Rapid Detection Of Microplastics And Pollutants Adsorbed On Their Surfaces

Microplastics?MPs?,as an emerging contaminant,exist widely in the water environment,and its availability in aquatic organisms is one of its main environmental risks.MPs have the characteristics of enriching trace hydrophobic organic contaminants?HOCs?in the surrounding water,and these adsorbed HOCs are the main source of chemical hazards of MPs.Currently,the most commonly used identification methods of MPs are Raman spectroscopy and Fourier transform infrared spectroscopy,which can provide accurate detection results.These methods require complex steps of sample pretreatment,especially MPs in aquatic animals,which must be digested to remove the organic matter on their surface.Furthermore,in these two methods,the samples can only be detected one by one.Therefore,in practical applications,the process of detection will consume a lot of time.The content of HOCs on MPs is often obtained by reverse calculation after detecting the concentration of HOCs in water with chromatographic analysis.This kind of detection method also has the problem that the sample pretreatment process is complicated and difficult to simplify.In view of the problems in the above conventional detection methods,a method for MPs detection in fish intestines base on hyperspectral imaging technology and a quantitative analysis method for HOCs detection on MPs or the water around them which is base on Surface-Enhanced Raman Scattering?SERS?technology are developed,respectively.When detecting MPs in the intestines of fish,a hyperspectral imaging system is used to scan the tiled intestinal contents to obtain hyperspectral data.After performing noise reduction and dimension reduction on the data,a classification model is established to identify MPs in hyperspectral images.In the experiment,in order to eliminate the interference of halation,the samples are dried.The comparison of the samples before and after drying shows that the detection effect of the samples after drying has been improved significantly.In addition,in order to study the effect of particle size,this study divides the particle size range of MPs into three types,and selects the region of interest from them used for classification model training.The results show that the model has the highest accuracy?96.15%?and recall rate?91.85%?when the region of interest is selected in a balanced range within the full range?0.1-1 mm?.Statistics on a large number of detection show that the detection limit of this method is 0.2 mm.In verification experiments,the average accuracy and recall rate of field MPs are similar to those of pure MPs.A total of 5 MPs are found in the verification experiments of 20 wild marine fishes,and there are no false detections or missed detections.In this study,on-site extraction and enhancement of fluoranthene on PE is achieved by placing PE with adsorbed fluoranthene on the prepared AgNPs@SiO₂ substrate and sequentially adding methanol and KI solution.In order to establish the quantitative relationship between the content of fluoranthene on PE and the corresponding SERS signal,the concentration of residual fluoranthene in water samples is detected firstly,and then the content of fluoranthene on PE is obtained by inverse calculation.Next,the prediction model is established through the interval partial least squares regression algorithm.The comparison shows that the determination coefficient of calibration?0.90?and determination coefficient of validation?0.82?of the developed model are the most appropriate when the model is established with the signal of the characteristic peak of fluoranthene at 563,672,789,1022,1267,1414,1456,and 1610 cm^-1.In order to establish the quantitative relationship between the content of fluoranthene on PE and the concentration of fluoranthene in corresponding water samples,multiple models are used to fit the experimental data.The results show that the Freundlich model has the highest determination coefficient?0.855?.Besides,when the content of fluoranthene adsorbed on PE detected by this method is 3.3 ng/g,it can be predicted that the concentration of fluoranthene in the corresponding water sample is 0.97 ng/mL,which indicates that this method can be used to predict the concentration of organic pollutants in water.