Article pubs.acs.org/est
Ubiquitous Occurrence of Chlorinated Byproducts of Bisphenol A and Nonylphenol in Bleached Food Contacting Papers and Their Implications for Human Exposure Yuyin Zhou, Mo Chen, Fanrong Zhao, Di Mu, Zhaobin Zhang, and Jianying Hu* Laboratory for Earth Surface Processes, College of Urban and Environmental Sciences, Peking University, Beijing 100871, China S Supporting Information *
ABSTRACT: The occurrence of bisphenol A (BPA), nonylphenol (NP), and their six chlorinated byproducts were investigated in 74 food contacting papers (FCPs) from China, the U.S.A., Japan, and Europe using a sensitive dansylation LC-MS/MS method. BPA ( > > > > > > > > > > > > > > > > > >
171b 156 171b 156 171b 156 171b 156 171b 156 171b 156 171b 156 156b 171 171b 156 156b 171 156b 171
cone voltage (V)
collision energy (eV)
60
30 40 30 55 50 60 58 60 58 60 47 60 70 60 60 45 60 60 60 60 60 45
60 52 55 55 68 62 50 50 45 40
a
BPA: bishphenol A; NP: nonylphenol; MCBPA: 4-chloro-BPA; DCBPA: dichloro-BPA (mixture of 2,6-dichloro-BPA and 2,6′-dichloro-BPA (1:0.25)); TCBPA: trichloro-BPA; TeCBPA: tetrachloro-BPA; MCNP: 2-chloro-NP; DCNP: 2,6-dichloro-NP. bMRM transition used for quantitation.
Migration of Chemicals from Food Contacting Paper to Coffee and Simulant. The migration of BPA, NP, and their chlorinated derivatives from coffee filter (FCP B4) to coffee solution was investigated. Samples were prepared by a stainless steel coffee maker. The stainless steel coffee maker was rinsed with acetone and hexane prior to use. One piece of FCP B4 (8.2 × 2.1 × 13 cm3) was put into the filter basket and 20 g of commercially available ground coffee and 150 mL water were added to make one cup of coffee. Then the coffee solutions were collected and concentrated by use of a previously developed method.9 Briefly, the samples were passed through glass fiber filters and then loaded onto Oasis HLB cartridges (6 cc, 200 mg, Waters), which were preconditioned with 20 mL of MTBE, 20 mL of MeOH, and 5 mL of ultrapure water, at a flow rate of 5−10 mL/min. Cartridges were then washed with 5 mL of ultrapure water and dried under a flow of nitrogen gas. A volume of 4 mL of MTBE/MeOH (1:1, v/v) was used to elute the analytes from the column. Extracts were purified and dansylated according to the method mentioned above in the analysis of FCPs and redissolved in 0.1 mL acetonitrile prior to UPLC−MS/MS analysis. Pure coffee, prior to contact with the coffee filter, was also processed and analyzed by use of the same method mentioned above. A total of six replicates were analyzed. Migration test from FCP B4 to simulant was also carried out as described in the SI. The migration rates in both tests were estimated by the following equation. MR(%) =
masstotal − mass pure conc.B4 × mass B4
six replicates were analyzed both in migration samples and samples uncontacted with FCP B4. UPLC−ESI−MS−MS Analysis. The UPLC apparatus was an Acquity Ultra Performance LC (Waters, Milford, U.S.A.). Target analytes were separated using a Waters Acquity UPLC BEH C18 column (100 mm × 2.1 mm × 1.7 μm) (Milford, U.S.A.). The column was maintained at 40 °C and a flow rate of 0.3 mL/min, and the injection volume was 5 μL. Acetonitrile and water containing 0.1% formic acid were chosen as the mobile phases. Gradient conditions were initiated with 60% acetonitrile followed by a linear increase to 75% acetonitrile in 0.5 min. After being increased to 80% in 6 min, acetonitrile was increased to 95% in 0.5 min and then to 100% in 2 min and kept isocratic for 2 min. Mass spectrometry was performed using a Premier XE tandem quadrupole mass spectrometer (Waters) equipped with a Z-Spray ionization (ESI) source and operated in the positive ion mode. The two most abundant multiselected reaction monitoring (MRM) transitions, cone voltages, and collision energies are summarized in Table 1. Common MS parameters were as follows: capillary voltage, 3.2 kV (ESI +); source temperature, 120 °C; desolvation temperature, 450 °C; source gas flow, 50 L/h; and desolvation gas flow, 800 L/h. Quantification and Quality Control. Identification of the target analytes was accomplished by comparing the retention time (within 2%) and the ratio (within 20%) of the two selected precursor ion-produced ion transitions with those of standards. Quantification was accomplished using the multiselected reaction monitoring (MRM) transitions and based on the recovery of internal standards (BPA-d4 for BPA; 2,2′dichloro-BPA-d12 for MCBPA, DCBPA, TCBPA, and TeCBPA; 4-n-NP for NP, MCNP, and DCNP). For substances that were not detected in some FCPs and coffee samples, the limits of detection (LODs) and limits of quantification (LOQs) were
× 100%
where MR is the mitigation rate; masstotal is the mass of a target analyte in the coffee or simulant which contacted with FCP B4; masspure is the mass of a target analyte in pure coffee or simulant without contacting with FCP B4; and massB4 is the mass of one piece of the coffee filter. For each migration test, 7220
DOI: 10.1021/acs.est.5b00831 Environ. Sci. Technol. 2015, 49, 7218−7226
Article
Environmental Science & Technology
Figure 1. UPLC−MS/MS MRM chromatograms of the ions selected for identification of analytes detected in samples of food contacting papers. BPA: bishphenol A; NP: nonylphenol; MCBPA: 4-chloro-BPA; DCBPA: dichloro-BPA (mixture of 2,6-dichloro-BPA and 2,6′-dichloro-BPA (1:0.25)); TCBPA: trichloro-BPA; TeCBPA: tetrachloro-BPA; MCNP: 2-chloro-NP; and DCNP: 2,6-dichloro-NP.
(RSD ≤ 5%) and correlation coefficients >0.995. The extent of the signal suppression and enhancement in UPLC-ESI-MS/MS analysis was evaluated by adding 20 μL derivatized acetonitrile spike with 0, 0.05, 0.5, and 2 ng of each analyte into 180 μL derivatized sample extracts. The standard addition curve was then compared to the working standards calibration curve to calculate matrix effects according to the following equation.
determined by spiking 0.05 ng/g of each analyte in FCPs and ground coffee samples. The LODs and LOQs of BPA and NP were calculated as three and ten times the standard deviation of the blank after the subtraction of the blank mean. Estimation on the LODs and LOQs were based on signal-to-noise ratio (S/N) observations with S/N of 3 and 10 for the transition reaction used for peak confirmation, i.e. the less sensitive one. The entire analytical procedures were checked for precision, intermediate reproducibility, blank contamination, linearity, and matrix effects. The recovery experiments were conducted by spiking three concentrations (0.05, 0.5, and 2 ng/g for analyzing method of FCPs and ground coffee) of all target analytes into the sample matrices (n ≥ 3 for each concentration level). Spiked samples were then subjected to the full processing and analytical procedures. Because BPA and NP are ubiquitous contaminants in the laboratory environment, only pretreated glassware and glass fiber filters (400 °C, 4 h) were used throughout the study. Aluminum foil was used in all plastic seals to avoid contact between sample (extract) and plastic. Silica SPE cartridges were prerinsed with 10 mL of water-saturated ethyl acetate and 10 mL of hexane/ethyl acetate (90:10, v/v) to minimize the contamination in the SPE procedure. In the analysis of FCPs, a total of four procedure blanks were analyzed. Procedural blanks were conducted by use of 6 mL of MeOH which was subjected to all processing and analytical procedures to determine background contamination. In the analysis of coffee samples, procedural blanks (n=3 for each) were prepared by substitution of 150 mL of Milli-Q water. Standard injections were done after five sample injections, and methanol was injected after each standard injection to monitor background contamination. The linearity of the instrument and method was demonstrated using a 12point (0.01, 0.05, 0.1, 0.5, 1, 5, 10, 50, 100, 500, 1000, and 5000 ng/L) calibration curve for each analyte with good precision
⎛ PA sa − PA unsp. ⎞ ME = ⎜ − 1⎟ × 100% PA ws ⎝ ⎠
where “ME” is the matrix effect, “PA” is the peak area, “sa” is the standard addition, “unsp.” is unspiked, and “ws” is the working standard. Quantification and quality control for analyzing method of simulant is shown in the SI. Statistical Analysis. Data analysis was performed with Statistical Product and Service Solutions (IBM), version 17.0. For statistical analysis, concentrations below the LOQ were substituted with a value equal to LOQ/2. Pearson’s correlation analysis was used to identify the significance of relationships between the concentrations of chlorinated byproducts and the concentrations of parent compounds. The statistical level of significance was set at p < 0.05.
■
RESULTS AND DISCUSSION Derivatization and Method Validation. In this study, a dansylation reaction with dansyl chloride (DNS) was used for determination of BPA, NP, and their chlorinated byproducts. Ionization and fragmentation of the isolated dansyl derivatives in electrospray tandem mass spectrometry resulted in protonated molecular ions of their dansyl derivatives. While the dansyle derivatiztion largely improved the sensitivity and decreased the matrix effects,9 the specificity would be not fully ensured since dansyl derivatization leads to the same product 7221
DOI: 10.1021/acs.est.5b00831 Environ. Sci. Technol. 2015, 49, 7218−7226
Article
Environmental Science & Technology
Table 2. Limits of Quantification (LOQs) and Recoveries of Target Analytes in Food Contacting Papers, Ground Coffee, and Simulanta food contacting papers
ground coffee
recovery(%) ± RSD analyte BPA BPA-d4 MCBPA DCBPA DCBPA-d12 TCBPA TeCBPA NP 4-n-NP MCNP DCNP
LOQs (ng/g)
0.05 ng/g (n = 3)
0.3
93 101 88 87 89 88 87 86 78 85 87
0.003 0.002 0.005 0.006 0.3 0.01 0.03
± ± ± ± ± ± ± ± ± ± ±
16 20 17 16 12 19 20 24 18 16 20
0.5 ng/g (n = 3) 99 103 96 94 92 90 90 84 81 86 84
± ± ± ± ± ± ± ± ± ± ±
18 10 10 14 11 12 6 12 6 2 6
simulant
recovery(%) ± RSD 2 ng/g (n = 3)
108 105 103 101 96 102 101 97 94 92 94
± ± ± ± ± ± ± ± ± ± ±
7 2 16 19 9 16 6 12 12 3 4
LOQs (ng/g)
0.05 ng/g (n = 3)
0.3
90 88 72 79 78 69 74 58 62 65 67
0.002 0.004 0.006 0.006 0.3 0.04 0.04
± ± ± ± ± ± ± ± ± ± ±
12 21 4 5 6 7 5 3 7 9 4
0.5 ng/g (n = 3) 98 93 86 84 82 80 80 79 81 76 74
± ± ± ± ± ± ± ± ± ± ±
16 14 9 8 10 11 16 12 6 12 11
recovery(%) ± RSD
2 ng/g (n = 3)
LOQs (ng/L)
± ± ± ± ± ± ± ± ± ± ±
5.4
104 105 103 101 96 102 101 90 89 82 84
16 2 16 19 9 16 6 12 13 9 7
0.01 0.01 0.01 0.03 4.3 0.02 0.02
0.05 ng/L (n = 3) 97 98 86 81 85 72 73 70 75 67 70
± ± ± ± ± ± ± ± ± ± ±
4 4 8 2 4 3 5 2 7 2 4
0.5 ng/L (n = 3) 94 95 85 89 86 72 75 77 75 70 72
± ± ± ± ± ± ± ± ± ± ±
5 7 6 12 10 5 10 9 6 4 5
2 ng/L (n = 3) 117 104 93 92 90 85 82 85 80 72 74
± ± ± ± ± ± ± ± ± ± ±
7 13 9 8 7 8 9 10 11 6 9
a
BPA: bishphenol A; NP: nonylphenol; MCBPA: 4-chloro-BPA; DCBPA: dichloro-BPA (mixture of 2,6-dichloro-BPA and 2,6′-dichloro-BPA (1:0.25)); TCBPA: trichloro-BPA; TeCBPA: tetrachloro-BPA; MCNP: 2-chloro-NP; and DCNP: 2,6-dichloro-NP.
ions, i.e., m/z 171 (cleavage of CS bond of the dansyl structure) and m/z 156 which is the m/z 171 losing one methyl group for all derivatized molecules (Table 1). In this study, target chemicals were identified by specific precursor masses and retention time of the derivatized molecules. The chemical structure of all analytes under survey as well as their dansyl derivatives were shown in SI Figure S1. In a previous publication, a dansylation reaction in buffer solution had been used for determination of BPA, NP, and their chlorinated byproducts in drinking water.9 While double DNS derivatives were observed for BPA, MCBPA, and DCBPA, only single DNS derivatives were observed at m/z 566 and 600 for TCBPA and TeCBPA, respectively.9 When the same dansylation method was used to detect TCBPA and TeCBPA, the analysis encountered interference due to matrix effects. To solve this problem, a previously developed dansylation reaction with DNS in acetonitrile under catalysis of 4-(dimethylamino)-pyridine for detection of fluorotelomer alcohols was used.35 In use of this method, double DNS derivatives were also observed at m/z 799 and 833 in the mass spectra of TCBPA and TeCBPA, respectively. To compare these two dansylation methods, both dansylation methods were attempted in the analysis of FCPs. The LOQs of dansylation reaction in the acetonitrile method were 0.003 ng/g, 0.002 ng/g, 0.005 ng/g, 0.006 ng/g, 0.01 ng/g and 0.03 ng/g for MCBPA, DCBPA, TCBPA, TeCBPA, MCNP, and DCNP, which were about 3−10 fold lower than those produced from the method which used a buffer solution (0.01 ng/g for MCBPA, 0.01 ng/g for DCBPA, 0.02 ng/g for TCBPA, 0.03 ng/g for TeCBPA, 0.1 ng/g for MCNP, and 0.3 ng/g for DCNP), while the LODs of BPA and NP were not altered. When using the newly established dansylation method to analyze the target chemicals in FCPs, we found that peaks of MCNP and DCNP overlapped with unidentified chemical peaks from the samples. Therefore, additional cleanup steps were used to eliminate potential interferences. The use of a silica cartridge following dansylation was able to further reduce interferences, and good chromatographic separation of MCNP and DCNP from unidentified chemicals in samples was achieved (Figure 1). As displayed in Table 2, the method recoveries for FCPs spiked with three concentrations of each analyte were 78−
101% with a RSD of 12−24% for 0.05 ng/g, 81−103% with a RSD of 2−18% for 0.5 ng/g, and 92−108% with a RSD of 2− 19% for 2 ng/g. The method recoveries for ground coffee spiked with three concentrations of each analyte were 58−90% with a RSD of 3−21% for 0.05 ng/g, 74−98% with a RSD of 6−16% for 0.5 ng/g, and 82−105% with a RSD of 2−19% for 2 ng/g. LODs and LOQs for FCPs were 0.0007−0.002 ng/g and 0.002−0.006 ng/g for chlorinated BPAs, and 0.003−0.01 ng/g and 0.01−0.03 ng/g for chlorinated NPs. LODs and LOQs for ground coffee were 0.0007−0.002 ng/g and 0.002−0.006 ng/g for chlorinated BPAs, 0.01 ng/g and 0.04 ng/g for chlorinated NPs. Less than 20% signal suppression or enhancement was observed for all target analytes. BPA and NP were both detected in procedural blanks (0.64 ± 0.03 ng/g for BPA and 0.46 ± 0.03 ng/g for NP in FCPs and 0.82 ± 0.03 ng/g for BPA and 0.67 ± 0.03 ng/g for NP in ground coffee), with LODs and LOQs of 0.09 ng/g and 0.3 ng/g for BPA, and 0.1 ng/g and 0.3 ng/g for NP in FCPs and ground coffee. The method recoveries, procedural blanks, LODs, and LOQs for simulant are shown in the SI. In general, the dansylation UPLC−MS−MS for analyzing BPA, NP, and their chlorinated derivatives were highly sensitive, though the derivatized method required relatively long analysis times compared to UPLC− MS−MS analysis without derivatization. Concentrations of Chlorinated BPAs and BPA in Food Contacting Papers. All target chemicals were all detected in FCPs as shown in the typical MRM chromatograms in Figure 1, and their concentrations in each sample are listed in SI Table S3. BPA was frequently detected (82%) in FCPs with mean concentration of 3.8 ng/g (
