arXiv1999
In order to obtain high quality, reliable electronic Raman spectra of a high-Tc superconductor compound, we have studied strongly under-doped HgBa2Ca2Cu3O8+d. This choice was made such as to i)minimize oxygen disorder in the Hg-plane generated by oxygen doping ii) avoid the need of phonon background subtraction from the raw data iii)eliminate traces of parasitic phases identified and monitored on the crystal surface. Under these experimental conditions we are able to present the pure electronic Raman response function in the B2g, B1g, A1g+B2g and A1g+B1g channels. The B2g spectrum exhibits a linear frequency dependence at low energy whereas the B1g one shows a cubic-like dependence. The B2g and B1g spectra display two well defined maxima at 5.6kBTc and 9kBTc respectively. In mixed A1g channels an intense electronic peak centered around 6.4 kBTc is observed. The low energy parts of the spectra correspond to the electronic response expected for a pure dx2-y2 gap symmetry and can be fitted up to the gap energy for the B1g channel. However, in the upper parts, the relative position of the B1g and B2g peaks needs expanding the B2g Raman vertex to second order Fermi surface harmonics to fit the data with the dx2-y2 model. The sharper and more intense A1g peak appears to challenge the Coulomb screening efficiency present for this channel. As compared to previous data on more optimally doped, less stoichiometric Hg-1223 compounds, this work reconciles the electronic Raman spectra of under- doped Hg-1223 crystals with the dx2-y2 model, provided that the oxygen doping is not too strong. This apparent extreme sensitivity of the electronic Raman spectra to the low lying excitations induced by oxygen doping in the superconducting state is emphasized here and remains an open question.