**3. Results and Discussion**

#### *3.1. Optimization of Chromatographic Separation*

The essence of the 2D-LC is through the LC1 flow of the target material and the matrix extraction column separation, which removes the impurities, then the elution target substances are enriched in the capture column for a further transfer, and then enter the LC2 flow path for detection through the analysis column. Among them, the interception window width of the LC1 flow path extraction column for the target substance and the LC2 peak separation degree are the key factors determining the transfer recovery rate, peak shape and time length of the two-dimensional analysis. This directly affects the accuracy, sensitivity and detection limit of the two-dimensional liquid chromatography detection method.

Our laboratory has previously established a two-dimensional liquid phase detection method for *Gelsemium* alkaloids in pig plasma, tissue and urine. Referring to the method described by Liu et al., we optimized the mobile phase and time program on this basis [31]. The interception window was 0.7–4.2 min when the mobile phase was an acetonitrile– phosphate solution. Excessive window width would cause a loss in the process of target material transfer, and more impurities would be transferred to the analysis column, resulting in a poor impurity removal effect. Eventually, the transfer recovery rate will be reduced, column efficiency will be reduced and the impurities will affect the target analysis. Only using methanol or acetonitrile and water as a mobile phase could not separate the three target substances well. Therefore, in order to adjust the viscosity and strength of the mobile phase and improve the separation effect and selectivity, we added methanol

based on the original acetonitrile–water as the mobile phase for fine tuning, and the separation degree was improved, but the target peak tailing was relatively serious. Thus, to improve the peak shape, methanol acetonitrile water was used as the mobile phase, and 1.6% phosphoric acid (85%)–1% acetic acid was added to improve the peak shape and reduce tailing. However, the peak time of humantenmine is 0.6 min, which may lead to the failure of the interception of the target substance. Finally, the peak time of humantenmine was delayed by adding auxiliary water. Finally, the LC1 time program was determined to be 0–0.4 min with 1.2 L/min auxiliary water, and the interception window was 1.3–3.2 min. The LC2 mobile phase was adjusted to avoid the overlap of the peak times of the three target substances and to maintain a good separation degree. By adjusting the flow rate ratio of bottle-A (organic phase), bottle-B (water phase—alkali) and bottle-C (water phase—acid) in pump A, the peak time is delayed when the water phase increases, and the increase in the water phase—alkali ratio will lead to wider peak deformation. Therefore, by controlling the ratio of alkali and increasing the ratio of acid, the three target peaks can achieve a better separation and a good peak shape can be obtained. Finally, the flow rate ratio of pump-A was determined to be A:B:C = 27%:10%:63%. The optimized time program is shown in Table 1.

**Table 1.** Time program for 2D-LC detection of three alkaloids.

