High speed PFM of lithium niobate

High Speed PFM of Lithium Niobate 

The Periodically Poled Lithium Niobate Test Sample (AR-PPLN, http://www.asylumresearch.com/Products/AR-PPLN/AR-PPLN.shtml) shown in this video consists of a 3 mm x 3 mm LiNbO₃ transparent die that is 0.5 mm thick. The active area is an alternating pattern of oppositely poled stripe domains that are parallel to one axis of the die and cover the entire die surface. The pitch of the domains is 10 µm. It is produced on a lithium niobate wafer. The process starts with poling the wafer with a very strong electric field to create a homogeneously poled substrate with a polarization axis orthogonal to the wafer surface. A single lithographic photomask is used to print a mask of parallel stripes on one side of the wafer. It is then polarized in the opposite direction, which affects only the unmasked striped areas and propagates through the entire thickness of the wafer. Because of the close spacing of the unmasked regions, the polarization can "wander" laterally through the wafer thereby making the 10 µm pitch periodic polarization best on one side of the wafer and somewhat random on the other side.

For these investigations, the sample was be electrically connected to a sample holder in the Cypher AFM including the optional High Voltage Amplifier (http://www.asylumresearch.com/Products/CypherOptions/CypherOptions.shtml). The HV-PFM option enables high sensitivity, high bias, and crosstalk-free measurements on piezoelectrics, including ferroelectrics and multiferroics. The cantilever was an Arrow UHF the was custom coated with a Pt film, making it electrically conductive.In this example, a short (~0.5 second) -100 Volt DC pulse was applied to the sample. This pulse distorted the domain boundary in the middle of the image. After applying the negative pulse, we applied a gradually increasing positive bias, starting from zero and ending at 30 V. This was sufficient to restore the domain to a state very close to the original. 

Astute observers will notice that the amplitude is not the same on either side of the domain boundary and in fact this asymmetry increases with the bias. We attribute this effect to the (undesirable) electrostatic bias effect as discussed in Kalinin and Bonnell, Physical Review B, Volume 65, 125408 (2002) and other references.