Recently, Xavier et al. claimed the existence of an insulating spin dimer state in the one-dimensional Kondo lattice model at quarter-filling amidst the paramagnetic metallic phase. In this comment we show that the dimer-dimer correlation decays algebraically at long distances and the charge gap vanishes in the bulk limit.
We revisit the problem of the quarter-filled one-dimensional Kondo lattice model, for which the existence of a dimerized phase and a nonzero charge gap had been reported by Xavier et al. [Phys. Rev. Lett. 90, 247204 (2003)]. Recently, some objections were raised claiming that the system is neither dimerized nor has a charge gap. In the interest of clarifying this important issue, we show that these objections are based on results obtained under conditions in which the dimer order is artificially suppressed. We use the incontrovertible dimerized phase of the Majumdar-Ghosh point of the ${J}_{1}\ensuremath{-}{J}_{2}$ Heisenberg model as a paradigm with which to illustrate this artificial suppression. Finally, by means of extremely accurate density-matrix renormalization-group calculations, we show that the charge gap is indeed nonzero in the dimerized phase.
The long-distance decay of correlation functions in the one-dimensional Hubbard model is determined for arbitrary band filling and correlation strength, using the exact solution of Lieb and Wu. In particular, for either infinitely strong on-site repulsion U, or in the close proximity of half filling for any U, spin-spin correlations decay like cos(2kFx)${\mathit{x}}^{\mathrm{\ensuremath{-}}3/2}$ ln1/2(x). For infinite U the results are generalized to the case of nonzero nearest-neighbor interaction. The behavior of the frequency-dependent conductivity is also discussed, in particular in the proximity of the metal-insulator transitions occurring for half and quarter filling.
We have studied the one-dimensional Kondo chain at half filling using a density matrix formulation of the numerical renormalization group. The charge gap is larger than the spin gap for all antiferromagnetic values of the exhange coupling J. A new type of excitation, a neutral spin singlet, consists of a particle and a hole which are repulsive for J\ensuremath{\gtrsim}5t and attractive for J\ensuremath{\lesssim}5t. As J\ensuremath{\rightarrow}0, RKKY interactions become more important and the staggered susceptibility \ensuremath{\chi}(q=2kF) diverges. We have also studied \ensuremath{\chi}(q) and the dispersion of the low-lying spin excitations as a function of wave vector q.
In order to better understand Kondo insulators, we have studied both the symmetric and asymmetric Anderson lattices at half filling in one dimension using the density-matrix formulation of the numerical renormalization group. The asymmetric case is treated in the mixed-valence regime. We have calculated the charge gap, the spin gap, and the quasiparticle gap as a function of the repulsive interaction U using open boundary conditions for lattices as large as 24 sites. We find that the charge gap is larger than the spin gap for all U for both the symmetric and asymmetric cases. Ruderman-Kittel-Kasuya-Yosida interactions are evident in the f-spin--f-spin correlation functions at large U in the symmetric case, but are suppressed in the asymmetric case as the f level approaches the Fermi energy. This suppression can also be seen in the staggered susceptibility \ensuremath{\chi}(q=2kF) and it is consistent with neutron scattering measurements of \ensuremath{\chi}(q) in CeNiSn.
The density-matrix renormalization-group method is applied to the one-dimensional Kondo-lattice model with the Coulomb interaction between the conduction electrons. The spin and charge gaps are calculated as a function of the exchange constant J and the Coulomb interaction Uc. It is shown that both the spin and charge gaps increase with increasing J and Uc. The spin gap vanishes in the limit of $J\ensuremath{\rightarrow}0$ for any Uc with an exponential form, ${\ensuremath{\Delta}}_{s}\ensuremath{\propto}\mathrm{exp}[\ensuremath{-}\frac{1}{\ensuremath{\alpha}({U}_{c})J\ensuremath{\rho}}]$. The exponent, $\ensuremath{\alpha}({U}_{c})$, is determined as a function of Uc. The charge gap is generally much larger than the spin gap. In the limit of $J\ensuremath{\rightarrow}0$, the charge gap vanishes as ${\ensuremath{\Delta}}_{c}=\frac{1}{2}J$ for Uc=0 but for a finite Uc it tends to a finite value, which is the charge gap of the Hubbard model.