QSAR aided design and development of biopolymer-based SPE phase for liquid chromatographic analysis of polycyclic aromatic hydrocarbons in environmental water samples

Abstract

A solid-phase extraction (SPE), using starch based biopolymer adsorbent, coupled with high performance liquid chromatography–fluorescence/UV detectors (HPLC-FLD/UV) method was developed for the determination of 16 polycyclic aromatic hydrocarbons (PAHs) in environmental water samples. Quantitative structure activity relationship was employed for assessing the suitability of epichlorohydrin, 1,6-hexamethylene diisocyanate and 4,4-methylene diphenyl diisocyanate (MDI) based starch, β- and γ-cyclodextrin biopolymers as the SPE sorbent phases, via density functional theoretical (DFT) modeling studies. The DFT parameters showed that MDI based biopolymers exhibited the highest retention potential as the SPE sorbent. The modelled materials were synthesized, characterized and evaluated for the extraction of (PAHs). The results of the screening SPE studies were in agreement with DFT quantum predictions. MDI-starch biopolymer was therefore selected as the best SPE phase. The calibration curves of extracted PAHs were linear in the range of 0.5–50.0 μg L⁻¹, with r² values of >0.99. The method attained good precisions values of 6.9–26.0 and 0.4–16.6% for UV and FLD detectors respectively; and the method detection limits values of 22.9–155.3 and 0.5–24.2 ng L⁻¹ for UV and FLD detectors respectively. The optimized method was successfully applied for the determination of 16 PAHs in real environmental water samples, and mean recoveries were predominantly >80% for tap and river water samples in both UV and FLD analysis. The values of SPE performance indicators showed that the developed biopolymer phase is better than similar studies reported in literature, and equally comparable to commercially available C₁₈ SPE phases.

---

1. Introduction

Robust pre-concentration/enrichment procedure is central to accurate and precise analysis of water pollution by toxic hydrophobic organic micro-pollutants. Levels of these pollutants in water are often below the detection capacities of most instrumental testing methods. Additionally, many instrumental techniques that incorporate a mass spectrometric (MS) detector like liquid and gas chromatography (LC/MS and GC/MS) are not very useful in detecting analytes in aqueous media; hence there is the need for change of the phase from aqueous to organic solvent. Also, considering the fact that high levels of other toxic and non-toxic components are found to be interfering with the analytes of concern, a clean-up step is often required. Among all the pre-concentration options, Solid Phase Extraction (SPE) is one of the few procedures that achieves the processes of enrichment, change of phase and clean-up, in one single step.^1–4

Amongst the hydrophobic micro-organic pollutants, aqueous pollution by polycyclic aromatic hydrocarbons (PAHs) has generated much public health concern due to their significant toxicity, relative persistence to biodegradation, potential mutagenicity, teratogenicity and carcinogenicity. Apart from being hydrophobic, the challenge of managing PAHs aqueous pollution is more tasking because PAHs exhibit mutagenic and carcinogenic toxicities even at ultra-trace levels.⁵ Consequently, there is continued reduction on the maximum permissible limit by most environmental regulatory bodies, which necessitates development of improved pre-concentration options.

To date, typical SPE adsorbents for PAHs are bonded silica phases (C₈, C₁₈, cyano and other groups), polystyrene-divinyl benzene copolymer (PS-DVB) based adsorbents, immunosorbents, metal oxides like magnetic oxide microspheres, sulphur microparticles, periodic mesoporous organosilica, magnetic carbon nitride, nanoparticles, molecularly imprinted polymers, and graphene related materials.^6–13 However, the application of these adsorbents for SPE pre-concentration of PAHs has been limited due to certain chemical and physical properties of these adsorbents, as well as prohibitive costs. For instance, the presence of residual surface silanol groups (even after an end-capping treatment) and a narrow pH stability range has posed great limitation to the continued application of bonded silica phases. Aside having biodegradation challenge, PS-DVB based SPE adsorbents have shown to be too hydrophobic, and hence, exhibit some level of irreversible interaction with PAHs, while the use of other adsorbents like immunosorbents is limited on cost basis, since they are very expensive. This situation therefore elicited strong research interest for development of SPE phases that are relatively cheap, biodegradable, and efficient for the pre-concentration and subsequent analysis of PAHs in aqueous media.

The polysaccharide based biopolymers represent an interesting and attractive alternative as adsorbents phases because of their fascinating structure, physicochemical characteristics, chemical stability, high reactivity and excellent selectivity towards aromatic compounds and metals, which emanates from the presence of chemical reactive groups (hydroxyl, acetamido or amino functional groups) in their polymer chains. Previous studies showed that starch and its derivative, cyclodextrin, have proved to be effective for the preparation of low-cost adsorbents for the removal of pollutants from wastewater.^14–17 Due to their level of hydrophilicity, which constitutes the major drawback in the utilization of starch and cyclodextrins adsorbent phases, modification of their chemical structure, to improve hydrophobicity, has been devised as a way out. Amongst other modification options, cross-linking process provides additional opportunity for tuning the affinity of the adsorbent phase to the analytes of interest, as a function of the nature of the cross linking agent as well as degree of cross linking process.

Despite the aforementioned advantages of starch and cyclodextrin's polymers, there are limited studies on the application of these biopolymers as SPE phases for chromatographic analysis. This is mainly because of the difficulty associated with the selection of suitable cross-linking agent which varies from one analyte to the other. Considering the array of crosslinking agents available to polymer scientists, the choice of selecting a suitable crosslinking agent is becoming increasingly difficult. However, the advent of Quantitative Structure Activity Relationship (QSAR) has transformed the search for optimal materials with desired properties based on chemical intuition and experience into a search with a mathematically founded and computerized form of approach. In recent times, QSAR modeling methods have gained acceptance as a veritable tool for computer aided design of polymer adsorbent phases.¹³ Generally, the ab initio computational methods such as the density functional theory (DFT) with various basis sets have been widely used, because they have proved to be adequate for pointing out changes in electronic structure of the analyzed system responsible for chemical interaction. Also, DFT gives exact values of the basic vital interaction parameters like orbital energies, energy gap, dipole moment, etc., for even huge complex molecules at relatively low computational cost.^18–20

In this study, starch and β-cyclodextrin based polymer were computationally designed based on the electronic structure by using the DFT approach, to select the most suitable crosslinking agent. The designed polymers were synthesized, screened and optimized as sorbent for the solid phase extraction of PAHs, and the obtained results were compared with those achieved using commercial reverse phase SPE sorbents.

2. Experimental

2.1 Molecular modelling study

The biopolymers were theoretically designed by cross-linking each of starch, β-cyclodextrin and γ-cyclodextrin with three different cross-linking agents. Fig. 1a showed reaction scheme for cross-linking reaction of starch with 4,4-methylene diphenyl diisocyanate (MDI) to generate MDI crosslinked starch biopolymer (MSB), while Fig. 1b represented the reaction of with MDI to generate MDI crosslinked β-cyclodextrin biopolymer (MBB). The structure of MDI crosslinked γ-cyclodextrin biopolymer (MGB) is expected to be similar to MBB except the difference in the carbon chain length in the different forms of the cyclodextrin. In a similar vein, each of starch, β-cyclodextrin, γ-cyclodextrin was separately cross-linked with each of the remaining two crosslinking agents viz.: hexamethylene-1,6-diisocyanate (HDI) and epichlorohydrin (EPI).

Fig. 1 Reaction scheme for the synthesis of (a) cross-linked starch biopolymer and (b) β-cyclodextrin biopolymer, using 4,4-methylene diphenyl diisocyanate (MDI) as cross-linking agent.

Being polymers that are constituted of repetitive monomer units as seen in Fig. 1, it is more convenient to represent each of the polymers with its monomer unit for modeling purpose. The cyclodextrin ring was not included as part of the monomer unit for cyclodextrin based polymers, because any effect (like the host–guest interaction) that may arise from the presence of the ring will occur in all the cyclodextrin biopolymers irrespective of the crosslinking agent. Based on this approach, the monomer units for MSB, MBB, and MGB are expected to be the same structure, as confirmed from the structures shown in Fig. 1. Similarly, the monomer units for HDI crosslinked starch biopolymer (HSB), HDI crosslinked β-cyclodextrin biopolymer (HBB) and HDI crosslinked γ-cyclodextrin biopolymer (HGB) were expected to be the same, while the monomer units for EPI crosslinked starch biopolymer (ESB), EPI crosslinked β-cyclodextrin biopolymer (EBB) and EPI crosslinked γ-cyclodextrin biopolymer (EGB) were expected to be the same. Based on this approach, a general scheme for the crosslinking reaction using the three different crosslinking agents to generate three distinctive monomer units is as shown in Fig. 2. In essence, the monomer unit for MDI based biopolymer represented MSB, MBB, and MGB; the monomer unit for HDI based biopolymer represented HSB, HBB, and HGB; while the monomer unit for EPI based biopolymers represented ESB, EBB, and EGB.

Fig. 2 General reaction scheme for the cross-linking of the starch/cyclodextrin.

Considering the large surface area of adsorbent phases, it is impossible to completely model the retention process on the entire surface in any surface calculation with density functional or ab initio methods, even with robust computer codes and highly efficient super computers.^21,22 Hence, a section of the polymer (the monomer units) was used to represent the surface of the sorbents, often terminating unfilled valencies with hydrogen atoms, to mimic a continuous surface. All the quantum calculations with full-geometry optimizations were performed using Spartan 10v1.0.1.suite of programs.²³ The monomer units were first subjected to energy optimization and the conformer with lowest energy (most stable) were adopted for the DFT calculations. The DFT method at B3LYP/6-31G level was used for calculating the relevant chemical interaction descriptors viz.: energy of highest occupied molecular orbital (E_HOMO), energy of lowest unoccupied molecular orbital (E_LUMO), energy gap, electron affinity, and dipole moment of adsorbents. The details for the calculation of energy gap, ionization potential and electron affinity can be assessed from the ESI.†

2.2 Chemical reagents and standards

The US EPA PAH mix standard stock solution (PAH solution 16–22, 200 mg L⁻¹) was obtained from O2Si smart solution® South Carolina, USA, and supplied by ANPEL Scientific Company Limited China. The US EPA PAH mix had the following PAHs at concentration of 200 mg L⁻¹ dissolved in acetonitrile: naphthalene, acenaphthylene, acenaphthene, fluorene, phenanthrene, anthracene, fluoranthene, pyrene, benzo[a]anthracene, chrysene, benzo[b]perylene, benzo[k]fluoranthene, benzo[a]pyrene, dibenzo[b]anthracene, benzo[g]perylene, indeno[c]perylene. LC grade of acetonitrile and methanol were purchased from Fisher Scientific, New Jersey, USA and MREDA Technologies Inc. USA, respectively. Epichlorohydrin was supplied by Aladdin Chemistry Company Ltd. Shanghai, China, while N,N-dimethylformamide (DMF), 4,4-methylene diphenyl diisocyanate (MDI) and 1,6-hexamethylene diisocyanate (HDI) were supplied by Aladdin Chemistry Company Ltd. Shanghai, China. Soluble starch was supplied by Guangfu Research Institute and β-cyclodextrin (analar) was supplied by Tianjin Kermel Fine Chemical Company, Tianjin. Ultrapure water was prepared using Milli-Q water purification system (Bedford, MA, USA).

2.3 Synthesis of the adsorbent phase and preparation of the SPE cartridge

The biopolymer based SPE phases were synthesized in one step by reticulation of the biopolymer precursors (starch, β- and γ-cyclodextrin) using different cross-linking agents (Fig. 2). The epichlorohydrin starch biopolymer (ESB), epichlorohydrin β-cyclodextrin biopolymer (EBB) and γ-cyclodextrin biopolymer (EGB) materials were prepared in one step by cross-linking starch (15.0 g), β-cyclodextrin (12.5 g) and γ-cyclodextrin (8.0 g) with 30.0, 25.0 and 20.0 mL of epichlorohydrin, respectively. The 1,6-hexamethylene diisocyanate starch biopolymer (HSB) was synthesized by dissolving 15.0 g aliquot of starch in 40 mL of DMF in a three-neck round bottom flask. Thereafter 5.9 mL of 1,6-hexamethylene diisocyanate (HDI) was gradually added while stirring at 70 °C using a magnetic stirrer. The viscosity of the solution increased so strongly and rapidly that it could not be stirred after about 1 h. At that point, acetone was added to stop the reaction, and the precipitated biopolymer product was filtered off, washed copiously with distilled water, and further purified in a Soxhlet extractor for 12 hours using acetone, and dried by lyophilization. The 1,6-hexamethylene diisocyanate β-cyclodextrin biopolymer (HBB) and 1,6-hexamethylene diisocyanate γ-cyclodextrin biopolymer (HGB) were prepared in a similar experiment by cross-linking β-cyclodextrin (12.5 g) and γ-cyclodextrin (8.0 g) with 5.2 and 3.5 mL of 1,6-hexamethylene diisocyanate, respectively; while the 4,4-methylene diphenyl diisocyanate starch biopolymer (MSB), 4,4-methylene diphenyl diisocyanate β-cyclodextrin biopolymer (MBB) and 4,4-methylene diphenyl diisocyanate γ-cyclodextrin biopolymer (MGB) were prepared in a similar procedure by cross-linking starch (15.0 g), β-cyclodextrin (12.5 g) and γ-cyclodextrin (8.0 g) with 17.6 g, 17.5 g and 5.28 g of 4,4-methylene diphenyl diisocyanate, respectively. The synthesized biopolymer products were characterized by FTIR and scanning electron microscopy. The FTIR spectra of the soluble starch, β-cyclodextrin, γ-cyclodextrin, and their crosslinked counterparts were recorded on a PerkinElmer (USA) Spectrum 1 FTIR spectrometer in the range 400–4000 cm⁻¹. The SEM images of the samples were taken with Hitachi S4800 scanning electron microscope.

Polyethylene frit was first set at the bottom of polypropylene SPE column, after which a 250 mg aliquot of the biopolymer adsorbent phase, was packed into the 6 mL capacity polypropylene SPE column. Another polyethylene frit (which served as guard frit) was set to hold the adsorbent in place, and thereafter, a subtle hand pressure was applied using a cylindrical glass rod, to make the packing compact. The prepared cartridge was thereafter considered to be ready for use in SPE studies. As a precaution, small particle sizes of the biopolymer phase (60 mesh size) were adopted for this study to reduce the level of void that may occur as a result of using hand pressure for packing the SPE cartridges.

2.4 Solid phase extraction standard procedure

In order to reduce the interferences of organic and inorganic contaminants, the entire SPE assembly was first successively washed with 50 mL each of dichloromethane, methanol, acetonitrile and 100 mL of ultrapure water before the first use. With the aid of adaptors, the outlet tip of the cartridge was connected to Solid-Phase Extraction Vacuum Manifold (Agilent, USA), and the inlet end of cartridge was connected to PTFE suction tube with the other end inserted into the sample solution. The cartridge was first conditioned successively with 10 mL aliquots of methanol (to remove air and leach impurity), and ultrapure water (to equilibrate the solid phase), prior to sample loading. Aliquots of water samples were loaded at a known flow rate, immediately after conditioning. After loading, the cartridge was dried in vacuum (using negative pressure) at room temperature for 25 min to remove residual water. The analytes retained on the cartridge were eluted by aliquots of elution solvent at a known flow rate. The eluents were collected into a test tube and evaporated to dryness using a rotary evaporator at 30 °C, and re-dissolved with 1 mL acetonitrile. The reconstituted extract was thereafter taken for high performance liquid chromatography-ultra-violet/fluorescence detection (HPLC-UV/FLD) analysis.

2.5 HPLC-UV/FLD instrumental analysis

The PAHs analysis was carried out by a PerkinElmer series 200 HPLC system (PE, Norwalk, CT, USA). The system was equipped with a Series-200 analytical micro pump, auto sampler, as well as fluorescence and UV detectors in series connection. PAHs separation was done on a Brownlee (PerkinElmer, USA) analytical PAH reverse phase column (150 × 3.2 mm, 5 μm, 110 Å), applying acetonitrile and water as mobile phase at the flow rate of 0.6 mL min⁻¹. The PAHs were separated with the following gradient program: maintaining 85% A for 8 min; followed by a linear gradient from 85% A at 8 min to 100% A at 30 min, and returning linearly to 85% A in 5 minutes, and thereafter maintaining 85% A for 5 min. The column temperature was 35 °C, and the chromatographic peaks for all the PAHs were detected at the wavelength of 254 nm for UV detector and respective excitation and emission wavelengths as shown in Table ESI.1 (ESI†) for fluorescence detector. The excitation and emission bandwidth applied for the fluorescence detector program was 10 nm. The injection volume was 10 μL. HPLC data acquisition and handling was managed with Totalchrom data acquisition program version 6.3.1 (PerkinElmer, USA).

2.6 Evaluation of the biopolymers as SPE sorbent phase

The first preliminary experiment was the adsorbent selection experiment which applied each of the synthesized biopolymer (ESB, EBB, EGB, HSB, HBB, HGB, MSB, MBB, and MGB) as SPE sorbent phase.

The second preliminary experiment was designed to re-calibrate the pump in flow rate units. This was done by evaluating the pump pressure that will generate a given flow rate, and thus establish the pressure-flow rate relationship for the SPE set-up. This was necessary as the vacuum pump was calibrated in pressure units. The preliminary study therefore applied known pressure values for a given time duration, and estimated the flow rate based on the volume of water that flowed through the SPE column within a given time duration. The experiment was done thrice, and the mean value was adopted as the flow rate at the given pressure.

Another preliminary experiment carried out was the parameter of drying time of the sorbent after sample loading. Drying times of 15, 20, 25, 30, and 35 minutes were studied. Dichloromethane (DCM) was adopted as elution solvent for drying experiment because DCM is immiscible with water, hence; any water carryover can be visually noticed. The drying time was established to be ≥20 minutes, which was applied for subsequent experiments.

2.7 Optimization of extraction conditions

The optimum extraction conditions were investigated by extracting PAHs with the packed SPE cartridges, from spiked water samples containing 2.0 μg L⁻¹ US EPA PAH mix, following the SPE standard procedure.

The effect of nature of elution solvent was assessed by eluting the loaded cartridges with dichloromethane, acetonitrile, methanol, acetone, and 50 : 50 mixture of dichloromethane/hexane, under the following conditions; sample volume: 500 mL, sample flow rate: 4.5 mL min⁻¹, elution volume: 15 mL, and elution flow rate: 1.0 mL min⁻¹.

The effect of sample volume was investigated with sample volumes of 500, 1000, and 1500 mL, under the following conditions; sample flow rate: 4.2 mL min⁻¹, elution solvent: dichloromethane, elution volume: 15 mL, and elution flow rate: 1.0 mL min⁻¹. The effect of sample flow rate was assessed by loading the spiked water samples at flow rates of 4.11, 4.55 and 5.47 mL min⁻¹, under the following conditions; sample volume: 500 mL, elution volume: 15 mL, elution solvent: dichloromethane and elution flow rate: 1.0 mL min⁻¹. To study the effect of sample modifier, methanol was chosen as sample modifier since most PAHs of environmental relevance are soluble in methanol. The spiked water samples were first modified with 0%, 5%, 10%, 15%, and 20% of methanol (analytical grade), before they were loaded on the SPE cartridges. Other SPE parameters were fixed as follows; sample volume: 500 mL; sample flow rate: 5.5 mL min⁻¹; elution volume: 15 mL; and elution flow rate: 1.0 mL min⁻¹.

The effect of volume of elution solvent was assessed by eluting the loaded cartridges with 5.0, 7.0, 10.0, 15.0, and 20.0 mL of dichloromethane. Other parameters were fixed as follows; sample volume: 500 mL, sample flow rate: 5.5 mL min⁻¹, and elution flow rate: 0.9 mL min⁻¹. The effect of elution flow rate was assessed by eluting the loaded cartridges with dichloromethane at flow rates of 0.456, 0.891 and 1.452 mL per minute, under the following conditions; sample volume: 500 mL, sample flow rate: 5.5 mL min⁻¹, elution volume: 15.0 mL.

2.8 Evaluation of method performance and validation

The performance of the developed pre-concentration method was assessed by evaluating linearity/linear range, recovery, precision, and detection limit. These parameters were evaluated according to guidelines for single-laboratory validation of analytical methods for trace-level concentrations of organic chemicals,²⁴ while validation was based on parameters defined in standard protocols describing chromatographic methods.²⁵

The performance evaluation and validation of the parameters were investigated by enriching real and US EPA PAH mix spiked water samples at suitable spike levels under the optimized SPE conditions: sample volume: 500 mL; sample flow rate: 5.5 mL min⁻¹; elution volume: 10 mL; and elution flow rate: 0.9 mL min⁻¹, sample modifier: 15% ethanol. The method detection limit, precision, linearity and linear range, as well as recovery studies were assessed based on standard analytical protocols^24–26 (see details in ESI†). The water samples used in the assessments, were tap water collected from isotope preparatory lab of centre for environmental remediation, IGSNRR, Beijing, and polluted water samples collected from Shahe river (40°7′43.36′′N, 116°19′9.43′′E), Beijing, China.

3. Results and discussion

3.1 Computational modelling and predictions

The factors that influence the retention process are the characteristics of the sorbent phase and analyte, the solution chemistry and the prevailing temperature. Since the solution chemistry and the prevailing temperature for this study is assumed to be the same for all the sorbent phases, the analyte–sorbent phase interaction was therefore considered as the dominant factor. Also, since this study adopted the same PAHs analytes for all the sorbents (for comparative purposes), the role of analyte properties is assumed to be same for all the sorbent phases.

The concept of Frontal Molecular Orbital (FMO) theory assumes that retention (adsorption) is essentially electric charge transport process between electron donor and acceptor; hence, the smaller the disparities of the frontier orbital levels between the analytes (adsorbates) and the sorbent phase, the stronger the retention affinity.²⁷ Molecular orbital energies give information about reactivity/stability of specific regions of the molecule. Among the molecular orbitals, a fundamental role is played by the frontier orbitals (Highest Occupied Molecular Orbital (HOMO) and Lowest Unoccupied Molecular Orbital (LUMO)), which are responsible for the formation of many charge transfer complexes such as those involved in SPE retention affinity. Except dipole moment, all other descriptors discussed in this study were based on molecular orbital energies (see eqn (S1)–(S4) of the ESI†).

Table 1 shows the computed values of these DFT molecular interaction parameters for the adsorbents. It has been established that the adsorption of PAHs is dominated by weak hydrophobic interactions like π–π and dipole–dipole interactions, which involve temporary sharing and/or intra-molecular movement of electrons.^28–30 Hence, considering the fact that ionization is built on the concept of complete removal of electron(s) from a molecule, it is pertinent that ionization potential is not suitable for describing hydrophobic interactions.

Table 1 DFT generated quantum chemical descriptors of the cross-linked adsorbents

Adsorbents/adsorbates	EPI phases	HDI phases	MDI phases
EHOMO (eV)	−6.45	−6.63	−5.57
ELUMO (eV)	1.06	0.81	−0.18
Ionization potential, I	6.45	6.63	5.57
Electron affinity, A	−1.06	−0.81	0.18
Energy gap, ΔE (eV)	7.51	7.44	5.39
Dipole moment (Debye)	3.30	6.54	6.71

---

The relative high values of HOMO and low LUMO energy of the studied sorbent phases are indicators for good retention affinity. Consequent upon this, the interaction trend based on HOMO energy, LUMO energy and electron affinity is MDI sorbent > HDI sorbent > EPI sorbent. However, neither ionization potential nor electron affinity alone can adequately describe adsorption interaction, since adsorption involves electron donation, acceptance and sharing. Therefore, HOMO–LUMO energy gap (ΔE), which incorporates both E_HOMO and E_LUMO is considered more adequate for describing retention affinity. The DFT calculation showed that the sorbent phases exhibited relatively low ΔE values. Low values of ΔE translate to high interaction activity and consequently low stability of a molecule, while high values translate to low interaction. Based on the ΔE values, the retention trend is MDI sorbent > HDI sorbent > EPI sorbent.

Also the concept of FMO established that effective overlap of molecular orbitals is necessary for molecular interaction. Therefore, the size, location and orientation of frontier orbitals play a crucial role in molecular interaction with other chemical moieties. In this vein, the location and orientation of LUMO (Fig. 1) and HOMO (Fig. S1†) orbitals showed that the orbitals of MDI based biopolymers are better positioned for interaction than EPI and HDI based biopolymers. This conclusion can be adduced from the fact that the orbitals of the MDI based biopolymers are centrally located on the cross-linker unit of the crosslinked polymer. Though the EPI and HDI based biopolymers showed large lobed orbitals far larger than their MDI counterpart, the location of the these orbitals around the anchor point of the cross-linker unit to the starch chain, limits the full utilization of these large lobed orbitals in effective orbital overlap with the orbitals of analyte of interest. Thus, effective orbital overlap for EPI and HDI based biopolymer is consequently far less than those of the MDI biopolymer.

Since dipole–dipole interaction plays a significant role in the retention of PAHs on SPE sorbents,^28,29 the dipole moment of the biopolymers can also be used to assess the retention interaction of the PAHs on the biopolymers. The values of dipole moment (Table 1) revealed a similar trend with that of energy gap: MDI based biopolymers > HDI based biopolymers > EPI based biopolymers. Considering the fact that π–π interaction is an electron rich interaction,^29,30 it is expected that electron affinity will have a positive relationship with π–π interactions. On this note, the values of electron affinity showed a similar interaction trend with those of energy gap and dipole moment.

3.2 Characterization of the biopolymer adsorbents

Comparative analysis of the FTIR spectra of soluble starch, γ-cyclodextrin and β-cyclodextrin, and their corresponding biopolymers (Fig. 4) showed the presence of the characteristic peaks of unmodified starch as well as the expected IR peaks of the biopolymers.

The IR spectra of starch (Fig. 4d), γ-cyclodextrin (Fig. 4h) and β-cyclodextrin (Fig. 4l) showed the major peaks characteristic of starch, γ-cyclodextrin and β-cyclodextrin. Among these peaks was the broad peak observed at 3434, 3392, and 3377 cm⁻¹ in starch, γ-cyclodextrin and β-cyclodextrin IR spectra, respectively, which represents O–H stretching vibrations of polymeric compounds especially polysaccharides, the peaks at 2930, 2928 and 2928 in starch, γ-cyclodextrin and β-cyclodextrin spectra, respectively, which represent C–H stretching vibration, as well as the peaks at 1468, 1457, 1458 in starch, γ-cyclodextrin and β-cyclodextrin spectra, respectively, which represent C–H bending vibration. The peaks at 1650, 1647, 1646 in starch, γ-cyclodextrin and β-cyclodextrin spectra, respectively, represent δ(O–H) bending of tightly bound water present. The peak at 1165 cm⁻¹ represents C–O bond stretching of the C–O–H group (and sometimes C–C contribution). The peaks at 1083, 1081 and 1081 cm⁻¹ in starch, γ-cyclodextrin and β-cyclodextrin spectra, respectively, represent the characteristic glucose bands that emanates from C–O bond stretching of the C–O–C group in the anhydroglucose ring. The peaks at 925, 939, and 942 cm⁻¹ in starch, γ-cyclodextrin and β-cyclodextrin spectra, respectively, indicate the skeletal mode attributable to the α-1,4 linkage of glucose molecules in starch and cyclodextrins. The peaks at 765, 756, and 757 in starch, γ-cyclodextrin and β-cyclodextrin spectra are indicative of C–O–C symmetrical stretching while the ones at 856, 852, and 860 cm⁻¹ respectively, represent C–H deformation.

The FTIR spectra of the MDI crosslinked biopolymers: MSB (Fig. 4a), MGB (Fig. 4e), and MBB (Fig. 4i) showed sharp peak at 1510, 1511, and 1507 cm⁻¹ signifying the N–H deformation of secondary amides (amide-II band), and 1649, 1643 and 1644 cm⁻¹ signifying C–O stretching vibration in secondary amides (amide-I band) in MSB, MGB and MBB spectra, respectively. These peaks are in consonance with the formation of amide groups (–CONH), as shown in the reaction scheme for the synthesis of the polymer (Fig. 1 and 2). The peak at 1595 cm⁻¹ signifying ring stretching vibration of benzene ring in aromatic compound, aromatic C–H (3027, 3029 and 3028 cm⁻¹ for MSB, MGB and MBB respectively), and C–C (1406, 1410 and 1409 cm⁻¹ for MSB, MGB and MBB, respectively) confirmed successful cross-linking with MDI molecule and successive incorporation of aromatic group into the chemical structural matrix of these biopolymers. The FTIR spectra of the HDI crosslinked biopolymers: HSB (Fig. 4b), HGB (Fig. 3f), and HBB (Fig. 4j) showed peaks that indicated similar functionalities with MDI crosslinked biopolymer, except the peaks attributable to the aromatic functionalities. Considering the structure of the crosslinked biopolymers, the spectra of EPI crosslinked biopolymers: ESB (Fig. 4c), EGB (Fig. 4g), EBB (Fig. 4k) expectedly displayed no new peaks when compared with their respective precursors. However, the comparative reduction in the intensity of O–H peaks in the biopolymers confirms reaction of the epichlorohydrin with the O–H group of starch and cyclodextrins. Also, the comparative increment in the intensity of C–H peaks is an indication of the additional –CH₂–, which is an evidence to the incorporation of the epichlorohydrin structure into the chemical structural matrix of the biopolymer. Interestingly, the retention of major characteristic peaks of starch and cyclodextrin (O–H, C–H and C–O stretching vibration peaks, peaks for α-1,4 linkage of glucose molecules, as well as characteristic broad peaks of anomeric C–H ring deformations) in their respective biopolymers, showed that both the starch and cyclodextrin structural backbones were retained in the cross-linked biopolymers.

Fig. 3 Lowest unoccupied molecular orbital (LUMO) of (a) EPI biopolymer (b) HDI biopolymer and (c) MDI biopolymer.

Fig. 4 FTIR spectra of (a) MSB (b) HSB (c) ESB (d) soluble starch (e) MGB (f) HGB (g) EGB (h) γ-cyclodextrin (i) MBB (j) HBB (k) EBB and (l) β-cyclodextrin.

The scanning electron micrograph of starch, γ-cyclodextrin, β-cyclodextrin and their respective crosslinked biopolymers are shown in Fig. 5a–f in a typical transversal section. The micrograph of starch (Fig. 5a) and MSB biopolymer (Fig. 5b) reveals that the surface of the soluble starch was smooth with no noticeable pores, while that of the MSB biopolymer is rough, uneven and consists of a number of void spaces of different sizes. The micrograph of γ-cyclodextrin (Fig. 5c) showed γ-cyclodextrin surface as being composed of big pebbles and boulders as against MGB biopolymer (Fig. 5d), which displays comparatively smaller particles and numerous voids on its surface. The comparison of surface morphology of β-cyclodextrin (Fig. 5e) and MBB (Fig. 5f) followed a similar trend with that of the γ-cyclodextrin. Hence, it can be deduced from the SEM analysis that the crosslinking of the biopolymer precursors enhanced the surface area and comparative functionality of the biopolymers.

Fig. 5 Scanning electron micrograph of (a) starch (b) 4,4-methylene diphenyl diisocyanate starch biopolymer (MSB) (c) γ-cyclodextrin (d) 4,4-methylene diphenyl diisocyanate γ-cyclodextrin biopolymer (MGB) (e) β-cyclodextrin and (f) 4,4-methylene diphenyl diisocyanate β-cyclodextrin biopolymer (MBB).

3.3 Screening of biopolymers for SPE phases

The preliminary SPE studies showed that epichlorohydrin (EPI) cross-linked biopolymers (ESB, EBB and EGB) and 1,6-hexamethylene diisocyanate (HDI) cross-linked biopolymers (HSB, HBB and HGB) showed very poor recoveries when they were applied as the SPE sorbents. However, the 4,4-methylene diphenyl diisocyanate (MDI) biopolymers (MSB, MBB and MGB) showed far better recoveries. This can be rationalized by the following two obvious reasons. The PAHs belong to the class of nonpolar organic compounds. According to the rule of like dissolves like, the more nonpolar a sorbent is, the more likely it will facilitate enrichment of PAHs. Apart from the hydrophobic interactions which constituted of dipole–dipole and van der Waal interactions, the MDI cross-linked biopolymers have additional π–π interaction by the virtue of the presence of aromatic rings in the molecular structure of these biopolymers. It is this additional interaction that conferred higher retention potentials to the MDI based biopolymers. The delocalized π-bond interaction between PAHs and MDI biopolymers is much stronger than the hydrophobic interaction between PAHs and HDI polymers, as well as PAHs and epichlorohydrin polymers. The second reason is the fact that previous studies have shown that MDI based biopolymers possessed higher surface area and pore characteristics than their HDI and EPI counterparts.^29,30 Considering the recovery values, the observed SPE performance of the studied biopolymers also seemed to be in accordance with the sorption coefficient of these adsorbents, as elucidated in the adsorption studies.²⁹ It was equally observed that this trend was more pronounced in MDI based adsorbents than the HDI and epichlorohydrin based biopolymers. The simple explanation for this observation still bordered on the surface and pore characteristics. With higher surface area, the elution solvent will have better access to the analytes, thus conferring the potential of higher elution (desorption) efficiency, hence all the analytes adsorbed are likely to be desorbed within the elution time frame. With poor surface characteristics, analytes of interest will penetrate into the available few pores over time, considering large volume of samples applied in SPE. However, due to the poor surface characteristics and the subsequent limitation of access of the elution solvent to the analytes of interest during elution process, the desorption efficiency will be low; thus, all the analytes retained are not likely to be fully eluted.

From the foregoing, all the MDI cross-linked starch, β- and γ-cyclodextrin biopolymers (MSB, MBB and MGB) showed greatest promise as SPE sorbent phase, as predicted in QSAR model studies. However, only the starch based MSB was selected for optimization. This is owing to the fact that starch is the cheapest amongst all the applied crosslinked biopolymer precursors considered in this study, which gave it the greatest comparative cost advantage.

3.4 Optimization of SPE parameters

3.4.1 Nature of elution solvent. The type of elution solvent is vital for the extraction efficiency. A complete desorption of analytes from the adsorbent significantly affects the sensitivity, and is closely related to the desorption solvent. In this vein, the solvent used for the PAH elution should be strong enough for the analytes to be desorbed from the strongly bounded adsorbent. So the choice of elution solvent was carefully taken into account. Fig. 6a shows the recoveries obtained from five different organic solvents/solvent mixtures: acetonitrile, methanol, acetone, dichloromethane and dichloromethane/n-hexane mixture (50 : 50 v/v) tested as elution solvents. The mean recovery of PAHs eluted by acetone was 72.58% with RSD of 17.23%, dichloromethane gave the mean recovery of 71.57%, with RSD of 15.07%, while acetone/DCM mixture gave mean recovery value of 64.46% with RSD of 20.42%. However, the application of methanol and acetonitrile as elution solvents gave much lower recoveries; 32.13% with RSD of 39.06% and 38.19% with RSD of 22.45%, for methanol and acetonitrile, respectively. Among the five solvents, acetone, dichloromethane and dichloromethane/n-hexane mixture provided comparable results for PAHs, about 30–40% higher than the recoveries obtained by using methanol or acetonitrile. Though acetone showed very good recovery, the acetone eluent seemed to have poor elution selectivity as indicated by multiple peaks in its chromatogram, and hence requires clean-up. In addition to the elution strength, DCM presents another advantage, which is the high vapor pressure necessary to achieve quick and effective solvent evaporation. Also, it was observed from recovery values of individual PAHs that DCM gave better recovery values for the high molecular weight PAHs than acetone. Considering the foregoing, DCM was selected as the best elution solvent.

Fig. 6 Bar charts showing the SPE optimization parameters for effect of (a) elution solvent (b) sample volume (c) sample flow rate (d) sample modifier.

3.4.2 Sample volume. The effect of sample volumes on the enrichment of PAHs is shown in Fig. 6b. For small molecular weight PAHs, it was found that the recoveries were steady in the range of 500–1000 mL but decreased when the sample volume exceeded 1000 mL. This may be attributed to the fact that in presence of larger amounts of solvent, it will result in less interaction between PAHs and the SPE sorbent. This effect was equally observed for high molecular weight PAHs, but was comparatively not as pronounced as those of small molecular weight. This can equally be attributed to the difference in the level of hydrophobicity between the PAHs. Overall, the effect of sample volume within the studied range (500–1000 mL) was minimal. On the basis of sample loading time and convenience of optimization, 500 mL was selected as the optimal sample volume.

3.4.3 Sample flow rate. Generally, sample loading time can be saved at a high flow rate while the possible analytes loss happens owing to an incomplete adsorption of PAHs by the sorbents; complete adsorption can be achieved at a low flow rate but it is time consuming. Therefore, a suitable flow rate for loading sample should be investigated to achieve high recovery and short loading time. Considering the sample volume applied for the study, relatively high sample flow rate is required to achieve enrichment in reasonable time, hence the choice of medium flow rate range of 4.1–5.5 mL min⁻¹. As shown in Fig. 6c, the PAH recoveries within the investigated flow rates were acceptable.

Considering that all the 16 PAHs could be efficiently adsorbed within this time span, the sample loading flow rate of 5.5 mL min⁻¹ was adopted as the optimum sample flow rate.

3.4.4 Sample modifier. Due to the low solubility of PAHs in water, it is necessary to add organic solvent as an organic sample modifier to prevent PAHs from adsorbing on the walls of the glass sample containers. Methanol was chosen for this study because it is relatively a good solvent for the 16 PAHs under study, relatively cheap, and less toxic. The content of the organic modifier (concentration in 500 mL water sample) investigated for this study was 0–20%. It can be observed from Fig. 6d, that incorporation of organic sample modifier significantly improved the recoveries of the PAHs, when compared to the recovery values of previous optimization experiments without organic solvent modification. For small and medium molecular weight PAHs, Fig. 4d showed that recovery of PAHs increased till it reached a maximum value, after which progressive increment in organic content led to decrease in recovery of PAHs. However, there was continuous increment in recovery values within the entire range (0–20%) for high molecular weight PAHs.

The observed general increment in recovery of PAHs with increment in the percentage of sample modifier can be explained on the basis of reduction or elimination of PAHs that had hitherto adsorbed on the walls of sample containers. However, with continuous increment in the percentage of organic solvent, the solubility of PAHs (especially the low molecular weight PAHs) in the sample will be significantly enhanced to the extent that it will cause reduction in interaction (and subsequently retention) of PAH molecules on the SPE sorbent, and hence cause reduction in recovery as observed for low molecular weight PAHs. Considering higher hydrophobicity of the high molecular weight PAHs, the increment in solubility as a result of increment in percentage of sample modifier might not be significant enough to induce reduction in PAHs interaction with the solid phase. Based on these facts, 15% (75 mL of methanol in 500 mL sample) was adopted as the optimum value of sample modifier to be added to enhance the recovery of PAHs.

3.4.5 Elution flow rate. Elution flow rate is an important parameter for enrichment efficiency and analyte recovery. The effect of the sample flow rate was investigated over the range from 0.45 to 1.45 mL min⁻¹. For a given elution volume, the analyte recoveries have been shown to have a positive correlation with the contact time of the solvent with the loaded SPE sorbent, up to a maximum (often referred as the optimum) value, after which further increment in the contact time does not translate to increment in analyte recoveries.³¹ However, for a given solvent volume, higher elution flow rate translates to shorter contact time, and vice versa. Previous studies have shown that for a given solvent volume, higher molecular weight PAHs require high contact time for better recoveries.³² For this study, a compromise was struck between flow rate and solvent volume. Since the elution volume in SPE is always relatively small, the use of low flow rates does not necessarily translate to much time, and hence the application of low flow rates in this study is justified. As shown in Fig. 7a, elution flow rate of 0.891 mL min⁻¹ (approximately 1.0 mL min⁻¹) was an optimal value for these 16 PAHs.

Fig. 7 Bar charts showing the SPE optimization parameters for effect of (a) elution flow rate and (b) elution volume.

3.4.6 Elution volume. It has been established that for a given elution flow rate, higher solvent volume translates to higher contact time and subsequently higher recovery, up to the optimum value after which further increment in volume does not result to a corresponding increment in analyte recovery.³¹ For small to medium molecular size PAHs, the recoveries increased with progressive increment of the eluent volume from 5 mL to 10 mL, after which increment in volume did not result to a reasonable increment in the recoveries, whereas the recoveries of higher molecular PAHs continued to increase with increment in elution volume up to 20 mL (Fig. 7b). This behaviour can be explained on the basis of PAHs hydrophobicity in aqueous media. The higher molecular weight PAHs are more hydrophobic; hence, they have stronger affinity towards the SPE phase, and therefore need more solvent to elute them. Looking at this scenario, the choice of elution volume was guided by the PAHs of priority concern, i.e. whether high or low molecular weight PAHs. However, since acceptable recoveries were recorded for the higher molecular weight PAHs when 10 mL elution volume was applied, 10 mL (5 mL each time and washed twice) dichloromethane was selected for this study.

3.5 Validation of SPE method

3.5.1 Method detection limit (MDL). The relevance of MDL in SPE method validation cannot be over emphasized as it is one of the major selection guidelines for selecting a method, in consideration with the expected concentration range of the analyte of interest. In the present study, method detection limit was determined using the standard deviation (n = 6) of the concentration at the selected low levels of enrichment. As shown in Tables 2 and 3, detection limits for the developed method were 22.9–155.3 ng L⁻¹ and 0.51–24.15 ng L⁻¹ for ultra-violet and fluorescence detectors, respectively.

Table 2 Method validation data for PAHs using UV detector (n = 6)

PAHs	Regression equation (y = a + bx)	Mean recovery	R^2	MDL (ng L^−1)	Precision (RSD)	Fortification μg L^−1
Naphthalene	y = 25 410x + 2034	67.69	0.9980	155.3	25.69	10.0
Acenaphthylene	y = 20 657x − 8684	83.10	0.9988	123.6	21.19	10.0
Acenaphthene	y = 7926x − 4513	101.62	0.9976	157.3	26.02	10.0
Fluorene	y = 154 517x − 39 955	84.91	0.9991	94.3	13.00	10.0
Phenanthrene	y = 454 750x − 27 118	90.26	0.9991	63.7	20.97	1.0
Anthracene	y = 49 796x + 281 634	80.37	0.9980	23.6	8.71	1.0
Fluoranthene	y = 126 437x − 39 230	86.25	0.9968	50.7	17.47	1.0
Pyrene	y = 109 378x − 53 288	71.21	0.9945	47.3	19.74	1.0
Benzo[a]anthracene	y = 217 338x − 75 594	84.61	0.9955	33.1	11.62	1.0
Chrysene	y = 404 855x − 30 163	93.67	0.9998	66.2	21.01	1.0
Benzo[b]fluoranthene	y = 244 919x − 110 252	106.49	0.9926	70.8	19.76	1.0
Benzo[k]fluoranthene	y = 192 612x + 34 810	108.98	0.9994	61.8	16.86	1.0
Benzo[a]pyrene	y = 233 160x − 116 549	99.43	0.9946	35.7	10.68	1.0
Dibenzo[b]anthracene	y = 55 464x − 19 809	97.24	0.9992	32.6	9.95	1.0
Benzo[g]perylene	y = 94 916x − 40 744	100.31	0.9982	26.4	7.83	1.0
Indeno[c]perylene	y = 268 245x − 119 733	97.89	0.9962	22.9	6.94	1.0

---

Table 3 Method validation data for PAHs using fluorescence detector (n = 6)

PAHs	Regression equation (y = a + bx)	R^2	Mean recovery	MDL (ng L^−1)	Precision (RSD)	Fortification (μg L^−1)
Naphthalene	y = 3417x + 136	0.9969	73.13	24.15	16.60	0.5
Acenaphthylene	ND	ND	ND	ND	ND	0.5
Acenaphthene	y = 4922x − 1331	0.9934	81.49	16.74	12.21	0.5
Fluorene	y = 5519x − 5933	0.9999	71.68	6.47	5.36	0.5
Phenanthrene	y = 7559x − 17 438	0.9996	81.36	3.50	2.55	0.5
Anthracene	y = 9096x + 151 641	0.9989	88.09	4.24	2.86	0.5
Fluoranthene	y = 86 434x − 23 229	0.9969	118.44	6.20	3.11	0.5
Pyrene	y = 9373x − 23 248	0.9995	94.29	1.94	1.22	0.5
Benzo[a]anthracene	y = 17 532x − 35 794	0.9979	72.65	0.51	0.41	0.5
Chrysene	y = 314 853x − 30 163	0.9998	86.38	0.88	0.60	0.5
Benzo[b]fluoranthene	y = 144 013x − 71 254	0.9997	83.91	1.04	0.74	0.5
Benzo[k]fluoranthene	y = 90 212x + 13 410	0.9984	92.10	0.97	0.63	0.5
Benzo[a]pyrene	y = 17 165x − 60 149	0.9986	84.73	0.63	0.44	0.5
Dibenzo[b]anthracene	y = 25 467x − 1180	0.9997	89.49	0.92	0.61	0.5
Benzo[g]perylene	y = 44 532x − 51 075	0.9968	102.49	2.41	1.39	0.5
Indeno[c]perylene	y = 128 542x − 39 735	0.9969	126.00	7.89	3.72	0.5

---

3.5.2 Precision. Precision is a measure of the random error associated with a series of repeated measurements of the same parameter within a sample. In other words, precision describes the closeness with which multiple analyses of a given sample agree with each other. In the validation study, precision was measured using relative standard deviations (RSD) at two concentration levels (1.0 and 10.0 μg L⁻¹) for UV detector, and one concentration level (0.5 μg L⁻¹) for fluorescence detector. The RSD values ranged from 6.94% for indeno[c]perylene to 25.69% for naphthalene when UV detector was used (Table 2), and from 0.41% for BaA to 16.60% for naphthalene, when fluorescence detector was used (Table 3). All RSDs were below 30%, the value considered acceptable for most chromatographic methods.³³

3.5.3 Linearity/linear range. The performance evaluation of an analytical method using linearity and linear range evaluates whether the calibration curve of the method is linear within the concentration range upon which the analyte of interest is likely to be present in the media under consideration. Linearity is more appropriately expressed by adjusted root mean square (R²) values. However, it can still be expressed as an equation of a straight line describing the calibration curve, whereas linear range is expressed as numerical concentration range being considered.

A five-point calibration curve constructed for each PAH analyte over the concentration range of interest (0.5–50 μg L⁻¹), showed that the regression coefficients (R²) of the calibration plots were higher than 0.99 in all cases (Tables 2 and 3). This indicated high level of linearity within the studied concentration range.

3.5.4 Recovery studies. In SPE method development, recovery studies are used to measure the accuracy of the measurements, which reflects the closeness of a measured value to the true value.

In this study, recoveries were measured at two fortification levels, 1.0 and 10.0 μg L⁻¹ for the UV detectors. At the first level (10.0 μg L⁻¹), recoveries varied from 67.69% for naphthalene to 101.62% for acenaphthene. At the second level (1.0 μg L⁻¹) recoveries were from 71.21% for pyrene to 108.98% for BkF. However, recoveries were measured at 0.5 μg L⁻¹ fortification level, for fluorescence detector. At this fortification level, the recoveries were from 71.68 to 126.00% (Table 3). In all the studies, except for the recovery value of naphthalenes, the efficiency of the SPE method was satisfactory, with recoveries in the range of 70–130%, which is considered acceptable for environmental samples.³³

3.6 Application of the developed SPE method to real water samples

The levels of PAHs in the tap and Shahe river water samples (unspiked samples), as analyzed using the MSB SPE, were as shown in Tables 4 and 5. All the investigated PAHs were detected in Shahe river water samples, while nine of the US EPA PAHs were detected in the tap water samples. The performance of the SPE preconcentration method was assessed by evaluating the recovery of PAHs in spiked real water samples (0.5 μg L⁻¹ spike level). Application of the developed method in the analysis of the spiked environmental water samples exhibited good recoveries. The mean values of recoveries using UV detector were 77.3 ± 11.4% and 83.9 ± 13.1%, for tap and Shahe water samples, respectively, while the recovery values for fluorescence detector were 81.4 ± 13.7% and 86.1 ± 14.6%, for tap and Shahe water samples, respectively. The validity of the recorded levels of PAHs was evaluated by comparing the recoveries of MSB SPE sorbent and that of standard C₁₈ silica (ENVI™18) SPE sorbent. The recovery values for both MSB and C₁₈ sorbents indicated that the analytical performance of the developed SPE method is comparable to that of commercially available C₁₈ SPE phase (Tables 4 and 5).

Table 4 Analysis of unspiked and spiked environmental water samples for comparative evaluation of MSB and C₁₈ silica (ENVI™) as SPE sorbent phases using UV detector

PAHs	Conc. in tap water (μg L^−1)	Recovery		Conc. in Shahe water (μg L^−1)	Recovery
		MSB	ENVI™ (C18)		MSB	ENVI™ (C18)
a Not detected.
Naphthalene	ND^a	74.47	70.04	0.1035	67.39	64.25
Acenaphthylene	ND	71.56	82.72	ND	70.70	70.01
Acenaphthene	ND	59.12	65.03	0.1008	68.26	77.45
Fluorene	ND	75.09	95.01	0.2300	70.67	89.46
Phenanthrene	ND	76.50	105.47	0.1441	75.53	97.73
Anthracene	ND	72.03	97.97	ND	75.23	94.94
Fluoranthene	0.2905	87.65	100.05	0.2806	86.06	98.13
Pyrene	ND	97.96	111.32	0.3260	89.75	103.32
Benzo[a]anthracene	0.2682	72.68	87.76	0.1045	80.46	86.19
Chrysene	0.1071	73.54	89.18	0.0725	80.79	93.37
Benzo[b]fluoranthene	0.1471	75.96	83.06	0.2351	106.48	81.20
Benzo[k]fluoranthene	ND	88.16	78.39	0.0604	104.97	85.97
Benzo[a]pyrene	ND	77.09	87.04	0.1932	98.75	86.70
Dibenzo[b]anthracene	ND^a	84.54	85.61	0.1247	89.72	84.61
Benzo[g]perylene	ND	104.49	80.92	0.1710	107.91	89.15
Indeno[c]perylene	ND	112.21	97.57	0.0319	105.36	100.74
Mean (x ± s)		81.4 ± 13.7	88.6 ± 12.3		86.3 ± 14.7	87.7 ± 10.7

---

Table 5 Analysis of unspiked and spiked environmental water samples for comparative evaluation of MSB and C₁₈ silica as SPE sorbent phases using fluorescence detector

PAHs	Conc. in tap water (μg L^−1)	Recovery		Conc. in Shahe water (μg L^−1)	Recovery
		MSB	ENVI™ (C18)		MSB	ENVI™ (C18)
a Not detected.
Naphthalene	ND^a	73.29	69.35	0.0235	63.85	60.23
Acenaphthylene	ND	69.38	77.90	0.0012	74.50	66.71
Acenaphthene	ND	52.64	64.26	0.0340	74.31	75.40
Fluorene	ND	73.74	90.78	0.1500	67.74	85.37
Phenanthrene	ND	73.39	107.04	0.3410	71.68	89.18
Anthracene	0.0117	73.39	97.07	0.0215	77.64	91.68
Fluoranthene	0.1404	86.87	92.74	0.3580	82.65	88.09
Pyrene	0.0852	102.16	113.90	0.2680	84.65	105.25
Benzo[a]anthracene	0.1366	68.11	89.72	0.0645	78.93	86.03
Chrysene	0.0890	70.84	80.75	0.0925	85.84	92.04
Benzo[b]fluoranthene	0.1471	75.96	78.11	0.1735	109.31	82.20
Benzo[k]fluoranthene	0.0448	86.31	80.62	0.0840	106.92	87.52
Benzo[a]pyrene	0.0576	77.56	82.24	0.1310	95.80	87.85
Dibenzo[b]anthracene	ND	81.48	82.11	0.1090	84.86	84.61
Benzo[g]perylene	ND	94.49	77.49	0.2010	86.60	89.29
Indeno[c]perylene	0.0642	77.81	92.46	0.0102	97.58	110.40
Mean (x ± s)		77.3 ± 11.4	86.0 ± 12.9		83.9 ± 13.1	86.4 ± 12.2

---

Also, the performance of the SPE preconcentration method using the developed material was equally compared with the performance of other methods reported in literature. In this vein, Table 6 showed the recoveries of PAHs as performance indicator for similar studies reported in literature. The merits of the present method are obvious; simple synthetic method for the adsorbents, high recoveries, low LODs, as well as fast and simple extraction procedure.

Table 6 Comparative evaluation of the analytical performance of the developed method and the values reported in literature^a

Preconcentration method	SPE* tap water	SPE* river water	SPE^# tap water	SPE^# river water	SPE subsoil water^34	SPE seawater^35	SPE seawater^8	SPE seawater^7	SPE seawater^10
a ND = not detected; NR = not reported; * present study (with fluorescence detector); ^# present study (with UV detector).
Naphthalene	73.29	63.85	74.47	67.39	35	ND	NR	NR	101
Acenaphthylene	69.38	74.50	71.56	70.70	46	ND	NR	NR	NR
Acenaphthene	52.64	74.31	59.12	68.26	105	ND	NR	NR	99
Fluorene	73.74	67.74	75.09	70.67	97	ND	NR	NR	96
Phenanthrene	73.39	71.68	76.50	75.53	102	52	81	92	97
Anthracene	73.39	77.64	72.03	75.23	86	10	81	88	87
Fluoranthene	86.87	82.65	87.65	86.06	113	34	97	NR	NR
Pyrene	102.16	84.65	97.96	89.75	112	32	97	95	NR
Benzo[a]anthracene	68.11	78.93	72.68	80.46	68	23	97	NR	NR
Chrysene	70.84	85.84	73.54	80.79	67	18	96	NR	NR
Benzo[b]fluoranthene	75.96	109.31	75.96	106.48	86	25	NR	NR	NR
Benzo[k]fluoranthene	86.31	106.92	88.16	104.97	73	18	NR	NR	NR
Benzo[a]pyrene	77.56	95.80	77.09	98.75	61	10	97	86	NR
Dibenzo[b]anthracene	81.48	84.86	84.54	89.72	63	22	13	NR	NR
Benzo[g]perylene	94.49	86.60	104.49	107.91	58	20	100	NR	NR
Indeno[c]perylene	77.81	97.58	112.21	105.36	67	16	99	NR	NR
Mean (x ± s)	77.3 ± 11.4	83.9 ± 13.1	81.4 ± 13.7	86.1 ± 14.6	77.4 ± 23.6	23.3 ± 11.6	85.8 ± 26.5	90.3 ± 4.0	96.0 ± 5.4

---

4. Conclusion

This study has successfully introduced QSAR as a veritable tool in the design and development of cross-linked biopolymer adsorbents. The observed affinity, and consequently the SPE performance trend were highly in agreement with modelling prediction. The import of this procedure is that the feasibility of using a given cross-linked biopolymer sorbent for a target analyte can be evaluated through computational simulation without necessarily going through synthesis and characterization of the adsorbents, as well as experimental SPE studies. This approach saves cost, time and labour.

The SPE pre-concentration method using MSB sorbent phase for enrichment of PAHs showed good analytical method performance in terms of detection limits, precision, linearity and linear range, as well as recoveries. The MSB sorbent performed creditably well when it was applied for enrichment of PAHs in real samples (tap and river water). The analytical performance of the SPE/HPLC method using the developed sorbent phase is very much comparable with other standard methods adopted by environmental regulatory agencies. In most cases, the observed analytical performance was better than the performance of most of the SPE sorbents reported in literature. Interestingly, the values of method detection limit achieved in this study for both UV and fluorescence detectors are lower than the maximum allowable levels for PAHs set by most environmental regulatory bodies, which make them suitable for monitoring studies for wide concentration range of PAHs pollution.

The availability of starch at relatively low cost, use of cheap and easily accessible cross-linking agents, and very simple synthetic procedure implies that the developed adsorbents can be easily prepared in the laboratory at a reasonable cost, and therefore can be applied for routine laboratory analysis. The biodegradable nature of cross-linked starch and cyclodextrin polymers confers a great advantage to the sorbent phases developed in this study, over polystyrene divinyl benzene (PS-DVB) based adsorbents which are non-biodegradable, but have proved to be the choice adsorbents for most aromatic compounds including PAHs.

Acknowledgements

The authors of this work acknowledge the support of the World Academy of Sciences (TWAS) and the Chinese Academy of Sciences (CAS) for providing fellowship to Peter ChukwunonsoOkoli (FR number: 3240240233) at the Institute of Geographic Sciences and Natural Resources Research (IGSNRR), CAS, Beijing, China where this research was carried out. We are equally grateful to Mr Jianli Wang, Guangxu Zhu, and ChungYu Wang of Centre for Environmental Remediation Unit, and the authorities of Central Research Lab of IGSNRR for their assistance in the laboratory instrumentation.

References

1. C. W. Huck and G. K. Bonn, Recent developments in polymer-based sorbents for solid-phase extraction, J. Chromatogr. A, 2000, 885, 51–72.
2. L. Chen, H. Wang, Q. Zeng, Y. Xu, L. Sun and H. Xu, et al., On-line coupling of solid-phase extraction to liquid chromatography—a review, J. Chromatogr. Sci., 2009, 47(8), 614–623.
3. B. Buszewski and M. Szultka, Past, present, and future of solid phase extraction: a review, Crit. Rev. Anal. Chem., 2012, 42(3), 198–213.
4. A. Andrade-Eiroa, M. Canle, V. Leroy-Cancellieri and V. Cerdà, Solid-phase extraction of organic compounds: a critical review (part I), TrAC, Trends Anal. Chem., 2016, 80, 641–654.
5. D. L. Diggs, A. C. Huderson, K. L. Harris, J. N. Myers, L. D. Banks and P. V. Rekhadevi, et al., Polycyclic aromatic hydrocarbons and digestive tract cancers: a perspective, J. Environ. Sci. Health, Part C: Environ. Carcinog. Ecotoxicol. Rev., 2011, 29, 324–357.
6. R. M. Marcé and F. Borrull, Solid-phase extraction of polycyclic aromatic compounds, J. Chromatogr. A, 2000, 885, 273–290.
7. J. Jin, Z. Zhang, Y. Li, P. Qi, X. Lu and J. Wang, et al., Enrichment of polycyclic aromatic hydrocarbons in seawater with magnesium oxide microspheres as a solid-phase extraction sorbent, Anal. Chim. Acta, 2010, 678, 183–188.
8. V. Khalili-Fard, K. Ghanemi, Y. Nikpour and M. Fallah-Mehrjardi, Application of sulfur microparticles for solid-phase extraction of polycyclic aromatic hydrocarbons from sea water and wastewater samples, Anal. Chim. Acta, 2012, 714, 89–97.
9. M. M. Abolghasemi, B. Karimi and V. Yousefi, Periodic mesoporous organosilica with ionic liquid framework as a novel fiber coating for headspace solid-phase microextraction of polycyclic aromatic hydrocarbons, Anal. Chim. Acta, 2013, 804, 280–286.
10. A. Mehdinia, N. Khodaee and A. Jabbari, Fabrication of graphene/Fe₃O₄@polythiophene nanocomposite and its application in the magnetic solid-phase extraction of polycyclic aromatic hydrocarbons from environmental water samples, Anal. Chim. Acta, 2015, 868, 1–9.
11. H. B. Zheng, J. Ding, S. J. Zheng, G. T. Zhu, B. F. Yuan and Y. Q. Feng, Facile synthesis of magnetic carbon nitride nanosheets and its application in magnetic solid phase extraction for polycyclic aromatic hydrocarbons in edible oil samples, Talanta, 2016, 148, 46–53.
12. Y. Cai, Z. H. Yan, N. Y. Wang, Q. Y. Cai and S. Z. Yao, Preparation of naphthyl functionalized magnetic nanoparticles for extraction of polycyclic aromatic hydrocarbons from river waters, RSC Adv., 2015, 5(69), 56189–56197.
13. M. Khajeh, M. Sharifirad, M. Bohlooli and M. Ghaffari-Moghaddam, Magnetic molecularly imprinted polymers–silver nanoparticle based micro-solid phase extraction for the determination of polycyclic aromatic hydrocarbons in water samples, RSC Adv., 2016, 6(60), 54702–54708.
14. E. V. Datskevich, Removal of Phenols from Water with Cross-Linked Starch Composites, Russ. J. Appl. Chem., 2009, 82(12), 2201–2209.
15. L. Guo, G. Li, J. Liu, P. Yin and Q. Li, Adsorption of Aniline on Cross-Linked Starch Sulfate from Aqueous Solution, Ind. Eng. Chem. Res., 2009, 48, 10657–10663.
16. L. Dehabadi and L. D. Wilson, Polysaccharide-based materials and their adsorption properties in aqueous solution, Carbohydr. Polym., 2014, 113, 471–479.
17. C. P. Okoli, Q. J. Guo and G. O. Adewuyi, Application of quantum descriptors for predicting adsorption performance of starch and cyclodextrin adsorbents, Carbohydr. Polym., 2014, 101, 40–49.
18. M. Tabandeh, S. Ghassamipour, H. Aqababa, M. Tabatabaei and M. Hasheminejad, Computational design and synthesis of molecular imprinted polymers for selective extraction of allopurinol from human plasma, J. Chromatogr. B: Anal. Technol. Biomed. Life Sci., 2012, 898, 24–31.
19. I. Bakas, N. B. Oujji, E. Moczko, G. Istamboulie, S. Piletsky and E. Piletska, et al., Computational and experimental investigation of molecular imprinted polymers for selective extraction of dimethoate and its metabolite omethoate from olive oil, J. Chromatogr. A, 2013, 1274, 13–18.
20. F. Ahmadi, E. Yawari and M. Nikbakht, Computational design of an enantioselective molecular imprinted polymer for the solid phase extraction of S-warfarin from plasma, J. Chromatogr. A, 2014, 1338, 9–16.
21. J. T. Hoeft, M. Kittel, M. Polcik, S. Bao, R. L. Toomes and J. H. Kang, et al., Molecular adsorption bond lengths at metal oxide surfaces: failure of current theoretical methods, Phys. Rev. Lett., 2001, 87, 086101.
22. T. S. Rantala, V. Lantto and T. T. Rantala, A cluster approach for modelling of surface characteristics of stannic oxide, Phys. Scr., 1994, 54, 252.
23. Y. Shao, L. Fusti-Molnar, Y. Jung, J. Kussmann, C. Ochsenfeld and S. T. Brown, et al., SPARTAN'10, build 1.0.1, Wavefunction Inc, Irvine CA, 2011.
24. A. Fajgelj and A. Ambrus, Guidelines for single-laboratory validation of analytical methods for trace-level concentrations of organic chemicals, 2007.
25. M. Thompson, S. L. Ellison and R. Wood, Harmonized guidelines for single-laboratory validation of methods of analysis (IUPAC Technical Report), Pure Appl. Chem., 2002, 74, 835–855.
26. US EPA (United States Environmental Protection Agency), Appendix B to Part 136 – Definition and Procedure for the Determination of the Method Detection Limit – Revision 1.11, 1990, 40 CFR 136.
27. J. N. Wang, A. M. Li, S. Zhong, W. B. Yan and Q. X. Zhang, Studies on the adsorption of phenolic compounds on a ps-based resin modified by nitro functional groups, Chin. J. Polym. Sci., 2008, 26, 713–721.
28. S. Kayillo, G. R. Dennis and R. A. Shalliker, An assessment of the retention behaviour of polycyclic aromatic hydrocarbons on reversed phase stationary phases: selectivity and retention on C₁₈ and phenyl-type surfaces, J. Chromatogr. A, 2006, 1126(1), 283–297.
29. C. P. Okoli, G. O. Adewuyi, Q. Zhang, P. N. Diagboya and Q. Guo, Mechanism of dialkyl phthalates removal from aqueous solution using γ-cyclodextrin and starch based polyurethane polymer adsorbents, Carbohydr. Polym., 2014, 114, 440–449.
30. C. P. Okoli, G. O. Adewuyi, Q. Zhang, G. Zhu, C. Wang and Q. Guo, Aqueous scavenging of polycyclic aromatic hydrocarbons using epichlorohydrin, 1,6-hexamethylene diisocyanate and 4,4-methylene diphenyl diisocyanate modified starch: pollution remediation approach, Arabian J. Chem., 2015 DOI:10.1016/j.arabjc.2015.06.004.
31. S. K. Poole, T. A. Dean, J. W. Oudsema and C. F. Poole, Sample preparation for chromatographic separations: an overview, Anal. Chim. Acta, 1990, 236, 3.
32. A. Kouzayha, M. Al Iskandarani, S. Mokh, A. R. Rabaa, H. Budzinski and F. Jaber, Optimization of a solid-phase extraction method using centrifugation for the determination of 16 polycyclic aromatic hydrocarbons in water, J. Agric. Food Chem., 2011, 59, 7592–7600.
33. J. R. L. Bispo, S. Navickiene and H. S. Dórea, Method Validation for SPE Applied to Determination of PAH in Petroliferous Industry Effluent Water, Am. J. Anal. Chem., 2011, 2, 971–978.
34. E. Martinez, M. Gros, S. Lacorte and D. Barceló, Simplified procedures for the analysis of polycyclic aromatic hydrocarbons in water, sediments and mussels, J. Chromatogr. A, 2004, 1047, 181–188.
35. A. Filipkowska, L. Lubecki and G. Kowalewska, Polycyclic aromatic hydrocarbon analysis in different matrices of the marine environment, Anal. Chim. Acta, 2005, 547, 243–254.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra10932b

‡ Present address: Chemistry Department, Vaal University of Technology, Vanderbijlpark 1900, South of Africa.

---