Reactive transport softwares are today one of the cornerstones of environmental research. They contain multiphysics with very complex algorithms, including flow, transport, chemical and sometimes heat transport, mechanical and/or biological algorithms. Because of this complexity, some parts of these algorithms still have not been sufficiently studied. In this work, we focus on algorithms for activity correction, a specific subset of equilibrium chemistry algorithms. We show that the most used algorithm (the inner fixed-point algorithm) or the most rigorous algorithm (the full Newton) might not be the most efficient, and we propose a new one, the outer fixed-point algorithm, which is more robust and faster than other algorithms.