Natural orbitals renormalization group approach to a Kondo singlet

TL;DR

This paper uses the natural orbitals renormalization group method to clarify the structure and spatial extent of the Kondo singlet, revealing a dominant active orbital and a short screening cloud.

Contribution

The study introduces the NORG method to analyze the Kondo singlet, providing a transparent view of the screening mechanism and the structure of the screening cloud.

Findings

Identified a single active natural orbital responsible for screening.

Showed the Kondo screening cloud is typically only a few nanometers.

Demonstrated the disentanglement of the impurity from the rest of the system.

Abstract

A magnetic impurity embedded in a metal host is collectively screened by a cloud of conduction electrons to form a Kondo singlet below a characteristic energy scale $T_K$, the Kondo temperature, through the mechanism of the Kondo effect. We have reinvestigated the Kondo singlet by means of the newly developed natural orbitals renormalization group (NORG) method. We find that, in the framework of natural orbitals formalism, the Kondo screening mechanism becomes transparent and simple, while the intrinsic structure of Kondo singlet is clearly resolved. For a single impurity Kondo system, there exits a single active natural orbital which screens the magnetic impurity dominantly. In the perspective of entanglement, the magnetic impurity is entangled dominantly with the active natural orbital, i.e., the subsystem formed by the active natural orbital and the magnetic impurity basically disentangles from the remaining system. We have also studied the structures of the active natural orbital respectively projected into real space and momentum space. Moreover, the dynamical properties, represented by one-particle Green's functions defined at impurity site with active natural orbital, were obtained by using correction vector method. In order to clarify the spatial extension of the Kondo screening cloud, the concept of Kondo correlation energy was introduced. With this concept we obtain a characteristic length scale beyond which the Kondo screening cloud is hardly detected in experiment. Our numerical results indicate that this characteristic length scale usually is just a few nanometers, which interprets why it is difficult to detect the Kondo screening cloud experimentally in a metal host.