The intensity ratios I(L)/I(K) of eight kinds of pure element's thin films from Ge to Sn have been measured in a TEM using X-ray EDS at 40, 80, 100, 120,160 and 200 kV. Using these I(L)/I(K) values, the Cliff-Lorimer factor KXSi and their interpolated values for the K lines, the KXSi values for L lines are obtained. In order to determine which Q ionization cross-section formula in literature is the best, the calculated values I(L)/I(K) using nine formulas of Q were compared with expermental data. It has been found that there are large discrepancies between the expermental data and calculated values. After considering the errors by various parameters used in calculation, it becomes evident that the discrepancy is mainly caused by inaccuacy of cross-section and the best formula of Q proposed by Fabre de la Ripelle must also be revised. The quantitative analysis for several specimens with known composition was done using our revised Q and EDAX-9100's progroms. The results snow that by using our revised formula of Q, the errors of analytical results from K-K and K-L lines at 100 and 200 kV could be reduced significantly.