Journal of Chromatography A, 1362 (2014) 110–118

Contents lists available at ScienceDirect

Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma

Analysis of phthalates in milk and milk products by liquid chromatography coupled to quadrupole Orbitrap high-resolution mass spectrometry Wei Jia a,b , Xiaogang Chu a,b,∗ , Yun Ling b , Junrong Huang a , James Chang c a

College of Chemistry & Chemical Engineering, Shaanxi University of Science & Technology, Xi’an 710021, China Institute of Food Safety, Chinese Academy of Inspection and Quarantine, Beijing 100123, China c ThermoFisher Scientific, 355 River Oaks Parkway, San Jose, CA 95134, United States b

a r t i c l e

i n f o

Article history: Received 30 April 2014 Received in revised form 5 August 2014 Accepted 7 August 2014 Available online 15 August 2014 Keywords: QuEChERS Q-Orbitrap Milk and milk products Phthalates

a b s t r a c t A new analytical method was developed and validated for simultaneous analysis of 27 phthalates in milk and milk products. Response surface methodology was employed to optimize a quick, easy, cheap, effective, rugged, and safe (QuEChERS) sample preparation method. Ultrahigh-performance liquid chromatography and electrospray ionization quadrupole Orbitrap high-resolution mass spectrometry (UHPLC/ESI Q-Orbitrap) was used for the separation and detection of all the analytes. The method was validated by taking into consideration the guidelines specified in Commission Decision 2002/657/EC and 2007/19/EC. The extraction recoveries were in a range of 90.7% to 104.6%, with coefficient of variation 0.99. The limits of detection for the analytes are in the range 0.32–2.6 ␮g kg−1 . This method has been successfully applied on screening of phthalates in 96 commercial milk and milk product samples. © 2014 Elsevier B.V. All rights reserved.

1. Introduction 1,2-Benzenedicarboxylic-acid esters, also known as phthalateacid esters (PAEs), have long been used as industrial plasticizer in a wide range of consumer products. The worldwide annual production of phthalates is approximately 6.0 million metric tons per year [1–3]. Because of their non-covalent bonding properties and extensive use, phthalates are released ubiquitous in the environment. The food packaging and food processing can also introduce these compounds in the food change. These may result in direct contamination of feed and food products, bioaccumulation in tissues, and transfer through the food chain [4–6]. A recent food safety concern has arisen from the unapproved use of certain phthalates as direct food additives in a broad range of food manufactured in China [7]. These phthalates were illegally substituted for food grad emulsifiers in formulating clouding agents meant to provide turbidity to selected food products, mainly

∗ Corresponding author at: Institute of Food Safety, Chinese Academy of Inspection and Quarantine, Beijing 100123, China. Tel.: +86 010 85778904; fax: +86 010 857707750. E-mail address: [email protected] (X. Chu). http://dx.doi.org/10.1016/j.chroma.2014.08.030 0021-9673/© 2014 Elsevier B.V. All rights reserved.

distilled spirits and beverages. Some of these products apparently might have been exported to various parts of the world [8]. Phthalates are considered to be potential mutagenic, carcinogenic activity and endocrine disrupters, with fetal animals being particularly sensitive [9,10]. Although the intake of phthalates may originate from many sources, there is special interest in monitoring the contamination of milk and milk products because they constitute a primary food source, especially for children [11–13]. For monitoring purposes, broad range analytical methods are needed to reduce analytical costs and allow for a more frequent monitoring of phthalates in milk and milk products. For the detection and quantification of phthalates, chromatographic techniques like high performance liquid chromatography (HPLC) with diode array detector (DAD) [14], gas chromatography (GC) and liquid chromatography (LC) coupled to mass spectrometry (MS) have been used [15–17]. Under GC–MS (electron ionization, EI) conditions, the fragment at m/z 149 is the common ion for most phthalates. This is a major limitation in using GC–MS for the determination of phthalates isomeric mixtures, primarily because of the occurrence of coeluting isomers with varying composition of alkyl substitution [18,19]. LC–MS is a suitable technique for the analysis of phthalates because no derivatization step is required as in GC–MS. Tandem quadrupole MS has been widely accepted

W. Jia et al. / J. Chromatogr. A 1362 (2014) 110–118

as the main tool in the structural characterization, identification, and quantitative analysis of phthalates owing to its efficiency, superior sensitivity and specificity [20–22]. However, LC–tandem quadrupole MS is not suitable for simultaneous screening of a large number of phthalates. Besides, false positives caused by complex food matrices are frequently encountered [23]; no studies were reported to simultaneously detect over 25 different phthalates in food samples. From last year the role of UHPLC-Q-Orbitrap is increasingly built up as enabling tool in food safety analysis for it can provide detailed structural information. In spite of the potential value of the application, to the best of our knowledge, so far no people has reported the application of Q-Orbitrap mass spectrometry combined with high performance liquid chromatography for simultaneous determination for a group of phthalates in foods [24]. The analysis of phthalates in milk and milk products is a difficult task because of the lipophilic properties of most phthalates. When sample extraction is performed by solvent mixtures of low polarity, fats are co-extracted together with phthalates. The chromatographic analysis requires the application of previous extraction and clean-up steps in order to remove lipids and proteins. A wide variety of sample preparation has been reported in literature for phthalates, such as liquid–liquid extraction (LLE), solid-phase extraction (SPE), solid-phase microextraction (SPME) and dilute-and-shoot (DAS) [25–28]. However, some of these methods still have some limitation, such as high variability in results, high requirement for clean sample, time consuming as well as expensive, which make them inadequate for routine analyses. Obviously, elimination or simplification of the sample preparation would reduce the risk of contamination. Hence, new straightforward approaches involving simpler and fewer steps would be welcome for a more effective clean-up of complex matrices such as milk and milk products samples. In this way, QuEChERS has been checked elsewhere for the extraction of mycotoxins, plant toxins, pesticide and veterinary drug residues in food and feed, but to date, no work focused on the determination of phthalates in milk and milk products using QuEChERS has been published [29–31]. In this paper, we describe the development of an easy-touse sample preparation based on QuEChERS for the simultaneous extraction of the 27 most important phthalates from milk (cow, fat content > 2%), milk beverages (cow, protein content > 0.7%) and yogurt (cow, fat content > 3%). Coupled with an optimized UHPLC/ESI Q-Orbitrap, this method was successfully applied on screening of phthalates in milk and milk products samples from local market.

2. Experimental 2.1. Chemicals and reagents All reagents were of analytical grade. Acetic acid, formic acid (FAc), ammonium formate, sodium acetate, sodium chloride and anhydrous magnesium sulfate (MgSO4 ) were purchased from Sigma–Aldrich (Steinheim, Germany). HPLC-grade acetonitrile (MeCN) and methanol (MeOH) were sourced from J.T. Baker (Deventer, Holland). BAKERBOND® octadecyl (C18 ), bondesil primary secondary amine (PSA), and ceramic homogenizers obtained from Agilent Technologies (Harbor City, USA). Ultrafree-MC centrifugal filter devices (0.22 ␮m) of Millipore (Millipore, Brussels, Belgium) were used. Trifluoroacetic acid was obtained from Fluka (Buch, Switzerland). Ultrapure Water (resistivity, 18.2 M) was purified on a Milli-Q Plus apparatus (Millipore, Brussels, Belgium). Standards of dimethyl adipate (DMeP), dimethyl phthalate (DMP), bis(2-methoxyethyl) phthalate (DMEP), bis(2-ethoxyethyl) phthalate (DEEP), diethyl phthalate (DEP), diallyl phthalate (DAP),

111

diisopropyl phthalate (DIPrP), dipropyl phthalate (DPrP), diphenyl phthalate (DPhP), dibutyl phthalate (DBP), diisobutyl phthalate (DIBP), bis(2-butoxyethyl) phthalate (DBEP), dibenzyl phthalate (DBeP), benzyl butyl phthalate (BBP), dibutyl adipate (DBuP), bisiso-pentyl ester (DIPP), dipentyl phthalate (DPP), dicyclohexyl phthalate (DCHP), bis(4-methylpentyl) phthalate (BMPP), dihexyl phthalate (DHXP), diheptyl phthalate (DHP), bis(2-ethylhexyl) adipate (DEeP), diisononyl-phthalate (DINP), dinonyl phthalate (DNP), diisodecyl-o-phthalate (DIDP), dioctyl phthalate (DNOP) and bis(2-ethylhexyl) phthalate (DEHP) were purchased from Dr. Ehrenstorfer (Augsburg, Germany). The purity of all references compounds were >98%. Stock solutions of individual compounds were prepared in MeOH (1000 mg mL−1 ) and stored at −20 ◦ C in the dark. Then, a multicompound working standard solution at a concentration of 100 mg L−1 of each compound was prepared by combining suitable aliquots of each individual standard stock solution and diluting them with appropriate amounts of MeOH and stored in screwcapped glass tubes at −20 ◦ C in the dark. Special care was taken to avoid the contact of solvents and reagents with plastic materials. To minimize the risk of secondary contamination, glass materials were used in place of plastic materials. All glassware was cleaned prior to the analysis according to the recommendations specified in U.S. EPA Method 506. All solvents were checked for the presence of phthalates before use. 2.2. Instrumentation The UHPLC/ESI Q-Orbitrap system consisted of an Accela 1250 LC pump and a CTC Analytics PAL open autosampler coupled with a Q Exactive mass spectrometer (Thermo Fisher Scientific, Bremen, Germany). The system was controlled by Exactive Tune 1.1 and Xcalibur 2.2 software (Thermo Fisher Scientific, San Jose, USA). 2.3. Analytical procedure 2.3.1. Sample preparation After homogenization on a Polytron PT-2000 (Kinematica, Switzerland) for 30 s, 15.0 g of each sample (milk, milk beverages or yogurt sample) was weighed in glass centrifuge tube (50 mL), fortified with the 27 different phthalates and let to stand for 15 min. 10 mL volume of MeCN with 1% acetic acid was added as an extraction solvent and the tube was tightly capped and vigorously mixed for 1 min using a vortex (Scientific Industries, New York, USA) mixer at maximum speed. MgSO4 (6 g), anhydrous sodium acetate (1.45 g) and ceramic homogenizers were added to the tube, to induce phase separation. After that, the tube was immediately shaken for 1 min, and then centrifuged for 5 min at 4000 rcf (relative centrifugation force) at 4 ◦ C (Beckman Couler, Brea, USA). Then the upper layer (8 mL) was submitted to a dispersive SPE clean up with a mixture of 1.2 g of MgSO4 , 405 mg of PSA and 95 mg of C18 . The glass tube was vortexed for 1 min and centrifuged for 5 min at 4000 rcf at 4 ◦ C. An aliquot of the final upper layer (200 ␮L) was transferred into a Mini-UniPrep vial, 300 ␮L MeOH and 500 ␮L 8 mM ammonium formate buffer were added. After the vial was capped, vortexed for 30 s. Finally the extract was taken for UHPLC/ESI Q-Orbitrap analysis. Methods blanks were prepared in the same way using prescreened water instead of milk sample. 2.3.2. Experimental design for response surface methodology (RSM) Response surface methodology (RSM) was employed to investigate the variations in recovery rates with respect to the preparation of conditions including extraction solvent volume, the amounts of sodium acetate, PSA, and C18 . The optimal composition of the four variables was determined by using a central composite design

112

W. Jia et al. / J. Chromatogr. A 1362 (2014) 110–118

Table 1 Variables and levels evaluated in the central composite design to optimize the extraction conditions. Independent variable

Unit

Symbol

Extraction solvent volume Na Acetate quantity PSA quantity C18 quantity

mL g mg mg

X1 X2 X3 X4

(CCD) approach. In this work, the full CCD consisted of (1) a complete two-factorial design; (2) n0 , center point (n0 > 1), and (3) two axial points on the axis of each design variable at a distance of ˛ = 2.000 from the design center. Hence, a total number of design points of N = 2k + 2k + n0 was used. The actual variable was coded to facilitate multiple regression analysis. The complete design consisted of 30 combinations including seven replicates of the center point with five degrees of freedom for calculation of errors in the experiments. The optimal values of response Y (individual recovery of interest compounds) were obtained by solving the regression equation and by analyzing the response surface contour plots. Table 1 indicates the coded and CCD-processed variables for the optimization of the QuEChERS method for samples. The resulting 30 experiments were carried out randomly. The goodness of fit of the regression model and the significance of parameter estimates were determined through appropriate statistical methods. Design Expert trial version 8.0.6.0 was used (Stat-Ease Inc., Minneapolis, MN). 2.3.3. Instruments and analytical conditions The analytical column used was a 100 mm × 2.1 mm, i.d., 2.6 ␮m, Thermo Accucore C-18 aQ connected to a 10 mm × 2.1 mm, Accucore C-18 aQ guard column (Thermo Fisher Scientific, San Jose, USA). A mobile phase consisting of eluent A (water, 0.1% FAc, 4 mM ammonium formate) and eluent B (MeOH, 0.1% FAc, 4 mM ammonium formate) was used at a flow rate of 0.3 mL min−1 . The following gradient was used: 0 min 100% A, 1 min 100% A, 7 min 0% A, 12 min 0% A, 13 min 100% A, until the end of the run at 15 min. The optimized sample injection volume was set at 5 ␮L. All the 27

Coded levels −˛

−1

2 0.5 0 0

6 1.0 200 50

0

+1

+˛

10 1.5 400 100

14 2.0 600 150

18 2.5 800 200

analytes eluting over 0–11 min while the last 4 min were used for column cleaning and re-equilibration (Fig. S1, Electronic Supplementary Material). Supplementary Fig. S1 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chroma.2014.08.030. The mass spectrometer Q-Orbitrap was equipped with a Heated Electrospray Ionization (HESI) source. The optimized HESI temperature was set at 350 ◦ C, the capillary temperature at 320 ◦ C, the electrospray voltage at 3.5 kV for positive mode. Sheath and auxiliary gas were 18 and 3 L min−1 . All quantitative data in this study were acquired using full MS scan mode. In full MS/dd-MS2 (Top 5), which is used for confirmatory purpose, the Q-Orbitrap performs data-dependent scans. As long as the targeted compounds were detected (a list of targeted accurate masses), precursor ions that were selected by the quadrupole were sent to the HCD collision cell of the Q-Orbitrap mass spectrometer. Here, they were fragmented with normalized collision energy (NCE) to obtain product ion spectra. At this stage, the mass resolution was set at 17,500 FWHM (m/z 200) and NCE 35%. Using this strategy, coeluting matrix compounds from the matrix or noisy peaks can be easily excluded, facilitating the identification and quantification of known or unknown analytes in a single run analysis. 3. Results and discussion 3.1. Optimization of the LC-Q-Orbitrap conditions The optimum mass spectrometric parameters for the identification and quantification of the 27 analytes were obtained from

Table 2 UHPLC/ESI Q-Orbitrap parameters of the 27 analytes. Peak

Compound

RT (min)

Elemental composition

Precursor ion

Exact mass (m/z)

Accurate mass (m/z)

Accuracy (ppm)

Fragment Theoretical 1 (m/z)

Fragment 2

Theoretical (m/z)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

DMeP DMP DMEP DEEP DEP DAP DIPrP DPrP DPhP DBP DIBP DBEP DBeP BBP DBuP DIPP DPP DCHP BMPP DHXP DHP DEeP DINP DNP DIDP DNOP DEHP

5.87 6.13 6.17 6.83 6.87 7.20 7.29 7.50 7.82 8.02 8.02 8.02 8.02 8.05 8.07 8.09 8.49 8.53 8.68 8.84 9.28 9.50 9.82 9.82 10.34 10.49 10.54

C8 H14 O4 C10 H10 O4 C14 H18 O6 C16 H22 O6 C12 H14 O4 C14 H14 O4 C14 H18 O4 C14 H18 O4 C20 H14 O4 C16 H22 O4 C16 H22 O4 C20 H30 O6 C22 H18 O4 C19 H20 O4 C14 H26 O4 C18 H26 O4 C18 H26 O4 C20 H26 O4 C20 H30 O4 C20 H30 O4 C22 H34 O4 C22 H42 O4 C26 H42 O4 C26 H42 O4 C28 H46 O4 C24 H38 O4 C24 H38 O4

[M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+H]+ [M+NH4 ]+ [M+NH4 ]+

175.09649 195.06519 283.11761 311.14892 223.09649 247.09649 251.12779 251.12779 319.09649 279.15909 279.15909 367.21152 347.12779 313.14344 259.19039 307.19039 307.19039 331.19039 335.22169 335.22169 363.25299 371.31559 419.31559 419.31559 447.34689 408.31084 408.31084

175.09680 195.06491 283.11796 311.14862 223.09632 247.09671 251.12764 251.12763 319.09618 279.15936 279.15891 367.21158 347.12770 313.14357 259.19046 307.19021 307.19035 331.19053 335.22177 335.22153 363.25324 371.31524 419.31573 419.31548 447.34726 408.31066 408.31094

1.76 1.43 1.22 0.98 0.75 0.88 0.60 0.62 0.98 0.97 0.64 0.15 0.26 0.43 0.27 0.58 0.13 0.42 0.23 0.49 0.69 0.94 0.33 0.26 0.82 0.43 0.25

C6 H7 O2 C9 H7 O3 C3 H7 O C4 H9 O C8 H5 O3 C11 H9 O3 C11 H13 O4 C8 H5 O3 C14 H9 O3 C8 H5 O3 C8 H5 O3 C6 H13 O C8 H5 O3 C8 H5 O3 C6 H7 O2 C8 H5 O3 C8 H5 O3 C8 H5 O3 C8 H5 O3 C8 H5 O3 C8 H5 O3 C6 H11 O4 C8 H5 O3 C8 H5 O3 C8 H5 O3 C8 H7 O4 C8 H7 O4

C6 H11 O2 C8 H5 O2 C11 H11 O4 C12 H13 O4 C10 H9 O3 C12 H21 O4 C8 H5 O3 C11 H13 O4 C6 H7 O C12 H13 O3 C12 H15 O4 C14 H17 O4 C7 H7 O C12 H13 O3 C5 H9 O2 C13 H15 O3 C9 H7 O3 C8 H7 O4 C8 H7 O4 C8 H7 O4 C8 H6 O3 C7 H11 O3 C17 H25 O4 C7 H5 O2 C20 H38 O2 C16 H23 O4 C8 H5 O3

115.07590 133.02895 207.06519 221.08084 177.05462 229.14344 149.02332 209.08138 95.04969 205.08647 223.09703 249.11268 107.04969 205.08647 101.06025 219.10212 163.03952 167.03443 167.03443 167.03443 150.03169 143.07082 293.17528 121.02895 310.28718 279.15963 149.02332

111.04460 163.03952 59.04914 73.06479 149.02332 189.05462 209.08138 149.02332 225.05517 149.02332 149.02332 101.09664 149.02332 149.02332 111.04460 149.02332 149.02332 149.02332 149.02332 149.02332 149.02332 147.06573 149.02332 149.02332 149.02332 167.03443 167.03443

W. Jia et al. / J. Chromatogr. A 1362 (2014) 110–118

113

Fig. 1. Effect of type of solvent on the extraction recovery (n = 7 batches) of phthalates in milk. The evaluation was done at 50 ␮g kg−1 equivalent in samples.

analyzing the compounds by flow injection analysis respectively. Sensitivity of target analytes was checked by recording chromatograms in full scan method in both positive and negative ionization mode. Due to adduct formation with FAc/ammonium formate buffer, some analytes exhibit strong ammonium adduct species ([M+NH4 ]+ ) which appear to be the most predominant ions in the mass spectrum. Table 2 summarizes the optimal parameters of the UHPLC/ESI Q-Orbitrap. In Full MS/dd-MS2 mode, the first scan event is a Full MS and the second is a dd-MS2 , where the ion specified in the inclusion list is selected by quadrupole. An isolation width of 1.5 Da was used for the quadrupole. In this study, N corresponds directly to loop count (N = multiplex × loop count, multiplex was set as 1.). Therefore, in Top 1 only triggered the mass in the inclusion list within 10 ppm mass deviation at a particular

retention time and above the threshold. A Top N approach was optimized such that in a given time, when N abundant masses in the inclusion list within 10 ppm range are detected in the survey scan, N times dd-MS2 are triggered to select the ion in the inclusion list, fragmented in HCD cell, collected in the C-trap and analyzed in the Orbitrap, however, due to the N multiple ions fragmentation all entered Orbitrap at once, the mixed fragmentation spectrum may not be able to compare to the known fragment pattern. AGC (Automatic Gain Control) target and maximum ion time is used to control the number of ions and maximum time taken to fill the C-trap, providing a duty cycle around 1.5 s. Chromatographic conditions were studied in order to achieve the best separation and retention for the compounds. First, several experiments were performed on different mobile phases

Table 3 Validation parameters of the developed method for milk sample. Compound

Extraction recoveries %a

DMeP DMP DMEP DEEP DEP DAP DIPrP DPrP DPhP DBP DIBP DBEP DBeP BBP DBuP DIPP DPP DCHP BMPP DHXP DHP DEeP DINP DNP DIDP DNOP DEHP

104.6 96.3 92.7 99.0 94.9 98.2 96.1 102.6 98.3 97.5 95.5 97.6 92.5 90.7 102.0 100.6 97.2 102.3 102.2 99.6 101.3 95.4 100.1 102.5 97.7 98.3 99.4

a b

Calibration equation (y=)

−965 + 33268x −796 + 17893x −867 + 48319x 653 + 51248x −392 + 32397x −430 + 22383x −649 + 31424x −494 + 23803x −380 + 33704x −667 + 44827x 503 + 34385x −783 + 20918x −684 + 21230x −730 + 41638x 447 + 39051x −663 + 20891x −886 + 15781x −389 + 32307x −601 + 40912x −845 + 28945x −489 + 46757x −325 + 28946x 4341 + 59070x 356 + 28935x −3123 + 19673x −564 + 30892x −934 + 11207x

Average of three concentration levels: CC␤, 2 CC␤, 4 CC␤. Interday precision.

Dynamic range (␮g kg−1 )

0.1–100 0.1–100 0.1–100 0.1–100 0.1–100 0.1–100 0.1–100 0.1–100 0.1–100 0.1–100 0.1–100 0.1–100 0.1–100 0.1–100 0.1–100 0.1–100 0.1–100 0.1–100 0.1–100 0.1–100 0.1–100 0.1–100 0.5–500 0.1–100 0.5–500 0.1–100 1–1000

CC␣ (␮g kg−1 )

CC␤ (␮g kg−1 )

r2

RSD% at three levels (n = 9 days)b

Level 1

Level 2

Level 3

0.15 0.19 0.25 0.16 0.26 0.13 0.29 0.26 0.17 0.20 0.27 0.32 0.30 0.28 0.34 0.27 0.31 0.20 0.33 0.30 0.14 0.24 0.96 0.18 1.04 0.30 0.16

0.38 0.48 0.62 0.40 0.64 0.32 0.72 0.66 0.42 0.50 0.68 0.80 0.74 0.70 0.84 0.68 0.78 0.50 0.82 0.76 0.36 0.60 2.40 0.46 2.60 0.74 0.40

0.9998 0.9993 0.9991 0.9992 0.9998 0.9993 0.9991 0.9992 0.9994 0.9999 0.9998 0.9996 0.9993 0.9994 0.9995 0.9992 0.9997 0.9998 0.9993 0.9999 0.9997 0.9995 0.9992 0.9993 0.9997 0.9999 0.9997

5.0 2.1 3.2 4.4 5.0 2.1 0.9 3.3 3.8 0.8 2.8 3.0 1.2 0.6 2.0 1.9 2.5 4.2 1.0 2.7 3.4 2.6 5.6 1.8 3.6 2.6 3.4

1.7 2.0 4.8 0.6 5.0 1.9 2.1 3.0 3.3 2.1 3.5 2.0 2.9 0.5 2.6 3.9 2.1 4.4 3.9 3.1 0.4 3.1 3.8 2.0 3.0 2.7 3.1

0.9 4.2 3.1 1.7 2.8 5.9 1.5 4.4 2.1 5.0 0.9 1.9 2.3 3.8 1.0 3.6 2.8 1.9 3.8 4.4 1.9 4.2 1.5 0.7 2.7 4.2 3.0

114

W. Jia et al. / J. Chromatogr. A 1362 (2014) 110–118

consisting of MeCN or MeOH as organic solvent and water as polar solvent with different concentrations of acetic and formic acid (from 0.05 to 0.5%), ammonium formate, ammonium acetate (from 1 mM to 10 mM). MeOH was chosen as the organic phase, because it less polar and achieved a better resolution than MeCN. On the other hand, the addition of FAc–ammonium formate provided better results than acetic acid–ammonium acetate and it was used to improve the ionization efficiency and achieve better chromatography. Finally, the best results were obtained when MeOH was used as organic modifier and aqueous solution of FAc (0.1%)–ammonium formate (4 mM) was employed. We evaluated various HPLC columns (Thermo Scientific Accucore: aQ C18 , RP-MS, C18 , PFP, Phenyl-Hexyl) in order to optimize the chromatographic separation of the above mentioned 27 analytes and the aQ C18 column yielded the best results. Several gradient profiles were studied, obtaining good response with the gradient described in Section 2. Under these conditions retention times of the analytes were constant, ranging from 5.87 (DMeP) to 10.54 (DEHP) min. 3.2. Optimization of the extraction procedure Despite QuEChERS uses MeCN acidified with acetic acid, and bearing in mind that conventional extraction procedures of phthalates from different samples use MeOH or acetone, the extraction solvent was evaluated first, using the extraction procedure described in Section 2.3, except that the type of solvent was varied. Thus, different solvents such MeCN, MeOH, and acetone were checked. Commission Decision 2002/657/EC and 2007/19/EC were used as guidelines for the calculation of recoveries and matrix effects. The obtained results are shown in Fig. 1. It can be observed many analytes had low recovery or could not be recovered with the mixture containing acetone. Furthermore, often a pronounced matrix-effect and RSD > 10% were observed. Probably, acetone causes a very rapid denaturation of proteins, with possible formation of aggregates that could adsorb/occlude the compounds of interest, reducing extraction reproducibility. In terms of matrixeffect and extraction recoveries, MeCN was found to be the most suitable extraction solvent across all tested milk, milk beverages and yogurt samples. First, MeCN, more polar solvent than MeOH and acetone, enables a better solubility and extraction efficiency of phthalate tightly bound to the matrix. Second, during the extraction process, MeCN with a relatively lower surface tension supported a more lasting suspension for interfacial transfer. The MeCN/water mixture performed worse in term of matrix interferences, the final extracts were turbid, probably due to insufficient protein precipitation and the amount of fat present in the final extract. Critical factors which could have an influence on the analyte recovery and method sensitivity were investigated by a central composite design (CCD). In general, higher amount of extraction solvent leads to a higher analytical signal in almost all analytes, but there was no preliminary study performed testing different ratios of mass (g) of sample per volume (mL) of extraction solvent. Sodium acetate has a dissolving effect on milk protein and fat globules, which could affect recovery rates. The PSA and C18 sorbent, as a dispersive medium, are used to retain those nondesirable components (organic acids, polar pigments, sugars and fatty acids) co-extracted from the matrix while target compounds remain in solution. Therefore, the factors included (1) volume of extraction solvent, (2) amount of sodium acetate, (3) amount of PSA, (4) amount of C18 . The individual recoveries of all analytes were introduced separately as the response in the statistical program. Fig. 2 shows the different response plots for DPhP (as an example) for different combinations of the parameters investigated. The optimum conditions were chosen by taking into consideration data obtained from the response surface plots and the regression coefficient

Fig. 2. Response surface plot for DPhP.

plots. Therefore, the response surfaces generated suggest that the best extraction conditions for DPhP were extraction volume 10 mL, sodium acetate 1.45 g, PSA 405 mg, C18 95 mg. Multiresidue methods for recovery were tested under the conditions optimized for DPhP, which gave the highest observed recovery in the single recovery tests and were therefore used to test the multiresidual compound quantification. The results of the multiresidue methods recovery test and validation information for each analyte are shown in Table 3. In summary, the resulting conditions allowed reliable simultaneous analysis of 27 of the target analytes with recoveries in the range of 90.7–104.6%. 3.3. Development of screening and confirmation methods In order to increase sample throughput, only one injection was carried out and fragment spectra were acquired in order to distinguish between negative and positive samples (screening method) and to confirm the presence of the analytes (confirmation method). Full MS scan mode allowed for screening and quantifying the analytes or retrospectively looking into unknowns. The identification of the target compounds was based on the measurement of the accurate mass and the retention time (RT). The RT window tolerance was set at a very conservative ± 3 SD (RT average minus or plus three times the standard deviation of the RT). The effect of the mass extraction window on selectivity for analytes in milk, milk beverages and yogurt matrices at a concentration 50 ␮g kg−1 (2–10 ppm

W. Jia et al. / J. Chromatogr. A 1362 (2014) 110–118

115

Fig. 3. Extracted ion chromatogram and spectra from a full MS/dd-MS2 experiment using UHPLC-Q-Orbitrap, corresponding to the theoretical extracted ion chromatograms (displayed as a stick per scan) the protonated molecule of DBP and DIBP (m/z 279.15909) and fragments in a spiked yogurt sample (50 ␮g kg−1 ).

mass deviations) was tested. The best results were obtained when mass extraction windows of 3.0 ppm were employed, providing a high selectivity and reduced probability of false positives. If a signal within the RT window showing a mass error < 3.0 ppm was found within the RT window, and the intensity of this signal was equal or above the lower value of the uncertain region (8.3 × 104 ), this sample was considered as a non-negative sample and it was reprocessed to confirm the results by the confirmation method. On the contrary, if no signal was found within the RT window or the measured mass error was higher than 3.0 ppm, the sample was considered negative and it was not re-processed by the confirmation method. In order to develop the confirmation method, it was necessary to establish fragments for each compound. When operated in full MS/dd-MS2 mode, a product ion spectrum with accurate mass measurement is obtained automatically according to inclusion list, and this defined as a data dependent acquisition (dd-MS2 ). The resolving power (R) is a critical parameter which has an impact on the correct assignment of the masses for the analytes. To eliminate matrix interferences the resolving power of the Q-Orbitrap was evaluated. With an R of 70,000 FWHM (17,500 FWHM for ddMS2 ) no overlapping in masses was observed between analytes and matrix components. Furthermore, it is important to notice, that despite some structural analogs and multiple isomers can coelute, the fragments used for their confirmation are characteristic of each one. Thus, DBP and DIBP are co-eluting multiple isomers but unequivocal fragments (m/z 223.09703, 205.08647, 95.04969, 59.04914) have been selected for their confirmation (Fig. 3). Specific fragments have been found for each analytes, minimizing the possibility of a wrong compound confirmation. On the other hand, DIPP and DPP have a common fragments (m/z 149.02332, 219.10212,

163.03952), but they do not co-elue (Table 2). In order to minimize probability of false positives, the non-negative samples were carried out using isotopic patterns. The relative isotope abundance (RIA) of the M+1 (being M the corresponding [M+H]+ ) peak, which is mainly due to the presence of 13 C, was useful to confirm analytes with a high number of carbons. 3.4. Validation of the proposed method One of the main problems involved in the determination of phthalates is laboratory contamination and the blank results were always subtracted to correct experimental values. All the tests were performed in triplicate and an internal quality control was analyzed with each batch of samples in order to measure background levels within the extraction method. Method validation was performed in terms of matrix-effect, specificity, linearity, trueness, precision, CC␣ and CC␤. Commission Decision 2002/657/EC and 2007/19/EC were used as guidelines for the validation studies. The matrix effect was tested by comparing the slopes of the matrix-free calibration curves to the matrix-matched calibration curves. Samples were first extracted according to the procedure as described the Experimental section. Matrix effect was investigated by calculating the percentage (C%) of signal enhancement or suppression, according to Eq. (1). C% =



1−

ss sm



× 100

(1)

where ss is the slope of calibration plot with matrix-matched calibration solutions and sm is the slope of calibration plot with calibration solutions in solvent.

116

W. Jia et al. / J. Chromatogr. A 1362 (2014) 110–118

Fig. 4. UHPLC/ESI Q-Orbitrap matrix effects of six milk and milk products. The evaluation was done at 500 ␮g kg−1 equivalent in samples.

The milk, milk beverages and yogurt matrix effects on analytes were not significant, only DINP, DNP and DMeP showed ion suppression of +50% [32]. Apparently, QuEChERS was as effective for phthalates extraction and sample cleanup for milk, milk beverages and yogurt. It is interesting to notice that the matrix effects from different types of milk, milk beverages and yogurt showed a similar ion enhancement or suppression profile (Fig. 4). Therefore, any milk, milk beverages and yogurt may be chosen as a standard matrix to construct matrixmatched calibration curves for quantification in routine practice. Extraction recoveries were assessed by spiking blank milk, milk beverages and yogurt samples before and after extraction at three concentration levels (CC␤, 2 CC␤, 4 CC␤) with five replicates at each level. The matrix matched standard calibration curves from the extracts were therefore evaluated with 11 calibration levels S0–S11, namely 0, 0.1, 0.5, 1, 10, 50, 100, 200, 500, 750, 1000 ␮g kg−1 , where “0” corresponded to the extract from the sample analyzed as such. For all analytes, r2 was greater than 0.9991 and the deviation of each point from the calibration line was lower than 15%. Specificity was assessed by verifying the presence of interference at the retention time of analytes greater than a signal-to-noise ratio of three. Within-laboratory reproducibility was assessed by spiking blank milk, milk beverages and yogurt samples at four different concentrations on nine different days. Trueness was calculated as the percentage of error between spiked and found concentrations. CC␣ was estimated from the calibration curve prepared by spiking blank milk, milk beverages and yogurt matrices at four concentration levels in the low concentration range. CC␣ is calculated as the concentration corresponding to the y-intercept plus 2.33 times its standard deviation. In the case of CC␤, the concentration corresponds to CC␣ + 1.64 s, s being

the standard deviation obtained at the CC␣ level. CC␣ ranged between 0.13 ␮g kg−1 and 1.04 ␮g kg−1 , and CC␤ ranged between 0.32 ␮g kg−1 and 2.60 ␮g kg−1 . Compared with traditional methods, the sensitivity was enhanced, and the accuracy was improved, leading to a powerful method for screening phthalates in foods. The results of this validation are summarized in Table 3. Three consecutive injections on-column at 50 ␮g kg−1 were carried out to detect any decrease in mass accuracy. No significant decrease was observed and the maximum mass deviation ranged from 0.1 to 2.0 ppm. This minor deviation in mass accuracy demonstrated the wide dynamic range of Q-Orbitrap for qualitative analysis at a resolving power of 70,000 FWHM. The CC␣ values of the proposed Q-Orbitrap mass spectrometer method are comparable to the limits of the previously reported triple quadrupole mass spectrometer methods. Comparing with the detection limits reported in literatures [33,34], the detection sensitivity and accuracy was improved more than five times. 3.5. Sample analysis Once the proposed method was optimized and validated, it was applied to investigate the occurrence of the 27 phthalates in a total of 96 commercial milk, milk beverages and yogurt products. The results indicated that the DEHP was found in almost all samples tested in this study, with levels ranging from 1 to 936 ␮g kg−1 . In total, 55% (53/96) of the samples were contaminated with two or more phthalates. In pasteurized and homogenized milk samples, DEP, BBP, and DEHP were detected quite frequently with levels over 10 ␮g kg−1 . Slightly higher levels of BBP and DINP were found in milk beverage samples. As shown in Table 4, DEP, BBP, DEHP, DIDP and DINP were detected in milk, milk beverages and yogurt

W. Jia et al. / J. Chromatogr. A 1362 (2014) 110–118

117

Fig. 5. Examples of typical UHPLC/ESI Q-Orbitrap MS chromatograms from a full MS/dd-MS2 experiment: (A1) extracted ion chromatogram (displayed as a stick per scan) of DEP [M+H]+ m/z 223.09651 in sample no. 13; (A2) dd-MS2 total ion chromatogram of DEP [M+H]+ m/z 223.09651 in sample no. 13.

Table 4 Quantification results for target analytes in positive milk, milk beverages and yogurt samples analyzed by UHPLC/ESI Q-Orbitrap. Samplea

Compound

Concentration (␮g kg−1 )

RSD% (n = 9)

No. 13

DEP BBP DEHP

13 85 57

0.9 2.9 3.1

No. 19

BBP DINP

97 5

2.3 4.3

No. 33

BBP DIDP

95 5

1.3 5.1

No. 35

DEP BBP DEHP

13 19 42

0.9 2.3 2.4

a Type of dairy products are indicated herein: milk, no. 13; milk beverages, no. 19; yogurt, no. 33 and no. 35.

products. Fig. 5 shows the typical chromatograms from a full MS/dd-MS2 experiment of analytes detected in positive samples. With this UHPLC/ESI Q-Orbitrap method, not only accuracy was enhanced, but also the low concentration preservative, this suggested that the UHPLC–ESI-Q-Orbitrap-MS method was appropriate for the screening of phthalates in foods. 4. Conclusions A new analytical method has been developed and applied in routine for screening and quantitation of phthalates in milk, milk beverages and yogurt samples. DEHP was found in almost all milk, milk beverages and yogurt samples tested in this study, with levels ranging from 1 to 936 ␮g kg−1 . In summary, by combining

QuEChERS extraction procedure and UHPLC/ESI Q-Orbitrap, an accurate and highly sensitive method was developed to screen 27 phthalates in foods. Acknowledgments The present research was financially supported by the grants from the project of General Administration of Quality Supervision, Inspection and Quarantine of the People’s Republic of China (Project number: 2012104002; 2013IK198), and the Graduate Innovation Fund of Shaanxi University of Science and Technology. References [1] Y. Ait Bamai, A. Araki, T. Kawai, T. Tsuboi, I. Saito, E. Yoshioka, A. Kanazawa, S. Tajima, C. Shi, A. Tamakoshi, R. Kishi, Associations of phthalate concentrations in floor dust and multi-surface dust with the interior materials in Japanese dwellings, Sci. Total Environ. 468–469 (2014) 147–157. [2] A. Guart, F. Bono-Blay, A. Borrell, S. Lacorte, Effect of bottling and storage on the migration of plastic constituents in Spanish bottled waters, Food Chem. 156 (2014) 73–80. [3] A.C. Ionas, A.C. Dirtu, T. Anthonissen, H. Neels, A. Covaci, Downsides of the recycling process: harmful organic chemicals in children’s toys, Environ. Int. 65 (2014) 54–62. [4] S. Langer, G. Bekö, C.J. Weschler, L.M. Brive, J. Toftum, M. Callesen, G. Clausen, Phthalate metabolites in urine samples from Danish children and correlations with phthalates in dust samples from their homes and daycare centers, Int. J. Hyg. Environ. Health 217 (2014) 78–87. [5] J. Lee, J.H. Lee, C.K. Kim, M. Thomsen, Childhood exposure to DEHP, DBP and BBP under existing chemical management systems: a comparative study of sources of childhood exposure in Korea and in Denmark, Environ. Int. 63 (2014) 77–91. [6] T.L. Pan, P.W. Wang, I.A. Aljuffali, Y.Y. Hung, C.F. Lin, J.Y. Fang, Dermal toxicity elicited by phthalates: evaluation of skin absorption, immunohistology, and functional proteomics, Food Chem. Toxicol. 65 (2014) 105–114. [7] I.J. Wang, C.C. Lin, Y.J. Lin, W.S. Hsieh, P.C. Chen, Early life phthalate exposure and atopic disorders in children: a prospective birth cohort study, Environ. Int. 62 (2014) 48–54.

118

W. Jia et al. / J. Chromatogr. A 1362 (2014) 110–118

[8] Q. Zheng, M. Feng, Y. Dai, Comparative antioxidant responses in liver of Carassius auratus exposed to phthalates: an integrated biomarker approach, Environ. Toxicol. Pharmacol. 36 (2013) 741–749. [9] V. Chopra, K. Harley, M. Lahiff, B. Eskenazi, Association between phthalates and attention deficit disorder and learning disability in U.S. children, 6–15 years, Environ. Res. 128 (2014) 64–69. [10] O. Favrat, B. Mangote, I. Tovena-Pécault, J. Néauport, Study of organic contamination induced by outgassing materials. Application to the Laser MégaJoule optics, Appl. Surf. Sci. 293 (2014) 132–137. [11] S.H.W. Schulpen, J.L.A. Pennings, E.C.M. Tonk, A.H. Piersma, A statistical approach towards the derivation of predictive gene sets for potency ranking of chemicals in the mouse embryonic stem cell test, Toxicol. Lett. 225 (2014) 342–349. [12] L. Trasande, S. Sathyanarayana, M. Jo Messito, R.S. Gross, T.M. Attina, A.L. Mendelsohn, Phthalates and the diets of US children and adolescents, Environ. Res. 126 (2013) 84–90. [13] W. Völkel, M. Kiranoglu, R. Schuster, H. Fromme, Phthalate intake by infants calculated from biomonitoring data, Toxicol. Lett. 225 (2014) 222–229. [14] Y.M. Ho, Y.K. Tsoi, K.S.Y. Leung, Ionic-liquid-based dispersive liquid–liquid microextraction for high-throughput multiple food contaminant screening, J. Sep. Sci. 36 (2013) 3791–3798. [15] G. Di Bella, A.G. Potortì, V. Lo Turco, M. Saitta, G. Dugo, Plasticizer residues by HRGC–MS in espresso coffees from capsules, pods and moka pots, Food Control 41 (2014) 185–192. [16] P. Gimeno, S. Thomas, C. Bousquet, A.F. Maggio, C. Civade, C. Brenier, P.A. Bonnet, Identification and quantification of 14 phthalates and 5 non-phthalate plasticizers in PVC medical devices by GC–MS, J. Chromatogr. A 949–950 (2014) 99–108. [17] S. Orecchio, R. Indelicato, S. Barreca, Determination of selected phthalates by gas chromatography–mass spectrometry in mural paintings from Palermo (Italy), Microchem. J. 114 (2014) 187–191. [18] J.H. Hu, T. Feng, W.L. Li, H. Zhai, Y. Liu, L.Y. Wang, C.L. Hu, M.X. Xie, Surface molecularly imprinted polymers with synthetic dummy template for simultaneously selective recognition of nine phthalate esters, J. Chromatogr. A 1330 (2014) 6–13. [19] M. Llompart, M. Celeiro, J. Pablo Lamas, L. Sanchez-Prado, M. Lores, C. GarciaJares, Analysis of plasticizers and synthetic musks in cosmetic and personal care products by matrix solid-phase dispersion gas chromatography–mass spectrometry, J. Chromatogr. A 1293 (2013) 10–19. ˜ ˜ Y. Pico, P. Campíns-Falcó, [20] A. Masiá, Y. Moliner-Martinez, M. Munoz-Ortu no, Multiresidue analysis of organic pollutants by in-tube solid phase microextraction coupled to ultra-high performance liquid chromatography–electrospraytandem mass spectrometry, J. Chromatogr. A 1306 (2013) 1–11. [21] G. Tranfo, B. Papaleo, L. Caporossi, S. Capanna, M. De Rosa, D. Pigini, F. Corsetti, E. Paci, Urinary metabolite concentrations of phthalate metabolites in Central Italy healthy volunteers determined by a validated HPLC/MS/MS analytical method, Int. J. Hyg. Environ. Health 216 (2013) 481–485.

[22] D. Xu, X. Deng, E. Fang, X. Zheng, Y. Zhou, L. Lin, L. Chen, M. Wu, Z. Huang, Determination of 23 phthalic acid esters in food by liquid chromatography tandem mass spectrometry, J. Chromatogr. A 1324 (2014) 49–56. [23] J. Li, Q. Su, K.-Y. Li, C.-F. Sun, W.-B. Zhang, Rapid analysis of phthalates in beverage and alcoholic samples by multi-walled carbon nanotubes/silica reinforced hollow fibre-solid phase microextraction, Food Chem. 141 (2013) 3714–3720. ˜ [24] H. Gallart-Ayala, O. Núnez, P. Lucci, Recent advances in LC–MS analysis of food-packaging contaminants, TrAC Trend. Anal. Chem. 42 (2013) 99–124. [25] M. Aragón, R.M. Marcé, F. Borrull, Determination of N-nitrosamines and nicotine in air particulate matter samples by pressurised liquid extraction and gas chromatography–ion trap tandem mass spectrometry, Talanta 115 (2013) 896–901. [26] A. Delgado, O. Posada-Ureta, M. Olivares, A. Vallejo, N. Etxebarria, Silicone rod extraction followed by liquid desorption-large volume injectionprogrammable temperature vaporiser-gas chromatography–mass spectrometry for trace analysis of priority organic pollutants in environmental water samples, Talanta 117 (2013) 471–482. [27] D. Huang, X. Wang, C. Deng, G. Song, H. Cheng, X. Zhang, Facile preparation of raisin-bread sandwich-structured magnetic graphene/mesoporous silica composites with C18-modified pore-walls for efficient enrichment of phthalates in environmental water, J. Chromatogr. A 1325 (2014) 65–71. [28] K. Servaes, S. Voorspoels, J. Lievens, B. Noten, K. Allaerts, H. Van De Weghe, G. Vanermen, Direct analysis of phthalate ester biomarkers in urine without preconcentration: method validation and monitoring, J. Chromatogr. A 1294 (2013) 25–32. [29] B.J.A. Berendsen, L.A.M. Stolker, M.W.F. Nielen, Selectivity in the sample preparation for the analysis of drug residues in products of animal origin using LC–MS, TrAC Trend. Anal. Chem. 43 (2013) 229–239. [30] M.J.M. Bueno, C. Boillot, H. Fenet, S. Chiron, C. Casellas, E. Gómez, Fast and easy extraction combined with high resolution-mass spectrometry for residue analysis of two anticonvulsants and their transformation products in marine mussels, J. Chromatogr. A 1305 (2013) 27–34. [31] N. León, M. Roca, C. Igualada, C.P.B. Martins, A. Pastor, V. Yusá, Wide-range screening of banned veterinary drugs in urine by ultra high liquid chromatography coupled to high-resolution mass spectrometry, J. Chromatogr. A 1258 (2012) 55–65. [32] J. Wang, W. Chow, D. Leung, Applications of LC/ESI-MS/MS and UHPLC/QqTOF-MS for the determination of 141 pesticides in tea, J. AOAC Int. 94 (2011) 1685–1714. [33] W. Dirwono, Y.S. Nam, H.M. Park, K.B. Lee, LC–TOF/MS determination of phthalates in edible salts from food markets in Korea, Food Addit. Contam. 6 (2013) 203–208. [34] X. Li, W. Xiong, H. Lin, L. Zhuo, S. Lv, X. Tang, M. Chen, Z. Zou, Z. Lin, B. Qiu, G. Chen, Analysis of 16 phthalic acid esters in food simulants from plastic food contact materials by LC–ESI-MS/MS, J. Sep. Sci. 36 (2013) 477–484.