Lifshitz phase transitions in the ferromagnetic regime of the Kondo lattice model

Abstract

We establish the low-temperature phase diagrams of the spin-1/2 and spin-1 Kondo lattice models as a function of the conduction-band filling n and the exchange coupling strength J in the regime of ferromagnetic effective exchange interactions (n 0.5). We show that both models have several distinct ferromagnetic phases separated by continuous Lifshitz transitions of the Fermi-pocket vanishing or emergence type: one of the phases has a true gap in the minority band (half metal), the others only a pseudogap. They can be experimentally distinguished by their magnetization curves; only the gapped phase exhibits magnetization rigidity. We find that, quite generically, ferromagnetism and Kondo screening coexist rather than compete, both in spin-1/2 and spin-1 models. We compute the Curie temperatures and establish a "ferromagnetic Doniach diagram" for both models. PACS numbers: 71.27.+a, 72.15.Qm, 75.20.Hr, 75.30.Kz, 75.30.Mb Materials with competing interactions, such as many lanthanide and actinide compounds, have complex lowtemperature phase diagrams with different ground states [1][2][3][4][5][6]. The Kondo lattice model (KLM) [7-9] describes a conduction band of itinerant electrons and a lattice of local moments on f shells, coupled at each site by an antiferromagnetic exchange interaction J. For large J, the moments are screened. The resulting paramagnetic state has Fermi liquid properties with strongly renormalized parameters. For small J, the conduction-band electrons are carriers of longrange magnetic interactions and the moments order. The two regimes are separated by a quantum phase transition at critical J * , as described by the Doniach diagram [10]. The Néel temperature increases at first quadratically with J, but then it peaks and decreases to zero at J * as the Kondo screening takes over. The simplest version of the KLM with spin-1/2 moments indeed has an antiferromagnetic (AFM) ground state (Néel order) for small J near half-filling [11,12]. The nature of the phase transition at J * has been investigated using a variety of methods, the most accurate of which confirm that the transition is second order (quantum critical) and indicate that it involves a change of the Fermi surface topology [13][14][15]. In the spin-1 KLM, there is no phase transition at half-filling and the AFM phase extends to large values of J. While most cerium compounds show AFM order, some are ferromagnetic (FM): CeRu 2 Ge 2 [16], CeIn 2 [17, 18], and CeRu 2 Al 2 B [19]. A number of uranium and neptunium heavy-fermion materials are also FM: UTe [20], UCu 0.9 Sb 2 [21], UCo 0.5 Sb 2 [22], NpNiSi 2 [23], Np 2 PdGa 3 [24], and UCu 2 Si 2 [25]. In addition, there are strong indications of robust coexistence of the Kondo effect and ferromagnetism, in particular in U compounds. In Refs. [25 -29] it has been proposed that an appropriate minimal model for this behavior is the spin-1 version of the KLM, where in the mean-field picture the conduction-band electrons underscreen the local moments, while the residual moments order ferromagnet...

A unified theory of ferromagnetic quantum phase transitions in heavy fermion metals

“…Right in between, the FM1 phase is featured with an indirect energy gap separating the E + (p) and E − (p) bands, and the chemical potential lies just within the gap. We have thus a spin selective Kondo insulator [7][8][9]35], in which the holon phase shift can be calculated from the quasiparticle dispersions via…”

“…Its resulting ferromagnetic (FM) or antiferromagnetic (AFM) long-range orders can often be suppressed by pressure, magnetic field, or chemical substitution and tuned to a heavy Fermi liquid (HFL) with an enhanced effective mass of Landau quasiparticles due to the Kondo screening by conduction electrons. However, unlike AFM Kondo lattices, whose magnetic quantum phase transitions (QPTs) and associated quantum criticality have been intensively studied in the past decades [3,4], the FM Kondo lattices are relatively less explored [5][6][7][8][9]. It has been believed that FM QPT in clean metallic systems should typically be of first order if not interrupted by other instabilities [10] because of non-analyticities in the free energy induced by certain soft modes related to the Fermi surfaces [11,12].…”

Large and small Fermi-surface spin density waves in the Kondo lattice model

“…Finally, we want to note that a similar Lifshitz transition has been observed in the ferromagnetic phase of the Kondo lattice model. 54 Thus, a change in the Fermi-surface topology when increasing the interaction strength seems to be a general property of the Kondo lattice model and not specific to the SDW phase close to half filling.…”

Lifshitz Transitions in Magnetic Phases of the Periodic Anderson Model

“…In the other phases, both spin bands have Fermi surfaces, and thus a transition to the half-metallic state from any of the other states inevitably accompanies a change in the Fermi surface topology. Indeed, such transitions to the half-metallic state have been found in the Kondo lattice model [24][25][26][27][28][29][30][31] and in the periodic Anderson model. [32][33][34][35][36][37][38] In these theoretical studies, while the models are similar, the antiferromagnetic and ferromagnetic cases are treated separately except for a Kondo lattice model with the explicit inclusion of antiferromagnetic and ferromagnetic Heisenberg interactions.…”