Determination of aryl-PFRs in indoor dust from different microenvironments in Spain and the Netherlands and assessment of human exposure

(1)

MSc Analytical Sciences

Chemistry

Master Thesis

Determination of aryl-PFRs in indoor dust from different

microenvironments in Spain and the Netherlands and assessment of

human exposure

By Maria K. Björnsdotter UvA: 11166835 VU: 2574756 June 2017 42 EC Period 2 to 6 Supervisor/Examiner: Examiner: Dr. A. Ballesteros-Gómez Dr. W.T. Kok Dr. H. Lingeman

Department of Analytical Chemistry Institute of Fine Chemistry and Nanochemistry

(2)

Abstract

Phosphate flame retardants (PFRs) are ubiquitous chemicals in the indoor environment. Among them, aryl-PFRs, such as triphenyl phosphate (TPHP), are frequently detected in indoor dust, which is an important route for human exposure to these contaminants. TPHP is an aryl-PFR and a plasticizer that is widely used in electric and electronic equipment. It has been shown to migrate from materials resulting in environmental contamination and it has been detected in indoor dust worldwide. Diphenyl phosphate (DPHP), the hydrolyzed metabolite of TPHP, has been used as a biomarker for monitoring the human exposure to TPHP. However, a lack of correlation between the levels of TPHP in indoor dust and DPHP in urine has been observed up to date. The high urinary concentrations of DPHP suggests additional sources of TPHP and DPHP and/or other aryl-PFRs that could also be metabolized into DPHP. In this study, DPHP (and TPHP) are measured in indoor dust in samples collected in the Netherlands (n=23) in June 2016 and in Spain (n=57) in March and April 2017 using liquid chromatography coupled with triple-quadrupole mass spectrometry (LC-QQQ-MS/MS). A suitable extraction/clean-up method based on salting-out extraction followed by dispersive solid phase extraction (SPE) was optimized and employed for this purpose. Additionally, the presence of other emerging aryl-PFRs was monitored by target screening of the samples.

TPHP and DPHP were present in all samples analyzed from Spain and the Netherlands (n=80) in the range 169-142459 ng/g and 106-79661 ng/g, respectively. The DPHP concentrations were strongly correlated to the TPHP concentrations (r=0.90, p<0.01). Estimated exposures for adults and toddlers in Spain to TPHP via dust ingestion were much lower than the reference dose, indicating no current health risk to the Spanish population. The estimated urinary DPHP levels for adults and toddlers in Spain as a result of exposure to TPHP And DPHP via indoor dust ingestion were too low to significantly contribute to the high urinary DPHP concentrations reported in the literature, indicating that other sources of DPHP may play an essential role in the urinary levels of DPHP. Other aryl-PFRs, namely Cresyl diphenyl phosphate (CDP), resorcinol bis(diphenyl phosphate) (RDP), 2-Ethylhexyl diphenyl phosphate (EDP), Isodecyl diphenyl phosphate (IDP) and Bisphenol A bis(diphenyl phosphate) (BADP), were all detected in indoor dust, however, with lower frequency.

(3)

1. Introduction

Flame retardants (FRs) are widespread in the environment due to their wide use in materials, such as furniture, electronics and textiles, in order to prevent ignition and to slow down the spread of an already initiated fire (EFRA, 2007). Concern has been raised considering their migration from materials as it affects the indoor air quality and is a route for human exposure (Kemmlein et al., 2003). Polybrominated diphenyl ethers (PBDEs) have been commonly used FRs until recently, when their use in electrical and electronic equipment was restricted due to their known toxicity, persistence and bioaccumulative properties (U.S. EPA, 2009). The European Union has banned the use of pentaBDE and octaBDE in 2004 (Directive 2002/96/EC) and the use of decaBDE in electric and electronic equipment in 2009 (European Court of Justice, 2008). This regulation has led to a phase-out of PBDEs in materials resulting in an introduction of alternatives, such as aryl-phosphate flame retardants (aryl-PFRs), onto the market. Studies have demonstrated an increase in the presence of alternative FRs in indoor dust, for which toxicity is still uncharacterized, in conjunction with the decrease of PBDE (Dodson et al., 2012; Tao et al., 2016).

1.1. Aryl-phosphate flame retardants (aryl-PFRs)

Phosphorus FRs (PFRs) include both inorganic and organic compounds and are widely used in plastics, polyurethane foams, thermosets, coatings and textiles (EFRA, 2007). Organic PFRs, commonly known as organophosphate FRs (OPFRs) mainly act by forming a polymeric structure of phosphoric acid formed from the reaction of phosphates under high heat. This layer, known as a char layer, shields from oxygen and prevents the formation of flammable gases and thereby lowers the risk of an initiated fire to spread (Schmitt, 2007). One of the most important groups of OPFRs are phosphate esters, which are mainly used as additive FRs in polyvinyl chloride and engineering plastics commonly used is electronic equipment (EFRA, 2007). Phosphate esters are derivatives from phosphoric acid with one, two, or three substituted groups (ATSDR) (Figure 1). The substituents might be aliphatic (alkyl-PFRs) or aromatic (aryl-PFRs) and are in many cases identical, which is the case for triphenyl phosphate (TPHP), a phosphate triester with three phenyl groups.

(5)

2

Figure 1. Phosphate esters, mono-, di-, and tri-substituted.

1.2. Triphenyl phosphate

Triphenyl phosphate (TPHP; CAS no. 115-86-6) (Figure 2) is an aryl-PFR mainly used as an additive in polymer mixtures used in electronic enclosure applications (LCSP, 2005). The use of TPHP has resulted in environmental contamination due to its migration from materials (Kemmlein et al., 2003).

Figure 2. Molecular structure of triphenyl phosphate (TPHP).

TPHP has been reported in the indoor environment in indoor dust collected from the floors of residences (<2-1798000 ng/g) (Garcia et al., 2007; Stapleton et al., 2009; Kanazawa et al., 2010; Bergh et al., 2011; Van den Eede et al., 2011; Ali et al., 2012a; Ali et al., 2012b; Dirtu et al., 2012; Dodson et al., 2012; Ali et al., 2013; Kim et al., 2013; Abdallah and Covaci, 2014; Araki et al., 2014; Brandsma et al., 2014; Cequier et al., 2014; Fan et al., 2014; Tajima et al., 2014; Brommer and Harrad, 2015; Hoffman et al., 2015; Mizouchi et al., 2015; Zheng et al., 2015; Ali et al., 2016; Canbaz et al., 2016; Cristale et al., 2016; Harrad et al., 2016; He et al., 2016; Wu et al., 2016; Kademoglou et al., 2017), in indoor dust from offices (11-50000 ng/g) (Bergh et al., 2011; Abdallah and Covaci, 2014; Brommer and Harrad, 2015; Cristale et al., 2016; Harrad et al., 2016; He et al., 2016; Wu et al., 2016; Kademoglou et al., 2017), in indoor dust from schools and daycare centers (10-90000 ng/g) (Bergh et al., 2011; Cequier et al., 2014; Fromme et al., 2014; Brommer and Harrad, 2015; Mizouchi et al., 2015; Cristale et al., 2016; Wu et al., 2016). TPHP has also been reported in dust from cars (<2-170000 ng/g) (Ali et al., 2013; Abdallah and Covaci, 2014; Brandsma et al., 2014; Brommer and Harrad, 2015; Ali et

P O OH R1 R2 P O OH R OH P O R3 R1 R2 P O OH O H OH

Phos phoric acid Phos phate monoester Phos phate diester Phos phate tri ester

P O O O O TPHP

(6)

3 al., 2016; Harrad et al., 2016), and from dust collected from public microenvironments such as shops, restaurants and supermarkets (14-34200 ng/g) (Van den Eede et al., 2011; Ali et al., 2012b; Abdallah and Covaci, 2014; Cristale et al., 2016; He et al., 2016). TPHP has also been reported in indoor air (0.19-5.7 ng/m3) (Björklund et al., 2004; Hartmann et al., 2004), in outdoor air (0.003 ng/m3) (Wolschke et al., 2016), sewage water influent (76-290 ng/L) and effluent (41-130 ng/L) and sewage sludge (52-320 ng/g dw) (Marklund et al., 2006), surface water (<LOD-10.3 ng/L) (Bollmann et al., 2012), sediment (5.6-253 ng/g) (Giulivo et al., 2016; Tan et al., 2016) and in fish (43-230 ng/g lw) (Giulivo et al., 2016; Matsukami et al., 2016). Furthermore, TPHP has been reported associated with airborne particles over the global oceans indicating a potential for long-range atmospheric transport towards the polar regions (Möller et al., 2012).

1.3. Toxicity and environmental concern of triphenyl phosphate

The widespread occurrence of TPHP in the indoor- and outdoor environment has led to concern regarding human health and the environment. The human toxicity of TPHP is considered low to high according to a recent alternatives assessment report (U.S. EPA, 2014). This is based on OncoLogic modeling studies showing a moderate potential for carcinogenicity and in vivo studies indicating a high potential for repeated dose effects based on reduced body weight in male rats administered TPHP in diet during four weeks (U.S. EPA, 2014). Furthermore, PFRs including TPHP may be associated with altered hormone levels and decreased semen quality in men (Meeker and Stapleton, 2010). The aquatic toxicity of TPHP is considered very high (Fish 96 h EC50=0.4 mg/L, fish 30-day LOEC=0.037 mg/L) and may cause long-term adverse

effects in the aquatic environment (U.S. EPA, 2014). The environmental persistence is considered low, although there is a moderate potential for bioaccumulation (U.S. EPA, 2014).

1.4. Exposure sources and pathways

FRs are commonly used as additives in consumer products such as furniture, electronics and textiles, i.e. they are not necessarily covalently bound in materials and tend to migrate into the surrounding environment (Kemmlein et al., 2003). Human exposure to FRs as well as other contaminants has been associated with inhalation and ingestion of contaminated indoor dust (Covaci et al., 2012). High levels of contaminants in indoor dust is posing a risk to human health, particularly vulnerable groups such as toddlers, which are especially prone to exposure to contaminants associated with dust since they encounter it more when crawling and playing on the floor as well as when they put items in their mouth (WHO, 2011). We spend most of

(7)

4 our time indoor in homes and offices and are continuously exposed to contaminants in indoor dust. Indoor dust has been considered one of the most important pathways for exposure to FRs (de Boer et al., 2016) and measuring levels of FRs in indoor dust therefore is considered a suitable approach for monitoring chronic exposure.

1.5. Monitoring human exposure

Recent research has been focusing on characterizing human exposure to aryl-PFRs by investigating correlations between aryl-PFRs in indoor dust and their metabolites in urine. For this purpose, several metabolites may be of interest, including diphenyl phosphate (DPHP), which is the hydrolyzed metabolite of TPHP (Figure 3) (U.S. EPA, 2014). DPHP has been used as a biomarker for assessing exposure to TPHP in indoor dust and has been widely reported in urine in the range <0.13-727 ng/mL (Cooper et al., 2011; Meeker et al., 2013; Van den Eede et al., 2013b; Hoffman et al., 2014; Hoffman et al., 2015; Van den Eede et al., 2015; Kosarac et al., 2016).

Figure 3. Hydrolysis of TPHP into DPHP.

Recent research, however, have found that the urinary levels of DPHP are strongly uncorrelated to TPHP concentrations in indoor dust (rS=0.04, (Meeker et al., 2013); rS=0.15, (Hoffman et

al., 2015)), indicating that TPHP in dust is not the only source for human urinary levels of DPHP. A possible additional source to explain the high urinary concentrations of DPHP could be the direct exposure to DPHP itself as it is used in other applications (Makiguchi et al., 2011; Zhao and Hadjichristidis, 2015) or direct exposure to DPHP via indoor dust ingestion as it may be present as an impurity and/or as a degradation product as a result of spontaneous or microbial hydrolysis of TPHP and/or other aryl-PFRs. DPHP has been reported in plastics from electrical and electronic equipment that contain high levels of resorcinol bis(diphenyl phosphate) (RDP) (Ballesteros-Gomez et al., 2016a; Ballesteros-Gomez et al., 2016b). Moreover, DPHP has been demonstrated to be a metabolite to some other aryl-PFRs such as 2-Ethylhexyl diphenyl phosphate (EDP) (Nishimaki-Mogami et al., 1988; Ballesteros-Gomez et al., 2015a), RDP (Ballesteros-Gomez et al., 2015b) and tert-Butylphenyl diphenyl phosphate

P O O O O TPHP DPHP P O O O OH

(8)

5 (BPDP) (Heitkamp et al., 1985). There is almost no data available about the presence of DPHP in the indoor environment and determining levels of DPHP in indoor dust could play an essential role in the understanding of the exposure sources and routes to TPHP and other aryl-PFRs as well as DPHP itself. To the best of our knowledge only one study has reported levels of DPHP (75-190 ng/g) in 4 dust samples collected in Australia (Van den Eede et al., 2015). In this study, a method for the quantitation of TPHP and DPHP in indoor dust was developed using liquid chromatography coupled with a triple-quadrupole mass spectrometer (LC-QQQ-MS/MS). The developed method was applied on indoor dust samples collected from households in the Netherlands in June 2016 (n=23) and in Spain in March and April 2017 (n=57) for the quantification of TPHP and DPHP. The levels of TPHP and DPHP were compared between different microenvironments and countries and the correlation between TPHP and DPHP concentrations were investigated. Human exposure to TPHP and DPHP via indoor dust ingestion was estimated using different exposure scenarios. Furthermore, to gain knowledge about the presence of other aryl-PFRs in indoor dust that could also contribute to the formation of DPHP, an in-house developed database was used for target screening of other emerging aryl-PFRs.

2. Experimental section

2.1. Chemicals and reagents

Acetonitrile and Methanol were acquired from VWR chemicals (Llinars del Vallès, Barcelona, Spain). Ammonium acetate was from Sigma Aldrich (Zwijndrecht, the Netherlands). Ultra-high-quality water was obtained from a Milli-Q water purification system (Millipore, Madrid, Spain). Standard reference material (SRM) 2585 (organic contaminants in house dust) were provided by the National Institute of Standards and Technology (NIST). TPHP, DPHP, TPHP-d15 and DPHP-d10 were obtained from Sigma Aldrich Chemie B.V. (Zwijndrecht, the

Netherlands). Cresyl diphenyl phosphate (CDP), Isodecyl diphenyl phosphate (IDP), EDP, RDP and Bisphenol A bis(diphenyl phosphate) (BADP) analytical standards were obtained from AccuStandard (New Haven, CT).

2.2. Method optimization

The method optimization was done by using the indoor dust reference material SRM 2585 (50 mg). Since the material already contained DPHP and TPHP at relatively high concentrations, the deuterated internal standards were employed for recovery optimization. The observed

(9)

6 average concentration of TPHP and DPHP in SRM 2585 (n=3) was 1075±151 ng/g and 4967±129 ng/g, respectively, which for TPHP is in accordance with previously reported concentrations by other authors ranging 980±60 (Harrad et al., 2016) to 1110±48 (Brandsma et al., 2013).

The reference material was spiked with 0.1 μg internal standard (IS) (TPHP-d15 and

DPHP-d10). The spiking was done before extraction, before clean-up or at the final reconstitution step

in order to assess the extraction efficiency, clean-up losses, matrix effects and total recoveries. When spiked before extraction, the SRM was left stand for 2 h prior to extraction to allow the solvent to evaporate in order to mimic as much as possible the interaction of the compound with the dust matrix.

The extraction of DPHP and TPHP in indoor dust was performed by salting-out extraction with acetonitrile and aqueous ammonium acetate (NH4Ac). A two-phase system was used to reduce

co-extraction of unwanted matrix components and thus achieve cleaner extracts. Due to the high polarity of DPHP it is expected to partly remain in the aqueous phase. Therefore, different salt concentrations were evaluated to increase the partition into the acetonitrile phase. After extraction, due to the complexity of the dust matrix, a clean-up step with QuEChERS (75 mg MgSO4, 25 mg PSA, 25 mg C18, 25 mg GCB) was assessed. Finally, two different LC columns

were evaluated to reduce interferences from co-eluting compounds, namely a Phenomenex Luna® C18 column (2.0 mm i.d., 100 mm length, 3.0 m particle size) and a Phenomenex

Luna® phenyl-hexyl column (2.0 mm i.d., 100 mm length, 3.0 m particle size) equipped with a Phenomenex SecurityGuardTM C18 guard column (2.0 mm i.d., 4.0 mm length). The LC

gradient and the MRM transitions were optimized.

An in-house database was developed for the targeted screening of CDP, IDP, EDP, RDP and BADP in indoor dust. MS/MS transitions and parameters were optimized.

2.3. Apparatus and sample analysis

The LC system used was an Agilent Technologies 1200 LC. A Phenomenex Luna® C18 column

(2.0 mm i.d., 100 mm length, 3.0 m particle size) was used for separation. The mobile phase consisted of 5 mM aqueous ammonium acetate (A) and methanol (B) at a flow rate of 0.25 mL/min. The gradient was as follows: initial 20% B, increased to 95% in 7.5 min and hold for 3 min and finally re-conditioning for 7 min. The MS/MS system was an Agilent Technologies 6420 Triple Quadrupole mass spectrometer equipped with LC-electrospray ionization (ESI) source. The source parameters were set as following: Gas temperature, 320°C; gas flow, 12.0

(10)

7 L/min; nebulizer, 50 psi; capillary voltage, +/-4000 V; MS1 heater, 100°C; MS2 heater, 100°C. The MRN transitions for target masses are given in Table 1. TPHP, BADP, RDP, IDP, EDP and CDP were analyzed in positive ionization mode and DPHP was analyzed in both negative and positive ionization mode.

Table 1. MRM transitions, dwell time, fragmentor voltage and collision energy. Quantifiers

for TPHP and DPHP are indicated in bold. Quantification of DPHP was performed with data acquired using negative ionization.

Compound Precursor ion

(m/z)

Product ion (m/z)

Dwell time (ms) Fragmentor (V) Collision energy (eV) Polarity TPHP 327.1 77.1 150 150 40 Positive TPHP 327.1 215.0 150 135 30 Positive TPHP-d15 342.2 82.2 150 135 30 Positive TPHP-d15 342.2 222.1 150 135 30 Positive DPHP 251.0 77.1 150 135 30 Positive DPHP 251.0 233.1 150 120 20 Positive DPHP-d10 261.1 81.1 150 135 30 Positive DPHP-d10 261.1 161.0 150 120 20 Positive DPHP 249.0 93.0 150 135 30 Negative DPHP 249.0 155.0 150 120 20 Negative DPHP-d10 259.1 97.9 150 135 30 Negative DPHP-d10 259.1 158.6 150 120 20 Negative CDP 341.1 65.2 150 116 89 Positive CDP 341.1 91.2 150 116 53 Positive IDP 391.2 251.1 150 81 15 Positive IDP 391.2 77.1 150 81 73 Positive EDP 363.1 251.1 150 71 11 Positive EDP 363.1 77.1 150 71 93 Positive RDP 575.1 77.2 150 151 115 Positive RDP 575.1 152.2 150 151 79 Positive BADP 693.2 367.2 150 156 47 Positive BADP 693.2 115.2 150 156 101 Positive

2.4. Sample collection and preparation

Sampling was performed using a filter (40 m) mounted in a nozzle adapted to a vacuum cleaner and were not further sieved. Dust samples were collected from residences in the Netherlands in June 2016 from floors (n=12) and from the surface of electrical equipment (n=11) and in Spain in March and April 2017 from the floors of living rooms (n=9), bedrooms (n=9) and offices (n=4), from the surface of electrical equipment (n=13), from cars (n=15) and

(11)

8 from public microenvironments (PMEs) (n=7) (two electronic shops, two clothing shops, one sport clothing shop, one decoration shop and one cafeteria). Approximately 50 mg dust were accurately weighed in 15 mL glass tubes and spiked with IS (TPHP-d15 and DPHP-d10, 0.1 g

each) prior to extraction. Due to the limitation of dust on top of electrical equipment, these samples were in the size of approximately 10-30 mg.

Salting-out extraction with acetonitrile was performed with aqueous NH4Ac (3 M):acetonitrile

(1:1 v/v) by vortex for 2 min followed by centrifugation at 3000 rpm for 5 min. After phase-separation, the acetonitrile layer was collected and transferred to a glass tube. The extraction was repeated 2 times and the acetonitrile layers (~ 6 mL) were combined and evaporated to approximately 1.5 mL (N2, 50°C). Sample clean-up was performed with QuEChERS (75 mg

MgSO4, 25 mg PSA, 25 mg C18, 25 mg GCB) by vortex for 2 min followed by

ultracentrifugation at 10 000 rpm for 5 min. The extract was then evaporated to near dryness (N2, 50°C) and reconstituted in 200 L MilliQ:acetonitrile (1:1 v/v) by vortex for 30 s followed

by ultracentrifugation at 10 000 rpm for 5 min. Extracts were transferred to LC vials and aliquots of 5 L were injected into the LC-MS/MS system.

2.5. Data processing

Quantification of TPHP and DPHP in indoor dust

Quantification of TPHP and DPHP in indoor dust was performed using the quantitative analysis MassHunter workstation software from Agilent Technologies. Linear calibration (1/x weighing, origin included) was employed. The method was evaluated based on extraction efficiency, clean-up losses, matrix effects and reproducibility. Method limits of detection (LOD) and quantification (LOQ) (ng/g) were estimated based on a signal-to-noise ratio of 3 and 10, respectively, taking into account the concentration factor of the method (sample size of 50 mg and final extract volume of 200 μL) and the actual total recoveries.

Statistics

One-way ANOVA was employed to investigate if the TPHP and DPHP concentrations were significantly different in dust collected in Spain and the Netherlands as well as in dust collected from different microenvironments. Pearson correlation was performed in order to investigate the correlation between DPHP and TPHP in indoor dust. For the statistical calculations, the microenvironments were divided into four groups: floor dust (bedrooms, living rooms and offices), dust collected from the surface of electronic equipment, car dust, and dust from the floors of PMEs.

(12)

9

Screening of aryl-phosphate flame retardants

Targeted screening of aryl-PFRs was performed using the quantitative analysis MassHunter workstation software from Agilent Technologies. An in-house database was built containing the masses of the [M+H]+ ion as well as two abundant fragment ions for each target compound. Criteria used for positives were: i) signal-to-noise ratio above 3, ii) qualifier ratio within 80-120% range of the ratio observed from injected authentic standards and iii) an absolute peak area larger than the area obtained from the lowest concentration authentic standard yielding a defined peak.

3. Results and discussion

3.1. Method optimization

The method for quantification of TPHP and DPHP in indoor dust was evaluated based on extraction recovery (%), clean-up recoveries (%), matrix effects (%), and reproducibility (RSD%).

Salting-out phase separation

Three different concentrations of NH4Ac (2, 3 and 4 M) were evaluated in order to increase the

extraction efficiency of DPHP and TPHP. These initial experiments were carried out without the presence of the dust matrix.

Extraction recovery (%) and related SD and RSD (%) for DPHP and TPHP are listed in Table 2 and illustrated in Figure 4. The extraction recoveries for DPHP were between 73% and 82% with the highest recovery obtained using 3 M NH4Ac. For TPHP, extraction recoveries were

in the range 75% - 87% with the highest recovery obtained using 3 M NH4Ac. For both DPHP

and TPHP, standard deviations between replicates were higher at 2 M (5% and 19% RSD for DPHP and TPHP, respectively) and 4 M (8% and 21% RSD for DPHP and TPHP respectively) NH4Ac compared to 3 M (3% and 1% RSD for DPHP and TPHP, respectively).

As expected, the salting-out phase separation was more efficient at higher salt concentrations resulting in extracts containing less water and so reducing the time required for sample evaporation in the pre-concentration step. The optimum salt concentration was considered 3 M, at this concentration the amount of remaining water in the organic phase was minimal and did not affect the evaporation step.

(13)

10

Table 2. Extraction recovery (%) and related SD and RSD (%) for DPHP and TPHP when

extracted with ACN:NH4Ac at different salt concentrations (2, 3 and 4 M).

NH4Ac concentration (M) (ionization mode) Extraction recovery (%) RSD (%) DPHP 2 (neg) 76 ± 4 5 2 (pos) 73 ± 4 5 3 (neg) 82 ± 2 3 3 (pos) 80 ± 2 2 4 (neg) 78 ± 7 8 4 (pos) 78 ± 7 9 TPHP 2 (pos) 83 ± 16 19 3 (pos) 87 ± 1 1 4 (pos) 75 ± 16 21

Figure 4. Extraction recovery (%) of DPHP and TPHP when extracted with ACN:NH4Ac at 2, 3 and 4 M.

Sample preparation recoveries

Sample preparation consisted in salting-out extraction, clean-up with dispersive SPE and an evaporation/reconstitution step. In order to assess losses of the target compounds during the procedure, the recoveries were estimated at the different sample preparation steps, namely extraction efficiency, up recovery, matrix effects, and total recovery (extraction + clean-up + matrix effects).

For the extraction efficiency, recoveries were calculated by spiking with IS before extraction and comparing the signals with those obtained when spiking with IS after the extraction. For the clean-up step, sample extracts were spiked before the clean-up and signals were compared to those obtained when spiking with IS at the reconstitution step. For assessing the matrix

76 82 78 ₇₃ 80 78 83 87 75 0 10 20 30 40 50 60 70 80 90 100

NH4Ac 2M NH4Ac 3M NH4Ac 4M NH4Ac 2M NH4Ac 3M NH4Ac 4M NH4Ac 2M NH4Ac 3M NH4Ac 4M

Negative ionization Positive ionization Positive ionization

DPHP TPHP

(14)

11 effects, samples were spiked at the reconstitution step and IS signals were compared with those obtained from spiked injection solvent. Finally, for total recoveries, samples were spiked before extraction and IS signals were compared with those obtained from spiked injection solvent. Experiments were done in replicates. The results are listed in Table 3 and illustrated in Figure 5.

Extraction recoveries for DPHP at the optimal salt concentration did not vary with the presence of dust (82±2 without dust and 80±3 with dust). The extraction recovery for TPHP was somewhat higher with the presence of dust (99±8) compared to without dust (87±1).

Regarding clean-up recoveries, values were 100±7% and 91±1% for DPHP and TPHP, respectively, so that losses in this step were minimal. For DPHP, matrix effects were not improved by the clean-up step. When operating in negative ionization mode, signal suppression was even slightly higher with the clean-up (87±2 and 94±6 with and without clean-up, respectively). However, when using positive ionization mode, a slightly larger suppression was observed without the clean-up step (56±3% against 62±0%). In general, it can be concluded that the clean-up step did not affect significantly the matrix effects for the analysis of DPHP. However, the negative ionization mode was clearly more selective and the signal less affected by matrix components than the positive ionization mode and was selected for the quantification of DPHP, despite being less sensitive (around 6 times less sensitive than positive ionization mode). For TPHP, the matrix effects were significant and improved when including clean-up (29±5%) compared to when the clean-up step was not included (13±2%). Although the improvement introduced by the clean-up step in terms of matrix effects was not optimal for the target compounds, extracts were clearly cleaner, which is beneficial for a good performance of the LC column and the MS source. The reproducibility in the TPHP analysis in terms of standard deviation (SD) was also improved by the clean-up (deviations were higher ranging 18% to 71% when not using clean-up and 7% to 20% when including clean-up).

Total recoveries (extraction + up + matrix effects) for DPHP were 69±2% (with clean-up, negative mode). As explained before, losses were mainly due to matrix effects and to extraction efficiency. Total recoveries for TPHP were 24±5% (with clean-up, positive mode), due mainly to strong matrix effects.

Table 3. Extraction recovery (%), clean-up recovery (%), matrix effects (%) and total recovery

(15)

12 Extraction recovery (%) RSD (%) Clean-up recovery (%) RSD (%) Matrix effects (%) RSD (%) Total recovery (%) RSD (%) DPHP Without clean-up (neg) 80 ± 3 3 - - 94 ± 6 6 75 ± 2 3 Without clean-up (pos) 77 ± 5 6 - - 56 ±3 5 43 ± 3 6 With clean-up (neg) 79 ± 3 4 100 ± 7 7 87 ± 2 2 69 ± 2 4 With clean-up (pos) 80 ± 2 2 94 ± 1 1 62 ± 0 0 47 ± 1 2 TPHP Without clean-up (pos) 99 ± 8 18 - - 13 ± 2 71 20 ± 4 18 With clean-up (pos) 92 ± 19 20 91 ± 6 7 29 ± 5 16 24 ± 5 20

Figure 5. Extraction recovery (%), clean-up recovery (%), matrix effects (%) and total recovery (%) and related SD for DPHP and TPHP obtained with and without clean-up.

Column and LC gradient

In order to reduce the signal suppression of TPHP, two different columns were evaluated (C18 and phenyl-hexyl) and the samples were analyzed using two different chromatographic gradients (Table 4).

Table 4. LC gradients.

Short gradient Long gradient

Time (min) %B Time (min) %B

0 20 0 10 0.5 20 0.5 10 8 95 20 95 11 95 23 95 11.10 20 23.10 10 18.10 20 30.10 10 80 79 ₇₇ 80 99 92 100 94 91 94 87 56 62 13 29 75 69 43 47 20 24 0 20 40 60 80 100 120

Without clean-up With clean-up Without clean-up With clean-up Without clean-up With clean-up

Negative ionization Positive ionization Positive ionization

DPHP TPHP

(16)

13 The matrix effects for DPHP were not improved when using a phenyl-hexyl column compared to a C18 column. However, the matrix effects were improved when using a longer gradient

(101±1 with long gradient and 83±1 with short gradient). For TPHP, the matrix effects were not improved when using a phenyl-hexyl column compared to a C18 column or when using a

longer gradient (Table 5, Figure 6). The C18 column and the short gradient were used for further

experiments in order to save time.

Table 5. Matrix effects (%), total recovery (%) and related SD and RSD (%) for DPHP and

TPHP obtained using two different columns and two different gradients.

Column Gradient (ionization

mode)

Matrix effects (%) RSD (%) Total recovery (%) RSD (%)

DPHP

C18 Short gradient (neg) 87 ± 2 2 69 ± 2 4

C18 Short gradient (pos) 62 ± 0 0 47 ± 1 2

Phenyl-hexyl Short gradient (neg) 83 ± 1 1 63 ± 0 0

Phenyl-hexyl Short gradient (pos) 55 ± 1 2 41 ± 0 1

Phenyl-hexyl Long gradient (neg) 101 ± 1 1 69 ± 1 1

Phenyl-hexyl Long gradient (pos) 74 ± 0 1 53 ± 1 1

TPHP

C18 Short gradient (pos) 29 ± 5 16 24 ± 5 20

Phenyl-hexyl Short gradient (pos) 29 ± 7 23 24 ± 8 31

Phenyl-hexyl Long gradient (pos) 28 ± 7 25 22 ± 5 22

Figure 6. Matrix effects (%) and total recovery (%) for DPHP and TPHP obtained using two different columns and with two different gradients.

3.2. TPHP and DPHP concentrations in indoor dust

For TPHP and DPHP, the instrument linear range was 0.005-5 µg/mL and 0.005-10 µg/mL, respectively. The instrument LOD and LOQ (TPHP and DPHP) were 0.0001 µg/mL and 0.005 µg/mL, respectively. Method LOD and LOQ were calculated based on signal-to-noise ratio of 3 and 10, respectively, considering sample amount, final extract volume, and total recovery.

87 62 83 55 101 74 29 29 28 69 47 63 41 69 53 24 24 22 0 20 40 60 80 100 120 Short gradient (neg) Short grdient (pos) Short gradient (neg) Short grdient (pos) Long gradient (neg) Long gradient (pos) Short gradient (pos) Short gradient (pos) Long gradient (pos) C18 Phenyl-hexyl C18 Phenyl-hexyl DPHP TPHP

(17)

14 The method LOD and LOQ for TPHP were 1.54 ng/g and 73.96 ng/g, respectively. For DPHP, the method LOD and LOQ were 0.38 ng/g and 19.23 ng/g, respectively. The total recovery in the real dust samples were 26 ± 14 and 104 ± 28 for TPHP and DPHP, respectively (calculated based on IS signal in samples (n=80) spiked prior to extraction and the average IS signal in spiked blanks (n=3)).

TPHP and DPHP were detected at high concentrations in all samples analyzed from the Netherlands and from Spain (Table 6, Figure 7). The highest concentrations of both TPHP and DPHP were observed in dust samples collected from the seats and dashboards of cars (142459 ng/g and 79661 ng/g for TPHP and DPHP, respectively) followed by dust collected from on top of electronic equipment (45330 ng/g and 21899 ng/g for TPHP and DPHP, respectively). To the best of our knowledge, only one study has reported DPHP in indoor dust in the range 75-190 ng/g (Van den Eede et al., 2015). In general, the TPHP concentration was higher than the DPHP concentration, commonly 2-3 times higher, in some cases up to 90 times higher. However, in some samples the DPHP concentration were up to 2 times higher than the TPHP concentration.

In dust collected from the Netherlands, TPHP concentration ranged 172-12853 ng/g and 285-45330 ng/g in dust collected from the floors of homes and offices and from on top of electronic equipment, respectively. The DPHP concentration ranged 151-4189 ng/g (homes and offices) and 218-6588 ng/g (on top of electronics). In samples collected from Spain, the TPHP concentrations ranged 265-18912 ng/g (living rooms), 211-1094 ng/g (bedrooms), 412-1353 ng/g (offices), 1270-26210 ng/g (on top of electronic equipment), 762-142459 ng/g (cars), and 169-1004 ng/g (PMEs). The DPHP concentrations ranged 111-461 ng/g (living rooms), 106-1031 ng/g (bedrooms), 408-1251 ng/g (offices), 299-21899 ng/g (on top of electronic equipment), 923-79661 ng/g (cars), and 263-556 ng/g (PMEs). The high concentrations of TPHP and DPHP found on top of electronic equipment in comparison to concentrations observed in dust collected from the floor in the same room (Figure 8) suggest that electronic equipment is a relevant source of TPHP and DPHP in the indoor environment. However, no correlation was observed between concentrations found in floor dust and in dust collected from the surface of electronic equipment (TPHP, r=0.18; DPHP, r=0.04).

One-way ANOVA revealed that there was no statistically significant difference in TPHP and DPHP levels in dust collected in Spain and in the Netherlands (TPHP, p=0.94, DPHP, p=0.62). The microenvironments were divided into four groups: floor dust (bedrooms, living rooms and

(18)

15 offices), dust collected on top of electronic equipment, car dust and dust collected from the floors of PMEs. Among these groups, no statistically significant difference in TPHP and DPHP levels were revealed except between car dust and floor dust. The concentration TPHP and DPHP in car dust where significantly higher than in floor dust (p<0.05), which could be explained by a high use of flame retardants in the manufacturing of car seats and dashboards and/or less frequently cleaning of cars in comparison to houses.

Table 6. TPHP and DPHP detection frequency (DF) and concentrations (ng/g) in indoor dust

from different microenvironments in Spain and the Netherlands.

DF (%) Mean ± SD Median Minimum Maximum

TPHP (Spain) Living rooms (n=9) 100 3161 ± 6051 944 265 18912 Bedrooms (n=9) 100 674 ± 297 734 211 1094 Offices (n=4) 100 760 ± 413 637 412 1353 On top of electronics (n=13) 100 5900 ± 7105 2416 1270 26210 Cars (n=15) 100 18305 ± 36362 4441 762 142459 PMEs (n=7) 100 665 ± 281 687 169 1004 DPHP (Spain) Living rooms (n=9) 100 241 ± 127 211 111 461 Bedrooms (n=9) 100 314 ± 284 197 106 1031 Offices (n=4) 100 771 ± 354 712 408 1251 On top of electronics (n=13) 100 3211 ± 5780 1753 299 21899 Cars (n=15) 100 8294 ± 19897 2311 923 79661 PMEs (n=7) 100 371 ± 103 357 263 556 TPHP (The Netherlands)

Homes and offices (n=12) 100 3073 ± 3789 1438 172 12853

On top of electronics (n=11) 100 10353 ± 12688 9786 285 45330

DPHP

(The Netherlands)

Homes and offices (n=12) 100 1199 ± 1227 742 151 4189

(19)

16

Figure 7. Median concentration (ng/g) TPHP and DPHP in indoor dust from different microenvironments in Spain and in the Netherlands.

Figure 8. TPHP (A) and DPHP (B) concentrations (ng/g) in dust collected from on top of electronic equipment and from the floor in the same room

The TPHP concentrations in indoor dust from homes in Spain and in the Netherlands are in line with those reported elsewhere (Table 7, Figure 9) (Garcia et al., 2007; Stapleton et al., 2009; Kanazawa et al., 2010; Van den Eede et al., 2011; Ali et al., 2012a; Ali et al., 2012b; Dirtu et al., 2012; Dodson et al., 2012; Ali et al., 2013; Kim et al., 2013; Abdallah and Covaci, 2014; Araki et al., 2014; Cequier et al., 2014; Fan et al., 2014; Tajima et al., 2014; Brommer and Harrad, 2015; Hoffman et al., 2015; Mizouchi et al., 2015; Zheng et al., 2015; Ali et al., 2016; Cristale et al., 2016; Harrad et al., 2016; He et al., 2016; Wu et al., 2016; Kademoglou et al., 2017). Same accounts for TPHP concentrations in dust collected from on top of electronic equipment as well as from floors of offices and PMEs (Figure 1, Table S-2) (Kanazawa et al., 2010; Bergh et al., 2011; Van den Eede et al., 2011; Ali et al., 2012b; Ali et al., 2013; Abdallah and Covaci, 2014; Araki et al., 2014; Brandsma et al., 2014; Tajima et al., 2014; Brommer and

100 1000 10000 Living rooms (n=9) Bedrooms (n=9) Offices (n=4) On top of electronics (n=13)

Cars (n=15) PMEs (n=7) Homes and offices (n=12)

On top of electronics

(n=11)

Spain The Netherlands

lo g (n g/ g) TPHP DPHP 100 1000 10000 100000 1 2 3 4 5 6 7 8 9 10 11 12 13 lo g TP H P (n g/ g) Room

On top of electronic equipment Floor

100 1000 10000 100000 1 2 3 4 5 6 7 8 9 10 11 12 13 lo g D PH P (n g/ g) Room

On top of electronic equipment Floor

(20)

17 Harrad, 2015; Ali et al., 2016; Ballesteros-Gomez et al., 2016a; Cristale et al., 2016; Harrad et al., 2016; He et al., 2016; Wu et al., 2016; Kademoglou et al., 2017). The median TPHP concentration observed in car dust (4441 ng/g) was however somewhat higher than reported before (135-3700 ng/g). Reported TPHP concentrations in house dust as well as in dust from other microenvironments span over a wide concentration range (<2-1798000 ng/g) with the highest concentration reported being observed in house dust from the U.S. (Stapleton et al., 2009). The lowest concentration was observed in house dust from Pakistan (Ali et al., 2012b; Ali et al., 2013) and in car dust from Kuwait (Ali et al., 2013). This high variation in TPHP concentrations, spanning several orders of magnitude, may be explained by different fire-safety regulations in different countries as well as different regulations regarding the production and use of PBDEs. Abdallah and Covaci (2014) reported levels of PFRs in house dust from Egypt which are among the lowest reported PFR levels worldwide (maximum concentration TPHP was 289 ng/g), which may be explained by a higher use of PBDEs and/or less strict fire-safety regulations in Egypt.

Table 7. Summary of TPHP concentration (ng/g) in dust from different microenvironments in

Spain compared to concentrations reported elsewhere.

Microenvironment n DF (%) Mean Median Minimum Maximum Country Reference

Houses (floors)

18 100 1918 782 211 18912 Spain Present study

12 100 3073 1438 172 12853 The Netherlands Present studya

8 100 2600 1850 290 9500 Spain García et al., 2007

5 100 1300 1102 580 2633 Spain Cristale et al., 2016

22 100 1171 230 70 18000 Germany Harrad et al., 2016

33 100 2020 500 40 29800 Belgium Van den Eede et al., 2011

32 - 10000 3300 490 110000 UK Brommer & Harrad, 2015

10 100 2737 1509 190 9549 UK Kademoglou et al., 2017

10 100 931 830 202 2922 Norway Kademoglou et al., 2017

48 100 1240 981 - 4850 Norway Cequier et al., 2014

47 96 1600 500 <20 22600 Eastern Romania Dirtu et al., 2012

20 100 101 67 8 289 Egypt Abdallah & Covaci 2014

15 100 1080 430 44 6890 Kuwait Ali et al., 2013

9 100 4511 3800 1200 9200 Kazakhstan Harrad et al., 2016

15 100 310 230 65 1200 Saudi Arabia Ali et al., 2016

31 100 107 94 <2 630 Pakistan Ali et al., 2012b

15 87 155 175 <2 330 Pakistan Ali et al., 2013

17 100 110 89 8.5 2500 Philipines Kim et al., 2013

20 100 73 71 13 440 Philipines Kim et al., 2013

(21)

18

7 - - 2500 122 16500 China Zheng et al., 2015

13 - - 3320 119 6030 China Zheng et al., 2015

13 - - 1740 31 6660 China Zheng et al., 2015

14 - - 9810 371 332000 China Zheng et al., 2015

6 100 560 600 150 1030 China He et al., 2016

21 100 540 376 122 1829 China Wu et al., 2016

8 63 160 150 ND 390 China He et al., 2016

41 76 - 5400 <1600 78400 Japan Kanazawa et al., 2010

48 60 - 870 - 2335 Japan Tajima et al., 2014

10 100 1400 820 230 6700 Japan Mizouchi et al., 2015

148 89 - 4510 <1600 245080 Japan Araki et al., 2014

32 100 10145 1200 490 110000 Australia Harrad et al., 2016

34 100 590 565 20 7510 New Zealand Ali et al., 2012a

14 100 8704 1600 20 37000 Canada Harrad et al., 2016

134 100 - 1700 260 63000 Canada Fan et al., 2014

50 98 7360 5470 <150 1798000 US Stapleton et al., 2009

53 100 1020 - 100 40350 US Hoffman et al., 2015

16 100 7999 2797 786 36463 US Dodson et al., 2012

Houses (on top of and around electronic

equipment as well as on surfaces e.g. tables, door frames etc.)

11 100 10353 9786 285 45330 The Netherlands Present study

8 - - 6500 1600 21000 The Netherlands Brandsma et al., 2014

30 100 7962 3721 222 50728 The Netherlands Ballesteros-Gómez et al.,

2016a

10 100 1600 1200 100 4200 Sweden Bergh et al., 2011

128 95 - 3130 - 27470 Japan Tajima et al., 2014

41 98 - 14300 <1600 175000 Japan Kanazawa et al., 2010

120 94 - 11540 <1600 889180 Japan Araki et al., 2014

Houses (coaches and mattresses)

10 100 2985 2350 180 8400 UK Harrad et al., 2016

220 >81 - 419 96 >95000 Sweden Canbaz et al., 2016

16 100 465 240 20 35190 New Zealand Ali et al., 2012a

Offices

12 100 8834 5752 1331 38094 UK Kademoglou et al., 2017

9 100 15708 5300 390 48000 Kazakhstan Harrad et al., 2016

23 100 4136 1928 31 38646 China Wu et al., 2016

(22)

19

63 76 2560 500 <300 64500 Germany Fromme et al., 2014

Daycare centers and schools

6 100 2400 1540 - 6150 Norway Cequier et al., 2014

16 100 868 531 41 3514 China Wu et al., 2016

9 100 269 140 10 1023 China Wu et al., 2016

18 100 6200 2200 350 62000 Japan Mizouchi et al., 2015

Cars

15 87 2165 1760 <2 7415 Kuwait Ali et al., 2013

15 100 665 245 2 4800 Pakistan Ali et al., 2013

Public

microenvironments (PMEs)

15 100 4700 1970 150 34200 Belgium Van den Eede et al., 2011

12 100 101 109 13.5 185 Pakistan Ali et al., 2012b

7 100 520 220 70 1840 China He et al., 2016

(23)

20

Figure 9. Reported median concentration TPHP (ng/g) in indoor dust from houses in different countries.

Figure 10. Reported median concentration TPHP (ng/g) in indoor dust from different microenvironments in different countries

3.3. Correlation between TPHP and DPHP concentrations in indoor dust

Pearson correlation was performed to investigate the correlation between TPHP and DPHP concentrations in indoor dust. Pearson correlation was calculated using all samples collected from the Netherlands and from Spain (n=80). The result indicates a strong and statistically significant positive correlation between the concentration of TPHP and DPHP in indoor dust (r=0.90, p<0.01) (Figure 11). Pearson correlation was also performed for individual microenvironments (Figure 12). Observations with standardized residuals larger than 2 (absolute value) were considered outliers and were excluded prior to regression (indicated in orange). Statistically significant positive correlations were observed in dust collected from floors of houses and offices (r=0.46, p<0.05) (Figure 12A), on top of electronic equipment (r=0.60, p<0.01) (Figure 12B) and cars (r=0.99, p<0.01) (Figure 12C). A strong correlation were also observed in dust collected from the floors of PMEs (r=0.72) (Figure 12D), however, not statistically significant (p=0.07). These findings indicate that the presence of DPHP in indoor dust is to a considerable extent related to the presence of TPHP suggesting that DPHP in indoor dust is mainly present as an impurity and/or a degradation product of TPHP.

7821438 1850 1102 230 500 3300 1509 830 981 ₅₀₀ 67 430 3800 230 600 94 175 89 71 1110 2500 3320 1740 9810 600 376 ₁₅₀ 5400 870 820 4510 1200 565 1600 1700 5470 2797 0 2000 4000 6000 8000 10000 12000 2416 9786 6500 3721 820 1200 3130 14300 11540 637 740 1500 43005752 5300 73 5300 1928 900 4441 180024001700 3300 135 1760 470 245 3700 687 4010 179 1970 629 ₁₀₉ ₂₂₀ 0 2000 4000 6000 8000 10000 12000 14000 16000

(24)

21 However, it cannot be ruled out that the presence of DPHP in indoor dust might also be a result of degradation of other aryl-PFRs.

Figure 11. Correlation between TPHP and DPHP concentration in indoor dust from Spain and the Netherlands.

1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 lo g D PH P (n g/ g) log TPHP (ng/g)

(25)

22

Figure 12. Correlation between TPHP and DPHP concentration in indoor dust in different microenvironments (A) floor dust (houses and offices), (B) on top of electronic equipment, (C) cars, and (D) PMEs.

3.4. Estimated exposure to TPHP and DPHP in indoor dust

Human exposure scenarios to TPHP and DPHP via dust ingestion in Spain were estimated using a method based on that described by Abdallah and Covaci (2014). Briefly, average and high dust ingestion rates (95th percentile) for adults (2.6 mg/day and 8.6 mg/day, respectively) and toddlers (41 mg/day and 140 mg/day, respectively) (Wilson et al., 2013) were used to calculate an average and a worst-case scenario exposure to TPHP and DPHP via indoor dust. Estimated exposure scenarios were calculated based on median and maximum concentrations in indoor dust in homes (bedrooms and living rooms), offices, cars and PMEs (different stores and one cafeteria) in Spain, taking into account the time spent in each environment according to the typical human activity patterns described by Abdallah and Covaci (2014) (i.e. 63.8% home, 22.3% office, 5.1% PMEs, 4.1% car and 4.7% outdoors, for adults, and 86.1% home, 5.1% PMEs, 4.1% car and 4.7% outdoors, for toddlers). Occupational exposure of drivers (e.g. taxi drivers and truck drivers) were estimated by using the concentrations in cars as representative concentrations for the working environment (i.e. time fraction spent in car was 4.1% + 22.3%). Following equation was used to calculate estimated exposure scenarios:

1.00 1.50 2.00 2.50 3.00 3.50 4.00 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 lo g D PH P (n g/ g) log TPHP (ng/g) 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 1.00 2.00 3.00 4.00 5.00 6.00 lo g D PH P (n g/ g) log TPHP (ng/g) 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 1.00 2.00 3.00 4.00 5.00 lo g D PH P (n g/ g) log TPHP (ng/g) 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60 2.80 3.00 1.00 1.50 2.00 2.50 3.00 3.50 lo g D PH P (n g/ g) log TPHP (ng/g) (D) (C) (B) (A)

(26)

23 𝐸𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 𝑑𝑎𝑖𝑙𝑦 𝑒𝑥𝑝𝑜𝑠𝑢𝑟𝑒 (𝑛𝑔/𝑑𝑎𝑦) = 𝐼𝑅× ∑ 𝐶_𝑖×𝐹_𝑖

where IR is the dust ingestion rate (g/day), Ci the concentration TPHP or DPHP in dust in

microenvironment i (ng/g) and Fi is the time fraction spent in microenvironment i.

The estimated exposure to TPHP and DPHP for different exposure scenarios including workers (offices), drivers, non-workers and stay-home toddlers are shown in Table 8. The estimated daily exposure to TPHP via indoor dust ingestion in Spain (based on average dust ingestion rates and median concentrations) were 2.2 ng/day, 4.4 ng/day, 2.3 ng/day and 36.5 ng/day for adult workers, drivers, non-workers and stay-home toddlers, respectively. These exposure scenarios are in line with those reported elsewhere (based on average dust ingestion rate and median concentration) which are in the range 0.9-58.5 ng/day, 7.0-30.2 ng/day and 3-75.4 ng/day for adult workers, non-workers and stay-home toddlers, respectively (Table 9) (Van den Eede et al., 2011; Ali et al., 2012a; Dirtu et al., 2012; Kim et al., 2013; Abdallah and Covaci, 2014; He et al., 2016; Wu et al., 2016; Kademoglou et al., 2017).

Worst-case scenario estimated daily exposure to TPHP via indoor dust ingestion (based on high dust ingestion rates and maximum concentrations) were 157.0 ng/day, 427.6 ng/day, 190.7 ng/day and 3104.5 ng/day for adult workers, drivers, non-workers and stay-home toddlers, respectively (Table 8). For adults, the calculated exposure estimates are in line with those reported elsewhere (based on high dust ingestion rate and 95th percentile or maximum concentration) which are in the range 13.0-953.2 ng/day and 70.0-506.1 ng/day for workers and non-workers, respectively. However, the worst-case scenario estimated daily exposure to TPHP via dust ingestion for stay-home toddlers were 3104.5 ng/day, higher than reported in previous studies (Table 9) (Van den Eede et al., 2011; Ali et al., 2012a; Dirtu et al., 2012; Kim et al., 2013; Abdallah and Covaci, 2014; He et al., 2016; Wu et al., 2016; Kademoglou et al., 2017). Despite the high estimated daily exposure for toddlers, all calculated exposure estimates for different scenarios are far below the reference dose of 164500 ng/day (adults) and 28905 ng/day (toddlers) calculated from the lowest reported chronic NOAEL, 23.5 mg/kg/day (U.S. EPA, 2015) divided by a safety factor of 10000 assuming body weights of 70 kg and 12.3 kg for adults and toddlers, respectively (U.S. EPA, 2008).

The estimated daily exposure to DPHP via indoor dust ingestion in Spain (based on average dust ingestion rates and median concentrations) were 1.0 ng/day, 2.0 ng/day, 0.8 ng/day and 11.8 ng/day for adult workers, drivers, non-workers and stay-home toddlers, respectively (Table 8). Worst-case scenario estimated daily exposure to DPHP via indoor dust ingestion

(27)

24 (based on high dust ingestion rates and maximum concentrations) were 36.4 ng/day, 186.8 ng/day, 36.0 ng/day and 585.5 ng/day for adult workers, drivers, non-workers and stay-home toddlers, respectively. To the best of our knowledge, this is the first study to report estimated daily exposure scenarios to DPHP via indoor dust ingestion.

Table 8. Estimated daily exposure (ng/day) for different exposure scenarios in Spain. Ingestion

rate

Workers Drivers Non-workers Stay-home toddlers

Median Maximum Median Maximum Median Maximum Median Maximum

TPHP Average 2.2 47.5 4.4 129.3 2.3 57.7 36.5 909.2 High 7.4 157.0 14.7 427.6 7.7 190.7 124.6 3104.5 DPHP Average 1.0 11.0 2.0 56.5 0.8 10.9 11.8 171.5 High 3.5 36.4 6.5 186.8 2.5 36.0 40.4 585.5

Table 9. Estimated daily exposure to TPHP (ng/day) reported elsewhere. Dust

ingestion rate

Adults (working) Adults (non-working) Toddlers

Median Maximum Median Maximum Median Maximum Country Reference

Average 1.9 - - - 4.8b _-

Egypt Abdallah & Covaci

2014

High 4.8 - - - 19.3b _-

Average 1.1 18.0a _- _- _3.8 _59.0a

Philipines

(residental area) Kim et al. 2013

High 2.8 44.0a _- _- _15.0 _240.0a Average 0.9 5.3a _- _- _3.0 _18.0a _Philipines (municipal dumping area) Kim et al. 2013 High 2.3 13.0a _- _- _12.0 _71.0a Average 58.5 381.3 30.2 191.0 75.4 477.4 UK Kademoglou et al. 2017 High 146.2 953.2 75.5 477.5 301.8 1909.8 Average - - 16.6 58.5 41.5 146.1

Norway Kademoglou et al.

2017

High - - 41.5 146.1 166.0 584.4

Average - - 7.0 28.0a _18.0 _69.8a

New Zealand Ali et al. 2012a

High - - 18.2 70.0a _71.9 _279.6a

Average 7.0 - 14.0 - 24.6 -

Belgium Van den Eede et al.

2011 High 28.0 357.0a _28.0 _147.0a _100.9 _500.6a Average 21.0 98.0a _- _- _21.6 _86.4a China Wu et al. 2016 High 49.0 245.0a _- _- _86.4 _344.4a Average - - - - China He et al. 2016 High - 141.5 - - - 195.0a Average - - 10.5 203.0a _26.3 _504.0a Eastern

Romania Dirtu et al. 2012

High - - 26.3 506.1a _105.0 _2016.0a

a₉₅th_percentile

(28)

25

3.5. Estimated urinary levels of DPHP

Estimated urinary levels of DPHP as a result of exposure to DPHP and TPHP via indoor dust were calculated based on the median and maximum levels of DPHP and TPHP in indoor dust in Spain as well as the time fractions spent in each microenvironment according to typical human activity patterns described in previous section. A method based on that described by Van den Eede et al. (2015) was employed. Briefly, an average and a high dust ingestion rate (95th percentile) for adults (2.6 mg/day and 8.6 mg/day) and toddlers (41 mg/day and 140 mg/day) (Wilson et al., 2013) was assumed. Other assumptions were the complete absorption of DPHP and TPHP after dust ingestion as well as the complete excretion of DPHP in urine and that DPHP is absorbed and excreted unchanged (Sudakin and Stone, 2011). The assumption that TPHP is metabolized into DPHP by liver enzymes at a rate of 20% was also included (Van den Eede et al., 2013a). Based on these assumptions and assuming a mean urinary output of 800 mL/day for adults and 600 mL/day for children, the estimated DPHP excretion rate (ng/day) and urinary levels (ng/mL) were calculated using following equation:

𝑈𝑟𝑖𝑛𝑎𝑟𝑦 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 (𝑛𝑔 𝑚𝐿) =

𝐼𝑅× ∑ 𝐶𝑖(𝐷𝑃𝐻𝑃)𝐹𝑖+ 0.2×𝐶𝑖(𝑇𝑃𝐻𝑃)𝐹𝑖

𝑈𝑟𝑖𝑛𝑎𝑟𝑦 𝑜𝑢𝑡𝑝𝑢𝑡 (_𝑑𝑎𝑦𝑚𝐿)

Where IR is the dust ingestion rate (g/day), Ci the concentration DPHP and TPHP in dust in

microenvironment i, and Fi is the time fraction spent in microenvironment i.

The estimated urinary DPHP levels for different exposure scenarios in Spain including workers (offices), drivers, non-workers and stay-home toddlers are shown in Table 10. The estimated urinary DPHP levels as a result of exposure to TPHP and DPHP via indoor dust ingestion (based on average dust ingestion rates and median concentrations) were 0.002 ng/mL, 0.004 ng/mL, 0.002 ng/mL and 0.032 ng/mL for adult workers, drivers, non-workers, and stay-home toddlers, respectively. These estimated urinary DPHP levels as a result of exposure to TPHP and DPHP via indoor dust ingestion are not high enough to significantly contribute to the high DPHP urinary levels reported in the literature ranging <0.13-727 ng/mL (Cooper et al., 2011; Meeker et al., 2013; Van den Eede et al., 2013b; Hoffman et al., 2014; Hoffman et al., 2015; Van den Eede et al., 2015; Kosarac et al., 2016).

Worst-case scenario estimated urinary DPHP levels for the different exposure scenarios (based on high dust ingestion rate and maximum concentration in dust) were 0.085 ng/mL, 0.34 ng/mL, 0.094 ng/mL, and 2.011 ng/mL for workers (offices), drivers, non-workers and

(29)

stay-26 home toddlers, respectively. The estimated urinary DPHP level in toddlers is 40 times higher than the worst-case scenario reported previously (0.05 ng/mL) (Van den Eede et al., 2015). Furthermore, the estimated worst-case scenario urinary DPHP levels are in the same range as the lower urinary DPHP concentrations reported previously (<0.13 ng/mL) (Kosarac et al., 2016).

Van den Eede et al. (2016) showed that serum enzymes are involved in the transformation of TPHP into DPHP and that the amount TPHP that reaches the liver after intake may be strongly reduced. Therefore, the metabolic transformation rate of TPHP into DPHP (by serum and liver enzymes) could be higher than 20% resulting in an underestimation of urinary DPHP levels. Same study also investigated the hydrolysis products of EDP by serum enzymes and results suggest an additional production of DPHP from EDP, however, at a much lower rate than for TPHP.

Additional sources of TPHP as well as direct exposure to DPHP from other sources in addition to exposure to other aryl-PFRs being metabolised into DPHP, may play an essential role in the high urinary DPHP levels.

It should be noted that the TPHP and DPHP concentrations in indoor dust varies over several orders of magnitude between different studied environments and between different homes and that the estimated urinary DPHP urinary levels in the present study therefore not can be compared directly to reported urinary levels elsewhere without a large degree of uncertainty.

Table 10. Estimated urinary DPHP concentration (ng/mL) for different exposure scenarios in

Spain.

Average ingestion rate High ingestion rate

Median Maximum Median Maximum

Workers 0.002 0.026 0.006 0.085

Drivers 0.004 0.103 0.012 0.340

Non-workers 0.002 0.028 0.005 0.093

Stay-home toddlers 0.032 0.589 0.109 2.011

3.6. Screening of aryl-phosphate flame retardants

TPHP and DPHP were detected in all samples analyzed from Spain (n=57) and the Netherlands (n=23). The other aryl-PFRs, namely CDP, IDP, EDP, RDP and BADP, were less frequently detected (Table 11). EDP was the most frequently detected aryl-PFR after TPHP and DPHP with a detection frequency of 64.9% and 65.2% in Spain and the Netherlands, respectively,

(30)

27 followed by IDP (50.9% and 43.5%), BADP (33.3% and 34.8%), CDP (3.5% and 8.7%) and RDP (0% and 4.3%). Detection frequencies of all aryl-PFRs included in the present study were similar in samples collected from Spain and the Netherlands. Furthermore, there were no observed differences in the presence of aryl-PFRs in different microenvironments (Table 12). However, due to the limited number collected from each microenvironment, these results are not conclusive.

Table 11. Compound name, CAS, molecular structure, chemical formula, monoisotopic mass

and detection frequency (%) of TPHP, DPHP, CDP, IDP, EDP, RDP, and BADP in indoor dust from Spain and the Netherlands.

Compound CAS

Molecular structure Chemical

formula

Monoisotopic mass (g/mol)

Detection frequency (%) Spain (n=57) The Netherlands (n=23)

Triphenyl phosphate (TPHP) 115-86-6 C18H15O4P 326.070801 100 100 Diphenyl phosphate (DPHP) 838-85-7 C12H10O4P 250.039490 100 100 Cresyl diphenyl phosphate (CDP) 26444-49-5 C19H17O4P 340.086456 3.5 8.7 Isodecyl diphenyl phosphate (IDP) 29761-21-5 C22H31O4P 390.195984 50.9 43.5 2-Ethylhexyl diphenyl phosphate (EDP) 1241-94-7 C20H27O4P 362.164703 64.9 65.2 Resorcinol bis(diphenyl phosphate) (RDP) 57583-54-7 C30H24O8P2 574.094666 0 4.3 O O O O P O H O O O P CH3 O O O O P C H3 CH3 O O O O P CH3 CH3 O O O O P O O O O O O O O P P

(31)

28 Bisphenol A bis(diphenyl phosphate) (BADP) 5945-33-5 C39H34O8P2 692.172913 33.3 34.8

Table 12. Detection frequency (%) of aryl-PFRs in indoor dust from different

microenvironments in Spain and the Netherlands.

Microenvironment TPHP DPHP CDP IDP EDP RDP BADP

Spain

Living rooms (n=9) 100 100 0.0 44.4 44.4 0.0 11.1

Bedrooms (n=9) 100 100 0.0 100 55.6 0.0 33.3

Offices (n=4) 100 100 0.0 50.0 75.0 0.0 0.0

Floor dust (bedroom + living room + office) (n=22)

100 100 0.0 68.2 54.5 0.0 18.2 On top of electronics (n=13) 100 100 7.7 46.2 92.3 0.0 61.5 Cars (n=15) 100 100 6.7 26.7 46.7 0.0 33.3 PMEs (n=7) 100 100 0.0 57.1 85.7 0.0 28.6 The Netherlands

Homes and offices (n=12) 100 100 8.3 33.3 66.7 0.0 0.0

On top of electronics (n=11) 100 100 9.1 54.5 63.6 9.1 72.7

4. Conclusions

Salting-out extraction with acetonitrile and 3 M ammonium acetate provided high extraction efficiencies for TPHP and DPHP in indoor dust. However, TPHP suffered from severe signal suppression that were somewhat improved when clean-up with QuEChERS was employed.

TPHP and DPHP were present at high concentrations with 100% detection frequency in all samples analyzed from Spain and the Netherlands. The highest maximum concentrations of TPHP and DPHP were observed in dust collected from the seats and dashboards of cars (142459 ng/g and 79661 ng/g for TPHP and DPHP, respectively), followed by dust collected from the surface of electronic equipment (45330 ng/g and 21899 ng/g for TPHP and DPHP, respectively). This suggest a high use of TPHP in the manufacturing of car interiors and electronic equipment. The lowest concentrations of TPHP (169 ng/g) and DPHP (106 ng/g) were observed in floor dust collected from PMEs and bedrooms, respectively.

O P O O O O P O O O CH3 C H3

(32)

29 TPHP concentrations in house dust in Spain and the Netherlands are in line with those reported elsewhere. However, the reported concentrations span over several orders of magnitudes with the lowest concentration being reported in house dust from Pakistan (<2 ng/g) (Ali et al., 2012b; Ali et al., 2013) and the highest concentration being reported in house dust from the U.S. (1798000 ng/g) (Stapleton et al., 2009). This wide range of TPHP concentrations being reported in house dust may be explained by different fire-safety regulations in different countries and/or regulations regarding the use of PBDEs.

DPHP concentrations were strongly and statistically significantly correlated to TPHP concentrations in indoor dust from Spain and the Netherlands (r=0.90, p<0.01). The strongest correlation was observed in dust collected from cars (r=0.99, p<0.01). These findings suggest that TPHP is a major source for DPHP in indoor dust. However, other possible sources for DPHP in indoor dust cannot be ruled out since DPHP has been suggested to be an impurity to and/or a degradation product of RDP (Ballesteros-Gomez et al., 2016a; Ballesteros-Gomez et al., 2016b) as well as a metabolite of EDP (Nishimaki-Mogami et al., 1988; Ballesteros-Gomez et al., 2015a), RDP (Ballesteros-Gomez et al., 2015b) and tert-Butylphenyl diphenyl phosphate (BPDP) (Heitkamp et al., 1985). Furthermore, DPHP is also used as an organocatalyst in polymerization processes (Makiguchi et al., 2011; Zhao and Hadjichristidis, 2015). More research would be desirable in order to investigate whether or not other sources of DPHP is also relevant.

The estimated average daily exposure to TPHP and DPHP in Spain is highest for toddlers (36.5 ng/g and 11.8 for TPHP and DPHP, respectively) followed by drivers (4.4 ng/g and 2.0 for TPHP and DPHP, respectively), which is far below the reference dose for TPHP of 164500 ng/day (adults) and 28905 ng/day (toddlers). However, the estimated worst-case scenario daily exposure to TPHP for toddlers were 3104.5 ng/g, which is less than 10 times below the reference dose for toddlers.

The estimated average urinary DPHP concentrations as a result of exposure to TPHP and DPHP via indoor dust ingestion is far below and insufficient to explain the DPHP concentrations reported in urine. Only the estimated worst-case scenario urinary DPHP concentrations are in the same range as the lower DPHP concentrations reported in urine, but still insufficient to explain the higher concentrations reported in urine. Other sources of TPHP exposure and/or the presence of other aryl-PFRs that are degraded and/or metabolised into DPHP may be a relevant

(33)

30 source for the high concentrations of DPHP reported in urine and further research is necessary in order to understand the high concentrations of DPHP in urine.

TPHP and DPHP were detected in all samples analyzed from Spain (n=57) and the Netherlands (n=23). CDP, IDP, EDP, RDP and BADP were less frequently detected. EDP was most frequently detected after TPHP and DPHP. In samples collected from Spain, EDP, IDP, BADP, CDP and RDP were detected in 64.5%, 50.9%, 33.3%, 3.5%, and 0.0% of the samples, respectively. In samples collected from the Netherlands, EDP, IDP, BADP, CDP and RDP were detected in 65.2%, 43.5%, 34.8%, 8.7%, and 4.3% of the samples, respectively. The presence of all aryl-PFRs included in the present study was similar in samples collected from Spain and the Netherlands and no differences could be observed between different microenvironments. However, due to the limited number collected from each microenvironment, these results are not conclusive.

Acknowledgments

A big thank you to everyone in the Supramolecular research group at the Department of Analytical Chemistry, University of Córdoba, for welcoming me with open arms into their group and for being patient and answering my many questions.

And many big thanks to my supervisor Ana Ballesteros-Gómez for giving me the opportunity to perform my research project abroad, for answering my questions and for taking the time of teaching me but also for giving me the chance of being independent and taking my own initiatives.