Mining data in massive databases to find useful data, frequent patterns (FP) and knowledge discovery, has been studied popularly in data mining research fields minding governments, industries and sales companies. In data mining, frequent patterns as growth tree are an efficient method for discovering knowledge and compressing information in a tree structure. Previous studies presented various methods to achieve frequent patterns which even require complex process and costs, particularly if the patterns overnumbered. In this paper we provide a divide and conquer algorithm based on FP-Growth Tree to create an initially sorted tree structure of the nodes that the most frequent patterns would be available at every moment of the tree construction procedure. In addition, we can consider a parameter to avoid inserting less frequent branches, thus, the tree would be sorted from the beginning of its formation.