5.2 Haldane material Y₂BaNiO₅

In Fig.38, we show the crystal structure of the inorganic Haldane material Y₂BaNiO₅ [97]. The structure is characterized by chains of compressed NiO₆ octahedra, which are separated by non-magnetic Y³⁺ and Ba²⁺ ions. The shortened Ni-O bonds along the chain (a-axis) lead to a relatively large super-exchange interaction $J\approx 285\sim340$ K, as estimated from the magnetic susceptibility [98,99]. The magnitude of the Haldane gap has been obtained as $E_{\rm g}\sim 100$ K with susceptibility [100,101,99] and inelastic neutron scattering measurements [102,103].

  

Figure 38: The crystal structure of Y₂BaNiO₅. $4\times 2\times 1$ unit cells are shown. The chain direction is the a-axis.

One valuable feature of this material is that both off-chain charge doping (Y³⁺$\rightarrow$Ca²⁺) and on-chain vacancy doping (Ni²⁺$\rightarrow$Zn²⁺, Mg²⁺) is possible [98,104,103]. Since charge-doping to the Haldane ground state is a unique feature, some previous measurements have been aimed to clarify the behavior of the doped charge.

DiTusa et al. [103] measured resistivity ($\rho$), X-ray absorption spectroscopy (XAS), and inelastic neutron scattering (NIS) of the Ca/Zn-doped systems, and found the following features:

(1)

the doped holes did not induce a clean metallic behavior, but a localized state with hopping ($\rho$).

(2)

the doped holes locate mainly at the 2p_z orbital of the oxygen ion, which has a lobe in the chain direction (XAS).

(3)

charge doping induces a density of states within the Haldane gap (NIS). With vacancy doping (Ni²⁺$\rightarrow$Zn²⁺), such `in-gap states' were absent.

The result (3) implies that hole-doping strongly perturbs the non-magnetic ground state, while vacancy-doping preserves the non-magnetic ground state.

Quantitatively, vacancy doped Y₂BaNiO₅ has shown a mysterious response. Ramirez et al. [104] measured specific heat of the vacancy-doped compounds (Ni²⁺$\rightarrow$Zn²⁺) and found that the number of doping-induced spins does not follow the Valence Bond Solid state scenario; the VBS picture predicts the creation of two S=1/2 spins for each vacancy (see Fig.35d), but the Zn-doped Y₂BaNiO₅ system exhibited one S=1 spin for two Zn ions. To understand this phenomenon, Ramirez et al. suggest a heuristic `singlet-triplet model', which assumes that half of the broken chains form a triplet S=1 and the other half, a singlet S=0. Neither the origin of the couplings between the chain-end spins nor the local structure of the doping-induced spins is clear at the present stage.

To investigate the ground state properties and dynamics of spin systems, $\mu$SR is a powerful technique as introduced in the previous chapter. The following presents $\mu$SR results of the nominally pure/doped Y₂BaNiO₅, as well as their susceptibility data in low magnetic fields. For our measurements, polycrystalline specimens of nominally pure Y₂BaNiO₅, Ca doped systems [(Y_2-xCa_x)BaNiO₅; x=4.5, 9.5, 14.9 and 30.5%] and Mg doped systems [Y₂Ba(Ni_1-yMg_y)O₅; y=1.7 and 4.1%] were prepared at the University of Tokyo, Superconductivity Laboratory, using a standard solid state reaction described below [105]:

• Stoichiometric mixtures of Y₂O₃, BaCO₃, NiO plus CaCO₃ and/or MgCO₃ were prepared.
• The mixture was baked in air at 1050$^{\circ}$C for 10 hours.
• After being pressed into a pellet, the sample was baked in air at 1250$^{\circ}$C for 10 hours.
• In order to reduce excess oxygen, the sample was annealed in an Ar atmosphere at 1200$^{\circ}$C for 10 hours.

Powder X-ray analysis of the samples did not detect any impurity phases. The Ca and Mg concentrations (x and y) were determined with the atomic light-absorption method, which was commercially available (Robertson Microlit Laboratory, NJ, USA).