Fluorescent Whitening Agents Natural Waters

Diss. ETH No. 12355

Fluorescent Whitening Agents • ID Natural Waters

A dissertation submitted to the SWISS FEDERAL INSTITUTE OF TECHNOLOGY ZURICH for the degree of Doctor of Natural Sciences

presented by JEAN-MARC ALAIN STOLL Dipl. Chem., University of Zurich born on November 9, 1966 citizen of Schaftbausen and Osterfingen (SH)

accepted on the recommendation of Prof. Dr. Walter Giger, examiner Dr. Robert P. Eganhouse, co-examiner Dr. Markus M. Ulrich, co-examiner

Zurich 1997 The following chapters have been published or submitted for publication:

Chapter 2 Stoll J. M. A. and Giger W. (1997). Determination of Detergent-Derived Fluorescent Whitening Agent Isomers in Lake Sediments and Surface Waters by Liquid Chromatogra- phy. Anal. Chem., 69, 2594-2599.

Chapter 3 Stoll J. M. A. and Giger W. Mass Balance for Detergent- Derived Fluorescent Whitening Agents in Surface Waters of Switzerland. Water Res. (in press).

Chapter 4 Stoll J. M. A., Ulrich, M. M., and Giger W. Dynamic Behavior of Fluorescent Whitening Agents in Greifensee: Field Measurements Combined with Mathematical Modeling of Sedimentation and Photolysis. Submitted to Environ. Sci. Technol.

Chapter 5 Stoll J. M. A., Poiger T. F ., Lotter A. F ., Sturm M., and Giger W. (1997). Fluorescent Whitening Agents as Molecular Markers for Domestic Waste Water in Recent Se- diments of Greifensee, Switzerland. In: Molecular Markers in Environmental Geochemistry (Eganhouse R. P. ed.), ACS Symposium Series 671, American Chemical Society: Washington, DC. pp. 231-241. I Table of Contents

Abstract ...... IV Z usammenfassung ...... VI

1 ll'J'fliOI>UCTIOl'l...... 1

1.1 Clean Clothing ...... 1 1.2 White Clothing ...... 3 1.3 Environmental Aspects of Laundry Detergents ...... 6 1.4 Environmental Aspects of Sewage Sludge ...... 8 1.5 Scope of this Work ...... 9 1.6 Outlook ..... "'························-··················································IO 1.7 Literature Cited ...... 11

2 ANALYTICAL METHODS ...... 15

2.1 Introduction ...... 17 2.2 Experimental Section ...... 20 2.2.1 Reagents ...... 20 2.2.2 Standard Solutions ...... 20 2.2.3 Samples ...... 21 2.2.4 Enrichment from Aqueous Samples ...... 21 2.2.5 Liquid Extraction of Solid Samples ...... 22 2.2.6 High-Performance Liquid Chromatography ...... 22 2.2. 7 Quality Assurance ...... 23 2.3 Results and Discussion ...... 25 2.3.1 Separation ...... 25 2.3.2 Detection and Quantitation ...... 26 2.3 .3 Applications ...... 30 2.4 Conclusions ...... 32 2.5 Literature Cited ...... 33 II

3 RIVER MONITORING ...... 35

3.1 Introduction ...... 37 3.2 Sampling Locations and Analytical Methods ...... 39 3 .3 Results and Discussion ...... 42 3.3.1 FWA Concentrations and Loads in Swiss Rivers ...... 42 3.3.2 FWA Mass Balance for Switzerland ...... 48 3.3.3 FWAs as Molecular Markers for Production Wastewater ..... 50 3.3.4 Elimination of FWAs in Lakes ...... 52 3.4 Conclusions ...... 53 3.5 Literature Cited ...... 54

4 LAKE MOI>EL ...... 57

4.1 Introduction ...... 59 4.2 Experimental Section ...... 61 4.2.1 Greifensee, the Study Site ...... 61 4.2.2 Sampling and Chemical Analyses ...... 63 4.2.3 Chemicals ...... 64 4.2.4 Mathematical Simulation Model...... 64 4.3 Results and Discussion ...... 65 4.3.1 Evaluation of FWA Inputs from STP Effluents ...... 65 4.3.2 Evaluation of Sorption/Sedimentation ...... 67 4.3.3 Evaluation of Photocheinical Degradation ...... 68 4.3. 4 Seasonal V aria ti on of Measured FWAs in Greifensee ...... 69 4.3.5 Computer Simulation, Model Validation ...... 73 4.3.6 Computer Simulation, Transport by Sedimenting Particles ... 76 4.3.7 Computer Simulation, Photodegradation in Greifensee ...... 77 4.3.8 Limits of the Model ...... 78 4.4 Literature Cited ...... 78

S LAKE SEDIMENT ...... 81

5.1 Introduction ...... 83 5.2 Study Site ...... 86 5 .3 Experimental ...... 88 5.3.1 Samples ...... 88 III

5 .3 .2 Reagents ...... 88 5.3.3 Liquid Extraction ...... 89 5 .3 .4 High-Performance Liquid Chromatography ...... 89 5 .4 Results and Discussion ...... 92 5. 4.1 FWAs as Molecular Markers for Domestic Waste Water ...... 92 5 .4.2 Sedimentary Archive and Emission History of FWAs ...... 94 5.5 Conclusions ...... 95 5 .6 Literature Cited ...... 96

Appendix ...... 99 IV Abstract

Detergents for laundry washing are among the classic chemicals of ci- vilization polluting surface waters. Their relevance for the aquatic environment is on the one hand due to their application with water and subsequent discharge to sewers. On the other hand the amounts used are large: A total of 70,000 tons, consisting of about 20 detergent components, were consumed in Switzerland in the year 1995. In order to evaluate the risk of laundry detergents, it is important to know the fate and behavior of every individual component in the aquatic environment. One component was chosen for this study whose behavior in the aquatic environment is known only to a limited extent: the fluorescent whitening agents (FWAs). The three most important detergent-FWAs were investigated: DAS 1 (a diw:nino£tilbene), DSBP (a di£tyrylhinhenyl) and BLS (a bleach~table compound). An analytical method for the FW As was developed as a prere- quisite for the investigation of the aquatic environment. This method con- sists of the extraction of FW As from both solid and aqueous samples with subsequent separation with reversed-phase HPLC. Limits of quantitation were 1 - 11 µg kg·1 and 0.2 - 3 ng L- 1 for dry sediment and aqueous samples, respectively. Recoveries ranged from 87 to 100 % with an analytical precision of 1 to 12 % (relative standard deviation). The method was then used to investigate the behavior and fate of detergent-derived FWAs in natural waters. The investigation was divided into three parts: 1. Occurrence and substance fluxes, 2. Transformation processes, 3. Enrich- ment in the benthos of a lake. The occurrence and the substance fluxes of the FWAs were investigated in the first part of this study. For this purpose, a monitoring program was conducted in Swiss rivers during one year. Concentrations of both DAS 1 and DSBP measured in the rivers were mostly between 10 1 1 and 120 ng L- , with maximum values of up to 1 µg L- • On the basis of these measurements, a mass balance was developed, indicating that 13 % of the FW As being used are discharged to surface waters. With an annual 1 consumption of 60 t in 1995 this corresponds to 8 t y· • The transformation processes of the FW As in Greifensee, a small lake in Switzerland, were investigated in the second part of this study. By v means of literature data and measurements in Greifensee, the following rates (for DAS 1 and DSBP, respectively) were estimated for the relevant processes: 0.0044 and 0.016 d- 1 (photodegradation), 0.0025 and 0.0020 d- 1 (sorption/sedimentation) and 0.0021 and 0.0018 d- 1 (flushing). On this basis, measured seasonal variations of FWA concentration depth profiles were modeled with a computer simulation. The successful modeling is a confirmation of the estimated processes and shows that photochemical degradation is the most important process for FWAs in Greifensee. However, this process is so slow that FWAs have to be considered as not readily degradable in Greifensee. The enrichment of FW As in the benthos of Greifensee was investigated in the third part of this study. Depth profiles from several locations of the lake benthos indicate that FW As are resistant to alteration after deposition. The concentrations measured in every particular layer can, therefore, be attributed to the inputs occurring at the time of sediment deposition. Thus, FWAs have a potential as molecular markers for domestic wastewater. By choosing this tripartite research concept, the fate and behavior of the FWAs in natural waters could be investigated extensively. The results of this study shall be the basis for further investigations on pollutants in natural waters. VI Zusammenfassung

Waschmittel gehoren zu den klassischen Zivilisationschemikalien, die unsere Gewasser verschmutzen. Ihre Umweltrelevanz basiert einerseits auf ihrem Gebrauch mit Wasser und dem daraus folgenden nahezu voll- standigen Eintrag ins Abwasser. Andererseits werden Waschmittel in grossen Mengen verwendet. In der Schweiz betrug der V erbrauch 1995 70'000 Tonnen, aufgeteilt auf etwa 20 Bestandteile. Um das Risiko ab- schatzen zu konnen, das von Waschmitteln ausgeht, ist es wichtig, das Verhalten von allen einzelnen Bestandteilen in Oberflachengewassem zu kennen. Ein solcher Bestandteil wurde fiir die vorliegende Arbeit ausge- wahlt, von dessen Verhalten in der Umwelt noch wenig bekannt ist: die optischen Aufheller. Die drei wichtigsten optischen Aufheller fiir Wasch- mittel wurden untersucht: DAS I (ein Di.Q,mino§tilben), DSBP (ein Di§tyrylhinhenyl) und BLS (ein Bleichmittel-§.tabiler Aufheller). Als Voraussetzung fiir die Gewasser-Untersuchung wurde eine analy- tische Methode fiir optische Aufheller entwickelt. Dabei wurden die optischen Aufheller aus Seesedimenten oder aus Oberflachenwasser extra- hiert und anschliessend mit Umkehrphasen-HPLC analysiert. Die Bestim- mungsgrenzen waren 1 - 11 µg kg- 1 fiir trockenes Sediment und 0.2 - 3 ng L- 1 fur wassrige Proben. Die Wiederfindungsraten lagen zwischen 87 und 100 % bei einer Prazision von 1 - 12 % (relative Standardabweichung). Diese Methode wurde anschliessend verwendet, um das Schicksal und das Verhalten der optischen Aufheller in Oberflachengewassem zu untersu- chen. Die Untersuchung wurde in drei Teile geteilt mit den folgenden drei Schwerpunkten: 1. Auftreten und Stoff-Fliisse, 2. Umwandlungsprozesse, 3. Anreicherung im Seegrund. In der ersten Teilstudie wurden das Auftreten und die Stoff-Fliisse der optischen Aufheller erfasst. Dazu wurde wahrend einem Jahr ein Uberwachungs-Programm in Schweizer Fliissen durchgefiihrt. Die in den Fliissen gemessenen Konzentrationen lagen in der Regel zwischen 10 und 1 1 120 ng L- , mit Extremwerten von bis zu 1 µg L- • Eine darauf basierende Massenbilanz zeigte, dass 13 % der verbrauchten optischen Aufheller in den Fliissen landen. Dies entspricht einer Menge von 8 t y- 1 (bei einem Verbrauch von 60 t im Jahr 1995). VII

In der zweiten Teilstudie wurden die Umwandlungsprozesse der optischen Aufheller im Greifensee untersucht. Aufgrund von Literaturda- ten und Messungen im Greifensee wurden dazu zunachst die relevanten Prozessraten (fiir DAS 1 bzw. DSBP) abgeschatzt: 0.0044 bzw. 0.016 d·1 (Photoabbau), 0.0025 bzw. 0.0020 d- 1 (Sorption/Sedimentation) und 0.0021 bzw. 0.0018 d·1 (Ausfluss). Auf dieser Basis wurden gemessene Konzentrationsverlaufe im Greifensee mit einem Computerprogramm si- muliert. Die erfolgreiche Computersimulation ist eine Bestatigung der abgeschatzten Prozesse. Sie zeigt,

Introduction

1.1 Clean Clothing

It has long been known that the washing power of water can be increased in various ways. Rainwater, for example, was found to be more suitable for washing than normal well water. Hot water also was found to have more washing power than cold water, and certain additives seemed to improve the water's effectiveness. Soap, the oldest of the surfactants, has been used for over 1000 years as a laundering agent, although for a long time only rich people could afford it. The first commercial detergent to appear on the German market in 1878, Henkel's "Bleichsoda", was based on a mixture of soda (Na2C03) and sodium silicate (Na20 · 2 Si02). Its water-softening effect resulted from the precipitation of calcium and magnesium ions, and it simultaneously eliminated iron salts, which had a tendency to turn laundry yellow. With the beginning of the 20th century and the introduction of the first self-acting detergent, Persil ( 1907), soap definitively took its place as one ingredient in multicomponent systems for the routine washing of textiles. In these, soap was combined with so-called builders, usually soda, sodium silicate and sodium perborate. The en- hanced washing power of these new agents substantially reduced the work entailed in doing laundry by hand. The next important development was the transition brought about by technology from the highly labor-intensive manual way of doing laundry to machine washing. This change in tum led to a need for appropriate changes in the formulation of washing agents. Soap, notorious for its sen- 2 Chapter 1 sitivity to water hardness, was gradually replaced by synthetic surfactants with their more favorable characteristics. By the middle of this century, products of the petrochemical industry (e.g. tetrapropylenebenzenesulfo- nate (TPS), linear alkylbenzenesulfonates (LAS)) had largely forced soap off the cleansing agent market in the industrialized nations. Today, detergents for laundry washing are very complex formulations containing several different types of substances. These can be categorized into the following major groups: surfactants, builders, bleaching agents and additives. Surfactants constitute the most important group of detergent components, and they are present in all types of detergents. Generally, these are water-soluble surface-active agents comprised of a hydrophobic portion (usually a long alkyl chain) attached to hydrophilic or solubility-enhan- cing functional groups. The most common agents in laundry detergents are anionic, e.g. linear alkylbenzenesulfonates (LAS) or secondary alkyl- sulfonates (SAS), although non-ionic surfactants of the ethylene oxide adduct variety have also acquired great importance (e.g. alcoholethoxy- lates). Cationic surfactant use is largely restricted to aftertreatment aids because of the fundamental incompatibility of these materials with anionic surfactants. Builders support detergent action in the course of the washing process and eliminate calcium and magnesium ions. The category of builders is comprised of several types of materials: specific alkaline substances such as sodium carbonate and sodium silicate; complexing agents like sodium phosphates or nitrilotriacetic acid (NTA); and ion exchangers, such as water-soluble polycarboxylic acids and insoluble zeolites. The most important bleaching agent in Europe is sodium perborate. Due to the alkaline medium during the washing process it is converted to the hydrogen peroxide anion, which is the active intermediate. By oxida- tive decomposition of chromophoric systems adhering to fibers, an increase in the reflectance of visible light is achieved. Additives are introduced only in small amounts to detergents, each to accomplish its own specific purpose. Examples are enzymes (capable of eliminating protein stains derived from sources such as milk, blood, egg and grass), soil antiredeposition agents (e.g. carboxymethyl cellulose (CMC)), foam regulators (e.g. fatty acid amides) and fluorescent whitening agents (FW As, compensating for the yellowish shade of white fibers) (Jakobi et al. 1987). General Introduction 3

1.2 White Clothing

White is the term used to describe the brightest achromatic color. In physical terms the ideal white would be represented by a substance, such as barium sulfate or magnesium oxide, which has virtually complete ( 100 % ) reflectance. From experience it is known that of two whites with equal luminosity or brightness, the one with a yellowish cast will appear less white than the other with a blue-violet cast. The reasons for this would seem to be purely subjective: blue and white are "cold" colors which in- tensify the whiteness, whereas yellow is a "warm" color and thus antago- nistic to the white (Liischer 1975). Man's earliest endeavors to enhance the brightness of white clothing led to the invention of various bleaching agents like potash, urine, sulfur dioxide, sodium hypochlorite and perborate. However, they have their limitations: if used too liberally they can cause some damage to the cloth. Moreover, the natural color of e.g. cotton cannot be whitened in this way. Understandably, the search was on for simpler, less harsh methods of ob- taining a good white. One advance came when it was discovered that the natural substance esculin (Figure 1.1) could be applied to textiles to give whites of unprecedented brilliance. Reflecting more visible light than ab- sorbing, esculin was the first fluorescent whitening agent (FW A) to be used commercially. Further research led to other compounds of the same

Figure 1.1 Structure of Esculin, the first fluorescent whitening agent. Extracted from leaves and bark of horse chestnut tree.

class, which were more suitable for industrial use and had better fastness properties (slower photodegradation). Since that time thousands of FW As have been described, but only a few are used worldwide (Siegrist et al. 1991 ). Two of the FWAs subject to the present investigation, DAS 1 (a 4 Chapter 1 diaminostilbene) and DSBP (a distyrylbiphenyl), are among the whiteners most commonly used in household detergents (Figure 1.2). Their worldwide production was estimated at 14,000 and 3,000 t, respectively, for the year 1989 (Kramer 1992). In Switzerland, a total of 59 t of DAS 1 and DSBP were consumed in 1995 (Gehri 1995, SWI 1995, Eugster 1997), corresponding to an average of 0.1 % of the total mass of detergents consumed. The third FW A investigated here, BLS (bleachstable ), was used in large-scale laundry facilities (e.g. in hospitals) until quite recently (Poiger 1994). BLS is not produced anymore.

DAS1

DSBP

BLS

Figure 1.2 Structures of the FWAs included in this study. Full names: DAS 1: 4,4'-bis[(4-anilino-6-morpholino-l ,3,5-triazin-2-yl)amino]stilbene-2,2'-disulfonate; DSBP: 4,4'-bis(2-sulfostyryl)biphenyl; BLS: 4,4'- bis(4-chloro-3-sulfostyryl )biphenyl. General Introduction 5

The three FWAs DAS 1, DSBP and BLS are based on one or two stilbene moieties. They all absorb UV-light in the wavelength range of 300 - 400 nm and emit visible blue light by fluorescence in the range of 400 - 500 nm (Kramer 1996). This means that not only is the yellowish shade compensated for but also the reflectance of visible light increases. FWAs act therefore as an additional visible light source, giving the colors more brightness. However, FW As cannot be used as replacements for cleaning or bleaching agents, since they are more effective the cleaner and whiter the fabric is. Further important properties of FWAs are their high affinity for cellulose, water solubility and stability with regard to other detergent ingredients. Stilbene-based FWAs are light-sensitive chemicals, especially when in dilute solution. Irradiation of FWAs with sunlight causes reversible E-Z isomerization of the stilbene moiety, as illustrated in Figure 1.3. Hence, DAS 1 containing one stilbene moiety occurs in two isomeric forms, herein called (E)-DAS 1 and (Z)-DAS 1. For FWAs containing two stilbene moieties (DSBP and BLS), three isomeric forms are possible, (E,E)-, (E,Z)- and (Z,Z)-FW A. FWAs are produced and added to laundry detergents in their fluorescent E- or E,E-isomeric forms. Photoisomerization to the corresponding Z-, E,Z-, or Z,Z-isomers leads to a complete loss of fluorescence (Gold 1975).

H H R hv

R hv R R

(EJ- FWA (ZJ- FWA

Figure 1.3 Reversible isomerization process of stilbene-type FWAs.

During the textile washing process, FWAs partly adsorb onto fabrics and, thus, maintain the whiteness of the prewhitened material. The FW As remaining in the washing liquor are discharged to sewers for treatment in municipal wastewater treatment plants. FWAs are partly retained on activated sewage sludge due to adsorption. The nonretained fraction of FWAs 6 Chapter 1 reaches surface waters (Ganz et al. 1975, Poiger 1994, Poiger et al. 1997), where photodegradation takes place with half-lives of several hours at the surface under summer noon sunlight (Kramer et al. 1996). Below the photic zone, FWAs can be assumed to be persistent because biodegradation has not been observed over a period of 30 days (Dojlido 1979, Kaschig 1996). Nevertheless, the concentrations found in sediments and surface waters are far below those expected to be an ecotoxicological risk (Zinkemagel 1975, Burg et al. 1977, Richner et al. 1997).

1.3 Environmental Aspects of Laundry Detergents

Until the middle of this century, general concern about the environmental behavior of household chemicals was small. Detergents had to provide clean clothing and were then discharged to rivers and lakes with no further treatment. Only around 1960 people began to realize that detergents can have an adverse effect on natural waters. As a result of the continuously growing consumption of synthetic surfactants, great masses of foam began to build up in the vicinity of dams and other obstructions. They were caused by insufficient biodegradation of TPS. Another observed impact on surface waters was the eutrophication of lakes and rivers caused by phosphates. The consequences of these two problems were the replacement of TPS by LAS (a surfactant class that is more readily biodegradable) and the construction or improvement of sewage treatment plants. The development of analytical methods for the determination of anthropogenic chemicals brought a general change in the concept of treating environmental pollution (Giger et al. 1991). While the two problems mentioned above had been observed in surface waters, harmful chemicals could now be detected before problems arose. Monitoring studies have thus been performed for major detergent compounds, such as LAS, alcohol ethoxylates (AE), alcohol ethoxy sulphates (AES), alcohol sulphates (AS) and nitrilotriacetate (NTA) as well as for nonylphenol (NP), a degradation product of a nonionic surfactant (nonylphenol polyethoxylate) (Ahel 1987, Alder et al. 1990, Sanches Leal et al. 1994, Feijtel et al. 1995, Holt et al. 1995, Schober! 1995, Houriet 1996, Matthijs et al. 1996). Also minor detergent ingredients (additives) have been monitored, e.g. FWAs in a sewage treatment plant (Poiger et al. 1997). General Introduction 7

However, most of the detergent compounds are applied as mixtures of different isomers and homologues, rendering the determination of individual molecules demanding and very costly. Other compounds are not derived only from detergents, but also originate from other sources (e.g. phosphates, EDTA), so that environmental occurrences cannot be attributed to specific sources. For a third group of detergent chemicals, analytical methods do not exist at all (polycarbonates) or detection limits are too high for environmental samples (phosphonates (Nowack 1997)). Thus, monitoring programs in surface waters have been limited to a small number of individual substances. In the last decade, the development of computers has provided a new possibility to investigate pollutants in surface waters. Instead of just monitoring chemicals in a given compartment, processes determined in a laboratory can now be taken into account in the evaluation in order to antici- pate possible problems and to avoid them (Giger 1995). By taking into account physico-chemical laws coupled with compound-specific data, it is possible to determine, for a given substance in a given environmental system, which of the many transformation and transport processes are expected to occur and how fast. With the help of such chemodynamic concepts, it is also possible to estimate unknown physico-chemical parameters, reac- tivities and the environmental behavior of new substances. These estimates are based on related substances with known properties. For this type of analysis, computers are needed to estimate substance parameters, to ascer- tain relevant processes and, in particular, to simulate the dynamic behavior of a substance in the environment. Such investigations are, for detergent ingredients as well as for anthropogenic pollutants in general, still rather scarce. Two illustrative examples of recent studies with this approach evaluated the dynamic behavior of tetrachloroethene (PER), atrazine and nitrilotriacetate (NTA) in Greifensee, a small lake in Swit- zerland (Ulrich et al. 1994, Muller et al. 1997). The main processes affecting the fate of these chemicals besides flushing are gas-exchange (PER) and biodegradation (NT A). Atrazine showed, except for a short time period during the summer, conservative behavior in the lake. Another process that might influence the fate of chemicals in natural waters, photodegradation, was studied recently with a similar approach in a river (Kari et al. 1995). In this thesis, photodegradation has been investigated in a lake together with sorption/sedimentation using FWAs as model substances (Chapter 4). 8 Chapter 1

FW As are degraded photochemically (Kramer et al. 1996) and they sorb onto particles. Chapter 5 reports on FWAs found in the benthos of Grei- fensee, which demonstrates that sorption/sedimentation is relevant in natural systems. In addition, a monitoring program was conducted in Swiss rivers (Chapter 3) in order to gain a complete picture of the behavior and fate of FWAs in surface waters. By this approach, we wanted to extend the knowledge of the behavior of detergent chemicals in general in surface waters. FWAs were chosen as model compounds, because they are rather reactive (photolysis, sorption) and, in contrast to most detergent ingredients, their consumption is known relatively precisely and individual molecules can be determined reliably (Chapter 2). Since the behavior of structurally similar compounds can often be related with reasonable accuracy, the data obtained for FW As should be useful for modeling a variety of other organic pollutants in lakes.

1.4 Environmental Aspects of Sewage Sludge

It is known that biodegradation of aromatic surfactants often leads to the formation of more persistent metabolites. High concentrations of 4- nony lphenol for instance were detected in digested sewage sludges (Giger et al. 1984). Moreover, very high concentrations of unaltered LASs were found in anaerobically digested sludges (McEnvoy et al. 1986), even though these compounds are considered to be readily degradable. The fate of sewage sludge is, therefore, important in regard to the environmental impact of sorbing chemicals. Scenarios to get rid of the contaminated sludge include discharge to farmland, incineration and dumping from barges to deep basins of the ocean. As reported by R. Eganhouse (Egan- house et al. 1988, Eganhouse 1997), 'the viability of the latter depends on the capacity of ocean waters to dilute, disperse and, ultimately, accommo- date waste-associated contaminants so that accumulation of these materials in the ecosystem above "unacceptable" levels does not occur. Ideally, one would like to monitor the short-term effects of dumping by directly measuring the concentrations of waste-derived contaminants in the water column. However, many toxic organic substances found in sewage sludge, such as polychlorinated biphenyls or polynuclear aromatic hydrocarbons, are not waste-specific. In fact, they may derive from a variety of sources and can be transported long distances in the atmosphere and ocean. Detec- General Introduction 9 tion of these compounds following sewage disposal can, thus, present a problem when "ambient concentrations" are relatively high. Hence waste- specific marker compounds are useful in assessing the impact of ocean- discharged sludge.' Detergent-derived FW As have the potential to serve as such markers. They are used only in laundry washing and, hence, are specific markers of wastewater. During sewage treatment they partly sorb onto sewage sludge, where no evidence for biodegradation was found (Poiger et al. 1997). Furthermore, they can be assumed to be persistent in the benthos of natural waters. One part of this thesis (Chapter 5) was therefore performed in the benthos of a lake in order to evaluate whether the use of FW As as molecular markers is feasible.

1.5 Scope of this Work

The scope of this study was to extend the knowledge of the behavior and fate of detergent-derived chemicals, especially FWAs. FWAs were chosen for this study, because their consumption is known relatively precisely and the detection of individual molecules is facile and highly sensitive. Besides, FW As have the potential to serve as probe molecules for other compounds, as they are affected by only two processes in a lake, i.e. photodegradation and sorption onto settling particles. In particular, they are not biodegraded nor transferred to the air. The investigations were divided into four parts:

1) In order to investigate the fate and behavior of FW As in natural waters, compound-specific analytical procedures had to be developed that allow the quantitative determination of FW As in natural waters. For this purpose, existing analytical methods for FW As in wastewater treatment plants were adapted for the lower concentrations occurring in natural waters. The methods and their validation are described in Chapter 2.

2) A monitoring program was conducted in 10 selected Swiss rivers during one year. Environmental occurrences of FW As were measured and compared with differences in the catchment areas, such as the presence of lakes, population densities and the presence of FW A manu- 10 Chapter 1

facturing plants. The scope of this study was to assess environmental concentrations of FWAs. The results are described in Chapter 3.

3) The dynamic behavior of DAS 1 and DSBP was evaluated quantitatively for Greifensee, a small lake in Switzerland, using field data and a simulation software (MASAS) for modeling organic pollutants in lakes. Over a period of one year the input, photodegradation and transport by settling particles as well as the seasonal variation in the vertical distribution were described by applying a one-dimensional model of 32 horizontally mixed boxes. This investigation should provide information about (i) particle transport in the lake and (ii) the behavior of detergent-derived chemicals in general. In addition, it should show the extent to which the results of laboratory experiments ·can be used to explain the behavior of chemicals in natural waters. In particular, it was interesting to determine whether the rapid photodegradation of FW As determined in laboratory experiments can be extended to the lake situation. These results are described in Chapter 4.

4) DAS 1, DSBP, and BLS were examined in sediment cores of Greifen- see, a small lake in Switzerland. The objectives of this study were to investigate whether FW As could serve as molecular markers for domestic wastewater and whether the sedimentary column could be used as a natural archive for the use of detergent chemicals by humans and for changes in the loading of natural waters with pollutants. This project is part of a larger research program that includes other components of laundry detergents such as LAS, a widely used synthetic surfactant. The results are described in Chapter 5.

1.6 Outlook

Two fields of interest can be derived for FWAs from this study: (i) FWAs have the potential to serve as probe molecules for photochemical degradation in surface waters. Since the behavior of structurally similar compounds can often be related with reasonable accuracy, the data obtained for FW As should be useful for modeling a variety of other organic pollutants in surface waters. (ii) The application of FWAs as molecular General Introduction 11 markers is very promising. Since FWAs are source-specific (domestic wastewater) and persistent in benthic environments, they can be used to estimate the contribution of sludge to ocean sediments and to identify the area affected by sludge dumping. Possible investigations could include deep ocean dumpsites or wastewater treatment plant outfalls to natural bodies of water. Besides the topics covered by this study, there are two compartments where further investigations are needed: (i) The monitoring study on FWAs in Swiss rivers provided only limited data about the processes affecting the fate of FW As. In order to evaluate the dynamic behavior of FWAs in rivers, a study similar to the one in Greifensee could be performed in a small section of a river. Such a study should combine a dense sampling program, spike experiments and mathematical modeling. (ii) In Switzerland, part of the sewage sludge originating from wastewater treatment plants is discharged to farmland. Processes occurring there, e.g. photochemical degradation, might be the subject of further investigations.

1. 7 Literature Cited

Ahel M. (1987). Biogeochemical Behaviour of Alkylphenol Polyethoxyla- tes in the Aquatic Environment. Ph. D. Thesis, University of Zagreb, Croatia. Alder A. C., Siegrist H., Gujer W. and Giger W. (1990). Behaviour of NTA and EDTA in Biological Wastewater Treatment. Wat. Res., 24(6), 733-742. Burg A. W., Rohovsky M. W. and Kensler C. J. (1977). Current status of human safety and environmental aspects of FWAs used in detergents in the United States. Critical Reviews in Environmental Control, 7, 91-120. Dojlido J. R. (1979). Investigations of biodegradability and toxicity of organic compounds. EPA-Report 600/2-79-163. Eganhouse R. P. (1997). Molecular Markers in Environmental Geoche- mistry. ACS Symposium Series 671, American Chemical Society, Wa- shington DC. Eganhouse R. P., Olaguer D. P., Gould B. R. and Phinney C. S. (1988). Use of Molecular Markers for the Detection of Municipal Sewage Sludge at Sea. Marine Environ. Res., 25, 1-22. 12 Chapter 1

Eugster H. (1997). Mifa, Frenkendorf, Switzerland. Personal communication. Feijtel T. C. J., Matthijs E., Rottiers A., Rijs G. B. J., Kiewiet A. and de Nijs A. (1995). AISICESIO Environmental Surfactant Monitoring Programme. Part 1: LAS Monitoring study in "de Meern" sewage treatment plant and receiving river "Leidsche Rijn". Chemosphere, 30/6, 1053-1066. Ganz C.R., Liebert C., Schulze J. and Stensby S. (1975). Removal of detergent FWAs from wastewater. J. Wat. Pollution Control Federation, 47' 2834-2849. Gehri K. ( 1995). Erfassung der schweizerischen Wasch- und Reinigungs- mittelindustrie verwendeten wichtigsten Rohstoffe im Jahre 1995. Verband der Schweizerischen Seifen- und Waschmittelindustrie (SWI), Zurich. Giger W. (1995). Spurenstoffe in der Umwelt. EA WAG News, 40, 3-7. Giger W., Brunner P. H. and Schaffner C. (1984). 4-Nonylphenol in Se- wage Sludge: Accumulation of Toxic Metabolites from Nonionic Sur- factants. Science (Washington, D.C.), 225, 623-625. Giger W., Schaffner C., Kari F. G., Ponusz H., Reichert P. and Wanner 0. (1991). Auftreten und Verhalten von NTA und EDTA in schweizerischen FlUssen. Mitteilungen der EA WAG, 32, 27-31. Gold H. (1975). The Chemistry of Fluorescent Whitening Agents. Major Structural Types. In: Fluorescent Whitening Agents (Anliker R. and Muller G. eds.). Georg Thieme Publishers, Stuttgart, pp. 25-46. Holt M. S., Waters J., Comber M. H. I., Armitage R., Morris G. and Newberry C. (1995). AISICESIO Environmental Surfactant Monito- ring Programme. SD/A Sewage Treatment Pilot Study on LAS. Wat. Res., 29/9, 2063-2070. Houriet J.-P. (1996). NTA dans Les eaux. Cahier de }'environment No 264 (Protection des eaux), OFEFP, Bern, Switzerland. Jakobi G. and Lohr A. (1987). Detergents and Textile Washing. VCH Verlagsgesellschaft mbH, Weinheim. Kari F. G. and Giger W. (1995). Modeling the Photochemical Degrada- tion of Ethylenediaminetetraacetate in the River Glatt. Environ. Sci. Technol., 29(11), 2814-2827. Kaschig J. (1996). Ciba Geigy AG, Basel. Personal communication. General Introduction 13

Kramer J. B. (1992). Fluorescent whitening agents. In: The Handbook of Environmental Chemistry (Hutzinger 0. ed.). Springer, Berlin, pp. 351-366. Kramer J. B. (1996). Photodegradation of FWAs in Sunlit Natural Waters. Ph.D. Thesis, ETH Zurich, No. 11934. Kramer J. B., Canonica S., Hoigne J. and Kaschig J. (1996). Degradation of FWAs in Sunlit Natural Waters. Environ. Sci. Technol., 30(7), 2227-2234. Lilscher M. (1975). Psychological Aspects of White. In: Fluorescent Whitening Agents (Anliker R. and Miiller G. eds.). Georg Thieme Publishers, Stuttgart, pp. 1-11. Matthijs E., Holt M. S., Kiewiet A. and Rijs G. B. J. (1996). Fate of surfactants in activated sludge waste water treatment plants. Belg. Chim. Oggi, 14/5, 9-10. McEnvoy J. and Giger W. (1986). Determination of Linear Alkylben- zenesulfonates in Sewage Sludge by High-Resolution Gas Chromato- graphy/Mass Spectrometry. Environ. Sci. Technol., 20(4), 376-383. Muller S. R., Berg M., Ulrich M. M. and Schwarzenbach R. P. (1997). Atrazine and Its Primary Metabolites in Swiss Lakes: Input Characte- ristics and Long-Term Behavior in the Water Column. Environ. Sci. Technol., 31(7), 2104-2113. Nowack B. (1997). Determination of Phosphonates in Natural Waters by Ion-Pair High Performance Liqud Chromatography. Submitted to J. Chromatogr. A. Poiger T. (1994). Behavior and fate of detergent-derived FWAs in sewage treatment. Ph.D. Thesis, ETH Zurich, No. 10832. Poiger T., Field J. A., Field T. M., Siegrist H. and Giger W. Behavior of Fluorescent Whitening Agents During Sewage Treatment. Water Res. (in press). Richner P., Kaschig J. and Zeller M. (1997). Latest Results from Monito- ring Studies and Environmental Risk Assessments (ERAs) of FWAs. Presentation at the 7th Annual Meeting of SETAC-Europe, Amster- dam. Sanches Leal J., Garcia M. T., Tomas R., Ferrer J. and Bengoechea C. (1994). LAS Removal. Tenside Surf. Det., 31/4, 253-256. Schober! P. (1995). LAS-Monitoring. Tenside Surf. Det., 32/( 25-35. 14 Chapter 1

Siegrist A. E., Eckhardt C., Kaschig J. and Schmidt E. (1991). Optical brighteners. In: Ullmann1s Encyclopedia of Industrial Chemistry. VCH Verlagsgesellschaft, Weinheim, pp. 153-176. SWI (1995). Jahresbericht 1995. Verband der Schweizerischen Seifen- und Waschmittelindustrie, Zurich. Ulrich M. M., Miiller S. R., Singer H.P., Imboden D. M. and Schwarzen- bach R. P. (1994). Input and dynamic behavior of the organic pollutants tetrachloroethene, atrazine and NTA in a lake - a study combi- ning mathematical modeling and field measurements. Environ. Sci. Technol., 28(9), 1674-1685. Zink.emagel R. (1975). FWAs in the environment. In: Fluorescent White- ning Agents (Anliker R. and Muller G. eds.). Georg Thieme Publi- shers, Stuttgart, pp. 129-142. 2

Analytical Methods

A method for the quantitative determination of three different fluorescent whitening agents (FW As) in lake sediments and surface waters is described. Stereoisomers of the two main laundry detergent FW As of the diaminostilbene type (DAS 1) and of the distyrylbiphenyl type (DSBP), as well as total BLS (a compound contained in detergents until a few years ago), were quantitated in sediments and water from Greifensee, a small lake in Switzerland. The freeze-dried sediments were extracted in an ultrasonic bath using methanol with tetrabutylammonium hydrogen sulfate as an ion-pairing reagent. Aqueous samples were extracted with C18 extraction disks, which were subsequently eluted by methanol with tetrabutylammonium hydrogen sulfate. Both extracts from solid and aqueous samples were analyzed by reversed-phase high-performance liquid · chromatography. Fluorescence detection was applied after postcolumn UV irradiation. Analytical reproducibility ranged from 1 to 12 % (relative standard deviation). Limits of quantitation were 1-11 µg/kg of dry matter and 0.2-3 ng/L for solid and aqueous samples, respectively. Recoveries ranged from 93 to 100 % and from 87 to 95 % in solid and aqueous samples, respectively. Concentrations of DAS 1 and DSBP ranged from 0.4 to 1.4 mg/kg of dry matter in top sediment layers and from 12 to 98 ng/L in lake water. Concentrations of BLS were between 0.02 and 0.08 mg/kg of dry matter in top sediment layers and <0.2 ng/L in lake water. 16 Chapter 2

Stoll J.M. A. and Giger W. (1997). Determination of Detergent-Derived Fluorescent Whitening Agent Isomers in Lake Sediments and Surface Wa- ters by Liquid Chromatography. Anal. Chem., 69, 2594-2599. Analytical Methods 17

2.1 Introduction

Detergents for laundry washing are mixtures of synthetic chemicals, which are used in very large quantities. Worldwide consumption of a major detergent ingredient, linear alkylbenzenesulfonates (LAS), reached 2.8 million tons/year in 1995.1 The environmental fate of major detergent components such as surfactants and builders has, therefore, been the subject of extensive research. Considerably less attention has been paid to the environmental fate of minor detergent components such as fluorescent whitening agents (FW As, Figure 2.1 ), which on average contribute only 0.15 % of the total mass of laundry detergents.2,3 The three most important detergent FW As are sold under the tradenames DAS 1, DSBP and BLS (diaminostilbene, distyrylbiphenyl, and bleachstable, respectively; full names are given in Figure 2.1 ). Worldwide production was estimated at 14 000 tons of DAS 1 and 3000 tons of others (predominantly DSBP) for the year 1989.2 BLS was used in large-scale laundry facilities (e.g., in hospitals) until quite recently.4 All FW As included in this study absorb UV light at 350 nm, with a molar extinction coefficient of over 50 000 M- 1cm- 1 and emit blue visible light at a maximum wavelength of 430 nm, with fluorescence quantum yields of 0.3 (DAS 1) and 0.8 (DSBP).5.6 Sulfonate groups increase the water solubility of the otherwise hydrophobic FW As. Intense blue fluorescence, an affinity for cellulose and water solubility are the main properties that render FW As suitable for whitening applications. Stilbene-based FW As are light-sensitive chemicals, especially when in dilute solution. Irradiation of FWAs with sunlight causes reversible E-Z isomerization of the stilbene moiety, as illustrated in Figure 2.2. Hence, DAS 1 containing one stilbene moiety occurs in two isomeric forms, herein called (E)-DAS 1 and (Z)-DAS 1. For FW As containing two stilbene moieties (DSBP and BLS), three isomeric forms are possible, (E,E)-, (E,Z)- and (Z,Z)-FW A. FW As are produced and added to laundry detergents in their fluorescent E- or £,£-isomeric forms. Photoisomerization to the corresponding Z-, E,Z- or Z,Z-isomers leads to a complete loss of fluorescence. 1 During the textile washing process, FW As partly adsorb o~to fabrics and, thus, maintain the whiteness of the prewhitened material. The FW As remaining in the washing liquor are discharged to sewers for treatment in municipal wastewater treatment plants. FW As are partly retained on the 18 Chapter 2 Q e )-~ H H f ' 038 N }-N =(N - e\ - QN N SO \ ;f rH b'- H N-{ 3 (J DAS1

803 DSBP

e c 803 e 03

BLS

Internal Standard

Figure 2.1 Structures of the FWAs included in this study. Full names: DAS 1: 4,4'-bis[(4-anilino-6-morpholino-l ,3,5-triazin-2-yl)amino]stilbene-2,2'-disulfonate; DSBP: 4,4'-bis(2-sulfostyryl)biphenyl; BLS: 4,4'- bis(4-chloro-3-sulfostyryl )biphenyl. · Internal standard: 4 ,4 '-bis(5-ethyl- 3-sulfobenzofur-2-yl )biphenyl. Analytical Methods 19 activated sewage sludge due to adsorption. The nonretained fraction of FWAs reaches surface waters,4,8-11 where photodegradation takes place with half-lives of several hours at the surface under summer noon sunlight.5 Below the photic zone, FWAs can be assumed to be persistent since biodegradation has not been observed over a period of 30 days,12.13 Nevertheless, the concentrations found in sediments and surface waters are far below those expected to be an ecotoxicological risk.14,15 However, in the light of general environmental chemistry considerations, two aspects are of significant interest: (i) FWAs can serve as molecular markers for domestic wastewater. Their history of input into lakes is recorded in the sediments (environmental archives). (ii) FWAs can be regarded as probe molecules for photochemical processes in surface waters.

H H R hv

R hv R R

(EJ- FWA (ZJ- FWA

Figure 2.2 Reversible isomerization process of stilbene-type FWAs.

The determination of FWAs was first performed by thin-layer chromatography (TLC).8,15-19 Since 1977, several authors have reported the use of HPLC for the determination of FWAs in detergents20-23 and in environmental samples.9,I0,24--28 In most of these methods, the isomers of the FWAs are not separated and thus, only total FWA concentrations can be measured. This is not a problem for detergent analyses or for river monitoring programs. However, if the behavior of FWAs in the environment is to be investigated, quantification of individual FWA isomer concentrations is critical, because E- and Z-isomers of the same FWA show different sorption behaviors4 and the E:Z-isomer ratio plays a major role in photodegradation.5 To the best of our knowledge, there are only two HPLC methods9,IO with which specific FWA isomers can be quantitated. They were designed for the FWA concentrations occurring in municipal 20 Chapter 2 wastewater and in rivers. Less contaminated samples, e.g., lake waters or lake sediments, cannot be analyzed by these methods because the detection limits and blank concentrations are too high. In this study, we present two methods for the determination of FWA isomers in samples of lake sediments and lake waters. These methods were adapted from the methods of Poiger et aI.9,IO in order to achieve lower blank values and detection limits. Besides minor changes in HPLC settings, such as column size and rinsing procedure, the main change was the replacement of plastic material with glassware for the sample enrichment steps. Like the methods of Poiger, our extraction procedure for FWAs from lake sediment is based on conventional liquid extraction, while dissolved FWAs are enriched by solid phase extraction. Individual FW A isomers are quantitatively determined by reversed-phase HPLC with fluorescence detection.

2.2 Experimental Section

2.2.1 Reagents

FWAs, as well as the internal standard (all technical grade), were obtained as sodium salts from Ciba-Geigy AG (Basel, Switzerland), with purities of 97 % (DAS 1), 90 % (DSBP) and 68 % (BLS). 100 % of the FWAs were present as E- or E,E-isomers. 4,4' -Bis(5-ethyl-3-sulfobenzofur-2-yl)biphenyl, a research compound, was used as an internal standard in this study. This compound has not been applied as a FWA and, therefore, does not appear in environmental samples, but shows fluorescence in the same wavelength region as the FW As. Ammonium acetate (analytical grade) was purchased from Merck ABS AG (Basel, Switzerland). Tetra- butylammonium hydrogen sulfate (TBA) was purchased from Fluka AG (Buchs, Switzerland). All solvents (HPLC grade) were purchased from FEROSA (Barcelona, Spain) and were used as received.

2.2.2 Standard Solutions

Standard solutions of FWAs were prepared in DMF/water (1:1). Trip- licate concentration series for external calibration curves of solid samples Analytical Methods 21 were prepared by dilution of the standard solutions in 0.27 M TBA in DMF/water (1:1), corresponding to concentrations of 30-300 and 300- 4000 µg of FW A/kg of dry sediment (for both DAS 1 and DSBP) and 3- 30 and 30-400 µg of FWA/kg of dry sediment (BLS). For aqueous samples, the standard solutions were diluted in 0.2 M TBA in DMF/water ( 1: 1), corresponding to concentrations of 2-100, 100-600, and 600-4000 ng/L (for both DAS 1 and DSBP) and 0.2-10, 10-60, and 60-400 ng/L (BLS). The response of the fluorescence detector was determined by linear regression and showed good linearity over the concentration ranges tested, indicated by correlation coefficients of 0.999-1.000.

2.2.3 Samples

Sediments from Greifensee, a small lake in Switzerland, were collected in February 1995 by means of a gravity coring device. A PVC tube (diameter, 6 cm) was pushed into the lake bottom, closed on the top and pulled out. On shore, the core was sliced horizontally into 5 cm segments corresponding roughly to 10 years of sediment accumulation. The samples were freeze-dried, homogenized with mortar and pestle and stored in the dark at 4 °C. Lake water was collected from the middle of Greifensee. Samples collected in February 1995, 10 m below the surface, were used for recovery and precision measurements. The samples used as application examples were collected in different depths in July 1995 and January 1996. All samples were immediately placed in the dark after collection, transported to the laboratory and analyzed within 2 h without filtration.

2.2.4 Enrichment from Aqueous Samples

Polypropylene filter assemblies (Millipore, Bedford, MA) attached to a vacuum manifold (Supelco, Bellafonte, PA) and equipped with glass funnels were used to support the solid-phase extraction disks (25 mm diameter, Cl8 Empore, Varian, Harbor City, CA). After cleaning with 2 mL of 0.05 M TBA in methanol, the disks were preconditioned using 10 mL of methanol followed by 30 mL of distilled water. The preconditioned disks were not allowed to go to dryness prior to sample application. 22 Chapter 2

Unfiltered water samples were homogenized by shaking prior to extraction so that the total concentration of PW As could be determined. A sample volume of 200 mL was passed through the extraction disk by vacuum. Air was passed through the disk for 5 min after the enrichment to remove excess water in the disk. The filter assemblies containing the extraction disks were separated from the vacuum manifold and from the funnels in order to minimize carryover effects. Syringes containing 2 mL of 0.05M TBA in methanol were connected to the filter assemblies. Elution was performed by first allowing the solvent to soak into the disks for 3 min before it was pushed through the disk with the syringe. The extract was collected in a glass vial, where the solvent was evaporated under a stream of nitrogen with mild heating (50 °C). A 0.5 mL mixture of water and DMF (1: 1) and I 0 µL of the internal standard (1 mg/L water:DMF (1:1)) were then added to the vial. An injection volume of 25 µL was used to analyze the extracts by HPLC.

2.2.5 Liquid Extraction of Solid Samples

Samples of 200 mg of dry sediment were mixed with 9 mL of 0.03 M TBA in methanol in screw-cap test tubes and briefly shaken, followed by sonication for 30 min. The samples were centrifuged at 2500 rpm for 5 min and decanted into 50 mL conical flasks. The extraction was repeated twice and individual extracts were combined. The extracts were evaporated to dryness under vacuum ( 100 mbar, 50 °C) and redissolved in 3 mL of a mixture of water and DMF (1: 1). After addition of 30 µL of internal standard (1 mg/L water:DMF (1:1)), the extracts were transferred to vials, centrifuged at 2500 rpm for 5 min and decanted into autosampler vials. Injection volumes of 10 µL were used to analyze the sediment extracts by HPLC.

2.2.6 High-Performance Liquid Chromatography

All analyses were performed using a Hewlett-Packard Model 1090L Series HPLC equipped with a gradient pump and an autosampler. The FWAs were separated on a reversed-phase microbore column (Hypersil Analytical Methods 23

ODS, 3 µm, 100 mm x 2.1 mm i.d. with precolumn, Hewlett Packard) operated at room temperature with an eluent flow rate of 0.4 mL/min. The mobile-phase solvents were a 2:3 mixture of methanol and acetonitrile (eluent A) and 0.1 M aqueous ammonium acetate buffer of pH 6.5 (eluent B). A 22 min linear gradient from 30 % A/70 % B to 60 % A/40 % B, followed by a 2 min linear gradient to 90 % A/10 % B, was used for analyses. After a washing time of 8 min, the initial eluent composition was reestablished by a 3 min linear gradient, followed by an equilibration time of 5 min. The outlet of the HPLC column was connected to a postcolumn UV irradiation apparatus (Beam Boost, ict AG, Basel, Switzer- land) equipped with a UV lamp having a maximum intensity at 254 nm and a 0.3 mm i.d. x 0.5 m Teflon capillary. This irradiation caused the isomerization of a part of the FW As. The irradiation time of 5 s was chosen in order to achieve photostationary conditions, i.e., a constant E:Z ratio,9 independent of whether the original isomer was in the E- or Z- form. In this manner, all FWA isomers could be detected as (E)-FWA by fluorescence, regardless of whether they had been present as E- or Z- isomers in the original sample (Figure 2.3). The fluorescence detector (Hewlett Packard, 1046A) was operated at an excitation wavelength of 350 nm and an emission wavelength of 430 nm.

2.2.7 Quality Assurance

Sediment blanks were determined by extracting an empty test tube along with every set of 11 samples. Parallel to every set of 16 aqueous samples, 4 samples of distilled water were analyzed. Unpolluted sediment from Greifensee dating from about 1900 was diluted with water and spiked with FW As in order to determine the limits of quantitation (LOQ).29 The spike quantities (DAS l/DSBP, 24 µg/kg of dry sediment; BLS, 2.5 µg/kg of dry sediment) were chosen in order to yield slightly higher concentrations than the expected LOQ. The sediment was stirred for 15 minutes and divided into eight aliquots. These eight aliquots were extracted and analyzed separately. For the water sample LOQs, distilled water was spiked with FW As. The FW A quantities (DAS l/DSBP, 4 ng/L; BLS, 0.4 ng/L) were chosen in order to yield slightly higher concentrations than the expected LOQ. The water was homogenized and divided into five aliquots. These five aliquots were enriched and 24 Chapter 2

fluorescence detection HPLC Aex = 350 nm Aem = 430 nm

postcolumn irradiation (.A = 254 nm, t = 5 s):

achievement of photostationary conditions (constant ratio of E- and Z-isomers)

E-isomer: detection of E-isomer

Z-isomer: also detection of E-isomer !!

Figure 2.3 Postcolumn irradiation and detection of FWAs.

analyzed separately. The LOQs for the E- and £,£-isomers were calculated from the standard deviations (s) of the measurements (LOQ = 10s). The values for the Z- and E ,Z-isomers were assumed to be similar. To evaluate method reproducibility, mean, standard deviation, and confidence intervals of the measurements were determined for individual isomers as follows: 150 g of sediment from the top 15 cm of the benthos from Greifensee was frozen, freeze-dried, homogenized, and divided into eight aliquots. These eight aliquots were extracted and analyzed separately. Eight equivalent samples from Greifensee water were extracted and analyzed separately. Recovery rates of (E)- and (E,E)-FW As were determined for both unpolluted sediment dating from about 1900 and recent sediment from the top 5 cm of the benthos of Greifensee. Both measurements were repeated three times. Sixteen equivalent samples were collected from Greifensee water, eight of which were spiked with FWAs. Measurements were Analytical Methods 25 carried out only for the E- and E,E-isomers, assuming the recovery rates of the Z- and E,Z-isomers to be similar. For the investigation of the behavior of individual FW A isomers, it is essential to prevent exposure of samples to UV or blue visible light emitted by normal laboratory light sources. For this reason, all sample preparation steps were conducted in a windowless room equipped with a special lamp (Philips TLD 36W/16 Yellow), and the extracts were stored in amber vials.

2.3 Results and Discussion

2.3.1 Separation

In contrast to previous reports, an HPLC column with a diameter of only 2 mm was applied for the separation of the FW As. This allowed the injection of smaller sample volumes and saved 60 % on solvents. To achieve baseline-separated peaks, the ratio of acetonitrile and methanol in the solvent had to be adjusted for the new column. Individual photoiso- mers of DAS 1 and DSBP, which occur under environmental conditions, i.e., (E)-DAS 1, (Z)-DAS 1, (E,E)-DSBP, and (E,Z)-DSBP, could be separated. The Z,Z-isomer of DSBP is not formed photochemically5 and, thus, does not have to be considered for natural sample analyses. The three isomers of BLS could not be separated chromatographically, so only total amounts of BLS are reported. The internal standard does not isome- rize and always yields one peak. To assign the measured peaks to the individual FW A isomers, a standard solution of every FWA was irradiated for 10 min by the sun and analyzed twice (illustrated for DSBP in Figure 2.4; results for DAS 1 are qualitatively equal). The first time, the postcolumn UV irradiation was switched off, so that only the fluorescent E- and £,£-isomers gave a signal. The second time, the UV irradiation was turned on again to visualize the nonfluorescent Z- and E,Z-isomers. The peaks that appeared only in the second measurements were assigned to the Z- and E,Z- isomers. In the case of BLS, no second peak appeared in the second measurement, which means that the individual isomers could not be separated. 26 Chapter 2

(E,E)-DSBP (E)-DAS 1 IS I ~ BLS . \,/ i------1 u\.J\.....__ ...... _____,,.,, ____ A •

Figure 2.4 HPLC analyses: (A) reference mixture of (E,E)-DSBP, (E)- DAS 1, and BLS, (B) reference mixture of (E,E)-DSBP and (E,Z)-DSBP without postcolumn irradiation, and (C) reference mixture of (E,E)- DSBP and (E,Z)-DSBP with postcolumn irradiation. For (Z)-DAS 1, see Figure 2.5.

2.3.2 Detection and Quantitation

Photoisomerization is a reversible process, and the photostationary state of a diluted solution depends only on the wavelength of the incident Analytical Methods 27 light and on the temperature. If a dilute solution of PWAs is irradiated for a time sufficient to reach a constant isomer ratio (photostationary state), the resulting E:Z ratios are independent of the initial isomeric composition. In the absence of light, isomerization does not occur, and isomer ratios remain constant. In this study, previously reported UV irradiation9 is applied on-line after HPLC separation and before fluorescence detection in order to minimize matrix effects. The isomer concentrations of all isomers are then determined on the basis of the E- or E,E-isomer only (Fi- gure 2.3). This is possible because (i) only the E- and E,E-isomers of the stilbene FWAs exhibit fluorescence, and the Z-, E,Z-, and Z,Z-isomers give virtually no signal, and (ii) the E:Z ratios are constant after UV irradiation. As described in the introduction section, the high limits of quantitation (LOQ) prevented the application of previously reported methods for PWA determination in lake sediments and surface waters. As the LOQs are strongly affected by the variability of blanks, a major effort was made to identify the sources of high blank values. In the case of aqueous samples, the main source was found in the repeated use of enrichment equipment. The highly sorbing PWAs were carried over from one sample to another by filter assemblies and by the vacuum manifold. Replacing polypropylene funnels with glass funnels and eluting the C18 disks directly in- to glass vials instead of through the vacuum manifold prevented a major part of the carryover. Due to technical problems, the polypropylene filter assemblies could not be replaced. Their substitution would probably have lowered the LOQs even more. Nevertheless, LOQs could be lowered by factors of 10-100 (to 3 ng/L for DAS 1 and DSBP, and to 0.2 ng/L for BLS). The blank signals never exceeded 2 % of the concentrations measured in surface waters for DSBP (E,E and E,Z) and (Z)-DAS I. Only for (E)-DAS 1 were the blank values 10-20 % of the concentrations measured in surface waters. The average of the measured blank values was subtracted from signals obtained for each sample. In the case of sediment samples, blanks were already very low. Only signals corresponding to (E,E)-DSBP could be detected in blank samples, and they never exceeded 3 µg of PWA/kg of dry sediment. The LOQs were lowered by a factor of 2-5 due to the use of an HPLC column with a smaller diameter (2 mm) and a smaller pore size (3 µm). The measured LOQs are 2 (DSBP and BLS) and 10 µg/kg (DAS 1) (Table 2.1 ). 28 Chapter 2

Table 2.1. Limits of Quantitation for FWAs in Sediment and Water Samples

(E)-DAS I (E,E)-DSBP BLS

Sediments (n = 8)a added amount 24.0 24.2 2.5 average measd value 22.6 22.7 2.4 standard deviation 1.05 0.21 0.14 LOQ 10.5 2.1 1.4

Water (n = 5)b added amount 4.13 4.17 0.43 average measd value 5.13 3.99 0.30 standard deviation 0.30 0.32 0.02 LOQ 3.0 3.2 0.2

a In µg of FWA/kg of dry matter. b In ng of FW A/L.

Recoveries (Table 2.2) were between 93 and 100 % and between 87 and 95 % for solid and aqueous samples, respectively. The recoveries reported here may only represent an upper limit. Especially in the case of sediments, one can assume that FW As will show a different binding behavior to particles in the benthos of a lake compared to artificially spiked matrices. FW As at the bottom of a lake have interacted with the particles for decades, whereas in the laboratory experiment interaction lasted only 15 minutes. The precisions of the individual FW A determinations were within 3-7 % (relative standard deviation) for sediment extracts and 1-12 % for aqueous extracts (Table 2.3). Confidence intervals (95 % ) were generally below 8 % of the mean concentration. Only FW A isomers with concentrations close to the detection limits showed poorer precision. The low limits of quantitation, good recoveries, and excellent precision of individual measurements allow for the determination of individual FW A isomers, even at the low ambient concentrations found in lake sediments and surface waters. Analytical Methods 29

Table 2.2. Recoveries of FW As in Sediment and Water Samples

(£)-DAS 1 (E,E)-DSBP BLS

Sediments (n = 3) background concna 0 264 3 318 0 11 spikeda 129 304 130 308 13 32 recovery (%) 99 93 97 96 100 97 standard deviation (%) 3 2 1 3 5 2

Water (n = 8) background concnb 8.5 66.7 0.0 spikedb 49.6 50.l 5.2 recovery ( % ) 88 87 95 standard deviation (%) 2 4 2 a In µg of FWA/kg of dry matter. h In ng of FWAJL.

Table 2.3. Precision of Individual FW A Measurements (n=8)

mean value standard deviation 95 % confidence interval conen con en % concn %

Sedimentsa (£)-DAS I 521.8 17.5 3 ± 41.4 ± 8 (Z)-DAS 1 94.0 3.3 4 ± 7.7 ± 8 (E,E)-DSBP 177.2 4.7 3 ± 11.2 ± 6 (E,Z)-DSBP 13.8 0.9 7 ± 2.1 ± 15 BLS 10.8 0.5 5 ± 1.1 ± 10

Wate~ (£)-DAS 1 8.5 1.0 12 ± 2.4 ±28 (Z)-DAS I 77.5 1.1 I ± 2.6 ± 3 (E,E)-DSBP 66.7 1.3 2 ± 3.1 ± 5 (E,Z)-DSBP 9.8 0.2 2 ± 0.5 ± 5 BLS < 0.2 ndc ndc ndc nd"

a In µg of FW A/kg of dry matter. b In ng of FW AJL. c Not determined. 30 Chapter 2

2.3.3 Applications

Concentrations of DAS 1 and DSBP isomers and total concentrations of BLS were determined in water and sediments of Greifensee, a small lake in Switzerland. Two examples of these analyses are shown in Figure 2.5. In the lake water, two different situations can be observed: (i) in summer, when the lake is stratified, and (ii) in winter, when it is mixed (Table 2.4). In summer, FW A concentrations are lower, due to stronger sunlight and presumably faster rates of photodegradation. As a consequence of stratification, concentrations in the sunlit top layer are lower than concentrations in the layers below 5 m, where no light is available. This effect is more pronounced for DSBP, because it is photochemically degraded more rapidly than DAS 1.5

(E,E)-DSBP ~ (E)-DAS 1

BLS (E,Z)-DSBP Ii (Z}-DAS 1 IS \ A ~ c:: Q) 0en ~ 0 ::I (E,E)-DSBP u::: (E)-DAS 1 IS ~ (E,Z)-DSBP (Z)-DAS 1 \ I B

0 5 10 15 20 25 Time (min)

Figure 2.5 HPLC analyses: (A) Greifensee sediment, top 5 cm (1985- 1995) and (B) Greifensee water, 20 m depth (July 26, 1995). Analytical Methods 31

A second effect is the change in isomer ratios. Isomer ratios are a function of temperature and light conditions.6,30 First, the isomerization rate from E- to Z-isomers depends on temperature, while the reverse isomerization from Z- to £-isomers is temperature-independent. Due to lower temperatures in winter (3 vs 25 °C at the surface), the equilibrium of isomers is, therefore, shifted toward (E)-DAS 1 and (E,E)-DSBP. Second, in both summer and winter, the isomer ratios in deeper waters are shifted towards (Z)-DAS 1 and (E,Z)-DSBP, because longer wavelengths are better transmitted by lake water and favor the Z-isomers. De- pending on how much and how fast the lake is mixed, the deeper layers of the water body are more or less important in defining the isomer ratios of the upper layers. Especially in summer, when the deeper water is colder than the surface water, the two observed trends have opposite effects on the isomer ratio. Further investigations are being carried out to explain the behavior of isomer ratios in lake water.3 I

Table 2.4. FWA Concentrations in Water and Sediments of Greifensee, Switzerland

DASI DSBP BLS (E) (Z) total (Ej(Z) (E,E) (E,Z) total (E,Ej(E,Z) total

Water (ng/L) summer'1 surface 6 47 53 0.13 10 2 12 4.6 <0.2 10 mdepth 5 54 59 0.10 47 9 56 5.3 <0.2 20 m depth 8 65 73 0.12 49 9 58 5.4 <0.2 winter' surface 17 81 98 0.21 63 8 71 8.2 <0.2 10 m depth 11 81 92 0.14 60 9 68 6.9 <0.2 20 m depth 12 83 95 0.15 63 9 71 7.0 <0.2

Sedimenr (µg/kg) 0.2 kmd 1220 200 1420 6.1 1130 14 1140 81 78 1.4 kmd 590 200 790 3.0 500 11 510 45 26 3.1 kmd 470 180 650 2.6 420 9 430 47 19

a July 26, 1995. b January 10, 1996. c Top layer (0-5 cm), February 27, 1995. d Distance from mouth of River Aa in Uster. 32 Chapter 2

In the sediments of Greifensee, a general decrease of FWA concentrations (Table 2.4) is observed with increasing distance from the mouth of the river that delivers FWAs into the lake (River Aa). £-Isomers have higher sorption coefficients 4,30 and, therefore, are transported more efficiently to the lake bottom than Z-isomers. Consequently, the ratios of E- to Z-isomers are about a factor of 10 higher in sediments than in water, an effect that is documented in Figure 2.5. For the same reason, the isomer ratios in the sediments decrease with increasing distance from the river mouth. By studying these effects more thoroughly, we hope to elucidate processes affecting the transport and fate of domestic wastes in natural waters. The concept of investigating one persistent and source- specific substance (a molecular marker) in order to gain information about particular sources of environmental pollution was discussed by several authors.32 Based on their environmental fate properties, fluorescent whitening agents have a high potential as molecular markers for domestic wastewaters.

2.4 Conclusions

The lake water and sediment concentrations of individual isomers of the two commonly used detergent-derived FWAs in Switzerland can be determined accurately by reversed-phase HPLC followed by postcolumn UV irradiation and fluorescence detection. The trends in isomer ratios found in Greifensee can be explained on the basis of the results of laboratory studies recently reported in the literature.6 We are using the methods presented here in a comprehensive research program investigating the temporal and spatial distribution of FWA concentrations in samples from Greifensee (water and sediments) in order to better understand the environmental behavior of detergent-derived FWAs.31 We are especially interested to learn how fast FWAs are degraded in a lake and the extent to which results of laboratory studies on photodegradation can be extrapo- lated to a lake situation. In that way, FW As can serve as probe molecules for photochemical processes in surface waters. In addition, we hope to apply these analytical methods for investigating the behavior of suspended particles and sediments in larger lakes and oceans. Because (i) FWAs sorb strongly to particles and (ii) they are detectable in low concentrations, their application as molecular markers appears to be feasible.33 The me- Analytical Methods 33 thods presented here are excellent tools for these purposes, because sample enrichment is facile, the separation is reliable, and detection is highly sensitive.

Acknowledgment

We thank the Chemical Industry in Basel, Switzerland, for their finan- cial support, Hewlett Packard for the donation of the HPLC equipment within the Rhine Basin Program, and Ciba-Geigy for providing the FWA reference compounds. We acknowledge M. Berg, S. Canonica, R. Egan- house, W. King, H. Kramer, S. Muller, and T. Poiger for their help during the preparation of this manuscript.

2.5 Literature Cited

(1) Ainsworth, S. J. Chem. Eng. News 1996, 74(4), 32-54. (2) Kramer, J. B. In The Handbook of Environmental Chemistry; Hutzinger, 0., Ed.; Springer: Berlin, 1992; Vol. 3, pp 351-366. (3) Anliker, R. In Fluorescent Whitening Agents; Anliker, R., Muller, G., Eds.; Georg Thieme Publishers: Stuttgart, 1975; Suppl. Vol. IV, pp 12-18. (4) Poiger, T. Ph.D. Thesis, ETH Zurich, No. 10832, 1994. (5) Kramer, J. B.; Canonica, S.; Hoigne, J.; Kaschig, J. Environ. Sci. Technol. 1996, 30, 2227-2234. (6) Canonica, S.; Kramer, J. B.; Reiss, D.; Gygax, H. Environ. Sci. Technol.1991,31, 1754-1760. (7) Gold, H. In ref 3, pp 25-46. (8) Ganz, C. R.; Liebert, C.; Schulze, J.; Stensby, S. J. Wat. Pollut. Control Fed. 1975, 47, 2834-2849. (9) Poiger, T.; Field, J. A.; Field, T. M.; Giger, W. Anal. Methods lnstrum. 1993, J, 104-113. (10) Poiger, T.; Field, J. A.; Field, T. M.; Giger, W. Environ. Sci. Technol. 1996, 30, 2220-2226. (11) Poiger, T.; Field, J. A.; Field, T. M.; Siegrist, H.; Giger, W. Water Res. (in press). 34 Chapter 2

(12) Dojlido, J. R. EPA Report 600/2-79-163; Environmental Protection Agency: Washington, DC, 1979. (13) Kaschig, J., Ciba Geigy AG, Basel, Switzerland; personal communication, 1996. (14) Zinkemagel, R. In ref 3, pp 129-142. (15) Burg, A. W.; Rohovsky, M. W.; Kensler, C. J. Crit. Rev. Environ. Control 1911, 7, 91-120. (16) Theidel, H. In ref 3, pp 94-103. (17) Lehmann, G.; Becker-Klose, M. Tenside Deterg. 1976, 13, 7-8. (18) Bloching, H.; Holtmann, W.; Otten, M. Seifen Ole Fette 1979, 105, 33-38. (19) Lepri, L.; Desideri, P. G.; Coas, W. J. Chromatogr. 1985, 322, 363-370. (20) Kirkpatrick, D. J. Chromatogr. 1977, 139, 168-173. (21) McPherson, B. P.; Omelczenko, N. J. Am. Oil Chem. Soc. 1980, 57, 388-391. (22) Micali, G.; Curro, P.; Calabro, G. Analyst 1984, 109, 155-158. (23) Jasperse, J. L.; Steiger, P.H. J. Am. Oil Chem. Soc. 1992, 69, 621- 625. (24) Abe, A.; Yoshimi, H. Water Res. 1979, 13, 1111-1112. (25) Uchiyama, M. Water Res. 1979, 13, 847-853. (26) Komaki, M.; Yabe, A. Nipon Kagaku Kiashi 1982, 5, 859-867. (27) Kato, K.; Mori, H.; Watanabe, N.; Yasuda, Y.; Nakamura, T. Gifu- ken Kogai Kenkyusho Nenpo 1982, 11, 40-43. (28) Abe, A.; Tanaka, K.; Fukaya, K.; Takeshita, S. Suishitsu Odaku Kenkyu 1983, 6, 399-405. (29) Taylor, J. K. Quality Assurance of Chemical Measurements; Lewis Publishers: Chelsea, MI, 1987. (30) Kramer, J. B. Ph.D. Thesis, ETH Zurich, No. 11934, 1996. (31) Stoll, J. M. A. This Ph.D. Thesis. (32) Eganhouse, R. P., Ed. Molecular Markers in Environmental Geochemistry; ACS Symposium Series 671; American Chemical Society: Washington, DC, 1997. (33) Stoll, J. M. A.; Poiger, T. F.; Lotter, A. F.; Sturm, M.; Giger, W. In ref 32, pp. 231-241. Chapter 5 of this Ph.D. Thesis. 3

River Monitoring

The two fluorescent whitening agents (FWAs) that are currently used in laundry detergents in Switzerland (DAS 1, a diaminostilbene, and DSBP, a distyrylbiphenyl) were studied in ten different Swiss rivers. Two-week- composite-samples were analyzed once a month for one year. In most of the investigated rivers, concentrations ranged from 6 - 120 ng L- 1 for DAS 1 and from 10 - 70 ng L- 1 for DSBP. Average per capita loads were 1.8 and 1.3 mg d- 1 inhabitanr1 for DAS 1 and DSBP, respectively. Higher concentrations were found in rivers with densely populated catchment 1 areas and below FWA manufacturing plants (DAS 1: 100 - 1000 ng L- ; 1 DSBP: 50 - 1100 ng L- ). Only sampling locations situated below FW A manufacturing plants showed significantly increased per capita loads compared to the other stations investigated (8.0 mg DAS 1 d- 1 inhabitanr1 at Weil (below Basel) on the river Rhine; 23.1 mg DSBP d- 1 inhabitanr1 at Porte du Scex on the river Rhone). Thus, these FWAs can be applied as molecular markers for production wastewater. The measurements made in Swiss rivers allowed for the development of a mass balance for FWAs. This mass balance indicates that 13 % of the FWAs being used are discharged to surface waters. 36 Chapter 3

Stoll J. M. A. and Giger W. Mass Balance for Detergent-Derived Fluores- cent Whitening Agents in Surface Waters of Switzerland. Water Res. (in press). River Monitoring 37

3.1 Introduction

Laundry detergents contain several synthetic chemicals and are important potential sources for the pollution of natural waters and soils. They are applied in large amounts and are discharged almost quantitatively to municipal and industrial wastewaters. Worldwide consumption of a major detergent ingredient, the linear alkylbenzenesulfonates (LAS), reached 2.8 106 t per year in 1995 (Ainsworth, 1996). It is therefore important to know the behavior and the effects of individual detergent components during sewage treatment and in the environment. Monitoring studies have been performed for major detergent compounds, such as LAS, alcohol ethoxylates (AE), alcohol ethoxy sulphates (AES), alcohol sulphates (AS) and nitrilotriacetate (NTA) as well as for nonylphenol (NP), a degradation product of a nonionic surfactant (nonylphenol polyethoxylate) (Matthijs et al., 1996; Feijtel et al., 1995; Ahel, 1987). One difficulty of these studies is that most of these compounds are applied as mixtures of different isomers and homologues, rendering the determination of individual molecules demanding and costly. Other compounds are not derived only from detergents, but also originate from other sources (e. g. phosphates, EDTA), so that environmental occurrences cannot be attributed to specific sources. For a third group of detergent chemicals, analytical methods don't exist at all (polycarbonates) or detection limits are too high for environmental samples (phosphonates; Nowack, 1997). Thus, monitoring programs in surface waters have been limited to a small number of individual substances such as NT A, EDTA, nonylphenol and substance groups such as LAS. Fluorescent whitening agents (FWAs, Figure 3.1) have the potential to be used as probe molecules and thus extend our knowledge of the behavior of detergent chemicals in surface waters, since their consumption is known relatively well, individual molecules can be determined reliably (Stoll and Giger, 1997), and FWAs represent a class of detergent compounds that is not biodegradable (but is degraded photochemically). So far, only a few environmental samples have been investigated for their FW A content (Poiger et al., 1996). Two compounds are currently used as FW As in Swiss laundry detergents: DAS 1 (Qif!mino§tilbene), and DSBP (gi§tyryl.Qiyhenyl) with an average mass of 0.1 % of the total detergent. In 1995, a total of 59 t of DAS 1 and DSBP were consumed in Switzerland (Eugster, 1997; Gehri, 38 Chapter 3 (J }-N~_H N 1N ~FN b

DAS 1

DSBP

Figure 3.1. Structures of the FWAs included in this study. Full names: DAS 1: 4,4'-bis-[(4-anilino-6-morpholino-1,3,5-triazin-2-yl)amino]stilbene-2,2'-disulfonate, DSBP: 4,4'-bis(2-sulfostyryl)biphenyl.

1995; SWI, 1995). Both DAS 1 and DSBP absorb UV-light at 350 nm with a molar extinction coefficient of 60,000 - 70,000 M- 1 cm- 1 (Kramer et al., 1996) and emit blue visible light at a maximum wavelength of 430 nm with a fluorescence quantum yield of 0.3 (DAS 1) and 0.8 (DSBP) (Cano- nica et al., 1997). This blue fluorescence is used for the application of FWAs in detergents (whitening fabrics) and it can also be exploited for their analytical determination. The sulfonate groups increase the water solubility of the otherwise hydrophobic FWAs, and the large carbon structures cause an affinity for cellulose. During textile washing, FWAs partly remain in the washing liquor and hence are discharged to sewers for treatment in municipal wastewater treatment plants. FWAs are partly retained on the activated sewage sludge due to adsorption. The non-retained fraction of FWAs reaches surface waters, as was reported by Poiger et al. ( 1997), who measured a dis- River Monitoring 39 charge to surface waters of 22 % of the consumed FWAs at one sewage treatment plant in Switzerland. In surface waters photodegradation takes place at the surface with half-lives of several hours in the top 15 mm layer under summer noon sunlight (Kramer et al., 1996). Below the photic zone FWAs can be assumed to be persistent, since bi ode gradation has not been observed (Kaschig, 1996; Dojlido, 1979). Concentrations that have been found in surface waters were always far below the PNEC (predicted no effect concentration) (Burg et al., 1977; Zinkernagel, 1975), so no effort has been made so far to investigate mass-flows in larger regions or during longer periods. In this study we have monitored the FWA concentrations for one year in ten Swiss rivers. This study was superimposed on a river monitoring program that has been carried out in Switzerland since 1975. Within this monitoring program (NADUF, Nationales Programm filr die f!Ilalytische Dauer!!nter- suchung der schweizerischen Fliessgewasser) three different classes of parameters are determined at 18 river stations: (i) general parameters of the rivers (discharge, temperature, pH, etc.), (ii) geochemical parameters (alkalinity, calcium, sulfate, etc.) and (iii) anthropogenic parameters (chloride, phosphate, dissolved organic carbon (DOC) etc.). Hence, the characteristics of these rivers are well known (Jakob et al., 1994). The scope of this study was to extend our knowledge of the behavior and fate of detergent-derived chemicals, especially FWAs. We wanted to see if differences in the catchment areas, such as the presence of lakes, population density, and the presence of FWA manufacturing plants, have an influence on the FWA concentrations in the rivers.

3.2 Sampling Locations, Collection of Samples and Analytical Methods

In order to develop a mass balance for FW As in Switzerland river samples from ten different stations were collected and analyzed during one year. The ten river stations chosen for this study represent three different types of catchment areas in Switzerland (Figure 3.2):

(I) Alpine rivers. These rivers show a smaller influence of human activity compared to other rivers since their catchment areas are 40 Chapter 3

I .I , I ~ I 50km I

Figure 3.2. Map of Switzerland showing major cities (squares) and surface waters with sampling locations (circles) at the rivers Rhine (IA, B, C), Thur (2), Glatt (3), Aare (4A, B), Saane (5), and Rhone (6A, B).

relatively large and have low population densities. Rain and melting snow cause distinct variations in the water discharge.

(II) Large rivers in the Swiss plateau that have lakes in their catchment area. The water of these rivers is contaminated by human activities, but the lakes have a purifying effect. Also fluctuations in their discharges are mitigated by the lakes.

(Ill) Small rivers in the Swiss plateau with highly populated catchment areas. The water quality in these rivers is strongly impacted by human activities.

Two of the sites were chosen to be directly downstream from the only two chemical plants manufacturing FW As in Switzerland: Station IC (Weil, DAS 1) and station 6A (Porte du Scex, DSBP). Hence the influence of the manufacturing site could also be investigated. River Monitoring 41

The sampling sites of the NADUF program consist of a small house be- neath the river. River water including all suspended particles is continuously pumped into the house, mixed proportionally to the discharge, and stored in glass bottles at 2-4 °C. Two-week composite samples were transported to the laboratory every 4 weeks and analyzed within one day. Dissolved and particle-bound FWAs were extracted simultaneously from samples of 200 mL river water by solid-phase extraction (C18 Empore discs: Varian, Harbor City, CA). After separation by reversed-phase HPLC (column: Hypersil ODS, 3 µm, 100 mm x 2.1 mm i.d. with precolumn; Hewlett Packard), individual FWAs were irradiated with a UV lamp (254 nm, 5 s) and detected with fluorescence (excitation: 350 nm, emission: 430 nm; Hewlett Packard, 1046A). Details of this method are described elsewhere (Stoll and Giger, 1997). Recovery rates were 87 - 88 % and standard deviations (n=8) 1 - 12 %. Both DAS 1 and DSBP could 1 be determined down to concentrations of 3 ng L- • In contrast to Stoll and

Table 3.1. Sampling Sites and River Characteristics River Location Nr Catchment Discharge Inhabts/Q lX)C o-Phos- 2 area (km ) a rateQ (inhabts/ (mg L-1) b phate (m3 8-1) b (m3 s·l)) a,c (f!~ L-1) b Group I (Alpine rivers)

Rhine Diepoldsau lA 6, 119 252 1,380 1.0 4 Saane Giimmenen 5 1,880 69 3,550 2.0 14 Rhone Porte du Scex 6A 5,220 217 1,600 0.9 11

Group II (Large rivers in the Swiss plateau)

Aare Bern 4A 2,969 146 2,000 2.2 5 Aare Hagneck 4B 5,140 216 3,900 1.5 7 Rhine Rekingen lB 14,718 526 5,570 2.1 20 Rhine Weil lC 36,472 1284 6,470 2.1 21 Rhone Chancy 6B 10,294 429 4,360 1.4 30

Group III (Small rivers with densely populated catchment areas)

Thur Andelfingen 2 1,696 64 7,140 2.9 74 Glatt Rheinsfelden 3 416 11 39,970 3.3 78

a Jakob et al. 1994, b BUWAL 1995, average values for 1995, c number of inhabitants from 1990, discharge: average 1961-1980, data given for information on the stations only, but not used for calculations. 42 Chapter 3

Giger (1997), no individual FW A isomers were quantified. Since the samples could not be stored completely in the dark before the analyses, an isomerization was to be expected. The isomer ratios determined under these conditions would most probably be different from the isomer ratios occurring in the rivers. The photochemical degradation that is occurring when FWAs are exposed to visible or UV light was expected to be small, since only indirect sunlight affected the samples for a period of less than two hours. Sorption onto the inner surface of the sampling tubes was also neglected because the possible sorption sites were expected to be in equilibrium due to the extensive usage of the tubes. Nevertheless, the concentrations reported here must be considered as minimum values. In addition to the FW A concentrations measured for this study data on discharge rates, DOC, and chloride concentrations were obtained for every individual sample (BUWAL, 1995). Weighted average values (proportional to the discharge) were used for calculations. Typical characteristics of the ten selected stations are listed in Table 3.1 for information and were not used for calculations.

3.3 Results and Discussion

3.3.1 FWA Concentrations and Loads in Swiss Rivers

Based on the measured DAS 1 and DSBP concentrations, the investigated rivers can be divided into two groups (Figures 3.3 and 3.4): The two rivers of group III (small rivers with highly populated catchment areas, sites 2 and 3) and the two sites, that are below the FW A manufacturing plants (site 1 C for DAS 1 and 6A for DSBP) show relatively high FW A 1 1 concentrations (DAS 1: 100 - 1000 ng L- ; DSBP: 50 - 1100 ng L- ). All 1 the other sites have clearly smaller concentrations (DAS 1: 6 - 120 ng L- ; DSBP: 10 - 70 ng L-1). Dilution seemingly has a strong impact on the observed concentrations. The latter fact is derived from the observation that concentrations are directly correlated with water discharge at all stations, i.d. high FW A concentrations at low discharge rates and vice versa (shown in Figure 3.5 for DAS 1 at the stations lA and 4A). On the other hand, a comparison with the temperatures in the investigated nvers showed no direct correlation to the FW A concentrations. River Monitoring 43

Table 3.2. FW A Loads and Content in the DOC (Stations Below Mann- factoring Plants are in Bold) Yearly loada Content in the noca.b (kg y-1) (ppm) GrouE Station DASl DSBP DASI DSBP I IA 120 150 10 20 5 160 60 40 10 6A 320 2,300 50 350 II 4A 170 100 20 10 4B 490 280 50 30 IB 1,000 720 30 20 lC 20,000 5,100 240 60 6B 980 550 50 30 III 2 260 220 40 40 3 150 210 130 180 a average of 13 measurements, b average load/average DOC load (BUW AL, 1995)

1.0 DAS 1 1.0 0.8 ,...... , 0.8 .,.... Ii- ..'...J Q) 0.6 3,; 0 3 I 0.6 ~ ~ E. Cf) I ~ <( 0.4 _, Cf) Cl 0.4 <( Cl 0.2 BA ..JI - JI 0.2 - -»--Jiil- 0 0 0 r-.. ~ C\I O> ~ ~ O> (0 ~ ~ r-.. r-.. ~ C") C\I C\I C\I C\I ~ ~ ~ ..... > 0 c:: ...... Q) a. 0 0 Q) co c:: .0 >. c:: :::J -;) Q) co a. :::J :::J Q) 0 co <( co ...., Cf) z Cl -;) LL ~ ~ -;) <(

Figure 3.3. Concentrations of DAS 1 measured in rivers from 1995 Ja- nuary 30 to 1996 January 1. Dates represent the middle of the sampling periods for two-week-composite-samples. Solid lines: rivers below FWA manufacturing sites (squares: JC, circles: 6A), dashed lines: small rivers ( V 2, L1 3 ), hatched area: all other stations. 44 Chapter 3

DSBP 1.0 3 1"- 2.0 '-....! ' I \ 1.6 0.8 I \ ..--...... I ~ _ti. ....I --tr--l:r \ ~ ~0.6 I \ 1.2 0 ...... \ ~ a...... c IIl \ a.. 0.8 IIl ~ 0.4 \ (J) 0 0.2 0.4

0 0 ,...._ ,...._ ,...._ ,.... ,.... 0 C\J O> ,.... O> co C') C\J C\J C\J C\J ,.... ,.... ,.... (.) .....(.) c ..... """..... """0) a. > """Q) ~ c .c >. c ::::J Q) 0 ~ Q) ~ a. ~ ::::J ::::J 0 ...., <( ...., (J) z 0 ...., LL ~ ~ ...., <(

Figure 3.4. Concentrations of DSBP measured in rivers from 1995 Ja- nuary 30 to 1996 January 1. Dates represent the middle of the sampling periods for two-week-composite-samples. Solid lines: rivers below FWA manufacturing sites (squares: 1 C, circles: 6A), dashed lines: small rivers ( V 2, .1 3 ), hatched area: all other stations.

Annually, DAS I loads ranged from I20 kg at station IA (0.3 I06 inhabitants in the catchment area) to 20,000 kg at station IC (6.9 I06 inhabitants). Annual DSBP loads were between 60 kg at station 5 (0.2 I06 inhabitants) and 5,IOO kg at station IC (Table 3.2). Daily per capita loads (Fi- gures 3.6 and 3. 7) showed much smaller variations throughout the year and between rivers. At all sampling sites (except 1C and 6A, the ones below the FWA manufacturing plants), loads of both DAS 1 and DSBP were between 0.3 and 4.2 mg d- 1 inhabitanr1 throughout the year. These relatively small differences in FWA loads during the year or from one site to another are attributed to unknown factors. The average of all values (except 1C and 6A) was used to calculate the average contribution of Swiss households. Averaged over the investigated year, and based on a population of 7 .06 106 in Switzerland, per capita DAS 1 and DSBP loads were 1 1 1.8 (±0.5) and 1.3 (+0.4) mg d- inhabitanr , respectively. This corres- River Monitoring 45

70- 0 DAS 1 60 - 0

50 - 0 ...- I -1 0 O> 40 B c: 0 0 --r- 0 0 (/) 30 0 0

Figure 3.5. Measured DAS 1 concentrations at the stations IA (squares) and 4A (circles) correlated with the water discharge (BUWAL, 1995). ponds to a total yearly load for Switzerland of 4.6 (+1.3) t DAS 1 and 3.3 (±1.0) t DSBP (95 % confidence interval calculated from the standard deviation). In other words, 13 (±4) % of the FWAs that were consumed during the year can be found in the rivers. The amount that is discharged by sewage treatment plants to surface waters, is probably greater, since FWA loads are reduced in surface waters between the discharge points and the sampling sites due to sorption/sedimentation and photochemical degradation. In contract to the value we obtained, the value of 22 % discharge to surface waters reported by Poiger et al. (1997) is based on a ten day investigation in one sewage treatment plant (STP) only. As the discharge from STPs may vary significantly depending on weather conditions and type of STP, the two values are in good agreement. In contrast to the uniform behavior of most of the investigated rivers, much higher per capita FW A loads were measured at the stations below FWA manufacturing plants (Table 3 .3 ). At sampling site 1C; below the plant manufacturing DAS 1 (Figure 3.6), an average of 8.0 (+1.3) mg 46 Chapter 3

DAS 1 ...... 12 I 1C (DAS 1 manufacturing site) -~ 10 :.a- / ctl ..c: c: 8

.....I "'C Cl 6 ...... E T""" 4 CJ) <( 0 2

0 ['.... ['.... T""" ()') T""" c.o C\J C\J > (.) .0 .... Cl 0.. t5 0 ""'"Q.) 0.. :::J Q.) Q.) -:> :::J 0 0 u.. <( <( CJ) z

Figure 3.6. Daily per capita loads of DAS 1 measured from 1995 January 30to1996 January 1. Dates represent the middle of the sampling periods for two-week-composite-samples. Solid lines: rivers below FWA manufacturing sites (squares: JC, circles: 6A), hatched area: all other stations.

DAS 1 d- 1 inhabitanr1 was found throughout the year. Assuming that the average per capita contribution of households is the same throughout Switzerland, an amount of approximately 6.2 (±1.4) mg d- 1 inhabitanr' (15.7 (±3.5) t DAS 1 per year) must have originated from the manufacturing process of DAS 1. At sampling site 6A, located below the plant manufacturing DSBP (Figure 3.7), the corresponding numbers are 23.1 (±13.3) mg DSBP d- 1 inhabitanr1 and 2.1 (±1.3) t DSBP per year. Thus, 70 % of the total FWA discharge to Swiss rivers can be attributed to FWA production discharge. However, this number is intriguingly high and it must be put into perspective by comparison to the more than 3000 t of FWA that are produced annually in Switzerland for export (Kaschig, 1996). Thus, the amount discharged to surface waters caused by the manufacturing process is only about 0.5 % of the quantity produced. River Monitoring 47

DSBP ...-. "I. 80 c ...... a:s :c 6A (DSBP manufacturing site) ~ 60 c / .- I "O Ol4Q ...... E a.. m en0 20

Figure 3.7. Daily per capita loads of DSBP measured from 1995 January 30to1996January1. Dates represent the middle of the sampling periods for two-week-composite-samples. Solid lines: rivers below FWA manufacturing sites (squares 1 C, circles: 6A), hatched area: all other stations.

The per capita loads of the FW As that are not manufactured above the stations 6A and 1C, respectively, are also slightly higher than at the other 1 1 1 sampling sites (6A: 3.2 mg DAS 1 d- inhabitanr ; 1C: 2.0 mg DSBP d- 1 inhabitanr ). In the case of DSBP this can be explained, because although

Table 3.3. FWAs Discharged to Surface Waters During the Manufacturing Process DAS 1 (IC) DSBP (6A) 1 1 FWA per capita loads (mg a inhabitanf ) Below manufacturing plants 8.0 ± 1.3 23.1 ± 13.3 All other sites 1.8 ± 0.5 1.3 ± 0.4 Contribution from manufacturing plants in units of mg d- 1 inhabitanr1 6.2 ± 1.4 21.7±13.3 in unitsofty"1 15.7 ± 3.5 2.1 ± 1.3 48 Chapter 3

DSBP is not produced above the sampling site IC, it is processed there. The relatively large changes in DSBP loads from one sampling period to the other indicate that this processing is interrupted from time to time, resulting in DSBP loads in the same range as in other rivers.

3.3.2 FWA Mass Balance for Switzerland

An FWA mass balance was developed for Switzerland in order to achieve an overall assessment of the environmental fate of FW As. Source, pathways, and sinks, shown in Figure 3.8, are described below. a) Source and Pathways of FWAs Estimates of the detergent producers (Eugster, 1997; Gehri, 1995) showed a consumption of 59 t of DAS 1 and DSBP in 1995. The confidence interval was estimated at ±5 t. Since DAS 1 and DSBP are only used in laundry detergents, there are two possible pathways to surface waters: 1. Directly from FWA manufacturing plants: During the manufacturing process, FWAs are precipitated from an aqueous solution by adding sodium chloride to the solution followed by filtration. A small fraction of PW As remains in the liquor and is discharged into the industrial wastewater and eventually to surface waters. By attributing the difference between the average FWA load in Swiss rivers and the PW A load below PW A manufacturing plants to the manufacturing process, the PW A discharge to surface waters by manufacturing plants is estimated as 18 (±5) tin 1995 (15.7 t of DAS 1 above site 1C and 2.1 t of DSBP above site 6A, confidence interval estimated with the standard deviation). 2. Through households, institutions and industrial laundries: Measurements in rivers of the study reported here indicate a total discharge of more than 8 (+2) t of FW As to surface waters in 1995. b) Sinks of FWAs Different processes reduce the amount of FW As in sewage effluents and eventually in surface waters. They are called "sinks" here, although only one of them (sorption onto particles in surface waters followed by sedimentation) is a final sink. The four main categories are: River Monitoring 49

1. Fabrics: The purpose of detergent-PW As is to replace FWAs on fabrics, which were photochemically degraded during wearing. One can therefore assume that fabrics are an important "sink" for FWAs. Nevertheless it is difficult to assess the amount of FWA that is really adsorbed onto fabrics during the washing process, because this fraction depends on the material of the fabrics and on the way the fabrics are washed (temperature, ratio of fabrics and water, detergent, water quality). A crude estimation of this "sink" was made by Poiger et al. (1997) by attributing the FWA load lost before introduction to the sewage treatment facility (56 %) to fabrics. For 1995 this would result in an amount of 33 t. Since the basis of this data is uncertain, the confidence interval was arbitrarily estimated at 30 ± 10 t. 2. Sewage sludge: According to Poiger et al. (1997), 32 % of the consumed FWAs are eliminated from the wastewater during sewage treatment due to sorption on- to sewage sludge. For 1995 this corresponds to 19 t, 6 t of which are inci- nerated. The rest is applied on agricultural land ( 13 t), where the further fate has not yet been investigated. Nevertheless, photochemical degradation is likely to occur. Since these data are based on one sewage treatment plant only, the confidence interval is arbitrarily estimated at + 50 %. 3. Lakes: Two processes reduce the FWA concentrations in lakes. FWAs can be (i) degraded photochemically (Kramer et al. 1996) and (ii) sorbed onto particles and sedimented to the lake bottom (Stoll et al., 1997). At this stage it is not possible to assess the relative importance of the two processes. 4. Export out of Switzerland: Part of the catchment areas of the sampling sites 1C and 6B lay outside of Switzerland. The per capita loads calculated for these two sites were mul- tiplied only by the number of inhabitants of the Swiss catchment area (5.2 6 6 10 and 1.1 10 , respectively; Jakob et al., 1994), reflecting only the fraction of FWAs originating from Switzerland. On the other hand, this result was enlarged by 4 %, in order to include the Swiss inhabitants not living in the catchment areas of stations 1C and 6B (Southern Switzerland). In this way the amount of FWAs exported out of Switzerland was assessed as 19 (±4) t for the year 1995 (15 t DAS 1, 4 t DSBP). The confidence interval was estimated with the standard deviation. so Chapter 3

Photodegradation Incineration

?1 ,, I\? ?Ir>. ) f\ ------FABRICS Soll I ~ ? ~ ? I i~ "C er/ l''<..l Q) 3000° I c: Q) .... 0 I 'O 'O :,...- Q) ,_ LO as~ -oo-Q) al I 0 - Q) al al ? ai :::i ('d(/) -al +I ~ I -a GO s~ ~ ~ GO ,_0 s ~ ..... t- (/) I CL '' '' ' I I SURFACE WATER f-t;> Export ~ Sediment ? 19±4d I _ B~undary of th:__system: Switzerland J

PROCESSES goods >

1 Figure 3.8. FWA mass balance for Switzerland (in units oft y· ), 1995. a detergent manufacturer (Eugster, 1997; Gehri, 1995), b Kaschig (1996), c Poiger (1997), d this project.

3.3.3 FWAs as Molecular Markers for Production Wastewater

As shown in Figure 3.9, except for the stations directly downstream from FW A manufacturing plants (DAS 1: 1C; DSBP: 6A), the average FW A concentrations correlated reasonably well (correlation coefficients of 0.95 (DAS 1) and 0.93 (DSBP)) with the chloride concentrations (BUW AL, 1995). As has been demonstrated in other studies (Muller et al., 1996; Jakob et al., 1994), chloride can be used in Switzerland as an anthropogenic indicator because the present chloride levels in rivers are significantly higher than the natural chloride levels anticipated from weathering. Only measurements from the stations directly downstream from FW A manufacturing plants showed FW A concentrations much River Monitoring 51 • 0.6 1C 3 I 0 .,.... I \ ...J 0 Cl 0.4 :i ...... 6A <( ~ • LL 0.2

Chloride (mg L-1)

Figure 3.9. Measured DAS 1 (o) and DSBP ( •) concentrations correlated with the measured chloride concentrations of river water samples (calculated both as weighted average (proportional to the discharge) of the year 1995). higher than expected from the chloride concentration. Thus, DAS 1 and DSBP can be applied as molecular markers for production wastewater. Concentrations of dissolved organic carbon (DOC; BUWAL, 1995) also correlated well with the measured FW A concentrations (correlation coefficients: DAS 1: 0.76; DSBP: 0.93). Consequently the results listed in Table 3.2 show that the two stations at which the FWAs represented the greatest proportion of the organic carbon are those below the manufacturing plants of the corresponding FWA (240 ppm of DAS 1 at 1C and 350 ppm of DSBP at 6A). However, in contrast to the correlation with chloride, one additional site (3), has also higher FWA contents in the DOC than the others. River 3 (Glatt at Rheinsfelden) is by far the smallest of the investigated rivers (with an average flow of 12 m3 s- 1 compared to 3 1 68 - 1320 m s- ) and has a highly populated catchment area. This special situation seems to be reflected in the FW A/DOC ratio. 52 Chapter 3

3.3.4 Elimination of FWAs in Lakes

In lakes, FWA concentrations in the water column are reduced by three processes, photodegradation (Kramer et al., 1996), sorption/sedimentation (Stoll et al., 1997), and flushing. Nevertheless, per capita loads measured at the locations of group II rivers (lakes in the catchment area, sites lB, 1C, 4A, 4B, 6B) are not significantly smaller than the ones of the other locations. The investigated river stations seem to be too far away from the lakes, so that the influence of the lakes is small compared to other factors. However, in a direct comparison of stations located before and after a lake, an influence of the lake can be documented by the FW A loads. Two such examples have been investigated within this study, Lake Constance ( 1A before and 1B after the lake) and Lake Geneva (6A before and 6B after the lake). The FWA loads measured at stations before and after these lakes are listed in Table 3.4 and show that FWA per capita loads are effectively decreased in lakes. The average DAS 1 per capita load decreases in Lake Geneva and remains constant in Lake Constance, while DSBP per capita loads decrease in both lakes. The effect is more pronounced in Lake Geneva, in which the water has a longer residence time (11 vs. 4 y). Thus, FWAs have more time to react photochemically or to sediment to the bottom. On the other hand, the effect is more pronounced for DSBP than for DAS 1. DSBP is degraded faster photochemically (Kramer et al., 1996), while DAS 1 sorbs more strongly to particulate matter (Stoll, 1997). The observation that per capita loads of DSBP decrease more than the ones of DAS 1 indicates that photochemical processes may be more important for the elimination of FWAs in lakes than sedimentation processes. The observed changes in FW A loads are not only due to degradation and sedimentation since, in both lakes investigated, other processes also influence the differences of the FWA loads. Additional rivers and sewage effluents flow into the two lakes and rivers between the sampling locations, so that only about half of the water analyzed at the second locations (lB and 6B) also passed by the first locations (IA and 6A). Furthermore, the average residence time of the water in the Lakes Geneva and Con- stance is several years (11 y and 4 y, respectively). This means that the loads measured before and after the lake during one period do not correspond to each other. Thus, only a rough comparison can be made and the conclusions of this paragraph have a speculative character. River Monitoring 53

Table 3.4. FWA Loads Before (A) and After (B) Lakes. Calculated as Average of 13 Measurements; 95 % Confidence Interval Derived from Standard Deviation. DASI DSBP Station load per capita load load per capita load 1 1 1 (t y-1) (mg d- inhabitanr ) (t y·I) (mg d-l inhabitanr ) Lake Constance IA 0.1 1.0 ± 0.1 0.2 1.3 ± 0.2 lB 1.0 1.1 ± 0.2 0.7 0.8 ± 0.2 Lake Geneva 6A 0.3 3.2 ± 0.4 2.3 23.l ± 13.3 6B 1.0 1.9 ± 0.4 0.5 1.0 ± 0.2

3.4 Conclusions

The present study shows that per capita loads of DAS 1 and DSBP do not vary significantly in the investigated rivers, neither due to changing seasons of the year nor to different catchment areas. Especially the occurrence of lakes in the catchment area did not have an influence on the measured per capita FW A laods. Obviously, only limited conclusions can be drawn from this monitoring study concerning processes affecting the fate of FWAs in surface waters. Averaged over the year, about 13 % of the FW As consumed in Switzerland are found in the rivers and were attributed to private households. Another 30 % are discharged by FWA manufacturing plants directly into the rivers. Therefore, DAS 1 and DSBP can serve as molecular markers for production wastewaters of the manufacturing plants. A good parameter for this purpose is the FW A concentration relative to the chloride concentration, as FW A load and number of inhabitants in the catchment area correlate significantly in rivers which are not influenced by effluents from FW A manufacturing plants.

Acknowledgments

We thank the Swiss Chemical Industry in Basel, Switzerland (Stipendien- fond der Basler Chemischen Industrie) for financially suppo~ting J.-M. Stoll, Hewlett Packard for the donation of the HPLC equipment within the Rhine Basin Program, and Ciba-Geigy for providing the FWA reference 54 Chapter 3 compounds. We acknowledge M. Berg, M. Clayton, R. Eganhouse, G. Henseler, R. Koblet, S. Miiller, T. Poiger and J. Zobrist for their help during the preparation of this manuscript.

3.5 Literature Cited

Ahel M. (1987) Biogeochemical Behaviour of Alkylphenol Polyethoxyla- tes in the Aquatic Environment. Ph.D. Thesis, University of Zagreb. Ainsworth S. J. (1996) Soaps and Detergents. Chem. Eng. News 74( 4), 32-54. Burg A. W., Rohovsky M. W. and Kensler C. J. (1977) Current Status of Human Safety and Environmental Aspects of FW As used in Detergents in the United States. Critical Reviews in Environmental Control 7, 91- 120. BUWAL (Bundesamt fiir Umwelt, Wald und Landschaft) (1995) Hydrolo- gisches Jahrbuch der Schweiz (Swiss Hydrological Yearbook). Eidge- nossische Drucksachen- und Materialverwaltung, Bern, Switzerland. Canonica S., Kramer J. B., Reiss D. and Gygax H. (1997) Photoisomeri- zation Kinetics of Stilbene-Type Fluorescent Whitening Agents. Envi- ron. Sci. Technol. 31, 1754-1760. Dojlido J. R. (1979) Investigations of Biodegradability and Toxicity of Organic Compounds. EPA-Report 60012-79-163. Eugster H. (1997) Personal communication. Mifa, Frenkendorf, Switzer- land. Feijtel T. C. 1., Matthijs E., Rottiers A., Rijs G. B. 1., Kiewiet A., de Nijs A. (1995) AIS/CESIO Environmental Surfactant Monitoring Pro- gramme. Part 1: LAS Monitoring Study in "de Meem" Sewage Treat- ment Plant and Receiving River "Leidsche Rijn". Chemosphere 30(6), 1053-1066. Gehri K. (1995) Erfassung der in der schweizerischen Wasch- und Reini- gungsmittelindustrie verwendeten wichtigsten Rohstoffe im Jahre 1995. Verband der Schweizerischen Seifen- und Waschmittelindustrie (SWI), Zurich, Switzerland. (Statistical yearbook of the detergent producers) Jakob A., Zobrist J., Davis J. S., Liechti P. and Sigg L. (1994) NADUF - Langzeitbeobachtung des chemisch-physikalischen Gewasserzustandes. River Monitoring 55

Gas Wasser Abwasser 74(3), 171-186. (Monitoring-program of Swiss rivers) Kaschig J. (1996) Personal communication. Ciba-Geigy AG, Basel, Swit- zerland. Kramer J. B., Canonica S., Hoigne J. and Kaschig J. (1996) Degradation of Fluorescent Whitening Agents in Sunlit Natural Waters. Environ. Sci. Technol. 30(7), 2227-2234. Matthijs E., Holt M. S., Kiewiet A. and Rijs G. B. J. (1996) Fate of Sur- factants in Activated Sludge Waste Water Treatment Plants. Belg. Chim. Oggi 14(5), 9-:-10. Muller S. R., Zweifel H. R., Kinnison D. J., Jacobsen J. A., Meier M. A., Ulrich M. M., and Schwarzenbach R. P. (1996) Occurrence, Sources, and Fate of Trichloroacetic Acid in Swiss Waters. Environ. Toxicol. Chem. 15(9), 1470-1478. Nowack B. (1997) Determination of Phosphonates in Natural Waters by Ion-Pair High Performance Liquid Chromatography. Submitted to J. Chromatogr. A. Poiger T., Field J. A., Field T. M., Siegrist H. and Giger W. Behavior of Fluorescent Whitening Agents During Sewage Treatment. Water Res. (in press). Poiger T., Field J. A., Field T. M. and Giger W. (1996) Occurrence of Fluorescent Whitening Agents in Sewage and River Water Determined by Solid-Phase Extraction and High-Performance Liquid Chromato- graphy. Environ. Sci. Technol. 30(7), 2220-2226. Stoll J.M. A., Poiger T. F., Lotter A. F., Sturm M. and Giger W. (1997) Fluorescent Whitening Agents as Molecular Markers for Domestic Waste Water in Recent Sediments of Greifensee, Switzerland. In: Mo- lecular Markers in Environmental Geochemistry, ACS Symposium Series 671 (Edited by Eganhouse R. P.). American Chemical Society, Washington DC, pp. 231-241. (Chapter 5 of this Ph.D. Thesis) Stoll J. M. A. and Giger W. (1997) Determination of Detergent-Derived Fluorescent Whitening Agent Isomers in Lake Sediments and Surface Waters by Liquid Chromatography. Anal. Chem., 69, 2594-2599. (Chapter 2 of this Ph.D. Thesis) Stoll J. M. A. (1997) Fluorescent Whitening Agents in Natur.al Waters. Ph.D. Thesis No. 12355, ETH Zurich, Switzerland. (This Ph.D. Thesis) 56 Chapter 3

SWI (Verband der Schweizerischen Seifen- und Waschmittelindustrie) (1995) Jahresbericht 1995. Zurich, Switzerland. (Association of the detergent producers) Zinkemagel R. (1975) Fluorescent Whitening Agents in the Environment. In Fluorescent Whitening Agents (Edited by Anliker R. and Muller G.), Sup. Vol. IV, pp. 129-142. Georg Thieme Publishers, Stuttgart. 4

Lake Model

The dynamic behavior of the two fluorescent whitening agents (FWAs) that are currently used in laundry detergents in Switzerland (DAS 1, a diaminostilbene, and DSBP, a distyrylbiphenyl) has been evaluated quantitatively for Greifensee, a small lake in Switzerland, by using simulation software (MASAS) for modeling organic pollutants in lakes. The one- dimensional simulation was based on (i) independently evaluated processes affecting the fate of FWAs and (ii) measured FWA concentrations in the lake. The measured FWA concentrations were between 50 and 120 ng/L (DAS 1) and between 10 and 110 ng/L (DSBP), with maximum values in the thermocline during summer. The relevant processes were input (114 and 156 g/d averaged over one year for DAS 1 and DSBP, respectively) and elimination by photodegradation, sorption/sedimentation and flushing. Elimination rates integrated over the whole lake and averaged over a year were 0.0044 and 0.016 d- 1 for photodegradation, 0.0025 and 0.0020 d- 1 for sorption/sedimentation, and 0.0021 and 0.0018 d- 1 for flushing (for DAS 1 and DSBP, respectively). Three major results of the simulation model may be pointed out: First, the successful description of seasonal variations in the vertical distribution confirms the previous evaluation of process rates. Second, the model showed that only a part of the particles present in Greifensee are actually sinking, and third, it was shown that 10 - 40 % of the FWA loading into Greifensee in summer occurs at subsurface levels of 4 - 8 m depth. 58 Chapter4

Stoll J. M. A., Ulrich M. M., and Giger W. Dynamic Behavior of Fluo- rescent Whitening Agents in Greifensee: Field Measurements Combined with Mathematical Modeling of Sedimentation and Photolysis. Submitted to Environ. Sci. Technol. Lake Model 59

4.1 Introduction

Laundry detergents contain several synthetic chemicals and are important potential sources for the pollution of natural waters. They are applied in large amounts and discharged almost quantitatively to municipal and industrial wastewaters. For example, worldwide consumption of a major detergent ingredient, the linear alkylbenzenesulfonates (LAS), reached 2.8 106 t per year in 1995 ( 1). It is, therefore, important to know the behavior and the effects of individual detergent components during sewage treatment and after release to the aquatic environment. Several studies have been performed to determine concentrations of specific detergent compounds in various compartments of the aquatic environment. These monitoring studies have provided important information on the occurrence and distribution of detergent-derived chemicals in sewage treatment plants, rivers and lakes. However, the data collected in such studies rarely provide quantitative information on the various processes that determine the fate of a given compound in the system being considered. Investigations focusing on the determination of in situ rates of transport, mixing and transformation processes as well as on the quantification of the inputs of specific organic chemicals to natural waters are, for detergent ingredients as well as for other anthropogenic pollutants, still rather scarce. Two recent studies of organic pollutants in surface waters evaluated the dynamic behavior of tetrachloroethene (PER), atrazine and nitrilotriacetate (NT A) in Greifensee, a small lake in Switzerland (2,3). The main processes affecting the fate of these chemicals are gas- exchange (PER), biodegradation (NTA), and flushing (all three). Another process that might influence the fate of chemicals in natural waters, photodegradation, was studied recently with a similar approach in a river ( 4). In the study presented here, photodegradation has been investigated in a lake together with sorption/sedimentation using fluorescent whitening agents (FW As, Figure 4.1) as probe substances. FW As are degraded photochemically (5) and they sorb onto particles. A recent study of FWAs in the benthos of Greifensee (6) demonstrates that sorption/sedimentation is relevant in natural systems. FWAs have the potential to extend our knowledge of the behavior of detergent chemicals in surface waters. They are rather reactive (photolysis, sorption) and, in contrast to most other detergent ingredients, the consumption of FWAs is known relatively precisely, and individual molecules 60 Chapter4

DSBP

Figure 4.1 Structures of the FWAs included in this study. Full names: DAS 1: 4,4'-bis-[(4-anilino-6-morpholino-1,3,5-triazin-2-yl)amino]stilbene-2 ,2 '-disulfonate, DSBP: 4 ,4 '-bis(2-sulfostyryl )biphenyl. can be determined reliably (7). Two compounds are currently used as FWAs in Swiss laundry detergents: DAS 1 (a gii!.ffiino~tilbene), and DSBP (a gi~tyrylhinhenyl). Normally, FWAs make up an average mass of 0.1 % of the total detergent. In 1995, a total of 59 t of DAS 1 and DSBP were consumed in Switzerland (8-10). Both DAS 1 and DSBP absorb UV-light with a maximum at 350 nm with a molar extinction coefficient of 60,000 · - 70,000 M·1cm·1 (5) and emit blue visible light at a maximum wavelength of 430 nm with a fluorescence quantum yield of 0.3 (DAS 1) and 0.8 (DSBP) (11). This blue fluorescence constitutes the function of FWAs in detergents (whitening the fabrics (12)), and it can also be exploited for their analytical determination. The sulfonate groups increase the water solubility of the otherwise hydrophobic FW As, and the large carbon structures cause an affinity for cellulose surfaces. Lake Model 61

During textile washing, FW As partly remain in the washing liquor and hence are discharged to sewers for treatment in municipal sewage treatment plants (STPs ). FW As are partly retained on the activated sewage sludge due to adsorption. The non-retained fraction of FW As, approximately 20 % of the consumed FW As, reaches surface waters ( 13-15), where FW As are photodegraded in the photic zone. In samples of Greifensee e.g., half-lives amount to several hours under summer noon sunlight in the top 1.5 cm (5). Below the photic zone, FWAs can be assumed to be persistent, because biodegradation has not been observed ( 16, 17). Con- centrations that have been found in surface waters were always far below the predicted no effect concentrations (PNEC) (18-20). The goal of this work was to quantitatively evaluate experimental data derived from several field investigations conducted at Greifensee, using literature data and computer simulation. A particular objective was to test whether both the temporal and vertical (spatial) variations of the FWA- concentrations found in Greifensee could be simulated by a computer model that only includes transport and transformation processes known to be relevant for the given substances and that employs only parameter values in reliable ranges. By understanding the important processes for the fate of FW As in a lake, we hope to learn about (i) particle transport in the lake and (ii) the behavior of detergent-derived chemicals in general. In addition, we wanted to evaluate the extent to which results of laboratory experiments can be used to explain the behavior of chemicals in natural waters. We were especially interested whether the ready photodegradabi- lity of FW As determined in the laboratory can be extended to the real lake situation.

4.2 Experimental Section

4.2.1 Greifensee, the Study Site

Greifensee is a small eutrophic lake located 10 km east of Zurich, Switzerland (Figure 4.2, Table 4.1 ). It is a holomictic and dimictic lake with regular overturn in December and March. Regular successions of oxic (winter/spring) and anoxic conditions (summer/fall) are observed in the hypolimnion of the lake. A long-term data set of particulate organic carbon (POC), temperature, oxygen, pH and global radiation were avai- 62 Chapter 4 lable (21,22). Water gauge recordings for the rivers Aabach, Aa and Glatt were received from the Water Pollution Control Agency of Zurich. The relatively high population density in the catchment area of Grei- fensee causes a significant input of anthropogenic chemicals into the lake. The effluents of seven sewage treatment plants (STPs) are discharged into Greifensee either directly or indirectly via the two major tributaries, Aa and Aabach (Figure 4.2). The only outflow of the lake is the Glatt river.

Switzerland in Europe GERMANY

50km

5km

Figure 4.2 Map of Greifensee, Switzerland, and its catchment area showing major inflows, outflow, residential areas (hatched areas), sewage treatment plants (black squares) and the sampling locations: 1, sampling station at the deepest site of the lake; 2-3, streams: 2, river Aa (Uster); 3, river Aabach (Monchaltorf); 4-6, effluents of sewage treatment plants: 4, Uster; 5, Monchaltorf; 6, Maur. Lake Model 63

Table 4.1. Characteristic Data of Greifensee, the Investigated Lake

3 volume a (m ) 1.51 108 2 surface area a (m ) 8.49 106 max. depth a (m) 32.6 mean depth a (m) 17.8 mean residence time of water a (yr) 1.1 3 1 5 mean throughtlow of water, Q a (m ct· ) 3.7 10 mean epilimnion depth (summer) a (m) 5 3 particulate organic carbon b (g m· ) 0.2 - 2.7 1 settling velocity of particles a (rn ct· ) 0.5 - 2.5 2 catchment area a (km ) 160 inhabitants in catchment area a 100,000 a ref. (2), b llli R. and Ribi B., EAWAG

4.2.2 Sampling and Chemical Analyses

Between April 1995 and April 1996 a total of 14 vertical sample profiles (13-16 depths) were collected above the deepest point of the lake in intervals of 4 weeks (location 1 in Figure 4.2). Concentration profiles at this location were assumed to be representative for the whole lake, because in general, horizontal mixing in Greifensee is fast as compared to elimination processes (2). After collection in a stainless steel sampling bottle (Friedinger, Lucerne, Switzerland), the lake water was transferred immediately into 1 L amber glass bottles, transported to the laboratory and analyzed within 2 hours. The two main tributaries, Aa and Aabach (locations 2 and 3), were sampled with time-proportional portable samp- lers (Manning Products, Texas). Based on water gauge recordings, flow- proportional one-day averaged samples were obtained and analyzed, thus recording the discharge of four of the STPs. The effluents of the other three STPs (locations 4-6, discharging directly into Greifensee) were sampled continuously over 24 h periods using flow-proportional sampling devices. One-day composite samples from both tributaries and STPs were collected on 13 days in 500 mL amber glass bottles, representing different seasons, weekdays and water discharges. After collection, samples were stored in the dark, transported to the laboratory and analyzed within 2 days. The FW As were extracted from samples of 200 mL (lake water) and 50 mL (rivers and STPs) by solid-phase extraction (C18 Empore discs: 64 Chapter4

Varian, Harbor City, CA). After separation by reversed-phase HPLC (column: Hypersil ODS, 3 µm, 100 mm x 2.1 mm i.d. with precolumn; Hew- lett Packard), individual FW As were irradiated in the column eluent with a UV lamp (254 nm, 5 s) and detected with fluorescence (excitation: 350 nm, emission: 430 nm; Hewlett Packard, 1046A). Recovery rates were 87 - 88 % and standard deviations (n=8) 1 - 12 %. Both DAS 1 and DSBP could be determined down to concentrations of 3 ng/L (7). Cylindrical sediment traps (78 cm length and 9 cm diameter (23)) were exposed mid-lake at location 1 at 10 and 30 m depth during the whole investigation period (April 1995 - April 1996). The particulate material in the traps was collected every four weeks and was subsequently freeze-dried, extracted in methanol and analyzed by HPLC (same procedure as for aqueous samples).

4.2 .3 Chemicals

FW As for calibration, all present as E- or £,£-isomers and technical grade, as well as the internal standard (technical grade), were obtained as sodium salts from Ciba-Geigy AG (Basel, Switzerland), with purities of 97 % (DAS 1) and 90 % (DSBP). 4,4' -Bis(5-ethyl-3-sulfobenzofur-2-yl)- biphenyl, a research compound, was used as an internal standard. This compound has not been applied as a FW A and, therefore, does not appear in environmental samples, but shows fluorescence in the same wavelength region as the FW As. Ammonium acetate (analytical grade) was purchased from Merck ABS AG (Basel, Switzerland). Tetrabutylammonium hydrogen sulfate (TBA) was purchased from Fluka AG (Buchs, Switzerland). All solvents (HPLC grade) were purchased from FEROSA (Barcelona, Spain) and were used as received.

4.2.4 Mathematical Simulation Model

The measured data were evaluated using the computer software MASAS (24) (Modeling of Anthropogenic Substances in Aquatic Sys- tems), which allows to construct mathematical models for describing the dynamic behavior of chemicals in lakes. The lake was described by aver- tical, one-dimensional model of 32 boxes, that assumed complete mixing Lake Model 65 within every box and within the epilimnion. However, near point sources (e.g., inflows of streams or STPs ), pollutant concentrations may have been somewhat higher; these parts of the lake were not the focus of this study. In contrast, vertical mixing below the epilimnion was explicitly described as a turbulent diffusion process with time- and depth-dependent diffusion coefficients. These coefficients were computed from independently measured temperature profiles by the heat budget method (25). For lake throughflow, outflow data were used. A more thorough description of MASAS is given in reference (24).

4.3 Results and Discussion

This study is based on three research concepts that are reflected in the text below: First, every relevant process was evaluated independently. In particular, reliable parameters were determined for the FWA input from STP effluents, for the sorption of FWAs to particles and their sedimentation and for the photochemical degradation of FWAs. Second, FWA concentration profiles were measured in the lake during one year, and third, the actual model was set up based on the two previous steps.

4.3.1 Evaluation of FWA Inputs from STP Effluents

In order to assess the loading of FWAs into Greifensee for every .day of the year investigated, total daily loads were measured on 13 days (sum of the sampling locations 2-6 in Figure 4.2; measurements in two smaller tributaries showed no measurable FWA concentrations). These total daily loads were correlated with the total daily sewage (sum of the water that was treated in all STPs of the catchment area) (Figure 4.3). The correlation determined by linear regression showed good linearity with correlation coefficients of 0.94 and 0.95 for DAS 1 and DSBP, respectively. This indicates that FWAs are eliminated from sewage in STPs more effectively when less water is treated (based on the assumption that the daily FW A amount discharged to STPs is constant). This observation can be explained, as FWAs are eliminated from sewage treated in STPs exclusively by sorption onto sewage sludge (13). Sludge is diluted during rainy periods which - in combination with the constant sorption coefficient - 66 Chapter4

300

,... DSBP u' x / .2? 200 Cf) ...... -,( ...... u ...... as ...... 0 ...... "" ...... 0 ...... x ~ ...... 2:- x ...... x:...... ~ :rg 100 as 8 ><..,...... DAS 1 ...... 0 ~

...... *"

0+-~~~~~~-r-~~~~~~--~~~~~~ 20 40 60 80 total daily sewage [ 1000 m3 d-1 J

Figure 4.3 Correlation of the total daily sewage in the catchment area of Greifensee and the total daily FWA loads measured at the sampling locations 2 - 6. Lines show the linear correlations: DAS 1 (crosses, dashed line) = 3.05 Q - 33.3 (r2 = 0.94); DSBP (circles, solid line) = 3.82 Q - 28.2 (r2 = 0.95). 95 % - confidence intervals: 26.6 (DAS 1), 31.8 (DSBP). leads to higher loads in the effluent. The correlation equations were assumed to be representative for the entire investigated year, since samples were collected on all days of the week, in all seasons and under dry as well as rainy conditions. Calculated loadings were 114 ± 27 and 156 ± 33 g/d for DAS 1 and DSBP, respectively, averaged over the whole investigation period (Table 4.2). The values obtained by this procedure were directly used in the model. Lake Model 67

4.3.2 Evaluation of Sorption onto Particles and Sedimentation

The FWA concentrations in the dried particles collected in sediment traps range from 0.2 to 2.2 mg/kg and from 0.1 to 1. 7 mg/kg for DAS 1 and DSBP, respectively. Sorption isotherms (Freundlich, equation 1) were determined by correlating these concentrations (Cs) with the FW A concentrations in the water of the corresponding depth at the end of the collection period of particles (Cw) and then used to assess apparent solid- water distribution ratios Kd following equation 2.

Cs= K·Cwn (eq. 1) C 5 n-1 (eq. 2) Kct =- =K·Cw c w

This procedure resulted in the following equations:

K (l) 107 OOO·C (µg)1.o3 DAS 1: d kg = ' w L (eq. 3) K (l) 24 500-C (µg)o.47 DSBP: d kg = ' w L (eq. 4)

The positive exponents indicate a situation in which previously sorbed molecules lead to a modification of the surface which favors further sorption. Such effects have been observed for surface active compounds like alkylbenzenesulfonates, where the sorbent becomes coated and increasing- ly exhibits a nonpolar nature (26). This seems to be the case for FWAs in the concentration range investigated(~ 0.1 µg/L). On the other hand, negative exponents (and smaller Kd-values) have been reported for higher concentrations (~ 10 mg/L), where a negative surface charge is built up with increasing adsorption, making the surface less attractive to other ne- gatively charged molecules (27). For the computer model, average distribution ratios were determined using equations 3 and 4, and average FWA concentrations (0.088 and 0.062 µg/L for DAS 1 and DSBP, respectively). Average log Kd values (L/kg) are 3.9 and 3.8 for DAS 1 and DSBP, respectively, with a 95 % confidence interval of 0.6 for both DAS 1 and DSBP (Table 4.2). This relatively big confidence interval is due to a large variety in composition and size of in situ particles and their. respective settling velocity. In addition, particles are farmed in some layers and dissolved in others. Bloesch and Sturm (28) concluded, therefore, that it is 68 Chapter4 not possible to calculate "real" settling velocities. This led us to vary the settling velocities used in the model over depth, as will be discussed below.

Table 4.2. Input Parameters for the Computer Simulation, Confidence Intervals are 95 % .

DASI DSBP

1 Loading from inflow (g d- ) a - minimum (October) 67±26 97 ± 31 - maximum (June) 239 ± 26 313±31 - average of 13 periods 114± 27 156 ± 33

Sorption/sedimentation 1 - log Kd (L kg- ) 3.9 ± 0.6 3.8 ± 0.6

1 Photochemical degradation in the perfectly mixed top meter, k (d- ) • - minimum (December) 0.002 0.013 - maximum (May) 0.036 0.156 - average of 13 periods 0.019 0.087 a Individual values for every period were used for the computer simulation (minimum and maximum given as examples). The average is given only for information and was not used for calculations.

4.3.3 Evaluation of Photochemical Degradation

Maximum photodegradation rate coefficients under natural light conditions have been determined for both DAS 1 and DSBP in filtered water from Greifensee (5). These were converted into daily rates proportionally to the flux of photons for every particular day. The new rates were subse- . quently reduced by a factor corresponding to the fraction of light reflected at the air-water interface (6 %, constant throughout the year, (29)) and by a factor corresponding to attenuation by particles in the water (5 - 29 %, estimated by the reduction of visible light in the top meter of Greifen- see ). During summer, periodic stratification of the top meter is likely to occur. Thus, the rate of vertical mixing is slow compared to photodegradation (DSBP e.g., is photodegraded by 50 % within 3 hours in the top 10 Lake Model 69 cm of the lake on a cloudless day in May). In order to compensate for smaller DSBP concentrations in the top meter, and in order to achieve a balanced model, DSBP photodegradation rates from March until Septem- ber were decreased by 10 - 30 % as the simulation model requires a completely mixed epilimnion. The effect for DAS 1 is five times smaller and was, therefore, neglected. Average degradation rates estimated in such a way for the top meter of Greifensee (degradation below 1 meter is virtually zero, as light with wavelengths below 400 nm is completely absorbed within this layer) are 0.019 and 0.087 d- 1 for DAS 1 and DSBP, respectively (Table 4.2). The values obtained by this procedure were directly used in the model.

4.3.4 Seasonal Variation of Measured FWAs in Greifensee

Figure 4.4 shows typical vertical concentration profiles of DAS 1 (dots) determined above the deepest point in Greifensee. As indicated by the temperature profiles (dotted lines), the lake was stratified between May and November, followed by complete overturn in winter. In winter, with less sunlight and less particles in the lake, total FW A concentrations generally increased. In summer, DAS 1 concentrations exhibit a maximum in the thermocline, whereas concentrations in the epilimnion and in the hypolimnion are distinctly lower. These findings are consistent with the assumption that DAS 1 is eliminated from the water column primarily by two processes, photodegradation in the epilimnion and sorption/sedimentation. We formulated two hypotheses about the origin of these peaks: (i) They are a consequence of subsurface loading of the FWAs (model A). This would mean that a part of the water discharged to the lake surface through tributaries and STPs is transported immediately to the depth with the corresponding temperature (and density). During summer this theore- tical intrusion depth is 4 - 8 m from the lake surface. During winter the temperature in the lake is always smaller than in the tributaries, thus giving no evidence for a subsurface loading. (ii) A second possibility is that the peaks in the thermocline are caused by particle dynamics (model B), because a part of the sinking particles is dissolved in the thermocline, thus transporting FWAs from the epilimnion to the thermocline. 70 Chapter4

--- -~ August 21 April 2 0 '.. 0 0 0 0,.... C\I 0,...... C\I ... 0 ,.... ,....0 -... - ...... -...... -...... _. __ ... -~-·~-----~---~~-----w-~•----- ... 0 0 0 0 0 0 0 0 0 0 0 0 ...... C\I ('t) ,.... C\I ('t)

I -0 0 July 26 0 0 0 ()) 0 ' ...... ' C\I 0 .. C\I ...... ca February 5 :::i • • • ••• E ,....0 0 ·c;; ...... _- ... __ ...... ---- ...... - .. _.,...... ---·-~----·------0 0 0 0 • 0 0 0 0 0 0 0 0 -0 ,.... C\I ('t) ...... C\I ('t) ()).... :::i December 13 en 0 June 26 0 0 0 ...... ct! 0 C\I 0 C\I ()) ...... u ...... L E ..... ()) --.•' 0 0 .... ' ' ...... :::i . ... ___ .. .., " ...... _.... _____ ...... --- . -~~------~~~------.. -- ....ct! ::i"' ()) 0 0 0 0 a. O'l 0 0 0 E 0 0 0 0 0 ()) -.s ,.... C\I ('t) ...... C\I ('t) ...... c: .... 0 ()) ~ 0 0 0 October 16 0 ca ...... ,....0 C\I 0 C\I 3: c: -- ' ...... ()) ' . 0 ' c: .-~- ,....0 0 0 ' ...... -..... _,.. _____ ...... 0 -- "' .... "' .. _- .. - ... -- .. -- ...... ,.... en 0 0 0 0

0 0 0 September 18 0 0...... C\I 0...... C\I

0 0 ...... ' ' ...... - ... _... _____ ...... -- ...... , __ ,,, .. ___ -... ~·-·---~------

Figure 4.4 Vertical DAS 1 concentration profiles (dots, scale at the top of graph) and temperature (dotted lines, scale at the bottom of graph) measured above the deepest point of Greifensee in 1995196. Solid lines: Computer simulation with fitted sedimentation velocities (starting April 3, 1995), Model A with depth loading. Lake Model 71

August 21 0 0 0 April 2 0 0 C\I 0 •• C\I T""" • • T""" • ... 0 0 . T""" .. -..... "" ...... _.. ___ T""" ------~------0 0 0 0 0 0 0 0 0 0 0 0 T""" C\I Ct) T""" C\I Ct) ·-. '"O 0 July 26 Q) .. 0 0 0 0 • C\I 0 •• C\I T""" •• .... ~ T""" -:J February 5 E 0 0 ...... _..... __ T""" T""" ·en ... -- ...... ---·-----··------0 0 0 0 • 0 0 0 0 0 0 0 0 '"O T""" C\I Ct) T""" C\I Ct) ....Q) :J 0 December 13 en June 26 0 o. 0 ...... m 0 C\I o, Q) T""" T""" C\I (.) E ...... • • • •• e...... Q) 0 0 ...... T""" - T""" :J - __ _ ---·-----··------.. -- "«;.... ::J Q) 0 0. C) 0 0 0 0 0 0 0 0 E c: 0 0 0 Q) -...... T""" C\I Ct) T""" Ct) C\I c: .... 0 -Q) :;::::: May30 m 0 0 0 October 16 0 m .... 0 C\I 0 C\I -S: c: T""" -·. T""" --...... -Q) ' . . . ••• 0 ' . • • • • • 0 ... c: ~ 0 0 ...... ~ ... __ T""" . T""" 0 •"' -- ...... ·-. ... -·---...... T""" Cl) 0 0 0 0 <( 0 0 0 0 0 0 0 0 a T""" C\I Ct) T""" C\I Ct)

0 Apri! 3 0 0 September 18 0 0 ... - ... ~ - C\I 0 C\I T""" ... --- T""" ...... • • ... ·-=-·· \~ • 0 . 0 T""" ...... __ .., ______... T""" ...... ______~------m 0 0 0 0 "'iD 0 0 0 0 0 0 0 0 ,, T""" C\I Ct) T""" C\I Ct) 0 :IE depth [m] depth [m]

Figure 4.5 Vertical DAS 1 concentration profiles (dots, scale at the top of graph) and temperature (dotted lines, scale at the bottom of graph) measured above the deepest point of Greifensee in 1995196. Solid lines: Computer simulation with fitted sedimentation velocities (starting April 3, 1995), Model B with all loading at the lake surface. 72 Chapter4

0 October 0 0 16 o October 16 0 ,__ C\J 0,__ C\J

... 0 T"""

0 f---,----i-""T---r--...--....,.....l. 0 0-1------..-~~~--.--1. o 0 ,__0 0 0 0 0 0 0 C\J CY) ,...... C\J CY)

September 18 o 0 September 0 0 18 --- C\J ...- ---. C\J --· \•"'. ..! • • ... • ~ • ...... _.. -......

0 +--...---r--r----r--r---r-1 0 0 0 0 0 0 0 0 T""" C\J CY) ,...... C\J CY) ... _.... 0 August 21 o 0 ' 0 ' ' C\J 0 ' T"""

,...... :r ...... _..... _.. 0, .s 0 +----.----.,...--...... ---.--..----.----l 0 c: 0 0 0 0 0 T""" C\J CY)

~..... 0 July 26 o c: 0 (]) C\J (.) c: • • 8 • • • 0 CL ..... ------.. -...... CD Cf) 0 +---r----.--.---...-----,...--_,...... 1. 0 0 +--..---.--,---..----,.----r-1 0 Cl 0 0 0 0 0 0 0 0 ,__ C\J CY) ,- C\J CY)

0 June 26 0 0 June 26 0 0 0 ..- C\J ..- C\J

0 0 ,- ...- --- ...... ~ ...... ~"' ......

o +----..---.--.---...--...--...,_L o m 0 +---..---.-.....---...----ir---r-1 0 0 0 0 0 G) 0 0 0 0 ..- C\J CY) "C ,- C\J CY) 0 depth [m] :!E depth [m]

Figure 4.6 Vertical DSBP concentration profiles (dots, scale at the top of graph) and temperature (dotted lines, scale at the bottom of graph) measured above the deepest point of Greifensee in 1995. Solid lines: Computer simulation with fitted sedimentation velocities (starting April 3, 1995), top: Model A with depth loading, bottom: Model B with all loading at the lake surface. Note the failure of model B confirming that subsurface loading is required to explain the observed behavior of DSBP. Lake Model 73

4.3.5 Computer Simulation, Model Validation

Two simulation models were established and calibrated with DAS 1, model A with subsurface loading and model B without subsurface loading. In both models, sorption coefficients of DAS 1 were kept constant and particulate organic carbon values (POC) were used as measured in the lake. Subsequently, the models were adapted to DSBP to test their validity. Only photodegradation rates and sorption coefficient were exchanged for the values obtained for DSBP. In that way, an independent control was achieved. The sedimentation velocity of the particles was used as a fitting parameter to generate the measured profiles. Subsurface loadings for model A were applied from June to October with increasing depths from 4 - 6 m to 6 - 8 m, corresponding to the depths with the same temperature as the tributaries. Best results with the simulation model were obtained by assuming between 10 % (June and October) and 40 % (July/ August) of the total loadings to enter the lake in the mentioned depths. Figures 4.4 (model A) and 4.5 (model B) indicate that both models can simulate the measured depth profiles of DAS 1. In order to examine the validity of the two models, the same parameters determined for DAS 1 were used to simulate the behavior of the second investigated FWA, DSBP. While model A (with depth loading, Fi- gure 4.6, top) reproduces the profiles of DSBP very well, model B (without depth loading, Figure 4.6, bottom) is not able to simulate the behavior of DSBP. Obviously, transport by particles alone cannot explain the increase of FWA concentrations in the thermocline. A second process, depth loading, must be involved. First order rate constants for the whole lake were determined for photodegradation and flushing by averaging the constants employed with model A over the volume of the whole lake. Corresponding constants for sedimentation were calculated with model A. A summary of these constants is given in Table 4.3. In addition to the validation of the model with DSBP, two controls were performed: First, the measured and the modeled seasonal variation of the total FWA amount in the lake were compared, showing good correspondence for model A (Figure 4. 7). A second control was achieved by comparing the measured and modeled FWA sedimentation fluxes at depths of 1O and 30 m (Figure 4.8). Considering the large confidence interval determined for the sorption coefficients (factor 4), measured and modeled data are in good correspondence. However, there seems to be a systema- 74 Chapter4 tical error yielding to big values in the simulation model. A possible explanation is that the distance of the sedimentation traps from the lake surface was depending on the water level of the lake, thus varying from 9 to 10 m and from 29 to 30 m, respectively. In summer, when the sedimentation velocity of model A has a strong gradient around 10 m depth (Figure 4.9), a reduction of the trap depth to 9 m causes a reduction of settling FWAs of 25 %. In fact, the depth of the upper sedimentation trap was 9.1 - 9.3 m during the periods with the biggest error (Apr. 3 - May 30 and Jun. 26 - Aug. 21). The extreme values of measured DSBP fluxes in period 5 (Jul. 26 - Aug. 21) were assumed to be artifacts and were not included in the calculation of the sorption coefficients.

Total FWA mass in Greifensee in 1995/96

DAS 1 15 ...... O> ...._.~ CJ) Q,. 1;;0'- ~ 10 1 '()... E "' 0

Apr3 Jun 26 Sep 18 Dec 13 Mar4 date in 1995/96

Figure 4. 7 Seasonal variation of the total DAS 1 and DSBP mass in Greifensee in 1995196; dots represent the measured total mass (black: DAS I; white: DSBP; calculated from the concentration profiles and the corresponding water volumes); lines represent the results of the computer simulation, model A (solid: DAS 1; dashed: DSBP ). Lake Model 75

FWAs in sedimentation traps in Greifensee, 1995/96

DSBP 10 m depth

10 OAS1 30 m depth B10 DSBP 30 m depth C\IB C\I E 8 .§ 8 ...... ~ 6 ...... ~ 6

sampling dates in 1995/96

Figure 4.8 Vertical DAS 1 (left) and DSBP (right) fluxes at depths of 10 and 30 m measured with sedimentation traps (filled boxes) and ~ calculated with the computer simulation, Model A (open bars).

Table 4.3. First Order Rate Constants Averaged for the Whole Lake (Pho- todegradation and Flushing) and Calculated by Model A (Sedimentation), and Their Relative Contribution to the Total FWA Reduction

a b k min k ma~ k average c contribution to total FWA 1 oo-3.d-1) oo-3.ct-I) 00·3.ct· ) reductionc

DAS I photodegradation 0.9 8.8 4.4 49% sorption/sedimentation 0.9 4.9 2.5 27% flushing 1.1 4.8 2.1 24% DSBP photodegradation 5.5 29.6 15.5 80% sorption/sedimentation 0.7 4.2 2.0 10% flushing 0.9 3.7 1.8 10% a periods 8/9 (October 16 - December 13), b period 3 (May 30 - June 26), c averaged over the whole year (April 1995 - April 1996) 76 Chapter4

4.3.6 Computer Simulation, Transport by Sedimenting Particles

As shown in Figure 4.9 for July, calculated sedimentation velocities of particles (dots) had to be modified for the computer simulation (solid line) in order to achieve the measured depth profiles. There is a good qualitative agreement with both calculated and adapted sedimentation velocities being highest between 10 and 25 m lake depth. While the quantitative differences are small in the epilimnion, they can be interpreted as being due to incorrect assumptions about the suspended particulate matter

0 • 5

10 ,...... , ..__.E 15 :aQ) "O._ Q) ca 20 ~

0 0.5 1.0 1.5 2.0 sedimentation velocity [m/d]

Figure 4.9 Sedimentation velocities of particles in Greifensee (July). Dots: calculated with the assumption of a constant particle flux. Bases of the calculations: measured suspended particle concentrations in the water column and measured sedimentation rates from sedimentation traps. Line: fitted in order to achieve the measured depth profiles with the computer simulation, Model A. Lake Model 77 concentration [S] in the hypolimnion. Equation 5 (2), where vs is the average particle settling velocity and hbox is the average depth of the box considered, shows the dependence of the sedimentation rate constant (ksed) from [S].

k - vs (1 1 ) sect - -h- . - 1 K · [S] (eq. 5) box + d

By using the measured POC-values for calculations, the value of [S] is over-estimated, as a part of the particles in the water column are not sinking and thus not transporting FWAs to the ground of the lake. In order to compensate for over-estimated [S]-values, the particle settling velocities had to be increased. The same observation was also made by Bloesch and Sturm (28), who increased the hypolimnetic particle settling velocity by a mean factor of three to 3 mid for a lake model of a lake similar to Grei- fensee.

4.3.7 Computer Simulation, Photodegradation in Greifensee

The computer simulation that could successfully be applied, is a confirmation for (i) the photodegradation rates measured in a laboratory (5) and (ii) their transformation to "effective" degradation rates in a lake by taking into account the water mixing and the absorption of photons by the water body. "Effective" photodegradation coefficients in Greifensee, averaged over the whole lake volume, vary from 0.9 to 8.8 10-3 d- 1 and from 5.5 to 30 10-3 d- 1 for DAS 1 and DSBP, respectively (Table 4.3). This means that less than 70 % of both DAS 1 and DSBP is degraded within 28 days by photochemical processes, even in summer with the strongest sunshine. DAS 1 and DSBP are therefore not ready photodegradable in Greifensee following the definition about degradability given by an EC directive (30). Note that this statement is somewhat arbitrary, as the EC directive does not specify the "aquatic environment". If only the photic zone of Greifensee is considered as "aquatic zone", as was done by Kaschig et al. (31 ), DAS 1 (in spring and summer) and DSBP (throughout the year) are ready photodegradable. However, we think it is more appropriate to define the whole lake as "aquatic environment". 78 Chapter4

Acknowledgments

We thank the Swiss Chemical Industry in Basel, Switzerland, for financially supporting J. M. Stoll, Hewlett Packard for the donation of the HPLC equipment within the Rhine Basin Program, and Ciba-Geigy for providing the FWA reference compounds. We acknowledge J. Bloesch, H. R. Biirgi, R. Eganhouse, G. Goudsmit, H. Kramer, and P. Krebs for their help during the preparation of this manuscript. In addition, we thank H. Biihrer, R. Illi and B. Ribi for their help during sampling and for providing ancillary data.

4.4 Literature Cited

(1) Ainsworth, S. J. Chemical & Engineering News 1996, 74(4), 32-54. (2) Ulrich, M. M.; Miiller, S. R.; Singer, H. P.; Imboden, D. M.; Schwarzenbach, R. P. Environ. Sci. Technol. 1994, 28, 1674-1685. (3) Muller, S. R.; Berg, M.; Ulrich, M. M.; Schwarzenbach, R. P. Environ. Sci. Technol. 1997, 31, 2104-2113. (4) Kari, F. G.; Giger, W. Environ. Sci. Technol. 1995, 29, 2814-2827. (5) Kramer, J. B.; Canonica, S.; Hoigne, J.; Kaschig, J. Environ. Sci. Technol. 1996, 30, 2227-2234. (6) Stoll, J. M. A.; Poiger, T. F.; Lotter, A. F.; Sturm, M.; Giger, W. In Molecular Markers in Environmental Geochemistry; ACS Sympo- sium Series 671: Washington, DC, 1997, pp 231-241. (Chapter 5 of this Ph.D. Thesis) (7) Stoll, J. M. A.; Giger, W. Anal. Chem. 1997, 69, 2594-2599. (Chap- ter 2 of this Ph.D. Thesis) (8) Eugster, H., Mifa, Frenkendorf, Switzerland; personal communication, 1997. (9) Gehri, K. Verband der Schweizerischen Seifen- und Waschmittelin- dustrie (SWI), Zurich, 1995. (10) SWI Verband der Schweizerischen Seifen- und Waschmittelindustrie, Zurich, 1995. (11) Canonica, S.; Kramer, J. B.; Reiss, D.; Gygax, H. Environ. Sci. Technol. 1991,31, 1754-1760. (12) Anliker, R.; Miiller, G. Fluorescent Whitening Agents; Georg Thieme Publishers: Stuttgart, 1975. Lake Model 79

(13) Poiger, T.; Field, J. A.; Field, T. M.; Siegrist, H.; Giger, W. Water Res. (in press) (14) Poiger, T. Ph.D. Thesis, ETH Zurich, No. 10832, 1994. (15) Ganz, C. R.; Liebert, C.; Schulze, J.; Stensby, S. J. Wat. Pollution Control Federation 1975, 47, 2834-2849. (16) Kaschig, J., Ciba Geigy AG, Basel; personal communication, 1996. (17) Dojlido, J. R. EPA-Report 600/2-79-163, 1979. (18) Richner, P.; Kaschig, J.; Zeller, M. Presentation at the 7th Annual Meeting of SET AC-Europe, Amsterdam, 1997. (19) Burg, A. W.; Rohovsky, M. W.; Kensler, C. J. Critical Reviews in Environmental Control 1977, 7, 91-120. (20) Zinkemagel, R. FWAs in the environment; Georg Thieme Publi- shers: Stuttgart, 1975. (21) Liechti, P. In Schriftenreihe Umwelt (Gewiisserschutz); BUW AL: Bern, Switzerland, 1994; Vol. 237, pp 131-138. (22) Biihrer, H., EA WAG, Diibendorf; personal communication, 1997. (23) Rosa, F.; Bloesch, J.; Rathke, D. E. In Handbook of techniques for aquatic sediments sampling. 2nd ed.; CRC Press: Boca Raton, Flori- da, 1994, pp 97-129. (24) Ulrich, M. M.; Imboden, D. M.; Schwarzenbach, R. P. Environmen- tal Software 1995, 10, 177 - 198. (25) Imboden, D. M.; Eid, B. S. F.; Joller, T.; Schurter, M.; Wethel, J. Schweizerische Zeitschrift far Hydrologie 1979, 4115, 177-189. (26) Schwarzenbach, R. P.; Gschwend, P. M.; Imboden, D. M. Environ- mental Organic Chemistry; John Wiley & Sons, Inc.: New York, 1993. (27) Kramer, J.B. Ph.D. Thesis, ETH Zurich, No. 11934, 1996. (28) Bloesch, J.; Sturm, M. In Sediments and Water Interaction; Springer Verlag: New York, 1986, pp 481-490. (29) Sauberer, F. Mitt. int. Ver. Limnol. 1962, 11, 1-77. (30) European Community Commision Directive 93/21/EEC (67/548/EEC, 18th ATP) OJ No Lll0/20, 1993, Apr. 27. (31) Kaschig, J.; Hochberg, R.; Zeller, M. Presented at the 4th World Surfactant Congress, Barcelona 1996, 5

Lake Sediment

Three different fluorescent whitening agents (FWAs) were examined in sediment cores of Greifensee, a small lake in Switzerland. Two of these FWAs (DAS 1 and DSBP) are currently used in domestic detergents. The third one (BLS) was contained in detergents until some years ago. During sewage treatment, FWAs are only partly eliminated, and, hence, residual amounts reach the aquatic environment. They are partly associated with particles (Kct = 30 - 440 L/kg) and can, therefore, settle to the bottom of surface waters. This behavior and the resistance of the FW As to biodegradation allows the application of FWAs as molecular markers for domestic wastewater. Total DAS 1 and DSBP inventories in the sediments 2 2 of Greifensee were found to be 18 - 270 mg/m and 7 - 80 mg/m , 2 respectively. Inventories of BLS ranged from 0.3 - 11 mg/m • With increasing distance from the discharge points, FWA inventories generally decreased. A sediment profile in Greifensee collected by means of a freeze core device shows the input history of FWAs from their first use in the mid 1960s. Because the equilibrium of FW As in lake sediments between dissolved and particulate fraction lays strongly on the particulate side, remobilization in the sedimentary core is assumed to be negligible. Thus, concentrations in a particular sediment layer can be attributed to the inputs occurring at the time of deposition. 82 Chapter 5

Stoll J. M. A., Poiger T. F., Lotter A. F., Sturm M., and Giger W. (1997). Fluorescent Whitening Agents as Molecular Markers for Domestic Waste Water in Recent Sediments of Greifensee, Switzerland. In: Molecular Mar- kers in Environmental Geochemistry (Eganhouse R. P. ed.), ACS Sympo- sium Series 671, American Chemical Society: Washington, DC, pp. 231- 241. Lake Sediments 83

5.1 Introduction

Detergents used for laundry washing and cleaning of surfaces are mixtures of different synthetic chemicals which are used in very large quantities. Worldwide consumption of detergent products was estimated at 31 million tons per year in 1984 (J). The most important components, on a weight basis, are surfactants, builders and bleaching agents. Other components are e.g. enzymes, foam regulators, dyes and perfumes. To improve the whiteness of textiles, fluorescent whitening agents (FWAs, Figure 5.1) are added to laundry detergents in relatively small amounts of about 0.15% on a dry weight basis (2, 3). Exact quantities of FWA use are unknown, but estimates showed a world wide production of 14,000 t of DAS 1 and 3,000 t of others (predominantly DSBP) in 1989 (2). FWAs contained in detergents serve to replace textile FWAs which are photochemically degraded during wearing or are washed out during the washing process (2). FWAs that are not attached to the fabric during the washing process are discharged with the washing water. In Switzerland, two compounds are currently being used as detergent FWAs: DAS 1 since the 1960s and DSBP since 1972 (4, 5). BLS, the third FWA included in this study, was used in large-scale laundry facilities (e.g. hospitals) until quite recently (6). The internal standard of this study is a fluorescent research compound that was never used as FWA and, therefore, does not occur in environmental samples. All FWAs included in this study are based on one or two stilbene moieties. Exposure of dissolved FWAs to sunlight causes reversible E-Z-isomerization of the stilbene moiety (Figure 5.2, (7)). Hence DAS 1 containing one stilbene moiety occurs in two isomeric forms, herein called (E)-DAS 1 and (Z)-DAS 1. With two stilbene moieties present in FWAs (DSBP and BLS), three isomeric forms are possible, (E,E)-, (E,Z)- and (Z,Z)-FWA. The Z,Z- isomer of DSBP is not formed photochemically. Thus, only amounts of (E,E)- and (E,Z)-DSBP are reported. The three isomers of BLS could not be separated chromatographically, so only total amounts of BLS are reported. FWAs are produced and added to laundry detergents in their fluorescent E- or E,E-forms. Photoisomerization to the corresponding Z-, E,Z- or Z,Z-forms leads to a complete loss of fluorescence (8). The environmental fate of DAS 1, DSBP, and BLS is summarized in Table 5 .1. Concentrations of FWAs in raw influent typically range from 10 to 20 µg/L (6, 9, JO). FWAs are partly retained in municipal waste water 84 Chapter 5

DSBP

BLS

Internal Standard

Figure 5.1 Structures of the FWAs included in this study. Full names: DAS 1: 4,4'-bis[(4-anilino-6-morpholino-J ,3,5-triazin-2-yl)amino]stilbene- 2 ,2 '-disulfonate; DSBP: 4,4'-bis(2-sulfostyryl)biphenyl; BLS: 4,4'-bis(4- chloro-3-sulfostyryl)biphenyl. Internal standard: 4,4'-bis(S-ethyl-3- sulfobenzofar-2-yl)biphenyl. Lake Sediments 85

R H H hv

R hv R R

(E)- FWA (Z)- FWA

Figure 5.2 Reversible isomerization process of stilbene-type FWAs.

treatment plants due to adsorption onto activated sludge, yielding sludge concentrations of 10 - 100 mg/kg dry matter. Neither biological degradation by activated sludge nor photochemical degradation were observed during sewage treatment (6, 9, 11). Both seem to be too slow compared to the residence times in the treatment plants. The concentration of DAS 1 in secondary effluents of four Swiss sewage treatment plants was found to range from 2.6 to 4.5 µg/L, which corresponds to an average elimination of 89% through activated sludge treatment. The corresponding values for DSBP and BLS were 3.3 - 8.9 µg/L (53% eliminated) and 0.01 - 0.15 µg/L (98%), respectively (6, 9). Average concentrations of DAS 1 in thirty-five rivers in the United States have been determined to range from 0.06 µg/L to 0.7 µg/L at locations above and below sewage outfalls, respectively (13). Recent studies in Swiss rivers showed concentrations for DAS 1 and DSBP ranging from 0.005-0.9 µg/L and 0.01-1.0 µg/L, respectively (15). BLS concentrations were too low to be quantified in these studies (< 0.2 ng/L). In surface waters, FWA concentrations are reduced by photodegradation (7) and sorption/sedimentation (9), whereas biodegradation was not observed over 28 days (4, 12 - 14). Sorption to sediment and little or no biodegradation indicate that FWAs may be present in sediments in detectable quantities. The objectives of this study were to investigate whether FWAs could serve as molecular markers for domestic wastewater and whether the sedimentary column could be used as a natural archive for the use of detergent chemicals by humans and for changes in loading of natural waters with pollutants. This project is part of a larger research program that includes other 86 Chapter 5 components of laundry detergents such as LAS, a widely used synthetic surfactant (16).

Table 5.1. Environmental Fate of Fluorescent Whitening Agents

DASI DSBP BLS Reference

Sewage treatment - biodegradation n. o. a n. o. a n. o. a (6, 9, 11) - photodegradation n. 0. a n. o. a n. o. a (6, 9) - sorption on sludge 89% 53 % 98% (6, 9) 41 - 81 % (10) - concentration ranges (µg/L) - before treatment 7.1 -11.4 6.9 - 21.3 0.27 -2.40 (6, 9) 14- 57 (10) - after treatment 2.6 - 4.5 3.3 8.9 0.01 - 0.15 (6, 9) 6-27 (10) - sludge concentrations, dry matter (mg/kg) 55 - 105 27 - 58 3 - 11 (6, 9)

Surface waters - biodegradation n. o. a n. o. a n. 0. a (4, 12-14) - photodegradation, tin 4.5 h 1.5 h n. d.b (7) - sorption (L/kg) - Kd (E or E,E-isomer) c 444 218 n. d.b (9) Kd (Z or E,Z-isomer) c 109 32 n. d.b (9) - concentration ranges (µg/L) - rivers in Switzerland 0.005 -0.9 0.01 1.0 <0.0002 (15) - rivers in U.S. A. 0.06-0.7 (13) - Greifensee, Switzerland 0.04- 0.1 0.01 - 0.1 <0.0002 (15)

• n. o. = not observed, b n. d. = not determined, c in river water and sediment

5.2 Study Site

Greifensee (Figure 5.3) is a highly eutrophic lake with regular overturn in winter (December - March). It is located I 0 km to the east of Zurich, 2 3 Switzerland, and has an area of 8.5 km , a volume of 0.15 km and a maximum depth of 32 m. Water enters the lake mainly through two tributaries (Aa and Aabach) and leaves through the outflowing river Glatt with a mean flow of 4.3 m3/s. The residence time of the water in the lake is I.I years. A high population density of 600 inhabitants per km2 (total I00,000) in the Lake Sediments 87

2 catchment area (167 km ) produces a significant amount of wastewater (17). The wastewater is treated by seven sewage treatment plants, five of which deliver their water into the two tributaries, Aa and Aabach. The two others, situated on the east and on the west lakeside, deliver their water directly into the lake, the one on the east side (Uster) being much larger. Summing up, three quarters of the wastewater produced in the catchment area enter the lake from the east side near Uster and one quarter is discharged to the lake through the tributary Aabach, coming from the south- east. The sewage treatment plant located on the western lakeside of Grei- fensee contributes only a few percent of the total wastewater input. In the catchment area of Greifensee, sewage treatment commenced in 1967 - 1971, and direct filtration was added to sewage treatment in 1981 - 1984. These changes resulted in two significant decreases of particle discharge to the lake.

Switzerland in Europe GERMANY

50km

5km

Figure 5.3 Map of Greifensee, Switzerland, and its catchment area showing major inflows, outflow, residential areas (hatched areas), and wastewater treatment plants (black squares). 88 Chapter 5

Since the middle of this century anoxic conditions in the hypolimnion have prevailed during summer, preventing higher life forms in contempora- ry sediments. Consequently, the surficial sediments are not bioturbated. In- dividual yearly sediment layers (varves) can be distinguished by the changing colors of the sediments (spring: gray, autumn: black), thus, permitting visual dating of sediment cores.

5.3 Experimental,

5.3.1 Samples

Two methods were used to collect sediments of Greifensee: (i) a conventional gravity coring device and (ii) a recently developed freeze corer (18). For the gravity coring method, a tube of PVC (polyvinylchloride, diameter: 6 cm) was charged with a weight and allowed to sink into the lake bottom, closed on the top, and pulled out. Onshore the core of 30 - 40 cm was directly divided into 5 cm sections corresponding each to roughly 10 years of sedimentation time. For the freeze core method, a metal sword was lowered 70 cm into the sediments and cooled down to -65 °C with carbon dioxide in methanol. The cold sword froze a layer of 4 cm of sediment around itself and was pulled out. The frozen sediment along the length of the sword was then divided into layers corresponding to periods of two years each. The conventional coring method was used to collect 16 cores from different sites in the lake. The freeze core technique was applied for collecting a sediment core from the center of the lake. The samples were placed in jars, freeze-dried, homogenized with pestle and mortar and stored at 4°C.

5.3.2 Reagents

FWAs, as well as the internal standard (all technical grade), were obtained as sodium salts from Ciba-Geigy AG, with purities of 97% (DAS 1), 90 % (DSBP) and 68 % (BLS). 100 % of the FWAs were present as E- or E,E-isomers. Ammonium acetate (analytical grade) was purchased from Merck ABS AG (Basel, Switzerland). Tetrabutylammonium hydrogen sulfate (TBA) was purchased from Fluka AG (Buchs, Switzerland). All sol- Lake Sediments 89 vents (HPLC grade) were purchased from F.E.R.O.S.A. (Barcelma, Spain) and were used as received.

5.3.3 Liquid Extraction

Samples of 200 mg dry sediment were mixed with 9 ml of 0.03 M TBA in methanol in screw-cap test tubes and briefly shaken, followed by sonication for 30 min. The sample was centrifuged at 2500 rpm for 5 min and decanted into a 50 ml measuring flask. The extraction was repeated twice, and the extracts were combined. The extract was evaporated to dryness under vacuum and re-dissolved in 3 ml of a mixture of water and DMF (di- methylformamide, 1: 1). After addition of 30 µl internal standard (1 mg/I 4,4' -bis(5-ethyl-3-sulfobenzofur-2-yl)biphenyl in a mixture of water and DMF (1:1)), the extract was transferred to a vial, centrifuged at 2500 rpm for 5 min and decanted into an autosampler vial. An injection volume of 10 µI was used to analyze the sediment extract by HPLC. To prevent isomerization or photochemical degradation, samples with FWAs in solution were not exposed to UV or blue light.

5.3.4 High-Performance Liquid Chromatography

All analyses were performed using a Hewlett-Packard model 1090L Series HPLC equipped with an autosampler and a ternary solvent delivery system. The FW As were separated on a reversed phase micro-bore column (Hypersil ODS, 3 µm, 100 x 2.1 mm i.d. with pre column, Hewlett Pac- kard) operated at room temperature with an eluent flow rate of 0.4 ml/min. The mobile phase solvents were a 2:3 mixture of methanol and acetonitrile (eluent A) and 0.1 M aqueous ammonium acetate buffer of pH 6.5 (eluent B). A 22 min linear gradient from 30% A I 10 % B to 60 % A I 40 % B followed by a 2 min linear gradient to 90 % A I 10 % B was used for analyses. The outlet of the HPLC column was connected to a postcolumn UV irradiation apparatus (Beam Boost, ict AG, Basel, Switzerland) equipped with a UV lamp with a maximum intensity at 254 nm. The eluate was irradiated in a 0.3 mm i.d. x 0.5 m Teflon capillary for 5 seconds (Figure 5.4). This irradiation time was sufficient to achieve photostationary conditions (6), thus transforming a fraction of the non fluorescent Z- and E,Z-isomers 90 Chapter 5 to fluorescent E- and E,E-isomers. The irradiated eluate was then monitored with a Hewlett Packard model 1046A fluorescence detector at an excitation wavelength of 350 nm and an emission wavelength of 430 nm. This detection method was adopted from (6,19) and will be published elsewhere. A typical result is shown in Figure 5.5.

fluorescence detection HPLC Aex = 350 nm Aern = 430 nm

postcolumn irradiation (.A= 254 nm, t = 5 s): achievement of photostationary conditions (constant ratio of E- and Z-isomers)

E-isomer: detection of E-isomer

Z-isomer: ~:::: also detection i--. of E-isomer !!

Figure 5.4 Postcolumn irradiation and detection of FWAs.

Standard solutions of FWAs were prepared in DMF/water (1: 1). Con- centration series for external calibration curves were prepared by dilution of the standard solution in 0.27 M TBA in D MF/water (1: 1), producing linear correlation from the limits of quantitation (DAS 1: 10 µg/kg dry sediment, DSBP and BLS: 2 µg/kg dry sediment) up to signals corresponding to concentrations of 4 mg/kg dry sediment (DAS 1 and DSBP) and 0.4 mg/kg dry sediment (BLS). Solvent blanks never exceeded Lake Sediments 91 a value corresponding to 2 µg FWA I kg dry sediment and were subtracted from the measured signals. Confidence intervals (95 %, n=8) of individual measurements were+ 8 % ((E)- and (Z)-DAS 1), ± 6 % ((E,E)-DSBP), ± 15 % ((E,Z)-DSBP) and ± 10 % (BLS). Recovery rates were 93 - 99 % (DAS 1), 95 - 97 % (DSBP) and 97 - 100 % (BLS) and were taken into consideration for reported results.

(E,E)-DSBP /

Q) (,) c (E)-DAS 1 Q) IS (,) BLS "'.._Q) (E,Z)-DSBP I 0 (Z)-DAS 1 :::::J 1993/94 II/ LL \

1977/78

1959/60

0 5 10 15 20 25 Time (min)

Figure 5.5 High performance liquid chromatograms of extracts from Grei- fensee sediments. IS: internal standard. Years given represent time of deposition. Fluorescence: IL(ex) = 350 nm, IL(em) = 430 nm. For chromatogra- phic conditions, see text. 92 Chapter 5

5.4 Results and Discussion

5.4.1 FWAs as Molecular Markers for Domestic Waste Water

Gravity cores were collected at 16 different locations of Greifensee. Every core was analyzed from the sediment water interface down to a layer where no more FW As could be found (normally below 20 cm). The concentrations were integrated over the whole core in order to obtain total 2 FWA inventories (in mg FWA/m ) for every core. As illustrated in Figure 5.6, the sample sites were divided into 3 groups, being under the influence of tributary Aabach (group A) and being under the influence of tributary Aa (group B towards south, group C towards north). Table 5.2 shows the

DSBP inventory Glatt 7-9 mg/m2 9 - 13 mg/m2 13- 40 mg/m2 >40 mg/m2

sampling sites O gravity cores ~ freeze core 1 km contour interval = 10 m

Figure 5.6 Sampling sites and spatial distribution of DSBP inventories in the sediments of Greifensee. A: Sampling sites under the influence of tributary Aabach, Band C: Sampling sites under the influence of tributary Aa towards south and north, respectively. Lake Sediments 93

FWA inventories, depending on the distance of the sample site from the mouth of the corresponding tributary. FWA inventories were generally found to decrease with increasing distance from wastewater inputs. To visualize this effect, contours to different DSBP concentrations were manually drawn on Figure 5.6. A second effect that can be seen in Table 5.2 is a change in isomer ratios with increasing distance from tributaries. Since (E)-DAS 1 and (E,E)-DSBP have higher sorption coefficients, they sediment faster to the lake bottom than the corresponding (Z)-DAS 1 and (E,Z)-DSBP. Hence (E)-DAS 1 and (E,E)- DSBP are enriched compared to (Z)-DAS 1 and (E,Z)-DSBP in areas close to the tributaries.

Table 5.2. FWA Inventoriesa in Sediments of Greifensee, Switzerland

distance DAS 1 DSBP BLS from tributary (£) (Z) total (E)/(Z) (E,E) (E,Z) total (E,E)/(E,Z) total

group A (Aabach) 0.3 km 243.4 21.9 265.2 11.1 77.0 1.2 78.2 63.1 11.5 0.6 km 35.6 6.6 42.2 5.4 21.7 0.8 22.5 28.2 1.8 0.7km 14.9 3.3 18.2 4.5 9.9 0.5 10.4 20.1 0.5 1.4km 19.7 5.6 25.4 3.5 8.0 0.4 8.4 18.4 0.3

group B (Aa towards south) 0.2km 157.9b 15.0b 172.9b 10.5 43.7b 2.6b 46.3b 16.7 3.8 0.5km 40.3 8.1 48.4 5.0 12.2 0.8 13.0 15.1 0.7 1.0km 28.5 6.7 35.2 4.2 10.3 0.7 11.0 14.8 0.5 l.Okm 27.5 5.7 33.2 4.8 9.6 0.8 10.4 12.8 0.4

group C (Aa towards north) 0.2km 157.9b 15.0b 172.9b 10.5 43.7b 2.6b 46.3b 16.7 3.8 1.1 km 49.8 7.6 57.4 6.6 16.0 0.9 16.9 17.3 1.0 l.4km 27.7 5.6 33.3 4.9 9.6 0.7 10.3 13.8 0.5 1.8km 24.3 6.4 30.7 3.8 7.5 0.5 8.0 15.2 0.3 2.2km 21.4 5.5 26.9 3.9 8.6 0.4 9.1 19.2 0.4 2.2 km 26.5 6.6 33.1 4.0 7.6 0.5 8.1 15.8 0.5 2.3km 20.9 5.9 26.8 3.6 6.9 0.6 7.5 12.1 0.3 2.9km 18.9 4.8 23.7 4.0 9.0 0.5 9.5 19.0 0.4 3.1 km 21.1 6.1 27.2 3.5 8.3 0.5 8.8 16.3 0.4

a Inventories of cores from the sediment/water interface down to a depth where no more FW As could be detected, unit: mg/m2 b Minimum values, FW As could still be detected in the lowest layer analyzed ( 40 cm) 94 Chapter 5

5.4.2 Sedimentary Archive and Emission History of FWAs

One freeze core was taken in the middle of the lake. Layers corresponding to periods of two years were analyzed. The so obtained vertical concentration and flux profiles of FWAs in the sediments of Greifensee docu- ment the emission history of the FWAs (Figure 5.7): DAS 1 is first found in sediment layers that were deposited in 1965, DSBP in layers from 1973. BLS is only found in layers dating from the 1980s. These findings match the very limited information which was available from the FWA manufacturing industry. Concerning the general development of FWA concentrations in the sediment, the two expected trend changes in 1967 /71 and in

sewage FWA FWA treatment concentration flux 0 95

2 -E A B 91 0 -Q) 0 .. ro 4 87 't: LI rl ...... Q) '< .S: 83 ...... 6 1.:-_, m.... c: ·~--- 0 Q) f ,1 E L r- 79 c. -('() 8 .I :.aQ) .-J "'O Cf) L- 75 ~ l:: ;:::;: -&110 '-,, 5· '-1 r-- 71 ::::J ;: ___I --- ;: ... ~-- .Q 12 67 Q) ..a ,-• DAS 1 -- DAS 1 ..c:: 63 DSBP - DSBP 0..14Q) 0 BLS ·••••• BLS 59 16 0 1 0 1 2 2 mg FWA I kg dry matter mg FWA/ m -y

Figure 5.7 Historic record of FWAs in the sediments of Greifensee measured as concentrations (A) and fluxes (B ). The years plotted on the right- side scales mark the borders of analyzed layers. Lake Sediments 95

1981/84 can be observed. Laundry detergent use increased continuously during this century, but FWA discharge to the environment, as recorded in the sediment, only increased until 1971 and then more or less stagnated until 1983, after which a decrease in FW A concentrations can be observed. The first change in 1971 corresponds to the beginning of sewage treatment in 1967171. The second one in 1983 reflects the addition of direct filtration to sewage treatment in 1981/84.

5.5 Conclusions

Detergent-derived FW As could be determined in sediment cores of Greifensee. The known properties of FW As indicate a conservative behavior in the sediments. Moreover, historic events like the introduction of a new FWA or the improvement of sewage treatment that can be documented by the measured FWA concentration profiles in the sediment make processes that would affect the FW A concentrations in the sediment core unlikely. Because FWAs are strongly sorbed to particles, sedimentation is a rapid process compared to the horizontal mixing in Greifensee, and, hence, concentrations in the sediments can provide information about the transport of particles in the lake. Another property that could provide information, is the ratio of E- and Z-isomers of the FWAs, since exposure to sunlight causes isomerization and different isomers show different sorption behavior. For all these reasons we propose the use of FWAs as molecular markers for domestic wastewater. The investigations and results reported here are part of a comprehensive study on the fate and behavior of FWAs in Greifensee (15). Concentra- tions of FW As have also been measured in samples from seasonal depth profiles of the water column, from the inflows to the lake and in particles which were collected in sediment traps deployed in the lake. These data, together with solid/aqueous phase distribution coefficients and photochemical reaction constants determined in the laboratory, will be fed into a mathematical simulation model in order to evaluate processes determining the fate of FW As in the aquatic environment. 96 Chapter 5

Acknowledgment

We thank the Chemical Industry in Basel, Switzerland for their finan- cial support, Hewlett Packard for the donation of the HPLC equipment within the Rhine Basin Program and Ciba-Geigy for providing the FWA reference chemicals. In addition, we thank R. Berger, A. Liick, J. Maag, R. Reiser, R. Stockli, and A. Zwyssig for their help during sampling. A. Alb- recht, J. Kaschig, and H. Kramer are acknowledged for their assistance during the preparation of this manuscript.

5.6 Literature Cited

1) Jakobi, G.; Lohr, A. Detergents and Textile Washing; VCH Verlagsgesellschaft mbH: Weinheim, D, 1987. 2) Kramer, J. B. In: The Handbook of Environmental Chemistry; Hutzinger, 0., Ed.; Anthropogenic Compounds, Detergents; Springer: Berlin, D, 1992; Vol. 3, F; pp 351-366. 3) Anliker, R. In: Fluorescent Whitening Agents; Anliker R. and Muller G., Ed.; Environmental Quality and Safety; Georg Thieme Publishers: Stuttgart, D, 1975, Sup. Vol. IV; pp 12-18. 4) Personal communication; Kaschig, J.; Ciba Geigy AG, Switzerland 1996. 5) Siegrist, A. E.; Eckhardt, C.; Kaschig, J.; Schmidt, E. In: Ullmann's Encyclopedia of Industrial Chemistry; VCH Verlagsgesellschaft mbH: Weinheim, D, 1991, Vol. A18, pp 153-176. 6) Poiger, T.; Field, J. A.; Field, T. M.; Giger, W. Anal. Methods Instrum. 1993, l, 104-113. 7) Kramer, J. B.; Canonica, S.; Hoigne, J.; Kaschig, J. Environ. Sci. Technol. 1996, 30, 2227-2234. 8) Gold, H. In: Fluorescent Whitening Agents; Anliker R. and Muller G., Ed.; Environmental Quality and Safety; Georg Thieme Publishers: Stuttgart, D, 1975, Sup. Vol. IV; pp 25-46. 9) Poiger, T. Ph.D. Thesis, ETH Zurich, No. 10832 1994. 10) Ganz, C. R.; Liebert, C.; Schulze, J .; Stensby, S. J. Wat. Pollution Control Federation 1975, 47, 2834-2849. 11) Katayama, M. Nippon Kasei Gakkaishi 1989, 40, 1025-1028. 12) Dojlido, J. R. EPA-Report 60012-79-163 1979. Lake Sediments 97

13)Burg, A. W.; Rohovsky, M. W.; Kensler, C. J. Critical Reviews in Environmental Control 1911, 7, 91-120. 14) Komaki, M.; Kawasaki, M.; Yabe, A. Sen-I Gakkaishi 1981, 37, 489- 495. 15) Stoll, J.M. A. This Ph.D. Thesis. 16) Reiser, R.; Toljander, H. O.; Albrecht, A.; Giger, W. In: Molecular Markers in Environmental Geochemistry; Eganhouse, R. P., Ed.; ACS Symposium Series 671, 1997, pp. 196-212. 17) Liechti, P.; Der Zustand der Seen in der Schweiz; Schriftenreihe Umwelt; BUWAL: Bern, CH, 1994; Vol. 237, pp 131-138. 18)Lotter, A. F.; Renberg, I.; Stockli, R.; Sturm, M. submitted to Aquatic Sciences 1997. 19) Poiger, T.; Field, J. A.; Field, T. M.; Siegrist, H.; Giger, W. Water Res. (in press). Appendix

Tables Al-2 FWA concentrations in Swiss rivers (Chapter 3)

Tables A3-6 FWA concentrations in the water of Greifensee (Chapter 4)

Tables A7-10 FW A concentrations in settling particles in Greifensee (Chapter 4)

Tables A 11-15 FW A concentrations in the tributaries of Greifensee (Chapter 4)

Tables A 16-17 Parameters used for the computer simulation of Greifensee (Chapter 4)

Tables Al8-21 FWA concentrations in the benthos of Greifensee (Chapter 5) 100 Appendix

Table Al. Concentrations of DAS 1 in Two-Week-Composite-Samples 3 Collected in Swiss Rivers in 1995/96 (µg/m ) station river Jan 23 Feb 20 Mar 20 Apr 17 May 15 Jun 12 Jul 10

Feb 6 Mar 6 Apr 3 May 1 May 29 Jun 26 Jul 24

Andelfingen Thur 108.3 105.0 93.3 118.0 99.0 153.4 169.5 Bern Aare 57.2 42.2 43.6 25.1 31.2 19.9 Brugg Aare 100.0 101.7 85.7 Chancy Rhone 78.1 77.1 86.2 70.7 55.9 48.0 25.7 Diepoldsau Rhine 35.0 26.2 32.7 14.2 6.1 8.0 6.0 Giimrnenen Saane 68.8 57.6 69.7 52.4 58.2 48.7 76.0 Hagneck Aare 93.3 77.8 79.8 74.1 63.4 41.9 49.3 Porte du Scex Rhone 72.3 67.3 72.9 75.8 34.2 34.3 23.3 Rekingen Rhine 87.4 58.1 71.0 70.1 59.7 45.7 45.9 Rheinsfelden Glatt 272.4 283.9 256.0 499.3 387.5 372.7 646.4 Weil Rhine 517.5 428.7 367.1 564.5 370.2 532.4

station river Aug 7 Sep 4 Oct 2 Oct 30 Nov 27 Dec 25

Aug 21 Sep 18 Oct 16 Nov 13 Dec 11 Jan 8

Andelfingen Thur 138.6 149.8 177.0 126.7 106.9 Bern Aare 27.7 37.9 33.5 49.1 66.6 Brugg Aare 0.0 103.8 113.6 130.7 Chancy Rhone 53.6 74.0 101.0 121.1 89.1 84.6 Diepoldsau Rhine 14.3 10.0 21.2 24.0 22.9 40.6 Giimrnenen Saane 88.3 77.4 67.5 79.9 92.2 77.4 Hagneck Aare 64.1 78.4 83.1 99.8 92.0 Porte du Scex Rhone 38.9 39.8 70.l 61.3 61.4 93.8 Rekingen Rhine 55.8 57.5 42.5 58.4 57.8 76.9 Rheinsfelden Glatt 625.4 557.2 498.2 581.3 436.9 255.8 Weil Rhine 688.0 505.2 600.6 986.2 745.8 278.1 Appendix 101

Table A2. Concentrations of DSBP in Two-Week-Composite-Samples 3 Collected in Swiss Rivers in 1995/96 (µglm )

station nver Jan 23 Feb 20 Mar20 Apr 17 May 15 Jun 12 Jul 10

Feb 6 Mar 6 Apr 3 May 1 May 29 Jun 26 Jul 24

Andelfingen Thur 102.6 115.5 47.1 103.8 95.0 102.7 129.1 Bern Aare 34.5 34.2 22.8 17.3 16.6 11.4 Brugg Aare 66.8 52.9 52.9 Chancy Rhone 39.5 39.3 42.9 35.9 32.3 27.5 18.9 Diepoldsau Rhine 35.7 30.8 35.6 24.8 11.3 12.7 11.4 Gilmmenen Saane 29.6 25.4 22.6 20.5 21.4 17.9 29.8 Hagneck Aare 65.7 56.1 46.4 34.9 31.1 27.5 24.0 Porte du Scex Rhone 696.4 300.6 627.7 250.2 59.0 821.1 243.8 Rekingen Rhine 65.8 53.1 60.8 48.8 34.2 32.5 25.7 Rheinsfelden Glatt 463.4 483.8 410.2 690.6 540.9 467.6 717.1 Weil Rhine 110.7 194.5 78.7 78.8 46.2 48.9

station river Aug7 Sep 4 Oct 2 Oct 30 Nov 27 Dec 25

Aug 21 Sep 18 Oct 16 Nov 13 Dec 11 Jan 8

Andelfingen Thur 115.7 130.0 190.8 145.3 93.1 Bern Aare 15.0 16.1 17.9 30.5 42.4 Brugg Aare 82.3 82.9 87.0 Chancy Rhone 27.6 40.2 56.7 71.2 62.6 47.8 Diepoldsau Rhine 20.7 15.6 19.3 20.4 26.3 40.8 Gilmmenen Saane 29.0 27.4 29.3 33.5 44.9 34.4 Hagneck Aare 36.3 39.2 42.6 53.8 55.0 Porte du Scex Rhone 159.1 219.9 227.3 417.6 299.1 336.6 Rekingen Rhine 32.4 33.5 28.7 43.6 46.3 54.7 Rheinsfelden Glatt 687.3 705.0 743.9 1090.7 900.4 339.8 Weil Rhine 210.6 161.8 81.3 202.1 539.5 62.6 102 Appendix

Table A3. Concentrations of (E)-DAS 1 in Lake Water Collected above the 3 Deepest Point of Greifensee in 1995/96 (µg/m )

depth Feb6 Mar6 Apr3. May2 May30 Jun 26 Jul 26 Aug21

Om 12.8 17.2 13.1 9.0 8.1 11.0 6.2 11.7 lm 10.2 12.1 7.8 7.5 5.5 6.8 4.3 6.8 2m 11.5 10.8 7.9 6.0 5.3 8.4 4.6 7.0 3m 5.8 8.4 8.2 5.2 6.7 4m 12.1 9.6 8.1 6.4 10.5 7.4 5.7 5m 5.8 8.5 6m 9.9 9.0 8.7 6.8 10.1 12.6 3.5 8.4 8m 11.6 11.2 8.8 7 .1 10.1 12.4 5.3 12.6 lOm 10.5 10.5 10.5 5.4 8.8 9.1 5.1 9.9 12m 9.9 13.3 10.2 5.7 5.4 8.3 7.4 14m 9.9 11.4 13.4 16m 9.6 5.4 8.2 5.4 7.0 20m 13.1 6.2 5.9 8.1 8.2 24m 12.2 10.0 5.0 8.4 5.7 3.6 8.5 26m 14.0 11.6 10.7 5.5 10.4 6.5 28m 13.8 12.7 11.6 4.5 9.4 7.4 6.0 6.6 30m 13.3 14.3 10.8 4.8 8.4 6.2 3.7 6.4 31 m 18.2 11.4 4.3 6.1 5.6 6.8

depth Sep 18 Oct 16 Nov 13 Dec 13 Jan 10 Feb5 Mar4 Apr2

Om 9.4 9.8 11.5 13.7 17.2 20.0 18.2 19.6 lm 8.9 18.4 2m 6.8 9.2 12.3 13.2 14.1 10.9 15.6 3m 8.9 16.5 4m 8.5 6.6 8.9 10.9 9.8 13.8 12.5 18.0 5m 10.2 6m 13.3 7.4 9.3 11.5 10.9 16.1 11.3 16.3 8m 19.1 11.0 9.6 11.2 12.2 11.6 11.6 13.7 9m 12.6 8.9 13.1 lOm 13.0 13.6 9.0 10.4 11.1 15.4 12.9 13.5 11m 12.2 9.2 10.2 12m 9.0 11.2 12.0 11.3 9.1 14.9 13.2 13.6 14m 9.3 9.7 11.6 13.3 20.7 15.8 16m 8.5 9.0 12.1 10.3 14.8 10.9 13.5 20m 7.8 9.7 8.3 10.0 12.2 14.4 11.3 13.8 22m 11.2 13.2 13.8 24m 8.2 8.4 8.1 9.4 12.6 10.1 14.5 13.0 26m 13.9 10.2 15.7 14.1 28m 10.9 9.7 9.2 11.3 12.3 17.1 12.7 30m 10.2 8.3 11.5 16.0 9.7 16.2 12.0 31 m 8.8 10.4 11.9 16.4 9.7 15.7 10.9 Appendix 103

Table A4. Concentrations of (Z)-DAS 1 in Lake Water Collected above the 3 Deepest Point of Greifensee in 1995/96 (µg/m )

depth Feb6 Mar6 Apr3 May2 May30 Jun 26 Jul 26 Aug21

Om 83.3 80.1 81.8 72.2 47.5 73.7 46.7 50.1 Im 89.3 90.2 88.2 73.5 51.1 80.1 52.8 52.2 2m 90.5 87.2 83.6 75.0 55.0 72.0 54.2 S2.4 3m 74.9 60.8 71.6 55.8 S5.7 4m 91.7 86.3 86.6 75.9 Sl.S 71.0 74.4 Sm 83.7 110.8 6m 86.1 86.7 84.1 78.6 43.2 7S.7 81.6 80.4 8m 92.S 92.0 89.3 79.0 S0.5 62.1 S6.0 SS.3 lOm 89.5 89.4 83.0 86.0 66.6 53.2 54.2 55.7 12m 93.S 87.9 83.0 88.2 71.3 62.6 S9.5 14m 86.6 87.2 8S.1 16m 85.4 83.9 68.8 64.7 66.0 20m 87.1 84.9 70.2 6S.3 69.0 24m 88.2 87.9 88.8 91.S 72.7 6S.O 68.2 26m 97.S 88.3 86.1 88.3 88.3 71.4 28m 9S.4 90.0 87.3 84.0 86.7 70.6 66.4 69.9 30m 88.4 82.8 87.4 84.6 83.5 73.5 68.0 70.6 31 m 84.4 86.7 86.0 74.7 70.2 73.9

depth Sep 18 Oct 16 Nov 13 Dec 13 Jan 10 FebS Mar4 Apr2

Om S9.5 68.6 73.6 77.0 81.1 82.7 75.0 74.2 lm 60.9 73.8 2m 62.6 72.3 77.5 78.8 83.6 79.4 76.8 3m 61.9 76.6 4m 63.2 75.8 78.7 79.9 77.4 85.9 79.9 76.0 Sm 61.0 6m 68.4 83.5 78.0 80.7 80.9 85.3 79.4 78.4 8m 73.2 79.8 79.5 80.9 81.3 89.7 82.1 79.3 9m 70.3 79.0 76.7 lOm 53.7 54.6 78.8 78.5 81.1 86.9 79.6 79.9 llm 58.3 77.8 80.9 12m 61.0 63.1 61.3 78.1 80.3 89.6 84.3 78.7 14m 69.3 64.0 76.6 8S.O 86.9 81.8 16m 64.6 70.3 69.9 78.1 83.7 83.2 78.7 20m 68.2 71.7 70.6 76.7 83.2 91.3 86.0 79.8 22m ' 82.1 90.5 88.4 24m 69.9 73.0 71.2 77.1 84.9 91.8 93.0 80.9 26m 85.0 94.8 94.0 82.6 28m 69.4 73.3 73.5 85.7 94.3 104.3 85.3 30m 70.8 72.2 75.9 88.5 100.8 101.4 92.9 31 m 71.3 78.2 78.3 87.1 102.6 103.7 98.0 104 Appendix

Table AS. Concentrations of (E,E)-DSBP in Lake Water Collected above 3 the Deepest Point of Greifensee in 1995/96 (µg/m )

depth Feb6 Mar6 Apr3 May2 May30 Jun 26 Jul 26 Aug21

Om 75.3 77.4 55.9 31.9 15.0 30.3 10.0 12.2 lm 74.7 77.0 54.4 30. l 14.7 33.4 12.6 12.6 2m 76.2 75.4 56.0 29.4 19.7 32.1 13.9 12.4 3m 30.0 34.0 33.6 16.2 13.5 4m 76.0 74.8 62. l 29.9 31.4 35.0 24.6 Sm 41.2 51.7 6m 72.9 74.6 60.8 45.2 38.9 69.2 66.7 24.6 8m 77.0 76.7 63.2 46.9 54.2 63.2 50.3 35.7 tom 73.9 77.3 65.5 56.9 50.0 51.9 47.4 41.0 12m 78.0 75.5 62.1 59.7 52.0 51. l 41.6 14m 75.7 76.1 63.5 16m 73.8 58. l 53.8 50.6 45.8 20m 76.8 60.2 54.0 48.6 47.6 24m 75.9 66.8 63.5 62.8 56.4 51.5 50.4 26m 89.4 77.4 65.8 64.5 61.6 56.3 28m 86.8 84.6 65.8 63.5 60.5 57.6 63.l 49.9 30m 85.1 84.9 65.6 64.3 58.3 56.2 48.8 48.1 31 m 87.9 66.7 62.8 55.3 50.8 49.7

depth Sep 18 Oct 16 Nov 13 Dec 13 Jan IO Feb 5 Mar4 Apr2

Om 29.7 28.1 42.5 53.3 62.8 62.5 53.0 52.1 lm 29.0 51.2 2m 29.3 29.0 40.4 54.1 62.1 50.8 49.9 3m 30.7 50.1 4m 32.6 29.7 41.0 53.8 60.0 62.0 52.8 51.0 Sm 34.9 6m 46.7 43.9 41.2 54.1 59.4 61.5 52.2 50.6 8m 61.4 48.9 41.9 54.2 59.2 62.3 54.6 50.7 9m 51.4 41.4 52.1 lOm 39.8 38.5 41.3 53.7 59.8 60.9 54.4 50.4 llm 38.7 40.7 53.4 12m 41.1 39.6 37.4 52.3 60.6 65.3 54.9 49.5 14m 41.5 39.2 51.2 61.9 61.4 55.1 16m 43.2 42.3 41.3 52.1 61.3 53.8 49.8 20m 44.9 44.7 42.3 51.2 62.5 66.5 57.5 50.2 22m 61.0 65.3 62.5 24m 47.0 45.2 43.4 52.8 65.7 69.l 67.4 49.7 26m 66.2 71.0 69.4 51.2 28m 46.4 44.7 39.2 73.0 71.0 77.8 55.1 30m 44.1 43.4 37.3 76.3 75.3 88.3 64.2 31 m 41.6 41.9 36.1 76.5 76.8 99.1 70.2 Appendix 105

Table A6. Concentrations of (E,Z)-DSBP in Lake Water Collected above 3 the Deepest Point of Greifensee in 1995/96 (µg/m )

depth Feb6 Mar6 Apr3 May2 May30 Jun 26 Jul 26 Aug21

Om 9.0 7.7 6.6 4.S 2.2a s.s 2.2a 2.Sa lm 10.3 10.2 9.0 S.3 2.9a 7.9 3.S 2.9a 2m 10.3 10.3 9.6 S.8 4.9 7.3 3.8 3.3 3m S.8 9.9 7.0 3.9 3.4 4m 10.9 10.2 8.9 S.2 9.1 7.2 9.4 Sm 13.4 19.8 6m 10.1 10.1 8.4 7.3 8.6 12.9 17.2 14.3 8m 10.S 10.8 8.6 7.S 7.7 11.3 9.7 9.1 lOm 10.7 10.4 7.8 8.S 7.8 9.2 8.9 8.S 12m 10.7 10.1 8.2 8.4 8.1 8.9 8.1 14m 10.1 9.7 7.7 16m 10.7 8.1 8.8 8.6 7.7 20m 9.9 8.0 8.7 9.0 8.7 24m 9.2 8.0 8.S 8.3 8.9 8.3 7.9 26m 11.8 10.7 8.0 8.S 8.1 8.9 28m 11.9 11.8 8.0 8.8 7.S 7.6 12.7 8.1 30m 11.7 11.7 8.S 8.8 7.4 8.7 8.2 7.3 31 m 11.9 8.1 8.6 8.S 8.2 7.7

depth Sep 18 Oct 16 Nov 13 Dec 13 Jan 10 FebS Mar4 Apr2

Om s.s S.6 6.8 7.1 7.7 8.4 6.7 7.0 lm S.6 7.0 2m 6.4 6.1 7.8 7.7 9.S 8.S 8.0 3m 6.1 7.7 4m 7.S 8.7 7.8 7.9 8.4 9.4 8.3 7.8 Sm 7.0 6m 8.S 11.S 7.8 7.8 8.3 8.9 8.8 7.9 Sm 11.4 10.3 7.9 7.9 8.S 8.2 8.6 8.3 9m 11.6 7.7 7.4 lOm 8.4 9.S 7.7 7.6 8.6 8.0 8.3 8.S 11m 9.7 7.7 7.8 12m 8.S 9.8 9.2 7.4 8.8 8.3 8.9 8.2 14m 10.0 9.3 7.8 8.S 8.3 8.8 16m 8.4 9.8 9.6 7.S 9.3 9.1 8.4 20m 8.S 10.1 9.6 8.2 8.9 9.0 9.2 8.2 22m 9.6 9.7 9.S 24m 8.7 9.9 9.6 7.7 9.S 10.0 10.3 8.6 26m 9.4 10.2 10.0 8.8 28m 8.1 9.S 8.S 10.0 10.1 12.1 9.1 30m 7.4 9.0 8.4 10.6 11.S 12.6 10.8 31 m 7.7 8.4 8.0 10.6 11.S 14.9 11.7

3 a below the limit of quantification (3 µg/m ) 106 Appendix

2 Table A7. Particles Collected in Sedimentation Traps (A = 63.6 cm ) Above the Deepest Point of Greifensee in 10 m Depth. Average Values of Two Measurements

date mass flux total C inorg. C org. C (mg) (g/m2d) (%) (%) (%)

413 - 512195 697.6 3.78 13.63 6.65 6.98 5/2- 5130195 339.6 1.91 16.57 7.15 9.42 5130- 6126195 1389.0 8.09 13.12 8.30 4.82 6126- 7/26/95 2369.1 12.42 14.56 10.05 4.50 7/26- 8/21195 1118.3 6.76 16.17 9.44 6.72 8/21 - 9/18/95 457.0 2.57 18.46 6.14 12.32 9/18 - 10/16/95 516.9 2.90 18.87 4.98 13.89 10/16 - 11113/95 369.4 2.07 18.42 8.00 10.42 11113 12113/95 184.7 0.97 17.68 4.88 12.80 12/13 - 1110196 291.9 1.64 13.33 4.45 8.87 1110- 215196 113.4 0.69 215 - 3/4/96 93.1 0.52 14.50 5.26 9.24 3/4- 4/2196 345.2 1.87 18.97 1.37 17.60

Table AS. FW A Concentrations in Particles Collected in Sedimentation 2 Traps (A == 63.6 cm ) Above the Deepest Point of Greifensee in 10 m Depth. Average Values of Two Measurements

date (E)-DAS 1 (Z)-DAS 1 (E,E)-DSBP (E,Z)-DSBP BLS (ng/g) (ng/g) (ng/g) (ng/g) (ng/g)

413 - 512195 455.2 108.4 231.8 < 2.0 6.9 512- 5130195 1697.9 458.0 500.9 3.2 4.4 5130- 6126195 240.6 331.1 148.1 < 2.0 4.1 6126- 7/26/95 143.3 30.6 101.0 < 2.0 1.8 7/26- 8/21/95 180.0 29.0 1735.1 4.5 1.4 8/21 - 9/18/95 1504.1 52.9 527.5 5.3 6.3 9/18 - 10/16/95 232.0 41.3 256.1 4.6 13.3 10/16 11113195 240.7 29.7 191.8 3.4 3.6 11/13 - 12113/95 999.9 163.7 832.8 7.7 15.0 12/13 - 1/10/96 691.9 155.3 800.5 7 .1 10.6 1110 - 215196 635.1 146.1 710.5 6.1 9.9 215 - 314196 791.8 121.1 742.3 4.6 12.7 3/4- 412196 1373.6 76.7 299.8 2.6 3.7 Appendix 107

2 Table A9. Particles Collected in Sedimentation Traps (A = 63.6 cm ) Above the Deepest Point of Greifensee in 30 m Depth. Average Values of Two Measurements

date mass flux total C inorg. C org. C (mg) (g/m2d) (%) (%) (%)

4/3 - 512195 911.9 4.94 13.11 5.99 7.12 512- 5130195 415.8 2.33 17.45 4.44 13.01 5130- 6126195 1376.9 8.02 13.67 7.80 5.87 6126- 7/26/95 1825.7 9.57 14.27 9.94 4.32 7/26- 8/21/95 803.7 4.86 16.15 9.03 7.12 8/21 - 9/18/95 383.6 2.15 18.89 6.34 12.55 9/18 - 10/16/95 330.3 1.85 19.53 5.05 14.47 10116 - 11/13/95 335.3 1.88 16.99 7.76 9.23 11113 - 12113/95 169.0 0.89 16.59 5.36 11.23 12113 - 1110196 373.7 2.10 13.62 4.43 9.19 1/10- 215196 174.0 1.05 12.64 4.10 8.54 215 - 314196 153.4 0.86 14.11 5.26 8.85 314- 412196 390.2 2.12 18.32 1.96 16.36

Table AlO. FWA Concentrations in Particles Collected in Sedimentation 2 Traps (A = 63.6 cm ) Above the Deepest Point of Greifensee in 30 m Depth. Average Values of Two Measurements

date (E)-DAS 1 (Z)-DAS 1 (E,E)-DSBP (E,Z)-DSBP BLS (ng/g) (ng/g) (ng/g) (ng/g) (ng/g)

413 - 512195 799.0 71.8 285.8 < 2.0 4.7 512- 5130195 1679.8 534.9 763.8 3.2 6.6 5130- 6126195 350.2 720.9 219.9 < 2.0 4.5 6126- 7/26/95 148.2 40.8 98.7 < 2.0 1.5 7/26- 8/21/95 258.3 36.4 1622.3 10.0 3.5 8/21 - 9/18/95 1141.5 65.2 586.9 7.4 4.6 9/18 - 10/16/95 268.1 89.5 378.2 8.3 10.0 10/16 - 11/13/95 210.5 57.9 251.4 7.4 6.0 11113 - 12113/95 861.4 151.0 840.2 19.7 10.6 12/13 - 1/10/96 656.6 151.5 851.2 9.1 10.6 1110- 215196 608.0 176.9 980.5 8.6 13.3 215 - 3/4/96 784.0 178.2 1141.8 10.0 . 14.2 3/4- 412196 1645.8 124.8 541.9 5.9 5.2 108 Appendix

Table All. FW A Concentrations in the Tributaries of Greifensee

Aa (Uster) Aabach (Monchaltorf) date DASl DSBP discharge DASl DSBP discharge 3 (µg/m3) (µg/m3) (10 m3/d) (µg/m3) (µg/m3) (103m3/d)

April 19, 1995 287 385 142.6 400 816 61.6 April 20, 1995 345 458 148.2 387 772 90.6 April 21, 1995 339 356 123.4 287 522 63.6 April 22, 1995 417 445 66.5 334 623 48.1 April 23, 1995 403 408 67.6 321 578 40.6 April 24, 1995 392 415 127.3 399 739 41.7 April 25, 1995 660 568 324.6 301 458 402.7

June 14, 1995 202 151 348.3 163 247 281.9 June 15, 1995 195 152 378.1 150 233 196.2

Sep 20, 1995 427 417 176.9 450 622 127.4 Sep 21, 1995 248 258 261.5 262 345 172.4

Dec 12, 1995 577 1137 20.6 Dec 13, 1995 526 1113 20.9

Jan 23, 1996 222 226 88.6 374 664 29.7 Jan 24, 1996 239 275 94.1 382 676 29.9

Table Al2. FW A Concentrations in River Aabach (MonchaltorO

date time DAS 1 DSBP discharge (µg/m3) (µg/m3) (103m3/d)

Dec 12, 1995 10.20 h 289 268 20.9 12.30 h 717 1435 20.9 15.30 h 594 1167 20.9 18.30 h 680 1476 20.9 21.30 h 558 1134 20.9 Dec 13, 1995 0.30 h 659 1335 20.9 3.30 h 571 1188 20.9 6.30 h 483 959 20.9 9.30 h 375 663 20.9 12.30 h 512 1108 20.9 15.30 h 572 1311 20.9 18.30 h 563 934 20.9 21.30 h 538 1209 20.9 Dec 14, 1995 0.30 h 569 1272 20.9 3.30 h 578 1276 20.9 6.30 h 503 1127 20.9 Appendix 109

Table Al3. FWA Concentrations in River Aa (Uster)

date time DASI DSBP discharge (µg/m3) (µg/m3) (103m3/d)

Dec 12, 1995 11.00 h 275 267 94.2 17.00 h 326 320 62.2 Dec 13, 1995 11.00 h 316 316 62.2

Jan 23, 1996 9.30 h 204 180 224.6 12.30 h 260 278 182.3 15.30 h 176 191 155.5 18.30 h 156 167 94.2 21.30 h 142 161 26.8 Jan 24, 1996 0.30 h 137 146 34.6 3.30 h 155 167 34.6 6.30 h 230 242 34.6 9.30 h 211 223 210.0 12.30 h 343 430 196.1 15.30 h 205 221 155.5 18.30 h 160 188 82.9 21.30 h 145 178 19.0 Jan 25, 1996 0.30 h 164 171 43.2 3.30 h 174 185 43.2 6.30 h 143 166 43.2

Table A14. FWA Concentrations in the Effluent of the Sewage Treatment Plant Uster

3 3 3 3 date DAS 1 (µg/m ) DSBP (µg/m ) discharge (10 m /d)

Apri1 19, 1995 2178 3259 14.2 Apri1 20, 1995 1960 2787 25.0 April 21, 1995 1927 2767 24.9 April 22, 1995 2078 3009 16.1 April 23, 1995 2107 3007 14.3 April 24, 1995 2191 3071 13.8 April 25, 1995 2524 3320 14.6

Sep 20, 1995 2551 3235 25.3 Sep 21, 1995 2385 2890 30.7

Dec 12, 1995 1880 2567 11.8 Dec 13, 1995 1794 2365 12.9

Jan 23, 1996 2103 2121 16.0 Jan 24, 1996 1935 2136 16.5 110 Appendix

Table A15. FW A Concentrations in the Effluents of the Sewage Treatment Plants Monchaltorf and Maur

STP Monchaltorf STPMaur date DAS 1 DSBP discharge DAS 1 DSBP discharge (µg/m3) (µg/m3) (103m3/d) (µg/m3) (µg/m3) (103m3/d)

April 19, 1995 3316 8421 1.0 2379 4938 3.4 April 20, 1995 3012 7136 1.3 1894 3897 3.1 April 21, 1995 2507 6142 0.9 1919 3352 2.4 April 22, 1995 2661 6637 0.8 2160 3633 2.0 April 23, 1995 2655 6269 0.7 2506 4152 1.7 April 24, 1995 3558 8072 0.8 2736 4675 2.1 April 25, 1995 2690 5651 1.9 2417 3861 5.6

Table A16. Epilimnion Depth, Input, and Photodegradation Rates used for the Computer Model photodegradation rates (top meter1) epilimnion depth input rawb reducedc (m) (g/d) oo-3 ct·!) (10-3d-I) period start end DASI DSBP DAS-1 DSBP DAS-1 DSBP 1 LO 2.5 129.0 174.6 32.04 178.5 23.58 114.7 2 2.5 2.5 127.5 172.7 41.25 229.8 35.74 135.1 3 2.5 2.5 239.2 312.6 33.12 184.5 27.33 152.3 4 2.5 2.5 102.2 141.0 47.36 263.9 34.71 131.3 5 2.5 2.5 87.2 122.2 40.06 223.2 31.73 155.8 6 2.5 5.5 121.3 164.9 25.68 143.1 20.47 92.5 7 5.5 4.5 97.2 134.8 20.45 113.9 16.68 92.9 8 4.5 10.0 67.4 97.5 14.00 78.0 12.50 69.7 9 10.0 25.0 91.4 127.5 5.20 29.0 4.01 22.4 10 25.0 32.6 143.5 192.8 3.24 18.l 2.28 12.7 11 32.6 25.0 79.0 112.0 6.00 33.4 4.83 26.9 12 25.0 20.0 103.6 142.8 14.90 83.0 10.24 57.1 13 20.0 30.0 91.9 128.2 25.59 142.6 17.10 63.8 a no photodegradation takes place below 1 m depth, b proportional to the photon flux; not used for the computer model, c reduced for reflection, attenuation by particles, and periodic stratification; used for the computer model Appendix 111

Table A17. Eddy Diffusion Coefficients K(z) used for the Computer 3 2 1 Model (Calculated with the Heat Budget Method, Units: 10" cm s- ) depth (m) period starr1 enda 0 - Sb 5 - 7.5 7.5 - 10 10 - 12.5 12.5 - 15 1 Apr3 May2 227.3 69.98 78.27 140.0 185.8 2 May2 May30 94.62 51.12 44.26 54.34 101.9 3 May30 Jun 26 53.00 38.78 41.51 24.70 31.26 4 Jun 26 Jul 26 25.44 3.649 2.039 11.83 25.69 5 Jul 26 Aug21 24.35 1.282 0.258 6.764 13.98 6 Aug21 Sep 18 21.40 21.40 8.144 7.685 11.02 7 Sep 18 Oct 16 5.817 5.817 5.817 2.285 7.663 8 Oct 16 Nov 13 2.841 2.841 2.841 2.841 11.45 9 Nov 13 Dec 13 10.65 10.65 10.65 10.65 10.65 10 Dec 13 Jan 10 363.2 363.2 363.2 363.2 363.2 11 Jan 10 Feb5 3868 3868 3868 3868 3868 12 Feb5 Mar4 243.7 243.7 314.6 314.6 314.6 13 Mar4 Apr2 243.7 243.7 314.6 314.6 314.6

depth (m) period 15 - 17.5 17.5 - 20 20 - 22.5 22.5 - 25 25 - 27.5 27.5 - 30 30 - 32.6 1 270.4 210.2 149.9 131.9 113.9 113.9 113.9 2 93.94 111.2 78.20 78.37 78.37 78.37 78.37 3 53.08 108.8 73.28 76.42 77.29 47.32 47.32 4 39.91 90.17 55.65 84.15 55.05 55.05 55.05 5 32.53 37.79 56.15 28.13 23.39 23.39 23.39 6 27.02 22.85 54.88 19.59 21.72 21.72 21.72 7 13.67 22.87 27.37 33.98 37.27 15.21 15.21 8 17.99 28.82 20.81 22.71 22.71 22.71 22.71 9 10.65 10.65 10.65 10.65 10.65 10.65 10.65 10 363.2 363.2 363.2 363.2 363.2 363.2 363.2 11 2331 1391 451.4 466.4 292.5 34.40 34.40 12 459.6 320.1 180.7 147.1 94.19 31.94 31.94 13 459.6 320.1 180.7 147.1 94.19 31.94 31.94 a in the years 1995/96, b the epilimnion is assumed to be completely mixed 112 Appendix

Table A18. Locations of the Sediment Cores Collected in Greifensee

core coordinates lake depth (m) distance from tributaries (km) Aa (Uster) Aabach (Monchaltorf)

A 692/050//247/100 14 3.1 5.7 B 69214501 /247 /350 12 2.9 5.7 c 693/025//246/800 15 2.2 5.0 D 6921750112461500 29 2.2 4.8 E 692/300//245/875 25 2.3 4.6 F 6921750//245/500 27 1.8 4.0 G 693/ 17 51 /2451725 30 1.4 4.0 H 693/725//246/075 18 1.1 4.0 I 6941350//245/150 19 0.2 2.9 J 6941100/ /2441850 30 0.5 2.7 K 6931775//2441500 21 1.0 2.6 L 694/225//244/ 175 23 1.1 2.1 M 694/600//243/600 15 1.6 1.4 N 6941700//242/575 9 2.6 0.6 0 6951200112421625 10 2.6 0.3 p 6951700112421925 10 2.6 0.5 freeze core 6931500//2441900 32 1.1 3.1 Appendix 113

Table A19. FW A Concentrations in Sediment Cores Collected in the Benthos of Greifensee on February 27, 1995.

core depth (E)-DAS 1 (Z)-DAS 1 (E,Z)-DSBP (E,E)-DSBP BLS (cm) (ng/g) (ng/g) (ng/g) (ng/g) (ng/g)

A 0-5 464.6 179.9 9.2 419.5 19.0 5 - 10 711.8 167.5 8.2 146.4 7.1 10- 15 56.2 23.7 11.5 0.0 0.0 15 - 20 0.0 0.0 < 2.0 0.0 0.0

B 0-5 376.9 95.7 7.1 427.4 17.7 5 - 10 614.8 147.8 6.2 158.4 6.2 10- 15 48.6 19.5 11.2 < 2.0 0.0 15 -20 0.0 0.0 < 2.0 0.0 0.0

c 0-5 509.0 159.1 7.7 433.9 27.8 5 - 10 781.4 162.8 10.8 150.8 4.2 10 - 15 < IO.O 14.0 6.9 3.5 0.0 15 -20 I0.8 < 10.0 < 2.0 5.4 0.0

D 0-5 525.6 221.5 8.6 425.9 23.4 5 - IO 987.7 196.7 6.5 220.6 14.0 10 - 15 315.2 55.3 17.2 5.9 < 2.0 15 - 20 < 10.0 13.3 < 2.0 3.4 0.0 20-25 < 10.0 10.6 < 2.0 < 2.0 0.0

E 0-5 532.2 207.1 7.9 421.0 20.2 5 - 10 799.8 176.9 9.4 160.4 3.8 10 - 15 59.6 30.2 17.9 0.0 0.0 15 -20 0.0 < 10.0 < 2.0 2.1 0.0

F 0-5 485.5 184.4 6.8 343.5 17.1 5 - IO 681.6 163.1 6.2 182.3 5.6 10- 15 279.8 52.7 15.0 3.5 0.0 15 - 20 13.9 < 10.0 < 2.0 < 2.0 0.0 20-25 0.0 0.0 < 2.0 0.0 0.0

G 0-5 585.6 195.6 10.6 499.3 25.6 5 -10 1022.1 169.2 9.2 239.1 13.7 10 - 15 204.3 27.0 23.3 5.2 0.0 15 -20 14.2 < 10.0 < 2.0 5.9 0.0

H 0-5 737.4 184.9 10.4 643.7 47.2 5 - 10 1337.l 153.4 9.1 356.5 20.9 10 - 15 650.5 75.5 26.2 12.0 0.0 15 - 20 15.1 < 10.0 2.2 5.9 0.0 20-25 15.7 17.0 < 2.0 3.2 0.0 114 Appendix

Table A19. Continued

core depth (E)-DAS 1 (Z)-DAS 1 (E,Z)-DSBP (E,E)-DSBP BLS (cm) (ng/g) (ng/g) (ng/g) (ng/g) (ng/g)

I 0-5 1216.2 194.5 14.2 1129.2 77.5 5 - 10 1602.1 190.9 9.6 714.9 110.4 10- 15 3382.4 217.5 21.3 573.5 17.6 15 - 20 1823.6 157.7 41.5 31.1 3.7 20-25 163.4 < 10.0 42.9 5.9 0.0 25-30 0.0 23.0 < 2.0 0.0 0.0 J 0-5 704.2 232.2 11.6 573.6 32.1 5 - 10 1103.1 169.4 7.1 329.5 18.1 10-15 747.0 147.8 31.7 20.0 < 2.0 15 - 20 10.9 < 10.0 < 2.0 2.9 0.0 K 0-5 521.6 188.7 8.9 485.0 25.2 5 - 10 858.8 166.0 9.9 180.8 8.1 10 - 15 148.2 31.0 17.5 < 2.0 0.0 15 -20 0.0 < 10.0 < 2.0 2.2 0.0 L 0-5 583.0 204.7 8.2 453.3 24.6 5 - 10 760.9 101.4 20.4 135.6 < 2.0 10- 15 11.7 12.8 5.7 < 2.0 0.0 15 - 20 0.0 < 10.0 < 2.0 < 2.0 0.0 M 0-5 523.6 174.0 7.2 496.2 23.8 5 - 10 658.0 158.2 11.8 95.1 0.0 10 - 15 24.9 13.6 5.3 < 2.0 0.0 15 - 20 0.0 < 10.0 < 2.0 < 2.0 0.0 N 0-5 332.5 73.2 7.6 308.1 12.9 5 - 10 365.3 69.5 5.6 215.8 13.3 10 - 15 65.2 26.2 10.2 7.8 0.0 15 -20 0.0 0.0 < 2.0 0.0 0.0 0 0-5 414.5 76.l 7.6 615.1 40.7 5 - 10 617.2 101.8 7.2 543.5 99.5 10- 15 1053.1 147.3 6.1 426.7 86.5 15-20 737.5 63.4 2.4 195.5 42.2 20-25 844.4 66.8 3.9 230.4 57.8 25 -30 674.8 55.2 2.5 189.l 40.S 30-35 1568.4 81.7 6.4 288.5 39.8 35-40 1904.1 149.9 7.0 332.5 4.0 p 0-5 386.3 89.4 6.9 427.9 24.0 5-10 445.0 80.0 6.9 318.0 29.8 10- 15 488.3 70.0 5.2 267.8 27.4 15 -20 413.7 73.1 8.2 108.1 8.0 20-25 26.8 16.8 9.1 6.3 < 2.0 25-30 0.0 0.0 < 2.0 6.1 0.0 Appendix 115

Table A20. Sediment Core Collected by Means of a Freeze Core in Greifensee on October 24, 1995.

year of lower limit thickness water content sedimentation deposition (cm) (cm) (%mass) (g/cm2·2y)

1993/94 1.46 1.46 89.1 0.246 1991/92 2.38 0.92 80.l 0.174 1989/90 3.36 0.97 80.2 0.233 1987/88 4.29 0.93 78.2 0.265 1985/86 5.18 0.90 76.4 0.205 1983/84 5.82 0.63 73.2 0.214 1981182 6.71 0.89 73.5 0.299 1979/80 7.49 0.78 69.3 0.287 1977178 8.20 0.71 69.1 0.250 1975176 8.90 0.70 68.9 0.273 1973/74 9.76 0.86 69.2 0.275 1971/72 10.45 0.69 71.6 0.231 1969170 11.25 0.80 72.8 0.255 1967/68 11.94 0.69 76.5 0.185 1965/66 12.84 0.90 72.6 0.264 1963/64 13.49 0.66 71.3 0.210 1961/62 14.03 0.54 69.1 0.220 1959/60 14.70 0.67 66.1 0.211

Table A21. FWA Concentrations in a Sediment Core Collected by Means of a Freeze Core in Greifensee on October 24, 1995.

year of (E)-DAS 1 (Z)-DAS 1 (E,Z)-DSBP (E,E)-DSBP BLS deposition (ng/g) (ng/g) (ng/g) (ng/g) (ng/g)

1993/94 473.7 200.9 13.1 775.4 9.7 1991/92 429.0 235.5 14.0 695.3 20.4 1989/90 456.6 214.7 7.0 406.3 34.4 1987/88 536.4 160.2 7.7 328.7 27.7 1985/86 651.5 127.1 4.6 200.7 20.5 1983/84 484.3 112.8 3.9 191.6 20.3 1981/82 895.4 134.4 8.0 285.8 27.0 1979/80 887.6 128.0 7.0 235.0 2.4 1977178 911.6 169.4 9.0 318.6 0.0 1975176 748.5 151.6 8.9 200.1 0.0 1973174 933.5 179.3 11.1 50.3 0.0 1971/72 1197.9 165.9 23.6 6.8 0.0 1969170 707.2 119.4 42.4 < 2.0 0.0 1967/68 231.8 64.8 59.5 0.0 < 2.0 1965/66 16.4 27.1 29.6 0.0 0.0 1963/64 0.0 < 10.0 16.0 0.0 0.0 1961162 0.0 0.0 6.5 0.0 0.0 1959/60 0.0 0.0 < 2.0 0.0 0.0 Danke ..

. . Walter Giger fiir die Leitung meiner Doktorarbeit, fur die Betreuung, das Vertrauen, die Diskussionen und Anregungen, und vor allem fur das Klima von Toleranz und Freundschaft in der Arbeitsgruppe.

. . Markus Ulrich und Bob Eganhouse fur die Ubemahme der Koreferate, fur die Diskussionen, die Ratschlage und fiir die Hilfe beim Publi- zieren.

. . Thomas Poiger und Hans Kramer fiir die gute Zusammenarbeit 1m Rahmen des TWTWT (The Whiter Than White Trilogy) .

.. Richard Illi, Franz Gunter Kari, Alfred Liick, Judith Maag, Rene Rei- ser, Bruno Ribi, Caroline Stengel und Alois Zwyssig fur ihre Gesell- schaft und tatkraftige Mitarbeit bei den Probenahmen sowie fiir die Hilfe bei Analysen im Labor.

.. Alfredo Alder, Beat Altenbach, Marianne Balmer, Christa Bilrgisser, Thomas Jabusch, Carlo Kanz, Katja Knauer, Eva Molnar, Sonja Rie- diker, Christian Schaffner, David Scheidegger, Blanca Schneider, Christine Schneider, Diana Soldo und Marc Suter fiir die gute Stim- mung in den Labors und Buros, fiir die vielen Griinde, etwas zu f eiem sowie die Umsetzung des letzteren .

.. Juerg Bloesch, Hans Rudolf Burgi, Peter Krebs, Stephan Muller, Peter Reichert, Michael Sturm und Alfred Wuest fiir die Beratungen.

. . Achim Albrecht, Michael Berg, Silvio Canonica, Maureen Clayton, Gerrit Goudsmit, Georg Henseler, Whitney King und Jiirg Zobrist fiir das Korrigieren meiner Papers .

. . Jurgen Kaschig und Peter Richner fiir optische Aufueller, Insider- Informationen und fur die Mithilfe bei der Geldbeschaffung .

.. den Herren Fuhrer, Kagi und Lehnherr fur Proben und Daten aus ihren Klaranlagen .

.. dem Stipendienfonds der Basler chemischen Industrie fiir die finanzielle Unterstiitzung. Curriculum vitae

Jean-Marc Alain Stoll

born on November 9, 1966 in Vevey, Switzerland

1972 - 1979 Primary school in Baltimore (USA) and Zurich

1979 - 1985 High school in Zurich and Wetzikon, Matura B

1986 - 1992 University of Zurich

1989 - 1990 High school teacher (Chemistry), Kantonsschule Zurcher Oberland, Wetzikon

1992 Diploma in Chemistry (Minor: Computer sciences)

1992 - 1993 Research assistant, Paul Scherrer Institute (PSI), Villigen

1993 - 1994 Travels through Tanzania (East Africa) and USA

1994 - 1997 Doctoral thesis, Swiss Federal Institute for Environmental Science and Technology (EA WAG), Diibendorf, and Swiss Federal Institute of Technology (ETH), Zurich