For decades, microelectronic industry has reduced components size while increasing their performance and reliability. Actual systems rely on multilayer nanostructures that require a perfect control of composition at atomic scales. In this study, we focus on detection of germanium contaminants in a thin silicon layer, using High Angle Annular Dark Field (HAADF) with Scanning Transmission Electron Microscopy. We proceeded to multislice calculations in order to extract quantitative information from HAADF measurements. Models consist of Ge substitutional impurities introduced in 10 nm thick Si layers. We analysed atomic column HAADF intensities along 100 zone axis, varying Ge impurities amount and position in an atomic column. Results show that for a reduced number of Ge atoms (n \textless 9), column intensity varies linearly with n. However depending on Ge atoms positions, a column containing n impurities can display a higher signal than a column containing n + 1 impurities. Therefore, a direct correlation between atomic column intensities and their Ge atoms content cannot be established without additional assumptions or complementary measurement.