About the role of boundary conditions on compositional imaging with a scanning electron microscope.

We consider the effects of different boundaries on the visibility of a specimen detail providing a compositional contrast in scanning electron microscopy, operating with backscattered electrons or secondary electrons. An object characterized by a gradual variation in composition, an As-doped region in Si, is investigated. The different boundaries in the cross-sectioned specimen correspond to the absence or presence of a poly-Si layer on top of the implanted region, deposited after the annealing treatment. It is shown that the interpretation model used for image formation is of paramount relevance for understanding the experimental results, indicating that the boundaries of the doped region are important in hindering or enhancing its visibility. The relevance of experimental parameters such as electron energy and probe dimension is also reported.

Low-energy STEM of multilayers and dopant profiles.

A conventional scanning electron microscope equipped with a LaB6 source has been modified to operate in a scanning transmission mode. Two detection strategies have been considered, one based on the direct collection of transmitted electrons, the other on the collection of secondary electrons resulting from the conversion of the transmitted ones. Two types of specimens have been mainly investigated: semiconductor multilayers and dopant profiles in As-implanted Si. The results show that the contrast obeys the rules of mass-thickness contrast whereas the resolution is always defined by the probe size independently of specimen thickness and beam broadening. The detection strategy may affect the bright field (light regions look brighter) or dark field (heavy regions look brighter) appearance of the image. Using a direct collection of the transmitted electrons, the contrast can be deduced from the angular distribution of transmitted electrons and their collection angles. When collecting the secondary electrons to explain the image contrast, it is also necessary to take into account the secondary yield dependence on the incidence angle of the transmitted electrons.

Electron microscope calibration for the Lorentz mode.

The calibration of a modern electron microscope for Lorentz microscopy observations has been performed using diffractogram, Fresnel diffraction fringe analysis and low-angle electron diffraction methods. An example related to the observations of electrostatic fields associated to a thinned reverse-biased p-n junction is also reported.

Backscattered electron imaging and scanning transmission electron microscopy imaging of multi-layers.

Experimental and theoretical results on image contrast of semiconductor multi-layers in scanning electron microscopy investigation are reported. Two imaging modes have been considered: backscattered electron imaging of bulk specimen and scanning transmission imaging of thinned specimens. The following main results have been reached. The image resolution of the multi-layers is, in both cases, defined by the probe size. The contrast, governed by density and atomic number differences, is affected by the size of the interaction volume in backscattered electron imaging and by the beam broadening in scanning transmission. Operating in the scanning transmission mode, the contrast of bright field images can be easily related to local variation in atomic number and density of the specimen while the dark field image contrast is strongly affected by electron beam energy, detector collection angles and specimen thickness. All these factors are able to produce contrast reversals that are difficult to explain without the support of a suitable simulation code.

Spatial resolution and energy filtering of backscattered electron images in scanning electron microscopy.

Numerical simulations of energy filtering effects on backscattered electron images of semiconductor multilayers are reported. The theoretical investigation has been performed for a wide range of energies, 1-40 keV, and for beam incidence angles between 90 degrees (normal incidence) and 20 degrees. Quite a general purpose of this research concerns the investigation of the optimum energy conditions and of their implications. It will be shown that the optimum energy defines an operating context suitable to ensure a compositional contrast enhancement; i.e. a minimum threshold current and a maximum resolution, without energy filtering, independent of the beam incidence angle. This optimum energy, depending on the specimen and its details, is, however, of the order of a few keV or less for specimen details having a size of the order of few nm. When the performance of the electron gun does not allow to work at low energy it is necessary to operate at an energy higher than the optimum one, the energy filtering can produce positive effects. Yet in those circumstances there is an optimum energy loss window suitable to minimise the threshold current. It spreads from 10-30%, depending on the primary energy and size of the compositional detail, for normal incidence, to a few per cent for high incidence angles and high energy. The simulation results for these last conditions are in agreement with the well-known experimental results obtained with the low-loss methods.

Comparison of spatial resolutions obtained with different signal components in scanning electron microscopy.

Comparative studies on the ultimate spatial resolution of the Scanning Electron Microscope, using different components of the electron signal have been performed on specimens providing compositional contrast. By operating the microscope in conventional way as well as with a specifically designed set-up we have ascertained that the delocalized components of the signal provide a spatial resolution of the order of the beam size, even if the practical use can be limited by the noise. To amplify the contribution of the delocalized components of the signal, as backscattered electrons by a bulk specimen or forward scattered electrons by a thin specimen, we used a device consisting of a plate of a material with high secondary yield placed above or below the sample. An important practical implication arises from this study. A detecting system consisting of a standard Everhart-Thornley detector coupled with a converter of backscattered or transmitted electrons represents a high performance detecting device for low voltage observations.