DisorderDriven nonFermi Liquid Behavior

1
Disorder-Driven non-Fermi Liquid Behavior in
Strongly Correlated Metals
Vladimir Dobrosavljevic Department of Physics and
National High Magnetic Field Laboratory Florida
State University,USA
Funding NHMFL/FSU Alfred P. Sloan Foundation NSF
grant DMR-9974311
Collaborators Maria Carolina de Oliveira Aguiar
(Campinas,FSU) Darko Tanaskovic (FSU) Eduardo
Miranda (Campinas) Gabi Kotliar (Rutgers)
2
Contents
3
Disorder-driven NFL behavior doped semiconductors

• NFL two-fluid thermodynamics

• Conventional FL transport
• (FL for disordered systems)

4
Disorder-driven NFL behavior Kondo alloys
Effects of disorder
T a-1
NOTE a 1 (marginal!!)
5
Phenomenological Kondo-Disorder Model (DKM)
Broad distribution of Kondo temperatures, Due to
random environment (random J-s) P(TK) fitted for
susceptibility
6
Microscopic Basis for DKM Dynamical Mean-Field
Theory (DMFT)

• Anderson lattice model, random Kondo couplings
• JK V2/Ef distribution from fitted P(TK)

Integate out all sites but one
f-sites
c-sites
Environment of Kondo spin hybridization function
determined self-consistently
DISORDER
7
Successes of the DMFT-DMK Model

• For UCu5-xPdx explained with no additional
  fitting parameters
• specific heat, dynamic neutron scattering,
  optical conductivity,
• resistivity, magnetoresistance,

8
Magnetic Griffiths phase (MGP) scenario (Castro-Ne
to Jones, 1999, 2001)

• Idea proximity to a magnetically ordered
  phasedisorder
• Clusters (droplets) form even before global
  ordering
• Numerical evidence of this in insulating random
  magnets
• (Huse, Bhatt, et al.)
• Metallic host quantum tunneling of clusters
  (cluster Kondo effect)
• Distribution of clusters sizes
  distribution of tunneling rates TK

• Phenomenology IDENTICAL
• as in the Kondo disorder model
• (only thermodynamics explored)
• P(TK) (TK)a-1 c(T) (T)a-1
• Non-universal exponent a

Rare, dilute magnetically ordered cluster
9
Why is the magnetic Griffiths phase is an
unlikely explanation?

• How much entropy is due to NFL? A lot !!!!
• Estimate from DKM fit of P(TK)
• (relevant for both models)
• There are 3-10 free two-level fluctuators
• (Kondo spins or droplets) per unit cell!!!
• (temperature range 10-50K)
• Average distance is 2-3 lattice spacings
• Clusters if existing CANNOT be large
• (as assumed in MGP scenario)

• Dissipation due to fermionic bath
• (as in Caldeira-Leggett problem)
• Millis, More, Schmalian (PRL 2001, PRB 2002)
• show that larger clusters are overdamped
• i.e. there is no tunneling!!!!
• (key assumption of MGP scenario)

cluster Kondo effect favors tunneling
10
Disorder-driven NFL - open questions
11
Localization-induced electronic Griffiths
phase (Miranda Dobrosavljevic, PRL 2001)
The physical picture
12
Localization interaction statDMFT approach

• Localization fluctuations of the cavity
• which is now site-dependent

Greens function of conduction electrons with
site j removed
Sum over z sites

• Quasiparticle Hamiltonian for conduction
  electrons (diagonalize numerically)

• Scattering potential from Kondo centers

• Efficient numerical implementation on Bethe
  lattices (recursion)

13

• Experiments distribution of
• distances (Kondo couplings JK)
• Compare results of
• Single impurity model
• DMFT-CPA
• StatDMFT (includes localization)
• Feedback effects of localization
• UNIVERSAL form of P(TK)
• Still too few low TK s, no NFL

14
(for Anderson lattice and Hubbard models)
15
(No Transcript)
16
(No Transcript)
17
(No Transcript)