998c20e9128d55e4.tex
1: \begin{abstract} 
2: We study by Wilson numerical renormalization group the spectral properties 
3: of a two-orbital Anderson impurity model in the presence of an exchange splitting 
4: which follows either regular or inverted Hund's rules. 
5: The phase diagram contains a non-Fermi liquid fixed point separating a 
6: screened phase, where conventional Kondo effect occurs, from an 
7: unscreened one, where the exchange-splitting takes care of 
8: quenching the impurity degrees of freedom. 
9: On the Kondo screened side close to this fixed point the impurity density of states 
10: shows a narrow Kondo-peak on top of a broader resonance. This narrow 
11: peak transforms  in the unscreened phase 
12: into a narrow pseudo-gap inside the broad resonance.  
13: Right at the fixed point only the latter 
14: survives. The fixed point is therefore identified by a jump of the density of states 
15: at the chemical potential.  
16: We also consider the effect of several particle-hole symmetry-breaking terms. 
17: We show that particle-hole perturbations which simply shift the orbital energies 
18: do not wash out the fixed point, unlike those perturbations 
19: which hybridize the two orbitals. 
20: Consequently the density-of-state jump 
21: at the chemical potential remains finite even away from particle-hole symmetry.
22: In other words, the pseudo-gap stays pinned at the chemical potential, although it 
23: is partially filled in. 
24: We also discuss the relevance of these results for lattice models which map 
25: onto this Anderson impurity model in the limit of large lattice-coordination.    
26: Upon approaching the Mott metal-insulator transition, 
27: these lattice models necessarily enter a region with a   
28: local criticality which reflects the impurity non-Fermi liquid fixed point.    
29: However, unlike the impurity, the lattice can get rid of the 
30: single-impurity fixed-point instability by spontaneously developing bulk-coherent 
31: symmetry-broken phases, which we identify for different lattice models. 
32: 
33: 
34: \end{abstract}