Specific recognition of cationic paraquat in environmental water and vegetable samples by molecularly imprinted stir-bar sorptive extraction based on monohydroxylcucurbit[7]uril–paraquat inclusion complex

Molecularly imprinted stir-bar coatings were created based on a hydroxylcucurbit[7]uril–paraquat inclusion complex. The inclusion complex that contained paraquat (PQ) as a template and monohydroxylcucurbit[7]uril ((OH)Q[7]) as a monomer was preassembled mainly through cavity inclusion interaction of (OH)Q[7] to form a one-dimensional self-assembly structure. The inclusion complex was anchored chemically on the surface of a glass stir bar with hydroxy-terminated poly(dimethylsiloxane) by the sol–gel technique to obtain a molecularly imprinted polymer-coated stir bar (MIP-SB). The molecularly imprinted coating showed specific adsorption for cationic PQ in aqueous media. Other quaternary amine compounds with a similar structure that coexisted in the solution, such as ethyl-viologen, diquat, and difenzoquat, were almost not extracted by the prepared MIP-SB. The sorptive capacity of the MIP-SB for PQ was nearly four times that of the non-imprinted stir bar (NIP-SB). The recognition mechanism indicated that the selectivity and extraction capacity resulted mainly from the imprinted cavity in the polymer that was formed by a one-dimensional assembly structure consisting of the (OH)Q[7]–PQ inclusion complex. The imprinted cavity was complementary to the PQ in shape, size, and functionality. A method to determine PQ in environmental water and vegetable samples was developed by combining MIP-SB sorptive extraction with HPLC-UV. The linear range was from 100 to 10,000 ng L⁻¹ with a 8.2 ng L⁻¹ detection limit for water samples and 0.02–0.85 mg kg⁻¹ with a 0.005 mg kg⁻¹ detection limit for vegetable samples. The limit of detection for both samples was lower than the EU-established maximum residual levels and that of other previously reported methods. The average recoveries were 70.0–96.1% with a relative standard deviation ≤ 7.6%, which showed the successful application in real sample analysis.