ABSTRACT Graphene‐based sorbents have been applied to bisphenol (BP) extraction, yet differences in material design and workflow complicate assessment of their analytical value. This critical narrative review evaluates 34 studies published between 2013 and 2025, identified through searches updated to July 2026, on graphene‐based sorptive extraction before chromatographic determination of BPs in aqueous and food‐related matrices. The analysis covers material design, retention mechanisms, extraction formats, analytical performance, and practical implementation. Graphene oxide was the most common graphene‐derived component, and magnetic solid‐phase extraction accounted for 20 studies. Environmental and drinking waters dominated the applications; BP A was included in every study, whereas individual analogues received less attention. Material engineering aimed to improve phase recovery, surface accessibility, and handling and to introduce additional binding environments. Mechanistic interpretations generally relied on sorbent composition and extraction behavior, with spectroscopic or computational support available for selected systems. Validated quantification of native, migration‐related, or leached BPs provided stronger evidence of real‐sample applicability than fortified recoveries alone. Reported sensitivity, enrichment, and recovery reflected the complete analytical procedure, including sample and extract volumes, detection platform, and matrix complexity. Limited matched comparisons did not establish universal superiority over conventional sorbents. Selectivity was often inferred from target retention, whereas direct comparative cleanup evidence was scarce. Reuse, fabrication reproducibility, carryover, post‐use stability, and workflow greenness were assessed unevenly. Future studies should prioritize controlled benchmarking, application‐relevant selectivity and cleanup assessment, reproducible fabrication, and evaluation of the practicality and greenness of complete sample‐preparation workflows.
ABSTRACT Nanoparticles (NPs) exhibit properties distinct from bulk materials of identical chemical composition, with their behavior and functionality intrinsically linked to parameters such as size, composition, surface functionality, and (surface) charge. Accurate characterization of these parameters is therefore essential for understanding and optimizing NP performance in a variety of applications, including coatings, cosmetics, sunscreens, drug delivery systems, and vaccines. Conventional analytical techniques for NP characterization include microscopy‐, scattering‐, field‐flow fractionation‐, and chromatography‐based methods, which are mainly employed for size determination. Infrared and nuclear magnetic resonance spectroscopy provide ensemble information on chemical composition, while mass spectrometry of NPs larger than 10 nm remains challenging. Among more established techniques, capillary electrophoresis (CE) represents an interesting alternative for NP characterization, due to its inherent selectivity toward size, charge, and surface properties. In this review, literature from the past seven years is surveyed to outline and discuss CE‐based approaches for NP analysis. It is demonstrated that, through appropriate calibration or use of theoretical models, CE can provide size information, (surface) charge, and zeta (ζ) potential. In addition, CE can be applied to assess NP stability, composition, and interactions with, for example, ligands. Importantly, owing to its separation capability, CE can reveal physicochemical heterogeneity within NP populations, rather than only ensemble‐averaged values. Furthermore, the coupling of CE with complementary separation or detection techniques, particularly asymmetric‐flow field‐flow fractionation (AF4) and single‐particle inductively coupled plasma mass spectrometry (SP‐ICP‐MS), offers comprehensive analytical platforms providing orthogonal information on NP size, composition, and charge simultaneously.
Capillary electrophoresis (CE) is an efficient, rapid, and sensitive separation and analysis technique that has been widely applied in biochemistry, pharmaceutical research, and environmental monitoring. However, charge-coupled device (CCD)-based CE fluorescence signals are susceptible to baseline drift, random noise, and channel crosstalk, which reduce the accuracy of peak recognition. To meet the demand for high-precision signal analysis, this study proposes a convolutional neural network (CNN)-bidirectional gated recurrent unit (BiGRU) peak recognition algorithm on the basis of wavelet transform and one-dimensional asymmetric least squares (ALS). In the proposed system, an FPGA-based platform is used for real-time signal acquisition and data transmission, whereas the host computer performs subsequent processing of the received spectral data. Experimental results show that the proposed method achieves high-accuracy peak recognition on the spectral dataset, with Precision, Recall, and F1-score reaching 98.44%, 99.93%, and 99.18%, respectively. These results demonstrate its good robustness and practical applicability, providing technical support for the miniaturization and intelligent development of portable gene sequencers.
Lithium resource serves as critical material for lithium-ion battery industry, and it becomes indispensable for global technological progress and economic development. Facing the challenge of high magnesium-to-lithium ratio in salt lake brines of China, electrodialysis (ED) exhibits great potential as an extraction technology. However, lithium-magnesium separation using ED is a complicated multi-physical process, and insufficient understanding of lithium extraction mechanism limits its efficiency improvement. Hence, the current work constructs a multi-physical model for lithium extraction from high magnesium salt lake brine using ED to reveal ion migration behavior and separation mechanism under synergistic fields. On the basis of a parametric analysis with optimization, the magnesium-to-lithium ratio in brine is dramatically decreased from initial state at 105.05 to 1.87 at the outlet of ED device under the temperature of 323.15 K, potential difference of 0.5 V, and membrane thickness of 0.35 mm. Furthermore, five neural-network models are developed as surrogates of the multi-physical model using 2784 simulation-generated samples without experimental data. Multidimensional dilated convolutional neural network (MDCNN) achieves the lowest in-domain mean absolute error (MAE), root mean square error (RMSE), and mean absolute percentage error (MAPE) of 0.0293%, 0.0395%, and 0.7735%, respectively, whereas R2 (>0.998) provides limited discrimination. In 24 additional out-of-domain simulation cases, MDCNN achieves the lowest MAPE (3.8846%), whereas convolutional attention neural network (CNN-Attention) achieves the lowest RMSE (0.2525) and the highest R2 (0.9494), demonstrating complementary generalization strengths. The current work offers a theoretical basis for optimizing lithium-magnesium separation performance using ED with neural-network surrogate modeling, laying foundation for big data analysis and development of experimental correlations.
ABSTRACT The electrophoretic technique is widely used to analyze and purify charged molecules based on their size, shape, and charge using an electric field for various applications, including DNA fingerprinting, genetic research, disease diagnosis, and drug development. However, electrophoretic data collection, processing, and analysis are hindered by experimental variability, instrument and imaging limitations, data processing difficulties, and analysis inconsistencies, all of which can distort band quality, impede accurate segmentation and quantification, and compromise reproducibility. Recent breakthroughs in data science have led to the development of powerful machine learning techniques, enabling models to learn from data, identify patterns, and make decisions or predictions. These techniques are being used to automate the collection, processing, and analysis of electrophoresis data. This review examines current applications of machine learning in electrophoresis, including the analysis and interpretation of electrophoresis‐derived data, prediction of electrophoretic properties and separation behavior, optimization of operating conditions, and integration with miniaturized sensing and point‐of‐care platforms. Cross‐cutting challenges related to dataset quality and independence, external validation, interpretability, computational requirements, reproducibility, and practical deployment are also discussed.
