Matthew L. Trawick,* Mischa Megens,* Christopher Harrison,§ Dan E. Angelescu,* Daniel A. Vega,† Paul M. Chaikin,*‡ Richard A. Register,†‡ Douglas H. Adamson‡
*Department of Physics, †Department of Chemical Engineering, and ‡Princeton Materials Institute, Princeton University, Princeton, New Jersey; §Polymers Division, National Institute of Standards and Technology, Gaithersburg, Maryland, USA
Full-text (for Scanning subscribers)
Summary: We describe a method for using polynomial mapping to correct scanning probe microscope images for distortion due to piezoelectric creep. Because such distortion varies from image to image, this method can be used when the actual locations of some features within an image are known absolutely, or in a series of images in which the actual locations of some features are known not to vary. While the general case of polynomial mapping of degree N requires the determination of 2(N+1)2 matrix elements by regression, we find that by understanding the mechanism by which piezoelectric creep distorts scanning probe microscope images, we can fix most of these coefficients at 0 or 1 a priori, leaving only 2(N+1) coefficients to be determined by regression. We describe our implementation of this strategy using the Interactive Data Language (IDL) programming language, and demonstrate our technique on a series of atomic force microscopy (AFM) images of diblock copolymer microdomains. Using our simplified scheme, we are able to reduce the effects of distortion in an AFM image from 5% of the scan width to a single pixel, using only five reference points.
Key words: scanning probe microscopy, image processing, distortion, piezoelectric creep, polynomial mapping
PACS: 07.79.-v, 07.79.Lh, 07.05.Pj, 68.37.-d, 83.80.Uv
This work was supported by the National Science Foundation through the Princeton Center for Complex Materials (DMR-9809483 and DMR-0213706), and by the donors of the American Chemical Society Petroleum Research Fund (35207-AC5,7).