Abstract: Using the Ensemble Empirical Mode Decomposition (EEMD) method, we accurately extracted the varying annual cycle of SPV (stratospheric polar vortex) from the ERA5 reanalysis dataset and the NOAA SIC (sea ice concentration) data. We further identified key sea ice regions associated with interannual changes in the build-up and break-up date of the SPV, and investigated how autumn sea ice in the key region influences SPV evolution through thermal and dynamic processes. The results show significant interannual variations in the SPV build-up and break-up dates, and the variation in break-up dates is more pronounced. Years with anomalously late (early) SPV build-up and early (late) break-up tend to have an anomalously strong (weak) SPV in mid-winter. Late SPV build-up is linked to warm temperature anomalies in the polar stratosphere from late winter to early spring, creating favorable conditions for early SPV break-up events. Further investigation showed a positive correlation between the October SIC in the Barents Sea and SPV build-up date, likely due to its influence on planetary wave modulation and reflected shortwave radiation in autumn. However, the statistical significance was relatively low, indicating the limited role of sea ice in the SPV build-up. Conversely, October SICs in the East Siberian Sea are significantly negatively correlated with the SPV break-up date. The primary mechanism behind this relationship is that an anomalously low SIC in the East Siberian Sea suppresses the upward propagation of planetary waves, leading to Eliassen-Palm (E-P) flux divergence anomalies and stronger stratospheric circumpolar westerlies. As such, the SPV break-up is delayed and dominated by slow diabatic processes. In contrast, a high SIC enhances planetary wave propagation, resulting in E-P flux convergence, weaker circumpolar westerlies, and an earlier SPV break up due to the rapid dynamic processes.