Scanning electron microscopy (SEM) is frequently used for the characterization of nanoparticles (NPs) and imaging with backscattered electrons (BSEs) is particularly interesting to reveal, e.g., contamination NPs in a NP-ensemble. However, the SEM contrast of samples with complex geometries, compared to flat bulk samples, cannot be quantitatively described by common theoretical models [1]. In this work we will show that a) the BSE SEM contrast of SiO₂ NPs on a complex substrate strongly depends on the primary electron energy E₀, working distance WD and the used substrate and b) that Monte Carlo (MC) simulations are well suited to model and optimize the NP-contrast.

For this purpose SiO₂ NPs with diameters from 50 nm to 110 nm were deposited on two different substrates. The first substrate is interesting for correlative SEM and light microscopy imaging and consists of glass slides coated by electrically conducting indium-tin-oxide (ITO) with 180 nm thickness [2]. The second substrate type consists of amorphous (glassy) carbon, which is covered by only 20 nm ITO. A FEI Quanta 650 FEG equipped with an annular semiconductor BSE detector mounted below the objective pole piece was used. E₀ between 3 and 17 keV and WDs between 4 and 12 mm were chosen. MC-simulations were performed with a modified version of NISTMonte program [3] employing screened Rutherford and Mott cross-sections (CSs) for comparison with the measured data. The baseline intensity I_black was recorded with blanked electron beam. The NP-contrast was calculated by C=(I_NP-I_sub) / (I_sub-I_black), where I_NP is the NP-intensity and I_sub the substrate intensity.

Figs. 1a,b show 5 keV BSE SEM images of SiO₂ NPs on the 180 nm ITO/glass substrate taken at WDs of 10 mm (Fig. 1a) and 4 mm (Fig. 1b). Although the same object is imaged, contrast inversion of SiO₂ NPs is observed. Fig. 1c shows a 5 keV BSE SEM image (WD = 10 mm) of SiO₂ NPs on the 20 nm ITO/carbon substrate where NP-contrast inversion can be observed compared to the 180 nm ITO/glass substrate (Fig. 1a). The images in Fig. 1 indicate that simple interpretation of BSE SEM images in terms of material contrast is not adequate for complex sample structures.

The experimental and simulated NP-contrast is in detail studied by systematically varying the WD for E₀ = 5 keV (cf. Fig. 2). While the NP-contrast for the 20 nm ITO/carbon substrate approaches zero with increasing WD, there is a contrast inversion for the 180 nm ITO/glass substrate at WD ~ 6 mm. We attribute this contrast inversion to the anisotropic angular BSE scattering characteristics, whereby the scattering angle range of collected BSEs is controlled by the WD.

The dependence of the NP-contrast on E₀ for a constant WD = 10 mm is presented in Fig. 3. Contrast reversal occurs at ~4.5 keV for SiO₂ NPs on 20 nm ITO/carbon and at ~10 keV for NPs on 180 nm ITO/glass. The NP-contrast for larger E₀ is in general higher on the ITO/carbon substrate due to the small ITO thickness and low BSE intensity from the carbon substrate below. Converging C-values for low E₀ indicate a) that the primary electrons do not even penetrate through the 20 nm ITO-layer anymore and b) that contrast inversion for the different substrates is related to the ITO-thickness. Another contrast inversion stands out, if both substrates are compared directly as highlighted in Fig. 3 by a red arrow at 5 keV. Additional MC-simulations are included in Fig. 3 assuming hypothetical substrates with 20 nm ITO on glass (dashed light-blue line) and 180 nm ITO on carbon (dashed purple line). The additional simulations demonstrate that the contrast inversion is also ITO-thickness dependent and not substrate-material dependent, because contrast inversion does not occur for ITO-layers with the same thickness on different substrates. MC-simulations with screened Rutherford CSs describe the NP-contrast well while simulations with Mott CSs (not shown here) show larger deviations from the experimental data.
To summarize, two unexpected effects were observed for BSE SEM contrast of SiO₂ NPs: a strong dependence on the used substrate, in our case especially the ITO layer thickness, and a “geometrical” contrast inversion which can be controlled by the WD. Optimum NP contrast is obtained for small E₀ and WD-values.

References

[1] H. Niedrig, J. Appl. Phys., 53 (1982), pp. R15-R49.

[2] H. Pluk, et al., J. Microsc, 233 (2009), pp. 353–363.

[3] N.W.M Ritchie, Surf. Interface Anal., 37 (2005), pp. 1006–1011.

Figures:

Fig. 1: 5 keV BSE SEM images of SiO2 NPs on a) 180 nm ITO/glass with WD = 10 mm, b) WD = 4 mm and c) on 20 nm ITO/carbon with WD = 10 mm. Brightness and contrast were adjusted for better visibility.

Fig. 2: Measured (solid symbols) and simulated (empty symbols) contrast of SiO2 NPs from BSE SEM images as a function of WD (E0 = 5 keV) on two different substrates (lines guide to the eye).

Fig. 3: Measured (solid symbols) and simulated (empty symbols) contrast of SiO2 NPs from BSE SEM images as a function of E0 (WD = 10 mm) on two different substrates (20 nm ITO on carbon and 180 nm ITO on glass) and additionally simulated NP BSE contrast on hypothetical substrates (20 nm ITO on glass (dashed light-blue line with “x” symbols) and 180 nm ITO on carbon (dashed purple line with “+” symbols)). Lines are guided to the eye.

« Back to The 16th European Microscopy Congress 2016

EMC Abstracts - https://emc-proceedings.com/abstract/backscattered-electron-sem-contrast-of-sio2-nanoparticles/