Metastability of the uniform magnetization in three-dimensional random-field Ising model systems. II. Fe0.47Zn

Abstract

Optical Faraday rotation was used to measure the uniform magnetization M versus temperature T, field H, and time t of the random-field Ising model system ${\mathrm{Fe}}_{0.47}$${\mathrm{Zn}}_{0.53}$${\mathrm{F}}_{2}$. The critical behavior of (\ensuremath{\partial}M/\ensuremath{\partial}T${)}_{H}$ versus T in fields up to 5 T confirms previous results at lower fields H\ensuremath{\le}1.9 T. The dynamical rounding temperature ${\ensuremath{\varepsilon}}^{\mathrm{*}}$ scales as ${H}^{2/\ensuremath{\varphi}}$ with \ensuremath{\varphi}\ensuremath{\sim}1.4, as predicted previously. Excess magnetization \ensuremath{\Delta}M is found in the field-cooled or field-decreased metastable domain state, respectively. \ensuremath{\Delta}M is concentrated at the domain walls and, hence, scales as ${H}^{2}$ at T\ensuremath{\sim}${T}_{c}$(H). On cooling \ensuremath{\Delta}M approaches zero in the low-H, broad-wall limit, but \ensuremath{\Delta}M is approximately constant for large H at all T${T}_{c}$(H), where vacancy pinning dominates. By decreasing from large H at constant low T, one subsequently finds \ensuremath{\Delta}M\ensuremath{\propto}[Tln(t/\ensuremath{\tau})${]}^{\mathrm{\ensuremath{-}}1}$. Both the behavior for T\ensuremath{\sim}${T}_{c}$ and for high H are essentially as predicted recently by Nattermann and Vilfan. The primary difference is that \ensuremath{\tau} is not simply a constant attempt time, ${\ensuremath{\tau}}_{0}$\ensuremath{\sim}${10}^{\mathrm{\ensuremath{-}}14}$--${10}^{\mathrm{\ensuremath{-}}10}$ s, but rather varies with T, approximately as \ensuremath{\tau}=${\ensuremath{\tau}}_{0}$exp(DT), with D\ensuremath{\sim}1.3 ${\mathrm{K}}^{\mathrm{\ensuremath{-}}1}$. This can be understood by considering the increasing influence of domain volume contributions to \ensuremath{\Delta}M as T approaches ${T}_{c}$. \ensuremath{\Delta}M0 is also found on reversing the T scan of a zero-field cooled sample below but close to ${T}_{c}$. \ensuremath{\Delta}M in this case is due to the freezing-in of very slow finite-size thermal fluctuations and does not indicate broken long-range order.