Algorithmic cooling is a method devised by Boykin, Mor Rowchodhury, Vatan and Vrijen (PNAS Mar '02) for initializing NMR systems in general and NMR quantum computers in particular. The algorithm recursively employs two steps. The first is an adiabatic entropy compression of the computation qubits of the system. The second step is an isothermal heat transfer from the system to the environment through a set of reset qubits that reach thermal relaxation rapidly. To allow experimental algorithmic cooling, the thermalization time of the reset qubits must be much shorter than the thermalization time of the computation qubits. We investigated the effect of the paramagnetic material Chromium Acetylacetonate on the thermalization times of computation qubits (carbons) and reset qubit (hydrogen). We report here the accomplishment of an improved ratio of the thermalization times from T1(H)/T1(C) of approximately 5 to around 15. The magnetic ions from the Chromium Acetylacetonate interact with the reset qubit reducing their thermalization time, while their effect on the less exposed computation qubits is found to be weaker. An experimental demonstrating of non adiabatic cooling by thermalization and magnetic ion is currently performed by our group based on these results.