Progress in Crystallographic Image Processing for Scanning Probe Microscopy
Peter Moeck and Jack Straton
Nano-Crystallography Group, Department of Physics, Portland State University, Portland, OR 97207-0751
Crystallographic Image Processing (CIP) originated with the electron crystallography community. Nobel
Laureate Sir Aaron Klug (OM, FRS) and coworkers pioneered the technique for the analysis of long-range
ordered biological materials in parallel illumination Transmission Electron Microscopes (TEMs).
Corrections for the effects of the TEM’s phase contrast transfer function and for less than optimal imaging
conditions are part of this kind of CIP. There are also “electron microscope independent” 2D crystallography
foundations to this kind of image processing.
Based on these foundations, we applied CIP to images of long-range ordered 2D periodic surface arrays that
were recorded with different kinds of scanning probe microscopes (SPMs) [1,2]. We amended our method
recently [3] to detect and correct frequently encountered artifacts in scanning probe microscopy, i.e. effects
of multiple mini-tips that collectively result in a blunt tip [4,5]. Loosely speaking, our version of CIP has the
effect of “sharpening up” a blunt scanning probe tip. This is achieved by the deconvolution of the prevailing
microscope’s point spread function from the SPM images. Although many scanning probe microscopists
have so far been content with ignoring these kinds of artifacts, there are also highly credible reports on
unambiguous observations on scanning probe tip changes during data recordings that led to blunt tip artifacts
in SPM images [6,7]. One of these reports proposes that multiple mini-tips cannot affect the character of the
observed translation symmetry in such an image while the 2D periodic motif may be smeared out [6].
Our theoretical analysis [4] confirms this idea so that one can confidently take “inconsistencies” between
observed 2D translation and point symmetries in SPM images (Fig. 1) as the hallmark of multiple mini-tip
artifacts. Our unambiguous determination of the underlying Bravais lattices of 2D periodic surface arrays [3]
on the basis of a geometric Akaike Information Criterion (AIC) [8] achieves the detection of multiple mini-
tip artifacts on a statistically sound basis. Figure 1 demonstrates the effectiveness of our version of CIP in
removing blunt tip artifacts from a simulated scanning tunneling microscope (STM) image of long-range
ordered 2D periodic arrays of cobalt-phthalocyanine molecules on a (001) oriented gold surface. Note that
the central circular area in the left part of Fig. 1 illustrates both (i) an inconsistency between 2D translation
and point symmetries and (ii) the fact that structural scanning probe tip changes during the recording of
experimental data cannot change the character of the observed translation symmetry in the resulting SPM
image.
Our recently developed unambiguous translation symmetry determination procedure [3] also constitutes
progress towards making CIP more objective in both electron crystallography by TEM and surface feature
assessments by SPM. This is because not all plane symmetry groups are disjoint. As it is well known, all of
the symmetry operations of a plane symmetry group are contained in its so called translationengleiche
minimal supergroup [9]. The application of the traditional quantifiers of the deviations of experimental
images from their plane symmetry enforced counterparts (as used in electron crystallography) can, therefore,
never be completely objective [1,4]. Because the practitioners of electron crystallography have typically
chemical intuition (and sometimes prior knowledge) about the atomic arrangement they are trying to solve
and often work with multiple 2D projections of the same 3D crystals, this lack of complete objectivity is
tolerable in that field. Scanning probe microscopists that are trying to utilize CIP in their field are less
fortunate as they need to work from one 2D projection only. Geometric AICs [8] have specifically been
developed to select the best model out of a set of non non-disjoint models so that their application in CIP for
Paper No. 0804
1611
doi:10.1017/S1431927615008831 © Microscopy Society of America 2015
Microsc. Microanal. 21 (Suppl 3), 2015