Atomic-scale Secondary-electron Imaging in the STEM and SEM

Since early days, the secondary-electron (SE) signal has provided the main imaging mode in a scanning electron microscope (SEM), providing topographical contrast due to variation in the number of secondaries generated within an escape depth of the surface. SE image resolution was originally no better than 10 nm (see Fig. 1) and thought to be limited by delocalization of the plasmon-mode inelastic scattering of primary electrons that gives rise to SE generation. But when field-emission STEM instruments were employed, SE resolution became closer to 1 nm [1,2] and as a result of the development of a Cs-corrected objective lens, single uranium atoms (on a thin carbon film) and atomic columns in YBCO superconductor could be resolved [3]. This atomic-scale resolution was interpreted [4] by assuming that the SE signal can be written as S = G T B D, where G is the number of SE generated by each primary electron, T is the fraction transmitted to the surface, B represents the fraction that overcome a surface-potential barrier, and D is the fraction of emitted SE that are detected. Each of these terms gives rise to a contrast mechanism (for example, T enables topographical contrast and B allows work-function contrast), and each has an associated resolution. For a 0.1-nm probe, contrast arising from the generation term (G) can have atomic dimensions because the point-spread function (PSF) for inelastic scattering has a narrow central peak and the other terms vary little on an atomic scale. As shown in Fig. 2a, several atomic shells contribute to SE emission, giving a full width at half maximum (FWHM ∼ 0.1nm) small enough to account for the observation of single heavy atoms on a thin substrate. The situation for atomic columns is more complicated and has been analyzed by several authors [5–7]. SE are produced at various depths within a crystal and are ejected in different directions; some excite other secondary electrons and those created within the escape depth contribute to the SE signal. In this case, the different atomic shells may contribute in proportion to their stopping power, rather than inelastic cross section, giving slightly better resolution; see Fig. 2b. The original observations were made using 200keV primary electrons and thin (<100nm) crystals but it is interesting to speculate whether atomic resolution could be achieved for a thick (bulk) specimen and at lower energies, using an aberration-corrected SEM. For this purpose, we prepared wedge-shaped silicon and strontium titanate (STO) samples, whose atomic columns were imaged using scattered electrons (recorded by a HAADF detector) and secondary electrons (recorded by a through-lens SE detector). As the thickness increased from 50 nm to 18 μm, the HAADF image contrast fell by a factor of more than 10, whereas the SE contrast remained nearly constant; see Fig. 3. To qualify as a bulk sample, the thickness should be at least half the primary-electron range, which for 200keV electrons is about 55 μm for Si and 27 μm for STO, so the results shown in Fig. 3 are encouraging. An important requirement for reliable SE imaging of atoms or atomic columns is a sufficiently clean surface. With no requirement for post-specimen lenses, the SEM can more easily provide sufficient space for in-situ specimen preparation and UHV pumping [8]. Secondary-electron resolution in SEM and STEM instruments, from 1950 to 2020. Relative contributions of different atomic shells to the total PSF (top blue curve) for inelastic scattering of 200keV electrons from a strontium atom, weighted by (a) the cross section and (b) the stopping power of each shell. HAADF (top) and SE (bottom) images of silicon, for thicknesses of 50 nm to 18 μm, with insets showing a diffractogram of each image.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.