Nanotechnology places increasing demands on techniques for sample characterisation on the sub-100 nm length scale. The scanning electron microscope (SEM) is one, widely used, technique for imaging and characterising nanomaterials using the intensity of secondary (SE) or backscattered electron (BSE) emission from a probed region of the nanomaterial to generate spatially resolved contrast in an image. The acquisition of the low-loss electron (LLE) signal [1] in the SEM provides an alternative method which may offer the advantage of improved spatial resolution compositional imaging.

Spatial resolution and contrast in compositional imaging of the LLE signal has been investigated by means of experimental measurements in a scanning electron microscope and Monte Carlo simulations for the case of a semiconductor superlattice structure comprising Si_0.85Ge_0.15 of 11.5 ± 0.4 nm separated by pure Si layers at a periodicity of 69.2 ± 0.2 nm. Both continuous slowing down approximation (CSDA) and discrete-loss based Monte Carlo models were considered (NISTMonte and PENELOPE) and it was found that the calculated contrast values were particularly sensitive to the choice of model in the low-loss regime. Experimental data were obtained using a purpose-built low-energy electron loss detector [2,3] comprising a retarding field analyser with an electron-optical input lens. The detector was attached to an FEI Sirion FEGSEM. Experimental data indicated that improved contrast was obtained as the maximum loss energy was lowered (fig 1a), and this trend was reproduced by the simulations (fig 1b). The results indicate that the LLE technique is a useful alternative to operating at low primary beam energies when performing compositional imaging on samples which have nanoscale compositional structure. Statistical noise considerations that affect the LLE signal are discussed.  In the case of spatial resolution, resolution metrics for compositional imaging are discussed. It was found that the LLE signal shows improved resolution compared with the backscattered electron signal (figs 2 & 3), however CSDA-based simulations predict better resolutions than simulations based on a discrete loss model. It is found that the energy-straggling has the most significant influence on the predicted resolution in the low-loss regime. The simulations suggest that a SEM with a high-quality small-diameter probe is required to fully appreciate the resolution benefits of the LLE signal. Experimental data indicates that certain samples (such as those with sub-surface compositional inhomogeneity or nanoscale topography) benefit from the LLE technique even when the SEM used has a more modest probe diameter.

[1] O C Wells. Appl. Phys. Lett. 19 232 (1971)

[2] I R Barkshire, R H Roberts, M Prutton. Appl. Surf. Sci. 120 129 (1997)

[3] C Bonet, A Pratt, M M El-Gomati, J A D Matthew, S P Tear. Microsc. Microanal. 14 439 (2008)

Figures:

Figure 1: (a) Experimentally measured LLE signal contrast and (b) Monte Carlo simulations of LLE contrast between Si and Si/Ge multilayers (full lines), and bulk Si and Si/Ge (dotted lines).

Figure 2. Comparison of NISTMonte simulated signal emitted over 2pi sr solid angle with that emerging in the direction of the LL detector for (a) BSE signal and (b) LLE signal with E(loss)=500 eV.

Figure 3. The NISTMonte simulated LLE signal emitted over 2pi sr solid angle for a range of loss energies compared with the BSE signal for a single SiGe layer.

« Back to The 16th European Microscopy Congress 2016

EMC Abstracts - https://emc-proceedings.com/abstract/spatial-resolution-and-compositional-contrast-in-imaging-using-the-low-loss-electron-signal-in-sem/