938 M.P. Nikiforov and D.A. Bonnell
2.3.2 “Nanotip” Models
In SPM simulations, the most common conception for Si and other
semiconductor surfaces has been that the main component of the tip–
surface interaction is the interaction of a dangling Si bond at the end
of the tip with the surface atoms. This dangling bond can be well
described using relatively small 4- or 10-atom Si clusters saturated by
H atoms.
26
Another approach is to assume from the outset that the tip
is ionic (MgO).
27,28
Calculations showed that if the bottom of the tip was
fl at, i.e., no nanotip (only a macroscopic tip), then the interaction with
the surface was averaged over several tip ions, and no contrast was
produced. When a nanotip was included, it had to extend signifi cantly
beyond the main part of the tip to achieve atomic resolution.
Silicon cluster tip models have been successful in developing a quali-
tative understanding of the origins of image contrast in SPM on metals,
semiconductors, and insulators; however, they fail in most cases to
qualitatively reproduce image contrast. The solution is to compare
images to simulations with a number of different tip models. Silicon
tips under experimental conditions are likely to be contaminated by
residual oxide, adsorbed hydrogen, and water,
29
or even materials
transferred from the surface. To compare the properties of clean and
contaminated silicon tips, the electronic structures of Si10 clusters with
adsorbed contaminant species were calculated using the density func-
tional theory by Sushko et al.
30
The results clearly showed that adsorbed
hydrogen has no effect on the potential gradient from the uncontami-
nated silicon cluster; however, adsorbed oxygen and hydroxyl groups
cause a signifi cant change in the potential gradient. Both potentials
decayed over a much longer distance than did that of the uncontami-
nated cluster, and the stronger gradients suggested a much stronger
interaction with the surface. Interestingly, the potential gradient from
an MgO cube corner with an O atom at the end was very similar to
that of the oxygen-contaminated silicon cluster, a strong negative
potential. An MgO cluster is a good model of a hard oxide tip, and has
the important advantage that reliable interatomic potentials exists for
MgO, alkali halides, and other oxides.
In an interesting experiment Bennewitz and co-workers atomically
resolved a copper (111) substrate, as well as a unit cell thick NaCl
grown on that surface.
31
An MgO “nanotip” of only a few atoms would
not atomically resolve the features observed in the experimental
images. Different “nanotip” models for SFM were compared to deter-
mine which most closely matched experimental results.
31
It was found
that a 64-atom MgO “nanotip” with an oxygen atom at the very end of
the tip imbedded in a macroscopic tip gives quantitative agreement
with image contrast. The MgO cube can also be oriented with the Mg
ion down, providing a strong positive potential. For many SFM experi-
ments on insulators, the ionic MgO tip model provides excellent quan-
titative agreement with the image. Recent achievements in instrumental
control reduce the possibility of tip contamination by sample material.
It has also become important to use conducting tip models when study-
ing insulating surfaces.