Figure 1a,b,c shows the FT-IR spectra of β-CD, β-CD-HMDI, and β-CD-TDI, respectively. The peak at 2270 cm⁻¹ (corresponding to isocyanate group) is absent in both Figures 1b,c, indicating that the polymerization reaction is complete [39]. The strong bands at 2860 and 2934 cm⁻¹ in Figure 1b correspond to the methylene group in HMDI, while the bands at 3364 and 1719 cm⁻¹ represent the NH and C=O groups. The stretching of NHCO, which indicates the formation of carbamate group, i.e., characteristic group of polymerization, is observed at 1570 cm⁻¹. The bands at 1535 and 1603 cm⁻¹ in Figure 1c represent the aromatic group in TDI. The FT-IR spectra obtain are similar to obtained by Bhaskar et al. [39]. Therefore, we conclude that polymerization between β-CD and cross-linker are completed and β-CD polymers are formed.

In order to evaluate the influence of cross-linker on the adsorption and the rate of adsorption for parabens, tests were carried out using polymers with different molar ratios of β-CD:cross-linker and types of cross-linker, and at contact times of two and 24 hours. All β-CD polymers prepared showed good adsorption capacities towards parabens (Figure 2a,b). Figure 2a shows that increasing the amount of HMDI and different contact times did not affect the adsorption of parabens onto β-CD-HMDI very much. However, there is a marked difference in the adsorption of parabens onto β-CD-TDI, especially with a higher amount of TDI, and there appears to be a bigger difference between the adsorptions at contact times two and 24 hours (Figure 2b). The results show that the adsorption of parabens on β-CD polymers is dependent on the amount of TDI used in the preparation of the polymers, but not on the amount of HMDI. However, β-CD-HMDI(a) and β-CD-TDI(b) were chosen for further studies since they provide relatively higher adsorption capacities amongst each of their series. The two selected polymers are also economically more viable since they contain lower amounts of cross-linkers.

The water absorption property of all polymers were then examined (Figure 3). The results show that, in general, β-CD-HMDI polymers absorbed more water than the β-CD-TDI polymers. This is probably because TDI has poor contact with aqueous solution due the presence of an aromatic group in TDI, and that TDI, as a cross-linker, can form a more non-polar polymeric network compared to HMDI. As the polymer becomes more hydrophobic, the diffusion of water is less, and hence the contact between the polymer and solute is slower. If there is not enough swelling of the polymer network, large molecules cannot penetrate across the adsorbent-cross linker chain [45]. Therefore, the adsorption of parabens is expected to decrease when the amount of TDI is increased, except at the contact time of 24 hours (Figure 2b). However, results in Figures 2a,b and Table 1 show that the adsorption of parabens on β-CD-TDI polymer is higher than on β-CD-HMDI polymer. This anomaly will be discussed in Section 2.3.

Table 1 shows the adsorption capacities of each paraben on β-CD-HMDI(a) and β-CD-TDI(b) polymers for single and mix solutions. In general, the adsorption capacities of each paraben in single solution are higher compared to the adsorption capacities of each paraben in mix solution. This is probably because in solution containing the mixture of parabens, there exists a competition among the four parabens to be adsorbed into a limited cavity. Adsorption capacity for β-CD-TDI(b) (0.555 mmol/g) was higher than β-CD-HMDI(a) (0.4080 mmol/g) as shown in Table 1. β-CD-TDI has higher adsorption capacity may due to their higher hydrophobicity compare to β-CD-HMDI.

When β-CD is immersed in an aqueous solution, the water molecules will fill in the apolar cyclodextrin’s cavity [34]. This polar-apolar interaction is energetically unfavorable [34]. Therefore, water molecules will be easily substituted by paraben molecules. Table 1 also shows that the adsorption capacity increases with decreasing of polarity of parabens in the order BP > PP > EP > MP (benzyl-, propyl-, ethyl-, and methyl-paraben, respectively). BP shows the highest removal either in single solution or a mixture of parabens (Table 1). This is because BP is the least polar and least soluble in water, therefore BP can easily adsorb into the cavity of β-CD due to the strongest solute-adsorbent interaction.

A good adsorbent is reusable and has high number of regeneration times. The reusability of an adsorbent can reduce the large amount of processing costs since adsorbent can be the most expensive consumable component in an experiment. Methods and materials used in the regeneration process are also important; they have to be simple and low cost. In this study, methanol was used since parabens have high solubility in methanol. Parabens can be desorbed from β-CD polymer by eluting 10 mL of methanol. The capacities of regenerated polymer are shown in Figure 4. The regenerated polymer could still achieve between 78% to 112% capacities after nine regenerations.

Application of the prepared polymers on real samples shows very positive results, especially for PP and BP. The results are shown in Tables 2 and 3. In general, the results are consistent with the adsorption of parabens from mix solution.

All reagents were used without further purification unless indicated otherwise. β-CD hydrate, 99% was purchased from Acros, Hungary. Methyl paraben (MP), ethyl paraben (EP), propyl paraben (PP), and benzyl paraben (BP) were supplied by Fluka, United Kingdom. Toluene-2,6-diisocyanate (TDI) (90%) was purchased from Fluka, Germany. 1,6-Hexamethylene diisocyanate (HMDI)(90%) was obtained from Aldrich, USA and pre-filtered methanol (HPLC grade) was supplied by Merck, Germany. N,N-Dimethylformamide (DMF) of analytical grade reagent from Fisher Scientific, UK. DMF was distilled before use. Single-solute stock solution of 100 mg/L was prepared in boiled ultrapure deionised water (Elga, USA). Mix-solute solution containing 100 mg/L each of MP, EP, PP, and BP was also prepared in the same manner. Working solutions were prepared daily by appropriate dilution of the stock solution.

HPLC was used to determine the concentration of parabens before and after removal. High performance liquid chromatography (HPLC) system that used consist of a LC-20AT pump, a SPD-M20A diode array detector, a SIL-20AHT auto sampler, a CTO-20AC column oven and CBM-20A communication bus module (Shimadzu, Japan). The parabens were separated on a reversed-phase Chromolith RP-18 monolithic column (100 mm × 4.6 mm i.d., Merck, Germany). LC Solution Software (release 1.23 SPI) was used to monitor the LC system and data acquisition. The mobile phase consisted of methanol (eluent A) and water (eluent B), gradient elution: 0 (zero) minutes, A 50% and B 50%, 9 minutes, A 79.8% and B 20.2%. Separations were accomplished at a flow rate of 1.00 mL/min at 30 °C and injection volume of 20 μL, at detection wavelength of 254 nm.

β-CD polymers were prepared according to method of Bhaskar et al. [39]. Briefly, 10 g of β-CD was first dissolved 30 mL of dry DMF at room temperature. A calculated amount of cross-linker (HMDI or TDI) was then added dropwise and the mixture stirred for 4 hours at 70 °C. The polymer formed was then precipitated by the addition of excess methanol, filtered, washed with methanol and dried overnight in an oven at 70 °C. The dried polymer was first ground and sieved using a 53 μm sieve before used.

Different molar ratios of β-CD:cross-linker were studied and the polymer with the highest adsorption capacity towards parabens in each series was chosen for further investigation. The polymers were prepared in two batches; one using HMDI as the cross-linking agent and the other using TDI. Eight different molar ratios of β-CD:cross-linker were used for each batch, namely, 1:4(a), 1:7(b), 1:10(c), 1:13(d), 1:16(e), 1:19(f), and 1:12(g).

Fourier transform infrared spectroscopy (FT-IR) spectra of polymers were taken on a Perkin-Elmer RX1 FT-IR spectrometer from 400–4000 cm⁻¹. The samples were prepared in pellet form using spectroscopic grade KBr powder.

Water retention property of the polymer was determined by packing it into an empty solid-phase extraction cartridge and dried in a desiccator for 24 hours. The cartridge was then attached to a vacuum manifold (Lichrolut, Merck, Germany) followed by the addition of ultrapure deionized water to soak the polymer. After soaking the polymer for 24 hours, the excess water was drained off at a flow rate of 2 mL/min. Weights of polymer before and after soaking was recorded. The average of triplicate was taken as show in Figure 3.

About 0.05 g of β-CD polymers was added into a flask containing solution of parabens mixture and the flask was shaken on an orbital shaker (YIHDER TS-560, Taiwan) at 150 rpm. The initial concentrations of parabens and shaking times were varied according to the parameters examined. The supernatant was withdrawn at pre-determined time intervals using a syringe fitted with a 0.2 μm pore size PTFE membrane filter. Three blank samples were treated in the same conditions for every bath of polymers. The adsorption capacity of parabens, q, is calculated as follows:

where C₀ and C denote the initial concentration of solution before removal and concentration of solution at equilibrium, both solutions were determined by using HPLC as discussed in Section 3.2. V is the volume of solution and m is mass of adsorbent used.

To study how effective the recycled polymer was for paraben removal, 0.2 g of polymer was stirred in 100 mL of 100 ppm parabens mixture solution for 5 hours and filtered. The concentrations of parabens in the filtrate were determined by HPLC. The polymer was then dried in an oven at 70 °C overnight. After drying, 10 mL of methanol was used to elute the adsorbed parabens and the polymer again dried overnight at 70 °C. The experiment was repeated for 9 additional cycles using the same polymer.

The polymers were applied on 7 samples to determine their removal ability in different matrices. The samples can be divided into 2 categories, i.e., water sample and commercial products. Water samples were tap water, bottled drinking water (purchased from hypermarket in Kuala Lumpur), swimming pool water (collected from the University of Malaya, Sports Centre’s swimming pool), river water (from a river flowing through the campus of University of Malaya), and treated wastewater effluent (from sewage treatment plant in Kuala Lumpur). 50 mL of each sample was used for the study. Soy sauce, and mouth rinse were used to represent commercial products. 1 mL/g of these samples was diluted by 50 mL of deionized water. Parabens found in the ingredient list of the commercialised samples except soy sauce.

All the samples were spiked with 100 μL of 100 mg/L parabens solution. Then, 0.05 g of polymer was added to the samples and was shaken overnight. Concentrations of parabens in the samples before and after removal were determined by HPLC.