Sample preparation is one of those fundamental processes in biochemical reactions. In order to obtain target concentrations properly, raw reactants are processed through a series of dilution operations. Since some rare reactants, such as infant blood or DNA evidence from a crime scene, are extremely difficult to acquire, it is important to minimize their consumption and waste during sample preparation. In this paper, we propose a multitarget sample preparation algorithm for reactant minimization on digital microfluidic biochips. Given a set of target concentrations, the proposed method first converts the reactant minimization problem into a network flow model, and then solves it through integer linear programming (ILP) accordingly. Experimental results demonstrate that our new algorithm can reduce the reactant consumption by up to 31% as compared with the current state-of-the-art.