LEEM is a well established method for the imaging of surfaces of materials using the impact of very slow electrons (down to zero energy ‐ mirror microscopy). On the other hand, scanning electron microscopy has not consistently solved the detection of electrons in the mirror mode, i.e. in the incident energy range of 3 to 0 eV. In contrast to LEEM, which forms an integral image of the specimen surface and projects it onto an image screen, the scanning method relies on the detection of signals from individual pixels of the specimen. Since mirror microscopy can be realized only when using immersion lenses (either electrostatic, or combined with magnetic immersion), the detector has to capture the fast electrons with energies in principle of the same value as of the primary beam. Such “signal” electron beam moves back through the optical system very close to its optical axis and standard rotationally symmetric electron optics is in principle not able to provide their detection, and signal electrons return towards the source. The higher magnification is used, the more signal electrons are lost.
One solution is to use a rotationally asymmetric imaging and detection system that provides separate beam lines of primary and signal beams [1]. Any separation of these two beams requires their deviation from the optical axis. Given that the energy of signal electrons is high (close to the energy of the primary electrons) in the mirror microscopy mode, the deflection of signal electrons without influencing the primary beam quality is not a simple problem. One of possible solutions is to use a Wien filter, which does not change the trajectory of the primary beam, but only of the signal beam. However, only relatively small deflection angles of the signal beam are achievable for fast electrons. Larger deflection angles cause non‐correctable defects of the primary beam. Another solution is to use magnetic prisms, which are able to compensate for any aberrations of the second order and also for energy dispersion in a symmetrical arrangement [2]. In the asymmetric arrangement, we can reach sufficiently large beam deflection angles (90°) having the energy dispersion on the order of units of micrometers per volt. By proper combination of homogeneous magnetic fields, the beam separator stigmatically transfers the primary beam back to the optical axis, while simultaneously allowing the detection of either energy non dispersed or dispersed signal electrons (for example of secondary or backscattered). Such a through‐the‐lens detector has zero optical power in the primary beam direction. In the “standard” operation mode with grounded specimen the through‐the‐lens detector can be (but does not have to be) switched off. The primary beam then passes through the detector rectilinearly. Signal electrons can be collected with any other standard detectors for SEM microscopy.
The first experiments verifying the correctness of the concept were made with the help of an assembly consisting of an (Schottky) electron source equipped with a magnetic gun lens followed by detector unit and electrostatic triode objective lens. The specimen is connected to a variable high voltage supply floating on the cathode potential. This arrangement allows to compensate the instability of the main high voltage supply in such a way that the incident energy remains well defined. Here we present the system with only electrostatic immersion objective lens. Experiments with a combination of electrostatic‐magnetic immersion lens are under preparation. The results presented here are demonstrating the ability of the detector to distinguish between BSE and SE electrons for landing energies as low as 0.5 eV. The integral detector is mixing both signals and shows the mirror character of the image.