations. However, due to the long turnover time and low removal rate

of carbamazepine, a stock of over 55 metric tons of the substance has

accumulated in the Baltic Sea.

In this thesis work, a developed grey box model predicted the analysed

environmental concentrations of carbamazepine in the Baltic Sea sub basins

and showed on a very long half-life of carbamazepine in the Baltic Sea, >3.5

years, which makes carbamazepine a good indicator of wastewater intrusion

in natural waters. The grey box model displayed to be a simple and useful tool

to predict environmental concentrations of organic substances in water, but

the inaccessibility of open data of meteorological, hydrological and substance

Results and discussion | 115

consumption made the simulation more troublesome and potentially more

inaccurate.

The discharge of untreated or treated wastewater into inland surface water

will bring pharmaceutical residues to freshwater, which in many places are

used as raw water in waterworks. Based on the current low concentrations of

carbamazepine and other pharmaceutical residues in drinking water, WHO

has concluded, that “appreciable adverse impacts on human health from

drinking water are very unlikely” 139.

Tests with advanced methods in pilot scale, at four regular WWTPs, showed

that it was possible to remove pharmaceutical residues from regular effluent

to a high degree, under different conditions in regular plant layout, process

performance and different concentrations of pharmaceutical residues. The

WWTPs had major differences in 1) sludge age 2) specific flow of wastewater

and share of stormwater 3) specific consumption of API. 4) wastewater

temperature and 5) hydraulic retention time, especially in the biological

treatment.

In the effluents of the four WWTPs, 22-26 of the originally selected 93-97 APIs

were regularly quantified, which reduced the evaluation set, of mainly

bioactive APIs, to 22-26 APIs, which were evaluated during the pilot tests. The

dilution factors in the receiving water, for the effluent wastewater from the

four WWTPs, were sufficiently large to dilute all APIs in the effluents below

the predicted critical environmental concentration (CEC) for

bioaccumulation. This result highlights that the release of the examined APIs

into these receiving waters, is not critical today, regarding aquatic

concentrations.

The designed stationary and mobile pilot plants, with nanofiltration, activated

carbon and ozonation, worked well after minor adjustments. The only

complement installed in the mobile pilot plant was the addition of sand filters,

to remove suspended solids from the regular effluents at Uppsala and Västerås

WWTPs. Henriksdal and Käppala WWTPs have already full-scale sand filters

installed. The installed sand filters prolonged the hydraulic operation time of

the GAC filters between backwashes, due to a lower load of particles, that

would otherwise have clogged the GAC filters. An alternative measure to

increase the hydraulic capacity of the GAC filter, was tested by changing GAC,

to a product with larger grains. The hydraulic capacity increased significantly

after the exchange, but the chosen product showed to have poor adsorption

capacity of APIs.

116 | Present investigation

In the pilot tests with nanofiltration, removal efficiencies of APIs varied

between 38-99%, with a median removal efficiency of 90% for the 32 APIs.

The total volume reduction (VRF) of the polluted hydraulic flow was 20,

resulting in a retentate stream of 5%, that must be further treated. The

removal efficiency of APIs by nanofiltration could have been higher than 90%,

if the specification of the applied membrane had been correct. The molecular

weight cut off for the applied membranes showed to be twice the value

declared, 300 instead of 150 Da. To better understand and predict the removal

of APIs in nanofiltration, a model was developed with the PLS methodology,

to predict the rejection in a nanofilter plant, treating effluent municipal

wastewater. In the development of the model, 59 physicochemical properties

were reduced by multivariate analysis to four descriptive parameters. To

predict the rejection of an API, four plant specific constants must be

determined, which are then multiplied with the four identified parameters:

polarizability, sphericity, charge and the ratio of hydrophobic and polar

accessible surface area of the molecule.

The pilot tests with ozonation, operated with ozone doses of 5-7 g/m3, reached

85-95% overall removal efficiency of APIs. Out of 26 APIs, 20 APIs were

removed by 80% or more by ozonation. Fluconazol and bupropion were the

most resistant APIs to ozonation. Biomarker responses in the ozonated

wastewater were lower than the responses in regular effluent. The effluent

standards of pharmaceutical residues will influence the applied ozone dose,

but to high doses can have adverse effects on aquatic organisms. The working

environment for the operators is important to protect from ozone residues and

leaks from the dosing system. During the pilot tests, an ozone detector

monitored the ambient ozone concentration, ready to stop the ozonation in

case of leakage.

In the pilot tests with activated carbon in form of GAC and PAC, 95% removal

efficiency was reached for almost all tested pharmaceutical substances. PAC

had generally higher removal efficiency of APIs, at a lower carbon usage rate

than GAC, but well-performing GAC products showed to approach the

performance of PAC. The fresh PAC dose was reduced by the recirculation of

used PAC. Depending on the effluent standards and PAC product, the contact

time and PAC dose can be varied to achieve different degrees of removal, both

with respect to the overall pharmaceutical removal and with respect to

individual substances. Memantine, fluconazole, irbesartan and clindamycin

were the poorest removed APIs by activated carbon. Biomarker responses to

Results and discussion | 117

unspecified chemicals were reduced in fish exposed to GAC treated effluent.

PAC-effluents were not subject for biotests.

An important finding from the present study is the large variations in specific

carbon usage rates between GAC and PAC products and between individual

APIs. A high consumption rate will increase the used amount of activated

carbon, shorten the operation time between exchange of GAC, increase the

ecological footprint and the running cost. Concerning GAC, evaluation of

products with respect to hydraulic capacity is also important. In conclusion,

screening of different GAC or PAC products should be done before acquisition

to a specific WWTP.

The implementation of PAC in existing WWTPs is probably the fasted and

most cost-effective solution, when the existing active sludge step can be used.

This work raised the awareness of the necessity of a separate PAC step at

WWTPs in Sweden and other countries, where the biological excess sludge is,

to different extent, used in agriculture, after sludge stabilization. A dosing of

PAC in the biological step would enrich APIs in the sludge and thereby ruling

out the recirculation of sludge from WWTPs to agriculture. The pilot tests

showed that ozonation, GAC and PAC all performed well at the different

WWTPs visited and the processes can be chosen independently of wastewater

matrix.

The removal efficiencies varied for different APIs and different treatment

methods. In most cases, a PAC system was the broadest working removal

process to achieve very high removal efficiencies. The effluent standards of

pharmaceutical residues will influence the selection of treatment method,

since the removal efficiencies differ for the individual APIs.

118 | Present investigation

5 Conclusion, future perspective and recommendations

The overall aim of the present work was to provide the appropriate knowledge

enabling the WWTPs to remove pharmaceutical residues according to the

demand of legislation and according to the present scientific and technical

frontline.

Selected biomarkers in fish, exposed to regular WWTP effluent, showed

effects on biological systems for degradation of harmful substances i.e.

detoxification in exposed fish. The work showed that some pharmaceutical

residues are slowly degraded and widespread in the aquatic environment.

Potential technologies for removal of pharmaceuticals were selected and

tested in pilot scale: nanofiltration, GAC and PAC treatment and ozonation.

In the pilot studies, each tested technology removed more than 80% of the

APIs on average. To reach more than 90% removal, activated carbon is

considered to be the first option with a necessity of screening for appropriate

PAC and GAC products before purchase.

The combination of chemical analysis of APIs, together with monitoring of

selected biomarkers in fish, was useful for the evaluation of treatment

technologies for removal of pharmaceutical residues. Ozonation and GAC

filtration significantly decreased both the numbers and the concentrations of

APIs present in regular effluent water, as well as biomarker responses to

unspecified chemicals in fish, exposed to ozonated or GAC-filtered effluent.

Under present circumstances, ozonation or adsorption onto activated carbon

were thus shown to be the main candidate technologies to increase the

removal of pharmaceutical residues from municipal wastewater.

The removal of pharmaceutical residues is considered today as the latest

launched addition to municipal WWTPs, but it will not be the last

improvement, in the still on-going development of water and wastewater

management, that started 6 000 years ago, as described in the introduction to

this thesis.

Future perspective and recommendations

Future research and development regarding pharmaceutical residues in

wastewater should include extended removal tests in pilot and full-scale

plants and modelling of the removal processes and mechanisms.

Conclusion, future perspective and recommendations | 119

Furthermore, additional long-term biotests should be undertaken in

combination with future pilot and full-scale tests. The complementary pilot

tests should include trials with control strategies for the dosing of ozone and

PAC respectively to the effluent wastewater. Input parameters to the control

strategies must reflect the content of dissolved and adsorbing organic

substances, but also the content and distribution of particles in the effluent

wastewater. All control parameters must preferable be measured on-line.

Measurements of ozone in the off-gas can also be used in a feed-back control

strategy of the ozone dose.

The complementary biotests should involve additional species and end-

points, but also include studies in situ of aquatic organisms and the associated

food web in the receiving waters of WWTPs. Today short term biotests,

reflecting acute toxicity are performed, but long-term tests to study chronic

effects or chronic toxicity from regular and advanced treated effluent, by

ozonation and activated carbon, should also be undertaken.

Furthermore, the commercial chemical analyses of pharmaceuticals in

wastewater, and in sludge particular, need to be improved to reach lower

detection limits. Additionally, it is desirable that these analyses can be offered

to a lower cost than today. The high cost limits the monitoring of WWTPs, but

also the number of samples and thereby the accuracy of lab and pilot studies.

Alternatively, a subset of indicator APIs can be selected and offered to the

market at lower cost.

Sweden’s first full-scale ozonation plant for removal of APIs was designed and

operated at Knivsta WWTP. The design was done by the author and was based

on the findings from the ozonation pilot tests of this study. Some knowledge

gaps for future studies were identified:

1) Strategies for the distribution of ozone to different zones in the contact

tanks.

2) Best solution for quenching surplus ozone and degradation of potential by-

products from ozonation - is the standard sand filter the best alternative or

can other type of filters lower the costs or improve the removal?

3) Up-scaling from pilot scale to full scale – is the first approach to maintain

the same water depth and hydraulic retention time correct?

120 | Present investigation

In Sweden, advanced treatment at WWTP is discussed in the wastewater

sector and political initiatives are taken to facilitate the developing of know-

how, ahead of a potential implementation in full-scale plants. In this process,

the author wants to give the following recommendations to the organizations

dealing with removal of pharmaceutical residues:

1) Increase the understanding of the present situation in the WWTP with

respect of the mass flow of specific pharmaceutical residues, by

measurement of influent and effluent flows. To perform this mapping,

there is a need of external access to advanced analysis of a multitude of

substances, present at very low concentration levels. This is a

considerable problem today, due to few available laboratories and high

cost.

2) Evaluate the collected data with respect to Environmental Quality

Standards (EQSs) and the EU Water Frame Directive (WFD), but to

provide a better foresight with respect to future regulatory initiatives,

results from the research frontier in the area must be taken into

account. For this to be successful, a discussion with representatives

from academy and wastewater organizations is crucial. There is a need

for development and application of less labor intensive ecotoxicity tests

to monitor also long-term effects.

3) Evaluate the present plant treatment technology performance versus

new requirements. Are process changes in the existing plant sufficient

to fulfill new requirements? How can additional technologies improve

the removal and is it possible to install them physically on the plant?

Preferable, the proposed technology must be tested at site.

4) Implement appropriate advanced treatment. Aim to install a flexible

technology, which works as broad and efficient as possible. The selected

technology must be adjustable to compensate for variation in pollution

load. Taking into account the consumption of resources, and the

production of by-products, that must be further processed.

Internationally, development and improvement of methods for removal of

pharmaceuticals are ongoing as well. A continuation and extension of the

international collaboration is valuable and desirable.

Acknowledgements | 121

6 Acknowledgements

I would like to thank the organisations and everyone who has supported me

during the years of studies, field experiments, evaluation and in the writing of

papers and this thesis.

Financial support was provided by Mistra, the Swedish Foundation for

Strategic Environmental Research (MISTRA) through the MistraPharma

research programme and the European Union's Central Baltic programme in

the Interreg framework through the Waterchain project. Their support is

gratefully acknowledged.

I am profoundly grateful to my main supervisor Gen Larsson for believing in

me and for her valuable guidance, support and constructive feedback during

the years.

My sincere thanks to my second supervisor Ton van Maris for valuable

discussions and the feedback on my writing, including parts of this thesis.

Stort tack Andres Veide för alla trevliga diskussioner genom åren och för din

engagerade förhandsgranskning av denna avhandling.

Till mina inspirerande lärare Jan Rennerfelt och Bengt Hultman vill jag

framföra mitt varma tack för er entusiasm, kunnighet och för allt det ni lärt

mig om vattenrening genom åren!

Tack Victor Kårelid, min doktorand och medförfattare, för din noggrannhet

på lab, i fält, vid bearbetning av data och författande av artiklar. Tack också

för att du ställde upp praktiskt, i ur och skur, vid flyttarna av

pilotanläggningen.

Jimmy Söderling, tack för att du med stor noggrannhet och tålmodighet,

planerade och monterade de olika delarna i den mobila pilotanläggningen.

Tack Anna Hjalmars och Pia Trygg för ert engagerade arbete med

pilotkörningar och tillhörande provtagningar, utförda i vått och torrt.

Tack ”Kocken” för all hjälp med ned- och uppmontering av pilotanläggningen

på de olika reningsverken och tack för alla skratt och nya ord jag fått lära mig.

Tack till de tre exjobbarna Louise Jansson, Niklas Dahlén och Viktor Tsiamis

som genom inledande labförsök bidrog med dimensioneringsunderlag till den

mobila pilotanläggningen, inklusive exponeringsakvarier.

122 | Acknowledgements

Mina närmaste doktorandkollegor: Lena Flyborg och Kristina Beijer, tack för

trevligt samarbete under försökskörningar, utvärderingar och författande av

artiklar.

Thanks to my colleagues at the Division of Industrial Biotechnology, both past

and present, for discussions, reflections and a nice companionship.

Thanks to every member in the MistraPharma programme-with special

thanks to: Christina Rydén for keeping the team together, Ingvar Brant and

Björn Brunström for nice and reflecting discussions and paper writing and last

but not least, Jerker Fick, Richard Lindberg and Mats Tysklind for the

priceless analysing of nearly an endless number of samples, but also our co-

authorship.

Tack till Stiftelsen Hållbara Hav och besättningarna på Briggen Tre kronor för

ert intresse och utförda provtagningar i Östersjön som resulterade i paper I.

Ett varmt tack till personalen på Käppala reningsverk och tack för att vi fick

vara hos er när vi byggde den mobila pilotanläggningen. Tack för goda råd och

hjälp med flyttar i berget, produktion av tankar, el- och nätverksinstallationer

och att vi fick låna lab- och kontorsplats. Speciellt tack till Andreas Thunberg,

Kenneth Hyllensved, Kenneth Jaktlund, Lennart Karlsson, Mats Schultze och

Michael Medoc för ert engagemang.

Tack till Hasse Holmström, Eric Cato och Ernst-Olof Swedling på Uppsala

Vatten för hjälpen och intresset för våra pilotkörningar hos er på

Kungsängsverket i Uppsala.

Tack till Agnieszka Juszkiewicz, Anna Lindkvist och Håkan Forsberg på

Mälarenergi för att vi fick köra pilotanläggningen även på Kungsängsverket i

Västerås och för all hjälp och ert stora intresse för vårt arbete.

Tack till Cajsa Wahlberg, Andreas Carlsson och Nicklas Paxéus för ett trevligt

och lärorikt samarbete kring läkemedelsrester och de omfattande försök vi

gjorde med alla reningstekniker som tänkas kan, inom ramen för Stockholm

Vattens läkemedelsprojekt 2006–2010. Därigenom fick jag en solid grund till

denna avhandling.

Sist men inte minst, mitt hjärtligaste tack till mina älskade: mamma Alice,

pappa Helge, fästmö Nette och dotter Hanna, tack för all kärlek, allt stöd, all

omsorg, all omtanke och att ni förgyllt mitt liv med er närvaro!

References | 123

7 References

1. Ternes TA. Occurrence of drugs in German sewage treatment plants and rivers1. Water Res. 1998;32(11):3245-3260. doi:10.1016/S0043-1354(98)00099-2

2. Loos R, Carvalho R, António DC, et al. EU-wide monitoring survey on emerging polar organic contaminants in wastewater treatment plant effluents. Water Res. 2013;47(17):6475-6487. doi:10.1016/j.watres.2013.08.024

3. Nikolaou A, Meric S, Fatta D. Occurrence patterns of pharmaceuticals in water and wastewater environments. Anal Bioanal Chem. 2007;387(4):1225-1234. doi:10.1007/s00216-006-1035-8

4. Verlicchi P, Al Aukidy M, Zambello E. Occurrence of pharmaceutical compounds in urban wastewater: Removal, mass load and environmental risk after a secondary treatment—A review. Sci Total Environ. 2012;429:123-155. doi:10.1016/j.scitotenv.2012.04.028

5. Hughes SR, Kay P, Brown LE. Global Synthesis and Critical Evaluation of Pharmaceutical Data Sets Collected from River Systems. Environ Sci Technol. 2013;47(2):661-677. doi:10.1021/es3030148

6. Huggett DB, Cook JC, Ericson JF, Williams RT. A Theoretical Model for Utilizing Mammalian Pharmacology and Safety Data to Prioritize Potential Impacts of Human Pharmaceuticals to Fish. Hum Ecol Risk Assess Int J. 2003;9(7):1789-1799. doi:10.1080/714044797

7. Schreiber R, Gündel U, Franz S, Küster A, Rechenberg B, Altenburger R. Using the fish plasma model for comparative hazard identification for pharmaceuticals in the environment by extrapolation from human therapeutic data. Regul Toxicol Pharmacol. 2011;61(3):261-275. doi:10.1016/j.yrtph.2011.08.006

8. Arnold KE, Brown AR, Ankley GT, Sumpter JP. Medicating the environment: assessing risks of pharmaceuticals to wildlife and ecosystems. Phil Trans R Soc B. 2014;369(1656):20130569. doi:10.1098/rstb.2013.0569

9. Coleman MD. Human Drug Metabolism: An Introduction. Chichester, UK: Chichester, UK: John Wiley & Sons, Ltd; 2010. doi:10.1002/9780470689332

10. Wahlberg C, Björlenius B, Paxeus N. Läkemedelsrester i Stockholms Vattenmiljö.; 2010. http://www.stockholmvattenochavfall.se/globalassets/pdf1/rapporter/avlopp/avloppsrening/lakemedelsrapport_slutrapport.pdf. Accessed June 6, 2017.

11. Cabrera E, Arregui F, Viollet P-L. Water Engineering and Management through Time: Learning from History. CRC Press; 2010.

124 | References

12. Vallet R. Habuba Kebira ou la naissance de l’urbanisme. Paléorient. 1996;22(2):45-76. doi:10.3406/paleo.1996.4636

13. Delougaz P, Hill H, Lloyd S. Private houses and graves in the Diyala region. 1967. https://oi.uchicago.edu/sites/oi.uchicago.edu/files/uploads/shared/docs/oip88.pdf. Accessed August 17, 2017.

14. Mays L. Ancient Water Technologies. Springer Science & Business Media; 2010.

15. Adams R. Heartland of Cities - Surveys of Ancient Settlement and Land Use on the Central Floodplain of the Euphrates.; 1981. https://oi.uchicago.edu/sites/oi.uchicago.edu/files/uploads/shared/docs/heartland_of_cities.pdf. Accessed August 17, 2017.

16. Hopkins JN. The Cloaca Maxima and the Monumental Manipulation of Water in Archaic Rome. publisher not identified; 2007. http://www3.iath.virginia.edu/waters/Journal4Hopkins.pdf. Accessed May 27, 2017.

17. Lanciani R. The ruins and excavations of ancient Rome; a companion book for students and travelers. https://archive.org/stream/cu31924028273997/cu31924028273997_djvu.txt. Published 1897. Accessed May 28, 2017.

18. Wentworth Rinne K. Aquae Urbis Romae: the Waters of the City of Rome. http://www3.iath.virginia.edu/waters/main.html. Published May 29, 2017. Accessed May 29, 2017.

19. Hodge AT. Aqueducts and Water Supply. In: Evans JD, ed. A Companion to the Archaeology of the Roman Republic. Blackwell Publishing Ltd; 2013:285-295. doi:10.1002/9781118557129.ch18

20. Frontinus SI. De Aquaeductu Urbis Romae.; 98. http://www.uvm.edu/~rrodgers/Frontinus.html. Accessed May 26, 2017.

21. Rodgers RH, ed. Frontinus on Water Supply. Class Rev. 2006;56(1):132-135.

22. Sarnıcı BB. Byzantium Period Water Architecture and a Masterpiece in Istanbul: The Big Basilica Cistern. Turk Neurosurg. 2014;24(6):823–827.

23. Passchier C, Sürmelihindi G, van Opstal D, Schram W. The Atlas Project of Roman Aqueducts (ROMAQ). http://www.romaq.org/. Published 2011. Accessed June 5, 2017.

24. Schram W, van Opstal D, Passchier C. Website on Roman aqueducts. http://www.romanaqueducts.info/index.html. Published 2017. Accessed June 5, 2017.

References | 125

25. Hansen RD. Water-related infrastructure in Medieval London. Nd Httpwww Waterhistory Orghistorieslondonaccessed 10 January 2015. 2009. http://www.waterhistory.org/histories/london/london.pdf. Accessed June 5, 2017.

26. Sabine EL. Latrines and Cesspools of Mediaeval London. Speculum. 1934;9(3):303-321. doi:10.2307/2853898

27. Meeder WL. Geschichte von Hamburg: vom entstchen der stadt bis auf die neueste zeit. zu haben beim verfasser; 1839.

28. Rambach JJ. Versuch einer physisch-medizinischen Beschreibung von Hamburg. Bohn; 1801.

29. Angelakis A, Snyder S. Wastewater Treatment and Reuse: Past, Present, and Future. Water. 2015;7(9):4887-4895. doi:10.3390/w7094887

30. Védry B, Gousailles M, Affholder M, Lefaux A, Bontoux J. From sewage water treatment to wastewater reuse. One century of Paris sewage farms history. Water Sci Technol. 2001;43(10):101-107.

31. Reid D. Paris Sewers and Sewermen: Realities and Representations. Harvard University Press; 1993.

32. Barles S. Urban metabolism and river systems: an historical perspective. Paris and the Seine, 1790-1970. Hydrol Earth Syst Sci. 2007;11:1757-1769.

33. SIAAP. Sewage treatment in the greater Paris area: from an indispensable hygienist approach to an environmental protection approach. In: ; 2006.

34. Crossness Engines Trust. The Engines. http://www.crossness.org.uk/history/the-engines.html. Accessed August 19, 2017.

35. Halliday S. The Great Stink of London. History Press; 2001.

36. Roberts B. Sir Joseph Bazalgette and London’s Intercepting Sewer System.; 2016. http://www.hevac-heritage.org/built_environment/pioneers_revisited/surnames_a-o/bazalgette.pdf. Accessed August 19, 2017.

37. Rennerfelt J. Kommunal Och Industriell Avloppsteknik : Bearbetad Och Förkortad Upplaga Av Rennerfelt-Ulmgren “Vattenreningsteknik.” Stockholm: KTH, biokemisk teknologi; 2000.

38. Ødegaard H, Rusten B, Westrum T. A new moving bed biofilm reactor - applications and results. Water Sci Technol. 1994;29(10-11):157-165.

126 | References

39. Hasselkus WN. EPA WASTEWATER TECHNOLOGY FACT SHEETS: We’re looking for a few good technologies. Proc Water Environ Fed. 2000;2000(10):802–822.

40. Gerardi MH. Troubleshooting the Sequencing Batch Reactor. Hoboken: Hoboken : Wiley; 2010.

41. Morling S. SBR-Technology - Use and potential applications for treatment of cold wastewater. 2009. https://www.diva-portal.org/smash/get/diva2:219078/FULLTEXT01.pdf. Accessed June 6, 2017.

42. Jenkins D, Wanner J. Activated Sludge - 100 Years and Counting. IWA Publishing; 2014.

43. Çeçen F, Aktaş Ö. Activated Carbon for Water and Wastewater Treatment: Integration of Adsorption and Biological Treatment. Weinheim, Germany: Wiley-VCH Verlag GmbH & Co. KGaA; 2011. doi:10.1002/9783527639441

44. Cronström A. Stockholms tekniska historia: Vattenförsörjning och avlopp. Stockholm: Kommittén för Stockholmsforskning; 1986.

45. Isgård E. I Vattumannens Tecken : Svensk VA-Teknik Från Trärör till Kväverening. Örebro: Ohlson & Winnfors; 1998.

46. Lundbeck H, Plazikowski U, Silverstolpe L. The Swedish Salmonella Outbreak of 1953. J Appl Bacteriol. 1955;18(3):535-548. doi:10.1111/j.1365-2672.1955.tb02112.x

47. Olin G. The Salmonella Epidemics of 1953, in Sweden and the Problems raised by them. Nord Med. 1956;55(17):577-585.

48. Tekniska verken. Nykvarnsverket Linköping Miljörapport 2017. Linköping; 2018:60.

49. Swedish Environmental Protection Agency. Wastewater Treatment in Sweden ISBN 978-91-620-8704-3. - 978-91-620-8704-3.Pdf. Stockholm; 2014. http://www.naturvardsverket.se/Documents/publikationer6400/978-91-620-8704-3.pdf?pid=13165. Accessed June 8, 2017.

50. Holmquist H, Ohlsson M, Anderberg S. Vattenverket 100 År : 1861-1961 : Minnesskrift. Stockholm: utg.; 1961.

51. Uddenberg N. Lidande Och Läkedom. 2, Medicinens Historia Från 1800 till 1950. [Lidingö]: Fri Tanke; 2015.

52. Bentivoglio M, Pacini P. Filippo Pacini: A determined observer. Brain Res Bull. 1995;38(2):161-165. doi:10.1016/0361-9230(95)00083-Q

References | 127

53. Norling B. Muggar Och Ställ : En Berättelse Om Gustavsbergs Sanitetsporslinsfabrik. Gustavsberg: Gustavsbergs porslinsmuseum i samarbete med Arena; 2010.

54. Käppalaförbundet. Käppalaförbundet 50 år. https://www.kappala.se/Documents/Diverse%20dokument/K%C3%A4ppalaf%C3%B6rbundet%2050%20%C3%A5r.pdf. Published 2007. Accessed June 7, 2017.

55. SYVAB. SYVAB - Vi värnar vårt vatten. file:///C:/Users/Berndt/SkyDrive/Referenslitteratur%20digital/VA-Historia/syvab_broschyr_final.pdf. Published 2009. Accessed June 7, 2017.

56. Michanek G, Zetterberg C. Den Svenska Miljörätten. Uppsala: Iustus; 2012. http://iustus.se/wp-content/uploads/2014/03/Supplement-Den-svenska-milj%C3%B6r%C3%A4tten-mars-2015.pdf.

57. Riksdagsförvaltningen. Miljöbalken 1998:808. https://www.riksdagen.se/sv/dokument-lagar/dokument/svensk-forfattningssamling/miljobalk-1998808_sfs-1998-808. Published 1998. Accessed August 22, 2017.

58. Carvalho RN, Ceriani L, Ippolito A, et al. Development of the First Watch List under the Environmental Quality Standards Directive. Luxembourg: Publications Office; 2015. http://bookshop.europa.eu/uri?target=EUB:NOTICE:LBNA27142:EN:HTML. Accessed July 24, 2017.

59. EUROPEAN PARLIAMENT AND OF THE COUNCIL. Directive 2013/39/EU.; 2013. http://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32013L0039&from=EN. Accessed July 24, 2017.

60. EUROPEAN PARLIAMENT AND OF THE COUNCIL. DIRECTIVE 2000/60/EC Water Frame Directive.; 2000. http://eur-lex.europa.eu/resource.html?uri=cellar:5c835afb-2ec6-4577-bdf8-756d3d694eeb.0004.02/DOC_1&format=PDF. Accessed August 19, 2017.

61. European Parliament and of the Council. DIRECTIVE 2008/105/EC.; 2008. http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2008:348:0084:0097:en:PDF. Accessed July 24, 2017.

62. Eggen RIL, Hollender J, Joss A, Schärer M, Stamm C. Reducing the Discharge of Micropollutants in the Aquatic Environment: The Benefits of Upgrading Wastewater Treatment Plants. Environ Sci Technol. 2014;48(14):7683-7689. doi:10.1021/es500907n

128 | References

63. Wahlberg C, Björlenius B, Ek M, Paxeus N, Gårdstam L. Avloppsreningsverkens förmåga att ta hand om läkemedelsrester och andra farliga ämnen redovisning av regeringsuppdrag: 512-386-06 Rm. Stockholm: Naturvårdsverket; 2008. http://www.naturvardsverket.se/Documents/publikationer/620-5794-7.pdf.

64. Ternes TA. Human Pharmaceuticals, Hormones and Fragances a Challenge for Urban Water Management. London: IWA Publ.; 2006.

65. Vieno N. Occurrence of Pharmaceuticals in Finnish Sewage Treatment Plants, Surface Waters, and Their Elimination in Drinking Water Treatment Processes. 2007. https://tutcris.tut.fi/portal/files/5469794/vieno.pdf. Accessed March 1, 2017.

66. Umweltbundesamt. Pharmaceuticals in the Environment – the Global Perspective.; 2016. https://www.umweltbundesamt.de/sites/default/files/medien/378/publikationen/pharmaceuticals_in_the_environment_0.pdf. Accessed June 15, 2017.

67. Grand view research. Active Pharmaceutical Ingredients (API) Market Size Report, 2014-2025.; 2017. http://www.grandviewresearch.com/industry-analysis/active-pharmaceutical-ingredients-market. Accessed June 15, 2017.

68. Fent K, Weston A, Caminada D. Ecotoxicology of human pharmaceuticals. Aquat Toxicol. 2006;76(2):122-159. doi:10.1016/j.aquatox.2005.09.009

69. Medical Products Agency - Sweden. Information från Läkemedelsverket nr 5 2016. file:///C:/Users/Berndt/SkyDrive/Referenslitteratur%20digital/Rapporter/L%C3%A4kemedel/Information_fran_Lakemedelsverket_nr_5_2016.pdf. Published 2016. Accessed June 16, 2017.

70. eHälsomyndigheten. Läkemedelsförsäljningen i Sverige 2014 No 2015/0424. https://www.ehalsomyndigheten.se/globalassets/dokument/statistik/lakemedelsforsaljningen-i-sverige-2014.pdf. Published 2015. Accessed June 14, 2017.

71. Socialstyrelsen. Läkemedel - statistik för år 2014.; 2015.

72. Cuklev Stern F. Transcriptomics and Bioconcentration Studies in Fish to Identify Pharmaceuticals of Environmental Concern [Elektronisk Resurs]. Göteborg: Department of Physiology/Endocrinology, Institute of Neuroscience and Physiology, The Sahlgrenska Academy, University of Gothenbrg; 2012. http://hdl.handle.net/2077/28251.

73. Hester RE, Harrison RM. Pharmaceuticals in the Environment. The Royal Society of Chemistry; 2016. doi:10.1039/9781782622345

References | 129

74. Larsson DGJ, de Pedro C, Paxeus N. Effluent from drug manufactures contains extremely high levels of pharmaceuticals. J Hazard Mater. 2007;148(3):751-755. doi:10.1016/j.jhazmat.2007.07.008

75. The European parliament and the council of the european union. REGULATION (EC) No 1831/2003 OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL of 22 September 2003 on additives for use in animal nutrition. http://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32003R1831&from=EN. Published 2003. Accessed June 14, 2017.

76. D’Silva J. The Meat Crisis: The Ethical Dimensions of Animal Welfare, Climate Change, and Future Sustainability. In: Sustainable Food Security in the Era of Local and Global Environmental Change. Springer; 2013:19–32. http://link.springer.com/chapter/10.1007/978-94-007-6719-5_2. Accessed June 14, 2017.

77. Boeckel TPV, Brower C, Gilbert M, et al. Global trends in antimicrobial use in food animals. Proc Natl Acad Sci. 2015;112(18):5649-5654. doi:10.1073/pnas.1503141112

78. Xu J, Wu L, Chang AC. Degradation and adsorption of selected pharmaceuticals and personal care products (PPCPs) in agricultural soils. Chemosphere. 2009;77(10):1299-1305. doi:10.1016/j.chemosphere.2009.09.063

79. European Parliament and Council. Directive 2000/76/EC incineration of waste. http://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=LEGISSUM:l28072&from=EN. Published 2000. Accessed June 13, 2017.

80. Brunner C, Brownell R, JR Rowland. Pharmacutical Plant Disposal by Incineration. In: Proceedings of... National Waste Processing Conference. American Society of Mechanical Engineers; 1984:451. http://www.seas.columbia.edu/earth/wtert/sofos/nawtec/1984-National-Waste-Processing-Conference/1984-National-Waste-Processing-Conference-41.pdf. Accessed June 13, 2017.

81. Kümmerer K. Pharmaceuticals in the Environment: Sources, Fate, Effects and Risks. Springer Science & Business Media; 2008.

82. Coats JR, Metcalf RL, Lu PY, Brown DD, Williams JF, Hansen LG. Model ecosystem evaluation of the environmental impacts of the veterinary drugs phenothiazine, sulfamethazine, clopidol, and diethylstilbestrol. Environ Health Perspect. 1976;18:167-179.

83. Hignite C, Azarnoff DL. Drugs and drug metabolites as environmental contaminants: Chlorophenoxyisobutyrate and salicylic acid in sewage water effluent. Life Sci. 1977;20(2):337-341. doi:10.1016/0024-3205(77)90329-0

130 | References

84. Purdom CE, Hardiman PA, Bye VVJ, Eno NC, Tyler CR, Sumpter JP. Estrogenic Effects of Effluents from Sewage Treatment Works. Chem Ecol. 1994;8(4):275-285. doi:10.1080/02757549408038554

85. Sumpter JP, Jobling S. Vitellogenesis as a biomarker for estrogenic contamination of the aquatic environment. Environ Health Perspect. 1995;103(Suppl 7):173-178.

86. Routledge EJ, Sheahan D, Desbrow C, Brighty GC, Waldock M, Sumpter JP. Identification of Estrogenic Chemicals in STW Effluent. 2. In Vivo Responses in Trout and Roach. Environ Sci Technol. 1998;32(11):1559-1565. doi:10.1021/es970796a

87. Schwaiger J, Ferling H, Mallow U, Wintermayr H, Negele RD. Toxic effects of the non-steroidal anti-inflammatory drug diclofenac. Aquat Toxicol. 2004;68(2):141-150. doi:10.1016/j.aquatox.2004.03.014

88. Triebskorn R, Casper H, Heyd A, Eikemper R, Köhler H-R, Schwaiger J. Toxic effects of the non-steroidal anti-inflammatory drug diclofenac. Aquat Toxicol. 2004;68(2):151-166. doi:10.1016/j.aquatox.2004.03.015

89. Memmert U, Peither A, Burri R, et al. Diclofenac: New data on chronic toxicity and bioconcentration in fish. Environ Toxicol Chem. 2013;32(2):442-452. doi:10.1002/etc.2085

90. Oaks JL, Gilbert M, Virani MZ, et al. Diclofenac residues as the cause of vulture population decline in Pakistan. Nature. 2004;427(6975):630-633. doi:10.1038/nature02317

91. De Lange HJ, Noordoven W, Murk AJ, Lürling M, Peeters ETHM. Behavioural responses of Gammarus pulex (Crustacea, Amphipoda) to low concentrations of pharmaceuticals. Aquat Toxicol. 2006;78(3):209-216. doi:10.1016/j.aquatox.2006.03.002

92. Mimeault C, Woodhouse AJ, Miao X-S, Metcalfe CD, Moon TW, Trudeau VL. The human lipid regulator, gemfibrozil bioconcentrates and reduces testosterone in the goldfish, Carassius auratus. 2005;73(1):44–54. doi:10.1016/j.aquatox.2005.01.009

93. Huber WG. Antibacterial drugs as environmental contaminants. Adv Env Sci Technol. 1971. http://agris.fao.org/agris-search/search.do?recordID=US201302262985. Accessed July 8, 2017.

94. Pendleton J, Gorman S, Gilmore B. Clinical Relevance of the ESKAPE Pathogens. Expert Rev Anti Infect Ther. 2013;3(11):297-308.

95. Bush K. Improving known classes of antibiotics: an optimistic approach for the future. Curr Opin Pharmacol. 2012;12(5):527-534. doi:10.1016/j.coph.2012.06.003

References | 131

96. Larsson DGJ. Pollution from drug manufacturing: review and perspectives. Phil Trans R Soc B. 2014;369(1656):20130571. doi:10.1098/rstb.2013.0571

97. Kümmerer K. Antibiotics in the aquatic environment – A review – Part II. Chemosphere. 2009;75(4):435-441. doi:10.1016/j.chemosphere.2008.12.006

98. Bengtsson-Palme J, Hammarén R, Pal C, et al. Elucidating selection processes for antibiotic resistance in sewage treatment plants using metagenomics. Sci Total Environ. 2016;572:697-712. doi:10.1016/j.scitotenv.2016.06.228

99. Sitruk-Ware R, Nath A. The use of newer progestins for contraception. Contraception. 2010;82(5):410-417. doi:10.1016/j.contraception.2010.04.004

100. Svensson J. Progestagenic Aquatic Contaminants Act as Potent Androgens in Fish : Experimental Studies in Three-spined Stickleback and Zebrafish. 2016. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-286535. Accessed April 19, 2016.

101. Kumar V, Johnson AC, Trubiroha A, et al. The Challenge Presented by Progestins in Ecotoxicological Research: A Critical Review. Environ Sci Technol. 2015;49(5):2625-2638. doi:10.1021/es5051343

102. Svensson J, Fick J, Brandt I, Brunström B. Environmental concentrations of an androgenic progestin disrupts the seasonal breeding cycle in male three-spined stickleback (Gasterosteus aculeatus). Aquat Toxicol. 2014;147:84-91. doi:10.1016/j.aquatox.2013.12.013

103. Brodin T, Fick J, Jonsson M, Klaminder J. Dilute Concentrations of a Psychiatric Drug Alter Behavior of Fish from Natural Populations. Science. 2013;339(6121):814. doi:10.1126/science.1226850

104. Ågerstrand M. From science to policy: improving environmental risk assessment andmanagement of chemicals. 2012.

105. Molander L, Ågerstrand M, Rudén C. WikiPharma – A freely available, easily accessible, interactive and comprehensive database for environmental effect data for pharmaceuticals. Regul Toxicol Pharmacol. 2009;55(3):367-371. doi:10.1016/j.yrtph.2009.08.009

106. Leeuwen CJ van, Vermeire TG. Risk Assessment of Chemicals: An Introduction. Springer Science & Business Media; 2007.

107. European Medicines Agency (CHMP). Guideline on the Environmental Riak Assessment of Medicinal Proucts for Human Use - WC500003978.Pdf.; 2006. http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2009/10/WC500003978.pdf. Accessed August 20, 2017.

132 | References

108. Ågerstrand M, Berg C, Björlenius B, et al. Improving Environmental Risk Assessment of Human Pharmaceuticals. Environ Sci Technol. 2015;49(9):5336-5345. doi:10.1021/acs.est.5b00302

109. Samuelsson LM, Björlenius B, Förlin L, Larsson DGJ. Reproducible 1H NMR-Based Metabolomic Responses in Fish Exposed to Different Sewage Effluents in Two Separate Studies. Environ Sci Technol. 2011;45(4):1703-1710. doi:10.1021/es104111x

110. Furuhagen S. Application and interpretation of biomarkers in ecotoxicology - from molecular to individual level responses. 2015. http://urn.kb.se/resolve?urn=urn:nbn:se:su:diva-120161. Accessed September 2, 2015.

111. Reutgard M, Furuhagen S. Linking sub-cellular biomarkers to embryo aberrations in the benthic amphipod Monoporeia affinis. Aquat Toxicol. 2016;173:36-42. doi:10.1016/j.aquatox.2016.01.005

112. Gunnarsson L, Jauhiainen A, Kristiansson E, Nerman O, Larsson DGJ. Evolutionary Conservation of Human Drug Targets in Organisms used for Environmental Risk Assessments. Environ Sci Technol. 2008;42(15):5807-5813. doi:10.1021/es8005173

113. World Health Organization. Biomarkers and Risk Assessment: Concepts and Principles. Geneva, Switzerland; 1993. http://apps.who.int/iris/bitstream/10665/42046/1/9241572043_eng.pdf. Accessed August 20, 2017.

114. Katsiadaki I, Scott AP, Hurst MR, Matthiessen P, Mayer I. Detection of environmental androgens: A novel method based on enzyme-linked immunosorbent assay of spiggin, the stickleback (Gasterosteus aculeatus) glue protein. Environ Toxicol Chem. 2002;21(9):1946-1954. doi:10.1002/etc.5620210924

115. Uno T, Ishizuka M, Itakura T. Cytochrome P450 (CYP) in fish. Environ Toxicol Pharmacol. 2012;34(1):1-13. doi:10.1016/j.etap.2012.02.004

116. Parkinson A, Ogilvie BW. Biotransformation of Xenobiotics. McGraw-Hill New York; 2001. http://184.168.171.185/BOOKS/CasarettAndDoullsToxicology_TheBasicScienceOfPoisons_6thEd/S2_Ch06_BiotransformationOfXenobiotics.pdf. Accessed August 20, 2017.

117. DeRuiter J. Non-Steroidal Anti-Inflammatory Drugs (NSAIDs), Principles of Drug Action, Vol. 2.; 2002. http://kubanni.abu.edu.ng:8080/jspui/handle/123456789/8827. Accessed July 23, 2017.

References | 133

118. Paxéus N. Analys Av Läkemedelsrester i Dricks- Och Avloppsvatten: Enprovningsjämförelese, Report: 10SV74. Stockholm; 2010. file:///C:/1_Verbatim1%20lkm/CW%20500009-101%20L%C3%A4kemedelsprojektet/Delrapporter/Interkalibrering/Interkalibrering.pdf. Accessed July 23, 2017.

119. Jessome LL, Volmer DA. Ion suppression: a major concern in mass spectrometry. Lc Gc N Am. 2006;24(5):498.

120. Joss A, Keller E, Alder AC, et al. Removal of pharmaceuticals and fragrances in biological wastewater treatment. Water Res. 2005;39(14):3139-3152. doi:10.1016/j.watres.2005.05.031

121. Nelson ED, Do H, Lewis RS, Carr SA. Diurnal Variability of Pharmaceutical, Personal Care Product, Estrogen and Alkylphenol Concentrations in Effluent from a Tertiary Wastewater Treatment Facility. Environ Sci Technol. 2011;45(4):1228-1234. doi:10.1021/es102452f

122. Plósz BG, Leknes H, Liltved H, Thomas KV. Diurnal variations in the occurrence and the fate of hormones and antibiotics in activated sludge wastewater treatment in Oslo, Norway. Sci Total Environ. 2010;408(8):1915-1924. doi:10.1016/j.scitotenv.2010.01.042

123. Björlenius B. Biologisk kvävereduktion i rejektvatten från slamrötning - Nitrogen Removal in Supernatant from Centrifugation of Digested Sludge. Vatten. 1988;44(3).

124. Fick J, Lindberg RH, Fång J, Magnér J, Kaj L, Brorström-Lundén E. Screening 2014: Analysis of Pharmaceuticals and Hormones in Samples from WWTPs and Receiving Waters. IVL Report number C 135; 2015. http://www.diva-portal.org/smash/record.jsf?pid=diva2:927576. Accessed March 20, 2017.

125. Fick J, Lindberg RH, Kaj L, Brorström-Lundén E. Results from the Swedish National Screening Programme 2010. Stockholm; 2011. http://www.ivl.se/download/18.343dc99d14e8bb0f58b542e/1443183072893/B2014.pdf. Accessed March 20, 2017.

126. Stockholms läns landsting. Läkemedelsrester i Akvatisk Miljö - Sammanställning Av Data.; 2007.

127. Stockholms läns landsting, Trossa AB. Sammanställning av läkemedelsprovtagningar-2005-2012. http://www.sll.se/Global/Om%20landstinget/Miljo/Rapport-lakemedelsprovtagningar-2005-2012.pdf. Published 2014. Accessed May 8, 2017.

134 | References

128. WHO. Collaborating Centre for Drug Statistics Methodology - ATC/DDD Index 2017. https://www.whocc.no/atc_ddd_index/?code=N03AF01&showdescription=yes. Published 2017. Accessed July 24, 2017.

129. FASS. Alvedon® - FASS Vårdpersonal. http://www.fass.se/LIF/product?userType=0&nplId=19581115000027. Published 2016. Accessed July 24, 2017.

130. Johnson AC, Williams RJ. A Model To Estimate Influent and Effluent Concentrations of Estradiol, Estrone, and Ethinylestradiol at Sewage Treatment Works. Environ Sci Technol. 2004;38(13):3649-3658. doi:10.1021/es035342u

131. German Federal Environment Agency. Erythromycin_EQS DATASHEET.; 2014. file:///C:/Users/Berndt/SkyDrive/Referenslitteratur%20digital/Rapporter/L%C3%A4kemedel/Erythromycin_EQS%20DATASHEET_2015_Feb27.pdf. Accessed July 24, 2017.

132. Sub-Group on Review of the Priority Substances List. Alpha-Ethinylestradiol EQS Dossier.; 2011. https://circabc.europa.eu/sd/a/efb00adc-ece7-4bec-a611-3b612cf1f942/EE2%20EQS%20dossier%202011.pdf. Accessed July 24, 2017.

133. Sub-Group on Review of the Priority Substances List. Beta-Estradiol EQS Dossier.; 2011. https://circabc.europa.eu/sd/a/c5356fa7-be0e-4d3b-b199-208d6e144a91/E2%20EQS%20dossier%202011.pdf. Accessed July 24, 2017.

134. Sub-Group on Review of the Priority Substances List. Diclofenac EQS Dossier.; 2011. https://circabc.europa.eu/sd/a/d88900c0-68ef-4d34-8bb1-baa9af220afd/Diclofenac%20EQS%20dossier%202011.pdf. Accessed July 24, 2017.

135. Loos R, Gawlik BM, Locoro G, Rimaviciute E, Contini S, Bidoglio G. EU-wide survey of polar organic persistent pollutants in European river waters. 2009;157(2):561–568. doi:10.1016/j.envpol.2008.09.020

136. Phillips PJ, Smith SG, Kolpin DW, et al. Pharmaceutical Formulation Facilities as Sources of Opioids and Other Pharmaceuticals to Wastewater Treatment Plant Effluents. 2010;44(13):4910–4916. doi:10.1021/es100356f

137. Sim W-J, Lee J-W, Lee E-S, Shin S-K, Hwang S-R, Oh J-E. Occurrence and distribution of pharmaceuticals in wastewater from households, livestock farms, hospitals and pharmaceutical manufactures. Chemosphere. 2011;82(2):179-186. doi:10.1016/j.chemosphere.2010.10.026

References | 135

138. Fick J, Söderström H, Lindberg RH, Phan C, Tysklind M, Larsson DGJ. Contamination of surface, ground, and drinking water from pharmaceutical production. Environ Toxicol Chem. 2009;28(12):2522-2527. doi:10.1897/09-073.1

139. World Health Organization. Pharmaceuticals in Drinking-Water. Geneva, Switzerland: World Health Organization; 2012. http://apps.who.int/iris/bitstream/10665/44630/1/9789241502085_eng.pdf. Accessed June 15, 2017.

140. Clara M, Kreuzinger N, Strenn B, Gans O, Kroiss H. The solids retention time—a suitable design parameter to evaluate the capacity of wastewater treatment plants to remove micropollutants. Water Res. 2005;39(1):97-106. doi:10.1016/j.watres.2004.08.036

141. Tambosi JL, de Sena RF, Favier M, et al. Removal of pharmaceutical compounds in membrane bioreactors (MBR) applying submerged membranes. Desalination. 2010;261(1-2):148-156. doi:10.1016/j.desal.2010.05.014

142. Falås P, Andersen HR, Ledin A, la Cour Jansen J. Impact of solid retention time and nitrification capacity on the ability of activated sludge to remove pharmaceuticals. Environ Technol. 2012;33(8):865-872. doi:10.1080/09593330.2011.601764

143. Morgenroth E, Sherden T, Van Loosdrecht MCM, Heijnen JJ, Wilderer PA. Aerobic granular sludge in a sequencing batch reactor. Water Res. 1997;31(12):3191-3194. doi:10.1016/S0043-1354(97)00216-9

144. Jin B, Wilén B-M, Lant P. A comprehensive insight into floc characteristics and their impact on compressibility and settleability of activated sludge. Chem Eng J. 2003;95(1):221-234. doi:10.1016/S1385-8947(03)00108-6

145. Rojas J. Relationship between the Sludge Settling Characteristics and the Parameters of the Activated Sludge System. December 2004. https://scholarworks.uno.edu/td/196.

146. Beun JJ, Hendriks A, van Loosdrecht MCM, Morgenroth E, Wilderer PA, Heijnen JJ. Aerobic granulation in a sequencing batch reactor. Water Res. 1999;33(10):2283-2290. doi:10.1016/S0043-1354(98)00463-1

147. Kreuk MK de, Heijnen JJ, Loosdrecht MCM van. Simultaneous COD, nitrogen, and phosphate removal by aerobic granular sludge. Biotechnol Bioeng. 2005;90(6):761-769. doi:10.1002/bit.20470

148. Amorim CL, Moreira IS, Ribeiro AR, et al. Treatment of a simulated wastewater amended with a chiral pharmaceuticals mixture by an aerobic granular sludge sequencing batch reactor. Int Biodeterior Biodegrad. 2016;115:277-285. doi:10.1016/j.ibiod.2016.09.009

136 | References

149. Zhao X, Chen Z, Wang X, Li J, Shen J, Xu H. Remediation of pharmaceuticals and personal care products using an aerobic granular sludge sequencing bioreactor and microbial community profiling using Solexa sequencing technology analysis. Bioresour Technol. 2015;179:104-112. doi:10.1016/j.biortech.2014.12.002

150. Xia Z, Xiao-chun W, Zhong-lin C, Hao X, Qing-fang Z. Microbial community structure and pharmaceuticals and personal care products removal in a membrane bioreactor seeded with aerobic granular sludge. Appl Microbiol Biotechnol. 2015;99(1):425-433. doi:10.1007/s00253-014-5984-0

151. Tchobanoglous G, Burton FL, Stensel HD, Metcalf & Eddy. Wastewater Engineering : Treatment and Reuse. 4th ed. / revised by George Tchobanoglous, Franklin L. Burton, H. David Stensel. Boston: Boston : McGraw-Hill; 2002.

152. Falås P, Longrée P, la Cour Jansen J, Siegrist H, Hollender J, Joss A. Micropollutant removal by attached and suspended growth in a hybrid biofilm-activated sludge process. Water Res. 2013;47(13):4498-4506. doi:10.1016/j.watres.2013.05.010

153. Falås P, Baillon-Dhumez A, Andersen HR, Ledin A, la Cour Jansen J. Suspended biofilm carrier and activated sludge removal of acidic pharmaceuticals. Water Res. 2012;46(4):1167-1175. doi:10.1016/j.watres.2011.12.003

154. IVL Svenska Miljöinstitutet AB, Swedish Environmental Agency. Enskilda avlopp som källa till läkemedelsrester och andra kemikalier. Stockholm: Naturvårdsverket; 2012.

155. Gros M, Blum KM, Jernstedt H, et al. Screening and prioritization of micropollutants in wastewaters from on-site sewage treatment facilities. J Hazard Mater. 2017;328:37-45. doi:10.1016/j.jhazmat.2016.12.055

156. Falås P. Removal of Pharmaceuticals in Biofilm and Activated Sludge Systems. Lund University (Media-Tryck); 2012. http://lup.lub.lu.se/record/3168560. Accessed June 9, 2017.

157. Blum KM, Andersson PL, Renman G, et al. 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. Sci Total Environ. 2017;575:265-275. doi:10.1016/j.scitotenv.2016.09.135

References | 137

158. Zhang W, Renman G. Screening of Different Types of On-Site Sewage Facilitiestreatment Function and Potential for Removing Micropollutants. Stockholm, Sweden: Dept. of Sustainable Development, Environmental Science & Engineering, KTH Royal Institute of Technology; 2016. http://uest.ntua.gr/swws/proceedings/pdf/Screening_different_types_of_OSSFs_Wen_Zhang.pdf. Accessed August 2, 2017.

159. Henry W. Experiments on the Quantity of Gases Absorbed by Water, at Different Temperatures, and under Different Pressures. Philos Trans R Soc Lond. 1803;93:29-276.

160. ChemIDplus. Chemical Nomenclature and Structure TOXicology data NETwork (TOXNET) U.S. National Library of Medicine. https://chem.nlm.nih.gov/chemidplus/chemidlite.jsp. Published 2017. Accessed July 22, 2017.

161. Hörsing M, Ledin A, Grabic R, et al. Determination of sorption of seventy-five pharmaceuticals in sewage sludge. Water Res. 2011;45(15):4470-4482. doi:10.1016/j.watres.2011.05.033

162. Ternes T, Janex-Habibi M, Knacker T, Kreuzinger N, Siegrist H, others. Assessment of technologies for the removal of pharmaceuticals and personal care products in sewage and drinking water facilities to improve the indirect potable water reuse. Contract No EVK1-CT-2000-00047. 2004.

163. Kümmerer K. The presence of pharmaceuticals in the environment due to human use – present knowledge and future challenges. J Environ Manage. 2009;90(8):2354-2366. doi:10.1016/j.jenvman.2009.01.023

164. Rojas MR, Leung C, Bonk F, et al. Assessment of the Effectiveness of Secondary Wastewater Treatment Technologies to Remove Trace Chemicals of Emerging Concern. Crit Rev Environ Sci Technol. 2013;43(12):1281-1314. doi:10.1080/10643389.2011.644221

165. Urase T, Kikuta T. Separate estimation of adsorption and degradation of pharmaceutical substances and estrogens in the activated sludge process. Water Res. 2005;39(7):1289-1300. doi:10.1016/j.watres.2005.01.015

166. Chang P-H, Jean J-S, Jiang W-T, Li Z. Mechanism of tetracycline sorption on rectorite. Colloids Surf Physicochem Eng Asp. 2009;339(1-3):94-99. doi:10.1016/j.colsurfa.2009.02.002

167. Noir ML, Guieysse B, Mattiasson B. Removal of trace contaminants using molecularly imprinted polymers. Water Sci Technol. 2006;53(11):205-212. doi:10.2166/wst.2006.354

168. de Ridder DJ, Villacorte L, Verliefde ARD, et al. Modeling equilibrium adsorption of organic micropollutants onto activated carbon. Water Res. 2010;44(10):3077-3086. doi:10.1016/j.watres.2010.02.034

138 | References

169. Kimura K, Amy G, Drewes JE, Heberer T, Kim T-U, Watanabe Y. Rejection of organic micropollutants (disinfection by-products, endocrine disrupting compounds, and pharmaceutically active compounds) by NF/RO membranes. J Membr Sci. 2003;227(1-2):113-121. doi:10.1016/j.memsci.2003.09.005

170. Lim M-H, Seah H. NEWater: A key element of Singapore’s water sustainability.

In:;2013.https://content.knovel.com/content/pdf/8261/00075_03.pdf?ekey=f2catq34hRpPeFNQX6qFQzorEw2X1dB6XAY_LrEPHMPko#page=2. Accessed July 28, 2017.

171. Marsh H, Reinoso FR. Activated Carbon. Elsevier; 2006.

172. Mailler R, Gasperi J, Coquet Y, et al. Removal of emerging micropollutants from wastewater by activated carbon adsorption: Experimental study of different activated carbons and factors influencing the adsorption of micropollutants in wastewater. J Environ Chem Eng. 2016;4(1):1102-1109. doi:10.1016/j.jece.2016.01.018

173. Yu J, Lv L, Lan P, Zhang S, Pan B, Zhang W. Effect of effluent organic matter on the adsorption of perfluorinated compounds onto activated carbon. J Hazard Mater. 2012;225-226:99-106. doi:10.1016/j.jhazmat.2012.04.073

174. Chowdhury ZK. Activated Carbon: Solutions for Improving Water Quality. American Water Works Association; 2013.

175. Knappe DRU, Snoeyink VL, Dagois G, De Wolfe JR. Effect of Calcium on Thermal Regeneration of GAC. J Am Water Works Assoc. 1992;84(8):73-80.

176. Margot J, Kienle C, Magnet A, et al. Treatment of micropollutants in municipal wastewater: Ozone or powdered activated carbon? Sci Total Environ. 2013;461-462:480-498. doi:10.1016/j.scitotenv.2013.05.034

177. Metzger S, Rößler A, Kapp H. Erweiterung Des Klärwerks Mannheim Um Eine Adsorptionsstufe Zur Verbesserung Der Abwasserreinigung - Spurenstoffbericht.; 2012.

178. Nicolet L, Rott U. Recirculation of powdered activated carbon for the adsorption of dyes in municipal wastewater treatment plants. Water Sci Technol. 1999;40(1):191-198.

179. Kim SD, Cho J, Kim IS, Vanderford BJ, Snyder SA. Occurrence and removal of pharmaceuticals and endocrine disruptors in South Korean surface, drinking, and waste waters. Water Res. 2007;41(5):1013-1021. doi:10.1016/j.watres.2006.06.034

180. Al-Rifai JH, Khabbaz H, Schäfer AI. Removal of pharmaceuticals and endocrine disrupting compounds in a water recycling process using reverse osmosis systems. Sep Purif Technol. 2011;77(1):60-67. doi:10.1016/j.seppur.2010.11.020

References | 139

181. Yoon Y, Westerhoff P, Snyder SA, Wert EC, Yoon J. Removal of endocrine disrupting compounds and pharmaceuticals by nanofiltration and ultrafiltration membranes. Desalination. 2007;202(1-3):16-23. doi:10.1016/j.desal.2005.12.033

182. Kosutic K, Dolar D, Asperger D, Kunst B. Removal of antibiotics from a model wastewater by RO/NF membranes. Sep Purif Technol. 2007;53(3):244-249. doi:10.1016/j.seppur.2006.07.015

183. Braeken L, Ramaekers R, Zhang Y, Maes G, Bruggen BV der, Vandecasteele C. Influence of hydrophobicity on retention in nanofiltration of aqueous solutions containing organic compounds. J Membr Sci. 2005;252(1-2):195-203. doi:10.1016/j.memsci.2004.12.017

184. Verliefde ARD, Heijman SGJ, Cornelissen ER, Amy G l., Van der Bruggen B, van Dijk JC. Rejection of trace organic pollutants with high pressure membranes (NF/RO). Environ Prog. 2008;27(2):180-188. doi:10.1002/ep.10272

185. Verliefde ARD, Heijman SGJ, Cornelissen ER, Amy G, Van der Bruggen B, van Dijk JC. Influence of electrostatic interactions on the rejection with NF and assessment of the removal efficiency during NF/GAC treatment of pharmaceutically active compounds in surface water. Water Res. 2007;41(15):3227-3240. doi:10.1016/j.watres.2007.05.022

186. Bellona C, Heil D, Yu C, Fu P, Drewes JE. The pros and cons of using nanofiltration in lieu of reverse osmosis for indirect potable reuse applications. Sep Purif Technol. 2012;85:69-76. doi:10.1016/j.seppur.2011.09.046

187. Yangali-Quintanilla V, Maeng SK, Fujioka T, Kennedy M, Amy G. Proposing nanofiltration as acceptable barrier for organic contaminants in water reuse. J Membr Sci. 2010;362(1-2):334-345. doi:10.1016/j.memsci.2010.06.058

188. Joss A, Siegrist H, Ternes TA. Are we about to upgrade wastewater treatment for removing organic micropollutants? Water Sci Technol. 2008;57(2):251-255. doi:10.2166/wst.2008.825

189. Schäfer AI, Nghiem LD, Waite TD. Removal of the Natural Hormone Estrone from Aqueous Solutions Using Nanofiltration and Reverse Osmosis. Environ Sci Technol. 2003;37(1):182-188. doi:10.1021/es0102336

190. Westerhoff P, Moon H, Minakata D, Crittenden J. Oxidation of organics in retentates from reverse osmosis wastewater reuse facilities. Water Res. 2009;43(16):3992-3998. doi:10.1016/j.watres.2009.04.010

191. Flyborg L, Björlenius B, Persson KM. Can treated municipal wastewater be reused after ozonation and nanofiltration? Results from a pilot study of pharmaceutical removal in Henriksdal WWTP, Sweden. Water Sci Technol. 2010;61(5):1113. doi:10.2166/wst.2010.029

140 | References

192. Huber MM, Göbel A, Joss A, et al. Oxidation of Pharmaceuticals during Ozonation of Municipal Wastewater Effluents:  A Pilot Study. Environ Sci Technol. 2005;39(11):4290-4299. doi:10.1021/es048396s

193. Huber MM, Ternes TA, von Gunten U. Removal of Estrogenic Activity and Formation of Oxidation Products during Ozonation of 17α-Ethinylestradiol. Environ Sci Technol. 2004;38(19):5177-5186. doi:10.1021/es035205x

194. José HJ, Gebhardt W, Moreira RFPM, Pinnekamp J, Schröder HF. Advanced Oxidation Processes for the Elimination of Drugs Resisting Biological Membrane Treatment. Ozone Sci Eng. 2010;32(5):305-312. doi:10.1080/01919512.2010.508194

195. Ternes TA, Stüber J, Herrmann N, et al. Ozonation: a tool for removal of pharmaceuticals, contrast media and musk fragrances from wastewater? Water Res. 2003;37(8):1976-1982. doi:10.1016/S0043-1354(02)00570-5

196. Zimmermann SG, Wittenwiler M, Hollender J, et al. Kinetic assessment and modeling of an ozonation step for full-scale municipal wastewater treatment: Micropollutant oxidation, by-product formation and disinfection. Water Res. 2011;45(2):605-617. doi:10.1016/j.watres.2010.07.080

197. Techcommentary. Advanced Oxidation Processes for Treatment of Industrial Waste EPRI Community Environmental Center Publ. No. 1.; 1996.

198. Ahuja S. Water Reclamation and Sustainability. Burlington: Burlington : Elsevier Science; 2014.

199. Van Leeuwen J, Prinsloo J. Ozonation at the Stander water reclamation plant. Water SA. 1980;6(2):96–102.

200. Arbetsmiljöverket. Hygieniska Gränsvärden (AFS 2015:7), Arbetsmiljöverkets Författningssamling.; 2015. https://www.av.se/globalassets/filer/publikationer/foreskrifter/hygieniska-gransvarden-afs-2015-7.pdf. Accessed July 26, 2017.

201. Bocci V. OZONE. 2nd ed.. Dordrecht : Springer Netherlands : Imprint: Springer; 2010.

202. Lee Y, von Gunten U. Oxidative transformation of micropollutants during municipal wastewater treatment: Comparison of kinetic aspects of selective (chlorine, chlorine dioxide, ferrateVI, and ozone) and non-selective oxidants (hydroxyl radical). 2010. http://ac.els-cdn.com/S0043135409008008/1-s2.0-S0043135409008008-main.pdf?_tid=f0c969ca-71e3-11e7-93ff-00000aab0f02&acdnat=1501061127_5685dd4ad7fa48c7e833dfa1fac50266. Accessed July 26, 2017.

203. Andreozzi R, Canterino M, Marotta R, Paxeus N. Antibiotic removal from wastewaters: The ozonation of amoxicillin. J Hazard Mater. 2005;122(3):243-250. doi:10.1016/j.jhazmat.2005.03.004

References | 141

204. Gottschalk C, Libra JA, Saupe A. Ozonation of Water and Waste Water: A Practical Guide to Understanding Ozone and Its Applications. John Wiley & Sons; 2009.

205. Zwiener C, Frimmel FH. Oxidative treatment of pharmaceuticals in water. Water Res. 2000;34(6):1881–1885.

206. Vosmaer A. Ozone, Its Manufacture, Properties and Uses. D. Van Nostrand Company; 1916.

207. American Water Works Res AWWR, Langlais B, Reckhow DA, Brink DR. Ozone in Water Treatment: Application and Engineering. CRC Press; 1991.

208. McArdell C. The first full-scale advanced ozonation plant in the Dübendorf WWTP running; the new Swiss water protection act approved. Norman Bull. 2015;4:36–37.

209. Glaze WH, Lay Y, Kang J-W. Advanced oxidation processes. A kinetic model for the oxidation of 1, 2-dibromo-3-chloropropane in water by the combination of hydrogen peroxide and UV radiation. Ind Eng Chem Res. 1995;34(7):2314–2323.

210. Glaze WH, Kang J-W, Chapin DH. The Chemistry of Water Treatment Processes Involving Ozone, Hydrogen Peroxide and Ultraviolet Radiation. Ozone Sci Eng. 1987;9(4):335-352. doi:10.1080/01919518708552148

211. Andreozzi R, Marotta R, Pinto G, Pollio A. Carbamazepine in water: persistence in the environment, ozonation treatment and preliminary assessment on algal toxicity. Water Res. 2002;36(11):2869-2877. doi:10.1016/S0043-1354(01)00500-0

212. Calisto V, Esteves VI. Psychiatric pharmaceuticals in the environment. 2009;77(10):1257–1274. doi:10.1016/j.chemosphere.2009.09.021

213. Cardoza LA, Knapp CW, Larive CK, Belden JB, Lydy M, Graham DW. Factors Affecting the Fate of Ciprofloxacin in Aquatic Field Systems. Water Air Soil Pollut. 2005;161(1-4):383-398. doi:10.1007/s11270-005-5550-6

214. Lindholm-Lehto PC, Ahkola HSJ, Knuutinen JS, Herve SH. Widespread occurrence and seasonal variation of pharmaceuticals in surface waters and municipal wastewater treatment plants in central Finland. Environ Sci Pollut Res. 2016;23(8):7985-7997. doi:10.1007/s11356-015-5997-y

215. Lindholm-Lehto PC, Ahkola HSJ, Knuutinen JS, Herve SH. Occurrence of pharmaceuticals in municipal wastewater, in the recipient water, and sedimented particles of northern Lake Päijänne. Environ Sci Pollut Res. 2015;22(21):17209-17223. doi:10.1007/s11356-015-4908-6

142 | References

216. Poiger T, Buser H-R, Müller MD. Photodegradation of the pharmaceutical drug diclofenac in a lake: Pathway, field measurements, and mathematical modeling. Environ Toxicol Chem. 2001;20(2):256-263. doi:10.1002/etc.5620200205

217. Vogna D, Marotta R, Napolitano A, Andreozzi R, d’Ischia M. Advanced oxidation of the pharmaceutical drug diclofenac with UV/H2O2 and ozone. Water Res. 2004;38(2):414-422. doi:10.1016/j.watres.2003.09.028

218. Andreozzi R, Caprio V, Marotta R, Vogna D. Paracetamol oxidation from aqueous solutions by means of ozonation and H 2 O 2/UV system. Water Res. 2003;37(5):993–1004.

219. Felis E, Marciocha D, Surmacz-Gorska J, Miksch K. Photochemical degradation of naproxen in the aquatic environment. Water Sci Technol. 2007;55(12):281. doi:10.2166/wst.2007.417

220. Hofmann J, Freier U, Wecks M, Hohmann S. Degradation of diclofenac in water by heterogeneous catalytic oxidation with H2O2. Appl Catal B Environ. 2007;70(1-4):447-451. doi:10.1016/j.apcatb.2005.11.023

221. Kanakaraju D, Glass BD, Oelgemöller M. Titanium dioxide photocatalysis for pharmaceutical wastewater treatment. Environ Chem Lett. 2014;12(1):27-47. doi:10.1007/s10311-013-0428-0

222. Lazar M, Varghese S, Nair S. Photocatalytic Water Treatment by Titanium Dioxide: Recent Updates. Catalysts. 2012;2(4):572-601. doi:10.3390/catal2040572

223. Serpone N. Relative photonic efficiencies and quantum yields in heterogeneous photocatalysis. J Photochem Photobiol Chem. 1997;104(1-3):1–12.

224. Calza P, Sakkas V, Medana C, et al. Photocatalytic degradation study of diclofenac over aqueous TiO2 suspensions. Appl Catal B Environ. 2006;67(3-4):197-205. doi:10.1016/j.apcatb.2006.04.021

225. Coleman HM, Abdullah MI, Eggins BR, Palmer FL. Photocatalytic degradation of 17β-oestradiol, oestriol and 17α-ethynyloestradiol in water monitored using fluorescence spectroscopy. Appl Catal B Environ. 2005;55(1):23-30. doi:10.1016/j.apcatb.2004.07.004

226. Hey G, Vega SR, Fick J, et al. Removal of pharmaceuticals in WWTP effluents by ozone and hydrogen peroxide. Water SA. 2014;40(1):165-174. doi:10.4314/wsa.v40i1.20

227. Kim IH, Tanaka H, Iwasaki T, Takubo T, Morioka T, Kato Y. Classification of the degradability of 30 pharmaceuticals in water with ozone, UV and H2O2. Water Sci Technol. 2008;57(2):195-200. doi:10.2166/wst.2008.808

References | 143

228. von Gunten U. Ozonation of drinking water: Part I. Oxidation kinetics and product formation. Water Res. 2003;37(7):1443-1467. doi:10.1016/S0043-1354(02)00457-8

229. WHO. Chlorine Dioxide, Chlorite and Chlorate in Drinking water Background document -WHO/ FWC/WSH/16.49. http://www.who.int/water_sanitation_health/water-quality/guidelines/chemicals/chlorine-dioxide-chlorite-chlorate-background-jan17.pdf?ua=1. Published 2016. Accessed July 27, 2017.

230. Aieta EM, Berg JD, Roberts PV, Cooper RC. Comparison of Chlorine Dioxide and Chlorine in Wastewater Disinfection. J Water Pollut Control Fed. 1980;52(4):810-822.

231. Svecevičius G, Šyvokienė J, Stasiūnaitė P, Mickėnienė L. Acute and Chronic Toxicity of Chlorine Dioxide (ClO2) and Chlorite (ClO2ˉ) to Rainbow Trout (Oncorhynchus mykiss) (4 pp). Environ Sci Pollut Res. 2005;12(5):302-305. doi:10.1065/espr2005.04.248

232. van Wijk DJ, Kroon SGM, Garttener-Arends ICM. Toxicity of Chlorate and Chlorite to Selected Species of Algae, Bacteria, and Fungi. Ecotoxicol Environ Saf. 1998;40(3):206-211. doi:10.1006/eesa.1998.1685

233. Hey G, Ledin A, Jansen J la C, Andersen HR. Removal of pharmaceuticals in biologically treated wastewater by chlorine dioxide or peracetic acid. Environ Technol. 2012;33(9):1041-1047. doi:10.1080/09593330.2011.606282

234. Huber MM, Korhonen S, Ternes TA, von Gunten U. Oxidation of pharmaceuticals during water treatment with chlorine dioxide. Water Res. 2005;39(15):3607-3617. doi:10.1016/j.watres.2005.05.040

235. Hey G, Grabic R, Ledin A, la Cour Jansen J, Andersen HR. Oxidation of pharmaceuticals by chlorine dioxide in biologically treated wastewater. Chem Eng J. 2012;185-186:236-242. doi:10.1016/j.cej.2012.01.093

236. Kitis M. Disinfection of wastewater with peracetic acid: a review. Environ Int. 2004;30(1):47-55. doi:10.1016/S0160-4120(03)00147-8

237. Nguyen L, Graffe A, Candy T. Peracetic acid as a method of effluent wastewater disinfection in Langley, BC. http://www.usptechnologies.com/wp-content/uploads/PAA-Langley-BC-WEFTEC-2014-1.pdf. Published 2014. Accessed July 27, 2017.

238. Koivunen J, Heinonen-Tanski H. Peracetic acid (PAA) disinfection of primary, secondary and tertiary treated municipal wastewaters. Water Res. 2005;39(18):4445-4453. doi:10.1016/j.watres.2005.08.016

239. Dell’Erba A, Falsanisi D, Liberti L, Notarnicola M, Santoro D. Disinfection by-products formation during wastewater disinfection with peracetic acid. Desalination. 2007;215(1-3):177-186. doi:10.1016/j.desal.2006.08.021

144 | References

240. Nödler K, Voutsa D, Licha T. Polar organic micropollutants in the coastal environment of different marine systems. Mar Pollut Bull. 2014;85(1):50-59. doi:10.1016/j.marpolbul.2014.06.024

241. Wahlberg C, Björlenius B, Paxéus N. Fluxes of 13 selected pharmaceuticals in the water cycle of Stockholm, Sweden. Water Sci Technol. 2011;63(8):1772-1780. doi:10.2166/wst.2011.124

242. Boxall ABA. The environmental side effects of medication. EMBO Rep. 2004;5(12):1110-1116. doi:10.1038/sj.embor.7400307

243. Grabic R, Fick J, Lindberg RH, Fedorova G, Tysklind M. Multi-residue method for trace level determination of pharmaceuticals in environmental samples using liquid chromatography coupled to triple quadrupole mass spectrometry. Talanta. 2012;100:183-195. doi:10.1016/j.talanta.2012.08.032

244. OECD. Health at a Glance 2015. OECD Publishing; 2015. doi:10.1787/health_glance-2015-en

245. Socialstyrelsen. Statistikdatabas för läkemedel. http://www.socialstyrelsen.se/statistik/statistikdatabas/lakemedel. Published 2017. Accessed March 15, 2017.

246. Shorvon S, Perucca E, Jr JE. The Treatment of Epilepsy. John Wiley & Sons; 2015.

247. Vida J. Anticonvulsants: Medicinal Chemistry, Vol.15. United States: Academic Press; 2013.

248. Zhang Y, Geissen S-U. Prediction of carbamazepine in sewage treatment plant effluents and its implications for control strategies of pharmaceutical aquatic contamination. ResearchGate. 2010;80(11):1345-1352. doi:10.1016/j.chemosphere.2010.06.030

249. Tlusta E, Ulmanova D, Vitasek Z, Vlcek J. Sledování spotřeby starších a nových antiepileptik v České republice v obdobi 1999-2004. Čes Slov Farm. 2006;56(1):37-41.

250. Nordiska läkemedelsnämnden. Nordisk Läkemedelsstatistik. 1975/1977. Vol. 1. Vol. 2, Läkemedelsregister Med Klassificering Och Definierad Dygnsdos. Helsingfors : Utg.; 1979.

251. Nordiska läkemedelsnämnden. Nordisk Läkemedelsstatistik. 1978/1980:1. Helsingfors: Utg.; 1982.

252. Tsiropoulos I, Gichangi A, Andersen M, Bjerrum L, Gaist D, Hallas J. Trends in utilization of antiepileptic drugs in Denmark. Acta Neurol Scand. 2006;113(6):405-411. doi:10.1111/j.1600-0404.2006.00639.x

References | 145

253. Sundhedsdatastyrelsen. Sundhedsdatastyrelsen - Statistikker. http://medstat.dk/en. Published 2017. Accessed March 15, 2017.

254. Estonian State Agency of Medicines, Latvian State Agency of Medicines, Lithuanian State Medicines Control Agency. Baltic Statistics on Medicines 2010-2012. Tartu; 2013. file:///C:/Users/Berndt/Dropbox/Briggen/API%20consumption/balticstatisticsonmedicines20102012.pdf. Accessed March 16, 2017.

255. National Institute for Health Development. Statistics on medicines. Health Statistics and Health Research Database. http://pxweb.tai.ee/PXWeb2015/pxweb/en/06Ravimistatistika. Published 2017. Accessed March 16, 2017.

256. Õun A, Haldre S, Mägi M. Use of antiepileptic drugs in Estonia: an epidemiologic study of adult epilepsy. Eur J Neurol. 2006;13(5):465-470. doi:10.1111/j.1468-1331.2006.01268.x

257. National Agency for Medicines. Finnish Statistics on Medicines 2007. Helsinki; 2008. file:///C:/Users/Berndt/Dropbox/Briggen/API%20consumption/Finland/Soumen%20laaketilasto07_2.pdf. Accessed March 16, 2017.

258. Finnish Medicines Agency Fimea. Finnish Statistics on Medicines 2010. Helsinki; 2011. file:///C:/Users/Berndt/Dropbox/Briggen/API%20consumption/Finland/20681_Suomen_laaketilasto_2010_netti.PDF. Accessed March 16, 2017.

259. Finnish Medicines Agency Fimea. Finnish Statistics on Medicines 2013. Helsinki; 2014. file:///C:/Users/Berndt/Dropbox/Briggen/API%20consumption/Finland/27596_SLT_2013_net.pdf. Accessed March 16, 2017.

260. Schwabe U, Paffrath D. Arzneiverordnungs-Report 1988-2014. Berlin; 2014. http://www.springer.com/de/book/9783662503508. Accessed March 16, 2017.

261. Zāļu valsts aģentūra. Zāļu Patēriņa Statistika-2011 - Statistics on Medicines Consumption. Riga; 2012. https://www.zva.gov.lv/doc_upl/zalu-paterina-statistika-2011.pdf. Accessed March 16, 2017.

262. Zāļu valsts aģentūra. Zāļu Patēriņa Statistika - Statistics on Medicines Consumption 2015. Riga; 2016. https://www.zva.gov.lv/doc_upl/zva-zstat-2015.pdf?b. Accessed March 16, 2017.

146 | References

263. Latvian State Agency of Medicines, Estonian State Agency of Medicines, Lithuanian State Medicines Control AgencyLithuanian State Medicines Control Agency. Baltic Statistics on Medicines 2013-2015. Riga; 2016. file:///C:/Users/Berndt/Dropbox/Briggen/API%20consumption/BalticStatisticsonMedicines20132015forweb.pdf. Accessed March 16, 2017.

264. Landmark CJ, Rytter E, Johannessen SI. Clinical use of antiepileptic drugs at a referral centre for epilepsy. Seizure. 2007;16(4):356-364. doi:10.1016/j.seizure.2007.02.006

265. Rønning M, Sakshaug S, Strøm H. Drug consumption in Norway 2005-2009. Oslo: Folkehelseinstituttet; 2010.

266. Sakshaug S, Strøm H, Berg C. Drug Consumption in Norway 2010–2014. Oslo: Norwegian Institute of Public Health; 2015. file:///C:/Users/Berndt/Dropbox/Briggen/API%20consumption/Legemiddelstatistikk_2015_1_WEB.pdf. Accessed March 15, 2017.

267. Apoteket. Svensk Läkemedelsstatistik 2000. Stockholm: Apoteket AB; 2001.

268. Apoteksbolaget. Svensk Läkemedelsstatistik 1991. Stockholm: Apoteksbolaget; 1991.

269. Apoteksbolaget. Svensk Läkemedelsstatistik 1995. Stockholm: Apoteksbolaget; 1995.

270. Apoteksbolaget. Svensk Läkemedelsstatistik 1996. Stockholm: Apoteksbolaget; 1996.

271. Ekama G., Siebritz IP, Marais G. Considerations in the process design of nutrient removal activated sludge processes. Water Sci Technol. 1983;15(3-4).

272. OECD. OECD.Stat Pharmaceutical Market. OECD.Stat. https://stats.oecd.org/Index.aspx?DataSetCode=HEALTH_PHMC#. Published 2018. Accessed August 24, 2018.

273. Loos R, Locoro G, Comero S, et al. Pan-European survey on the occurrence of selected polar organic persistent pollutants in ground water. Water Res. 2010;44(14):4115-4126. doi:10.1016/j.watres.2010.05.032

274. Elmgren R. Understanding Human Impact on the Baltic Ecosystem: Changing Views in Recent Decades. Ambio. 2001;30(4/5):222-231.

275. Korpinen S, Meski L, Andersen JH, Laamanen M. Human pressures and their potential impact on the Baltic Sea ecosystem. Ecol Indic. 2012;15(1):105-114. doi:10.1016/j.ecolind.2011.09.023

References | 147

276. Tixier C, Singer HP, Oellers S, Müller SR. Occurrence and Fate of Carbamazepine, Clofibric Acid, Diclofenac, Ibuprofen, Ketoprofen, and Naproxen in Surface Waters. Environ Sci Technol. 2003;37(6):1061-1068. doi:10.1021/es025834r

277. Helbing W, Zagarese H. UV Effects in Aquatic Organisms and Ecosystems.; 2003. doi:10.1039/9781847552266

278. Piazena H, Häder D-P. Penetration of Solar UV Irradiation in Coastal Lagoons of the Southern Baltic Sea and Its Effect on Phytoplankton Communities. Photochem Photobiol. 1994;60(5):463-469. doi:10.1111/j.1751-1097.1994.tb05135.x

279. Zerefos CS, Bais AF. Solar Ultraviolet Radiation: Modelling, Measurements and Effects. Springer Science & Business Media; 1997.

280. Lindim C, van Gils J, Cousins IT, et al. Model-predicted occurrence of multiple pharmaceuticals in Swedish surface waters and their flushing to the Baltic Sea. Environ Pollut. 2017;223:595-604. doi:10.1016/j.envpol.2017.01.062

281. Fick J, Lindberg RH, Tysklind M, Larsson DGJ. Predicted critical environmental concentrations for 500 pharmaceuticals. Regul Toxicol Pharmacol. 2010;58(3):516-523. doi:10.1016/j.yrtph.2010.08.025

282. Oekotoxzentrum. Proposals for Acute and Chronic Quality Standards. http://www.ecotoxcentre.ch/. Published 2018. Accessed October 26, 2018.

283. Vieno NM, Tuhkanen T, Kronberg L. Seasonal Variation in the Occurrence of Pharmaceuticals in Effluents from a Sewage Treatment Plant and in the Recipient Water. Environ Sci Technol. 2005;39(21):8220-8226. doi:10.1021/es051124k

284. Kiso Y, Kon T, Kitao T, Nishimura K. Rejection properties of alkyl phthalates with nanofiltration membranes. J Membr Sci. 2001;182(1):205-214. doi:10.1016/S0376-7388(00)00567-6

285. Kiso Y, Sugiura Y, Kitao T, Nishimura K. Effects of hydrophobicity and molecular size on rejection of aromatic pesticides with nanofiltration membranes. J Membr Sci. 2001;192(1):1-10. doi:10.1016/S0376-7388(01)00411-2

286. Ozaki H, Li H. Rejection of organic compounds by ultra-low pressure reverse osmosis membrane. Water Res. 2002;36(1):123-130. doi:10.1016/S0043-1354(01)00197-X

287. Kimura K, Toshima S, Amy G, Watanabe Y. Rejection of neutral endocrine disrupting compounds (EDCs) and pharmaceutical active compounds (PhACs) by RO membranes. J Membr Sci. 2004;245(1):71-78. doi:10.1016/j.memsci.2004.07.018

148 | References

288. Xu P, Heberer T. Rejection of Emerging Organic Micropollutants in Nanofiltration–Reverse Osmosis Membrane Applications. Water Environ Res. 2005;77(1):9.

289. Yoon Y, Lueptow RM. Removal of organic contaminants by RO and NF membranes. J Membr Sci. 2005;261(1):76-86. doi:10.1016/j.memsci.2005.03.038

290. Yoon Y, Westerhoff P, Snyder SA, Wert EC. Nanofiltration and ultrafiltration of endocrine disrupting compounds, pharmaceuticals and personal care products. J Membr Sci. 2006;270(1):88-100. doi:10.1016/j.memsci.2005.06.045

291. Boussu K, Vandecasteele C, Van der Bruggen B. Relation between membrane characteristics and performance in nanofiltration. J Membr Sci. 2008;310(1):51-65. doi:10.1016/j.memsci.2007.10.030

292. Chemicalize - Instant Cheminformatics Solutions. https://chemicalize.com/. Accessed April 26, 2018.

293. ChemSpider | Search and share chemistry. http://www.chemspider.com/. Accessed April 26, 2018.

294. Verliefde ARD, Cornelissen ER, Heijman SGJ, et al. Construction and validation of a full-scale model for rejection of organic micropollutants by NF membranes. J Membr Sci. 2009;339(1):10-20. doi:10.1016/j.memsci.2009.03.038

295. Mulder M, Antakyali D, Ante S. Costs of Removal of Micropollutants from Effluents of Municipal Wastewater Treatment Plants - General Cost Estimates for the Netherlands Based on Implemented Full Scale Post Treatments of Effluents of Wastewater Treatment Plants in Germany and Switzerland STOWA and Waterschap the Dommel, The Netherlands.; :55.

296. Björlenius B. Fördjupning Avloppsreningsprocesser. Stockholm: Stockholm Vatten; 2008. http://www.stockholmvattenochavfall.se/globalassets/pdf1/avloppsvatten/reningsprocessen/fordjupning-avloppsreningsprocess. Accessed March 29, 2018.

297. Jonsson L, Plaza E, Hultman B. Experiences of nitrogen and phosphorus removal in deep-bed filters in the Stockholm area. Water Sci Technol. 1997;36(1):183-190. doi:10.1016/S0273-1223(97)00351-X

298. Nam S-W, Choi D-J, Kim S-K, Her N, Zoh K-D. Adsorption characteristics of selected hydrophilic and hydrophobic micropollutants in water using activated carbon. J Hazard Mater. 2014;270:144-152. doi:10.1016/j.jhazmat.2014.01.037

References | 149

299. Snyder SA, Adham S, Redding AM, et al. Role of membranes and activated carbon in the removal of endocrine disruptors and pharmaceuticals. Desalination. 2007;202(1-3):156-181. doi:10.1016/j.desal.2005.12.052

300. Grover DP, Zhou JL, Frickers PE, Readman JW. Improved removal of estrogenic and pharmaceutical compounds in sewage effluent by full scale granular activated carbon: Impact on receiving river water. J Hazard Mater. 2011;185(2):1005-1011. doi:10.1016/j.jhazmat.2010.10.005

301. Boehler M, Zwickenpflug B, Hollender J, Ternes T, Joss A, Siegrist H. Removal of micropollutants in municipal wastewater treatment plants by powder-activated carbon. Water Sci Technol. 2012;66(10):2115. doi:10.2166/wst.2012.353

302. Mailler R, Gasperi J, Coquet Y, et al. Study of a large scale powdered activated carbon pilot: Removals of a wide range of emerging and priority micropollutants from wastewater treatment plant effluents. Water Res. 2015;72:315-330. doi:10.1016/j.watres.2014.10.047

303. Ridder DJ de, Verliefde ARD, Heijman SGJ, et al. Influence of natural organic matter on equilibrium adsorption of neutral and charged pharmaceuticals onto activated carbon. Water Sci Technol. 2011;63(3):416-423. doi:10.2166/wst.2011.237

304. Meinel F, Zietzschmann F, Ruhl AS, Sperlich A, Jekel M. The benefits of powdered activated carbon recirculation for micropollutant removal in advanced wastewater treatment. Water Res. 2016;91:97-103. doi:10.1016/j.watres.2016.01.009

305. Meinel F, Sperlich A, Jekel M. Pilot-scale study of powdered activated carbon recirculation for micropollutant removal. Water Sci Technol. 2016;74(4):927-934. doi:10.2166/wst.2016.273

306. Antoniou MG, Hey G, Rodríguez Vega S, et al. Required ozone doses for removing pharmaceuticals from wastewater effluents. Sci Total Environ. 2013;456-457:42-49. doi:10.1016/j.scitotenv.2013.03.072

307. Lee Y, Kovalova L, McArdell CS, von Gunten U. Prediction of micropollutant elimination during ozonation of a hospital wastewater effluent. Water Res. 2014;64:134-148. doi:10.1016/j.watres.2014.06.027

308. Tsiamis V. Biomarker Response in Rainbow Trout (Oncorhynchus Mykiss) Exposed to Additionally Treated Sewage Treatment Plant Effluent. Ekotoxikologiska avdelningen, Uppsala universitet; 2013. http://clamator.its.uu.se/uploader/858/BIOMSc-13-040-Tsiamis-Viktor-report.pdf. Accessed August 12, 2017.

150 | References

309. Cuklev F, Gunnarsson L, Cvijovic M, et al. Global hepatic gene expression in rainbow trout exposed to sewage effluents: A comparison of different sewage treatment technologies. Sci Total Environ. 2012;427–428:106-114. doi:10.1016/j.scitotenv.2012.04.009

310. Gunnarsson L, Adolfsson-Erici M, Björlenius B, Rutgersson C, Förlin L, Larsson DGJ. Comparison of six different sewage treatment processes—Reduction of estrogenic substances and effects on gene expression in exposed male fish. Sci Total Environ. 2009;407(19):5235-5242. doi:10.1016/j.scitotenv.2009.06.018