Rapid Determination of Vitamin D3 in Aquatic Products by Polypyrrole-Coated Magnetic Nanoparticles Extraction Coupled with High-Performance Liquid Chromatography Detection

A method using polypyrrole-coated Fe3O4 (Fe3O4@PPy composites) based extraction coupled with high performance liquid chromatography was developed for adsorption and detection of trace vitamin D3 (VD3) in aquatic products. The fabricated Fe3O4@PPy composites were characterized by scanning electron microscopy, transmission electron microscopy, X-ray diffraction, Fourier transform infrared spectroscopy, and thermogravimetric analysis. Fe3O4@PPy composites showed efficient adsorption of VD3 at pH 9.0 and 25 °C with a dose of 25 mg per 10 mL of sample solution and an adsorption time of 11 min. Methanol was selected as the desorption solvent to recover VD3 from Fe3O4@PPy composites after 3 min of static treatment. Fe3O4@PPy composites can be used for VD3 adsorption at least two times. The developed method showed a good linearity for VD3 determination in the range of 0.1–10 μg/mL with a correlation coefficient of 0.9989. The limits of detection and quantification were 10 ng/mL and 33 ng/mL, respectively. The recovery of VD3 in a spiking test was 97.72% with a relative standard deviation value of 1.78%. The content of VD3 in nine aquatic products was determined with this method. Our results show that Fe3O4@PPy composites provide a convenient method for the adsorption and determination of VD3 from the complex matrix of aquatic products.


Introduction
Vitamin D (VD), a group of fat-soluble secosteroids, plays important roles in the physiological activity of humans. The traditional role of VD is to maintain calcium and phosphorus homeostasis and normal bone function and structure [1]. Recently, more and more studies have reported that various chronic diseases, including insulin resistance, diabetes, and cardiovascular disease are linked to a VD deficiency [2,3]. In nature, VD mainly exists in two physiological forms based on different side-chains. Ergocalciferol (VD 2 ) is mainly found in plants and cholecalciferol (VD 3 ) comes from animals [4]. It is considered that fatty fish, fish liver, and fish oil are excellent supplementation sources for natural VD 3 [5,6]. Other foods, such as meat and egg yolk, also contain high amounts of VD 3 [5].
The complexity of food matrices means that measuring the content of VD 3 needs appropriate pretreatment before instrumental analysis. Appropriate extraction techniques can remove substances, including proteins, polysaccharides, and lipids that can interfere with VD 3 detection, and therefore, enhance the accuracy and detection limit of the method [7]. Popular extraction techniques are liquid-liquid extraction, solid phase extraction (SPE), dispersive liquid-liquid microextraction, magnetic solid phase extraction (MSPE), and enzyme linked immunosorbent assay [8][9][10]. Considering the amount of organic solvent for 10 min at 6000 rpm. Finally, the supernatant was collected and used for the VD 3 adsorption experiments.

VD 3 Concentration Measurement
VD 3 concentration was determined by de Azevedo's method [25] with slight modifications. Briefly, 10 µL of sample filtrate was loaded onto a 1260 Agilent HPLC system (Waldbronn, Germany) with a C 18 column (4.6 × 250 mm, 5 µm) (Elite, Dalian, China) at 40 • C. The mobile phase was 100% methanol with a flow rate of 0.6 mL/min and detection at 264 nm. To calculate sample VD 3 concentration, VD 3 standards were prepared in methanol with concentrations ranging from 0 to 10.0 µg/mL and assayed under the same conditions. VD 3 concentration was determined from the calibration curve for the VD 3 standards (y = 15.38 × −0.0583, R 2 = 0.9975).

Preparation of Fe 3 O 4 @Polymerization of Pyrrole (Fe 3 O 4 @PPy) 2.4.1. Fe 3 O 4 Nanoparticles (Fe 3 O 4 NPs) Preparation
Fe 3 O 4 NPs were prepared according to the method of Nalle, Wahid, Wulandari and Sabarudin (2019) [26] with slight modifications. In brief, 0.18 g of FeCl 2 ·4H 2 O and 0.3 g of FeCl 3 ·6H 2 O were dissolved in 15 mL of deionized water degassed by ultrasonic treatment, and stirred for 15 min at 55 • C. Then, 7.2 mL of 3 mol/L NaOH was rapidly added to the mixture and continuously stirred for 40 min. After 30 min of incubation at 90 • C in a water bath, the reaction solution was cooled to room temperature. The resulting black sediment was washed repeatedly with deionized water until a pH of 7.0 was achieved, then dried in a vacuum oven at 60 • C overnight. The generated Fe 3 O 4 NPs were stored for further experiments.

Synthesis of Fe 3 O 4 @PPy Composites
Fe 3 O 4 NPs were coated by polymerization of pyrrole (PPy) using the method of Zhang et al. (2020) [23] with few modifications. In brief, 0.028 g of sodium dodecyl sulfate and 0.2 g of Fe 3 O 4 were added to 80 mL of deionized water and sonicated for 20 min to obtain a homogeneous dispersion. Subsequently, pyrrole monomer at a ratio of 1:1, 3:1, 5:1 (v/w) with respect to Fe 3 O 4 NPs content was added and stirred for 10 min. Then, 10 mL of 1 mol/L FeCl 3 ·6H 2 O was added slowly to the reaction and stirred for 12 h at room temperature. The generated Fe 3 O 4 @PPy particles coated with different PPy ratios were recovered using an external magnetic field, washed with deionized water three times, and finally dried in a vacuum oven at 60 • C overnight. The VD 3 adsorption rate and particle size of these freeze dried Fe 3 O 4 @PPy were then determined.

Adsorption Rate for VD 3
Fe 3 O 4 @PPy powder (50 mg) was added to a saponified solution (10 mL resulting from 4 g of Penaeus sinensis by-products). After 20 min of static adsorption at room temperature, the Fe 3 O 4 @PPy composites were separated from the mixture using an external magnet, washed with 2 mL of deionized water and 2 mL of ethanol. Then, the collected Fe 3 O 4 @PPy particles were added to 2 mL of methanol and desorbed for 10 min in a standing state, followed by separation of the Fe 3 O 4 @PPy particles using the action of a magnet. The remaining solution was evaporated at 45 • C to remove the methanol, and then redissolved in 500 µL of methanol. After filtration through a 0.22 µm filter, the concentration of VD 3 was determined as described in Section 2.3. The adsorption rate of VD 3 was calculated according to Equation (1) as follows: where c represents the concentration of VD 3 (µg/mL), v represents the total volume of the filtrate (mL), and m 0 (µg) represents the theoretical amount of VD 3 extracted with liquid-liquid extraction (LLE) using hexane as the solvent, calculated by multiplying the concentration of VD 3 in the LLE by the volume. The Fe 3 O 4 @PPy composites fabricated with the ratio of PPy that showed the highest adsorption rate of VD 3 were used for further experiments.

Particle Size Measurement
The Fe 3 O 4 @PPy composites prepared with different ratios of PPy (1:1, 3:1, 5:1, v/w) were dispersed by sonication in pure water for 10 min. Then, 1.0-1.5 mL of the dispersion was dropped into the sample pool of a Zeta-sizer Nano-ZS90 (Malvern Instruments, Worcestershire, UK). The mean particle diameter (z-average) was determined in triplicate at 25 • C.

X-ray Diffraction (XRD) Analysis
XRD patterns for the Fe 3 O 4 @PPy composites or Fe 3 O 4 NPs were determined by an X-ray diffractometer (MiniFlex600, Rigaku, Japan) using Cu-Kα radiation in the region of 2θ from 10 • to 70 • at a scanning rate of 0.02 • /s.

Thermogravimetric Analysis (TGA)
The thermogravimetric property of the Fe 3 O 4 @PPy composites was measured using a thermogravimetric analyzer (DTG-60, Shimadzu, Japan) under a N 2 atmosphere at temperatures ranging from room temperature to 600 • C. Fe 3 O 4 NPs were assayed in parallel for comparison.

Adsorption Conditions
The adsorption experiments were carried out in a 100 mL conical flask containing 10 mL of the VD 3 saponification solution at various pHs (6.0-14.0) with a range of adsorbent doses (0.005-0.05 g Fe 3 O 4 @PPy composites), adsorption temperatures (25-55 • C), and adsorption times (3-30 min). After adsorption, the Fe 3 O 4 @PPy composites were collected from the mixture with an external magnet. Subsequently, VD 3 desorption from Fe 3 O 4 @PPy composites was carried out using methanol as the desorption solvent according to the conditions described in Section 2.5. The concentration of VD 3 was determined by HPLC as described in Section 2.3. The amount of VD 3 adsorbed to the Fe 3 O 4 @PPy composites was calculated as q e (µg/g) using Equation (2) as follows: where c represents the concentration of VD 3 (µg/mL), v represents the total volume of the filtrate (mL), and m p is the mass of the Fe 3 O 4 @PPy composites used in the experiment (g).

Desorption Conditions
To develop a satisfactory desorption method, the desorption solvent and desorption time were investigated further. After VD 3 adsorption, the collected Fe 3 O 4 @PPy composites were added to 2.0 mL of desorption solvent (ethanol, acetonitrile, and methanol) and sonicated or allowed to stand for 1 to 15 min at room temperature. After magnetic separation of the Fe 3 O 4 @PPy composites, the resulting solution was evaporated and then redissolved in methanol as described in Section 2.4.2. Finally, 10 µL of filtrate was analyzed for VD 3 content using HPLC as described in Section 2.3. The desorption rate of VD 3 from Fe 3 O 4 @PPy composites was calculated according to Equation (3) as follows: where c represents the VD 3 concentration (µg/mL), v represents the total volume of the filtrate (mL), and m a is the total amount of VD 3 adsorbed by the Fe 3 O 4 @PPy composites (µg).

Reusablility of Fe 3 O 4 @PPy Composites
To investigate the recyclability of the Fe 3 O 4 @PPy composites as adsorbents for VD 3 in saponified aquatic products, one batch of Fe 3 O 4 @PPy composites was used as the adsorbent to conduct the adsorption and desorption experiments. The recovery of VD 3 from recycled Fe 3 O 4 @PPy composites was compared after repeated use. In addition, the characteristics of recycled Fe 3 O 4 @PPy composites were determined using SEM, TEM, XRD and FTIR measurements.

Method Evaluation
Quantitative parameters for the HPLC determination of VD 3 after Fe 3 O 4 @PPy composites extraction, including linearity, coefficient of determination (r 2 ), limits of detection (LOD), limits of quantification (LOQ), accuracy and precision, were evaluated under optimal adsorption and desorption conditions. The sensitivity of the method was evaluated by the LOD and LOQ at a signal-to-noise ratio of 3 (S/N = 3) and 10 (S/N = 10), respectively. The accuracy of recovery was assessed by spiking saponified samples with a VD 3 standard (2 µg) and calculating recovery according to Equation (4). These samples were analyzed six times per day and the precision of the method was evaluated by intra-day relative standard deviation (RSD).

Application of Fe 3 O 4 @PPy Composites for VD 3 Detection in Aquatic Products
Aquatic products, mentioned in 2.1., were homogenized and saponified as described in Section 2.2. After the pH of the saponified sample solution was adjusted to 9.0, 25 mg of the Fe 3 O 4 @PPy composites were added for VD 3 extraction. After standing at room temperature (25 • C) for 11 min, the adsorbent was separated from the mixture using magnets. The adsorbent was rinsed with 2.0 mL of deionized water and then desorbed statically with 2.0 mL of methanol for 3 min. This methanol solution was magnetically separated from the Fe 3 O 4 @PPy particles and evaporated to dryness using a vacuum rotary evaporator, then redissolved in 500 µL of methanol. Finally, the methanol solution was filtered through a 0.22 µm organic filter, and 10 µL of the filtrate was analyzed by HPLC for VD 3 determination. The amount of VD 3 in aquatic products was expressed as µg per 100 g.

Statistic Analysis
All experimental results were expressed as the mean ± standard deviation. Analyses were performed with a one-way analysis of variance (ANOVA) and Tukey's test using SPSS ® software 19.0 (Chicago, IL, USA) to determine significant differences at p ≤ 0.05.

Effect of PPy to Fe 3 O 4 NPs Ratio on the Adsorption Rate of VD 3
As is shown in Figure 1a, the VD 3 adsorption rate for the Fe 3 O 4 @PPy composites increased with the dose of PPy. At ratio of 5:1 for PPy and Fe 3 O 4 (v/m), the adsorption rate for VD 3 reached 100% (p < 0.05). The particle size of Fe 3 O 4 @PPy composites also increased with the amount of PPy added (Figure 1b). This indicated that an increase in pyrrole monomer led to a thickening of the coating on the surface of Fe 3 O 4 NPs and provided more binding sites, thus improving the adsorption of VD 3 . In this study, Fe 3 O 4 @PPy composites prepared at ratio of 5:1 (pyrrole/Fe 3 O 4 , v/w) had the highest adsorption rate for VD 3 , and therefore, were selected for subsequent experiments.    (Figure 3a), which were accordance with previous reports on Fe 3 O 4 NPs [28]. After coating with PPy, the XRD pattern of the Fe 3 O 4 @PPy composites had similar peaks to those detected in Fe 3 O 4 NPs, indicating the presence of Fe 3 O 4 NPs. Similar results were reported for the characteristics of iron oxides in Fe 3 O 4 @PPy [10]. However, the intensities of these peaks were all decreased to some extent. Furthermore, a broad peak was observed in the low range of 20-30 • , which was ascribed to the typical amorphous structure of polypyrrole [21]. The findings in Figure 3a further provide further evidence of the existence of a PPy coating on the surface of the Fe 3 O 4 NPs, consistent with TEM images (Figure 2d).

FTIR Analysis
The FTIR spectra of the Fe 3 O 4 NPs and Fe 3 O 4 @PPy composites are compared in Figure 3b. The typical band at 567 cm −1 resulted from the stretching vibration of the Fe-O bond in Fe 3 O 4 [29]. After interaction with PPy, the bands at 779 cm −1 and 897 cm −1 related to =C-H out-of-plane vibration of pyrrole rings [18] were observed in the Fe 3 O 4 @PPy composites. Furthermore, some of the typical bands associated with PPy, such as 1298 cm −1 (C-H in-plane vibration), 1165 cm −1 (=C-H in plane vibration), and 1062 cm −1 (C-N stretching vibration) [30,31] were detected in the Fe 3 O 4 @PPy composites. The appearance of a band at 1635 cm −1 might be ascribed to the red shift of basic C=C stretching of the Py ring due to slight over-oxidation [27,31]. Additionally, the band at 2744 cm −1 related to C-H stretching vibration was dramatically increased in the Fe 3 O 4 @PPy composites. All the typical bands corresponding to PPy, as well as the missing band at 567 cm −1 , suggest that the Fe 3 O 4 NPs were enveloped by a PPy coating. Figure 3c shows the TGA curves for Fe 3 O 4 NPs and Fe 3 O 4 @PPy composites. The weight loss from Fe 3 O 4 NPs was 3.02% after heating from 25 • C to 100 • C, which was related to the evaporation of a small amount of water. The weight of Fe 3 O 4 NPs (94.96%) was stable at temperatures above 100 • C. By comparison, three stages were detected in the TGA pattern for the Fe 3 O 4 @PPy composites. In the first stage, the evaporation of water and a slight degradation of PPy could be responsible for the observed weight loss of 6.58% when Fe 3 O 4 @PPy was heated from 100 • C to 200 • C. In the second stage, the Fe 3 O 4 @PPy composites decomposed dramatically at about 250 • C, which is consistent with the TGA results for a Fe 3 O 4 -PPy composite with a Fe 3 O 4 content of 34% described by Chen et al. (2003) [30]. In the third stage, the TGA pattern for Fe 3 O 4 @PPy showed that 34.87% of the core content (Fe 3 O 4 ) was left behind at 450 • C. The results in Figure 3c should be sufficient to prove that the fabricated Fe 3 O 4 @PPy composites have a core-shell structure.

Effect of pH
The pH value of the adsorption environment is a critical factor in MSPE methods, since it can change the surface net charges for both the magnetic materials and the target compounds [23]. In this study, the initial pH of the sample solution (saponified Penaeus sinensis by-products) was 14.0. By decreasing pH, the adsorption (q e ) of VD 3 increased from 5.91 µg/g at pH 14.0 to 31.64 µg/g at pH 9.0 for the Fe 3 O 4 @PPy composites (Figure 4a). However, when the pH was lowered further, the adsorption capacity decreased sharply and was reduced to zero at pH 6 and pH 5. These findings indicate that an acidic environment is not suitable for the adsorption of VD 3 to the Fe 3 O 4 @PPy composites. Under acidic conditions, some proteins with an isoelectric point of about 5.0 present in saponified aquatic products could precipitate, which might trap the originally released VD 3 , thus resulting in poor adsorption of VD 3 to Fe 3 O 4 @PPy. To reveal the role of pH on the adsorption capacity of the Fe 3 O 4 @PPy composites, the zeta potential was measured at different pH values. As shown in Figure 4a, the Fe 3 O 4 @PPy composites were negatively charged when pH conditions were greater than 9.0 (zeta potential < 0). In contrast, the Fe 3 O 4 @PPy composites were positively charged when the pH was below 9.0 (zeta potential > 0). It should be noted that the Fe 3 O 4 @PPy composites had a net charge close to neutral at pH 9.0 and demonstrated the highest adsorption capacity (q e ) for VD 3 . The results in Figure 4a indicate that the driving force for adsorption of VD 3 onto the Fe 3 O 4 @PPy composites does not rely on charge interactions, but rather, hydrophobic interactions and/or π-π stacking.

Effect of Adsorbent Dose
To achieve a good adsorption efficacy with a minimal dose of the Fe 3 O 4 @PPy composites, different amounts were applied to extract VD 3 from saponified Penaeus sinensis by-products. As the adsorbent dose increased from 5 mg to 50 mg, the amount of VD 3 adsorbed by the Fe 3 O 4 @PPy composites increased with each increase in dosage up to 25 mg, when the maximum adsorption was achieved, and remained stable at higher doses (Figure 4b). Increasing the amount of adsorbent can provide more adsorption sites for the target ingredient, which helps improve the adsorption speed [23]. However, the efficiency of the adsorbent should also be considered because the amount of adsorbent comes at a cost. In this study, the adsorption capacity of the Fe 3 O 4 @PPy composites for VD 3 decreased gradually with each increase in dose, which could be ascribed to the existence of excess amounts of adsorbent. This could explain why the adsorption dose increased gradually, however the adsorption efficiency did not increase accordingly. Considering the high content of VD 3 adsorbed and a relatively higher q e compared to other doses, 25 mg was chosen as the optimum dose for the subsequent study.

Effect of Adsorption Temperature
VD is sensitive to heat and degrades easily under high temperatures [32]. The extraction efficiency of VD 3 from milk by Fe 3 O 4 @PPy decreased as the extraction temperature increased [10]. However, in this study, when the adsorption temperature ranged from 25 • C to 55 • C, the adsorption capacity of the Fe 3 O 4 @PPy composites for VD 3 remained stable (Figure 4c). A difference in the sample matrix between the saponified aquatic products used in this study and the unsaponifiable milk samples in the literature might explain the differential effect of temperature on VD 3 adsorption. Similar to our results, Zhang et al. (2020) [23] described that temperature had little effect on the extraction of 11 antiseptic ingredients with Fe 3 O 4 @PPy composites. Considering the energy saving and a convenient operation, 25 • C was selected as the adsorption temperature for subsequent experiments.

Effect of Adsorption Time
According to previous studies, compared to a conventional SPE method, a quicker equilibrium between the target component and Fe 3 O 4 @PPy nanoparticles can be reached as because of their high surface area and short diffusion route [33]. Figure 4d shows that the adsorption capacity of the Fe 3 O 4 @PPy composites for VD 3 increased rapidly from 0 to 11 min, then remained stable thereafter. This meant that 11 min was sufficient time to adsorbVD 3 molecules from the sample matrix. A similar adsorption time of 10 min was reported for VD extraction from milk with Fe 3 O 4 @PPy [10]. Therefore, we selected 11 min as the appropriate time for VD 3 adsorption to Fe 3 O 4 @PPy composites.

Desorption Conditions
Two desorption methods, namely ultrasonic and static treatment, were used to recover VD 3 from the Fe 3 O 4 @PPy composites (Figure 5a). No significant difference was found for the two desorption methods (measured as the VD 3 content in the desorbed solution) (p > 0.05). Acetonitrile, methanol, and ethanol were used as desorption solvents under static conditions to recover VD 3 from the Fe 3 O 4 @PPy composites (Figure 5b). Clearly, acetonitrile and methanol were more efficient than ethanol for desorbing VD 3 (p < 0.05). After a consideration of cost, applicability, and safety, we selected methanol as the desorption solvent. Subsequently, the desorption time with methanol was tested for the Fe 3 O 4 @PPy composites as shown in Figure 5c. The amount of VD 3 released was found to increase over time from 0.5 min to 3 min, and then remained stable after 3 min. Therefore, we chose 3 min as the desorption time for the follow-up study.

Regeneration of Fe 3 O 4 @PPy Composites
To determine if the Fe 3 O 4 @PPy composites prepared in this study could be reused after adsorption of VD 3 in aquatic products, the performance of the particles was evaluated after repeated adsorption and recovery of VD 3 . After one cycle of adsorption and desorption of VD 3 , the Fe 3 O 4 @PPy composites were dried in a vacuum oven and used for another round of adsorption and desorption. As is shown in Figure 6, the recovery of VD 3 decreased rapidly each time the Fe 3 O 4 @PPy composites were recycled. At the end of two cycles, the recovery of VD 3 was 82.25%, but this dropped to 57.48% after three cycles. Our results suggested that the Fe 3 O 4 @PPy composites could be reused at least twice for a VD 3 recovery over 80%. The decrease in VD 3 recovery might be due to damage of the PPy shell with repeated use, thus leading to incomplete adsorption of VD 3 . To investigate possible reasons for the decreased VD 3 recovery, the characteristics of recycled Fe 3 O 4 @PPy composites were compared using SEM and TEM, as well as XRD and FTIR analysis. No obvious microstructural changes were observed for the recycled Fe 3 O 4 @PPy composites with TEM (Figure 7c,d) when compared to its original condition (as shown in Figure 2d). In contrast, SEM identified some irregular areas wrapped in spherical particles after the composites that had been reused twice (see dotted outlines in Figure 7b), which were not present in unused composites (as shown in Figure 2c) or those that had been recycled once (Figure 7a). These changes might reduce the contact area for VD 3 adsorption, which might be responsible for the reduction in VD 3 adsorption.  The specific peaks in the XRD spectrum of the Fe 3 O 4 @PPy composites related to the Fe 3 O 4 NPs were not affected by adsorption of VD 3 (Figure 7e). A similar phenomenon was found for the reused magnetic composites. However, the broad peak in the range of 20-30 • that was associated with the typically amorphous structure of PPy shifted obviously to smaller angles, suggesting structural changes in the PPy shell after the Fe 3 O 4 @PPy composites had been reused twice. Regarding the FTIR spectra, the band at 1635 cm −1 associated with C=C stretching vibration of PPy disappeared from the Fe 3 O 4 @PPy com-posites after adsorption of VD 3 (Figure 7f), which implied that the functional C=C group is involved in a hydrophobic interaction between Fe 3 O 4 @PPy and VD 3 . However, the band representing C=C stretching vibration in reused Fe 3 O 4 @PPy composites was not restored after rinsing with methanol (the desorption solvent). This means this functional group might be irreversibly damaged by the desorption solvent. Furthermore, the reused Fe 3 O 4 @PPy composites showed increased intensities for the bands at 1062 cm −1 , related to C-N stretching vibration, and 1165 cm −1 , assigned to =C-H in plane vibration [30]. Similarly, the intensities of the bands at 897 cm −1 and 779 cm −1 , attributed to the out-of-plane vibration of the =C-H in pyrrole rings [18], increased each time Fe 3 O 4 @PPy was recycled. In addition, the functional group at 2744 cm −1 shifted towards blue after adsorption of VD 3 compared to a red shift that increased each time the Fe 3 O 4 @PPy composites were reused. These changes suggest that the band related to the C-H stretching vibration in PPy might be crucial for VD 3 adsorption and desorption. The increases in the intensities of the typical peaks related to the PPy ring indicate that the PPy coating might partially fall off the surface of the Fe 3 O 4 @PPy composites with VD 3 loading, which cannot be recovered by external magnetic adsorption, thus reducing VD 3 recovery.

Method Validation and Application
Good linearity for the VD 3 assay was achieved in the range of 0.1-10 µg/mL in saponified solutions of shrimp by-products with correlation coefficients (r 2 ) reaching 0.9989. The limit of detection (LOD) (S/N = 3) and the limit of quantification (LOQ) (S/N = 10) were 10 ng/mL and 33 ng/mL, respectively. In the spiking test, the recovery of VD 3 was 97.72%, and the relative standard deviation (RSD) value was 1.78% (Table 1). The VD 3 content of Penaeus sinensis by-products and other aquatic products was determined using Fe 3 O 4 @PPy composites-based extraction coupled with HPLC detection. The results in Table 2 show significant variations in VD 3 content for the various species and parts tested in this study. Similarly, some authors have also reported significant differences in VD 3 content for fish, both between species and within species [6]. For example, the average content of VD 3 in mahi-mahi was only 1.11 µg/100 g, while it reached 45.3 µg/100 g in tilapia [34]. Baltic salmon had significantly higher VD 3 content (26.5 µg/100 g) than farmed Norwegian salmon (5.9 µg/100 g) [6]. In this study, low average VD 3 content was detected in squid (Loliolus japonica) meat (2.86 µg/100g) and in silver pomfret (Pampus argenteus) (4.65 µg/100 g). By comparison, the tested clams Paphia undulata, Cyclina sinensis, and razor clam (Sinonovacula constricta) had high VD 3 content (>70 µg/100 g), suggesting these shellfish are a good source of dietary VD 3 . Furthermore, it was noted that the average VD 3 content in the by-products of Penaeus sinensis (Solenocera crassicornis), Pacific white shrimp (Litopenaeus vannamei), and cuttlefish (Sepia esculenta) was higher than 10 µg/100 g, which exceeds the recommended daily intake of 5 µg/d [6]. In addition, the content of VD 3 was higher in the by-products than that the corresponding muscle tissue (p < 0.05). These findings also suggest that the by-product of Pacific white shrimp should be considered a good raw material for VD 3 extraction. Note: "*" represents the VD 3 content, expressed as the mean ± standard deviation (SD) (n = 3). Different letters (a-i) against the mean content of VD 3 for each species suggest a significant difference (p < 0.05).

Conclusions
In the present work, we fabricated an effective adsorbent composed of Fe 3 O 4 NPs functionalized with a PPy coating for VD 3 extraction. The results of SEM, TEM, XRD, FTIR and TGA prove that the Fe 3 O 4 @PPy composites have a core-shell structure. The adsorption of VD 3 from saponified Penaeus sinensis by-products to Fe 3 O 4 @PPy composites was optimal under the following conditions: pH 9.0 with a 25 mg dose at 25 • C and 11 min adsorption time. The adsorbed VD 3 could be effectively desorbed from the binding sites of the Fe 3 O 4 @PPy composites with methanol after static contact for 3 min. The accuracy and reproducibility of the developed method for VD 3 extraction and detection were quite satisfactory as evidenced by a high linear correlation coefficient and a low intra-day RSD. Compared to other conventional methods, the proposed method is more rapid since it does not require complicated extraction and concentration procedures. Instead, VD 3 can be separated from complex samples within minutes by quick and easy magnetic separation. Furthermore, this method is environmentally friendly since it requires less organic solvent. In addition, the results for VD 3 content in the tested aquatic products will provide important reference information for the rational selection of products for the development of VD 3fortified foods.