Friday, April 20, 2012

Beam Energy and Topography: SE1 and SE2

This is an SEI (secondary electron) image of the exterior surface of the EDS window. The Moxtek AP3.3 window consists of a silicon support with a hexagonal mesh to provide ~ 77% open area covered by ultra-thin layers of polymer, aluminum (to prevent charging) and a proprietary sealant called DuraSeal. The EDS window is on the order of nm in thickness. As such, if it is thicker than the IMFP of the low energy secondary electrons used to generate this image-- why is the grid visible?

There are two categories of secondary electrons. There are those that originate from the interaction of the primary beam with the sample. These are called SE1 and come from the vicinity of where the SEM beam interacts with the sample. It is also possible for elastically scattered electrons-- backscattered electrons-- to generate secondary electrons as they leave the surface. These are called SE2. Since the silicon mesh is of a higher density than the ultra-thin window, there are more SE2 electrons being generated near the mesh-- and thus the mesh is visible in SEI mode. We're really not seeing through the window!

While the 30kV image was dominated by information from the grid due to the increased SE2 signal, at 2.5 kV the image is entirely driven by SE1 signal. In some SEM's the SE1 and SE2 signals can be discriminated against by changing the parameters of the ET detector, but that is not possible with the JEOL 5900. Dropping the beam energy to 2.5 kV has the same effect.

The morphology of the window dominates, and one can see little "pillows" of window in the regions of the mesh pores. This window failed by a blow-out when the vacuum of the detector was suddenly vented to air, blowing the window into the chamber (the present sample was found on the bottom of the SEM chamber).

This image shows the BEI TOPO image-- topographic backscattered electron image-- taken from the exterior surface of the grid at 30 kV. The regions corresponding to the "pillows" seen above in SEI at 2.5 kV are evident by subtle contrast in the BEI TOPO image. Each pillow region is light on the left side and dark on the right side suggesting that the window is pushed out away from the silicon grid supporting it. This would be consistent with the window failing by blow-out into the SEM chamber.

The final image tests the morphology inferred by the SEI image at 2.5 kV and the BEI TOPO image at 30 kV. It is an SEI image at 2.5 kV at very high tilt. The window can clearly be seen pushing outwards from the pores in the grid. The fact that the window is still intact in these regions is a testament to the high tensile strength of the window material. The EDS detector window is spec'd to withstand 10,000 cycles to atmosphere where the SEM chamber is vented with the detector interior at vacuum. This represents the extreme opposite case.

The point of this example is to illustrate the beam energy and choice of detector can have a profound effect on the morphology observed in SEM. If this system were imaged only at 30 kV, the morphological change of the window due to the blow-out would not be seen.

EDS Detector Window

This is an image if the blown EDS window. The right portion shows the exterior surface (towards the chamber) and the left portion shows the interior surface (towards the detector). The interior surface shows clearly the bright hexagnoal silicon grid for supporting the actual window.

Thursday, April 12, 2012

EDS Detector Window Absorption

This graph shows the transmission through several different EDS detector windows. The solid line is for the Moxtek AP1.3, the dot-dashed line for the Moxtek HT2.2 and the dashed line for the Moxtek AP3.3. Our EDS detector has an AP3.3 window. The graph is borrowed from F. Scholze and M. Procop, X-Ray Spectrom. 2005; 34: 473–476. There are all light element EDS windows-- meaning they are for doing very low energy spectroscopy-- and contain C, N and O in a polymer film that is coated with a very thin layer of Al and often a B compound. The absorption edges of B (0.18 keV), C (0.28 keV), N (0.40 keV), O (0.53 keV) and Al (1.56 keV) are all visible in this transmission data measured at the beamline at Physikalisch-Technische Bundesanstalt in Berlin. These absorption edges must be considered in modeling X-ray transmission in standardless EDS. The complexity of this low energy transmission function itself argues for the use of standards.

This last image shows the transmission through an AP3.3 window that is contaminated with water at a level of 2 µg/cm2. The AP3.3 window is shown by the solid circles, the contaminated window by the solid line. Note the drastic changes in window transmission. While the bare window data can be used to determine the absorption of any window contamination layer, small chemical shifts can change the absorption edge structure and thus the detection efficiency in the near the edges.

Kaersutite: Low Energy EDS

This is EDS taken from a sample of graphite coated Kaersutite which has the formula NaCa2(Mg4Ti)Si6Al2O23(OH)2 as an example of the low energy performance of the EDS detector with its new Moxtek AP 3.3 window. Unresolved O Kα and Ti Lα are clearly seen, as are Na, Mg, Al and Si Kα. Some of the Fe L lines can be seen between Ti Lα-O Kα and Na Kα.

Optimal WD for EDS

The EDS detector does not have any inherent focusing properties. However, there is an optimal working distance that allows the maximum number of X-rays to enter the EDS detector. This is entirely a geometric effect and reflects the limitation of solid angle entering the EDS detector electron trap.

In this graph Cu Kα signal was measured from Cu foil as a function of working distance. The count rates per nA are quite small given that the EDS detector was operated at only 10% dead time. There is an obvious plateau around 15 mm. Previously the EDS detector was optimized for a working distance of 10 mm. This optimal working distance was lowered slightly to allow for larger EDS elemental maps. Clearly there is little significant change in signal over a large range of working distance from 12 mm to 18 mm.

It should be noted that this does slightly change the effective take-off angle, ψ, of X-rays leaving the sample-- by about 1.5°. This does have a slight impact on quantitation where the A or absorption correction is a function of the cscψ. The differential effect on the A-correction factor is no more than 3.5%. In general, in standardless quant this is of little concern, but for those doing quant with standards, working at a working distance of 10-12 mm is recommended.

Monday, April 9, 2012

Digital Processing Using Convolutional Kernels

One of the advantages of a fully digital SEM such as the JEOL 5900 is the ability to use digital processing tools to manipulate images. This manipulation may simply be noise filtering. Image granularity may be removed by averaging longer per pixel or by increasing the probe spot-size (in cases where image is not degraded by increasing the probe diameter). Such noise can also be removed by frame averaging provided that there is no significant drift of the image field between subsequent frames. However, the digital format of the image allows the application of digital image processing techniques.

As an example, the above image of latex spheres can be processed using a Laplacian edge-detection kernel. Each pixel can be averaged with its nearest neighbors using a special weight to produce a second spatial derivative of the image. Such a spatial derivative will serve to emphasize the edges of structures and thus the Laplacian kernels act as image edge enhancement tools. An example of a Laplacian kernel is:

[0, -1, 0]
[-1, 4, -1]
[0, -1, 0]

This image is produced using a Laplacian kernel (but not the one above). Note that it acts to enhance the edges of structures. Isolated spheres now look like hollow rings, where spheres that were closely packed together with little secondary electron contrast between them somewhat merge together.

Other possibilities are the sharpening and smoothing kernels. These can be applied separately or together to a digital image. The final image combines both to make a sharper and cleaner image. Since these are convolutional filters, the more nearest neighbors used in filtering will reduce image graininess, but like all convolutional filters, will result in some blurring of the image as high frequency spectral information is lost. Custom kernels can also be made that combine sharpening, derivative and smoothing functions. In the case of SEM images where there is a fast and slow scanning direction, it is not uncommon to have different types of noise in the vertical and horizontal directions of the image. In such cases, novel asymmetric kernels can be of great utility.