Random quantum spin systems

Author(s)

José Abel Hoyos Neto

• Document type: Doctoral Thesis
• Press: Campinas, SP.
• Institution: Universidade Estadual de Campinas (UNICAMP). Instituto de Física Gleb Wataghin
• Defense date: 2005-11-22

Examining board members

Eduardo Miranda; Mucio Amado Continentino; Sílvio R. A Salinas; Guillermo Gerardo Cabrera Oyarzún; Amir Ordacgi Caldeira

• Advisor: Eduardo Miranda

Abstract

The purpose of this thesis is the study of the role of quenched disorder in low-dimensional strongly interacting quantum spin systems. From the theoretical point of view, spin chains are extremely attractive due to their unconventional behavior that originates in the competition between magnetic ordering and quantum fluctuations. The introduction of disorder, ubiquitous in experimental realizations, is an element that can destabilize the clean phases giving rise to new physical behavior. That is the main motivation of this study. In this thesis, we study 4 random antiferromagnetic spin systems: (i ) the spin-1/2 two-leg and zigzag ladders, (ii ) the isotropic SU(N) spin chains, (iii ) the anisotropic SU(4) spin chain, and (iv ) we also revisit the spin-1/2 chain. For such a task, we use generalizations of the strong disorder real-space renormalization group method. Concerning the first systems, we show that the ladders are always renormalized to well-known spin chains. The two-leg ladder is renormalized to a random dimerized antiferromagnetic spin-1/2 chain, hence exhibiting two phases. For strong dimerization or equivalently weak disorder the system is in the gapful Haldane phase where disorder is irrelevant. Otherwise, the Haldane gap closes and the system is driven into a nonuniversal Griffiths phase, where the thermodynamical quantities are controled by the dynamical exponent z. In contrast, the zigzag ladder is renormalized either to a random antiferromagnetic spin-1/2 chain or to a random spin chain with both ferro- and antiferromagnetic couplings. If the randomness and frustration are sufficiently weak, the ladder is renormalized to the former chain, but otherwise it belongs to the same universality class of the latter one. In addition, we related the dynamical exponent of the ferro- and aniferromagnetic spin chain with its fixed point coupling constant distributions. Moreover, through simple qualitative arguments, we determined the phase diagram of the zigzag ladder with correlated disorder. That calculation clearly showed that frustration is responsible for the appearance of ferromagnetic couplings, which place the system in the basin of attraction of the fixed point of the ferro- and antiferromagnetic spin chains. With respect to theSU(N) spin chain, we developed a generalization of the strong-disorder renormalization group method to the case of an antiferromagnetic isotropic spin chain whose spins belong to the totally antisymmetric irreducible representations of the SU(N) group, with N greater than or equal to 2. We solved the flow equations analytically and found that such chains belong to a new universality class whose fixed point distributions are characterized by infinite disorder, rendering our results asymptotically exact. The characteristic exponents of these fixed points are universal, i. e., independent of the bare disorder, and depend only on the symmetry group rank. Due to the similarities of the spin clustering rules between the ferro- and antiferromagnetic spin chain and the isotropic SU(N) spin chain in the limit of N ® µ, we were able to analytically calcu- late, through a 1/N expansion, the mean correlation function of the former chain. In the case of the SU(4) spin chain, we modified the generalization of the renormalization group method to take into account the coupling anisotropy. We determined the phase diagram through analytical and numerical calculations. All fixed points found are universal and of infinite-randomness type. However, the characteristic exponents depend in a nontrivial fashion on the anisotropy. Finally, we revisited the antiferromagnetic spin-1/2 chain. We calculated the amplitude of the mean correlation function and related it with the amplitude of the entanglement entropy of the chain. In addition, we gave arguments in favor of the universality of these amplitudes (AU)