3.1 Growth ZIF-8 on CC for EE-TFME of TCs
ZIF-8, abounding rich π systems and large pores (11.6 Å) with exceptional thermal and chemical stability, can extract TCs via porosity effect, hydrogen bonding, charge-complementary and π-π interaction. After treating CC, ZIF-8 was modified on CC via in situ growth method. Then ZIF-8/CC was applied the positive voltage for EE-TFME of TCs. ZIF-8/CC used as work electrode and platinum wire counter electrode were inserted into solution to construct a full circuit, TCs would enrich through electric field (Fig. 2). After the extraction, ZIF-8/CC was eluted by 1% acetic acid methanol solution. Finally, samples would be analyzed by HPLC-PDA.
3.2 Characterization of ZIF-8/CC
The SEM images showed that ZIF-8/CC still had many bunchy weaves (Fig. 3A), and almost every carbon fiber was covered with ZIF-8. In addition, the crystal anchored on the CC fiber presented a prismatic structure (Fig. 3B), which was consistent with the morphology of ZIF-8. ZIF-8 irregularly accumulated on CC surface, which increases the contact area between the electrolyte and the electrode surface, thus possibly promoting electrochemical behavior (Huang et al. 2006; Sun et al. 2022).
The Fourier transform infrared (FT-IR) spectra of ZIF-8 and ZIF-8/CC showed the peak at 3133 cm− 1 was due to C-H extension in methylimidazole ring and at 2500–3000 cm− 1 were due to O-H, C-H and N-H (Fig. 4A). The absorption peaks of 1306 and 991 cm− 1 were C-N stretching bending mode of methylimidazole ring, and the characteristic peak of Zn-O bond appeared at 528 cm− 1 (Lin et al. 2014). The FT-IR of ZIF-8/CC was slightly burr due to the presence of CC, while the peaks were almost the same as that of ZIF-8, which indicated that ZIF-8/CC was successfully synthesized.
To explore the crystal structure of ZIF-8 and ZIF-8/CC, XRD patterns were given (Fig. 4B). The peaks of ZIF-8 were in conformity to the images in CCDC database and the previous reports (Asadian et al. 2020). The main diffraction peaks appeared at 2θ = 7.3°, 10.4°, 12.7°, 14.7°, 16.4°, 18.0°, 24.5° and 26.6°, which indicated the synthesized ZIF-8 crystal has good crystallinity and purity. Due to the existence of CC, the peaks of ZIF-8/CC had burrs. ZIF-8 has been successfully loaded onto the surface of CC.
The porosity and specific surface area of ZIF-8/CC were investigated by N2 adsorption-desorption at 77 K (Fig. 4C). The N2 adsorption capacity increased sharply at 0 ~ 0.05 P/P0 and 0.9 ~ 1.0 P/P0, which may be due to the formation of the second layer with the increase of relative pressure. The specific area of ZIF-8/CC was 1829.160 cm2/g, the pore volume was 0.73 cc/g, and the DFT pore size distribution was 0.966 nm (Fig. 4D), which were consistent with the previous reports (Liu et al. 2019).
3.3 Properties of ZIF-8 for EE-TFME
The extraction efficiency of CC and ZIF-8/CC was assessed (Fig. 5A). Compared with CC, the extraction efficiency of TCs by ZIF-8/CC was improved within 15 min, and the extraction efficiency of ZIF-8/CC was 2.27 times for OTC, 1.92 times for TC and 2.07 times for DC. This was attributed to the rich π system of ZIF-8, which was conducive to the adsorption of TCs and thus improved the extraction efficiency.
The conductivity of CC and ZIF-8/CC were measured by cyclic voltammetry (Fig. 5B). The redox current of ZIF-8/CC was higher than that of CC, with the oxidation peak of 2.19 mA and the reduction peak of -2.19 mA, which were nearly 5.1 times and 4.8 times higher than those of CC (0.43 mA and − 0.44 mA). This may be due to the excellent specific surface area of ZIF-8, and the irregular accumulation of ZIF-8 on the surface area of CC resulting in many gaps, which increased the surface area and thus improve the electrical activity (Zeng et al. 2019).
The sorbents reached a dynamic equilibrium in the sample solution. Under the external voltage, the recoveries of TCs gradually increased with the extraction time change from 5 min to 20 min. (Fig. 5C). However, the recoveries decreased from 20 min to 25 min, TCs may electrolyze into other substances (Yexin et al. 2021). In addition, the extraction efficiency of ZIF-8/CC on TCs without external electric field was tested. With the extraction time increased from 5 min to 30 min, DC reached the extraction equilibrium in 25 min, while OTC and TC both reached the extraction equilibrium in 30 min (Fig. 5D). The fast extraction equilibrium was due to the large contact area of TFME and the application of external electric field. ZIF-8/CC had excellent conductivity (Zeng et al. 2019), which promoted the movement of TCs to extractant and the strong interaction between ZIF-8 and TCs.
3.4 Optimization of EE-TFME procedure
3.4.1 Effect of pH
A key factor affecting the adsorption process is the pH. To determine the best pH, the recoveries of ZIF-8/CC for TCs were measured at pH 4.0–9.0 (Fig. 6A). The recoveries increased with the increase of pH at 4.0–6.0, while at pH 7.0–9.0, the recoveries decreased. This may be due to the instability of ZIF-8 under acidic conditions, and TCs were easily dissolved in water and degraded into other substances under alkaline conditions, and both TCs and ZIF-8 were negatively charged and repel each other (Sereshti et al. 2021). Therefore, pH 6.0 was chosen as the best value.
3.4.2 Effect of voltage
TCs would be deprotonated and negatively charged at pH 6.0, which made them move to ZIF-8/CC through charge-complementary and electrophoresis. The applied voltage varying from 0.2 to 0.9 V was studied (Fig. 6B). The extraction efficiency of the TCs increased with the voltage increased from 0.2V to 0.6V, which may result from an electric double layer at the electrode/solution interface. However, when the voltage further exceeded 0.6 V, the recoveries gradually decreased. TCs were directly oxidized to ketones and other substances on the surface of the ZIF-8/CC electrode at higher voltage (Sheberla et al. 2017; Zheng et al. 2019). Therefore, 0.6 V was chosen as the best voltage.
3.4.3 Effect of desorption solution
To improve the elution efficiency of TCs, the eluting solvent must be easier to adsorb the target than the adsorbent. The desorption capacity of PB solution (pH 9.0), acetonitrile and methanol were tested (Fig. 6C), and methanol had the best elution efficiency. ZIF-8/CC was synthesized in methanol system, the interference of its surface groups could be reduced in methanol, thus obtaining higher recoveries (Liu et al. 2017). Further changing the pH of methanol, 1% acetic acid methanol was prepared and tested, the recoveries of OTC and TC were higher than those of in methanol. Finally, 1% acetic acid methanol solution was selected as eluent.
3.4.4 Effect of elution time
Appropriate elution time is helpful for complete elution. The effect of elution time between 5 min and 30 min was studied (Fig. 6D). The recoveries of OTC reached the maximum at 20 min, while TC and DC reached the maximum at 25 min. Then the recoveries of OTC and DC decreased slightly, probably because TCs were re-adsorbed on ZIF-8/CC. Therefore, 25 min was selected as the reasonable elution time.
3.5 Analytical method validation
The validation of the proposed method was evaluated by linear range, LOD, LOQ, inter-day and intra-day RSD. Each sample was supplemented with 50 ng/mL standard sample solution, and each sample was measured five times in parallel (Table 1). The linear ranges of this method were 2.5–200.0 ng/mL, R2 ≥ 0.9916, LODs were 2.0–2.5 ng/mL, LOQs were 8.0–8.5 ng/mL, and the intra-day and inter-day RSD are less than 6.7% and 7.1% respectively, which indicated that this method had good reproducibility. According to the MRL of TCs in China, this method can be used to detect TCs in honey and milk.
Table 1
Analytical performance and parameters of the proposed method
Targets | R2 | LOD (ng/mL) | LOQ (ng/mL) | Linear range (ng/mL) | RSD (50 ng/mL, %) |
Inter-day | Intra-day |
OTC | 0.9916 | 2.5 | 8.5 | 10.0–200.0 | 4.2 | 7.1 |
TC | 0.9958 | 2.5 | 8.0 | 10.0–200.0 | 3.9 | 4.8 |
DC | 0.9962 | 2.0 | 8.0 | 8.0–200.0 | 6.7 | 5.3 |
This method was compared with the sample pretreatment and detection methods of TCs in other literatures, including extraction time, adsorbent, LOD and linear range (Table 2). Ultrasonic-assisted extraction combined with solid-phase extraction (UAE-SPE) took 40 min to reach the extraction equilibrium (Zhou et al. 2009), and dispersion liquid-liquid microextraction (DLLME) combined with hollow fiber also took 40 min to reach the extraction equilibrium (Xu et al. 2017). The LODs obtained by SPE combined with molecular imprinting were 20–40 ng/mL (Feng et al. 2016), which were higher than this method. The LODs of magnetic-solid phase extraction (MSPE) using carbon nanotubes with limited access and magnetic nanoparticles coated with carbon nanofibers were higher than that of this method, and the extraction time was both 30 min (de Faria et al. 2017; Vuran et al. 2021). In addition, the LODs and extraction time of EE-TFME were equivalent to those of electrochemical control solid phase micro-extraction (EC-SPME) (Sereshti et al. 2021).
Table 2
Comparison with other reported methods for determining TCs
Method | Adsorbent | LOD (ng/mL) | Linear range (ng/mL) | Extraction time/min | Reference |
UAE-SPE | Oasis HLB | - | 100–5000 | 40 | (Zhou et al. 2009) |
SPE-HPLC-DAD | Molecularly imprinted polymer | 20–40 | 100–1000 | - | (Feng et al. 2016) |
DLLME-HPLC-UV | Hollow fiber membranes | 0.96–3.6 | 5–2500 | 40 | (Xu et al. 2017) |
HPLC-DAD | Restricted access carbon nanotubes | 7.5–13.2 | 50–200 | 30 | (de Faria et al. 2017) |
SPE-HPLC-FLD | electrospun graphene oxide–doped poly (acrylonitrile-co-maleic acid) nanofibers | 20.4–44.8 | 5–500 | - | (Weng et al. 2019) |
MSPE-HPLC-PDA | nanofiber coated magnetic particles | 3.55 | 10–600 | 30 | (Vuran et al. 2021) |
EC-SPME-HPLC-UV | polyaniline-graphene oxide nanocomposite | 2.4–7.6 | 8–750 | 20 | (Sereshti et al. 2021) |
EE-TFME-HPLC-PDA | ZIF-8/CC | 2.0–2.5 | 8.0–200.0 | 20 | This work |
HPLC-DAD: high-performance liquid chromatography-diode array detection; HPLC-UV: high-performance liquid chromatography-ultraviolet; HPLC-FLD: high-performance liquid chromatography-fluorescence detection |
3.6 Analysis of real samples
To further evaluate the method, the method was verified in real samples. TCs in honey and two kinds of milk samples were detected and analyzed under the optimal pretreatment conditions. The HPLC-PDA chromatograms of TCs in blank honey and two kinds of milk samples and their spiked (50 ng/g) samples was showed in Fig. 7. DC was detected in two kinds of milk samples, and none of the other samples were detected. The content of DC was lower than LOQ, which was also lower than MRL of EU and China. To verify the accuracy of the method, the spiked recovery experiments of 50 ng/g and 100 ng/g were carried out on three real samples, and their spiked recoveries and RSDs were calculated respectively (Table 3). All recoveries of TCs ranged from 82.4–102.9%, and RSDs were less than 8.2%. Therefore, this method can be used to analyze and detect trace TCs in honey and milk.
Table 3
Concentrations, RSD (%, n = 5) and spike recoveries (%) of TCs in real sample
Real samples | Spiked concentration (ng/g) | Recoveries (RSD) |
OTC | TC | DC |
Honey | 0 | ND | ND | ND |
50 | 102.9 (3.5) | 98.6 (4.6) | 93.4 (4.6) |
100 | 91.9 (3.1) | 96.4 (8.0) | 93.1 (4.1) |
Milk1 | 0 | ND | ND | ༜LOQ |
50 | 94.1 (6.8) | 101.6 (4.6) | 94.8 (6.2) |
100 | 91.6 (2.7) | 95.2 (6.1) | 86.4 (5.8) |
Milk2 | 0 | ND | ND | ༜LOQ |
50 | 93.2 (2.8) | 90.8 (3.6) | 91.4 (8.2) |
100 | 82.4 (6.4) | 83.7 (5.1) | 86.9 (4.2) |