Grape juices and wines are rich in numerous groups of polyphenolic compounds which require a dedicated separation technique for such complex samples. LC × LC is considered the best technique for the analysis of such samples as it can achieve better resolution and higher peak capacity compared to 1D LC. The ever-growing demand for protecting the environment necessitates reducing or eliminating hazardous solvents to improve the environmental friendliness of analytical procedures. In this study, propylene carbonate was used as an eco-friendly mobile phase component in comprehensive two-dimensional liquid chromatography to analyze phenolic compounds in grape juices and a dealcoholized wine sample. Novel green RPLC × RPLC-DAD and RPLC × RPLC-MS methods were developed for the first time to identify phenolic compounds in five samples (two red grape juice samples, two white grape juice samples, and one dealcoholized wine sample). Four different RPLC × RPLC systems were developed; three systems were connected to a diode array detector (RPLC × RPLC-DAD), while the fourth system was connected to DAD and MS detectors (RPLC × RPLC-DAD-ESI-MS). Solvent X (propylene carbonate:ethanol, 60:40) was adopted as a green organic modifier in the first dimension (1D) and methanol in the second dimension (2D). The practical peak capacity and the surface coverage were calculated as metrics to measure the separation performance of all proposed systems. The orthogonality values for the setups ranged from 0.64 to 0.92 when calculated by the convex hull method, and from 0.54 to 0.80 when calculated by the asterisk equations method. The practical peak capacity production rate ranged from 14.58 to 22.52 peaks/min. The results revealed that the phenolic compounds were separated efficiently with good coverage of the 2D separation space and high peak capacity. A total of 70 phenolic compounds were detected based on MS data and information from the literature.

Humic acid-based carbon dots (HACDs) have excellent properties and are widely used in environmental detection, bioimaging, and optoelectronic materials. Herein, we investigated the structure-activity relationship between the morphology and optical properties of HACDs, and reported on a novel strategy for metronidazole (MNZ) and ornidazole (ONZ) sensing in multiple real samples. It was found that the average particle size decreased from 3.28 to 2.44 nm, optimal emission wavelength was blue-shifted from 500 to 440 nm, and the quantum yield (QY) improved from 5 to 23% with the temperature increasing from 110 to 400 °C. Under the oxidation of hydrogen peroxide (H2O2) and potassium permanganate (KMnO4), the UV-vis spectra of HACD aqueous solution showed time-dependent behavior, and the fluorescence emission of HACDs achieved spectrally tunable multi-color luminescence in the temporal dimension. The surface of HACDs contained a large number of hydroxyl (-OH) and carboxyl (-COOH) fluorophores, resulting in excellent pH sensing. Meanwhile, the synthesized HACDs revealed sensitive response to MNZ and ONZ with the limit of detection (LOD) of 60 nM and 50 nM in aqueous solutions, which had also been successfully applied in various actual samples such as lake water, honey, eggs, and milk with satisfactory results because of the inner filter effect (IFE). Our research is advantageous to enhance the potential applications of HACDs in advanced analytical systems.

According to previous published works, precise modeling of the first dimensional (1D) peak in comprehensive two-dimensional gas chromatography (GC × GC) requires at least 3 modulated sub-peaks (MSP). This requirement is sometimes difficult to meet, e.g., in case of undersampling modulation. In the present work, the feasibility of modeling of the 1D peak with only 2 MSP was demonstrated. The effects of modulation phase (ϕ), modulation period (PM), the peak width (1σ), and the peak shape of the original 1D peak on the accuracy of the proposed method were explored. When employing PM ranging from 6 s ~ 3 s to modulate original peaks with 1σ = 1.2 s ~ 0.6 s, the maximal error of the modeled 1tR is 1.08 s, which is far less than the error generated by employing the largest MSP to estimate the 1tR. The deviation of modeled 1tR increases with the increase of peak shape distortion, and this deviation is ≤ 0.67 s when tailing factor (Tf) in the range of 0.8 to 1.5. The application of the proposed method was demonstrated by assisting identification of a monoterpene in Myrrh sample. The proposed approach could improve the accuracy in calculation of 1tR or 1I and enhance the reliability of compound identification in GC × GC analysis with undersampling modulation.