Decomposition Peak Method of Overlapping Chromatographic Peaks Based on Forward-Backward Fitting

Aiming to the chromatograph existing overlapped peaks that impacted the quantitative analysis, the method of overlapped peak resolution based on forward-backward fitting was researched. The method included two procedures: forward fitting and backward fitting. It started the first forward fitting from the point which was far away from the overlapping area, the peak point and back edge of back single peak were modified by the gradual fitting. Then, based on results obtained by the first forward fitting, the first backward fitting was conducted with the similar process and the peak point and front edge of front single peak were also modified. Multiple times of complete forward fitting and backward fitting were implemented with above steps. After multiple iterations, the results continuously approached front edge of the front single peak and back edge of the back single peak. When the calculation error reached to the set value, the iteration was stopped. Then, according to the similarity principle of chromatographic overlapping peaks, the back edge of the front single peak and the front edge of the back single peak were obtained respectively. So the overlapping peak was separated into two single peaks. Because the ratios of peak height, resolutions and tailed factors had great influence on the overlapping chromatographic peaks. The simulation experiments were designed from these three aspects to verify the forward-backward fitting method. At the same time, in order to further validate the effectiveness of the forward-backward fitting algorithm, separation experiments of p-xylene and m-xylene measured by gas chromatography mass spectrometry were designed. At last, the forward-backward fitting method with perpendicular-drop method, intersection vertical method and proportional distribution method were compared. The results suggested that the error of the forward-backward fitting method was less than 1.8%, while the maximal error of other three methods reached to 29.92%. In addition, from the point of view of the processes of four methods, the proposed method didn’t rely on two maximums of overlapped peaks and could handle various types of the overlapping peaks with a higher accuracy. The proposed method can counterbalance the shortcomings of three methods. Therefore, with wider applicable peak shapes and higher accuracy, the forward backward method has certain quantitative analysis advantages and can be better applied to quantitative analysis of complex components.