Polishing effects on Silicon Carbide

John Wilson
The Institute of Optics, Opt 307
Rochester, NY

Abstract

The effects of grinding and polishing silicon carbide (SiC) are examined by use of several different instruments. Sub-surface damage is viewed using a scanning electron microscope. Micrographs of the crystalline structure are recorded for different production methods. X-ray analysis is used to confirm the purity of the sample and identify anomalies.

Introduction

Silicon Carbide (SiC) is a polycrystalline material that is hard and has uses as an abrasive, mirror, and semi-conductor. It can occur naturally but is typically produced in a manner of different ways. Different means of production will result in different surface structures that should be visible on an SEM. Magnetorheological Finishing (MRF) is a deterministic polishing technique that can be used to polish samples of SiC. To quantify the surface of the samples, the RMS roughness will be examined. This is a measure of how far the peaks and valleys are from the mean surface level. Different amount of polishing times will result in different RMS roughness values that can be measured.

Even though it can be used as a semi-conductor (and looks like a stereotypical metal), charging can be a problem for SEM imaging. Typical images of SiC are blurry and there is significant trouble in obtaining clear images with magnifications greater than 10,000x. The structure of SiC is made of grains which are typically 10-20 microns in size. Relatively low magnifications can be used to image these grains which mitigates the imaging troubles.

Experimental Procedure

Three main pieces of SiC were obtained which had been ground to various levels of smoothness. Table 1 shows the three samples along with their average RMS roughness and the grinding tool used. The three main samples had spots that had been polished by MRF for different amounts of time. The longer the amount of time polished, the more material removed from the surface.

Sample nameGrinding toolRMS roughness (nm)
Rough ground~40 micron bronze tool131
Medium ground10-20 micron bronze tool86
Fine ground2-4 micron resin tool29

Table 1- Main sample properties

The easiest way to minimize charging is to coat samples with a small layer (angstrom sized) of a conductive material, for example gold. The samples used in this project were on loan so no changes could be made to the surface other than fiducial marking and gentle cleaning. Charging was controlled by using small apertures (to limit the amount of electrons reaching the surface) and low accelerating voltages (typically 2 KeV).

Images and Results

Overview

A Leica light microscope was used to initially examine the surfaces to get a sense of what could be found there. Table 2 shows images of the three main samples taken outside the polishing spots. The top images were taken with the light microscope using nomarksi objectives at 100x (highest magnification available). The SEM micrographs on the bottom row were taken with SE and are at a fairly low magnification for the SEM. All SEM images in this project are SE since BSE images did not contain any useful images due to the purity and uniformity of the sample. Table 2 shows the two samples of SiC that were created in different ways. The light microscope image of sample #2 is taken at the edge of a polished spot found on the surface.

Rough groundMedium groundFine ground

Table 2- Main sample overview

Sample #1Sample #2

Table 3- Secondary sample overview

The RMS roughness was measured for four different spots on each of the three main samples. Five measurements were taken at each spot with a white light interferometer (Zygo Newview 100) at the location of deepest penetration. After four minutes of polishing, the change in surface roughness leveled out. Sixteen minutes did little to improve the roughness and actually made it worse for the rough ground sample (compared to four minutes). Figure 1 shows the data plotted on a log scale graph.

Figure 1- RMS roughness values for polishing times

X-ray analysis was performed on all of the samples. No significant difference was found between samples and a typical qualitative plot is shown in Figure 2. The two main elements are silicon and carbon which is what the sample should have been made of. The trace amounts of oxygen were not always present and represent some impurities in the sample. X-ray mapping showed uniform distribution between the carbon and silicon throughout the samples. Changes in accelerating voltages resulted in different counts for silicon and carbon but did not excite any new elements (4-20 KeV range used).

Figure 2- Typical X-ray counts

Sub-Surface Damage (SSD)

The three main samples have three distinct types of sub-surface damage visible from the overview pictures. Table 3 shows examples of the different types of damage. The first type is the long trenches which are created from the grinding tools. The distance between trenches and their depth is related to the size of the grinding tool used on the surface. These trenches are also made during the polishing process although these are shallower and more densely packed (related to the size of the abrasive used in the polishing process). The pits are visible sub-surface damage which are also caused by the grinding process. Chunks of material are chipped away during grinding leaving these pits. The semi-circular shape of the pits comes from the rotating tool used in grinding. Tooling marks can clearly be seen in the fourth image (d) as periodic curved lines. The least common damage observed was lateral cracks. These are most likely caused by pressure from the grinding tools. The frequency of these cracks is related to the fracture toughness of the material.


a) Grinding tool trenches (medium ground)

b) Grinding tool trenches (rough ground)

c) Trenches from grinding and polishing inside spot (medium ground)

d) Tooling marks visible (rough ground)

e) Lateral cracks (medium ground)

f) Pits and shallow trenches (rough ground)

Table 4- Sub-surface damage

Crystalline Structure

Also visible in these images are the different crystalline structures of the material. These are present as clearly defined lines between light and dark areas on the sample. Figure 1 shows this structure inside the 2 minute spot on the medium ground sample. The structures are easier to see inside the spots since there is less SSD to obscure the image.

Figure 3- Crystalline structure

Of the two secondary samples of SiC that were produced in different ways, the second sample had clearly defined crystalline structures. An interesting effect of the way this sample was created is a radial "flower". These "flowers" are shown in Table 5 c,d and are very common across the entire surface.


a) Crystalline structure of sample #2

b) Higher magnification of crystalline structure

c) "Flower" pattern in lower right section

d) Close up of "flower" pattern

Table 5- Crystalline structures

Phenomenon

On the outer edge of the fine ground sample was a small section of the surface that had not be processed in any way after arriving from the manufacturer. This outer edge had several interesting artifacts that were not found on any of the surfaces that had been ground or polished. Interspersed about the rough surface were particle like objects approximately 10 microns in size. Table 6 shows a collection of micrographs showing these particles. Using x-ray analysis it was shown that these particles were made of silicon. Also found throughout the surface, although less frequent, were highly spherical objects only a few microns in size. X-ray analysis showed these particles to be made of iron. An example of one can be seen in the third (c) and fourth (d) picture in Table 6.


a) Overview of rough surface

b) Valley with several silicon particles present

c) Silicon particles with small iron particle on top

d) Close up of silicon particle with iron particle on top

Table 6- Silicon and iron particles

X-ray maps

An x-ray map was constructed which scanned a sample surface and recorded the location of x-ray counts. Several regions of interest were selected which corresponded to elements likely to be present. This produced a map of the surface that shows where the different elements are located. Table 7 shows the maps for each element of interest on a surface that included both a silicon particle and an iron particle. The absence of iron and oxygen at the upper middle section is most likely caused by an obstruction further up (north) which is where the x-ray detector was located. The less energetic x-rays emitted by the interactions with the carbon and oxygen atoms could not penetrate the obstruction and thus were not detected.


a) Secondary electron image

a) Carbon

a) Silicon

a) Iron

a) Oxygen

Table 7- X-ray maps

The images in Table 8 help to further illustrate the effect topology has on x-ray analysis. An x-ray map was made over an area that included three large valleys. No x-ray counts are obtained from these holes since the x-rays can not make it through the obstructions to the detector on the left hand side. An anaglyph was made to show that these are indeed valleys and helps visualize the problem.


a) Secondary electron

b) Carbon

e) Anaglyph

c) Silicon

d) Oxygen

Table 8- X-ray maps and anaglyph

Depth of Field

One of the main advantages of SEM over light microscopes is the greater depth of field. This is clearly demonstrated in Table 9 where the micrograph on the right was taken with the SEM and the image on the left was taken with a light microscope. The SEM can easily bring the entire particle into focus however the light microscope can only focus on a small section at a time. To see a movie of the light microscope changing focus on the particle, view the movie here.

a) Light microscope image

b) SEM micrograph

Table 9- Depth of field difference

Nomarski

Differential interference contrast imaging was used were used for all the light microscope images in this project. This was accomplished by using a Nomarski objective whose main component is a prism made of a birefringent material (usually calcite) which splits an incident beam into two closely spaced beams of different polarizations. These beams interact with the surface and in places where refractive index changes rapidly (or height) the beams will interfere differently when recombined. This can be used to give images a three dimensional effect since different heights can have different colors. Figure 4 shows a 16 minute spot polished on the rough ground sample of SiC. Since the spot was polished for so long, there is a large difference in height between the spot and sample surface. The nomarski objective gives this height gradient a color.

Figure 4- Light Microscope image with Nomarski objective

Anaglyphs

Anaglyphs provide a way of seeing three dimensional data using two dimensional micrographs. Two images of an object are taken at slightly different angles. These images are overlaid with each other and colored blue and red. Special glasses are then used which allow the brain to construct a 3-d image. Normally the center point of the sample is located and the working distance adjusted until it is at the eucentric working distance. The day these anaglyphs were made the motorized stage was broken so another method was used. A micrograph of the image was taken at 0° tilt and key features of the object were circled on the monitor using a dry erase marker (permanent markers are not recommended). The sample was tilted 3°-5° and the circles on the monitor were lined up with their associated features. The images were made in Photoimpact using the instructions provided by Brian McIntyre. To see the three dimensional image use a pair of glasses that has a red filter on the left eye and a blue filter on the right eye.



Conclusions

The grain structure of SiC was clearly visible on several different samples. These grains were within the expected size range and were better observed on smoother surfaces. The poly-crystalline structure of SiC was also demonstrated by the appearance of different shades on the surfaces. According to the x-ray analysis these surfaces were uniformly made of carbon and silicon so the shades were not a result of different atomic numbers. Sub-surface damage was apparent in the samples that had been ground and polished.

The as received surface at the edge of the fine ground sample contained many artifacts that were not observed on any processed surface. The particles of silicon and iron are most likely debris left over from either the creation process or from the environments it was exposed to. With enough cleaning it should be possible to completely remove these artifacts.

Acknowledgments

I would like to thank Shai Shafrir for several things: giving me the idea for the project, loaning me the samples, and offering advice. Brian McIntyre played a pivitol role in this project and deserves thanks for such things as: teaching me how to use an SEM, reminding me how to use an SEM, and offering many good ideas on how to take better pictures.

Please enter any comments, criticisms, questions, etc. below.

Your name:

Email address: