Coupling mesoporous imprinted polymer based DGT passive samplers and HPLC: A new tool for in-situ selective measurement of low concentration tetrabromobisphenol A in freshwaters

Coupling mesoporous imprinted polymer based DGT passive samplers and HPLC: A new tool for in-situ selective measurement of low concentration tetrabromobisphenol A in freshwaters

Science of the Total Environment 685 (2019) 442–450 Contents lists available at ScienceDirect Science of the Total Environment journal homepage: www...

2MB Sizes 0 Downloads 2 Views

Science of the Total Environment 685 (2019) 442–450

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Coupling mesoporous imprinted polymer based DGT passive samplers and HPLC: A new tool for in-situ selective measurement of low concentration tetrabromobisphenol A in freshwaters Zhongmin Feng a,b, Yun Wang a, Lan Yang a, Ting Sun a,b,⁎ a b

College of sciences, Northeastern University, Shenyang 110819, China School of Materials Science and Engineering, Northeastern University, Shenyang 110819, China

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• Diffusive coefficient was measured by experiment and simulated by mathematical model. • Novel selective binding agent based on mesoporous imprinted polymer in DGT technique • A new coupling method between DGT and HPLC for in-situ sampling of TBBPA • Reliable results of deployments in natural waters and synthetical seawater

a r t i c l e

i n f o

Article history: Received 20 March 2019 Received in revised form 18 May 2019 Accepted 20 May 2019 Available online 21 May 2019 Editor: Shuzhen Zhang Keywords: Diffusive gradients in thin films technique TBBPA Mesoporous imprinted polymer Passive sampling

a b s t r a c t Accurate measurement of tetrabromobisphenol A (TBBPA) is very important because of its widespread environmental pollution. Diffusive gradients in thin films technique (DGT), an in-situ passive sampling method, is regarded as a reliable and robust measurement technique. A new DGT technique based on mesoporous imprinted polymer was combined with high-pressure liquid chromatography (HPLC) method for sampling, preconcentration and monitoring low concentration TBBPA in natural waters. The diffusion coefficient of TBBPA through the diffusive gel was measured by diffusion cell test and simulated using mathematical expression. The effects of different ambient conditions were tested under laboratory conditions and the performance of DGT sampler was validated in natural waters. The diffusion coefficient of TBBPA in the diffusive gel was measured as 2.18 × 10−6 cm2 s−1 and simulated as in the range 1.41–3.48 × 10−6 cm2 s−1 by Amsden model. Comparison of experimental and theoretical data, the validity of the experimental method can be verified by the mathematical model. The binding agent with mesoporous imprinted polymer showed selective affinity to TBBPA and its adsorption rate met the requirement of DGT device. The DGT method detection limit was at the level of ng L−1 for 7 days deployment. DGT sampler was suitable for application in aquatic environment with a range of pH (4.5–7.6), ionic strength (1 × 10−4 - 0.5 mol L−1), and dissolved organic matter (DOM) concentration

⁎ Corresponding author at: College of sciences, Northeastern University, Shenyang 110819, China. E-mail address: [email protected] (T. Sun).

https://doi.org/10.1016/j.scitotenv.2019.05.297 0048-9697/© 2019 Published by Elsevier B.V.

Z. Feng et al. / Science of the Total Environment 685 (2019) 442–450

443

(0–10 mg C L−1). The coupling method of DGT and HPLC was a promising technique for in situ sampling, preconcentration and monitoring low concentration TBBPA in most typical natural waters. © 2019 Published by Elsevier B.V.

1. Introduction Brominated flame retardants (FRs) are ubiquitous industrial chemicals that are widely utilized as additive or reactive chemicals in electronic equipment, polymers, textiles, and furniture. Tetrabromobisphenol A (TBBPA) is one of the most often used brominated flame retardants which accounts for 60% of global commercial consumption (Liu et al., 2016). TBBPA is a relatively persistent organic pollutant and poses a potential threat to the soil, water, atmosphere and living organisms. Due to the structural resemblance of TBBPA and thyroid hormone, TBBPA can interfere with estrogen signaling (Zhang et al., 2018). International Agency for Research on Cancer (IARC) has listed TBBPA into group 2A (Shen et al., 2018). Lots of evidence shows TBBPA occurrence in air, soil, natural water, dust, sediment and biological samples, suggesting TBBPA is a widespread contaminant (McAvoy et al., 2016). Consequently, for its ubiquitous use and ecotoxicity, accurate measurement and monitoring of TBBPA in aquatic ecosystems is critical for understanding its ecotoxicity, transport and possible risk to aquatic biota and human beings. Because of its moderately high octanol/water partition coefficient (log Kow = 5.9), TBBPA exhibits a poor water solubility (Liu et al., 2016). Moreover, its solubility depends upon the aqueous pH and the ionization state. Low concentration of TBBPA has been reported in rivers ranged from 0.3 to 4870 ng L−1 (Bao and Niu, 2015; Gao et al., 2016). At present, low concentration of TBBPA in the complex environment is usually determined by HPLC, LC-MS, GC or GC–MS method based on pretreatment and enrichment processes (Frederiksen et al., 2007; Liu et al., 2019). Routinely used enrichment methods require large solvent consumption, have risk of contamination, and need laborious procedures. Passive sampling methods, such as the diffusive gradient in thin films technique (DGT) and the polar organic chemical integrative sampler (POCIS), have developed as a sufficiently quantitative sampling tool and overcome the above limitations. Passive sampling methods can accumulate targets during in situ deployment, provide time-weighted average concentration, and avoid repeated spot grab sampling. Some authors have shown that o-DGT device serves as a robust passive sampling tool for organic pollutant and seems to be preferable to POCIS sampler for in complex environment applications (Challis et al., 2018). Thus, combination method of DGT passive sampler and HPLC should be an ideal candidate for in situ preconcentration and measurement of low concentration TBBPA in natural waters. DGT technique is composed of a diffusive agent and a binding agent. Various sorbents have emerged as the binding agent (Chen et al., 2017; Guan et al., 2018). The binding agent also can be specially designed for pollutants and then it has selectivity for pollutants, such as poly (acrylamidoglycolic acid-coacrylamide) hydrogel for Cu (Li et al., 2002), copper ferrocyanide for Cs (Li et al., 2009), 3-mercaptopropyl functionalized silica gel for AsIII (Bennett et al., 2011) and Hg (Hong et al., 2011), ion and molecular imprinted polymer for inorganic Sb (Fan et al., 2016) and 4-chlorophenol (Dong et al., 2014). Nowadays, mesoporous silica based imprinted polymer has been successfully prepared and showed good selective affinity to TBBPA molecules. So, mesoporous imprinted polymer can be introduced as a new DGT binding agent for sampling of TBBPA. Besides the binding agent, the diffusive agent (usually a hydrogel) is also an indispensable component of DGT devices. According to DGT theory, it is necessary to measure the diffusion coefficient D of pollutant through the diffusive agent. Classical steady-state diffusion cell test is by far the most frequently used method

for diffusion coefficient measurement. However, due to low solubility of non-polar organic matter, large measurement error may be obtained in dilute solution. Therefore, it is necessary to model pollutant diffusivity in hydrogels using mathematical expressions and examine the validity of the diffusion cell experimental data by comparing with the theoretical data. The purpose of this study is to evaluate the coupling between mesoporous imprinted polymer based DGT passive sampler and HPLC method, in order to in situ preconcentrate and monitor TBBPA in natural waters. The diffusion coefficient of TBBPA through the diffusive gel was measured by diffusion cell test and simulated using mathematical expression. The effects of different ambient conditions were tested under laboratory conditions and the performance of DGT sampler with mesoporous imprinted polymer was validated in natural waters. 2. Experimental procedure 2.1. Reagents, material and solutions The chemicals used in this study were of analytical reagent grade or higher. All plastic containers and DGT devices involved in this experiment were acid-cleaned before use. Tetrabromobisphenol A (TBBPA, Maya Reagent, 98%) was used to prepare the TBBPA stock solution (200 mg L−1) in methanol. Purified water (Hangzhou Wahaha Group Co., Ltd., China) was utilized to dilute the stock solution. Acetonitrile, methanol and acetic acid (HPLC grade) were purchased from Starmark Science and Technology Development Co., Ltd. of China. Details of four membrane filters and regents in the selectivity test are given in the Supporting information (SI). 2.2. Instrumentation The properties of fibrous silica and mesoporous imprinted polymer were characterized by scanning electron microscopy (SEM, SU8010, Hitachi), transmission electron microscope (TEM, Tecnai G20, FEI), and Fourier-transform infrared spectroscopy (FTIR, Vertex 70, Bruker). Nitrogen sorption analysis was operated by a full-automatic physical adsorption instrument (ASAP 2020 HD88, Micromeritics) at 77 K. The surface areas and the pore size distribution were calculated by the Brunauer–Emmett–Teller (BET) and Barrett–Joyner–Halenda (BJH) model. 2.3. Analysis TBBPA in the eluent solution was analyzed by high-performance liquid chromatography (HPLC, Hitachi Primaide, Japan) equipped with Diamonsil® C18 column (5 μm particle size, 4.6 × 250 mm), thermostated column compartment (30 °C), and UV-detector operated at wavelength of 235 nm. A mobile phase contained acetonitrile /water (85:15, v/v) was used at a flow rate of 0.6 mL min−1. TBBPA in the spiked natural water was diluted with equivalent volume of acetonitrile and determined using the same method. 2.4. Diffusive and binding gel preparation Diffusive gel (d = 0.6 mm) was prepared with 2.5% agarose solution. The gel solution was boiled until it became transparent, pipetted into two pre-heated glass plates separated by a certain thick spacer, and

444

Z. Feng et al. / Science of the Total Environment 685 (2019) 442–450

left to cool. The agarose gels were cut into discs of ~1.8 cm diameter and stored in 0.01 mol L−1 NaNO3 solution. Fibrous silica supported mesoporous molecularly imprinted polymer nanospheres were used as the binding agent. Fibrous nano-silica and mesoporous imprinted polymer was prepared according to previous report with some modifications (Deng et al., 2008; Yang et al., 2015). Fig. 1 shows the schematic illustration of the procedure for synthesis of mesoporous imprinted polymer. Detailed information for preparation of mesoporous molecularly imprinted polymer is available in SI. 1.8 g mesoporous imprinted polymer was dispersed into 8 mL gel solution contained acrylamide (28.5%, W/v) (Amresco, purityN99%) and bis-acrylamide (1.5%, W/v) (Aladdin, purityN99%). Under stirring for 20 min, 15 μL N,N,N′,N′-tetramethylethylenediamine (TEMED, SCRC, AR grade) as catalyst and 188 μL freshly prepared ammonium persulfate (SCRC, AR grade) solution (10%, W/v) as initiator were added into the above gel solution. The mixture was well stirred, pipetted into two glass plates separated by a 0.5 mm thick spacer, and incubated at 45 °C for 1.5 h. Then the gel sheet was immersed in deionized water for 24 h and extracted with methanol for 48 h in a Soxhlet extractor to remove template TBBPA. After extraction, the mesoporous imprinted polymer gels (MIP-gels) were stored in deionized water for use. Blank gels were obtained by repeating the procedure in the absence of template TBBPA molecules. 2.5. Binding properties of MIP gels and gel elution The agarose gels, polyvinylidene fluoride (PVDF), poly(tetrafluoroethylene) (PTFE), polyethersulfone (PES) and nylon membrane filter were placed in the solutions to assess their possible uptake of TBBPA. The MIP gels were soaked in 10 mL of 2 mg L−1 TBBPA solutions for static adsorption. The adsorption capacity of the MIP gels was evaluated by placing these discs in 20 mL TBBPA solution at concentrations ranging from 0.5 to 20 mg L−1 for 24 h (pH 6.0). The specific recognition properties of MIP gels for TBBPA, bisphenol A (BPA), trichlorophenol (TCP), naphthalene (NAP), and amino-diphenylmethane (MDA) were

investigated and compared with that of blank gels. The selectivity coefficient (β) of the MIP gels for template TBBPA is defined as the ratio of the imprinting factor of template TBBPA and the competition molecules. Detailed information for selectivity tests is available in SI. Because the TBBPA standard solution was prepared with methanol, the TBBPA working solutions contained 10% methanol (methanol/water, v/v, 10:90) to avoid co-solvent effects and enhance hydrotropy. All experiments were done in three times unless otherwise stated. The harvested MIP gels were soaked in 5 mL acetonitrile or methanol or a mixed solution (4.5 mL methanol 0.5 mL acetic acid or 4.5 mL acetonitrile 0.5 mL acetic acid) for 24 h. The elution efficiency of MIP gels was obtained by the ratio between the desorbed mass of TBBPA and the accumulated mass of TBBPA. A typical DGT sampler which obeys to Fick's first law of diffusion comprises a binding gel, a diffusive gel and a membrane filter. After a deployment time(t), the DGT-predicted concentration of TBBPA(CDGT) is calculated by Eq. (1): CDGT ¼

M Δg DAt

ð1Þ

Where M (μg) is the mass of molecule TBBPA accumulated on the MIP gels, Δg (mm) is the agarose gel thickness, D (cm2 s−1) is the diffusion coefficient of TBBPA in the agarose gel, and A (cm2) is the area of exposure window of samplers. The molecule TBBPA is rinsed from MIP gels with suitable solution and M is calculated by Eq. (2): M¼

  Ce Ve þ Vgel fe

ð2Þ

Where Ce (mg L−1) is the TBBPA concentration in the eluent, Ve (mL) and Vgel (mL) are the volume of the eluent solution and the MIP gel, respectively, and fe is the elution coefficient. The measurement process of the mass of TBBPA accumulated on the MIP gels is illustrated in Fig. 2.

Fig. 1. Schematic illustration the procedure for synthesis of mesoporous imprinted polymer.

Z. Feng et al. / Science of the Total Environment 685 (2019) 442–450

445

Fig. 2. The measurement process of the mass of TBBPA accumulated on the MIP gels.

2.6. Measurement of diffusion coefficient of TBBPA Diffusion coefficient of TBBPA through the diffusive gel was measured using a diffusion cell as previously stated with minor modification (Guan et al., 2018), and the details are available in SI.

(DOM) on MIP-DGT performance was evaluated by soaking the DGT devices in 10 L solutions for 24 h containing 1.0 mg L−1 of TBBPA and different concentrations of DOM (5,10, 20 mg C L−1). The DOM solutions were prepared by sodium humate, sodium citrate, sodium oxalate, sodium benzoate and sodium tartrate, respectively.

2.7. Effects of different ambient conditions 2.8. Application in natural waters and synthetical seawater The accumulation of TBBPA molecules by MIP-DGT devices was evaluated in 20 L solution (pH = 6.2, 0.01 mol L−1 NaCl) containing 520 μg L−1 TBBPA for times ranging from 4 to 48 h. To evaluate the effect of pH and ionic strength on the performance of MIP-DGT devices, DGT samplers were immersed in TBBPA solutions of different pH (4.5–8.5; 0.01 mol L−1 NaCl) and various ionic strength (0.0001–0.5 mol L−1 NaCl; pH = 6.2) for 24 h. The effect of dissolved organic matter

The performance of the MIP-DGT samplers was investigated in natural waters (Hunhe river sample and Nanhu lake sample, Shenyang, China) and synthetical seawater spiked with 500 μg L−1 TBBPA for 48 h. The composition of the synthetical seawater was prepared according to Panther et al. (2010). Compositions of the two freshwater samples and the synthetical seawater are available in Table S1 (see SI).

Fig. 3. SEM images of silica (a), fibrous silica (b) and fibrous silica supported imprinted polymer (c); TEM images of silica (d), fibrous silica (e) and fibrous silica supported imprinted polymer (f).

446

Z. Feng et al. / Science of the Total Environment 685 (2019) 442–450

3. Results and discussion 3.1. Characterization of fibrous silica and mesoporous imprinted polymer SEM and TEM images of silica, fibrous silica and mesoporous imprinted polymer are presented in Fig. 3. Silica microspheres were sphere-shaped with an average size of about 30 nm (Fig. 3a, d). The surface morphology of fibrous silica was shaped like a Hydrangea flower (Fig. 3b). The average size of fibrous silica microspheres was about 300 nm and their pores were radiated outward from the center (Fig. 3e). After polymerization, the surface of mesoporous imprinted polymer was completely different from fibrous silica and shaped like a highly crosslinked polymer network (Fig. 3c, f). The observation indicated the successful combination of the imprinted polymer on the surface of fibrous silica. The silica, modified silica and mesoporous imprinted polymer were characterized using FT-IR instrument in the purpose of confirming the successful modification and polymerization (Fig. 4a). All the changes of adsorption bands suggested that the successful preparation of silica supported mesoporous imprinted polymer for TBBPA molecules. Details of discussion about FT-IR characterization are available in SI. The N2 adsorption-desorption isotherms of fibrous silica and mesoporous imprinted polymer were measured to investigate their surface areas and pore size distributions. It can be seen that both of them exhibited a typical type IV profile with an H3 hysteresis loop, suggesting the presence of typical mesoporous structures (Fig. 4b, c). For imprinted

polymer, the adsorption branch was noncoincidence to the desorption branch even at low pressure, which was also observed in other organic polymers (Ding et al., 2015; Kim et al., 2018; Zhuang et al., 2014). After polymerization, the BET surface area was reduced from 90 m2 g−1 to 40 m2 g−1, which may be explained by the existence of large organic group on the mesopores. The pore volume vs. pore diameter plots of the fibrous silica, obtained from adsorption isothermal line via BJH model, presented a sharp peak at around 3 nm. In comparison, the imprinted polymer showed narrow pore-size distribution centered at 3 nm and 9 nm, respectively. These results suggested that mesopores were the predominant pores for fibrous silica and imprinted polymer. Moreover, compared with the fibrous silica, the proportion of mesopores of the imprinted polymer at 9 nm increased a lot. The imprinted polymer was polymerized in the presence of TBBPA molecules and extracted with methanol to remove template TBBPA molecules. Thus, many empty cavities were remained in the mesoporous imprinted polymer, which provided recognition sites to the target molecules. According to the previous report, the dynamic radius of TBBPA molecule is 3.91 nm (Ding et al., 2015). The size of recognition site matched the size of target molecules, which caused the proportion of mesopores of the imprinted polymer at 9 nm increased. 3.2. Characteristics of MIP-gels in solutions The binding kinetics of TBBPA onto the MIP-gel was evaluated by soaking it in 10 mL TBBPA solutions of 2 mg L−1 for static adsorption. It

Fig. 4. (a) Comparison between FT-IR spectra of silica, modified silica and mesoporous imprinted polymer, (b) N2 adsorption-desorption isotherm and pore size distribution of fibrous silica, (c) N2 adsorption-desorption isotherm and pore size distribution of mesoporous imprinted polymer.

Z. Feng et al. / Science of the Total Environment 685 (2019) 442–450

447

Fig. 5. (a) Uptake of TBBPA by the MIP-gel as a function of time in 10 mL solutions containing 2 mg L−1 TBBPA, (b) uptake of TBBPA by the MIP-gel as a function of initial concentration, and (c) selectivity properties of TBBPA on MIP and blank gels relative to BPA, TCP, NAP and MDA.

was observed that accumulation of TBBPA by the MIP-gel increased linearly with time for the first 4 h (Fig. 5a). After 6 h, the adsorption efficiency was above 95%. The average adsorption rate over the first 4 h was 0.69 ng cm−2 s−1. Assuming a DGT sampler was soaked in a high concentration solution of 2 mg L−1, the flux was calculated as 0.06 ng cm−2 s−1 at 25 °C according to DGT equation (Eq. (1)), which was much lower than the adsorption rate. The difference indicated that uptake of TBBPA onto the MIP-gels was fast enough to meet the requirement of DGT theory. In a high concentration of 20 mg L−1, the adsorption capacity of MIPgel for TBBPA was measured as 172 μg per disc (Fig. 5b). To investigate the adsorption selectivity of the TBBPA molecule, four structural analogs (BPA, TCP, NAP, and MDA) were selected as controls and their chemical structures are available in Fig. S1. Fig. 5c shows the uptake efficiency of TBBPA and other competition molecules onto the MIP-gels and blank gels. The relative selectivity coefficient (β) values between the TBBPA and other competition molecules (BPA, TCP, NAP, and MDA) were 1.8, 1.9, 2.0 and 2.0, respectively. The MIP-gels showed a good selectivity and affinity to TBBPA molecules, which can be explained by that the target TBBPA molecules can easily match the profile and size of recognition sites. A high and reliable elution efficiency of TBBPA from MIP-gels is required in DGT equation. Four different eluents (acetonitrile, methanol, methanol acetic acid, and acetonitrile acetic acid) were tried to eluted TBBPA from the MIP-gels (Table S2). The desorption efficiency varied with the types of eluent. Consequently, 5 mL acetonitrile was effective in removing TBBPA from the MIP-gels. The obtained elution coefficient for 24 h was 82.3 ± 8.3% (n = 6).

Moreover, the uptake efficiency of TBBPA onto the agarose gel and four different membranes was investigated and the results are shown in Fig. S2. Consequently, the agarose gel and PVDF membrane were chosen to be applied in the DGT samplers and details are available in SI. 3.3. Diffusion coefficient of TBBPA through the diffusive gel To calculate the concentration predicted by DGT equation, it is necessary to measure the TBBPA diffusion coefficient D. The TBBPA diffusion coefficient obtained from Fig. S3a was 2.02 × 10−6 cm2 s−1 at 22 °C with a 0.45 μm PVDF filter covered on the agarose gel. The standard diffusion coefficient of TBBPA at 25 °C was calculated to be 2.29 ± 0.15 × 10−6 cm2 s−1 (n = 6) using the Stokes-Einstein equation (Zhang and Davison, 1995). For a constant ionic strength, the TBBPA diffusion coefficient was not appreciably affected by the pH of source solution in the range of 5–8 (Fig. S3b). Due to the viscosity effect, the TBBPA diffusion coefficient slightly decreased as NaCl concentration increased from 0.005 to 0.5 mol L−1 (Fig. S3c). The relative standard deviation of these values was b10%. Thus, the average TBBPA diffusion coefficient (2.18 × 10−6 cm2 s−1) was applied in the DGT equation. 3.4. Comparison of experimental and theoretical diffusion coefficient for TBBPA The diffusion of solutes in hydrogels (eg. agarose gel and polyacrylamide gel) has been widely applied in various fields. In order to model solute diffusion in hydrogels, many mathematical expressions have been developed. As previously reported, the solute diffusivity in hydrogels tends to be lower than that in aqueous solution, and some

448

Z. Feng et al. / Science of the Total Environment 685 (2019) 442–450

factors such as the solute molecular diameter, polymer chain mobility, and size of the polymer chains openings have an effect on the diffusion of solutes. Amsden model (Amsden, 1998) based on obstruction theory was developed to describe the solute diffusion in agarose gel as follows,  2 ! Dg rs þ r f ¼ exp −π D0 ks φ−0:5 þ 2r f

ð3Þ

where Dg and D0 are the solute diffusion coefficients in the agarose gel and in the aqueous solution, respectively. rs and rf are the dynamic radius of solute and the radius of polymer chains, respectively. φ is the fiber volume fraction, and ks is a constant. Thus, the theoretical diffusion coefficient of TBBPA through the agarose gel also can be evaluated by Eq. (3). According to the results of molecular dynamics simulations for TBBPA, the dynamic radius of TBBPA molecule is 3.91 nm (Ding et al., 2015). Th average radius of agarose hydrogel fibers is taken to be 19 Å and the value of constant ks is 10.6 Å (Dai et al., 2011). The fiber volume fraction φ for agarose hydrogel is in the range from 0.01 to 0.05(Kong et al., 1997). Then the calculated ratios of Dg/D0 from Eq. (3) are between 0.234 and 0.576. The solute diffusivity in the free solution D0 (m2 s−1) can be described by Wilke-Chang model as follows (Wilke and Chang, 1955), 1 T 2 0:6

D0 ¼ 7:4  10−15 ð∅MB ÞμVA

ð4Þ

where ϕ is the association parameter of solvent (for water, ϕ = 2.6), MB is the molar mass of solvent molecules (for water, MB = 18 kg kmol−1), μ is the solvent viscosity (for water, μ = 0.8937 × 10−3 Pa·s, 25 °C), T(K) is the solution's temperature and VA is the molar volume of solute (for TBBPA, VA = 257 cm3 mol−1)(Joonsung Yoon and Lesser, 2009). Thus, the TBBPA molecule diffusivity in the aqueous solution D0 was calculated as 6.04 × 10−6 cm2 s−1 according to Eq. (4). Substituting the value of D0 into Eq. (3), the theoretical diffusion coefficient of TBBPA through the agarose gel Dg was calculated as in the range 1.41–3.48 × 10−6 cm2 s−1. The diffusive gel of DGT devices was made from 2.5% agarose gel solution. The experimental TBBPA diffusion coefficient D was measured as 2.18 × 10−6 cm2 s−1 at 25 °C. From previous report, in terms of accounting for change caused by polymer chain flexibility, Amsden model was best at providing consistence to the experimental data in applicability to large molecule solute diffusion situations. Comparison of experimental and theoretical diffusion coefficient, Amsden model based on obstruction theory was consistent with the experimental data. At the same time, the validity of the experimental data obtained by diffusion cell test was verified by the theoretical data obtained by mathematical model. 3.5. Effect of deployment time and DGT method quantitation limits According to the principle of DGT (Eq. (1)), a linear relationship is expected between the mass (M) accumulated in the DGT devices and deployment time (t), providing that the binding capacity has not been exceeded. The uptake of TBBPA by MIP-DGT devices was evaluated in 20 L solution containing 520 μg L−1 TBBPA. The DGT devices were taken out at intervals of 4–12 h and the MIP-gels were eluted with acetonitrile. The eluate was determined by HPLC method to calculate the uptake mass of TBBPA(M). Fig. S4 presents the linear accumulation of TBBPA with the deployment time over a 48 h period. It was observed that the mass of TBBPA uptake by DGT devices agreed well with the calculated uptake mass of TBBPA (dashed line in Fig. S4). Meanwhile, the spiked water sample was also directly measured by HPLC method. Fig. 6 shows the comparation of HPLC spectrogram between the water sample and the eluent solutions after different deployment time. The retention time of TBBPA in water sample was 10.5 min

Fig. 6. Comparison of HPLC spectrogram between the water sample diluted with the same volume acetonitrile and the eluent solutions desorbed from the binding gels after different deployment time.

with a very weak signal response for its low concentration. The retention time of TBBPA in eluate shifted to 11.5 min. Moreover, the value of signal response increased as a function of deployment time, suggesting the calculated mass of TBBPA in DGT samples increased with the deployment time. Compared with water sample, the value of signal response increased about 10 times after 48 h deployment. According to Eq. (1), the longer the deployment time, the bigger the accumulated mass onto the MIP-gels and the larger the HPLC signal response in the eluent solution. Theoretically, the prolonging of the deployment time is beneficial to sampling of low concentration TBBPA in the environments. Table S3 shows the DGT blank, instrument limit of quantification (LOQ) and DGT method detection limit (MDL). The obtained gel blank of TBBPA without deployment was 27.49 ± 0.14 ng per device. The obtained LOQ of TBBPA was 274.97 ng. MDL for DGT of TBBPA was calculated from LOQ and the deployment time using Eq. (1). The longer the deployment time, the lower the value of MDL. Assuming a deployment time of 1 d, 3 d and 7 d, the calculated MDL for DGT method was 56.29, 131.35 and 394.10 ng L−1, respectively. These results indicated that the DGT passive sampling method coupled with HPLC determination method can improve sensitivity and accuracy of in situ monitoring low concentration TBBPA in environments. 3.6. Effect of pH, ionic strength and different dissolved organic matter The environmental pH varies as the environment changes. The surface charge density of mesoporous imprinted polymer and the charge of TBBPA species are directly related to the solution pH. Thus, it is necessary to evaluate the impact of pH on the performance of MIP-DGT devices for sampling of TBBPA. In the pH range 4.5–7.6, the TBBPA concentration predicted by DGT sampler (CDGT) had a good coincidence with the concentration measured directly by HPLC method (CSol) (Fig. 7a). When the ambient pH was above 8, the ratio of CDGT/CSol was decreased to b0.9. TBBPA has two slightly acidic protons, and the values of pK1 and pK2 of TBBPA are 7.5 and 8.5, respectively (Kuramochi et al., 2008). Depending upon the ambient pH, TBBPA can be present as either phenol or ionized forms. Species distribution as a function of pH for the TBBPA solution is shown in Fig. 7c. At lower pH (b8), TBBPA exists essentially in the phenol form. Above pH 8, partial TBBPA presents as in the ionized form, which was not beneficial for the accumulation in the MIP-gels. These results indicated that the MIP-DGT samplers can be used to measure the molecular species of TBBPA because of electrostatic repulsion between the anionic species of TBBPA and the negatively charged surface of MIP-gels. The ratios of CDGT/CSol under various ionic strength is presented in Fig. 7b. With NaCl concentrations varying from 1 × 10−4 to 0.5 mol L−1,

Z. Feng et al. / Science of the Total Environment 685 (2019) 442–450

449

Fig. 7. Effect of ambient pH (a) and ionic strength (b) on the performance of MIP-DGT samplers. The dotted horizontal lines represent acceptable values range from 0.9 to 1.1. Error bars state the standard deviation of triplicate samples measurement; (c) species distribution as a function of pH for the TBBPA solution: temperature = 25 °C, initial concentration = 0.001 mol L−1.

the ratios of CDGT/CSol was in the range 0.9–1.1. Even at high ionic strength of 0.5 mol L−1, no significant impact on MIP-DGT determination was observed. This was consistent with the previous research using imprinted polymer-based materials, indicating the MIP-DGT samplers offered potential application in seawater or high salinity water (Dong et al., 2014). DOM is a widespread component in surface waters. The concentration of DOM varies over a broad range for the various degrees of organic matter contamination in different natural waters. Therefore, it is necessary to investigate the effect of different DOM on the performance of MIP-DGT sampler for sampling of TBBPA in freshwaters (Fig. S5). Details of DOM effect are available in SI and the results indicated that MIP-DGT devices were not suitable in seriously contaminated natural waters (N10 mg C L−1). 3.7. Deployment in natural waters and synthetical seawater The performance of the MIP-DGT samplers was investigated in two natural waters and synthetical seawater spiked with 500 μg L−1 TBBPA for 48 h. After deployment, the values of CDGT/CSol were 0.81 ± 0.22 (n = 3), 0.79 ± 0.12, and 0.89 ± 0.15, respectively, in Nanhu lake water, Hunhe river water and synthetical seawater (Table S4). From the above results, MIP-DGT devices were not suitable in seriously contaminated and high pH natural waters. The pHs of the two natural waters and synthetical seawater were nearly neutral (Table S1). DOM concentrations in the Nanhu lake and Hunhe river water samples were 8.8 and 12.1 mg C L−1, respectively. The ratios of CDGT/CSol for the two natural waters were not within the ideal range, which may be explained by as follows: competition between DOM and TBBPA or complex formation

between TBBPA and other matter. Some parts of TBBPA may be interacted with DOM or be absorbed on the metal-oxyhydroxide colloid particles in the natural waters (Sun et al., 2008). Thus, it is necessary to investigate the relationship between the labile form and different species of TBBPA (molecular form, ionized form or complex form) by DGT method in future work. 4. Conclusion This work established a new coupling method between the diffusive gradient in thin films technique (DGT) with mesoporous imprinted polymer and high efficiency liquid chromatography (HPLC) to in situ sample, preconcentrate and monitor low concentration TBBPA in the aquatic environment. The binding agent with mesoporous imprinted polymer showed selective affinity to TBBPA and its adsorption rate met the requirement of DGT device. The diffusion coefficient of TBBPA in the diffusive gel was measured as 2.18 × 10−6 cm2 s−1 and simulated as in the range 1.41–3.48 × 10−6 cm2 s−1 by mathematical model. Comparison of experimental and theoretical data, the validity of the experimental method can be verified by the mathematical model. The DGT method detection limit was at the level of ng L−1 for 7 days deployment. DGT sampler was suitable for application in aquatic environment with a range of pH (4.5–7.6), ionic strength (1 × 10−4–0.5 mol L−1), and DOM concentration (0–10 mg C L−1). The poor uptake of TBBPA in high pH solution can be attributed to electrostatic repulsion between the ion species of TBBPA and the negatively charged surface of MIP-gels, indicating that the MIP-DGT samplers can be used to measure the molecular species of TBBPA. The coupling method between DGT and HPLC was a promising technique for in situ sampling, preconcentration

450

Z. Feng et al. / Science of the Total Environment 685 (2019) 442–450

and monitoring low concentration TBBPA in most typical natural waters. Acknowledgement This project received partial financial supported by the National Natural Science Foundation of China (No. 21777021), China Postdoctoral Science Foundation (No. 2018M631813), the Fundamental Research Funds for the Central Universities (No. N170504024) and the Doctoral Scientific Research Foundation of Liaoning Province of China (No. 20170520066). Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.scitotenv.2019.05.297. References Amsden, B., 1998. Solute diffusion within hydrogels. Mechanisms and models. Macromolecules 31, 8382–8395. Bao, Y., Niu, J., 2015. Photochemical transformation of tetrabromobisphenol A under simulated sunlight irradiation: kinetics, mechanism and influencing factors. Chemosphere 134, 550–556. Bennett, W.W., Teasdale, P.R., Panther, J.G., Welsh, D.T., Jolley, D.F., 2011. Speciation of dissolved inorganic arsenic by diffusive gradients in thin films: selective binding of As-III by 3-mercaptopropyl-functionalized silica gel. Anal. Chem. 83, 8293–8299. Challis, J.K., Stroski, K.M., Luong, K.H., Hanson, M.L., Wong, C.S., 2018. Field evaluation and in situ stress testing of the organic-diffusive gradients in thin-films passive sampler. Environ. Sci. Technol. 52, 12573–12582. Chen, W., Li, Y.Y., Chen, C.E., Sweetman, A.J., Zhang, H., Jones, K.C., 2017. DGT passive sampling for quantitative in situ measurements of compounds from household and personal care products in waters. Environ. Sci. Technol. 51, 13274–13281. Dai, H.J., Wu, J.J., Wang, Y.X., Tan, S.X., Liang, S.M., Jiang, B., et al., 2011. Diffusion of levofloxacin mesylate in agarose hydrogels monitored by a refractive-index method. J. Appl. Polym. Sci. 122, 3000–3006. Deng, Y., Qi, D., Deng, C., Zhang, X., Zhao, D., 2008. Superparamagnetic highmagnetization microspheres with an [email protected] core and perpendicularly aligned mesoporous SiO2 shell for removal of microcystins. J. Am. Chem. Soc. 130, 28–29. Ding, K., Zhang, H., Wang, H., Lv, X., Pan, L., Zhang, W., et al., 2015. Atomic-scale investigation of the interactions between tetrabromobisphenol A, tetrabromobisphenol S and bovine trypsin by spectroscopies and molecular dynamics simulations. J. Hazard. Mater. 299, 486–494. Dong, J., Fan, H., Sui, D., Li, L., Sun, T., 2014. Sampling 4-chlorophenol in water by DGT technique with molecularly imprinted polymer as binding agent and nylon membrane as diffusive layer. Anal. Chim. Acta 822, 69–77. Fan, H.-T., Liu, A.-J., Jiang, B., Wang, Q.-J., Li, T., Huang, C.-C., 2016. Sampling of dissolved inorganic Sb-III by mercapto-functionalized silica-based diffusive gradients in thinfilm technique. RSC Adv. 6, 2624–2631. Frederiksen, M., Vorkamp, K., Bossi, R., Svensmark, B., 2007. Analysis of HBCD and TBBPA by GC-MS versus LC-MS-MS - indications of systematic differences in obtained results. 4th International Workshop pn Bromonated Flame Retardants. Gao, Y., Pang, S.-Y., Jiang, J., Ma, J., Zhou, Y., Li, J., et al., 2016. Transformation of flame retardant tetrabromobisphenol A by aqueous chlorine and the effect of humic acid. Environ. Sci. Technol. 50, 9608–9618.

Guan, D.-X., Li, Y.-Q., Yu, N.-Y., Yu, G.-H., Wei, S., Zhang, H., et al., 2018. In situ measurement of perfluoroalkyl substances in aquatic systems using diffusive gradients in thin-films technique. Water Res. 144, 162–171. Hong, Y.S., Rifkin, E., Bouwer, E.J., 2011. Combination of diffusive gradient in a thin film probe and IC-ICP-MS for the simultaneous determination of CH3Hg and Hg2 in oxic water. Environ. Sci. Technol. 45, 6429–6436. Joonsung Yoon, T.J.M., Lesser, Alan J., 2009. Self-reinforcing isotactic polypropylene prepared using crystallizable solvents. J. Appl. Polym. Sci. 113, 3564–3576. Kim, H.-C., Huh, S., Lee, D.N., Kim, Y., 2018. Selective carbon dioxide sorption by a new breathing three-dimensional Zn-MOF with Lewis basic nitrogen-rich channels. Dalton Trans. 47, 4820–4826. Kong, D.D., Kosar, T.F., Dungan, S.R., Phillips, R.J., 1997. Diffusion of proteins and nonionic micelles in agarose gels by holographic interferometry. AICHE J. 43, 25–32. Kuramochi, H., Kawamoto, K., Miyazaki, K., Nagahama, K., Maeda, K., Li, X.-W., et al., 2008. Determination of physicochemical properties of tetrabromobisphenol A. Environ. Toxicol. Chem. 27, 2413–2418. Li, W., Zhao, H., Teasdale, P.R., John, R., 2002. Preparation and characterisation of a poly (acrylamidoglycolic acid-coacrylamide) hydrogel for selective binding of Cu2 and application to diffusive gradients in thin films measurements. Polymer 43, 4803–4809. Li, W., Wang, F., Zhang, W., Evans, D., 2009. Measurement of stable and radioactive cesium in natural waters by the diffusive gradients in thin films technique with new selective binding phases. Anal. Chem. 81, 5889–5895. Liu, K., Li, J., Yan, S., Zhang, W., Li, Y., Han, D., 2016. A review of status of tetrabromobisphenol A (TBBPA) in China. Chemosphere 148, 8–20. Liu, A., Shen, Z., Yuan, L., Xu, M., Zhao, Z., Liang, X., 2019. High-performance thin-layer chromatography coupled with HPLC-DAD/HPLC-MS/MS for simultaneous determination of bisphenol A and nine brominated analogs in biological samples. Anal. Bioanal. Chem. 411, 725–734. McAvoy, D.C., Pittinger, C.A., Willis, A.M., 2016. Biotransformation of tetrabromobisphenol A (TBBPA) in anaerobic digester sludge, soils, and freshwater sediments. Ecotoxicol. Environ. Saf. 131, 143–150. Panther, J.G., Teasdale, P.R., Bennett, W.W., Welsh, D.T., Zhao, H., 2010. Titanium dioxidebased DGT technique for in situ measurement of dissolved reactive phosphorus in fresh and marine waters. Environ. Sci. Technol. 44, 9419–9424. Shen, J., Huang, G., An, C., Xin, X., Huang, C., Rosendahl, S., 2018. Removal of tetrabromobisphenol A by adsorption on pinecone-derived activated charcoals: synchrotron FTIR, kinetics and surface functionality analyses. Bioresour. Technol. 247, 812–820. Sun, Z., Mao, L., Xian, Q., Yu, Y., Li, H., Yu, H., 2008. Effects of dissolved organic matter from sewage sludge on sorption of tetrabromobisphenol A by soils. J. Environ. Sci. 20, 1075–1081. Wilke, C.R., Chang, P., 1955. Correlation of diffusion coefficients in dilute solutions. AICHE J. 1, 264–270. Yang, J., Li, Y., Wang, J., Sun, X., Cao, R., Sun, H., et al., 2015. Molecularly imprinted polymer microspheres prepared by Pickering emulsion polymerization for selective solidphase extraction of eight bisphenols from human urine samples. Anal. Chim. Acta 872, 35–45. Zhang, H., Davison, W., 1995. Performance characteristics of diffusion gradients in thinfilms for the in-situ measurement of trace-metals in aqueous solution. Anal. Chem. 67, 3391–3400. Zhang, H., Liu, W., Chen, B., He, J., Chen, F., Shan, X., et al., 2018. Differences in reproductive toxicity of TBBPA and TCBPA exposure in male Rana nigromaculata. Environ. Pollut. 243, 394–403. Zhuang, Y., Seong, J.G., Do, Y.S., Jo, H.J., Cui, Z., Lee, J., et al., 2014. Intrinsically microporous soluble polyimides incorporating troger's base for membrane gas separation. Macromolecules 47, 3254–3262.