2025 Verified technology Utilization of secondary electron contrast for dopant analysis
Developed within TN02000020 Centre for advanced electron and photon optics
ID: TN02000020/014-V02
Authors:
Adam Očkovič (CEITEC VUT)
Petr Wandrol (Thermo Fisher Scientific)
Ilona Müllerová (ÚPT AVČR)
Miroslav Kolíbal (CEITEC VUT)
Description of the samples used to build the technology:
The technology was developer utilizing two samples, both being silicon carbide (SiC) multiplayer on top of bulk SiC, grown by chemical vapor epitaxy. The n-dopant is nitrogen, dosed to the reactor during the growth process.
Fig. 1: Sample A, cross-section that was used in most analyses, consists of four epitaxial layers. The color intensity reflects the dopant concentration level.
Hardware requirements:
The methodology is optimized for use in Thermo Fisher Scientific (TFS) electron microscopes with Elstar electron column, optionally equipped with energy filtering application. The dopant contrast is carried out by secondary electrons (SE). The methodology has been developed for Through-the-lens (TLD) detection system of VERIOS 5 microscope.
Dopant contrast definition:
The normalized dopant contrast is defined as:
where the is SE signal intensity of reference bulk substrate and is SE signal intensity of any other differently doped layer. Contrast is normalized so it can reach values from 0 to 1.
1. Setting the optimum imaging conditions
Figure 2 shows the data collected on sample A to identify the optimum imaging conditions with respect to getting the maximum dopant contrast between layers with the highest and the lowest doping to reach minimal signal-to-noise ratio. The three critical parameters judged are (a) landing energy of the electrons, (b) working distance and (c) electron beam current.
- Landing energy (LE): the lower the LE the lower the signal-to-noise ratio. This criterion is critical, as the dopant contrast shows similar values if LE is above 1 keV.
- Working distance (WD): the lower WD the better contrast is achieved
- Beam current (BC): large beam currents (>0.76 nA) result in significant charging effects which cannot be mitigated. On the other hand, the lower the BC the worse signal-to-noise ratio. The optimum imaging window has been selected between 50 pA and 0.5 nA, which is reasonably wide to allow contrast tuning for specific use case.
Fig. 2: Setting the optimum imaging conditions. Left: dependence of the image contrast on the LE of electrons. Middle: Dependence of image contrast on WD. Right: Typical SE images obtained using different BC. Images obtained using higher currents are often compromised by sample charging artifacts, usage of low currents comes with worse signal-to-noise ratio. The optimum imaging window (blue arrow) has been identified as 50 pA and 0.5 nA.
2. Energy filtering of the SE electrons
The dopant contrast can be enhanced by energy filtering of the SE electrons utilizing the energy filtering module developed by TFS. The general principle of the technique lies in application of electric potential on the sample and choosing a specific potential of electrodes in the TLD detection system. The dopant contrast is carried by electrons with certain energies, hence, choosing appropriate energy window results in dopant contrast enhancement. Calculated SE energy spectra for default and optimized energy filtering settings of the TLD detector are shown in Fig. 3. Optimized settings of the detector allow to increase the contrast 2.5 times with respect to the default conditions.
Fig. 3: The dopant contrast enhancement by energy filtering of SEs is enabled by different collection efficiency of the used detector for distinct kinetic energies of SEs.
3. Image processing
The acquired images are further processed – a block diagram of this process is shown in Fig. 4. The processing consists of several steps, as summarized here. The image processing is image-dependent and has to be adjusted for each case, hence only general step description is given here. Certain processing steps are not part of this technology.
- Detector-background removal – capturing/calculating dark image and its subtraction from the dopant contrast image
- Noise filtering and digital scaling of the image
- Identification of relevant maxima in the signal
- Grey level masking of the selected areas of interest
- Contrast calculation (see Dopant contrast definition) and result plotting/exporting if more than one images are processed
Fig. 4: Block diagram of image processing for dopant contrast quantification.
Quantitative considerations
The dopant contrast could be utilized for quantification of dopant concentration, assuming that the layer stack under study contains at least one layer of known dopant level. However, this statement is valid only for calibrated detector (and microscope) settings. Fig. 5 shows that semi-log plot of SE contrast versus dopant concentration is nearly linear, similar to what has been achieved previously on silicon.
Fig. 5: SE contrast dependence on dopant concentration (calibrated by different technique) on a log scale exhibits nearly linear dependence, which could be used for quantification of SE images.