ARPES STUDIES IN THE NORMAL AND SUPERCONDUCTING STATE OF Biz Sr 2
CaCuZOfJ,J. C. CAMPUZANO,H. DING,A. BELLMAN
2003-01-01
Abstract:High resolution ARPES measurements were carried out on Bi2212 samples [T,=87 K]. Most of the results have either appeared in print or will appear soon; here we briefly summarize the various points together with the relevent references. Due to the low cross sections in Bi2212, we set the energy resolution in this experiment to a FWHM= 18.8 meV, equivalent to a gaussian of (T= 8 meV, and the angular resolution was f l”, equivalent to 1/22nd of the Brillouin zone edge, or 1/32nd of the zone diagonal. Normal state spectra [ 1,2] show only one peak corresponding to the CuO2 planar band; no resolvable bi-layer splitting is seen above T,. All other spectral features can be attributed to umklapp bands corresponding to a superlattice with Q = (0.2 1,0.21 )TC which has been independently seen in structural studies. We also find some evidence for “shadow bands” which are (7r, TT) fold-backs of the main bands; details will be presented elsewhere [2]. The main CuO2 Fermi surface corresponds to a doping of 0.17 holes per Cu atom. While the Fermi surfaces and the dispersion of the peak positions appear to be like that in band theory, the spectra themselves are anomalously broad in the normal state. With decreasing T the spectra sharpen up enormously [3]. We have carried out careful studies of the lineshape, and by studying sum rules, came to the conclusion that our ARPES data can be interpreted simply in terms of the one-particle spectral function A(k, w). Thus the intensity can be expressed as I(k, w) = Zo(k)f(w)A(k, w) where f(w) is the Fermi function and ZO includes all the matrix element effects. Even though the lineshape changes a lot with temperature, the integrated intensity at k~ does not. We note that the sharpening up of the spectra below T, is not due to a pile-up of states below the gap, as previously interpreted in the photoemission literature, since ARPES measures the spectral function, and not the density of states. The sharpening of the spectra is due instead to a dramatic increase of the quasiparticle lifetime in the superconducting state, as observed in optical experiments. As the lifetime increases, Z”(k, w) decreases, decreasing the width of the spectrum. Using the sum rule j-+,“f(w)A(k, w) = n(k) where n(k) is the momentum distribution, together with the fact that n(k~) is independent of temperature, we find that the peak intensity must increase. We note in passing that the energy-integrated ARPES intensity can be used to experimentally measure the momentum distribution n(k)