Usual and Unusual Effects in ADF-STEM Imaging of Dopant Atom in Crystals
Anudha Mittal, Aloysius A. Gunawan, and K. Andre Mkhoyan
Department of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, MN
55455.
Interest in imaging and identifying individual dopant atoms inside crystalline specimens has a long history in
electron microscopy. While both conventional transmission electron microscopes (TEMs) and scanning
transmission electron microscopes (STEMs) are capable of imaging an individual atom [1], detecting an
individual dopant atom inside the crystal appears to be easiest using an annular dark field detector in STEM.
With recent advances in lens aberration correction ADF-STEM imaging has become applicable even for
crystalline sample with < 0.1 nm spacing between atomic columns.
Imaging dopant atoms in crystalline specimens is complicated by sample-sensitive changes in the
incident electron beam due to channelling [2]. However, despite these challenges, several groups have
successfully imaged dopant atoms inside a host: Voyles et al. [3] observed Sb atoms inside Si, Varela et al.
[4] studied La-doped CaTiO
3
, Shibata et al. [5] imaged Y atoms in Al
2
O
3
, and Lupini et al. [6] studied Bi
dopant atoms in Si. Yet, the number of cases reported in literature is too few to develop a systematic view on
parameters and conditions that govern visibility of individual dopant atom in ADF-STEM images.
Understanding the roles of the microscope parameters and specimen conditions are instrumental in designing
experiments to detect individual dopant atoms and determine their location as precisely as possible to
reconstruct the atomic structure of the doped material.
We have investigated the limits of ADF-STEM imaging to determine presence and position of
individual dopant atoms inside the host crystal by analysing simulated using the multislice method ADF-
STEM images. Specimen features, such as thickness and crystallographic orientation of the host material,
position of a dopant atom inside a specimen, Z-difference between dopant element and host, and probe
parameters that affect the visibility of a dopant atom have been examined. This analysis provides a guide for
optimization of the conditions for improving detection of a dopant atom and assessment of the conditions
under which a dopant atom is not detectable at all in ADF-STEM images.
The relationship between the position of a dopant atom and its visibility in ADF-STEM image is not
simple: several different locations of dopant atom can lead to the same contrast between doped and non-doped
atomic columns. For instance, an Sn dopant atom located on the entrance surface in 3 nm thick Si [110] has
66% visibility. Sn dopant atom, located 2.7 below the entrance surface, in 5 nm thick Si [110] has also 66%
visibility. Moreover, different types of dopant atoms can also have the same calculated visibility. For
example, a Ge dopant atom, located 1.2 nm below the entrance surface in a 2 nm thick Si oriented long [110]
direction also has 66% visibility. Thus, determination of dopant atom position requires at least knowledge of
dopant identity and specimen thickness. However, even with specimen thickness, dopant element, and probe
conditions known, two different dopant atom positions can still lead to the same visibility (see Fig. 1).
Another result showing the extent to which intensity of doped columns can be counter-intuitive is the
presence of negative values of visibility at certain specimen thicknesses although Z
Dopant
>Z
Host
. For instance,
Sn dopant atom at 9 nm depth in 25 nm thick Si [111] has -4 \% visibility (see Fig. 2). Comparison of beam
intensity profiles of doped and pure columns shows that the beam intensity of a doped column is less than the
intensity of a pure column at certain specimen thicknesses, as shown in Fig. 3. Atoms exposed to fewer
incident electrons will contribute less to ADF image intensity. Hence, a doped column can have lower
intensity than the pure host column even though Z
Dopant
>Z
Host
. Similarly, a doped column can have higher
intensity than a pure column when Z
Dopant
<Z
Host
(see Fig. 4) [7].
References
[1] A.V. Crewe, J. Wall, and J. Langmore, Science 168 (1970) 1338.
[2] J. Fertig and H. Rose, Optik 59 (1981) 407.
1302
doi:10.1017/S1431927611007380
Microsc. Microanal. 17 (Suppl 2), 2011
© Microscopy Society of America 2011
https://doi.org/10.1017/S1431927611007380 Published online by Cambridge University Press