The depth of the imaginary part of the optical potential is derived from the assumption that, at a given energy and for each partial wave L, it is proportional to the compound nucleus density level up to a given excitation energy above the yrast level corresponding to the angular momentum L, and remains a constant for smaller values of L. The prescription is successfully tested for the system 160 + 28Si at nine different projectile energies between 33 and 81 MeV; it fails however at 141.5 MeV, as expected, because other channels, besides elastic scattering and fusion, are important.In heavy-ion physics the large decrease of the elastic cross section at energies where the nuclei begin to penetrate each other points to a large absorption out of the elastic channel. This is usually described by means of the imaginary potential in the optical model. During the last decade much effort has been made to derive this potential from different approaches [l-4] . In this paper, following a suggestion by Arima and Hodgson [5], we present a method to construct the imaginary part of the optical potential. This is inspired, in part, in the model proposed by Helling et al. [6] where that potential is given by the transition probability from the elastic channel into a precompound state which is a doorway state for the formation of the compound nucleus. This transition probability is given in first order by Fermi's golden rule:where p(E* , L) is the level density of the precompound nucleus with excitation energy E* and angular momentum L. ~iin+ is the interaction between the pre-_-. I compound state Gcomp and the elastic channel +elas. To estimate the magnitude of the transition matrix element it is necessary to introduce microscopic wave functions. Because the nucleons in the overlap region contribute most to the transition probability, Fink et al. [7] assume that the radial dependence of the square of the matrix element in (1) is proportional to 14 the number of nucleons in the overlap region. This suggests a factorization of the imaginary potential:where W(r) contains the radial dependence and W(E, L) the energy and angular momentum dependence through the level density p(E* , L) of the compound nucleus. This approximation implies that the formation of the compound nucleus is the dominant reaction mechanism. Such an assumption is reasonable only for not too high energy and not very heavy ions. This is one first limitation of our model. The yrast level is the state of highest possible angular momentum for a given excitation energy. Since angular rotations have usually the highest angular momentum, for a given energy the yrast level is reached if all excitation energy is converted into rotational energy. Bellow the yrast line of the compound nucleus the imaginary potential (2) is exactly zero since no compound states can be reached. Based on these ideas the imaginary part (2) of the optical potential may be expressed as follows: W(r) = { 1 + exp [(r -R)/a] }-l , R =Q,@;'~ +Ati3),