Early work by the author with Prof. Ishibashi [Scott et al., J. Appl. Phys. 64, 787 (1988)] showed that switching kinetics in ferroelectrics satisfy a constraint on current transients compatible with d = 2.5 dimensionality. At that time with no direct observations of the domains, it was not possible to conclude whether this was a true Hausdorff dimension or a numerical artefact caused by an approximation in the theory (which ignored the dependence of domain wall velocity upon domain diameter). Recent data suggest that the switching dimensionality is truly fractal with d = 2.5. The critical exponent β characterizing the order parameter P(T) can be written as a continuous function, which is exact within hyperscaling; here ν and η are the exponents characterizing the pair correlation function G(r,T) and the structure factor S(q,T). For d=2.5 the estimate is that β is approximately ¼.
I. IntroductionTwenty years ago the author used current transients i(t) = dD/dt to determine details about switching kinetics in ferroelectric thin films; here t is time and D is the displacement vector, approximately equal to polarization P in a ferroelectric. The initial study [1] used potassium nitrate, KNO 3 , but similar studies of lead zirconate titanate PZT followed shortly thereafter [2]. Later work by Shur [3] extended this particular modelling technique. The theory used for this data-fitting was developed originally by Ishibashi and Takagi [4,5], and is based upon the analogous problem of grain growth in crystallization due to Avrami [6]. The theory is falsifiable in a nice way by the fact that it affords constraints that test the self-consistency of the model; in particular the dimensionless ratio i(p) t(p)/P s = Λ (1.)