In this paper, we propose a new 2-Dimensional Underwater Barrier Deployment Algorithm (2D-UBDA) ensuring the barrier detection of toxic substances in a river. Our objective is to guarantee a full detection of chemical pollutant sources, while minimizing the deployment cost. To achieve this, first 2D-UBDA determines the potential deployment areas within a predefined target field installation and this, for each pollution source, by using a 3D-propagation model of a substance to predict its molarity in any point within the river. Then, based on an integer linear programming algorithm, 2D-UBDA selects the minimum number of sub-areas in which chemical sensors will be deployed by taking into consideration the intersections between the potential deployment zones of all pollution sources located upstream of the target field installation. To validate our proposal, the Pamplonita river located in Amazon rainforest is used as a case of study. Based on extensive simulations, 2D-UBDA outperforms the basic deployment strategies in terms of number of chemical sensors and successful detection of pollutant.