Abstract text (incl. figure legends and references)

A Monte Carlo tool called Geant4SEM is presented for the simulation of secondary electron (SE) emission, back scattered electron (BSE) trajectories and transmission signal in a scanning electron microscope (SEM). The tool is based on the powerful open source platform Geant4 [1]. Although initially it was developed for high energy physics simulations, the toolkit is well suited for analyzing interactions of particles with matter in many areas and has been used for similar purposes before [2]. Within Geant4, physics models can easily be implemented and tested individually.

Thus far, four physics models have been implemented into the framework of Geant4. Mott cross-sections based on ELSEPA [3] have been used for the elastic scattering model. A dielectric function theory approach for cross sections of inelastic scattering and generation of SEs coming with JMONSEL [4] according to Shinotsuka et al. [5] has been adopted. Additionally, an inelastic scattering model for electron-phonon interaction based on Ganachaud and Mokrani [6] has been implemented as well as a model for boundary crossing of electrons. The detection of electrons can be set to occur right after exiting the sample or at a distance and can be filtered by electron properties such as kinetic energy and emission angle.

Aforementioned models have been validated against experimental results. Figure 1 shows a comparison of simulated SE yields of gold at multiple primary electron energies with other simulations and experimental results of Walker et. al. [7].
Fig 2 displays signal profiles of an SEM image of a nominally 100 nm high and a 250 nm wide silicon step on a silicon wafer at a beam energy of 1.3 keV and an equivalent simulation. To achieve a quantitative comparison between the two, the exponential decay of the SEM signal has been fit to the simulated data and to the over multiple lines averaged measured data from the left edge towards the center of the silicon step sample. The exponential factor of our simulation lies within the margin of error of the experimental data's fit.

It is demonstrated that Geant4SEM is capable of quantitative predictions of experimental SEM images. It deepens our understanding of the SEM imaging process and enables accurate measurements of nanosize structures such as semiconductor structures, fibers and particles of varying sizes, shapes and materials.

[1] S. Agostinelli et al., Nucl. Instrum. Methods Phys. Res. Sect. A 506, 250-303 (2003)

[2] E. Kieft, E. Bosch, J. Phys. D: Appl. Phys. 41, 215310 (2008)

[3] F. Salvat et al., omput. Phys. Commun. 165, 157-190 (2005)

[4] J. Villarubia et al., Ultramicroscopy 154, 15-28 (2015)

[5] Shinotsuka, Tanuma, Powell, Surf. Interface Anal. 51, 427-457 (2019)

[6] J.P. Ganachaud, A. Mokrani, Surface Science 334, 329-341 (1995)

[7] Walker et al. Scanning. 30. 365-80. (2008)