Abstract
A new method for the analysis of four target flavonoids in two kinds of citrus samples by ultra-high performance supercritical fluid chromatography (UHPSFC) method was developed. Main variables affecting the UHPSFC separation were optimized, and under the optimized conditions the four target compounds (tangeretin, nobiletin, hesperetin and naringenin) can be separated within 10 min. The UHPSFC method allowed the determination of the four target compounds in the diluted stock solutions with limit of detection (LOD) ranging from 1.08 to 2.28 μg mL−1, and limit of quantification (LOQ) ranging from 1.45 to 4.52 μg mL−1, respectively. The coefficients of determination (R 2) of the calibration curves were higher than 0.9950. The recoveries of the four target compounds at three different concentrations were in the range of 82.4–117.6%. The validation results demonstrated that the proposed method is simple, accurate, time-saving and environment friendly, and it is applicable to a variety of complex samples such as medicine-food dual purpose herbs and functional foods.
1 Introduction
Citrus is an important crop worldwide and is mainly used in the food industry for its fresh juice and pericarp, because citrus and their peel are important source of bioactive compounds that are important to human nutrition [1], these major active compounds include flavonoids, coumarins, alkaloids and limonoids [2]. Citrus peel was the important products of citrus sources and has a long application history in China and other countries due to the high medicinal and nutritional values. A number of citrus peel has been officially recorded in Chinese Pharmacopeia, such as Citri reticulatae pericarpium (chenpi in Chinese) and Aurantii fructus (zhike in Chinese). Due to their benefit chemical components and multiple bioactivities, chenpi can be used as medicine-food dual purpose herb and zhike can be used as functional foods [3].
In the past decades, due to their numerous bioactivities, the research on flavonoids has greatly expanded. A number of methods for the separation and analysis of flavonoids in citrus peel have been established and reported, including high-performance liquid chromatography (HPLC) [4], capillary electrophoresis with electrochemical detection (CE-ED) [5], gas chromatography-mass spectrometry (GC-MS) [6]. In recent years, ultra-high performance supercritical fluid chromatography (UHPSFC) technique is arousing a growing interest. The technique uses compressed carbon dioxide (CO2) mixed with a certain amount of organic solvent as mobile phases, which allows for higher linear velocities due to faster diffusion and lower viscosities. As a powerful and environment-friendly technique, it has attracted increasing interest from researchers and has been successfully applied for natural product metabolomics analysis [7], food analysis [8], pharmaceutical analysis [9], clinically samples analysis [10] and so on.
Literature reports only a few UHPSFC method for the separation and quantitative determination of flavonoids in citrus samples. Morin [11] determined five polymethoxylated flavones (sinensetin, nobiletin, tangeretin, heptamethoxyflavone and 5,6,7,4ˊ-Tetramethoxyflavone) in citrus oils by a laboratory-made SFC equipment; Dugo et al. [12] proposed a SFC method for rapid analysis of six polymethoxylated flavones (tangeretin, heptamethoxyflavone, nobiletin, 5,6,7,4ˊ-Tetramethoxyflavone, hexamethoxyflavone and sinensetin) from citrus oils; Li et al. [13] developed an efficient method for separation four polymethoxyflavones (nobiletin, tangeretin, 3,5,6,7,8,3′,4′-heptamethoxyflavone and 5,6,7,4′-tetraoxyflavone) from sweet orange (Citrus sinensis) peel; Jiang et al. [14] developed a SFC method for analysis of six flavonoids (tangeretin, 3,5,6,7,8,3′,4′-heptamethoxyflavone, nobiletin, sinensetin, didymin and hesperidin) in Citri Reticulatae Pericarpium. Although, these researches reported the efficient SFC methods for separation different flavonoids in citrus samples, there are a wide range of flavonoids and real samples that contain flavonoids, therefore, it is still interesting and necessary to investigate the fast and accurate methods to the analysis of flavonoids in different samples.
The aim of the present work was to develop and optimize a simple and fast UHPSFC method for the determination of four flavonoids target compounds (tangeretin, nobiletin, hesperetin and naringenin) in chenpi and zhike samples. To the best of our knowledge, there is no report of the use of UHPSFC for the determination of the four flavonoids in the above samples. During method development, different chromatographic conditions such as stationary phases, additives in organic solvent and injection volume were tested. After the optimization of the separation conditions, the complete separations were obtained and the developed method was validated. This study might provide a feasible and simple method for quantitative determination of multiple target compounds in citrus samples.
2 Materials and methods
2.1 Chemicals and materials
Tangeretin, Nobiletin, Hesperetin and Naringenin were used as reference standards, and the chemical structures of the four target compounds were listed in Fig. 1. Tangeretin (purity higher than 98%), Hesperetin (purity higher than 98%) and Naringenin (purity higher than 98%) were obtained from Chendu Must Bio-technology Co., LTD (Chendu, China), Nobiletin (purity higher than 98%) was obtained from Meilunbio. Chenpi and Zhike samples were purchased from local drug stores (Jiangmen, China). HPLC-grade methanol was obtained from Fisher, and formic acid was purchased from Macklin. Food grade carbon dioxide for the UHPSFC was purchased from a local company (Jiangmen, China), and the purity is higher than 99.999%.
The Chemical structures of the four target compounds considered in this work
Citation: Acta Chromatographica 2022; 10.1556/1326.2022.01010
The Chemical structures of the four target compounds considered in this work
Citation: Acta Chromatographica 2022; 10.1556/1326.2022.01010
The Chemical structures of the four target compounds considered in this work
Citation: Acta Chromatographica 2022; 10.1556/1326.2022.01010
2.2 Standard solutions and sample preparation
The stock solutions of the four standards were prepared in absolute methanol. Then, the work solutions were freshly prepared by diluting proper amounts of the above stock solutions with methanol to determine the calibration curves. Each analyte in mixed standard solutions and real samples was identified by comparison between the retention time of the peaks of the standard reference compounds and UV absorption value.
The chenpi and zhike sample (0.4 g) was heating reflux extracted with 40 mL 50% ethanol for 1 h, then the sample was cooled at room temperature and the supernatant was transferred to a 50 mL volumetric flask. For the final analysis, 3 mL of each solution was transferred to a 5 mL volumetric flask.
Each solution was filtered through a 0.22 µm nylon membrane prior to analyze and was measured three times in parallel, then the average peak area was used for the further analysis.
2.3 Equipment
The separation and determination of the four target compounds were performed on a Waters Acquity Ultra Performance Convergence Chromatography (UPC2) system that equipped with a convergence chromatography manager, a sample manager, a binary solvent manger, a column heater manager and a photodiode-array detector. Data acquisition and processing was performed by using the Empower 3 program. The mobile phase A was CO2 and phase B was organic solution.
Six columns were evaluated in this experiment, the names and known properties of the columns used are presented in Table 1. Among the six columns, Trefoil CEL1 column is a chiral column and the others are achiral columns.
Six columns employed in the present study
| Column name | Particles | Bonded ligand | Dimensions | Particle size |
| Viridis BEH column | Fully porous particles | Bare hybrid silica | 3.0 × 100 mm | 1.7 μm |
| Trefoil CEL1 column | Chiral stationary phase | Functionalized celluloses | 3.0 × 150 mm | 2.5 μm |
| Torus 2-PIC column | Fully porous particles | 2-Picolyl-amine | 2.1 × 150 mm | 1.7 μm |
| Torus Diol column | Fully porous particles | Propanediol | 2.1 × 150 mm | 1.7 μm |
| Viridis BEH-2P column | Fully porous particles | 2-Ethylpyridine | 2.1 × 150 mm | 1.7 μm |
| Virids CSH Fluoro-Phenyl | Fully porous particles | Pentafluorophenyl | 2.1 × 150 mm, | 1.7 μm |
2.4 Method validation procedures
The developed method was validated in terms of linearity, limits of detection (LODs), limits of quantification (LOQs), precision and accuracy. Calibration curves were obtained by running the mixed standards of different concentrations in triplicate, and the correlation coefficient was obtained by a linear regression model. The regression equation form of the calibration curves is y=ax+b, where y is the peak area and x is the concentrations of the analytes. The LOD and LOQ for each compound were estimated to have signal-to-noise ratios (S/N) of 3 and 10, respectively.
Relative standard deviations (RSD) were calculated as a measure of precision. The intra-day and inter-day precision were employed to study the repeatability and reproducibility of the developed method. For the intra-day precision tests, a mixed standard sample was analyzed six times over a single day and for the inter-day tests, the same sample was analyzed over three consecutive days. Accuracy was determined by spiking chenpi and zhike samples with different concentrations (low, medium and high spike) of the four target compounds. The recovery percentage (%) for the four target compound was calculated using the following equation: (Detected amount-Original amount)/Spiked amount × 100 [15].
3 Results and discussion
3.1 Method optimization
3.1.1 Column screening
In the UHPSFC analysis, the low polarity of CO2 as a mobile phase limits its application. Therefore, an organic modifier is generally added to enhance elution strength. Methanol is commonly used organic modifier due to its complete miscibility with CO2 over a wide range of temperatures and pressures [16]. Therefore, in this study, methanol was employed as the modifier at first.
With the aim of acquiring optimal separation conditions with a short analysis time, six columns with various types of stationary phases described in Section 2.1 were tested. In consideration of the gradient program (time and flow-rate) should have been adapted between the six columns to provide meaningful comparison, the initial chromatographic conditions for column screenings were set as listed in Table 2. The oven temperature was kept at 40 °C, the back pressure was set at 2000 psi and the injection volume was 1 µL. The wavelength range of the PDA detector was 210–400 nm, and the detection wavelength was set at 280 nm for analysis after the comprehensive consideration of the maximum absorption of the four standards.
Chromatographic elution gradient programs of different columns for UHPSFC analysis
| Viridis BEH | Trefoil CEL1 | Torus 2-PIC Torus BEH 2-EP Torus Diol |
Virids CSH Fluoro-Phenyl | ||||||||
| Time (min) | Flow rate (mL min−1) | B (%) | Time (min) | Flow rate (mL min−1) | B (%) | Time (min) | Flow rate (mL min−1) | B (%) | Time (min) | Flow rate (mL min−1) | B (%) |
| Initial | 1.0 | 3.0 | Initial | 1.0 | 10.0 | Initial | 0.6 | 3.0 | Initial | 0.5 | 3.0 |
| 10 | 1.0 | 8.0 | 15 | 1.0 | 40.0 | 10 | 0.6 | 8.0 | 10 | 0.5 | 8.0 |
| 18 | 1.0 | 30.0 | 16 | 1.0 | 10.0 | 18 | 0.6 | 30.0 | 12 | 0.5 | 20.0 |
| 19 | 1.0 | 3.0 | 17 | 1.0 | 10.0 | 19 | 0.6 | 3.0 | 40 | 0.5 | 45.0 |
| 20 | 1.0 | 3.0 | 20 | 0.6 | 3.0 | 41 | 0.5 | 3.0 | |||
| 42 | 0.5 | 3.0 | |||||||||
The chromatograms in Fig. 2 show the separation of the four target compounds on six columns under the initial chromatographic conditions. As can be seen from Fig. 2, the retention times obtained from 2-PIC, Diol and BEH-2P were longer than those obtained using other columns, this is because the column is longer and the particle size is smaller. Considering the pressure that the instrument can withstand when these three columns were used, lower flow rate was set. Since the chemistry of different stationary phases had specific selectivity to the target compounds, the peaks shapes and retention times of the target compounds varied between different columns. As illustrated in Fig. 2, the good separation efficiency and the shorter retention time was achieved in the chromatogram obtained by using BEH column. On the CEL1 column, two peaks were presented for naringenin (peak 4, Fig. 2B). Figure 2C–E showed that on 2-PIC, Diol, and BEH 2-EP column, the peak shapes of tangeretin (peak 1) and nobiletin (peak 2) were poor. On CSH Fluoro-Phenyl column (Fig. 2F), hesperetin (peak 3) and naringenin (peak 4) have poor separation resolution and peak shapes were extremely poor. Therefore, the results of the above comparison show that the BEH column was the most suitable for the separation of the four target compounds as there was relatively high resolution with a shorter analysis time, and the BEH column was applied for the following experiments.
Chromatograms and elution order of target compounds on six tested columns. Analytes: 1- Tangeretin, 2- Nobiletin, 3- Hesperetin, and 4- Naringenin
Citation: Acta Chromatographica 2022; 10.1556/1326.2022.01010
Chromatograms and elution order of target compounds on six tested columns. Analytes: 1- Tangeretin, 2- Nobiletin, 3- Hesperetin, and 4- Naringenin
Citation: Acta Chromatographica 2022; 10.1556/1326.2022.01010
Chromatograms and elution order of target compounds on six tested columns. Analytes: 1- Tangeretin, 2- Nobiletin, 3- Hesperetin, and 4- Naringenin
Citation: Acta Chromatographica 2022; 10.1556/1326.2022.01010
3.1.2 Determination of organic modifier
After the Viridis BEH column was determined as the optimal column, the organic modifier should be investigated. In supercritical fluid chromatography, a small amount of organic solvent is often added to CO2 to modify the chromatographic retention, selectively, peak shape and sample solubility [17]. Methanol is the most polar alcohol and lead to the higher polarity of the bulk mobile phase, resulting in lower retention of most analytes [18]. Therefore, in the present study, methanol has been employed firstly. What makes us happy is that an acceptable separation can be obtained using methanol, therefore, the methanol was determined as the organic modifier and the other organic solvents were not tried.
3.1.3 Effect of additives in the modifier
As can be seen in Fig. 2A, the peak shapes of 3-hesperetin and 4-naringenin were not symmetrical. Therefore, in order to obtain good peak shapes, the influence of the percent of formic acid as additives was examined. In the chromatographic experiments, due to experimental variations, such as small variations in mobile phase composition and other unknown instrument shifts can cause the retention time shifts. In the present study, there is also a small retention time shift. In order to observe the effect of additive on the peak shapes more directly, the Interval Correlation Shifting algorithm (icoshift) method was employed to align the retention time shifts [19]. The aligned results were illustrated in Fig. 3.
Comparison of chromatograms obtained using different percent of formic acid additive in the methanol on BEH column. (A) the retention time shifts aligned chromatograms obtained using different percent of formic acid; (B) the enlarged view of A-1 region; (C) the enlarged view of A-2 region. Analytes: 1- Tangeretin, 2- Nobiletin, 3- Hesperetin, and 4- Naringenin
Citation: Acta Chromatographica 2022; 10.1556/1326.2022.01010
Comparison of chromatograms obtained using different percent of formic acid additive in the methanol on BEH column. (A) the retention time shifts aligned chromatograms obtained using different percent of formic acid; (B) the enlarged view of A-1 region; (C) the enlarged view of A-2 region. Analytes: 1- Tangeretin, 2- Nobiletin, 3- Hesperetin, and 4- Naringenin
Citation: Acta Chromatographica 2022; 10.1556/1326.2022.01010
Comparison of chromatograms obtained using different percent of formic acid additive in the methanol on BEH column. (A) the retention time shifts aligned chromatograms obtained using different percent of formic acid; (B) the enlarged view of A-1 region; (C) the enlarged view of A-2 region. Analytes: 1- Tangeretin, 2- Nobiletin, 3- Hesperetin, and 4- Naringenin
Citation: Acta Chromatographica 2022; 10.1556/1326.2022.01010
As can be seen from Fig. 3A, the addition of formic acid has barely effect on the peak shapes of 1-tangeretin and 2-nobiletin, probably because a minor change of the pH environment could not give rise to alter their ionization state [20]. As can be seen from Fig. 3B (the enlarged view of Fig. A-1 region) and Fig. 3C (the enlarged view of Fig. A-2 region), with the increasing of the percentage of formic acid, the peak shapes of 3-hesperetin and 4-naringenin have been improved to a certain extent. Considering that if the percentage of the formic acid continues to increase, the polarity and acidity of the mobile phase will be changed significantly thus affecting the overall analysis effect [21], in the present study, 1% formic acid has been determined.
3.1.4 Effect of injection volume
Under the same conditions, a higher response can be obtained using a larger injection volume. Therefore, the injection volume of 2.0 µL was characterized firstly. When the injection volume was 2.0 µL, the peaks of tangeretin and nobiletin are no longer sharp (chromatograms were not showed here). Therefore, in the presented study, the injection volume was determined at 1.0 µL.
After the above optimization, the initial chromatographic condition was further optimized to achieve a faster analysis purpose. The final conditions were as follows: CO2 (A) and 1% formic acid in methanol (B), the gradient elution was programmed as follows: 3% B at 0 min, 6% B at 7 min, 35% B at 12 min and held at this composition for 1 min, 3% B at 14 min and then the column was equilibrated for 1 min under the composition. Flow rate, column temperature and back pressure were set to 1.0 mL min−1, 40 °C and 2000 psi, and the injection volume was 1 µL.
3.2 Method validation
The UHPSFC conditions described in Section 3.1 were employed, and the newly proposed analytical method was validated by linearity, LOD, LOQ, precision and recovery.
As illustrated in Table 3, under the optimized UHPSFC method, 13 mixed standard solutions were analyzed, the calibration curve showed good linearity of 0.9982, 0.9985, 0.9972 and 0.9950 for tangeretin, nobiletin, hesperetin and naringenin, respectively within the corresponding concentration range, which showed quite a linear correlation for each target compound. In order to easily check the performance of the established calibration curves, the relationship between the concentrations and average areas of the four target compounds were illustrated in Fig. 4.
Calibration curves, LODs, LOQs and precision of the UHPSFC method
| Compound | Linearity | Precision (RSD%) | |||||||
| Calibration curve | R 2 | Range (µg mL−1) | LOD (µg mL−1) |
LOQ (µg mL−1) |
Inter-day (n = 6) | Intra-day (n = 6) | |||
| Peak area | Retention time | Peak area | Retention time | ||||||
| Tangeretin | y = 2866.3x–4119.4 | 0.9982 | 5.42–162.72 | 1.08 | 1.45 | 2.10 | 0.38 | 3.65 | 0.89 |
| Nobiletin | y = 2089.8x–2146.6 | 0.9985 | 4.13–110.08 | 1.10 | 1.38 | 1.16 | 0.24 | 5.06 | 1.26 |
| Hesperetin | y = 2826.8x–9805.2 | 0.9972 | 4.92–147.60 | 2.28 | 3.92 | 2.90 | 0.06 | 5.27 | 1.04 |
| Naringenin | y = 3064x–10575 | 0.9950 | 5.52–165.60 | 1.84 | 4.52 | 2.72 | 0.19 | 3.58 | 0.62 |
The relationship between the concentrations and average areas of the four target compounds
Citation: Acta Chromatographica 2022; 10.1556/1326.2022.01010
The relationship between the concentrations and average areas of the four target compounds
Citation: Acta Chromatographica 2022; 10.1556/1326.2022.01010
The relationship between the concentrations and average areas of the four target compounds
Citation: Acta Chromatographica 2022; 10.1556/1326.2022.01010
The LOD and LOQ were determined using diluted stock solutions at S/N of 3 and 10, respectively. The LOD and LOQ were in the range of 1.08–2.28 μg mL−1 and 1.38–4.52 μg mL−1.
The variations including retention times and peak areas were carried out by RSD. As can be seen, the RSD values of peak area and retention time of each analyte were less than 2.90 and 0.38%, 5.27 and 1.26% for inter-day and intra-day, respectively. The obtained values confirmed the precision of the UHPSFC assay method.
Recovery test was carried out to investigate the accuracy of the proposed method, as illustrated in Table 4, the recovery of the developed method was acceptable with good accuracy in the range of 82.4–117.6%.
Recovery (%) of the UHPSFC assay method
| Sample | Spike | Analytes | |||
| Tangeretin | Nobiletin | Hesperetin | Naringenin | ||
| Chenpi | Low | 87.8 | 97.6 | 93.8 | 83.1 |
| Medium | 98.3 | 94.1 | 87.8 | 98.0 | |
| High | 94.8 | 87.5 | 117.6 | 96.9 | |
| Zhike | Low | 90.1 | 95.4 | 82.4 | 102.9 |
| Medium | 99.7 | 98.8 | 91.8 | 101.2 | |
| High | 93.3 | 96.1 | 116.5 | 90.7 | |
3.3 Application of the established method
The validated method was applied to determine the content of the four target compounds in chenpi and zhike samples. The contents of the four target compounds were not given here, the reason is that we did not optimize the extraction conditions, it is of little value to show the content values. To the best of our knowledge, there is no report of use of UHPSFC method for simultaneous determination of the four target compounds. Literature research yields some analytical methods for the separation and quantitative analysis of active compounds in citrus peel. Among these analytical methods, the most reported is HPLC method, for example, Wang et al. [22] proposed LC method for quantitative analysis of polymethoxyflavones in citrus peel extracts, Tan et al. [23] proposed HPLC method for simultaneous quantitative analysis of multiple constituents in different Chinese medicinal materials from Citrus genus, Xu et al. [24] proposed a HPLC-DAD method to obtain chemical fingerprint for quality control of Fructus Aurantii Immaturus, and many others not listed here. Compared with these HPLC methods, the proposed UHPSFC method is faster and needs less consumption of organic solvent. Therefore, combining the analytical speed and the validation data, the proposed UHPSFC method has the advantages of being faster and greener.
4 Conclusion
In the present study, a rapid and environment-friendly UHPSFC method was developed, optimized and validated for the separation and simultaneous determination of four target compounds in citrus samples for the first time. The ability of different columns, organic solvent, additives and injection volume led to rapid and efficient separation of the four target compounds was evaluated, and the optimization separation condition was obtained. The method was validated and showed good linearity and precision for the target compound. The study indicates that the proposed method has high separation efficiency and might have a prospective future in other samples, such as food samples, environmental samples, and biological samples and so on.
Conflict of interest
The authors declare there is no conflict of interest associated with this publication.
Acknowledgements
The authors gratefully acknowledge the financial support from the Project of Young innovative talents in Colleges and Universities in Guangdong Province (No: 2021KQNCX101), Jiangmen Program for Innovative Research Team (No: 2018630100180019806) and the Jiangmen City Science and Technology Basic Research Project (2020030102060005412) and the team project of Wuyi University (2019td09).
References
- 1.↑
Tang, X. ; Zhao, H. ; Jiang, W. ; Zhang, S. ; Guo, S. ; Gao, X. ; Yang, P. ; Shi, L. ; Liu, L. Food Funct. 2018, 9, 5880–5890.
- 3.↑
He, Y. J. ; Zhu, M. ; Zhou, Y. ; Zhao, K. H. ; Zhou, J. L. ; Qi, Z. H. ; Zhu, Y. Y. ; Wang, Z. J. ; Xie, T. Z. ; Tang, Q. ; Wang, Y. F. ; Luo, X. D. J. Sep. Sci. 2020, 43, 3349–3358.
- 6.↑
Duan, L. ; Guo, L. ; Dou, L. L. ; Zhou, C. L. ; Xu, F. G. ; Zheng, G. D. ; Li, P. ; Liu, E. H. Food Chem. 2016, 212, 123–127.
- 7.↑
Perrenoud, A. G. -G. ; Guillarme, D. ; Boccard, J. ; Veuthey, J. -L. ; Barron, D. ; Moco, S. J. Chromatogr. A. 2016, 1450, 101–111.
- 9.↑
Desfontaine, V. ; Guillarme, D. ; Francotte, E. ; Novakova, L. J. Pharm. Biomed. Anal. 2015, 113, 56–71.
- 10.↑
Storbeck, K. H. ; Gilligan, L. ; Jenkinson, C. ; Baranowski, E. S. ; Quanson, J. L. ; Arlt, W. ; Taylor, A. E. J. Chromatogr. B 2018, 1085, 36–41.
- 12.↑
Dugo, P. ; Mondello, L. ; Dugo, G. ; Heaton, D. M. ; Bartle, K. D. ; Clifford, A. A. ; Myers, P. J. Agricul. Food Chem. 1996, 44, 3900–3905.
- 14.↑
Jiang, Z. M. ; Wang, L. J. ; Liu, W. J. ; Wang, H. Y. ; Xiao, P. T. ; Zhou, P. ; Bi, Z. M. ; Liu, E. H. J. Chromatogr. B 2019, 1133, 121845.
- 15.↑
Xu, R. ; Chen, X. ; Wang, X. ; Yu, L. ; Zhao, W. ; Ba, Y. ; Wu, X. Food Chem. 2019, 298, 125067.
- 16.↑
Plachka, K. ; Chrenkova, L. ; Dousa, M. ; Novakova, L. J. Pharm. Biomed. Anal. 2016, 125, 376–384.
- 17.↑
Huang, Y. ; Feng, Y. ; Tang, G. ; Li, M. ; Zhang, T. ; Fillet, M. ; Crommen, J. ; Jiang, Z. J. Pharm. Biomed. Anal. 2017, 140, 384–391.
- 23.↑
Tian, F. ; He, X. F. ; Sun, J. ; Liu, X. D. ; Zhang, Y. ; Cao, H. ; Wu, M. H. ; Ma, Z. G. J. Sep. Sci. 2020, 43, 736–747.
