Nucleation Mechanism of Polarization Switching
on Ferroelectric Surfaces
Peter Maksymovych,1 Stephen Jesse,1 Mark Huijben,2 Ramamoorthy Ramesh,2 Anna Morozovska,3 Samrat Choudhury,4 Long-Qing Chen,4 Arthur P. Baddorf,1 and Sergei V. Kalinin1
for Nanophase Materials Sciences, Oak Ridge National Laboratory;
2Department of Materials Sciences and Engineering and Department
University of California Berkeley; 3Lashkaryov Institute for
Semiconductor Physics, National Academy of Science of Ukraine; 4Department
of Materials Science and Engineering, Pennsylvania State University
We have determined the thermodynamic mechanism of nanoscale polarization switching in the near-surface region of representative ferroelectric materials (BiFeO3 and Pb(Zr0.2Ti0.8)O3) . The ferroelectric phase transition was observed via a piezoresponse hysteresis, which was recorded locally using ultrahigh vacuum atomic force microscopy in a previously inaccessible temperature range from 50K to 300K. Contrary to prior assumptions of thermally-activated local switching, we have discovered a very weak temperature dependence of the tip bias required to induce local switching (see Figure). The observed behavior can be rationalized by assuming only the intrinsic nucleation mechanism, where the driving energy for the phase transition is supplied solely by the electric field and not involving thermal fluctuations. This result echoes the notorious Landauer paradox, posed almost 50 years ago, which states that the activation energy for ferroelectric domain nucleation in the defect free material is many orders of magnitude larger than kT in most experiments. The paradox has led to a widely accepted assumption that domain nucleation in bulk materials occurs at defect centers. In our measurements, the paradox is resolved via the strength of the electric field which is several orders of magnitude larger in the local probe geometry than in more common capacitor experiments. The assignment of the intrinsic local mechanism in local polarization switching is backed by a combination of analytical and phase-field analysis. The analysis also revealed that the model based on Landauer’s approximation overestimates the temperature dependence of the nucleation voltage, signaling its limited applicability to the description of the local phase transitions due to an incorrect description of the transition state.
Exploring the evolution of phase transitions with increasing spatial confinement is one of the key directions in nanoscience. This pathway will reveal the nanoscale origin of variability in macroscopic phase transitions, uncover intrinsic rather than defect-mediated properties of materials and will pave the way to new applications through local modulation of material properties via engineered defects and controlled doping. Ferroelectric materials are a model system to explore phase transitions, since the order parameter (spontaneous polarization) can be reversibly switched by applied electric field and without chemical degradation. The nanoscale confinement of the ferroelectric phase transition is achieved through a highly localized electric field created by a biased metal tip in contact with a dielectric surface. The materials chosen for our measurements are of immediate interest to prospective low-dimensional applications such as ferroelectric control of electron transport, coupling of multiple order parameters, nanoactuation, supercapacitance etc. The results obtained here will also be pertinent to related bias-induced phase transitions, e.g. solid-state reactions and local electrochemistry.
 P. Maksymovych, S. Jesse, M. Huijben, R. Ramesh, A. N. Morozovska, S. Choudhury, L.-Q. Chen, A. P. Baddorf, S. V. Kalinin, Phys. Rev. Lett. 102, 017601 (2009).
Research sponsored by the Center for Nanophase Materials Sciences (PM, AB, SJ, SK) and the Division of Materials Sciences and Engineering (Y.H.C., T.Z., and R.R.), Basic Energy Sciences, DOE, and the NSF (S.C., L.-Q.C.).
Topography, amplitude and phase of piezoresponse of 50 nm BiFeO3(100) film used in this study. The figure shows averages of 50-100 local hysteresis loops obtained with the same physical cantilever in ultrahigh vacuum and after venting the chamber to ambient.
Temperature dependence of local ferroelectric switching showing the average hysteresis loops of BiFeO3 as a function of sample temperature obtained using switching spectroscopy with a piezoresponse force microscope in 100-200 places on the surface for each loop. A step feature around zero-bias is an experimental artifact.
and calculated temperature dependence of the nucleation bias. The
y-axis is a nucleation bias normalized to the room temperature
values. (a) Comparison of experiment with rigid model; (b) Comparison
of experiment with phase-field calculations and the analytical LGD
(Landau-Ginzburg-Devonshire) model. The values of the nucleation
bias obtained from phase-field modeling of local ferroelectric switching
on BiFeO3(100) using a Lorentzian distribution of electric field
on the surface with a characteristic width of 25 nm. The numerical
parameters for the LGD-based values (Tc = 974 K and U0 =
-0.0077 V/K) were obtained by fitting to phase-field data.