1

Johnson, A. C., Bowes, M. J., Crossley, A., Jarvie, H. P., Jurkschat, K., Jürgens, M. D., Lawlor, A. J., Park,

B., Rowland, P., Spurgeon, D., Svendsen, C., Thompson, I. P., Barnes, R. J., Williams, R. J. and Xu, N.

(2011). An assessment of the fate, behaviour and environmental risk associated with sunscreen

TiO2 nanoparticles in UK field scenarios. Science of the Total Environment 409(13): 2503-2510.

DOI: 10.1016/j.scitotenv.2011.03.040

An assessment of the fate, behaviour and environmental risk associated with sunscreen TiO2 nanoparticles in UK field

scenarios

Andrew C. Johnson1, Michael J. Bowes1, Alison Crossley2, Helen P. Jarvie1, Kerstin Jurkschat2, Monika D. Jürgens1, Alan J. Lawlor3, Barry Park.4, Phillip Rowland3, David Spurgeon1, Claus Svendsen1, Ian P. Thompson5, Robert J. Barnes5, Richard J. Williams1, Nan Xu1*

1Centre for Ecology and Hydrology, Wallingford, OX10 8BB

2Department of Materials, Oxford University Begbroke Science Park, Sandy Lane, Yarnton, Oxford OX5 1PF

3Centre for Ecology and Hydrology, Lancaster, LA1 4AP

4GBP Consulting Ltd, Purton, Swindon, SN5 4EJ

5Department of Engineering Science, Oxford University Begbroke Science Park, Sandy Lane, Yarnton,

Oxford OX5 1PF

*Current address School of Environment and Energy, Peking University Shenzhen Graduate School, 518055, Shenzhen, China

Email contact: ajo@ceh.ac.uk

ABSTRACT

______________________________________________________________________________________

The fate of Ti was examined in an activated sludge plant serving over 200,000 people. These

studies revealed a decrease of 30 to 3.2 µg/L of Ti<0.45 µm from influent to effluent and a

calculated Ti presence of 305 mg/kg DW in wasted sludge. Thus, using sludge as a fertiliser would

result in a predicted deposition of up to 250 mg/m2 of Ti to soil surfaces using a recommended

maximal agricultural application rate. Given the major use of TiO2 in many industrial and domestic

applications where loss to the sewer is possible, this measured Ti was presumed to have been

largely TiO2, a proportion of which will be nanoparticle sized. To assess the behaviour of

engineered nanoparticle (ENP) TiO2 in sewage and toxicology studies, Optisol (Oxonica Materials

Ltd) and P25 (Evonik Industries AG), which are representative of forms used in sunscreen and

cosmetic products, were used. These revealed a close association of TiO2 ENPs with activated

sludge. Using commercial information on consumption, and removal rates for sewage treatment,

predictions were made for river water concentrations for sunscreen TiO2 ENPs for the Anglian and

Thames regions in Southern England. The highest predicted value from these exercises was 8.8

µg/L for the Thames region in which it was assumed one in four people used the recommended

application of sunscreen during a low flow (Q95) period. Ecotoxicological studies using potentially

vulnerable species indicated that 1,000 µg/L TiO2 ENP did not affect the viability of a mixed

community of river bacteria in the presence of UV light. Direct exposure to TiO2 ENPs did not

impair the immuno-effectiveness of earthworm coelomocyte cells at concentrations greatly above

those predicted for sewage sludge.

Key words

TiO2 nanoparticle, sunscreen, sewage, exposure, prediction, river water, ecotoxicity, earthworms

2

1. Introduction

A recent study comparing TiO2, ZnO, Ag, Carbon nanotubes and fullerene nanoparticles

concluded that TiO2 nanoparticles where the most likely to enter the natural environment in the

largest quantities (Gottschalk et al., 2009). One application for TiO2 engineered nanoparticles (ENP)

has been in sunscreen products where they have been used for many years. A number of studies

have indicated they do not penetrate the skin to any great extent suggesting they would be relatively

easy to remove from the skin during washing and so enter the wastewater stream (Mavon et al.,

2007; Nohynek et al., 2008). They therefore represent an example of an ENP where widespread

contamination of the natural environment is possible. Over recent years a number of

ecotoxicological studies on TiO2 ENPs in (largely) aquatic organisms such as bacteria, algae,

zooplankton and fish have been carried out (Adams et al., 2006; Federici et al., 2007; Aruoja et al.,

2009; Lee et al., 2009; Ramsden et al., 2009). The value of such ecotoxicity data would be

immeasurably increased if it could be compared against potential exposure levels in the

environment. Risk assessment traditionally is based on a comparison of the measured

environmental concentration (MEC), or predicted environmental concentration (PEC) with an effect

concentration, such as a no observable effect concentration (NOEC) or predicted no effect

concentration (PNEC) so in this case whilst we now have some information to derive NOECs and

PNECs, there is still little to go on for PECs.

Until plentiful measurements of TiO2 ENPs in aquatic environments become available, it

will be necessary to have recourse to predictive (modelling) approaches. Such approaches begin

with obtaining commercial information on consumption of the product. Recently, multi-media

modelling approaches have been attempted to assess the annual mass of TiO2 ENPs that would

reach US, European and Swiss aquatic and terrestrial environments (Mueller and Nowack, 2008;

Gottschalk et al., 2009). However, to assess the contribution of the sunscreen component of TiO2

ENPs more information is needed on their particular consumption and use characteristics. To reach

the wider environment, the most important route for all TiO2 ENPs will be via sewage treatment

plants (STPs). Some laboratory fate studies have suggested that metal oxide ENPs will be attracted

to sludge particles (Limbach et al., 2008) and there is laboratory and field evidence for this with

TiO2 ENPs (Kiser et al., 2009). For chemicals and substances emanating from point sources there

are enormously wide variations in loadings and dilutions along rivers, between regions and between

nations (Anderson et al., 2004; Ort et al., 2009; Williams et al., 2009). This variation needs to be

considered when generating realistic PECs. However, where a conservative (worst case) PEC is

substantially below a NOEC/PNEC further efforts to improve model precision are unwarranted

(Johnson et al., 2008). Following capture in sewage biosolids, discharge to soil may be the most

important route for TiO2 ENPs to the wider environment, particularly in countries where such

3

products are used as a soil fertiliser. This situation also raises the possibility of indirect

contamination of water courses following leaching from the biosolids during surface runoff.

Whilst the discharge of TiO2 ENPs, including the sunscreen variety, to the wider

environment may raise concerns, it is worth recalling that TiO2 itself is not a novel molecule. Ti is

a major oxide in rocks and is typically present at 0.5% in the continental crust (Taylor and

McLennan, 1985) as the naturally occurring minerals rutile, anatase, or brookite (Greenwood and

Earnshaw, 1997). An assessment of river bed-sediments has shown Ti is common across the UK

with a median concentration of 5.9 g/kg TiO2 with a maximum value of 91 g/kg (Johnson and

Breward, 2004). The industrial use of TiO2 began in 1923 with current worldwide production

estimated at over 4 million tonnes (http://chemlink.com.au/titanium.htm). The three major TiO2

markets comprise paints, accounting for up to 60% of tonnage, with plastics and paper accounting

for approximately 25% and 15% respectively, leaving TiO2 ENPs as a fraction of a percent of the

overall tonnage (http://chemlink.com.au/titanium.htm). The major markets for TiO2 ENPs are in

cosmetics, plastics and coatings applications of which about 5,500 t/yr are produced and consumed

within the US and Europe (Gottschalk et al., 2009). Recent research has demonstrated the

propensity for TiO2 (20-300 nm) to leach from exterior paints and this is likely to be an important

anthropogenic source to the environment (Kaegi et al., 2008). Thus, colloidal if not nanoscale TiO2

has been a natural part of soil and river sediments for millennia and must also have been discharged

by man as TiO2 whitening pigments into the aquatic environment for at least the last 50 years. This

presence in the environment will certainly make distinguishing TiO2 ENPs from the geogenic and

anthropogenic background problematic but it also suggests that wildlife already has a long history

of exposure to this molecule. Indeed there is an argument that river aquatic wildlife is already

exposed to natural nanoscale metal oxides such as TiO2 (Wigginton et al., 2007).

To improve our risk assessment of TiO2 ENPs, and the sunscreen forms in particular, this study

attempted to address the following questions:

• How might TiO2 ENPs behave in the sewage environment?

• What aquatic, or terrestrial, PECs in a UK setting can be expected?

• In what circumstances and at what concentration might sunscreen TiO2 ENPs harm a free-

living river bacterial community and earthworms?

• How do PECs compare to concentrations that might harm in potentially vulnerable aquatic

and terrestrial organisms?

4

2. Materials and methods

2.1. TiO2 ENPs used in the study and the microscopic examination of their association with

activated sludge

For this study TiO2 ENPs from Oxonica Materials Ltd and Evonik Industries AG were

examined (Table 1). All studies on the fate and behaviour of TiO2 ENPs, or monitoring of colloidal

Ti particles was carried out at, or using material from, a large activated sludge sewage treatment

plant (STP) in the UK. This STP receives 60,000 m3/d wastewater from a human population of

219,000 with a trade waste component equivalent to a further 105,000 people. The aeration tank is

predominantly plug flow with submerged diffuse aeration and a hydraulic retention time of around

10 h. The behaviour of TiO2 ENP and fresh activated sludge was examined with the aid of

transmission and scanning electron miscrocopy. A suspension of 100 mg/L Optisol (Table 1) and

fresh activated sludge were examined separately by transmission electron microscopy (TEM) before

being mixed together for 30 min and allowed to settle before further analysis. Carbon coated Cu

TEM grids were dipped into the water, or activated sludge suspensions, blotted on filter paper, and

TEM examination followed after 1-3 hours of drying. The samples were analysed on a JEOL 2010

analytical TEM at 200kV. This instrument has a resolution of 0.19 nm and an electron probe size

down to 0.5 nm. The instrument is equipped with an Oxford Instruments LZ5 windowless energy

dispersive X-ray spectrometer (EDS). Thus, confirmation of the titanium content of the observed

objects was available from the integral X-ray spectrometer.

Table 1

Summary of TiO2 ENP products used in the study

Company Product Advertised

size

TiO2 content

Oxonica Materials

Ltd.

Optisol™ UV Absorber

Titanium Dioxide modified with

0.67% Manganese to reduce free

radical generation on exposure to

UV light

Average 70 nm > 99%

Evonik Industries

AG

AEROXIDE® TiO2 T805

Fumed Titanium Dioxide treated

with octylsilane to achieve

hydrophobic surface

Average 21 nm ≥ 97%

Evonik Industries

AG

AEROXIDE® TiO2 P25

Hydrophilic Fumed Titanium

Dioxide

Average 21 nm ≥ 99.5%

5

2.2. Measurement of Ti in sewage sludge solids

To assess the total amount of Ti going into sewage sludge, samples were taken from influent,

activated sludge tank and final effluent of the selected STP. The influent samples were taken after

the screen at the entrance to the STP but before the primary settlement tank, the activated sludge

from about the first 1/3 of one of the activated sludge tanks and the effluent at the final discharge of

the sewage treatment plant. In all cases only the solid fraction was analysed but on the different

sampling occasions slightly different methods were used to separate the solids from the liquid: The

Oct 2006 activated sludge sample was stored frozen and filtered through glass fibre filters (Fisher,

MF 100, ca. 1.6 µm) in 2009 and the solids scraped off the dry filter. In Jan 2008, due to the

relatively low solid content of raw sewage, the method involved concentrating by centrifugation for

10 min at 17,000 g, discarding most of the supernatant and re-suspending the solids in the

remaining liquid. The concentrated influent was then filtered through 0.2 µm filters (cellulose

nitrate, Whatman) and the dried solids scraped off for analysis. The activated sludge sample from

Jan 2008 was filtered through the 0.2 µm filter before scraping off and analysis. In Feb 2009

influent, activated sludge and effluent were taken and the sample preparation was simplified such

that no centrifugation was used prior to 0.2 µm filtration and the whole dried filters with attached

solids were digested directly. Unused filters were used as a method blank and one of the two

activated sludge samples was prepared by centrifugation (30 min, 3000 g, discarding the

supernatant) instead of filtration.

Each sample (6 mg -1.6 g dry weight depending on matrix) was added to a long neck boiling

tube, followed by 10 ml of mixed 50/50 H2SO4/H2O2 digestion mixture. The tube was heated to

approx 360°C on a hot plate until all visible organic material was digested, this corresponded to a

digestion time of approximately 2.5 hours (adapted from Lomer et al., 2000). After digestion, a

clear solution was obtained which contained small amounts of undigested inorganic material,

possibly silica. The completed digests were made up to a final volume of 50 ml with ultrapure water

(resistivity > 18 MΩ cm-1

). The sulphuric acid digestion was considered appropriate to ensure

complete dissolution of TiO2 in such a complex matrix. Due to the propensity of S molecular

species, arising from the sulphuric acid matrix, to interfere with Ti isotopes, inductively coupled

plasma mass spectrometry (ICP-MS) could not be used. Therefore, Ti concentrations were

determined using inductively coupled plasma optical emission spectrometer (ICP-OES) calibrated

with matrix matched standards (5% H2SO4/H2O2) over the range 0 – 10 mg/L Ti. The calibration

was linear over this range of Ti concentration. A separate, independent analytical quality control

(AQC) solution at a concentration of 2.5 mg/L Ti was prepared from a 1 g/L stock (Spex Certiprep)

and analysed with the digest solutions. The analysis of the AQC gave a mean of 2.54 mg/L Ti, with

6

a standard deviation of 0.02 mg/L Ti. The efficiency of the H2SO4/H2O2 digestion was checked

using a reagent blank spiked with 1 mg of TiO2 ENPs (P25). The recovery of Ti from P25 was

around 70% indicating acceptable dissolution using the H2SO4/H2O2 digestion described.

2.3. Removal of < 0.45 µm Ti fraction passing through an activated sludge plant

Wastewater samples (1 litre) were taken at 1 h intervals by dip sampling between 10:00 and

14:00 on 29. July 2008. Samples were taken at three sites through the treatment process; from the

raw sewage influent, from the liquor following primary settlement, and the final effluent following

activated sludge treatment and secondary settlement. A 60 mL sub-sample was filtered

immediately through a 0.45 µm cellulose nitrate membrane (47 mm) on site. Thus, in this fraction

only colloidal (or dissolved) forms of Ti were collected and examined. The samples were acidified

to a final concentration of 1% HNO3 in the field and returned to the laboratory in a cool box where

they were stored in the dark at 4°C until analysis. The Ti concentration of each sample was

determined by inductively coupled plasma mass spectrometry (ICP-MS) (Perkin Elmer DRC II). A

separate, AQC at 5 µg/L Ti (Spex Certiprep) was prepared and analysed with the filtered effluent

solutions. The analysis of the AQC gave a mean of 4.98 µg/L Ti, with a standard deviation of 0.11

µg/L Ti. 1% HNO3 is preferred for ICP-MS analysis as there is no major molecular interference

with the Ti analytical isotope.

2.4. Predicting UK soil and surface water concentrations of sunscreen TiO2 ENPs

To predict possible soil concentrations the actual content of Ti measured in sewage sludge

in this study was used. An assessment was made on the proportion of this Ti that would be present

as TiO2 ENPs based on information on theTiO2 ENP market. Potential soil concentrations could

then be calculated based on recommended sludge applications to land.

To derive per capita effluent concentrations two separate approaches were used. In the first

approach, concentrations were based on industry figures provided by Oxonica Ltd for annual

consumption of sunscreen TiO2 ENPs, and secondly a scenario was built based around

recommended consumer sunscreen advice and an assumption on the market penetration of TiO2

ENPs sunscreen products (information from Oxonica Ltd). Thus, the first scenario was reliant on

estimates of consumption of the products together with assumptions on the period of year they

would be used and discharged. The second scenario was independent of consumption estimates and

was based on predicted consumer behaviour. The next stage was to obtain information on

7

behaviour in STPs to derive effluent concentrations. This was obtained from this study and also

from literature studies.

To calculate river water concentrations information on the human population and the

available dilution in the Thames and Anglian regions of the UK was used using information from

the LF2000-WQX model (Williams et al., 2009). The LF2000-WQX software extension combines

hydrological models estimating the magnitude and variability of flows across a catchment (Young

et al., 2003)with a range of water-quality models, including a catchment-scale water-quality model.

Overall these prediction approaches could be seen as being relatively robust in their consideration

of sewage removal and dilution in water but preliminary in their assumption of theTiO2 ENP market

and sunscreen use.

2.5. Impact of TiO2 ENPs and UV light on river water bacteria

A Phillips TEM2 FEI Tecnai 80 kV TEM was used to look at the bacterial cells and

bacterial cells plus P25 TiO2 ENPs. All samples for the TEM were prepared following standard

procedures for fixing and embedding of biological samples (Bechtel and Bulla, 1976).

Particle size distribution was measured using a CPS Disc Centrifuge (Model DC24000, CPS

Instruments, Inc.), operated at a speed of 18,500 rpm. A sucrose gradient was injected into the

spinning disc using 24% (w/v) and 8% (w/v) sucrose solutions and subsequently sealed by injecting

0.5 ml of dodecane. A polyvinyl chloride (PVC) standard solution of 0.377 µmol/L in deionised

water was used for calibration.

River water was sampled from the River Thames near Oxford (UK National Grid Reference:

SP495083) on 5. May 2009. A liquid culture of the river water bacteria was grown aerobically in

nutrient broth (Oxoid) at 24°C on a rotating platform at 120 rpm for 3 days. Bacteria were

harvested by centrifugation at 3063 × g for 30 minutes, washed twice in sterile phosphate-buffered

saline (PBS) (pH 7.4), and re-suspended in PBS. The bacteria concentration for experimental work

was determined by a viable count procedure on nutrient agar (Oxoid) plates after serial dilutions of

the culture in PBS. A concentration was chosen that would yield approximately 100 bacterial

colonies per plate. A stock solution of this bacterial concentration (103 CFU/ml) was then made up

and used for all experimental work.

In the photocatalytic experiments, stock aqueous P25 TiO2 ENP suspensions (20 g/L, 2 g/L,

0.2 g/L and 0.02 g/L in deionised water) were prepared and immediately sonicated for 1 hour in the

dark prior to the experiments. 2 ml of each stock suspension was then added to 100 ml sterile glass

beakers containing 38 ml of bacterial suspension to give TiO2 ENP concentrations of 1 g/L, 0.1 g/L,

8

0.01 g/L and 0.001 g/L. A UV control was also set up using 2 ml sterile deionised water in place of

the aqueous TiO2 ENP suspension. Three replicates were used for the control and each TiO2 ENP

concentration. The beakers were placed on a magnetic stir plate with continuous stirring (to ensure

maximal mixing and to prevent settling of the TiO2 ENPs) and illuminated with a UV lamp (LF-

206.LS, UVIlite ultraviolet lamp, Uvitec, UK) from above for 1 hour, a duration similar to previous

studies (Wei et al., 1994; Maness et al., 1999). The peak wavelength of the lamp was 365 nm

which is in the middle of the UVA band (315-400 nm). UVA is the predominant UV component of

sunlight. UV light intensity was measured with a low power photodetector (918 series, Newport,

USA) and picoammeter (model 6485, Keithley, USA). The light intensity reaching the surface of

the suspensions was approximately 4.2 W/m2. This intensity was selected as it did not kill the

bacteria but was sufficient to produce reactive oxygen species in the presence of TiO2 ENPs. The

temperature of the suspensions was taken at the first and last time point to test that this parameter

remained constant. A control (no TiO2 or UV), and dark controls (TiO2 no UV) for each TiO2 ENP

concentration, were also set up. At each time point (0, 10, 20, 30, 40, 50 and 60 minutes), 100 µl of

suspension was removed from the beaker and spread on nutrient agar plates. Three replicates were

produced for each plate. Plates were incubated at 24±2°C for 7 days.

2.6. Impact of TiO2 ENPs on earthworm immune (coelomocyte) cells

To assess potential TiO2 ENP effects on earthworms, as representatives of the soil

community, Eisenia veneta (Rosa) coelomocytes (phagocytotic cells involved in innate immunity

within the earthworm coelome) were exposed in vitro to three separate TiO2 ENPs (Table 1). After

exposure, cells were assayed to assess potential effects on their immune system activity, by

challenging them with foreign cells (rabbit blood cells). E. veneta was selected for testing as this

species favours an organic rich habitat and so is relevant to the sewage sludge exposure scenario.

Direct exposure of coelomocytes to ENP in vitro was used as a worst case, as previously these cells

have been shown to have a high sensitivity to a wide range of chemicals including metals (Goven et

al., 1993; Fugere et al., 1996; Giggleman et al., 1998).

E. veneta used were obtained from a sheltered outdoor culture, where they were reared in a

reconstituted soil made up of 33 % black top soil (Madingly Mulch, Cambridge, UK), 33 %

Sphagnum peat and 33 % composted bark (LBS Horticultural, Colne, UK). Adult worms were

maintained under test conditions (12 ± 1.5°C in constant darkness) for at least a week before the

start of the experiments. The physiological Ringer’s solution used in the studies was prepared two

days before the test. The quality of the Ringer’s solution was tested by incubating freshly extracted

earthworm coelomocytes for 2 hours, and subsequently checking cell viability using the Eosin Y

9

method described below. One day prior to use, a rabbit red blood cell stock solution was made

from a bottle of freeze-dried rabbit red blood cells (Sigma No. R-1629) and 10 ml sterile Ringer’s.

Rapidly and repeatedly passing the solution within the bottle, through a 0.25 mm gauge needle

ensured that a homogenous suspension was formed. This stock solution was stored at < 4°C.

Immediately prior to use, the rabbit blood cells were re-suspended again before preparing the 30-

fold dilute suspension for incubation with the exposed coelomocytes.

Coelomocytes for TiO2 exposure were harvested from five adult worms using a hyperdermic

syringe preloaded with 50 µl Ringer’s solution into which the cells where drawn. Harvested

coelomocyte cells were exposed to a single aqueous concentration of each TiO2 ENP for 24 h at

12°C. Nominal test concentrations were 7.8 mg/L Optisol, 25 mg/L P25 and 17.5 mg/L T805 in

distilled water. Frequency distributions of particle size in these suspensions were characterised

using NanoSight® and particle aggregation pattern by TEM. The in vitro TiO2 ENP exposures were

created by mixing a volume of the suspended phase (without any agitation) of the TiO2 ENP

solutions, with an equal volume of coelomocyte containing Ringer’s solution.

After 24 h exposure to the ENP, the coelomocyte cells were incubated for a further 24h with

equal volumes of a dilute suspension of rabbit red blood cells to stimulate an immune response to

the rabbit cells. Immune activity was measured using the procedure of Goven et al. (1993), which

monitors the ability of earthworm coelomocytes to illicit normal innate immune responses to the

foreign red blood cells. The immune status of coelomocytes were scored dependent upon the

different response types such as phagocytosis of non-self material, and rosette formation of foreign

cells adhering to the cell surface of the coelomocytes (Eyambe et al., 1991). However, as no

differential sensitivity among the different immune reactions was ever observed, immune activity

was expressed as the percentage of the total cells exhibiting any type of activity.

To ensure that the process of coelomocyte extraction and all subsequent handling did not cause

damage to the coelomocytes, a test for cell viability was undertaken using Eosin Y solution (2

mg/ml in distilled water). A 20 µl sub-sample of coelomic fluid was placed on a microscope slide

and 1 µl of Eosin Y solution was added. Healthy living cells stained green and dead cells stained

red when viewed under the microscope at x 400 magnification.

3. Results

3.1. Association of Titanium and TiO2 ENPs with sewage sludge

10

When mixed together for 30 min it could be observed by eye that the activated sludge

cleared the water column of the opaque Optisol suspension, whereas without activated sludge the

suspension remained opaque. TEM revealed the TiO2 ENPs to be intimately associated with the

sludge matrix (Fig. 1) and apparently absent from the water phase. Identification was confirmed

using the integral X-ray spectrometer.

Fig. 1: TiO2 ENP associated with settled sludge as identified by X-ray spectrometer.

11

Table 2

Ti content in solids taken from a UK activated sludge plant (ASP) and associated calculations,

square brackets are references for the calculations

sample

measured Ti [µg/g

DW] (ave ±SEM) [1]

suspended solids

(MLSS) g/L [2]

Ti on solids as µg/L:

[1] x [2]

Raw sewage influent

01/08 (n=2) 135 ± 3 0.26 35

Raw sewage influent

02/09 (n=4) 196 ± 15 0.33 65

average influent 165 [3] 0.29 50 [4]

Activated sludge (AS)

10/06 (n=1) 541 2.3 1244

AS 01/08 (n=2) 422 ± 43 3.4 1435

AS 02/09 (n=2) 650 ± 26 2.4 1530

average activated

sludge 538 [5] 2.7 1400

effluent 02/09 (n=4) 568 ± 23 0.010 5.7

Sewage flow rate at this ASP 60,000 m3/d [6]

Calculated Ti input into this ASP: [4] x [6] 3 kg/d

Predicted Ti in wasted sludge assuming ratio of 5:3 primary to secondary

sludge: (5 x [3] + 3 x [5]) / 8 305 µg/g DW [7]

Sludge DW production in UK in 2004 (Defra, 2009) 1,136,800 t/year [8]

Projected Ti in total UK (2004) sludge: [7] x [8] 347 t/year

Despite being collected at different times, with slightly different methods, whether collected from

the influent, or from within the activated sludge tank, Ti concentrations appeared to be relatively

similar and consistent (Table 2). The average Ti content in activated sludge of 538 µg/g was slightly

less than the 900 µg/g found in a US study (Kiser et al., 2009). The activated sludge mixed liquor

and solids had a higher Ti content than the influent primary sludge (Table 2). This is likely to be a

12

feature of sludge re-circulation as the majority of the secondary settled sludge is returned several

times to the aeration tank before being wasted. This process would concentrate TiO2 on the

returning floc particles. The previous laboratory study indicated a strong affinity for activated

sludge flocs by the TiO2 ENPs could be expected. Based on a typical primary to secondary sludge

mix that would make up waste sludge, the product of this STP might be expected to contain about

305 mg/kg dry weight (DW) Ti. In a recent study it was predicted that 100-433 mg/kg TiO2 ENPs

would be present in European sewage sludge (Gottschalk et al., 2009). In this case, if extrapolated

across the UK, this would imply a discharge of about 350 t/yr Ti based on the sludge production in

2004 (Defra, 2009). Given the high commercial use of TiO2 in products likely to reach the sewer, it

is probable that the majority of this Ti found in the sludge originated from TiO2. This hypothesis is

supported by a non-quantitative SEM analysis of Ti particles found in biosolids and effluent where

a high proportion appeared to be TiOx species as determined by energy dispersive x-ray analysis

(Kiser et al., 2009)

3.2. Removal of colloidal Ti passing through an activated sludge plant

The Ti<0.45 µm fraction entering the activated sludge plant in the raw sewage between 10:00

and 14:00 had a mean concentration of 30 µg/L (values ranged between 17 and 40 µg/L, Table 3)

which is similar to the 35 µg/L filtered Ti reported entering an ASP in the USA (Kiser et al., 2009).

There was only a very slight (<10%) decline in concentration following primary settlement. This

corresponds with laboratory simulations of primary wastewater treatment, which showed that

unfunctionalized oxide nanoparticles did not undergo removal by flocculation and sedimentation

during the primary settlement stage (Jarvie et al., 2009). The greatest decline followed the

biological stage of activated sludge giving a final mean concentration of 3.2 µg/L (Table 3). Thus,

a removal of 89.5% of the Ti<0.45 µm fraction occurred in this STP, very similar to that predicted

by Mueller and Nowack (2008) but more than the Ti<0.7 µm filtered fraction which fell from 30

µg/L to 10-20 µg/L in a US study (Kiser et al., 2009). From experiments both here and in the

literature the mean agglomeration size of TiO2 ENPs such as P25 in aqueous matrices at near

neutral pH is in the region of 300-400 nm (Adams et al., 2006; Battin et al., 2009; Boncagni et al.,

2009). So it seems reasonable to presume that the <0.45 µm fraction Ti captured in these studies

would have been largely TiO2, and would have included TiO2 ENPs. Given the population served

by this plant, and its flow, these values would suggest that 8 mg/d/capita of Ti<0.45 µm is

discharged by the local inhabitants in July which would translate to 1 mg/d/capita of Ti<0.45 µm

present in the effluent following sewage treatment.

Table 3

13

Ti<0.45 µm present in the liquor within an ASP sampled in July 2008 (mean of 5 samples)

Raw sewage influent Following primary

settlement

Final effluent

Mean (µg/L) 30.5 26.7 3.2

Std Dev. 11.8 7.5 0.4

3.3. Predicting UK surface water concentrations of sunscreen TiO2 ENPs

The worldwide production of TiO2 ENPs has been suggested to be 5,000 t/yr (Mueller and

Nowack, 2008) and its production and consumption in Europe 3,400 t/yr (Gottschalk et al., 2009).

Of this commercial information suggests the world market for sunscreen TiO2 ENP is about 1500

t/yr and the major sunscreen TiO2 ENP products sold each year in the UK represent 130 t TiO2 ENP

(data provided by Oxonica Ltd). If we assumed 1/3 of sunscreens purchased in UK are used within

the UK, then ca. 42 t/yr of TiO2 ENP is potentially available to be washed off the body after

application. This may be directly to the sea, river, lake or is removed by subsequent washing either

off the body directly or after towelling and hence could find its way ultimately into the sewage

system.

Thus, potentially about 42 tonnes/yr TiO2 ENP in sunscreen is used within the UK, or with a

61.4 million population (Estimate for mid 2008, Office for National Statistics, August 2009) 0.7

g/yr/capita. If this was spread throughout the year this could be considered as 1.9 mg/d/capita.

However, this would be unlikely to be the case in the British climate! It is more probable that

sunscreen application in the UK would be throughout the 3 summer months of the year yielding 7.5

mg/d/capita during this period and 0 mg/d/capita for the other 9 months of the year. As the UK

wastewater discharged per person is about 160 L/day

(http://www.defra.gov.uk/sustainable/government/progress/regional/summaries/16.htm) the

maximum effluent concentrations could be around 47 µg/L (assuming that all of the applied

nanoparticles are washed off and none are lost on route) during this summer period.

A recent chemicals (estrogens) modelling study identified the rivers in the Anglian, Thames

and Midland regions of England and Wales as being the most exposed with respect to sewage inputs

and available dilution (Williams et al., 2009). This risk is a combination of population density and

available diluting rainfall. In terms of the UK both the Anglian and Thames regions are areas of

relatively high sunlight hours where sunscreen might be expected to be applied within the British

Isles. As the Anglian Region has an overall 3 m3/d/capita dilution (mean flow), and using the per

capita consumption for the sunscreen TiO2 ENP described above for the summer period, with 90%

14

removal in sewage treatment (based on this and other studies, Mueller and Nowack, 2008; Kiser et

al., 2009), a regional PEC of 0.25 µg/L would be predicted. Following the same calculation for the

Thames Region with a 1.5 m3/d/capita available dilution (mean flow) this would generate 0.5 µg/L

TiO2 ENP regional PEC for the same 3 summer month period. An issue which may complicate the

prediction of river water concentrations is the possibility of TiO2 ENPs being transported into water

courses from sewage sludge applications to land. However, as the TiO2 ENPs would be bound to

sludge which itself would be buried within the soil an assessment of the colloidal/particulate

movement into water courses is difficult. Studies on metals in sewage sludge often show limited

leaching, particularly when the pH is neutral (Kidd et al., 2007). Runoff losses from land applied

sludge would not be a major issue in the drier summer months of the year which are the object of

this modelling study.

Rather than scaling down from known annual sales of the product, a different approach to

predicting a regional sunscreen TiO2 ENP water concentration is to calculate instead what might

happen during a weekend heat wave (when most consumers are not at work, or school, and so

would have opportunity for sun exposure). If the recommended sunscreen application is followed

by a person applying sunscreen to half their body area (2 m2), this would lead to 600 mg/d/capita

given a typical content of 3% in TiO2 ENP containing formulations. However, this value must be

reduced as commercial information indicates only 22% of UK sunscreen market is taken by TiO2

ENP containing products. In this UK heatwave scenario it was considered plausible that 1 in 4

people would apply sunscreen and follow the recommended application rate. This being the case,

in the Anglian Region, with 3 m3/d/capita available dilution (mean flow) and taking into account

90% removal in sewage treatment the predicted concentration would be 1.1 µg/L across the region

on that day. Following the same assumptions for Thames Region this would give 2.2 µg/L TiO2

ENP across the regions rivers. But these values are based on mean flow, not low flow. For the

River Thames at the most downstream gauging point at Kingston mean flow is 78.2 m3/sec but Q95

(flow value exceeded 95% of the time) is only 18.8 m3/sec which indicates 4 times less available

dilution, so in these circumstances of heat wave and low flow a PEC of 8.8 µg/L sunscreen TiO2

ENP is predicted across the region (Table 4). For comparison, a prediction for Switzerland for all

TiO2 ENPs likely to be discharged to surface water gave values of 0.7 to 16 µg/L (Mueller and

Nowack, 2008), but more recently 0.01 to 0.06 µg/L were seen as more likely for European surface

waters (Gottschalk et al., 2009). These concentrations were based on a simplistic approach to

dilution in that the volume was a function of the surface area of the region’s water bodies with all

assumed to be 3 m depth.

15

Table 4

Summary of predicted sunscreen concentrations TiO2 ENP in selected UK surface waters with an

effluent value for comparison.

Scenario Anglian

Region for

3 month

summer

period

Anglian

Region for

1 d

heatwave

Thames

Region for

3 month

summer

period

Thames

Region for

1 d

heatwave

Thames

catchment 1

d heatwave

Q95 low

flows

Measured

sewage

effluent values

for Ti<0.45

µm in July

2008*

Presumed

sunscreen

TiO2 ENP

(µg/L)

0.25 1.1 0.5 2.2 8.8 3.2

*This refers to all dissolved or colloidal Ti, not just TiO2 ENP products

3.4. Predicting soil exposure

Values provided by the water industry indicate 62% of sewage sludge in the UK is spread to

land as a soil conditioner (WaterUK, 2006). Based on the sludge measurements these data suggest

there are 305 g/t DW Ti in sewage sludge (Table 2). Good agricultural practice advises limiting

total N applications to 250 kg/ha/yr N, so as sludge is considered to contain a minimum of 3% N by

dry weight (Hogan et al., 2001) up to 8.3 tonnes DW/ha sludge may be applied. As we predict 305

g Ti/t DW sludge this would result in an inadvertent application of 2.5 kg Ti/ha, or 250 mg/m2, of

Ti to soil. Another possibility for soil contamination is atmospheric deposition of TiO2 following

sludge incineration. However, incineration only accounts for around 20% of sludge disposal

(Defra, 2009), thus, 69 t of our predicted total 347 t/yr discharge of Ti in sludge (Table 2) could be

involved. If this were to be redistributed evenly over the land surface of the UK (244,820 km2) this

would deposit only an additional 0.28 mg/m2 to the soil. If we supposed this Ti in sludge was

anthropogenic TiO2 this would now represent 534 g/t DW (534 mg/kg). If TiO2 ENPs were present

in the sludge in proportion to our consumption of total TiO2 this should be 0.1% or 0.5 mg/kg

sludge DW. If this proportion of TiO2 ENP as a proportion of the whole is correct then this would

introduce no more than 0.44 mg/m2 into the soil. The apparently strong affinity of TiO2 ENPs with

sludge floc particles observed in this study (Fig. 1) lends support to the assumption of commercial

TiO2 ENPs being present in this total Ti in the biosolids.

3.5. Impact of TiO2 ENPs and UV light on river water bacteria

The photocatalytic reaction of TiO2 with UV light giving rise to reactive oxygen species is

well known (Linsebigler et al., 1995; Maness et al., 1999). These reactive species may oxidise

16

organic contaminants, and can cause fatal damage to microorganisms by disrupting or damaging

various cell functions or structures (Linsebigler et al., 1995; Jacoby et al., 1998; Maness et al.,

1999). River microbial communities are thus potentially vulnerable to the effects of TiO2 ENP

exposure and so are suitable organisms for investigation.

In the experimental setting it was noted that particle agglomeration in the solutions caused a

significant amount (> 80%) of nanoparticle agglomerates to have a diameter > 100 nm. The P25

TiO2 ENP suspension ranged from 25– 1874 nm, with a mean average size of 368 nm, similar to

that observed by others with these particles (Adams et al., 2006; Battin et al., 2009; Boncagni et al.,

2009). TEM observations showed that the enriched river water bacterial community comprised a

diverse group of bacterial cells and also illustrated the small size of the agglomerated TiO2 ENPs

(100 – 500 nm) in comparison to the much larger bacterial cells (1.5 – 3 µm). The TiO2 ENP

aggregates largely did not adhere to the bacterial cells, but remained detached (Fig. 2).

Fig. 2. Biological TEM A) Enriched river water bacterial cells B) Enriched river water bacterial

cells plus TiO2 ENPs

B A

17

0

20

40

60

80

100

120

140

0 10 20 30 40 50 60

Time (minutes)

Survival ratio (%)

Fig. 3. The survival of river water bacterial cells versus illumination time [Control ( ), UV

control ( ), 1 g/L TiO2 no UV ( ), 0.001 g/L TiO2 plus UV ( ), 0.01 g/L TiO2

plus UV ( ), 0.1 g/L TiO2 plus UV ( ), 1 g/L TiO2 plus UV ( )]. Error bars are

1 x SD (9 replicates).

Exposure to UV, or TiO2 ENPs alone, did not affect the viability of the culturable bacterial cells

(Fig. 3). However, UV illumination with 1 g/L, 0.1 g/L, or 0.01 g/L TiO2 ENP resulted in a 99%,

88%, 27%, kill respectively. In similar experiments with pure cultures, significant kill was

observed at 0.1 g/L (Wei et al., 1994; Maness et al., 1999), but the impact of UV activated TiO2

ENPs at lower concentrations was not investigated. UV illumination with only 0.001 g/L (1,000

µg/L) TiO2 did not affect cell viability. This experiment indicated that 0.01 g/L (10,000 µg/L) of

TiO2 ENP with UV illumination is required for significant kill of an enriched river water bacterial

community with no observable effect occurring at 1,000 µg/L TiO2 ENP.

3.6. Impact of TiO2 ENPs on earthworm coelomocyte (immune) cells

The effects of soil borne TiO2 ENPs was assessed in a worst case in-vitro exposure system

using earthworm coelomocyte cells that are known to be sensitive to chemical exposure. A quality

control assessment of the freshly extracted coelomocyte cells from unexposed worms using Eosin Y

indicated in-vitro cell viability of 90-95% over 24 h. This indicated that the cells could be

18

maintained in-vitro during the experimental period without compromising their viability or

function.

Analysis of the TiO2 ENPs in Ringer’s solution indicated a tendency for them to aggregate.

For Optisol the most frequent sized aggregates were within the 350 to 400 nm range, while both

P25 and T805 tended to form smaller 250 to 300 nm aggregates; T805 also had 32 to 100 nm

agregrates. Since Ringer’s is designed to mimic the ionic strength of coelomic fluid this aggregation

is likely to be relevant to physiological conditions within the earthworm coelome.

In isolated unexposed coelomocytes, 92% (±standard dev. 6%) of cells showed some form

of immune activity towards co-incubated rabbit red blood cells. The immune activity levels for the

dosed coelomocytes was 85% (±26%), 80% (±8%) and 90% (±12%) following exposure to the

concentrations of 7.8 mg/L Optisol, 25 mg/L P25 and 17.5 mg/L T805 used. One way ANOVA

indicated there was no significant difference between the total immune activity of unexposed

coelomocytes with those exposed to any of the TiO2 ENPs. As this experiment was a worst case

exposure for coelomocyte cells, this suggests that TiO2 ENPs are unlikely to harm the earthworm

immune system in the real environment.

4. Discussion

TiO2 is present naturally in the environment and is certainly widespread in UK river bed

sediments (Johnson and Breward, 2004). There appears to be a considerable quantity of Ti

colloids/nanoparticles (whether intentionally engineered or not) entering the sewers and ultimately

reaching STPs. The TiO2 ENPs in sunscreens will be an irregular presence of this larger load. The

TEM observations and STP measurements (here and also (Kiser et al., 2009)) suggests that the

majority of this load will be partitioned off into the sludge and not pass through into final effluent.

Thus, the greatest environmental exposure will be terrestrial following sewage sludge application

(common in the UK). Sludge may also be incinerated, or disposed as landfill, so other entry points

are possible (Mueller and Nowack, 2008), but the greatest direct environmental challenge would

appear to be through the use of sludge as a fertiliser product (Gottschalk et al., 2009). Calculations,

based on the presence of Ti in sludge in this study, suggest that 250 mg/m2 total Ti application to

soil would be possible at the highest recommended application rates. Taking current assumptions

of the market share into account, this suggests no more than 0.5 mg/kg of TiO2 ENP would be

found in sludge introducing 0.4 mg/m2 into soil when used as a fertiliser. For this route of

exposure, earthworms are clearly relevant, but from the direct exposure of their immune system

cells, it appears that this receptor group would be unlikely to be compromised by even the most

extreme exposure to TiO2 ENPs. Further reassurance has come from a recent feeding study with a

19

terrestrial isopod where no deleterious effects were observed on the organism at up to 1000 mg/kg

TiO2 ENPs in dry food (Drobne et al., 2009). A TiO2 ENP PNEC of 1 mg/kg has been suggested

for soil organisms (Gottschalk et al., 2009) but further studies are needed.

Using current assumptions, the highest plausible regional water concentration predicted was 8.8

µg/L TiO2 sunscreen ENP for the Thames region during a heat wave. Given the value of 3.2 µg/L

Ti<0.45 µm fraction measured in sewage effluent in this study, 3.5 µg/L TiO2 ENP predicted for

European effluent (Gottschalk et al., 2009) and 20 µg/L in a US study (Kiser et al., 2009) finding

such a high concentration in river water itself with its greater dilution would be rather exceptional.

These predicted TiO2 ENP and measured colloidal TiO2 concentrations remain reassuringly below

the lowest reported direct toxic effect of a TiO2 ENP with aquatic organisms such as 100 µg/L with

a fish, (Federici et al., 2007), 980 µg/L for an algae (Aruoja et al., 2009) and 1,000 µg/L NOEC for

river bacteria reported here. Because we cannot know the toxic effect on all forms of aquatic life it

is common to apply a safety factor such as 1000 (Gottschalk et al., 2009) which would give a PNEC

of around 1µg/L. Such a level could be exceeded for a limited period in the high sunscreen use

scenarios predicted here. We still do not know if TiO2 ENPs, which seem likely to accumulate in

sediment (Boncagni et al., 2009), may cause problems for benthic organisms. When a bacterial

biofilm was challenged with 5,300 µg/L TiO2 ENP (P25, or Hombikat-100) and UV light for 2 d,

harmful effects were observed (Battin et al., 2009). Such a concentration is two to three orders of

magnitude greater than has been measured so far in sewage effluent here, or by others (Kiser et al.,

2009). However, during exposure times of a few months it may be possible to imagine that a

biofilm in a shallow un-shaded ditch receiving neat effluent may accumulate an equivalent TiO2

ENP load to that received in this experiment (Battin et al., 2009). However, given the natural

turnover in bacterial cells and the enormously challenging environment of sewage effluent in

general, it seems unlikely that TiO2 ENP would represent the greatest challenge to the survival of

biofilms in such environments. Biofilms in larger rivers would face lower concentrations due to

dilution and in many cases less UV light due to the depth of water, shading, or water colouration.

5. Conclusions

• Around 90% of the <0.45 µm fraction Ti was removed from the water stream in an activated

sludge treatment plant yielding 3 µg/L in the effluent . This is likely to reflect the fate of the

associated TiO2 ENPs

• The highest plausible sunscreen TiO2 ENP concentration predicted for a UK water region

was 8.8 µg/L (River Thames).

• River bacteria were found to be insensitive to the presence of 1,000 µg/L of a TiO2 ENP

solution in the presence of UV light over a 1 h period.

20

• Using measured sludge Ti concentrations from this study, under the highest soil application

scenario of sewage sludge, no more than 250 mg/m2 of total Ti would be introduced into

soil. On the basis of TiO2 ENP sales as a proportion of total TiO2 consumption, this would

suggest no more than 0.4 mg/m2 of TiO2 ENP would be introduced into soil. Because of the

way these compounds partition in sewage treatment, the greatest environmental exposure to

TiO2 ENPs will be in soil, following sludge applications. More ecotoxicological tests on soil

organisms would be helpful.

• Earthworm coelomocytes, which act as the major components of their immune system, are

unaffected by direct exposure to TiO2 ENPs.

• Whilst there is as yet little data available, current evidence suggests TiO2 ENPs discharged

down the drain, particularly from sunscreens, do not currently represent a major threat to

free swimming aquatic organisms in UK rivers. This assessment could of course change if,

for example, our consumption of these products were to rise by at least 100%, or if

organisms 100% more sensitive than those tested so far were found.

Acknowledgements

This work was funded within the Centre for Ecology’s research core budget recieved from

the Natural Environment Research Council.

6. References

Adams LK, Lyon DY, Alvarez PJJ. Comparative eco-toxicity of nanoscale TiO2, SiO2, and ZnO water suspensions. Water Res 2006; 40: 3527-3532.

Anderson PD, D'Aco VJ, Shanahan P, Chapra SC, Buzby ME, Cunningham VL, Duplessie BM, Hayes EP, Mastrocco FJ, Parke NJ, Rader JC, Samuelian JH, Schwab BW. Screening analysis of human pharmaceutical compounds in US surface waters. Environmental Science & Technology 2004; 38: 838-849.

Aruoja V, Dubourguier HC, Kasemets K, Kahru A. Toxicity of nanoparticles of CuO, ZnO and TiO2 to microalgae Pseudokirchneriella subcapitata. Sci Total Environ 2009; 407: 1461-1468.

Battin TJ, Kammer FVD, Weilhartner A, Ottofuelling S, Hofmann T. Nanostructured TiO2: Transport behaviour and effects on aquatic microbial communities under environmental conditions. Environ Sci Technol 2009; 43: 8098-8104.

Bechtel DB, Bulla LA. Electron-Microscope Study of Sporulation and Parasporal Crystal-Formation in Bacillus-Thuringiensis. J Bacteriol 1976; 127: 1472-1481.

Boncagni NT, Otaegui JM, Warner E, Curran T, Ren JH, De Cortalezzi MMF. Exchange of TiO2 Nanoparticles between Streams and Streambeds. Environmental Science & Technology 2009; 43: 7699-7705.

Defra. e-Digest statistics about waste and recycling: Sewage sludge http://www.defra.gov.uk/evidence/statistics/environment/waste/wrsewage.htm#wrtb11, Defra, 2009.

Drobne D, Jemec A, Tkalec ZP. In vivo screening to determine hazards of nanoparticles: Nanosized TiO2. Environmental Pollution 2009; 157: 1157-1164.

Eyambe GS, Goven AJ, Fitzpatrick LC, Venables BJ, Cooper EL. A Noninvasive Technique for Sequential Collection of Earthworm (Lumbricus-Terrestris) Leukocytes During Subchronic Immunotoxicity Studies. Lab Anim 1991; 25: 61-67.

Federici G, Shaw BJ, Handy RD. Toxicity of titanium dioxide nanoparticles to rainbow trout (Oncorhynchus mykiss): Gill injury, oxidative stress, and other physiological effects. Aquatic Toxicology 2007; 84: 415-430.

21

Fugere N, Brousseau P, Krzystyniak K, Coderre D, Fournier M. Heavy metal-specific inhibition of phagocytosis and different in in vitro sensitivity of heterogeneous coelomocytes from Lumbricus terrestris (Oligochaeta). Toxicology 1996; 109: 157-166.

Giggleman MA, Fitzpatrick LC, Goven AJ, Venables BJ. Effects of pentachlorophenol on survival of earthworms (Lumbricus terrestris) and phagocytosis by their immunoactive coelomocytes. Environmental Toxicology and Chemistry 1998; 17: 2391-2394.

Gottschalk F, Sonderer T, Scholz RW, Nowack B. Modeled Environmental Concentrations of Engineered Nanomaterials (TiO2, ZnO, Ag, CNT, Fullerenes) for Different Regions. Environmental Science & Technology 2009; 43: 9216-9222.

Goven AJ, Eyambe GS, Fitzpatrick LC, Venables BJ, Cooper EL. Cellular Biomarkers for Measuring Toxicity of Xenobiotics - Effects of Polychlorinated-Biphenyls on Earthworm Lumbricus-Terrestris Celomocytes. Environmental Toxicology and Chemistry 1993; 12: 863-870.

Greenwood NN, Earnshaw A. Chemistry of the elements. Butterworth-Heinemann, Oxford, 1997, 1600 p. Hogan F, McHugh M, Morton S. Phosphorus availability for beneficial use in biosolids products. Environ

Technol 2001; 22: 1347-1353. Jacoby WA, Maness PC, Wolfrum EJ, Blake DM, Fennell JA. Mineralization of bacterial cell mass on a

photocatalytic surface in air. Environmental Science & Technology 1998; 32: 2650-2653. Jarvie HP, Al-Obaidi H, King SM, Bowes MJ, Lawrence MJ, Drake AF, Green MA, Dobson PJ. Fate of Silica

Nanoparticles in Simulated Primary Wastewater Treatment. Environmental Science & Technology 2009; 43: 8622-8628.

Johnson AC, Ternes T, Williams RJ, Sumpter JP. Assessing the concentrations of polar organic microcontaminants from point sources in the aquatic environment: Measure or model? Environmental Science & Technology 2008; 42: 5390-5399.

Johnson CC, Breward N. G-BASE: Geochemical Baseline Survey of the Environment. Report CR/04/016N. British Geological Survey, Keyworth, 2004, 14 p.

Kaegi R, Ulrich A, Sinnet B, Vonbank R, Wichser A, Zuleeg S, Simmler H, Brunner S, Vonmont H, Burkhardt M, Boller M. Synthetic TiO2 nanoparticle emission from exterior facades into the aquatic environment. Environmental Pollution 2008; 156: 233-239.

Kidd PS, Dominguez-Rodriguez MJ, Diez J, Monterroso C. Bioavailability and plant accumulation of heavy metals and phosphorus in agricultural soils amended by long-term application of sewage sludge. Chemosphere 2007; 66: 1458-1467.

Kiser MA, Westerhoff P, Benn T, Wang Y, Pérez-Rivera J, Hristovski K. Titanium Nanomaterial Removal and Release from Wastewater Treatment Plants. Environmental Science & Technology 2009; 43: 6757-6763.

Lee SW, Kim SM, Choi J. Genotoxicity and ecotoxicity assays using the freshwater crustacean Daphnia magna and the larva of the aquatic midge Chironomus riparius to screen the ecological risks of nanoparticle exposure. Environ Toxicol Pharmacol 2009; 28: 86-91.

Limbach LK, Bereiter R, Müller E, Krebs R, Gälli R, Stark WJ. Removal of Oxide Nanoparticles in a Model Wastewater Treatment Plant: Influence of Agglomeration and Surfactants on Clearing Efficiency. Environ Sci Technol 2008; 42: 5828-5833.

Linsebigler AL, Lu GQ, Yates JT. Photocatalysis on Tio2 Surfaces - Principles, Mechanisms, and Selected Results. Chemical Reviews 1995; 95: 735-758.

Lomer MCE, Thompson RPH, Commisso J, Keen CL, Powell JJ. Determination of titanium dioxide in foods using inductively coupled plasma optical emission spectrometry. Analyst 2000; 125: 2339-2343.

Maness PC, Smolinski S, Blake DM, Huang Z, Wolfrum EJ, Jacoby WA. Bactericidal activity of photocatalytic TiO2 reaction: Toward an understanding of its killing mechanism. Applied and Environmental Microbiology 1999; 65: 4094-4098.

Mavon A, Miquel C, Lejeune O, Payre B, Moretto P. In vitro percutaneous absorption and in vivo stratum corneum distribution of an organic and a mineral sunscreen. Skin Pharmacol Physiol 2007; 20: 10-20.

Mueller NC, Nowack B. Exposure modeling of engineered nanoparticles in the environment. Environ Sci Technol 2008;

Nohynek GJ, Dufour EK, Roberts MS. Nanotechnology, cosmetics and the skin: Is there a health risk? Skin Pharmacol Physiol 2008; 21: 136-149.

Office for National Statistics. Population Estimates. In. Statistical Bulletin. August 2009. Ort C, Hollender J, Schaerer M, Siegrist H. Model-Based Evaluation of Reduction Strategies for

Micropollutants from Wastewater Treatment Plants in Complex River Networks. Environmental Science & Technology 2009; 43: 3214-3220.

Ramsden CS, Smith TJ, Shaw BJ, Handy RD. Dietary exposure to titanium dioxide nanoparticles in rainbow trout, (Oncorhynchus mykiss): no effect on growth, but subtle biochemical disturbances in the brain. Ecotoxicology 2009; 18: 939-951.

Taylor SR, McLennan SM. The continental crust: its composition and evolution. Geoscience texts. , Blackwell Scientific Publications., Oxford, 1985, 312 p.

22

WaterUK. Recycling of biosolids to land. Water UK, London. 2006, 28. Wei C, Lin WY, Zainal Z, Williams NE, Zhu K, Kruzic AP, Smith RL, Rajeshwar K. Bactericidal activity of TiO2

photocatalyst in aqueous-media - toward a solar-assisted water disinfection system. Environmental Science & Technology 1994; 28: 934-938.

Wigginton NS, Haus KL, Hochella MF. Aquatic environmental nanoparticles. J Environ Monit 2007; 9: 1306-1316.

Williams RJ, Keller VDJ, Johnson AC, Young AR, Holmes MGR, Wells C, Gross-Sorokin M, Benstead R. A national risk assessment for intersex in fish arising from steroid estrogens. Environmental Toxicology and Chemistry 2009; 28: 220-230.

Young AR, Grew R, Holmes MGR. Low flows 2000: a national water resources assessment and decision support tool. Water Sci Technol 2003; 48: 119-126.