As the real samples contain large amount of interferents, such as nucleic acid, protein, fat, etc., which would ruin the separation and detection, in this section, kinds of organic solvents, including methanol, acetonitrile, and acetone (the percent of organic solvent was set to 70%) as well as trichloroacetic acid (TCA), were tested for their abilities to reduce the possible interference. Our data showed that all these four sample treatment protocols could perform satisfied cleaning results. However, the recoveries (present via peak heights) obtained from these treatment methods differed (Figure 1, the shown data throughout this paper was obtained based on average of at least three parallel samples). It could be seen that, acetone treated sample showed the highest peak height among all the selected sample treatment methods and was regarded as the best one for further usage.

In this section, the ratios between acetone and real sample were investigated in a certain range (from 30% to 70%, percent of acetone in total volume). The electropherograms were shown in Figure 2. It could be seen that when the percent of acetone was less than 50%, the MEL could not be separated from interferents (arrow pointed in Figure 1); on the contrary, when it was larger than 50%, interferents were well separated from MEL. For the selection of acetone percent, a 60% was considered as the best one which maintained better Rs (3.57) and highest theoretical plate number (N) value (123,000) and less dilution to samples.

The pH value of run buffer was one of the important parameters that affect the separation and determination efficiency, as it influenced electro-osmotic flow (EOF) and the status of charged electrical species. Yan et al investigated the influence of pH in the range of 2.4 to 4.0, however, it's not adequate, as MEL contains three amino groups and the pKa values is around 5.0 - 5.5. In our works, a wider pH value ranging from 2.4 to 6.0 was carefully studied (Figure 3). It could be seen that, in pH range of 2.7 - 3.8, MEL (# in Figure 3) could be well separated from the interferent (* in Figure 3). It also could be seen that the migration times were similar to each other which may due to the slightly increase of EOF and decreased of positive charges of MEL when the pH increased. Furthermore, a change of peak order between MEL and the interferent was firstly observed in our works when the pH values went upper than 4.0 (at pH3.8, the peak of interferent appeared at the right side of MEL peak, but at pH4.5, it moved to the left side). For selection of best pH, the resolution (Rs, between MEL and interferent) combined with N values were evaluated. Our data (not shown) indicted that the Rs values went up as the pH values increased from 2.4 to 3.1, and went down when the pH increased continuously. Finally, a buffer pH of 3.1 was selected for its better N value (~115,000 ± 11,400, n = 3) and largest Rs value (3.71).

The viscosity of the buffer and the thickness of the electric double layer were strongly related to buffer concentration, as a result, buffer concentration could affect EOF and hence the separation result. Normally, for selection of buffer concentration, the lowest concentration which provided a good separation was preferred, as the high concentration buffer may cause extra Joule heat generation, which in turn, caused peak broaden and decreasing in separation efficiency. In this section, a concentration range of 10 mmol/L to 100 mmol/L phosphate buffer at pH3.1 was investigated. Our works indicated that when the buffer concentration was higher than 10 mmol/L better separation could be easily obtained and when the buffer concentration was higher than 80 mmol/L, the current was too high (>200 μA) to maintain good separation (data not shown). The influences of buffer concentrations on migration times and N values were shown in Figure 4. It could be seen that the migration times increased while the buffer concentration became lager, (resulted in decreased EOF); And that the N values got its maximum value when the concentration set to 60 mmol/L (higher concentration led to higher Joule heat generation and poor efficiency), based on this, a buffer concentration of 60 mmol/L was selected for further studies.

In theory, higher voltage reduces analysis time, however, it may lead to significant losses of resolution and peak efficiencies if excessive Joule heat generated within the capillary. Based on this, the choice of separation voltage should be optimized with considering total analytical times, separation efficiency and other factors, like buffer

concentration etc, so that an acceptable level of current could be controlled. In this section, several voltages were investigated and the results were shown in Figure 5. As was shown, N values dropped while the voltages increased, especially when the voltages became larger that 17 kV (the current reached 180 μA). It’s also could be seen that the migration times dropped with increasing separation voltages and the total analytical times shortened, however, it’s at the expense of decreased N values and Rs (data not shown). A voltage of 17 kV was finally selected as its maintained shorter analysis time and satisfied N values (~155,400).