Analysis of Organophosphorus Flame Retardants in Sewage Treatment Plant Sludge Optimization of an analytical detection method for TPP, RDP and BDP

(1)

University of Amsterdam

Institute for Biodiversity and Ecosystem Dynamics

Analysis of Organophosphorus Flame

Retardants in Sewage Treatment Plant Sludge

Optimization of an analytical detection method for TPP, RDP and BDP

BSc. Thesis Chemistry

July 1, 2015

Author

Yasmine Keehnen

Student ID

10069445

Supervisors

(2)

POPULAR SCIENTIFIC ABSTRACT

Flame retardants, as already suggested by their name, are components added to- or build into materials, to prevent the formation or the spread of fire. The components are incorporated into all sorts of products, for instance computers, textiles and furniture. There are four different groups of flame retardants, with brominated flame retardants (BFRs) used the most

commercially. The presence of BFRs is even confirmed in human breast milk and fish. Unfortunately, some of these BFRs are toxic and harmful for man and environment, which led to the bans and restrictions of the use of these compounds. These restrictions required

alternatives, which led to the marketing of organophosphorus flame retardants (OPFRs). However, information on environmental behavior and toxicity is very limited. In this study, a method for the detection of three OPFRs is developed: triphenylphosphate (TPP), resoricinol bis(diphenyl phosphate) (RDP) and bisphenol A bis(diphenyl phosphate (BDP). With this method, three types of samples from a sewage water treatment plant (water influent, effluent and sludge) were tested for the presence of the targeted compounds. The presence of these compounds was indeed confirmed for all sample types, but to obtain more significant data, the experiments should be repeated.

(3)

ABSTRACT

Flame retardants (FRs) are used to prevent combustion or delay the spread of fire. These chemicals are added to- or build in products composed of polymers with a flammable nature, for instance computers and textiles. FRs can be classified into four groups: inorganic-, organohalogenated-, organophosphate- and nitrogen based flame retardants.

Organohalogenated flame retardants, especially brominated flame retardants (BFRs), are widely applied due to their low costs and effectiveness. However, the negative effects on the environment of some BFRs led to bans and restrictions. As a result, organophosphorus flame retardants (OPFRs) are increasingly used as an alternative although information on their toxicity and environmental behavior is minimal. The focus of this study is on a trio of compounds: triphenylphosphate (TPP), resoricinol bis(diphenyl phosphate) (RDP) and bisphenol A bis(diphenyl phosphate (BDP). The aim of this study was to develop and optimize an analytical methodology for the targeted compounds, to determine their presence in water influent, effluent and sludge from a sewage water treatment plant. A solid phase extraction method was established and analysis was carried out by LC-MS/MS. Extraction had proven best under acidic conditions (pH 2) with hydrophilic-lipophilic-balanced (HLB) sorbent material, 200 mL sample volume and triple laboratory material extraction with hexane. Overall recoveries for TPP, BDP and RDP were 62%, 58% and 73% respectively. The analytes of interest, all within de limit of quantification (LOQ), were detected in the water in- and effluent with TPP significantly most present. The extraction method proved not yet suitable for the water treatment plant sludge, nevertheless the presence of all three

compounds was also confirmed in one sludge sample. More research is recommended to repeat the methodology on water in- and effluent, and to develop a more suitable extraction method for BDP, RDP and the sludge to obtain more conclusive results.

(4)

1. Introduction

1.1 Flame Retardants

To prevent the combustion and delay the spread of fire upon ignition, flame retardants (FRs) are added to- or build in products with a flammable nature. These products, for instance furniture, computers and textiles, are composed of petroleum-based polymers. Advances in polymer science over the last 50 years has led to the presence of anthropogenic polymers in nearly everything surrounding us. The concept of flame retardants however, is not from recent years; the Egyptians (450 BC) and Romans (200 BC) were known to use alum and vinegar to decrease the combustibility of wood. 1

In present time, the flame retardants in use can be classified into four groups: inorganic-, organohalogenated-, organophosphate- and nitrogen based flame retardants. 2_{These FRs act at}

different stages in the fire cycle, depending on their chemical makeup, in order to stop or postpone fire. 3_{For instance, nitrogen based FRs stop the decomposition process and hinder}

the release of flammable gasses by facilitating the formation of stable molecular compounds. These FRs also release nitrogen gases that prevent combustion. Organophosphate FRs, in the presence of a heat source, release phosphoric acid. This causes the material to form a thick layer of carbon, better known as charring. The char prevents the release of flammable gasses and blocks the fuel from the flame to form a barrier between the material and the heat source. Phosphorus based FRs can also act in the vapor phase trough flame poisoning. 4

Halogens on the other hand function as captors of free radicals, highly oxidizing agents essential for the flame to aggrandize, that are produced during the combustion process.

1_{Due to its low costs, the ability to be added to materials without altering their properties and}

effectiveness at preventing flame formation as well as retaining the fire, over 75 different reactive and additive brominated FRs are applied commercially. 5

1.2 Brominated Flame Retardants

According to the European Flame Retardants Association (EFRA), the total consumption of FRs in Europe in 2006 was 465,000 tons, of which 10% were brominated FRs. 6_{The majority}

of the BFR market is composed of poly brominated diphenyl ethers (PBDEs), and tetrabromobisphenol A (TBBP-A), 67,440 and 119,600 metric tons worldwide in 2001 respectively. 5_{Data on toxicity and exposure of these BFRs and of biphenyls (PBBs) and}

hexabromocyclododecanes (HBCDs), despite much lower production volumes, is most available. These BFRs are all highly persistent and can accumulate in the environment. 7,8

The effectiveness of BFRs is positive in preventing fire, but negative towards health and environment. Reactive brominated FRs are covalently bonded to the polymer matrix. Additive brominated FRs are not covalently bonded but added to the polymer and therefore, these flame retardants may easily be discharged from the products into the environment. Due to resistance to environmental degradation and hydrophobicity their presence in the

environment is confirmed by various studies.

Toxic effects and amounts on man and wildlife vary among and within BFR groups. For instance, the presence of PBDEs has been confirmed in the blood of Swedish personnel of a plant for dismantling electronics from airborne uptake. 8-11_{However, most BFRs end up in}

adipose tissue and other parts of the human body via contaminated foods of animal origin, as much as 98%, with an average dietary uptake of 1.4-1.2 ng/kg body weight/day. 12,13

(6)

endocrine disruption. 8,10_{Also, the presence of PBDEs has been determined in human breast}

milk by research in multiple countries; concentrations showed a strong time dependent increase, with a doubling in levels within five years. 1,8-11,13,14

Besides abiotic and nutritional presence, quantities of BFRs are found in our

environment: levels of PBDEs and derivatives in sediments from The Baltic Sea, rivers in the UK and Belgium and The Wadden Sea, in sewage treatment plant sludge samples from Germany, Sweden, The Netherlands and biota in many different countries. 5,11,15-18

1.3 Organophosphorus Flame Retardants

Restrictions and limitations were introduced within Europe and the United States because of the aforementioned negative impact of some BFRs. 7,13_{For instance, the use and import of}

products containing over 0.1% penta- or octobromo diphenylethers are banned from the European Union since 2004. 19_{Organophosphorus based FRs (OPFRs) are increasingly used}

as an alternative; in Western Europe alone ca. 91,000 tons/year in 2006, even though the information on toxicity and environmental behavior is minimal. 20_{Hence, it is of great}

importance to assess whether these halogen free FRs are a suitable and safe alternative. OPFRs are considered suitable alternatives for BFRs if they are less persistent, less bio accumulative and less toxic to humans and the environment.

The focus of this study is on a trio of compounds, as seen in figure 1:

triphenylphosphate (TPP), resoricinol bis(diphenyl phosphate) (RDP) and bisphenol A bis(diphenyl phosphate) (BDP). The additive organophosphate flame retardant TPP is solely active in the gas phase. During thermal degradation, phosphoric acid is formed, which reacts further to form pyro phosphoric acid. This acid acts as a heat transfer barrier in the condensed phase. The primary mechanism of RDP is the solid phase mechanism but in addition, a gas phase mechanism is assumed. 6_{BDP is active in the condensed- and gas phase, oppose to most}

OPFRs this compound does not cause charring.

Some chemical properties of these targeted compounds are listed in table 1. The solubility decreases by increasing molar mass. The log KOC is the soil adsorption coefficient and the Log

KOW gives a lipophilic-hydrophilic value measured with octanol versus water. To some extent,

this KOW value is a coefficient to determine the bioaccumulation. Compounds with higher

(7)

boiling points are considered better alternatives; they are less volatile and therefore less likely to be released into the environment. Contrariwise, compounds with higher boiling points accumulate more readily.

Table 1. Physiochemical properties of TPP, BDP and RDP, values are indicatory. 6

Compoun d Melting point [°C] Boiling point [°C] Flash point [°C]

Log KOW Log KOC Solubility

[mg/L]

TPP 49 370 220 4.59 3.72 1.9

BDP -19.83 680 377 4.5 4.53 0.42

RDP - 587 322 7.41 4.63 1.1*10-4

1.4 Environmental occurrence and toxicity

Similar to some BFRs, OPFRs are added to the polymer and as a result, they can easily diffuse into the surrounding environment to be distributed further through water and air. The significant presence of OPFRs is identified in man, wildlife, and the in- and outdoor

environment.

Various studies established traces of OPFRs in sediment, water, air and dust. 6,20-22_High

concentrations of Tris(chloro-isopropyl)phosphate (TCPP) were found in sewage treatment plant sludge, influent and effluent from twenty plants in Germany, with mean values of 5100 ng/g dry weight, 520 ng/L and 380 ng/L respectively. 23

Nonetheless, information on the environmental distribution of TPP, RDP and BDP is very limited. A research by Rodil et al. of aqueous samples and sewage sludge revealed no significant presence of RDP and BDP. 24_{On the other hand, TPP was found in air, waste- and}

surface water in Germany, Sweden and Japan. An average concentration of TPP in settled indoor dust is 11.4 μg/g, and an average of 10 ng/m3_{is found in indoor air.} 20

1.5 Goal

To determine whether RDP, BDP and TPP can be considered suitable alternatives for BFRs more information about these compounds on persistence, toxicity and bioaccumulation is necessary. Further review of the previous conducted analytical detection and quantification method is necessary. In a previous study by Helmus in 2012, based on experimental work by Rodil et al., a method was drafted to obtain the sample with Solid Phase Extraction (SPE) and quantification with LC-MS/MS.

The aim of this study is to investigate the presence of RDP, TPP and BDP in sewage treatment plant sludge samples. By analyzing the sludge, adsorption, persistence and

bioaccumulation can be identified. Furthermore, a suitable analytical detection method will be developed by optimization of the extraction method by Helmus. The best suitable SPE

(8)

2. Methodology

Firstly, in this study, a method was developed for extraction of the three targeted compounds, with stock solutions with fixed concentrations. Different solid phase extraction cartridges were tested based on- and in combination with the methodology developed by Helmus. 2

Different sample volumes were tested to determine possible break through. Not only was the SPE method optimized, adsorption to laboratory materials was evaluated, since strong adsorption may be expected due to the relatively high KOW values (table 1).

Once the methodology was optimized for SPE and adsorption to laboratory materials was minimized, it was applied on sludge samples and water influent and effluent, gathered from a sewage treatment plant.

2.1 Chemicals and Materials

A stock solution containing a mix of TPP (99%, Sigma Aldrich, 115-86-6), BDP (≥80%, [21], ICL, 5945-33-5) and RDP (≥80%, [21], ICL, 125997-21-9) in acetonitrile was used (200 mg/L). To function as internal standard (IS), a stock solution of deuterated TPP (dTPP, 98%, Sigma-Aldrich, 1079-02-3) in acetonitrile was used (500 mg/L).

Solvents were obtained from Biosolve (Valkenswaard, the Netherlands) and were ULC/MS grade (acetonitrile, methanol, ultra pure water) or specified as analytical reagent (acetone). Moreover, methanol, acetone, formic acid (Biosolve, 99% purity, ULC/MS grade), hexane (Glass distilled grade, Rathburn, Scotland) ammonium acetate (99.999% purity, Sigma-Aldrich, Zwijndrecht, the Netherlands) and acetic acid (100% purity, Merck, Darmstadt, Germany), were used.

To collect the water and sludge samples and fit larger sample volumes for solid phase extraction, 200 mL polypropylene bottles were used. Solid phase extraction elutions were collected in 15 mL plastic tubes. Glass vials (1-2 mL, Bester, Amstelveen, the Netherlands) were used for analysis with LC-MS/MS.

2.2 Sludge sample preparation

Ten sludge samples of approximately 200 mL were gathered from a sewage treatment plant. In order to form a suitable sample for solid phase extraction, these samples needed to be threated. In attempt to obtain suitable SPE samples, they were centrifuged for two hours at 2000 rpm and afterwards, the water was discarded. The sludge was then freeze-dried over night to remove any remaining water. dTPP was added as internal standard.

2.3 Water sample preparation

In addition to the sludge samples, ten water- influent and effluent samples of approximately 200 mL were collected from the water treatment plant. These twenty samples were acidified (pH 2) with formic acid to prevent any possible degradation. Eight samples, four influent and four effluent, were then centrifuged for two hours to accelerate sedimentation. Prior to the centrifugation, two samples each were spiked with the OPFR stock solution to determine analyte recoveries. Solid phase extraction was then preformed directly from the bottles, followed by hexane extraction of the bottle and solid residue. The hexane was evaporated and the collected elution fraction was added to the tube for redissolvement.

(9)

Also, the water samples were all spiked with the internal standard dTPP. Afterwards, one more of the water influent and effluent samples was spiked with the OPFR mix, one sample of each remained unspiked (see figure 2).

Figure 2. Treatment of four water influent (left) and four water effluent (right) samples with dTPP (IS) and the OPFR stock solution, before and after solid phase extraction.

2.4 Solid Phase Extraction

The extraction method is based on research by Helmus (2012), who developed his method based on research by Rodil et al. (2005). Two columns that proved best in solid phase extraction by Helmus (2012) were retested (table 2). Recoveries were highest under acidic conditions; therefore the extraction will be preformed at pH 2.

The separation in the first column was based on hydrophilic-lipophilic balance. The sorbent in the second column consists of strong cation exchangers and separation was caused by different affinity with the ion exchanger. From prior research it appeared that the mixed mode cation exchange cartridge (MCX) under acidic conditions retained the highest extraction recovery. 2_{This is noteworthy since the separation in the column is based on}

particle charge and the neutral compounds are not expected to undergo pH dependent interactions.

To begin the extraction, the column was conditioned with 5 mL methanol, to wet it thoroughly. To optimize analyte retention, the column was equilibrated with 5 mL of the 2% aqueous formic acid (FA) solution. Next, the column was loaded with the sample and washed with the acidic solution of water and 2% FA. After washing, the column was dried under vacuum for 30 minutes. Finally, the column was eluted with methanol under vacuum to collect the sample. The extraction was applied at a flow rate of approximately one drop per second. The elution volume consisted of 13 mL.

For sample volumes up to 25 mL, addition to the column was by pipet from an glass Erlenmeyer flask. Then, the loading volume was scaled up in a stepwise manner from 25 mL, 50 mL, 75 ml, 125 mL to 200 mL to obtain a higher concentration in the extract. Samples were prepared and stored in 200 mL polypropylene bottles. Sample workup of the larger volumes was realized by connecting the cartridge to the polypropylene bottles by tube. In case

4 Water influent 4 Water effluent Approx. 200 mL 4 Water influent 4 Water effluent Approx. 200 mL - dTPP, 0.5 g

- Spiked before SPE with OPFR mix, 0.2 g

- dTPP, 0.5 g

- dTPP 0.5 g

- Spiked after SPE with OPFR mix, 0.2 g

- dTPP 0.5 g

- Spiked after SPE with OPFR mix, 0.2 g

- dTPP 0.5 g - Unspiked

(10)

of break trough, when saturation of the sorbent is reached, a larger cartridge would have been necessary. Fortunately, no break trough was encountered in this study.

Table 2. Columns used for solid phase extraction. HLB: hydrophilic-lypophilic-balanced, MCX: mixed mode cation exchange.

Manufacturer Column Sorbent (mg) Material Interactions

Waters Oasis HLB 150 HLB vdW, polar

Waters Oasis MCX 150 MCX vdW, polar, SCX

2.5 Extraction of laboratory materials

The extraction of the polypropylene bottles was preformed twice with 15 mL solvent to test for hexane/acetone ratios, and three times with 10 mL hexane after SPE. Between and after the extractions, the organic solvents were evaporated under nitrogen at a temperature of 45 °C. After evaporation, the analytes were redissolved in water/methanol. Furthermore, the polypropylene bottles were shaken with the organic rinsing solvents and sonicated, to dissolve as much analyte as possible.

2.6 LC-MS/MS

To analyze the obtained samples, reversed phase chromatography was used in this study as well as a tandem mass spectrometer to determine ion masses.

Before analysis, sample material was transferred to 2 mL glass vials. Chromatographic separation was be preformed at 0.25 mL/min, with an apolar C18 stationary phase, at 45 °C

(Luna C18(2), 3 m, 100 Å, 150x3.0 mm ID, Phenomenex, Tottance, CA, USA). Two solutions, eluens A and B, were used to vary the proportion of the water- and organic phase. The first solution A, consisted of 50% methanol, 50% semi-boiled water, 0.016% acetic acid and 5.0 mmol ammonium acetate. The second solution (B) consisted of 100% methanol, 0.016% acetic acid and 5.0 mmol ammonium acetate. The solutions A and B were filtered (Glass Fibre, GF/F, Whatman) and degassed by sonification before use.

For the MS/MS, transitions were applied as listed in table 3. Corresponding fragmentation of the molecular cations are found in figure 3. Analytes were detected by electrospray ionization tandem mass spectrometry operated in the positive ion, with nitrogen as ionisation- and collision gas.

Table 3. Transitions for tandem MS.

Compound Transition Proposed product ion

TPP Transition 1 327.000  215.000 [(PhO)2PO-H2O]+ TPP Transition 2 327.000  153.000 [C12H9]+ BDP Transition 1 693.000  367.000 [(M+H) – OP(OPh)3]+ BDP Transition 2 693.000  327.000 [HOP(OPh)3]+ RDP Transition 1 575.000  419.000 [(M+H) – PO3Ph]+ RDP Transition 2 575.000  481.000 [(M+H) –C6H5OH]+ dTPP Transition 1 342.000  243.000 dTPP Transition 2 342.000  164.000

(11)

Figure 3. Proposed fragmentation pathways of molecular cations of (a) TPP, (b) RDP and (c) BDP, under tandem MS conditions. 24

Inaccurate sample concentrations may be registered as a result of matrix effects. When a component co-elutes with the analyte of interest leading to a decrease (suppression) or increase (enhancement) in ionization efficiency, with respect to- and based on the chemical properties of the analyte, these effects are encountered. 25_{To determine the presence of a}

matrix effect, spiked sample recoveries were compared to samples with no matrix. A set of eight calibration standards was prepared in acetonitrile, with a wide concentration range; to maintain linearity for the low concentrated as well as the highly concentrated samples (table 4).

Table 4. Composition of calibration standards.

Calibration level

OPFR standard concentration (g/L) OPFR dTPP (IS) 0 0 500 1 1 500 2 10 500 3 25 500 4 50 500 5 100 500 6 500 500 7 1000 500

(12)

2.7 Quantification

All targeted compounds were quantified with external calibration. A linear regression model was established individually for all transitions. Deviating measurements were removed manually and the model was forced through zero. The linearity of the calibration model (R2₎

was at least 0.99 for all transitions.

To correct for analyte losses, dTPP was used as internal standard since this compound is similar in structure to TPP, and to a certain degree to RDP and BDP. Through the addition of fixed concentrations of IS to the calibration standards (see table 4) and analyte samples, the analyte concentration was then calculated from the IS concentration.

In addition to the internal standard, control samples were used in solid phase

extraction experiments to calculate relative recoveries. The control sets were constructed from the OPFR stock solution and methanol (50% and 75%) and acidified with 2% formic acid. Because the controls were not extracted by solid phase extraction, recoveries were corrected for analyte losses trough SPE.

To determine the quality of the results, samples were spiked with an OPFR matrix solution. This solution containing the three targeted compounds was added in fixed concentrations to the samples to calculate analyte recoveries.

Reported analyte recoveries and concentrations were derived by averaging the results of both transitions, for each compound. Furthermore, the limit of quantification was based on the concentrations of the calibration standards, as seen in table 4.

(13)

3. Results and Discussion

3.1 Method Optimization

3.1.1 SPE - Cartridges

For the solid phase extraction, two types of columns were tested in triplicate; the Waters Oasis HLB 150 and Waters Oasis MCX 150, as mentioned in the methodology (2.3) and seen in table 2. Four different fractions were collected and analysed with LC-MS/MS. The

recoveries of all fractions combined represent an overview of analyte distribution after solid phase extraction. The analyte concentrations were normalized with respect to their sample volume and compared to a set of control samples to calculate relative recoveries, as described under quantification (2.7). As a result of over correction for analyte loss, with recoveries well over 100%, the internal standard was not used to obtain analyte recovery data.

Sufficient total analyte recoveries suggest an accurate characterization of the distribution of analytes (see figure 2). For all three targeted compounds, the Waters Oasis HLB 150 column provided a higher yield compared to the Waters Oasis MCX 150 column, around 80% and 50% respectively. Since the three compounds are neutral of charge, no stronger retention was expected for MCX due to lack of affinity with its strong cationic exchanging sorbent material. On the other hand, it is noteworthy that the performance of the HLB columns improved under acidified conditions, as suggested by Helmus 2_{, remarkably}

since these compounds are not expected to undergo pH dependent interactions.

The second collected sample, the wash, indicated no presence of analytes. A small amount of BDP (4.4 %) was found in the loading sample. From the significant presence of analytes in the second elution it appeared that a larger extraction volume, compared to the first (6 mL) and second elution (6 mL) combined, could provide higher recoveries for all compounds. Out of all three compounds, the retention of BDP after the first extraction was somewhat lower. However a relative large quantity of BDP was collected in the second elution. The necessity of the larger extraction volume could indicate that BDP adsorbs most to the column, which can be explained by its high KOW value and low solubility.

0 10 20 30 40 50 60 70 80 90 100

Analyte recoveries for tested cartridges

Elution 2 Elution 1 Wash % R ec ov er y

(14)

Figure 4. Average recovery percentages of the analytes, normalized with respect to controls, in four collected fractions after solid phase extraction for the Waters Oasis MCX 150 (left) and Waters Oasis HLB 150 (right) cartridges.

3.1.2 SPE - Increase of sample volumes

In order to test the column materials, a sample loading volume of 10 mL was used.

Nonetheless, to develop an SPE method that is of a wider general application and useful for samples with low concentrations, five different loading volumes (25 mL, 50 mL, 75 mL, 125 mL and 200 mL) were examined in duplicate. All volumes were loaded from polypropylene bottles through connecting tubes, under vacuum. One volume (25 mL) was also added to the column by pipet from a glass Erlenmeyer flask to test for possible analyte losses in the connecting tubes during the loading of the sample. The extraction volume was increased to 18 mL to promote analyte elution. Extraction concentrations were normalized with respect to the original sample volumes and recoveries were determined with control samples.

From the graph in figure 7, appendix A it appears that analyte recoveries of samples inserted to the column by pipet are approximately 10% higher compared to these analytes loaded by connecting tube. For both methods, recoveries of TPP and BDP are quite similar but deviate from RDP by more than 10%. Recoveries of the 50 mL and 75 mL sample volumes are comparable with yields of 15% for TPP and approximately 25% - 28% for BDP and RDP. Recoveries corresponding to the 200 mL are quite low: 13% for TPP and 23% for BDP and RDP. Out of all five tested sample volumes, the best analyte recoveries were reached with a 125 mL loading volume ranging from 18% (TPP) to 32% (RDP).

The aforementioned results suggest that RDP and BDP both elute more readily than TPP, which is remarkable, since data from the literature and the previous experiment suggest the opposite. These relative low recoveries were most likely caused by adsorption to the polypropylene bottles and other laboratory materials, since the recoveries were determined based on the concentrations of the not extracted controls.

3.1.3 SPE - Variation in ratios of solvent mixture

Given the fact that recoveries from solid phase extractions varying in sample volume were relatively low, it is desirable to examine different methodologies to enhance analyte yield. TPP, BDP and RDP are increasing in KOW and decreasing in solubility (table 1). As a result,

these compounds have the tendency to adsorb to the polypropylene bottles but dissolve in methanol (or another organic solvent) very well. Various amounts of methanol were added to a test solution to see whether losses due to adsorption could be minimized. The ratio between the two solvents should be selected so that the methanol enhances the presence of analytes in the extraction and does not wash out the compounds directly during the sample workup.

The samples with different solvent mixture ratios (200 mL) were prepared in

polypropylene bottles in duplicate. The internal standard was added (20 μg) as well as 50 μg of the OPFR stock mix.

The proportions between the recoveries of TPP, BDP and RDP for 50:50 are similar to recovery proportions for 60:40 water/methanol, with an overall exceeding recovery for the latter (see figure 8, appendix B). TPP is least present after extraction, which can be ascribed to the fact that it is less hydrophobic than RDP and BDP and therefore the methanol, prevents interaction with the cartridge. For the two aforementioned solvent ratios, BDP appeared to be the most abundant in extract, 62% for 50:50 and 79% for 60:40. The KOW of BDP is close to

the value of TPP but it is significantly less soluble; hence the methanol was sufficient to remove this compound to a great extent, but did not flush it out prior to the extraction.

(15)

In conclusion, the highest analyte recovery was achieved with a water/methanol ratio of 60:40. Compared to the results from the previous experiment on sample volumes (see 3.1.2 and appendix 5.1)this solvent mixture increased recoveries for TPP (9%), BDP (47%) and RDP (8%). However, this may not be the best practical method; the methanol would dilute the sample concentration, which could complicate the detection of analytes.

3.1.4 Adsorption to laboratory materials

The KOW values and solubility of TPP, RDP BDP as mentioned above and in table 1, data

from literature and previous results, emphasize the review of analyte losses due to adsorption to laboratory materials. Four mixtures of the organic solvents hexane and acetone were tested in duplicate to select a suitable solvent to extract analytes from laboratory materials. The internal standard was added (20 μg) as well as 50 μg of the OPFR stock mix.

The results after double extraction are displayed in figure 9 in appendix C. For all hexane/acetone ratios, TPP had the highest recovery followed by RDP and BDP. Recoveries of BDP and RDP showed almost no difference between the different extraction solvent ratios 17.5% - 19% and 28% - 33% respectively. The yield of TPP was approximately 10% higher after extraction with pure acetone.

The aforementioned recoveries are relatively low for all analytes. The higher yield for TPP compared to RDP and BDP may indicate that these tested organic extraction solvent mixtures were not strong enough for proper extraction of the laboratory materials, for the latter two compounds. Also, between the two extractions the bottles were shaken with the organic extraction solvent for fifteen minutes. Possibly, this amount of time was to little or there was not enough contact between the solvent and the bottle, for BDP and RDP to dissolve, due to their lower solubility and higher KOW.

Overall, the differences in recoveries between extraction solvent mixtures, needed for LC-MS/MS measurements, were minimal. The evaporation was a time-consuming step but progressed better for hexane, hence pure hexane appeared as the better extraction solvent choice.

In order to determine overall analyte recoveries, solid phase extraction was preformed

followed by triple extraction of the polypropylene bottles with pure hexane. To gain the most significant data, the experiment was preformed in quartet. Furthermore, to avoid analyte loss during SPE sample workup, the organic extraction solvent (methanol) was first used to rinse the polypropylene bottles. Recoveries were determined with the internal standard method.

From the results in figure 5 it appears that 10%-20% of the original analyte

concentration is retrieved during the hexane extraction. For BDP a relative low recovery is found after the extraction compared to the other two compounds. The extraction of the polypropylene bottles brought the total recoveries for the targeted compounds to approximately 60% for TPP and BDP and over 70% for RDP.

The recoveries after solid phase extraction are significantly lower compared to the recoveries of the analytes in 3.1.1, which could be explained by the quantification method; the use of internal standard resulted in overcompensation with recoveries over 100%, therefore the controls were used to quantify. Nevertheless, the internal standard provided representative analyte distribution in this experiment but lower recoveries, possibly because of more

accurate representation of analyte losses by quantification with the IS. The unexpected slightly inverse trend in recoveries compared to the solubility of the analytes may be ascribed to their molecular sizes. The considerable sizes of BDP and RDP might impair these

compounds from interaction with the polypropylene bottles. Another possibility might be that more analytes were present due to improper cleaning of the laboratory materials for BDP and

(16)

TPP BDP RDP 0 10 20 30 40 50 60 70 80

Analyte recoveries of SPE and hexane extraction combined

EXTR SPE Compound % R ec ov er y

Figure 5. Distribution of analytes after solid phase extraction and triple extraction with 10 mL hexane.

3.1.5 Overall method summary

In summary, the Water Oasis HLB 150 cartridge preformed best analyte retention under acidified conditions (pH 2). The maximum recoveries were obtained with sample volumes of 125 mL. Nonetheless, these recoveries were only slightly elevated from recoveries with 200 mL. The latter volume was designated as the better choice; this way, the absolute analyte quantity in the extract was the highest and thus the most suitable for solutions with low detection concentrations.

The solvent mixture of water/methanol proved best with a ratio of 60:40. For extraction of the polypropylene bottles, the extraction with pure acetone provided a slightly higher recovery for TPP compared to other organic solvent ratios. However, due to the time-consuming evaporation of the solvent, extraction with pure hexane appeared to be the best option.

After implementation of the aforementioned method combined, overall recoveries for TPP, BDP and RDP were 62%, 58% and 73% respectively.

3.2 Method application

3.2.1 Water influent and effluent

Water influent and effluent samples, of a 24 hours composite, were collected from a water treatment plant and treated as described in 2.2 and treated with the aforementioned optimized methodology. Three samples of both water in- and effluent were spiked with the OPFR stock solution (0.2 μg); two samples prior to- and the third sample after solid phase extraction (see figure 2). In addition to the solid phase- and laboratory material extraction, to obtain the most OPFR analytes, the sediment in the water influent is thrice extracted by liquid-liquid

extraction with pure hexane and evaporated under nitrogen. The extraction fractions were combined and the volume (13 mL) was reduced for further concentration to approximately 1-2 mL. The internal standard dTPP was used (0.5 μg) for all samples to quantify OPFR concentrations and recoveries.

Because the starting volumes of the water samples were approximately 200 mL and the final volumes of the samples ranged from 1 mL to 2 mL, the spiking of these samples with

(17)

the OPFR stock solution was not completely accurate. Quantification was based on the internal standard method, but these results remain merely rough estimations.

TPP BDP RDP 0 20 40 60 80 100 120

Recovery OPFRs of water in- and effluent

Influent Effluent Blank Compound % R ec ov er y TPP BDP RDP 0 0.1 0.2 0.3 0.4 0.5 0.6

Concentration OPFRs in water in- and effluent (unspiked)

Compound C on ce n tr at io n μ g /L

Figure 6. Recoveries of the analytes of water in- and effluent samples (left). Concentrations in μg/L of TPP, BDP and RDP in water influent and effluent samples (right).

The recoveries of the analytes of interest are found in the left diagram in figure 6. For TPP, the recovery of the influent sample was 105% followed in close proximity by RDP (95%). The effluent recoveries of these compounds were roughly half of the influent: 60% and 53% for TPP and RDP respectively. The influent recovery for BDP (35%) is significantly lower compared to the other two compounds, also lower in value than its effluent recovery (44%). The blank samples had recoveries of 4%, 1% and 3% for TPP, BDP and RDP respectively.

As expected based on its relative high solubility and low KOW, the recovery of TPP

was highest for both in- and effluent. On the contrary, both recoveries of BDP were below 50% and significantly lower compared to TPP and RDP; this is noteworthy since BDP has higher solubility and lower KOW compared to RDP and therefore could be expected to elute

more readily. Conceivably, the extraction method was not compatible enough for BDP. The presence of TPP, BDP and RDP was detected in the (unspiked) water in- and effluent samples, as seen in the right diagram in figure 6, all within limit of quantification (LOQ). Therefore all concentrations were established to be significant. TPP was found considerably more present in the influent (0.53 μg/L) and effluent (0.26 μg/L) compared to BDP (0.0210-1 _{μg/L, 0.0610}-1 _{μg/L) and RDP (0.02 μg/L, 0.04 μg/L). Since the latter}

compounds are more hydrophobic, they may have been more accumulated and therefore less present in the water samples. On the other hand, as a result of the mentioned hydrophobicity and as previously suggested, perhaps the extraction method is not yet optimal for BDP and RDP.

Contradictory to the line of expectation, since it may be expected that the analytes would accumulate in the sediment due to their hydrophobicity, the influent recoveries for TPP and RDP were higher compared to their effluent recoveries. This remarkable difference could not be ascribed to the additional liquid-liquid extraction for the water influent since a

comparable difference was present among the in- and effluent samples spiked after the extraction (see figure 10, appendix D). Possibly, components present in the water influent sediment enhanced the LC-MS/MS ionization efficiency, with a matrix effect as result.

(18)

these materials. Furthermore, it was possible that some components were very persistent and acquired a longer elution time.

3.2.2 Sewage treatment plant sludge

Ten primary sludge samples were collected from a sewage water treatment plant. These samples were freeze-dried and attempted to extract. However, the hardened solid would not disperse in the organic extraction solvent (hexane). Thus the composed method for sludge sample preparation prior to the solid phase extraction turned out to be unsuitable.

Nonetheless, the extraction of the solid sludge with hexane was conducted to some extend, for one sample. The single final sample containing 60 mL of the organic solvent hexane was sonicated and evaporated under nitrogen, then redissolved in pure methanol to test for the presence of OPFRs.

The results as seen in figure 11, appendix D, indicate the presence of OPFRs in the tested sample. For TPP, 0.75 μg OPFR/kg sludge was found. For BDP and RDP amounts of 0.044 μg and 0.033 μg per kg of sludge were found respectively. The concentrations of these last two compounds were 4.1 μg/L (BDP) and 3.1 μg/L (RDP), thus all within the range of quantification.

TPP was found in most abundance. Most likely this compound dissolved more readily in the extraction solvent compared to BDP and RDP, due to its higher solubility and lower KOW. A second possibility could be that, in sewage treatment plant sludge, BDP and RDP

were less present than TPP. This would remarkable since the latter two compounds

accumulate more readily as a result of their aformentioned physiochemical properties, thus perhaps BDP and RDP are less commerically applied.

Compared to the organophosphorus flame retardant TCPP found in sludge from twenty plants in Germany with an amount of 5100 μg/kg (see 1.4), the presence of TPP is relatively small. However, it is necessary to review the sludge sample preparation method further in order to collect more accurate data. Since the hardened sludge did not disperse in the hexane, analytes could have remained undissolved. Therefore, the actual amount of TPP, RDP and BDP is plausibly much higher.

3.3 Conclusion

In this study, a methodology was developed and optimized for a trio of organophosphorus flame retardants: TPP, RDP and BDP. This method consisted of solid phase extraction under acidic conditions (pH 2) with hydrophilic-lypophilic-balanced sorbent material, a sample volume of 200 mL and extraction with hexane of the laboratory materials.

The composed methodology was tested on water treatment plant influent, effluent and sludge samples for the detection of the targeted compounds. The results, although indicatory, showed the presence of all three OPFRs in water in- and effluent. Recoveries for the spiked water influent were 105%, 35% and 95% for TPP, BDP and RDP respectively. Effluent recoveries amounted to 60%, 53% and 44% for the aforementioned analytes. The

concentrations of TPP of the water in (0.53 μg/L)- and effluent (0.26 μg/L) was significantly higher compared to BDP and RDP. Remarkably, the presence of analytes was also detected in the blank samples, possibly indicating the presence of OPFRs in the laboratory materials prior to the extraction.

The preparation method for water treatment plant sludge was not yet suitable. Nonetheless, an extraction was preformed and presented the presence of TPP (0.75 μg OPFR/kg sludge) in higher amounts compared to BDP (0.04 μg OPFR/kg sludge) and RDP

(19)

(0.03 μg OPFR/kg sludge). Due to their hydrophobicity, the latter two compounds were expected to be more accumulated in the sludge sediment.

3.4 Future research

The developed methodology as described in 3.1.5 could be retested and possibly be improved to obtain a higher recovery of BDP.

The aforementioned methodology should be repeated on water in- and effluent samples, obtain more accurate recoveries and concentrations. As mentioned in 3.2.1, more precise spiking and evaporation of the extracted samples would enhance the quality of the results. In addition, water samples collected over a greater range of time could also provide more representable data.

For the water treatment plant primary sludge, the described preparation method (see 2.4), especially the freeze-drying the sludge, proved unusable; the solid sediment did not disperse into the extraction solvent. Therefore, a more suitable method should be developed to prepare the samples for solid phase extraction. The dispersion of the solid residue could be improved by freeze-drying the samples over a longer range of time. The developed

methodology should then ideally be conducted on samples gathered over a longer period of time to obtain the most accurate results.

(20)

4. References

1. Alaee, M.; Arias, P.; Sjödin, A.; Bergman, Å An overview of commercially used brominated flame retardants, their applications, their use patterns in different countries/regions and possible modes of release. Environ. Int. 2003, 29, 683-689. 2. Helmus, R. Analysis and Biodegradation of Organophosphorus Flame Retardants in

Aquatic Environments, University of Amsterdam, Amsterdam, 2012. 3. Troitzsch, J. H. Overview of flame retardants. Chemistry today 1998, 16.

4. Pawlowski, K. H.; Schartel, B. Flame retardancy mechanisms of triphenyl phosphate, resorcinol bis (diphenyl phosphate) and bisphenol A bis (diphenyl phosphate) in

polycarbonate/acrylonitrile–butadiene–styrene blends. Polym. Int. 2007, 56, 1404-1414. 5. Birnbaum, L. S.; Staskal, D. F. Brominated flame retardants: cause for concern? Environ.

Health Perspect. 2004, 112, 9-17.

6. van der Veen, I.; de Boer, J. Phosphorus flame retardants: properties, production, environmental occurrence, toxicity and analysis. Chemosphere 2012, 88, 1119-1153. 7. Covaci, A.; Harrad, S.; Abdallah, M. A.; Ali, N.; Law, R. J.; Herzke, D.; de Wit, C. A. Novel brominated flame retardants: a review of their analysis, environmental fate and behaviour. Environ. Int. 2011, 37, 532-556.

8. Darnerud, P. O. Toxic effects of brominated flame retardants in man and in wildlife.

Environ. Int. 2003, 29, 841-853.

9. Sjodin, A.; Hagmar, L.; Klasson-Wehler, E.; Kronholm-Diab, K.; Jakobsson, E.; Bergman, A. Flame retardant exposure: polybrominated diphenyl ethers in blood from Swedish workers. Environ. Health Perspect. 1999, 107, 643-648.

10. Sjödin, A.; Patterson, D. G.; Bergman, Å A review on human exposure to brominated flame retardants—particularly polybrominated diphenyl ethers. Environ. Int. 2003, 29, 829-839.

11. de Wit, C. A. An overview of brominated flame retardants in the environment.

Chemosphere 2002, 46, 583-624.

12. Bocio, A.; Llobet, J.; Domingo, J.; Corbella, J.; Teixido, A.; Casas, C. Polybrominated diphenyl ethers (PBDEs) in foodstuffs: human exposure through the diet. J. Agric. Food

Chem. 2003, 51, 3191-3195.

13. Darnerud, P. O.; Eriksen, G. S.; Johannesson, T.; Larsen, P. B.; Viluksela, M.

Polybrominated diphenyl ethers: occurrence, dietary exposure, and toxicology. Environ.

Health Perspect. 2001, 109 Suppl 1, 49-68.

14. Meironyté, D.; Norén, K.; Bergman, A. Analysis of polybrominated diphenyl ethers in Swedish human milk. A time-related trend study, 1972-1997. Journal of Toxicology and

(21)

15. Zegers, B. N.; Lewis, W. E.; Booij, K.; Smittenberg, R. H.; Boer, W.; de Boer, J.; Boon, J. P. Levels of polybrominated diphenyl ether flame retardants in sediment cores from Western Europe. Environ. Sci. Technol. 2003, 37, 3803-3807.

16. Jakobsson, K.; Thuresson, K.; Rylander, L.; Sjödin, A.; Hagmar, L.; Bergman, Å Exposure to polybrominated diphenyl ethers and tetrabromobisphenol A among computer technicians. Chemosphere 2002, 46, 709-716.

17. Öberg, K.; Warman, K.; Öberg, T. Distribution and levels of brominated flame retardants in sewage sludge. Chemosphere 2002, 48, 805-809.

18. Law, R. J.; Allchin, C. R.; De Boer, J.; Covaci, A.; Herzke, D.; Lepom, P.; Morris, S.; Tronczynski, J.; De Wit, C. A. Levels and trends of brominated flame retardants in the European environment. Chemosphere 2006, 64, 187-208.

19. Fromme, H.; Körner, W.; Shahin, N.; Wanner, A.; Albrecht, M.; Boehmer, S.; Parlar, H.; Mayer, R.; Liebl, B.; Bolte, G. Human exposure to polybrominated diphenyl ethers (PBDE), as evidenced by data from a duplicate diet study, indoor air, house dust, and biomonitoring in Germany. Environ. Int. 2009, 35, 1125-1135.

20. Reemtsma, T.; Quintana, J. B.; Rodil, R.; Garcı, M.; Rodrı, I. Organophosphorus flame retardants and plasticizers in water and air I. Occurrence and fate. TrAC Trends in

Analytical Chemistry 2008, 27, 727-737.

21. Carlsson, H.; Nilsson, U.; Becker, G.; Östman, C. Organophosphate ester flame retardants and plasticizers in the indoor environment: Analytical methodology and occurrence.

Environ. Sci. Technol. 1997, 31, 2931-2936.

22. Fries, E.; Püttmann, W. Occurrence of organophosphate esters in surface water and ground water in Germany. Journal of Environmental Monitoring 2001, 3, 621-626. 23. Bester, K. Comparison of TCPP concentrations in sludge and wastewater in a typical

German sewage treatment plant-comparison of sewage sludge from 20 plants. Journal of

Environmental Monitoring 2005, 7, 509-513.

24. Rodil, R.; Quintana, J. B.; Reemtsma, T. Liquid chromatography-tandem mass spectrometry determination of nonionic organophosphorus flame retardants and plasticizers in wastewater samples. Anal. Chem. 2005, 77, 3083-3089.

25. Hall, T. G.; McKearn, D.; Smukste, I.; Bresciano, K. R.; Savage, R. E.; Wang, Y.

Identifying and overcoming matrix effects in drug discovery and development; INTECH

(22)

5. Appendix

A.

25 mL Pipet 25 mL Tube 50 mL 75 mL 125 mL 200 mL 0 5 10 15 20 25 30 35

Analyte recoveries for different methods and volumes

TPP BDP RDP Volume & method % R ec ov er y

Figure 7. Distribution of analytes after solid phase extraction, normalized with respect to the control samples, with a variety of sample volumes loaded through connecting tubes and in addition, 25 mL loaded by pipet.

B.

50:50 60:40 75:25 0 20 40 60 80 100 120

Analyte recoveries for different solvent mixtures

TPP BDP RDP Water/methanol % R ec ov er y

Figure 8. Distribution of analytes after solid phase extraction for a variety of water/methanol solvent ratios.

(23)

C.

0 10 20 30 40 50 60

Analyte recoveries after extraction with different organic solvent ratios

Hexane/Acetone % R ec ov er y

Figure 9. Distribution of analytes after double extraction with a variety of solvent acetone/hexane mixtures.

D.

TPP RDP BDP 0 0.5 1 1.5 2 2.5 3 3.5 4

Concentration OPFRs water in- and effluent, before and after SPE

Infl spike before SPE Infl spike after SPE Eff spike before SPE Eff spike after SPE Axis Title C on ce n tr at io n μ g /L

Figure 10. The concentrations of OPFRs of water inffluent and effluent samples, spiked prior to- and after solid phase extraction.

(24)

E.

TPP BDP RDP 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Amount OPFR in microgram per kg sludge

μg O P F R /k g sl ud ge

Analysis of Organophosphorus Flame Retardants in Sewage Treatment Plant Sludge Optimization of an analytical detection method for TPP, RDP and BDP

University of Amsterdam

Institute for Biodiversity and Ecosystem Dynamics