Type of presentation: Poster

IT-16-P-6106 Improving quantification in Scanning Electron Microscopy by comparing transmission experiments with Monte Carlo Simulation.

Walker C. G.1, Mika F.1, Frank L.1, Konvalina I.1, Müllerová I.1
1Institute of Scientific Instruments of the ASCR, v. v. i., Brno, Czech Republic
chris@isibrno.cz

Introduction
   The quantification of signals from Scanning Electron Microscopes (SEMs) is still an area of research activity despite the longevity and widespread use of the SEM in the scientific literature. Knowledge of the transmission of electrons through thin films can provide important information regarding the thickness of the films. Hence a series of experiments were carried out to compare the transmission of electrons through thin films of various materials in the range 15-30keV to establish a quantitative basis for further investigations.

                                                                                                                                                                                    
Experiment & Monte Carlo simulation
  The experiments were carried out using an FEI ultra-high resolution Magellan 400 SEM [1] and a home built scanning low energy electron microscope (SLEEM)[2]. The Magellan SEM was equipped with a multi-annular semiconductor STEM detector (Fig. 1a) which is divided into a central disk and five concentric annuli, allowing imaging in the bright field (BF) mode, four dark field modes (DF1-4) and a high angle annular dark field (HAADF) mode. The detector was placed under the sample with the BF region on the axis of the electron column. Signals that were detected through a thin film were compared with signals with no film. The films were either Si or Au and 100nm thick and each signal was normalized to the intensity of the BF signal. The Magellan SEM was operated in high resolution mode, so magnetic fields were present in the sample region. These were taken into account in the simulation.

   The SLEEM microscope was equipped with a YAG single channel scintillator screens for detecting both backscattered and transmitted electrons. The sample had zero bias and no magnetic field was applied, so the sample region was regarded as field free.
   Electron transport simulations within each material was carried out using Monte Carlo (MC) calculations and modelling of the trajectories of the electrons in the magnetic field after the electrons had been emitted from the sample using the program EOD [3].

Results & Conclusions
   Good agreement is found for the results from the SLEEM instrument (see Fig 1b). In contrast, the distribution of signals on the multi-annular detector show much greater intensities for the HAADF region in the MC simulations as compared to the experiment (see Figs. 2 and 3). This is especially the case for high Z elements. However, the overall trend across the different detectors is the same for experiment and simulation.

References
[1] www.fei.com
[2] Müllerová I., et al. Microsc. and Microan. 18 (2012), p. 996.
[3] Zlámal J., Lencová B., Nucl. Instr. and Meth. A. 645 (2011), p. 278.

The work is supported by the TA CR (TE01020118), MEYS CR (LO1212), its infrastructure by MEYS CR and EC (CZ.1.05/2.1.00/01.0017) and by ASCR (RVO:68081731)

Fig. 1: Figure 1a. The multi-annular detector. BF = Bright Field, DF = Dark Field, HAADF = High Angle Annular DF. Figure 1b. SLEEM Instrument results (100nm Si). BSE = Backscattered Electrons, TE = Transmitted electrons. The discrepancy for BSE is believed due to difficulties in determining the 100% experimental signal and unknown detector response.

Fig. 2: Figure 2. Results from the Magellan - Si 15kV 100nm. All signals are normalised to the Bright Field (BF) signal.

Fig. 3: Figure 3. Results from the Magellan - Au 30kV 100nm. All signals are normalised to the Bright Field (BF) signal.