arXiv1995
We consider the Eliashberg theory in the coupling region where some fundamental qualitative deviations from the conventional BCS-like behaviour begin to appear. These deviations are identified as the onset of a cross-over from BCS superconductivity to Bose condensation. We point out that the beginning of this cross-over occurs when the gap $Δ_g$ becomes comparable to the boson energies $Ω_{ph}$. This condition is equivalent to the condition of Ref. \cite{Strinati} $k_Fξ\approx 2π$ and traduces the physical constraint that the distance the paired electron covers during the absorbtion of the virtual boson, cannot be larger than the coherence length. The frontier region of couplings is of the order of $λ\approx 3$, and high-$T_c$ materials are concerned. A clear qualitative indication of the occurence of a cross-over regime should be a dip structure above the gap in the density of states of excitations. Comparing our results with tunneling and photoemission experiments we conclude that high-$T_c$ materials (cuprates and fullerides) are indeed at the beginning of a cross-over from BCS superconductivity to Bose condensation, even though the fermionic nature still prevails. Taking into account the analysis of Ref. \cite{Strinati}, we predict a dip structure in heavy fermion and organic superconductors. Non-adiabatic effects beyond Migdal's theory are considered and give insight on the robustness of Eliashberg theory in describing qualitatively this cross-over regime, although for the quantitative interpretation of the results the inclusion of non-adiabatic corrections can be important.