The elimination of memory effect in SIMS (Secondary Ion Mass Spectrometry) has become one of the dominant factors contributing to the reduction of depth profiling accuracy. Most atoms sputtered from the sample by primary ion beam irradiation are deposited on the adjacent mechanical surfaces, especially on the secondary ion extractor. Some of these deposited materials are resputtered by bombardment of energetic ions, such as accelerating secondary ions and primary reflected ions, and deposited again on the analytical sample surface. Therefore, if the elements found in a prior analysis is the same as the elements in the sample to be analyzed, the background signal from the prior analysis will reduce the accuracy of the analysis. This paper gives a detailed quantitative assessment of the elimination of memory effect. In order to reduce the memory effect, the authors evaluated a technique of covering the background source by in-situ sputtering of non-interference materials for several hours prior to analysis. The memory effect was evaluated quantitatively by measuring the detection limit of phosphorus implanted into Si as a function of the number of atomic layer deposited on the top of secondary ion extracting electrode covered with stainless steel (SUS 304). The phosphorus deposited in the prior analysis of a GaP sample was used as the origin of the memory effect. From the experimental results, it was found that the detection limit decreased as a function of the number of atomic layer of the material deposited on the extracting electrode. As a results, the memory effect is eliminated by depositing about 20 atomic layers of the SUS 304.