使用全管柱偵測法探討層析理論板數之計算

Abstract

在液相層析中常使用理論板數(N)來表示其分離效率。然而，計算理論板數的方式常常被誤用。在非線性層析實驗條件下，相較於原始定義以空間為單位計算得到的理論板數，以時間為單位所計算得到的理論板數值常被高估。此外，由於層析波峰壓縮效應(peak compression effect)，樣品在管柱中的波寬會被壓縮變窄。而這種效應也將影響理論板數計算。 在此，我們使用全管柱偵測系統（WCD）觀測樣品在階梯式沖提和連續式沖提中的流動訊號。利用全管柱偵測系統可同時獲得空間和時間的信號，並用來計算理論板高與板數。本研究中有兩種修正方式。首先，由樣品遷移速度概念，將波峰寬度由時間單位轉換為空間單位，以滿足最初的定義。由空間計算的板數作為基準，此修正後的板數大幅降低了誤差，從 1000%降低至40％。第二種修正涉及層析波峰壓縮效應。波寬壓縮發生在兩動相的交界處，當由沖提能力較弱的動相改變沖提能力較強的沖提動相時，因分佈在背部動相(靠近管柱進口端)樣品的遷移速度比在前面動相(靠近管柱出口端)中的樣品快，導致波峰寬度變窄。此效應會造成理論板數增加，第二次修正板數則考量此效應進行修正。在進一步考量波壓縮效應所修正後的板數，其誤差值甚至可低於 10％，。在連續式梯度沖提所提修正板數方程式中，所有的參數皆可自一般層析實驗中，安裝在管柱出口端傳統的單通道探測器得到。

Separation efficiency is normally displayed using the theoretical plate number (N) in liquid chromatography. However, the way how to calculate this number has often been misused. Under non-linear chromatographic processes, the N values computed using chromatograms recorded on the time coordinate have almost always highly over-estimated compared with the values originally defined by the spatial peak profile. In addition, analyte bandwidth in the column would be reduced due to a peak compression effect. This effect will also influence the plate number calculations. A whole-column detection (WCD) system was applied to investigate the flowing analyte signals in the column during the step-wise and linear gradient elutions. Both the spatial and temporal signal data obtained using WCD were used to adjust the equations for calculating the plate number and plate height. Two correction measures were made. First, analyte migration speed was introduced in converting the peak bandwidth of the temporal scale into the spatial scale in order to satisfy the original definition. This adjustment reduced the calculation errors of the temporal data drastically from 1000% to 40%, verified by the spatial data. The second modification involved the peak compression effect. This effect resulted in peak bandwidth narrowing when the elution solvent of the greater solvation power replaced that of the lower solvation power. Solute molecules distributed in the back solvent migrated faster than those in the front solvent, therefore a bandwidth reduction occurred at the solvent boundary. The plate number increase due to this effect was considered in the second modification. This additional correction made the error lower than 10% for all the experiments performed in this study. All the parameters in the equation derived for the linear gradient elution could be obtained experimentally using the conventional single-channel detector installed at the column outlet.