The calculation of the ground state energy of a biexciton is quite involved, and that of the excited states even more so. Karp and Moskalenko [1] (KM) proposed a model, which is intuitively very appealing, to get around this difficulty. They proposed to represent the interaction between two excitons by a Morse potential [2], in analogy with a diatomic molecule, but recognized that the Morse potential must be generalized to the case of an arbitrary ratio s (s ¼ m * e =m * h ) of the electron and hole masses. This ratio appears in the expression for the interaction potential because of the averaging of the energy of the Coulomb interaction of electrons and holes in two excitons over the functions of relative motion in excitons. This approach is justified if the binding energy of an exciton is greater than the binding energy of a biexciton. The generalized Morse potential should be invariant under the replacement of s with 1=s, which reflects the symmtery of the Hamiltonian under the replacement of electrons with holes and of m * e with m * h . KM further required that this potential should satisfy the Feynman theorem for the ground-state energy [3,4], give the correct values of the binding energy between hydrogen molecules [5], and yield the binding energy of a biexciton which agrees with the results of Akimoto and Hanamura [6] for arbitrary s. With these stipulations, they constructed the following expression for the generalized Morse potential:where r 0 ¼ 1:4a s =S; a ¼ d ffiffiffi S p ; and S ¼ ð1 þ s 2 Þ=ð1 þ sÞ 2 . a s is the exciton radius and I is the ionization energy of an exciton with a given value of s. j and d are parameters which have the same value as for a hydrogen molecule. KM found their values to be j = 0.351, and d = 1.44. At least when s ( 1, this model ought to work.Separating the three coordinates of the centre of gravity of a biexciton, KM obtained the Schrö -dinger equation for the relative motion of excitons in a biexciton, which takes place in the central force field given by the Eq. (1). The usual separation of the variable in problems with central forces can also be applied to the present case, and this led to the following Schrö dinger equation:where uðrÞ ¼ ryðrÞ, m s is the exciton mass, n is the total quantum number and ' is the angular momemtum quantum number. For ' ¼ 0, the Schrö dinger equation for the Morse potential can be solved analytically [2, 7] to a high degree of approximation [8]. In an analogous fashion, KM obtained the following expression for the energy of ' ¼ 0 states:However, for ' > 0, KM had to use an approximate method to calculate the energies. They obtained results for a number of levels for biexcitons in CuCl, CuBr and CuI.We shall represent the ratio of the binding energies of the biexciton and the exciton by R. Thus R ¼ Eð0; 0Þ=ðÀIÞ.Recently, Usukura, Suzuki and Varga [9] have calculated the binding energies and other properties of excitonic complexes by a precise variational method. A comparison of their ðR; sÞ curve with the one obtained from Eq. (2) with th...