Scientific breakthroughs possess the transformative potential to reshape research trajectories and scientific paradigms. However, there is limited systematic evidence on how these breakthroughs influence the evolution of scientific knowledge. Building on the concepts of disruption and consolidation in science, we categorize forward-citing papers into two distinct categories: Disruptive Citing Papers (DCP) and Consolidating Citing Papers (CCP). Analyzing the dynamic patterns of DCP and CCP in Nobel Prize–winning papers, we find that in the early postpublication phase, scientific breakthroughs generate more consolidating citations than disruptive citations. Additionally, CCP in this early phase demonstrate higher scientific impact. However, in the long-term phase, scientific breakthroughs generate more disruptive citations, with DCP often involving larger and more diverse teams. Linguistic analysis also uncovers nuanced differences between CCP and DCP. Furthermore, the dynamic patterns of knowledge flow in scientific breakthroughs differ significantly from control groups. Collectively, our results reveal that scientific breakthroughs initially consolidate knowledge before disrupting it in later phases, offering profound insights into the mechanisms driving scientific progress.