In parallel with the recent measurement by Méot et al. [1] of the 239Pu(n, 2n)238Pu reaction cross section in the [7-10] MeV neutron-incident energy range, a modeling work was undertaken to support the new measurement. This theoretical framework is here dedicated to study the impact of the type of level density for the 239Pu residual nucleus (RN) on the final shape of the low-energy tail of the (n, 2n) excitation function. For this purpose, the AVXSF-LNG program [4] has been upgraded to model second-chance reactions and coupled to the TALYS-ECIS06 nuclear reaction codes [3] that are used to provide the compound nucleus (CN) cross section to AVXSF-LNG as a function of the selected pre-equilibrium model . The spin-dependent population of the RN obtained after CN emission is commonly advertised when using an exciton-based model for calculating the total pre-equilibrium cross section. On the other hand a population based on a particle-hole state density is rather expected when using the MSD/MSC quantum mechanical description in place of the exciton-based model. As far as a spectrum of a few number of individual excitations is favored, the combinatorial Quasi-Particle-Vibrational-Rotational Level Density method [4] implemented in the AVXSF-LNG computer program is well suited to model the various alternatives. Several RN level densities have been tested in this work. It includes the testing of the Quasi-particle Random Phase Approximation-based result of Ref. [6]. As a guide line, the amount of pre-equilibrium flux in the total (n,2n) excitation function is being studied as a function of the type of level density selected for the residual nucleus.