Improving Backscattered Electron Images

Backscattered electrons (BSEs), the result of elastic scattering between the electron beam and sample, carry information about the constitution of a sample.

The collection of good quality data using a backscattered electron detector (BSD) depends on several different factors such as the BSD used, the detector electronics and also the sample’s constitution, shape and conductivity. This article will explore using the same BSD to image the same sample to explore the different factors that impact image quality. These include the beam intensity, the number of imaging frames, the chamber pressure and the working distance.

BSE Imaging in an SEM

The elastic scattering of electrons in the primary beam, which closely pass the nuclei of atoms in the sample, produce BSEs. The image quality varies depending on which BSD is used, and its internal electronics, and the samples conductivity. However, if a specific system is used with a fixed detector what other factors influence BSE image quality?

Comparing the Quality of BSE Images

The peak to signal noise ratio (PSNR) is a useful way of comparing the quality of different BSE images.

Where nx and ny are the pixel dimensions, 10log10 is the conversion in dB, and t(x,y) and r(x,y) are the test and reference images. The formula takes the maximum pixel intensity in the reference image (i.e. the signal) and divides it by the intensity difference between the test and reference image (i.e. the noise) to provide a signal:noise ratio.

If more integrated frames are used when capturing an image there will be lower noise. Figure 1 shows images of the same area captured using 1, 4, 16 and 32 integrated frames. The more integrated frames used the better the data quality/PSNR:

The reference image for the PSNR equation was set as the 32 frame image. The results are shown below.

BSE images of the same area acquired with the same beam settings with different number of integrated frames, from 1 to 32.

Figure 1: BSE images of the same area acquired with the same beam settings with different number of integrated frames, from 1 to 32.

The Impact of the Probe Current

The quality of a BSE image varies depending on the current of the electron beam with images taken at higher currents producing sharper images than images taken at lower currents. Figure 2 shows four images of the same area which were taken at the same beam voltage (10 kV) but with different beam intensities – 180 pA, 330 pA, 0.9 nA and 5.7 nA

BSE images of the same area acquired with the same beam voltage and number of integrated frames, at different beam intensities: low, image, point and map.

Fiureg 2: BSE images of the same area acquired with the same beam voltage and number of integrated frames, at different beam intensities: low, image, point and map.

The reference image for the PSNR calculation was set using the image at the highest beam intensity (5.7 nA). The results are shown below.

The results agree with the hypothesis that a greater beam current provides a sharper image with a better PSNR.

The Impact of the Working Distance

The working distance itself has no impact on how many BSEs are produced by the sample-beam interaction. However, the working distance (which is the distance between the sample and BSD) does impact which BSEs can be detected.

Figure 3 illustrates this point showing the difference between when a sample is placed close to the detector (WD1) and further away (WD2). The angle over which the BSD can collect BSEs (akin to its sight) is greater for WD1 (α) than for WD2 (β).

Schematics of different geometry, where the working distance is varied, from short (WD1) to large (WD2). With the short working distance, the collection angle is larger than that of the large working distance.

Figure 3: Schematics of different geometry, where the working distance is varied, from short (WD1) to large (WD2). With the short working distance, the collection angle is larger than that of the large working distance.

The increase in working distance affects the collection angle of the BSE detector, as shown in Figure 4. As α is the collection angle at the shortest working distance, the ratio between the angle α and β, the angle at WD2, increases linearly with the working distance.

Meaning that the further away the sample is positioned, the smaller the collection angle of the backscattered electron detector, the “noisier” the image, as shown in Figure 5.

Ratio between the angle at the shortest working distance α and the angle at large working distance β for different sample position.

Figure 4: Ratio between the angle at the shortest working distance α and the angle at large working distance β for different sample position.

The effect of the working distance on the noise in the BSE image. The scale bar is 5 µm.

Figure 5: The effect of the working distance on the noise in the BSE image. The scale bar is 5 µm.

If different images of the same area are taken using the same beam settings, but different working distances, it is possible to determine the noise increase associated with increasing the working distance.  Figure 6 shows four different images taking at working distances ranging between 7.69 mm and 10.7 mm.

On the left, the reference image is taken at a working distance of 7.69 mm. The other three images of the same area are taken with the same beam settings at different working distance (8.47mm, 9.48mm and 10.7mm). The horizontal field of view in the four micrographs is 179 µm.

Figure 6: On the left, the reference image is taken at a working distance of 7.69 mm. The other three images of the same area are taken with the same beam settings at different working distance (8.47mm, 9.48mm and 10.7mm). The horizontal field of view in the four micrographs is 179 µm.

The PSNR can be calculated by using the image taken at the smallest working distance as a functional reference (Figure 7). As predicted the PSNR ratio decreases (i.e. there is more noise) as the working distance is increased – the larger the working distance, the less clear an image will be.

PSNR calculated from the images in Figure 6.

Figure 7: PSNR calculated from the images in Figure 6.

The Impact of the Pressure Chamber

The pressure of the pressure chamber also has a significant effect on BSE image quality.

Under high vacuum conditions more BSEs emitted from the sample succeed in reaching the detector without scattering from gas molecules in the chamber (left image, Figure 8).

Under low vacuum conditions more gas molecules are present which means more BSEs are scattered and don’t reach the detector (right image, Figure 8). For this reason when everything else is constant, high vacuum conditions result in sharper images than low vacuum conditions.

The effect on the collection of BSEs by the detector due to high chamber pressure (left) and low chamber pressure (right).

Figure 8: The effect on the collection of BSEs by the detector due to high chamber pressure (left) and low chamber pressure (right).

To illustrate this point images were collected (Figure 9) at different chamber pressures (1 Pa, 10 Pa and 60 Pa) and different acceleration voltages (5 kV and 10 kV). All images were taken using four frames other than the reference image which used 32 frames at 1 Pa pressure.

BSE images acquired in the same area, at the same working distance, by varying the chamber pressure (1Pa, 10Pa and 60Pa) for 5Kv and 10kV acceleration voltage. The reference image is acquired at the same settings as that at 1Pa, but with a larger number of integration frames. The horizontal field of view in these micrographs is 26.9 µm.

Figure 9: BSE images acquired in the same area, at the same working distance, by varying the chamber pressure (1Pa, 10Pa and 60Pa) for 5Kv and 10kV acceleration voltage. The reference image is acquired at the same settings as that at 1Pa, but with a larger number of integration frames. The horizontal field of view in these micrographs is 26.9 µm.

The PSNR calculated using the images demonstrates that for chambers at higher pressures there is more noise (Figure 10). For the 10 kV beam over the course of a pressure increase from 1 Pa to 60 Pa the PSNR decreases by 6%, whereas for the 5 kV beam the effect is more pronounced with the PSNR decreasing by 30%.

PSNR calculated for images acquired at different chamber pressure for 5kV and 10kV incident beam, shown in Figure 9.

Figure 10: PSNR calculated for images acquired at different chamber pressure for 5kV and 10kV incident beam, shown in Figure 9.

This information has been sourced, reviewed and adapted from materials provided by Phenom-World BV.

For more information on this source, please visit Phenom-World BV.

Tell Us What You Think

Do you have a review, update or anything you would like to add to this article?

Leave your feedback
Submit