Molecularly imprinted polymer thin-film as a micro-extraction adsorbent for selective determination of trace concentrations of polycyclic aromatic sulfur heterocycles in seawater

Journal Pre-proof

Molecularly imprinted polymer thin-film as a micro-extraction adsorbent for selective determination of trace concentrations of polycyclic aromatic sulfur heterocycles in seawater Hassan Y. Hijazi , Christina S. Bottaro PII: DOI: Reference:

S0021-9673(19)31295-6 https://doi.org/10.1016/j.chroma.2019.460824 CHROMA 460824

To appear in:

Journal of Chromatography A

Received date: Revised date: Accepted date:

16 October 2019 28 November 2019 21 December 2019

Please cite this article as: Hassan Y. Hijazi , Christina S. Bottaro , Molecularly imprinted polymer thin-film as a micro-extraction adsorbent for selective determination of trace concentrations of polycyclic aromatic sulfur heterocycles in seawater, Journal of Chromatography A (2019), doi: https://doi.org/10.1016/j.chroma.2019.460824

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier B.V.

Highlights     

A procedure to prepare molecularly imprinted polymer thin-film (MIP thin-film) on a glass slide is proposed. Role and effect of various porogens on thin-film formation was investigated and studied. The porosity of thin-film was improved by adopting the principle of interpenetrating polymer networks (IPNs) Analysis of seawater carried out without pre-treatment using PASHs-MIP thinfilms with GC-MS. The size and shape of MIP slides and the ease of preparation makes them efficient for off-site and on-site environmental analysis when combined with a suitable detection system.

1

Molecularly imprinted polymer thin-film as a micro-extraction adsorbent for selective determination of trace concentrations of polycyclic aromatic sulfur heterocycles in seawater

Hassan Y. Hijazia, Christina S. Bottaroa* a

Department of Chemistry, Memorial University of Newfoundland, St. John's, Canada, A1B 3X7. E-mail: [email protected]; Fax: +1-709-864-3702; Tel: +1709-864-8088

ABSTRACT A tailor-made porous molecularly imprinted polymer (MIP) thin-film was prepared by in situ photo-radical polymerization on a glass slide and used as a microextraction adsorbent. Detection was carried out using gas chromatographymass spectrometry (GC-MS) to afford a method suitable for the selective determination of trace concentrations of polycyclic aromatic sulfur heterocycles (PASHs) in seawater. PASHs are one of the most problematic aromatic organic pollutants, as they are considered more persistent and toxic compared to other analogous aromatic compounds in the environment. The optimized thin-film consisted of a 2-thiophenecarboxaldehyde pseudo-template with 1-vinylimidazole (1-Vim) as the functional monomer, bisphenol A dimethacrylate (BPADMA) as the cross-linker, acetonitrile as the porogen, and polyethylene glycol to boost porosity through formation of interpenetrating polymer networks. The adsorption behaviours of the thin-film, including adsorption kinetics, binding isotherms, and selectivity of MIP thin-film were investigated in detail. The highest imprinting

2

factors (2.3-3.0) and adsorption capacity for targeted PASHs were achieved at a template:monomer:cross-linker ratio of 1:4:8. The method with no sample or film pretreatment showed very good reproducibility for the extraction of PASHs from spiked seawater samples (RSDs ≤ 6.0%, n = 3), was linear (R2>0.9960) over a range of 0.5-40 µg L-1, and gave limits of detection n the range of 0.029-0.166 µg L-1. Keywords: molecularly imprinted polymers, thin-films, microextraction, seawater, polyaromatic sulfur heterocycles, interpenetrating polymer networks

3

1. Introduction Every year, 380 million gallons of oil enter the world’s oceans and coastal waterways from natural and human sources, including shipping traffic, oil spills, and the controlled discharge or intentional dumping [1]. Sulfur-containing compounds are common in petrogenic oils at as much as 10% of total mass depending on the source; mainly occurring as polyaromatic sulfur-heterocycles (PASHs) [2]. Because of the formation of sulfur oxides upon burning, they are usually removed by hydrodesulfurization process, in which hydrocarbons and hydrogen sulfide are formed [3]. This makes them a good marker for differentiating oil from produced water or natural seeps from inputs of processed petroleum products. PASHs, such as dibenzothiophene (DBT), are considered the most problematic components in petroleum products due to their possible mutagenicity, carcinogenicity, and acute toxicity, particularly when compared to other analogous compounds in the environment [4–7]. There are few methods of analysis in the literature that selectively target PASHs; most of these methods were developed for PAHs analysis. The methods for PASHs usually involve the use of non-selective solid-phase extraction (SPE), which uses significant volumes of organic solvents for pre-conditioning and elution [8–10]. Other species in the sample may be preconcentrated with target analytes, which can decrease the extraction efficiency and make subsequent analysis more complex. Selective adsorbents, including molecularly imprinted polymers (MIPs), can be used to circumvent some of these issues.

4

MIPs are common artificial molecular recognition material, formed by polymerization of a functional monomer in the presence of a template molecule, which is normally the target molecule or a similar analogue. Template removal after polymerization vacates binding sites selective for molecules with size and functionality similar to the template. Therefore, analytes with correct geometry and functionality should rebind selectively to these sites [11–13]. The polymeric framework of the MIPs must be rigid enough to maintain the structure of the cavities after template removal [13]. Most MIPs are also stable over a range of temperatures and pressures, and in acids, bases, and organic solvents [14]. They can also be made easily at low cost compared with other selective materials [12]. Optimization of the MIP composition is very important, particularly when non-covalent imprinting is used. The key to success in this process is the initial selection of a suitable monomer that offers the best possible interaction with the template, followed by the optimization of the polymer components ratios. Theoretically, any template can be used for imprinting, but template bleeding during analysis can artificially elevate the sample background and thus limits the application of MIP for trace analysis [15–17]. To overcome this problem, a pseudo-template which has an analogous structure to the original template can be used. Traditional MIP preparation methods, such as bulk polymerization, usually produce selective polymers with high affinity, but sometimes features low capacity and low mass transfer because of small or discontinuous pores.

5

Moreover, the grinding process produces particles with irregular shapes and sizes, and subsequent sieving is wasteful. All these limitations could lead to a MIP with poor analytical performance [13]. More uniform particles can be made using precipitation polymerization or grafting onto silica or solid core particles, although precipitation polymerization needs a large amount of template in the preparation process because of the high dilution factor and grafting method is complicated and time-consuming [13,14]. MIPs can be prepared as thin-films for solid-phase microextraction or incorporation into sensing devices. Such films can be made with high selectivity and surface area using spin coating and “sandwiching” techniques [18–20] with controlled and uniformed thicknesses and desirable mass transfer behaviour. However, spin-coating can has limitations, such as low viscosity and high volatility of components of the pre-polymerization mixture, which frequently make this method unsuitable [18,21]. “Sandwiching” and drop-casting techniques are simple and effective alternatives methods for thin-film preparation. The role of the porogen (solvent) is significant, and careful selection of a proper porogen is necessary to produce a porous polymer. Some experimental studies found that the highest analyte-polymer rebinding during the analysis can be obtained from the same solvent used in the imprinting process [22,23]. For example, water has been used as a porogen with a water-compatible polymer for the detection of 1-methyladenosine in human urine samples [24]. Unfortunately,

6

imprinting in water can be difficult due to the limited range of water-soluble monomers and cross-linkers [25]. To our knowledge, there is no published work in the literature on the preparation or use of MIP thin-films for the analysis and detection of PASHs in aqueous media. Most MIPs prepared are used to remove these compounds from fuel or petroleum products [26–28]. In this research, a novel porous MIP thin-film was prepared on a glass substrate using an approach that takes advantage of interpenetrating polymer networks (IPNs) during polymerization [29]. The adsorption behavior and selectivity of the MIP thin-films were investigated in detail using the Freundlich isotherm model. This model was chosen due to its capability of evaluating the heterogeneity of MIPs. Reproducible MIP thin-films with relatively good imprinting properties and selectivity were used successfully to extract PASHs from water samples with no pre-treatment with analysis by GCMS. 2. Materials and methods 2.1. Materials Benzothiophene dibenzothiophene

(≥99%),

(≥95%),3-methylbenzothiophene 4,6-dimethyldibenzothiophene

(96%), (97%),

2-

thiophenecarboxaldehyde (98%), p-cresol (99%) and indole (98%), fluorene-d10 (99%), methacrylic acid (99%), 4-vinylpyridine (95%), acrylamide (≥99%), 1vinylimidazol (≥99%), acrylonitrile (≥99%), allylamine (≥99%), ethylene glycol dimethacrylate (98%), bisphenol A dimethacrylate (˃98%), 2.2-dimethoxy-27

phenylacetophenone (99%), polyethylene glycol (average MW 20,000), 3(trimethoxysilyl)propyl methacrylate (98%) were all purchased from Sigma Aldrich (Oakville, Canada). Methanol, acetonitrile, chloroform, hexane, toluene, 1-octanol and hydrochloric acid (37% w/w) were ACS reagent grades and purchased from ACP chemicals. Glass slides (25 × 75 mm2) and micro cover glasses (18 × 18 mm2) were purchased from Fisher Scientific (Toronto, ON, Canada). Deionized water (DI) was ultrapure water (18 MΩ.cm) was produced by Barnstead Nanopure Diamond water purification system (Lake Balboa, CA, USA). 2.2. Pre-treatment and derivatization of glass slides The glass microscope slides were cut (25 x 22 mm2), cleaned and functionalized before coating with MIP films. Cut slides were soaked in a solution of methanol/hydrochloric acid (1:1) for 30 min, and then rinsed with DI water, and dried at room temperature. The cleaned slides were placed in a silanizing solution of 2% (v/v) of 3-(trimethoxysilyl) propyl methacrylate in toluene overnight. The derivatized slides were washed with methanol and then dried under nitrogen. The prepared slides were stored in a dark place until use. 2.3. Preparation of MIP thin-films Preparation

of

MIP

thin-film

was

carried

out

via

photo-radical

polymerization of a thin layer of liquid pre-polymerization solution sandwiched between a derivatized glass slide and a quartz cover glass slide as shown in Fig. 1. Modification of the glass surface by silanization is necessary to improve the polymer adhesion to the surface, allowing for covalent bonding of the polymer 8

network to the glass [30,31]. The pre-polymerization mixture of each MIP thin-film was prepared according to Table S1. The solutions were degassed in a sonicator for 5 min, mainly to remove dissolved air, which can interfere with radical polymerization. Each MIP slide was prepared using 8-µL of the prepolymerization solution pipetted onto the surface of a derivatized glass slide and covered immediately by a quartz cover slide. Once the solution formed an evenly distributed layer, it was exposed to a UV-light (λ=254 nm, 6W) for 45 min. The cover slide was immediately removed using a sharp blade. The glass slide with the MIP coating was placed in methanol for 3 h to extract the template and unreacted components. This template extraction step was repeated for 1 h with fresh methanol. The MIP coating was rinsed with methanol followed by DI water and dried at room temperature. The non-imprinted polymer (NIP) was prepared in the same way but in the absence of the pseudo-template. The mass of the film (mfilm, ~ 4 mg) was obtained by subtracting the mass of the glass slide before polymerization from the mass with the coating. An analytical balance (Mettler Toledo XS 105) with an accuracy of 0.01 mg was used for these measurements. 2.4. Characterization of the thin-films Scanning electron microscopy (SEM) images were taken using FEI Quanta MLA 650F SEM (Hillsboro, OR, USA), the acceleration voltage of 15 kV, and at a magnification of 50,000x. The film thickness was determined using the instrument software from the cross-section images for the edges of MIP thin-film

9

which taken with a 36° tilt angle at 10 kV accelerating voltage. All samples were sputtered with gold prior to analysis. Analytical performance and optimization of the MIP thin-films were characterized by quantifying the bound analytes, extracted and analyzed with an Agilent 6890 gas chromatograph (Mississauga, ON, Canada) equipped with an Agilent 7683 auto-sampler and coupled to an Agilent 5973 mass selective detector (MSD). Instrument control and data analysis were performed using Agilent ChemStation software version D.01.00 Build 75. An Agilent fused-silica capillary column (DB-5MS, 30 x 0.25 i.d., 0.25 µm stationary film thickness) was used with helium (purity 99.999%) as the carrier gas with a flow rate of 1.2 mL min-1. The initial oven temperature was held at 50 °C for 1 min, and then increased to 280 °C at 20 °C min-1, and was held at this temperature for 1 min. The total analysis run time was 13.5 min. The injector temperature was set at 280 °C, and the injection was performed in the splitless mode. 2.5. Optimization and rebinding experiments Several porogens, specifically, chloroform, methanol, acetonitrile, 1octanol, and 1-octanol: methanol (1:1) were tested. To select the best monomer, six MIP films (MIP1-6, Table S1) were prepared first and studied. The template:monomer:cross-linker (T:M:C) in a typical mole ratio of 1:4:20 was used in the preparation of these compositions. MIP7 and 8 were prepared to study the effect of using different cross-linker and the addition of PEG (MW 20,000) polymer on the binding capacity and porosity of the polymers (Table S1). The

10

T:M:C ratio was optimized by testing different ratios for the 1-Vim-BPADMA polymer (MIP9-11, Table S2). DMPA was used as a photoinitiator in all MIP preparations. In the rebinding experiments, each MIP slide was placed in a 100-mL beaker containing 40 mL of water spiked with a mixture of PASHs at a concentration of 100 μg L-1 with magnetic stirring at room temperature for 2 h. The solutions for spiking PASHs were prepared in acetonitrile and stored in the refrigerator at 4 °C. Working solutions were prepared daily by appropriate dilution of the spiking solutions with distilled water. The amount of analyte adsorbed in the MIP on each slide Qt (µg g-1) at time t was calculated using the following equation:

Qt =

(C0 Ct )

(1)

where C0 (µg L-1) is the initial concentration of PASH compound in the water sample, Ct (µg L-1) is the concentration of PASH compound in the water sample at time t, V (L) is the volume of the sample and mfilm (g) is the mass of the polymer film. 2.6. Adsorption kinetics and binding isotherm experiments Binding experiments were performed to examine the adsorption kinetics and binding isotherms for the optimum composition of MIP thin-film (MIP9, Table S2). In the adsorption kinetics experiments, each MIP slide was placed in a 250mL beaker containing 200 mL of seawater spiked with PASHs at a concentration

11

of 50 µg L-1 with magnetic stirring. Uploading was studied for specific periods (124 h) at room temperature. The amount of adsorbed analyte in each slide Q t (µg g-1) at time t was calculated using Equation (1). The adsorption binding isotherm experiments were carried out by placing MIP (or NIP) films in 100-mL beakers, each containing 40 mL of seawater sample spiked with PASHs at initial concentrations ranging from 5 to 80 µg L-1. The samples were left to upload for 15 h (equilibration time) at a stirring speed of 500 rpm. The binding capacity of analyte in each slide at equilibrium (Q, µg g-1) was calculated using the following equation:

Q=

(C0 C)

(2)

The heterogeneity index (m) and fitting parameter constant (a, µg g-1) were determined from the Freundlich isotherm (FI) curve, which was obtained from plotting of the following linearized form of the equation: log Q = m log C

log a

(3)

where C is the free concentration of the analyte in the sample at equilibrium (µg L-1). The imprinting factor (IF) values of targeted PASHs were calculated using the following equation: IF =

Q Q

( )

12

2.7. Analysis of Seawater samples Seawater samples were collected from St. John’s Harbour, NL, in 4-L precleaned amber glass bottles. No preservation method was used in the sampling process. The bottles were filled to the top to eliminate headspace, stoppered with screw caps, and sealed with Parafilm and stored in a cold place (~4 °C) until use. All water and seawater samples were analyzed using MIP slides without pre-treatment. Each MIP slide was placed in a beaker containing 100 mL of spiked seawater at a known concentration (μg L−1) of PASHs with stirring (500 rpm) for 2 h. Later, the slide was removed from the beaker and rinsed with DI water and left to dry at room temperature for ~5 min. After drying, each slide was placed in a beaker containing 10 mL of hexane for extraction with stirring for 2 h. After extraction, the slide was removed, and the extract spiked with the internal standard (fluorene d-10) and then reduced to 1 mL under nitrogen. A 1-µL portion of this solution was injected into the GC-MS for analysis. The mass spectra were recorded in a selected ion monitoring (SIM) mode, using 70 eV in the electron ionization source. The quantifier ions for the SIM were: m/z = 134 for BT; m/z = 148 for 3-MBT; m/z = 184 for DBT; m/z = 212 for 4,6-DMDBT; m/z = 107 for pcresol and m/z = 117 for indole. The limit of detection (LOD) for each analyte was calculated as three times the standard deviation of measurements of the lowest detectable concentration over the slope of the calibration curve. All MIP films analyses were conducted in triplicate.

13

3. Results and discussion 3.1. Optimization of MIP thin-film composition 2-Thiophenecarboxaldehyde (Fig. S1) was used as pseudo-template in the preparation of MIP films to avoid complications of template bleeding, which may occur during the analysis. The selection of this template was made based on the potential for π-π, sulfur-nitrogen, sulfur-oxygen, and hydrogen bonding interactions with the tested monomers (Fig. S2). These monomers were used in excess compared to the selected template to shift the equilibrium towards the formation of the template-monomer complex, which typically leads to the formation of more cavities with selective recognition in the MIP. Chloroform, methanol, acetonitrile, and 1-octanol were examined to find the optimal porogen for the preparation of MIP films with the desired performance characteristics. Among these porogens, 1-octanol gave a stable and opaque film with a uniform surface, while chloroform, methanol, and acetonitrile produced films that were brittle and flaky, with a glassy appearance (Fig. S3 a, b and c). The features of the film formed with 1-octanol can be attributed to its poor ability to solvate the growing polymer, which leads to early phase separation whereas the other porogens are better solvents for the forming polymer resulting in later phase separation with a micro-porous morphology [18,32], which for our work gives a glassy polymer film with poor stability. Although the use of 1-octanol gave the best thin-film appearance, it showed poor imprinting factors (IF ≤1.13) for all PASHs. A high value of IF is an indicator of successful imprinting and evidence

14

that the selective interactions exceed the non-selective sorption capacities intrinsic to the polymer components. The low IF values for PASHs may be a consequence of low porosity or low surface area and thus limited access to selective sites [11,33]. An attempt was made to change the point of phase separation and increase the porosity of the polymer using a mixture of 1octanol:methanol (1:1). An improvement in the IFs (Table S3) was observed, especially using the 1-vinylimidazole monomer, which produced a MIP (MIP4) with the highest binding capacity among the monomers tested (Fig. S4). As a result, 1-vinylimidazole was selected as the best monomer for PASHs-MIP. Although an improvement in the IFs was observed using 1-octanol:methanol, the reproducibility was unsatisfactory (RSDS 5.0-18.0%). Thus, a new approach was taken to improve porosity and reproducibility of MIP thin-films. PEG (MW 20,000) was added to the pre-polymerization mixture at 12% (w/w) of the total mass of the pre-polymerization mixture (MIPs 7 and 8, Table S1). For this formulation, acetonitrile was used as the porogen; although it is not an ideal porogen on its own, it was chosen for its ability to solvate all the prepolymerization

components.

PEG

is

thought

to

create

a

system

of

interpenetrating polymer networks, which helps to support a porous structure as the polymer grows, mediating the influence of the porogen on phase separation and porosity. The pre-polymerization solution is also made more viscous with PEG, which is desirable in forming an even layer between the slides in the dropcasting fabrication technique. Following polymerization, the template and PEG

15

are removed, to yield a porous wettable MIP similar in appearance (uniform thickness and opacity) to MIPs using 1-octanol as a porogen (Fig. S3 e). Most importantly imprinting factors and binding capacity were improved, but further optimization of the polymer composition was required. In an effort to further improve the adsorption properties of our MIP film, the cross-linker EDGMA was replaced with bisphenol A dimethacrylate (BPADMA) to prepare a new MIP (MIP8, Table S1). The binding capacities of this polymer were compared with the binding capacity of the polymer prepared with EDGMA (MIP7), as seen in Fig. S5. The binding capacity increased more than doubled for BT and increased by ~30% for 3-MBT. This improvement may be partially attributed to additional π-π interactions of the thiophenic ring with BPADMA. Due to the size of the template, these selective sites are not easily accessible by the larger DBT and 4,6-DMDBT, and very little change was seen in their binding capacity. The effect of the degree of crosslinking in MIPs is widely acknowledged. Typically, this is studied by keeping the amount of monomer fixed and varying the amount of crosslinker added to the pre-polymerization solution. However, because the molar mass of BPADMA is relatively high, changing the amount of the cross-linker can have a large effect on the mass of polymer film formed. Since the binding capacity is dependent on the mass of the polymer, this makes it difficult to decouple changes in mass from the effect of imprinting. As such, we kept the mass of cross-linker constant and varied the amount of template and monomer, while keeping the template to monomer ratio constant at 1:4 (MIP9-

16

MIP11, Table S2). The percentage of PEG was also increased from 12% to 14% (w/w) in the preparation of MIP9-MIP11 to have a more viscous prepolymerization solution and control the formation of even layer between the glass slides. The result is a reasonably consistent mass of polymer on each slide. Binding capacities of each analyte for the MIPs and NIPs and IFs in are presented in Table 1. As the amount of monomer is decreased, the binding capacity decreases only slightly for the MIPs, whereas for the NIPs show an increase. The NIPs with less monomer should have a higher degree of crosslinking, but also be slightly less dense (based on the mass of polymer components). Thus, we attribute the increase in binding capacity to increased porosity and higher surface area, as non-selective interactions dominate. In this context, the effect of the presence of the template and the imprinting effects are dramatic. For the highest amount of monomer (MIP/NIP 9), the inclusion of the template more than doubles the binding capacity. As the monomer content decreases, even increases in non-selective binding cannot offset the loss in binding associated with the selective interactions, thus the binding capacities for MIPs decrease. This results in a loss in selectivity and lower IFs. The best imprinting factors were achieved with MIP 9 for all analytes (BT, 3-MBT, DBT, 4,6-DMDBT, 3.0, 2.8, 2.9, and 2.3, respectively). Further attempts to improve the performance by increasing the template to monomer to cross-linker ratios higher than 1:4:8 resulted in unstable polymers that gave irreproducible results.

17

Table 1. Binding capacities, imprinting factors range, and relative standard deviations for 1-Vim- BPADMA MIP at different T:M:C ratios. -1

Qt (µg g ) RSD Polymer

T:M:C

BT

3-MBT

DBT

4,6-DMDBT

IFs range (%)

NIP9

0:4:8

166

155

258

77

-

≤ 5.2

MIP9

1:4:8

495

431

762

178

2.3-3.0

≤ 4.9

NIP10

0:2:8

182

162

311

86

-

≤ 3.7

MIP10

0.5:2:8

452

411

613

127

1.5-2.5

≤ 2.9

NIP11

0:1:8

266

227

383

101

-

≤ 1.4

MIP 11

0.25:1:8

424

380

697

141

1.4-1.8

≤ 2.3

3.2. Morphology of the MIP thin-films SEM images were taken of the surface of the MIP9 and NIP9 thin-films (Fig. S6 a and b). The MIP films also showed a consistent thickness measurement at the edge of ~13.5 µm (Fig. S6 c). Both MIPs and NIPs exhibited porous surfaces, but the surface of the MIP appeared to be significantly more porous, with a more uniform surface.

18

Although the addition of PEG gives stable porous polymers, the differences in morphology between MIPs and their analogous NIPs illustrate the importance of the template-monomer complex in the formation of the pore structure. This effect is not limited to the macro-scale porosity, but it was also observed in binding study data (Table 1), which showed higher binding capacities for MIP9 compared to NIP9. 3.3. Optimization of MIP thin-film extraction procedure Several factors have been recognized as having an influence on the SPME method sensitivity and the figures of merit; we have limited our studies to stirring speed, sample volume, desorption solvent, and desorption time. Extraction time is also critical, and thus it extraction kinetics is presented as a separate study. Higher stirring speed reduces the time required to reach the equilibrium by enhancing the diffusion of the analyte towards the MIP film and produces consistent mixing for the sample. Various speeds of stirring (125, 350, and 500 rpm) for 2 h uploading was tested. The most reproducible and highest recoveries for all PASHs were obtained at a stirring speed of 500 rpm. No further increase was observed after stirring speeds higher than 500 rpm. Higher sensitivity can be achieved by increasing the sample volumes analyzed or using longer extraction times. A study of the effect of the sample volume is presented in Fig. S7. Although some of the features of these plots may

19

seem irregular, we repeated the experiments and the responses reported were reproducible. Between 10 mL and 50 mL, a sharp increase in the PASHs responses was observed, with only small changes in adsorption between 50 mL to 200 mL, which we believe is related to the saturation of the highest affinity imprinted sites and subsequent population of lower affinity sites with different partition coefficients. Although the amount of analyte adsorbed tends to increase with volume, the extraction efficiency decreases, and thus free analyte concentrations increase. These factors lead incremental increases in adsorption for all but DBT, which appears to be displaced by the analytes with more favorable affinity for the adsorbent. For volumes between 200 and 800 mL there is a large increase in the masses of analytes available and the behaviour is consistent with the adsorbent becoming saturated. Further volume increases will only have modest effects on sensitivity. Although using larger volumes than 40 100 mL will provide gains, large sample volumes are impractical for both sample collection and sample handling in the lab. Nevertheless, these results reveal that the MIP films are suitable as an adsorbent at low and high sample volumes. The nonpolar solvent hexane was selected as the desorption solvent due to its ability to extract all PASHs from MIP film and volatility, which makes the analyte extracts easy to concentrate. We also made preliminary tests of other solvents, for example dichloromethane, but none gave better results or were more environmentally friendly than hexane. It was found that 2 h desorption of analytes from the MIPs gave the highest and most reproducible recoveries.

20

3.4. Adsorption kinetics of MIP thin-film The effect of adsorption time on the loading of PASHs onto the MIP film was investigated over 24 h. Fig. 2 shows that the binding of all PASHs increased rapidly within the first 8 h, with a slower increase until equilibrium was reached at ~15 h. No further increases were detectable with longer upload times. The nonalkylated thiophenes, BT and DBT, showed the highest binding, and this cannot be explained by hydrophobicity. We attribute this to molecular recognition that preferentially binds these compounds. The other explanation that has been considered is differences in diffusion rates, which would be higher for the lightest compounds. However, the similarity in time to equilibrium and the higher binding for DBT compared to 3MBT has led us to dismiss this theory. The linear uptake within the first 8 h shows that the MIP thin-films are suitable for the 2 h uptake times we have selected. Shorter (or longer) times can be used depending on the sensitivity needed with excellent reproducibility. Although 2 h may not be considered rapid, there is no pre-treatment of the sample or pre-conditioning of the film required, which makes it easy to process many samples simultaneously, increasing throughput.

3.5. Adsorption isotherms for PASHs in seawater Several binding isotherm models have been used to evaluate the binding equilibrium behavior of MIPs [34]. Since it has been recognized that MIPs made through non-covalent imprinting typically have heterogeneous binding sites, the 21

FI model, which has a heterogeneity parameter (m), was the best model to study the MIP and NIP binding behavior [34–36]. FI curves (Fig. 3) were obtained by plotting log Q versus log C according to Equation (3), and m and a was determined from the slope and intercept, respectively (Table 2). The m values can range between 0 and 1; as m goes to 1, the binding sites become more homogeneous. The fitting parameter, a, is related to the average binding affinity, K0 (K0 = a1/m). Initial conclusions drawn from the plots confirm that the MIP films exhibit higher adsorption capacities for the PASHs than the NIP films. The values of m and a indicate that the MIPs exhibit more homogeneous binding site energies and these sites bind the analytes much more strongly (i.e., higher affinity), particularly for the non-alkylated PASHs. The high IF values for BT and DBT are a consequence of these features. The differences between the MIP and the NIP also exemplified in the goodness of fit for the model (R2), where the MIP showed

better

fitting

to

the

FI

model

for

all

the

analytes.

Table 2. Freundlich fitting parameters to the adsorption isotherm for MIP and NIP

Fitting parameter

a (µg g-1)

m

K0 (µg g-1)

R2

Analyte

MIP

NIP

MIP

NIP

MIP

NIP

MIP

NIP

BT

0.976

0.893

26.903

5.675

29.222

6.985

0.9918

0.9863

3-MBT

0.974

0.795

12.274

5.665

13.135

8.862

0.9990

0.9810

DBT

0.969

0.797

25.698

8.087

28.511

13.791

0.9963

0.9913

4,6-DMDBT

0.765

0.696

8.039

4.152

15.277

7.726

0.9645

0.9618

22

3.6. Method validation Method accuracy and precision (intra and inter-day) were evaluated using a recovery study for seawater samples spiked with a mixture of PASHs at three concentrations (2 µg L-1, 15 µg L-1 and 30 µg L-1). All analyses were conducted in triplicate. The figures of merits including linear range, correlation coefficient (R2), limit of detection (LOD), method accuracy and precision, are summarized in Tables 3 and 4. The results showed very good linearity (R2 ≥ 0.9962) for each PASH compound. The limits of detection values in seawater were 0.029, 0.040, 0.068, 0.166 µg L-1 for BT, 3-MBT, DBT, and 4,6-DMDBT, respectively. Accuracy was in the range of 77 to 121%, intra-day precision ranged from 1.2% to 5.8% and the inter-day precision from 2.1% to 13.5%. These results reflected the suitability of the developed method for the determination of PASHs in seawater.

3.7. Analysis of seawater samples To evaluate the suitability of the new MIP films for the extraction of PASHs from real samples, natural seawater samples were spiked with PASHs at concentrations in the range of 0.5-40.0 µg L-1. Analysis of seawater samples was carried out without pre-treatment, and the results showed very good linearity (R2) for each PASH compound (Table 4). The limit of detection values in seawater sample were 0.029, 0.040, 0.068, 0.166 µg L-1 for BT, 3-MBT, DBT, and 4,6DMDBT respectively. Acceptable reproducibility was obtained with low relative

23

standard deviations (RSDS ≤ 6.0 %). We do note that PASHs were undetectable in the raw seawater samples (non-spiked). The recorded peak signals in the selected chromatograms (Fig. S8) for spiked seawater (at low and high concentrations) versus non-spiked seawater shows good sensitivity for the detection of PASHs in seawater and demonstrate the goodness of the device presented in this work. There are few methods of analysis for PASHs in the literature that specifically target them; most of the methods reported are for analysis of PASHs with PAHs. These methods usually involve the use of non-selective solid-phase extraction (SPE) with a range of detection methods [8–10]. The analytical performance of the method reported here is comparable to other methods reported in the literature (Table S4). Although these methods show comparable figures of merit, our methods show some advantages over these methods. For example, the sample treatment required is simple and requires no special equipment (e.g., syringe filters, a vacuum manifold, etc.), and minimal amounts of reagents are consumed. The MIP materials can be used directly without conditioning of the solid phase. As with stir bar SPE, once the analyte has been extracted no further manipulation is required as analytes are desorbed directly from the polymer coating. With the -stir bar SPE method, carryover is a concern, since the bar must be reused several times to make it cost-effective. The developed method improves the selectivity of analysis and simplifies the extraction process which reduces the need for sample pre-treatment. A

24

comparison between the adsorption recovery values of PASHs in deionised (DI) water and seawater (SW) was made to assess matrix effect on the adsorption process (Table 5). The results indicate that the effect of matrices on the adsorption of PASHs by MIP was acceptable (93-97%), except in the case of 4,6DMDBT, where at 85% the effect was more pronounced. The observed decrease in the adsorption recovery of 4,6-DMDBT can be explained by the nature of the interactions between this analyte and the MIP in water, where molecular recognition is driven by geometry and π-π interactions. 4,6-DMDBT has two methyl substituents on the benzene rings; these donating groups may weaken the π-π interactions [37] with the MIP, making the hydrophobic interaction more dominant. Thus, the matrices present in seawater could compete and disrupt the hydrophobic interaction, causing a decrease in the adsorption recovery of the 4,6-DMDBT.

25

Table 3. Calibration curves for determination of PASHs in DI water using MIP thin-films Analyte

Linear range -1 (µg L )

2

Equation

R

%RSD

y = 0.0594x+ 0.0018

0.9990

≤ 3.6

y = 0.0218x+ 0.0004

0.9981

≤ 5.1

DBT

y = 0.0408x+ 0.0016

0.9987

≤ 2.4

4,6-DMDBT

y = 0.0099x+ 0.0015

0.9977

≤ 5.6

BT 3-MBT 0.5-40

Table 4. Method validation and figures of merit of for determination of PASHs in seawater using MIP thin-films

Analyte

Linear range

Equation

2

R

-1

-1

(µg L )

BT

% Accuracy (± %RSD)

%RSD

Intra-day

Inter-day

LOD (µg L )

2 µg L

-1

15 µg L

-1

30 µg L

-1

2 µg L

-1

-1

-1

15 µg L

30 µg L

y = 0.0575x+ 0.0022

0.9968

0.029

83 (±5.7)

112 (±2.7)

98 (±3.1)

10.0

2.3

2.1

y = 0.0195x+ 0.0006

0.9962

0.040

106 (±4.3)

104 (±5.8)

108 (±3.5)

13.5

3.5

2.9

DBT

y = 0.0402x+ 0.0019

0.9980

0.068

77 (±3.7)

103 (±3.2)

99 (±1.2)

7.3

4.8

1.8

4,6-DMDBT

y = 0.0081x+ 0.0015

0.9968

0.166

121 (±4.8)

110 (±3.9)

104 (±4.1)

8.4

5.1

3.6

3-MBT

0.5-40

26

Table 5. Percent recovery values PASHs obtained from the analysis of 100-ml sample of spiked DI water and seawater samples at the same concentration (50 µg L-1) using MIP thin-film. %Recovery (SD, n = 3)

Matrix effect (% of DIW recovery)

Analyte DI

SW

BT

16.8 (0.3)

16.3 (0.4)

97

3-MBT

13.4 (0.4)

12.5 (0.5)

93

DBT

14.8 (0.3)

14.2 (0.4)

96

4,6-DMDBT

7.9 (0.3)

6.7 (0.6)

85

3.8. Selectivity of PASHs-MIP thin-film MIP selectivity against interferents is an important advantage. The selectivity is dependent on the specificity of the interactions of the MIP with the analytes, as well as geometric factors. Hydrophobic, π-π interactions, and often hydrogen bonding are typically bonding modes of for non-covalent imprinting. We carried out a selectivity study using two likely interferents that feature functionalities and or shapes similar to the PASHs. This type of study also probes the nature of the analyte-MIP adsorption chemistry. Specifically, p-cresol (a common polar aromatic pollutant) and indole (a nitrogen-heterocycle analog of benzothiophene) were spiked with the PASHs in seawater at the same concentration (100 µg L-1), and then analyzed with the MIP-GC-MS method. Binding capacities were calculated and are presented in Fig. 4. The results show that the MIP film exhibited higher selectivity toward all PASHs compared to other interferents, except for 4,6-DMDBT, which showed adsorption equivalent to indole. This may be attributed to the steric effect of the two methyl groups in 4,6-

27

DMDBT, which restricts its access to selective cavities of the MIP. If the hydrophobic effect dominants in the MIP-analyte interactions, then the sorption should trend with log Kow values (Table S5 [38,39]), i.e., the binding capacity should be higher for heavier PASHs like 4,6-DMDBT, which has the highest log Kow. Since the best binding capacity was observed for BT, which has the lowest log Kow value, it can be concluded that the mechanism of uptake of PASHs is not dominated by hydrophobic interactions but is likely the result of a confluence of factors including steric effects and π-π interactions between the MIP and aromatic rings in PASHs. BT is the least sterically hindered and has the highest diffusion rate of the PASHs, which may, in part, explain why it shows the highest binding efficiency. 4. Conclusions In this research, a novel MIP thin-film on glass substrate was prepared as an adsorbent for the selective extraction of PASHs in water. No sample preparation or pre-treatment was needed prior to use as a sorbent. The MIP thinfilms displayed good binding behavior, high selectivity, reproducibility, linearity, and low LOD values for PASHs. The small size of the MIP film device and the ease of preparation makes them efficient for off-site and on-site environmental analysis when combined with a suitable detection system. The preparation of thin-films in this work is be promising for use in environmental sensors and other applications.

28

Conflicts of interest There are no conflicts to declare. Author contributions The first author (Hassan Y. Hijazi) declares that he contributed to all sections of the submitted paper (“Molecularly imprinted polymer thin-film as a microextraction adsorbent for selective determination of trace concentrations of polycyclic aromatic sulfur heterocycles in seawater”) including: literature review, carrying out all experiments, collecting and analyzing the data, developing and designing new experiments, performing all calculations, and writing and editing the initial draft of manuscript. The corresponding author (Christina S. Bottaro). Dr. Bottaro provided guidance for experiments, helped to interpret the data and to develop the discussion in the paper. There are no other co-authors were involved in the preparation of this manuscript.

Acknowledgments This work was supported by the Petroleum Research Newfoundland and Labrador (PRNL), Atlantic Innovation Fund (AIF), Natural Sciences and Engineering Research Council of Canada (NSERC), Department of Chemistry and School of Graduate Studies (SGS) at Memorial University of Newfoundland (MUN).

29

5. References [1]

J.W. Farrington, J.E. Mcdowell, Mixing Oil and Water, in: Science (80-. )., ProQuest Central pg, 2002: pp. 2156–2157.

[2]

James G. Speight, The chemistry and technology of petroleum, in: Fuel Process. Technol., CRC Press, Taylor and Francis, 1982: pp. 325–326.

[3]

B. Castro, M.J. Whitcombe, E.N. Vulfson, R. Vazquez-Duhalt, E. Bárzana, Molecular imprinting for the selective adsorption of organosulphur compounds present in fuels, in: Anal. Chim. Acta, 2001: pp. 83–90. https://doi.org/10.1016/S0003-2670(01)00799-1.

[4]

Y. Gui-peng, Photochemical oxidation of benzothiophene in seawater, Chinese J. Oceanol. Limnol. 18 (2000) 85–91. https://doi.org/10.1007/bf02842547.

[5]

K.G. Kropp, P.M. Fedorak, A review of the occurrence, toxicity, and biodegradation of condensed thiophenes found in petroleum., Can. J. Microbiol. 44 (1998) 605–22. https://doi.org/10.1139/cjm-44-7-605.

[6]

Analysis and Fate of Dibenzothiophene Derivatives in the Marine Environment, Int. J. Environ. Anal. Chem. 27 (1986) 81–96. https://doi.org/10.1080/03067318608078392.

[7]

A. Croisy, J. Mispelter, J. ‐M Lhoste, F. Zajdela, P. Jacquignon, Thiophene analogues of carcinogenic polycyclic hydrocarbons. Elbs pyrolysis of various aroylmethylbenzo[b]thiophenes, J. Heterocycl. Chem. 21 (1984) 353–359. https://doi.org/10.1002/jhet.5570210217.

30

[8]

P. Avino, I. Notardonato, G. Cinelli, M. Russo, Aromatic Sulfur Compounds Enrichment from Seawater in Crude Oil Contamination by Solid Phase Extraction, Curr. Anal. Chem. 5 (2009) 339–346. https://doi.org/10.2174/157341109789077696.

[9]

C. Yu, Z. Yao, B. Hu, Preparation of polydimethylsiloxane/βcyclodextrin/divinylbenzene coated “dumbbell-shaped” stir bar and its application to the analysis of polycyclic aromatic hydrocarbons and polycyclic aromatic sulfur heterocycles compounds in lake water and soil by hig, Anal. Chim. Acta. 641 (2009) 75–82. https://doi.org/10.1016/j.aca.2009.03.031.

[10] R.A. Gimeno, A.F.M. Altelaar, R.M. Marcé, F. Borrull, Determination of polycyclic aromatic hydrocarbons and polycylic aromatic sulfur heterocycles by high-performance liquid chromatography with fluorescence and atmospheric pressure chemical ionization mass spectrometry detection in seawater and sediment samp, J. Chromatogr. A. 958 (2002) 141–148. https://doi.org/10.1016/S0021-9673(02)00386-2. [11] K. Haupt, K. Mosbach, ChemInform Abstract: Molecularly Imprinted Polymers and Their Use in Biomimetic Sensors, ChemInform. 31 (2010) no-no. https://doi.org/10.1002/chin.200042261. [12] K.D. Shimizu, C.J. Stephenson, Molecularly imprinted polymer sensor arrays., Curr. Opin. Chem. Biol. 14 (2010) 743–50. https://doi.org/10.1016/j.cbpa.2010.07.007. [13] H. Yan, H.R. Kyung, Characteristic and synthetic approach of molecularly 31

imprinted polymer, Int. J. Mol. Sci. 7 (2006) 155–178. https://doi.org/10.3390/i7050155. [14] S. Scorrano, L. Mergola, A. Scardino, M.R. Lazzoi, R. Del Sole, G. Mele, G. Vasapollo, Molecularly Imprinted Polymers: Present and Future Prospective, Int. J. Mol. Sci. 12 (2011) 5908–5945. https://doi.org/10.3390/ijms12095908. [15] L.I. Andersson, A. Paprica, T. Arvidsson, A highly selective solid phase extraction sorbent for preconcentration of sameridine made by molecular imprinting, Chromatographia. 46 (1997) 57–66. https://doi.org/10.1007/BF02490930. [16] J. Matsui, K. Fujiwara, T. Takeuchi, Atrazine-selective polymers prepared by molecular imprinting of Trialkylmelamines as dummy template species of atrazine, Anal. Chem. 72 (2000) 1810–1813. https://doi.org/10.1021/ac9911950. [17] X. Liu, J. Liu, Y. Huang, R. Zhao, G. Liu, Y. Chen, Determination of methotrexate in human serum by high-performance liquid chromatography combined with pseudo template molecularly imprinted polymer, J. Chromatogr. A. 1216 (2009) 7533–7538. https://doi.org/10.1016/j.chroma.2009.06.018. [18] R.H. Schmidt, K. Haupt, Molecularly imprinted polymer films with binding properties enhanced by the reaction-induced phase separation of a sacrificial polymeric porogen, Chem. Mater. 17 (2005) 1007–1016. https://doi.org/10.1021/cm048392m. 32

[19] P.C. Chou, J. Rick, T.C. Chou, C-reactive protein thin-film molecularly imprinted polymers formed using a micro-contact approach, Anal. Chim. Acta. 542 (2005) 20–25. https://doi.org/10.1016/j.aca.2004.12.074. [20] D.R. Kryscio, N.A. Peppas, Surface imprinted thin polymer film systems with selective recognition for bovine serum albumin, Anal. Chim. Acta. 718 (2012) 109–115. https://doi.org/10.1016/j.aca.2012.01.006. [21] M. Jakusch, M. Janotta, B. Mizaikoff, K. Mosbach, K. Haupt, Molecularly Imprinted Polymers and Infrared Evanescent Wave Spectroscopy. A Chemical Sensors Approach, Anal. Chem. 71 (1999) 4786–4791. https://doi.org/10.1021/ac990050q. [22] D. Spivak, M.A. Gilmore, K.J. Shea, Evaluation of binding and origins of specificity of 9-ethyladenine imprinted polymers, J. Am. Chem. Soc. 119 (1997) 4388–4393. https://doi.org/10.1021/ja963510v. [23] M. Kempe, K. Mosbach, Binding studies on substrate- and enantioselective molecularly imprinted polymers, Anal. Lett. 24 (1991) 1137–1145. https://doi.org/10.1080/00032719108052959. [24] S. Scorrano, L. Longo, G. Vasapollo, Molecularly imprinted polymers for solid-phase extraction of 1-methyladenosine from human urine, Anal. Chim. Acta. 659 (2010) 167–171. https://doi.org/10.1016/j.aca.2009.11.046. [25] A.G. Mayes, M.J. Whitcombe, Synthetic strategies for the generation of molecularly imprinted organic polymers, Adv. Drug Deliv. Rev. 57 (2005) 1742–1778. https://doi.org/10.1016/j.addr.2005.07.011. [26] L.A. Tom, C.L. Gerard, C.M. Hutchison, A.S. Brooker, Development of a 33

novel molecularly imprinted polymer for the retention of 4,6dimethyldibenzothiophene, Microchim. Acta. 176 (2012) 375–380. https://doi.org/10.1007/s00604-011-0730-0. [27] P. Xu, W. Xu, X. Zhang, J. Pan, Y. Yan, Molecularly-Imprinted Material for Dibenzothiophene Recognition Prepared by Surface Imprinting Methods, Adsorpt. Sci. Technol. 27 (2010) 975–987. https://doi.org/10.1260/02636174.27.10.975. [28] H. Li, W. Xu, N. Wang, X. Ma, D. Niu, B. Jiang, L. Liu, W. Huang, W. Yang, Z. Zhou, Synthesis of magnetic molecularly imprinted polymer particles for selective adsorption and separation of dibenzothiophene, Microchim. Acta. 179 (2012) 123–130. https://doi.org/10.1007/s00604-012-0873-7. [29] L.H. Sperling, Interpenetrating Polymer Networks, Springer Netherlands, Dordrecht, 2004. https://doi.org/10.1007/978-94-007-6064-6_8. [30] M. Zayats, A.J. Brenner, P.C. Searson, Protein imprinting in polyacrylamide-based gels, Biomaterials. 35 (2014) 8659–8668. https://doi.org/10.1016/j.biomaterials.2014.05.079. [31] L.M. Kindschy, E.C. Alocilja, Development of a molecularly imprinted biomimetic electrode, Sensors. 7 (2007) 1630–1642. https://doi.org/10.3390/s7081630. [32] R.H. Schmidt, A.S. Belmont, K. Haupt, Porogen formulations for obtaining molecularly imprinted polymers with optimized binding properties, in: Anal. Chim. Acta, 2005: pp. 118–124. https://doi.org/10.1016/j.aca.2005.03.064. [33] C. Alvarez-lorenzo, C. Angel, Handbook of Molecularly Imprinted Polymers, 34

(2013) 402. [34] J.A. García-Calzón, M.E. Díaz-García, Characterization of binding sites in molecularly imprinted polymers, Sensors Actuators, B Chem. 123 (2007) 1180–1194. https://doi.org/10.1016/j.snb.2006.10.068. [35] A.M. Rampey, R.J. Umpleby, G.T. Rushton, J.C. Iseman, R.N. Shah, K.D. Shimizu, Characterization of the Imprint Effect and the Influence of Imprinting Conditions on Affinity, Capacity, and Heterogeneity in Molecularly Imprinted Polymers Using the Freundlich Isotherm-Affinity Distribution Analysis, Anal. Chem. 76 (2004) 1123–1133. https://doi.org/10.1021/ac0345345. [36] R.J. Umpleby, S.C. Baxter, M. Bode, J.K. Berch, R.N. Shah, K.D. Shimizu, Application of the Freundlich adsorption isotherm in the characterization of molecularly imprinted polymers, Anal. Chim. Acta. 435 (2001) 35–42. https://doi.org/10.1016/S0003-2670(00)01211-3. [37] S.E. Wheeler, Understanding substituent effects in noncovalent interactions involving aromatic rings, Acc. Chem. Res. 46 (2013) 1029–1038. https://doi.org/10.1021/ar300109n. [38] G. Durell, T. Røe Utvik, S. Johnsen, T. Frost, J. Neff, Oil well produced water discharges to the North Sea. Part I: Comparison of deployed mussels (Mytilus edulis), semi-permeable membrane devices, and the DREAM model predictions to estimate the dispersion of polycyclic aromatic hydrocarbons, Mar. Environ. Res. 62 (2006) 194–223. https://doi.org/10.1016/j.marenvres.2006.03.012. 35

[39] D.R. Luellen, D. Shea, Calibration and field verification of semipermeable membrane devices for measuring polycyclic aromatic hydrocarbons in water, Environ. Sci. Technol. 36 (2002) 1791–1797. https://doi.org/10.1021/es0113504.

Figures captions:

Fig. 1. Preparation steps of MIP thin-film on a glass slide.

36

Fig. 2. Adsorption kinetics of MIP thin-film for PASHs in 200 mL of DI water spiked at an initial concentration of 50 µg L-1 (n=3). Error bars represent ±1 SD.

Fig. 3. Adsorption isotherm of MIP and NIP for PASHs fitted to Freundlich isotherm (equilibration time15 h). Error bars represent ±1 SD.

37

Fig. 4. Selectivity of MIP thin-film for PASHs versus p-cresol and indole (n=3). Error bars represent ±1 SD.

38