Kink Structure in the Quasiparticle Band of Doped Hubbard Systems

Abstract

By making use of the self-consistent projection operator method with high-momentum and high-energy resolutions, we find a kink structure in the quasiparticle excitation spectrum of the two-dimensional Hubbard model in the underdoped regime. The kink is caused by a mixing between the quasiparticle state and excitations with short-range antiferromagnetic order. We suggest that this might be the origin of the strong concentration dependence of the 'kink' found in La2−xSrxCuO4 (x = 0.03 − 0.07).KEYWORDS: kink, quasiparticle excitations, Hubbard model, ARPES, cuprate, LSCO Recent high-resolution photoemission experiments show that there is a well-defined 'kink' in the quasiparticle band dispersion of high-T c cuprates, whose energy scale is ω kink = 60 − 70 meV in both the normal and superconducting states.1-3 The kink along the nodal direction was found to have a universal feature, 2 i.e., the Fermi velocity v F in the low-energy region (|ω| < ω kink ) is not sensitive to the type of cuprates, doping concentration, and isotope substitution, 4 although v F in the high-energy regime (|ω| > ω kink ) strongly depends on the latter. Various theoretical explanations for the kink have been attempted from two different points of view. One type of theory relies on the coupling of an electronic quasiparticle to the spin fluctuation resonance mode observed in inelastic neutron scattering experiments.5, 6 Another relies on a coupling to phonons, particularly to the longitudinal optical phonon mode found in neutron experiments.2, 3 The latter approach raised again the fundamental question on the mechanism underlying high-T c superconductivity in cuprates, i.e., electron-or phononmediated. Although the reliability of these theories is under debate and the improvements of these theories are in progress, 7, 8 it has not yet been seriously studied whether or not the kink in the quasiparticle state can be solely of electronic origin. We deal with this problem in the present letter and report the appearance of a kink due to long-range electron correlations in the two-dimensional (2D) Hubbard model for small doping concentrations.The difficulty in the present problem is that the perturbation approach is not applicable to cuprates because of strong electron correlations. Therefore, advanced theories such as the Lanczos method, 9 the quantum Monte-Carlo (QMC) method, 10-12 and dynamical cluster approximation (DCA), 13,14 have been applied to the cuprate system. They clarified the global structure of the single-particle excitation spectrum in the 2D Hubbard model. A detailed structure of the low-energy excitations at low temperatures, however, has not been derived because of the limited resolutions in both momentum and energy and the limited range of intersite * E-mail address: yok@sci.u-ryukyu.ac.jp † E-mail address: fulde@mpipks-dresden.mpg.de electron correlations inherent in the cluster approaches.In particular, a high resolution is indispensable for theoretical investigations of the kink. In order to overcome these difficulties, we h...

Renormalization of Bulk Magnetic Electron States at High Binding Energies

“…Furthermore, our interpretation is in line with quite a number of theoretical works, e.g. on cuprates, in which the emergence of kinks has been attributed to the coupling of QPs and bosonic (antiferromagnetic) spin excitations [25,26,27].…”

Paramagnon-induced dispersion anomalies in the cuprates

“…In fact, this scattering is a precursor to the antiferromagnetically ordered state near half-filling and it is responsible for opening the magnetic gap. In contrast, a number of authors have related the presence of waterfall-like effects near half-filling to "Mott" physics associated with ͑ , ͒ AFM fluctuations, [29][30][31] but have difficulty explaining why these effects persist into the overdoped regime. …”

Nonlocal Excitation Spectra in Two-Dimensional Doped Hubbard Model

“…Preliminary results on these topics have been published recently. 28,29 It should be noted that the validity of numerical calculations on these problems strongly depends on the momentum and energy resolutions. We therefore have recalculated the spectra increasing the number of mesh from 80 × 80 to 160 × 160 in the first Brillouin zone and the energy mesh by a factor of two.…”