Abstract
We report the results of neutron elastic scattering measurements between -250oC
and 620oC on the lead-free relaxor Na1/2Bi1/2TiO3 (NBT). Strong, anisotropic, elastic
diffuse scattering intensity decorates the (100), (110), (111), (200), (220), and (210)
Bragg peaks at room temperature. The wavevector dependence of this diffuse
scattering is compared to that in the lead-based relaxor PbMg1/3Nb2/3O3 (PMN) to
determine if any features might be common to relaxors. Prominent ridges in the
elastic diffuse scattering intensity contours that extend along <110> are seen that
exhibit the same zone dependence as those observed in PMN and other lead-based relaxors. These ridges disappear gradually on heating above the cubic-to-tetragonal
phase transition temperature TCT = 523oC, which is also near the temperature at which
the dielectric permittivity begins to deviate from Curie-Weiss behavior. We thus
identify the <110>-oriented ridges as a relaxor-specific property. The diffuse
scattering contours also display narrower ridges oriented along <100> that are
consistent with the x-ray results of Kreisel et al. (2003); these vanish below 320oC
indicating that they have a different physical origin. The <100>-oriented ridges are
not observed in PMN. We observe no equivalent relaxor-specific elastic diffuse
scattering from the homovalent relaxor analogues K0.95Li0.05TiO3 (A-site disordered)
and KTa0.95Nb0.05O3 (B-site disordered). This suggests that the <110>-oriented
diffuse scattering ridges are correlated with the presence of strong random electric
fields and invites a reassessment of what defines the relaxor phase. We find that
doping NBT with 5.6% BaTiO3, a composition close to the morphotropic phase
boundary with enhanced piezoelectric properties, increases the room temperature
correlation length along [1 1 0] from 40 Å to 60 Å while doubling the associated
integrated diffuse scattering. Similar behavior was reported by Matsuura et al. (2006)
for compositions of PMN doped with PbTiO3. Finally, we comment on the recent
observation of monoclinicity in NBT at room temperature by placing a strict bound on
the strength of the (½½½) superlattice reflection associated with the Cc space group
based on the atomic coordinates published in the x-ray study by Aksel et al. (2011) for
NBT. We argue that a skin effect, analogous to that reported in the relaxors PZN and
PMN-10%PT, can reconcile our single-crystal data with the powder data of Aksel et
al. We believe this represents the first evidence of the relaxor skin effect in a
lead-free relaxor.