Evolution of nonlinearity with different film thicknesses. (A) Spatial maps. (B) Histograms (areas not normalized). (C) Macroscopic dielectric measurements and averaged PFM data (no free parameters).

A collaborative group of researchers from Oak Ridge National Lab, Penn State, the Univ. of Sheffield (U.K.) and EPFL (Switzerland) have made a discovery that overturns some 100-year-old assumptions about the behavior of polycrystalline and other ferroelectric materials. In brief, the nonlinear materials responses and associated enhanced electromechanical properties that occur at the microlevel stop develop in clusters on the micron length scale. Their work is published in the Proceedings of the National Academy of Sciences.

Let me explain their conclusions in something of a roundabout way. Over the last decade, the science and engineering community has perfected the creation of microelectromechanical systems and applications. The usefulness of these MEMS, commonly found in anything from the accelerometers in iPhones to defense-oriented chemical- and biosensors, is now being improved by taking advantage functional materials (such as ferroelectrics) with low-field hysteretic (nonlinear) responses, and associated enhanced materials properties

Hysteresis – think of it as material “memory,” such as the tendency of iron to remain magnetized – has been studied for decades. Perhaps its most ubiquitous use is in the binary memory storage system used to store data on computer hard drives. Generally speaking, hysteresis results from the collective nonlinear activity at the level of individual domains within the material.

Piezoelectricity in many ferroelectric materials is amplified by a hysteretic behavior: A mechanical stress (or an applied electric field) triggers a realignment of a material’s domains and enhances the polarization (or strain). Because the domains don’t return to their original position, the response is associated with hysteresis.

As mentioned above, these behaviors have been well-studied and utilized at the mesolevel for decades, and more recently at the microlevel. The piezoelectric materials in many medical ultrasound systems, for example, show the “Rayleigh” nonlinear behavior that occurs when a small external field applied. The field causes domain walls move and domains parallel to the external field start to grow. The unique aspect of this low field hysteretic response is that they strongly enhance – by tens of percents – the effective electromechanical properties.

But, now, lets say that instead of making bulk devices, we want to make MEMS or NEMS – nanoelectromechanical systems – that might be useful in extremely small sensors and bioinspired devices. It would be natural to assume that all we have to do, assuming we had the right tools and techniques, is scale everything down to a fraction of its previous size.

The Rayleigh behavior is just going to continue, but at a tinier scale, right? The aforementioned research group essentially said, “Well, let’s see.”

They wanted keep things simple, and use some of the oldest, but still not fully understood, materials to keep the focus on getting some basic insights. “We just wanted to look at what happens in Rayleigh (low-field) behaviors as we move from thin films to thicker films,” said ORNL’s Sergei Kalinin, one of the principle investigators. They chose a polycrystalline lead zirconate titanate capacitor system with small grains and domains that was explored using piezoresponse force microscopy.

What they did see is the origins of the Rayleigh behavior. The nonlinearity evolves in clusters that are well beyond the scale of the microstructure – demonstrating that domain walls move in concert. As the film thickness increases, the clusters merge and grow larger. For films of about 4 μm in thickness, the material is uniformly nonlinear at the measurable length scale.

This formation of nonlinearity clusters clearly is one of the factors that affects the capability to fabricate nanoscale electromechanical devices, and also provides insight into emergent collective behaviors in disordered systems. The authors note that the hysteretic responses are common for systems with structural (glasses, polymers), magnetic (spin glasses) or polar (dipole glass) disorder. The present studies are enabled by the fact that polarization reorientation does not change the underlying crystallographic lattice, and hence is (potentially) reversible. At the same time, the strong coupling between polarization and strain (reversible lattice deformation) allows us to study the dynamics locally using PFM. This makes polycrystalline ferroelelctrics an ideal model for studying general relaxation processes in disordered systems, which also include structural glasses, polymers (in which dynamics is irreversible) and spin-glasses and magnetic clusters (in which mapping local magnetization is a challenge).

Besides Kalinin, the principle investigators on this project are Stephen Jesse (ORNL) and Susan Trolier-McKinstry (Penn State).

Share/Print