This is part of a series of blog posts – looking into the appearance and composition of commercially available sharpening stones. If you are interested in the previous episodes, check out the archive for them.
If you have some suggestion on what I should look at next, or want to share your super secret DIY stones, I could be persuaded to open the bag of analytical devices… hit me up on Instagram under @marvgro for that.
Disclaimer: I’m not for sale. Every review you see on this blog is bought with my own money. I have no affiliation to any manufacturer.
Review
Today’s sharpening stone is another PDT – this time, their “silver bond” CBN stone, at 650 grit, which they state as 28/20 µm. It uses the superabrasive CBN. According to the manufacturer homepage, it is a “vitrified metal” bond. I strongly suspect, this is the case of using the marketing buzzword of “vitrified”.
Let’s take a look under the microscope:
Optical micrographs of the PDT Silver stone. Instrument: Leica Emspira
The stone is very silver in colour – not at all like the regular bronze coloured metal bond stones. We can also make out a decent amount of black CBN particles. Let’s further look into this under the SEM:
SEM micrographs of the PDT silver CBN stone. Instrument: Zeiss GeminiSEM 560.
We can spot some very large, darker particles, but also a lot of smaller grains, of which the majority is in the size of the abrasive. This matrix looks like a regular metal bond, no signs of vitrified bond are visible here. To be fair, one can start to be very picky about the classification here: Both a metal bond as well as a vitrified bond are typically created by taking a low melting point matrix, and raising the temperature to a point where the bond matrix starts to fuse together. On metal bonds, one would call this sintering, as the “vitrification” typically implies a glassy phase, that one does not achieve with metals. It is save to say, that this stone is not vitrified, but instead a sintered metal bond.
EDS analysis of the PDT Silver 650 grit stone. Instrument: Oxford Ultim Max ∞ 40mm2 EDS sensor. Note that our EDS sensor doesn’t show elements lighter than boron.
This stone is once again a colourful firework of different elements! Let’s dig into what we can see: The largest visible particle is some titanium (blue, uppermost corner, around 8%). Titanium is typically added to CNC abrasive bonds to make them tougher. We can see the CBN grains (B, red) which are distributed with a slight tendency to agglomeration all over the image. Moreover, there’s a bit of silver in the bond (about 1%), and a high tin content (24%) compared to the copper content (25%). This explains the silver colour of the stone! Overall, the addition of these elements makes the bond harder (higher tin content), quite significantly so! The downside is, it also increases the brittleness- which is probably why there are some SiC as well as the titanium particles. Overall, it looks to me like the mix between SiC and CBN is about 1:5, making this mostly (but not exlusively!) a CBN stone. Overall, I this will be a very hard, long lasting stone. I expect it to have quite some pressure and push the material more around than it is cutting.
In order to evaluate the sharpening performance and material removal mode of this stone, a blade was sharpened with it. I am using a standardised testing procedure, read about it here. Nevertheless, it’s 65 HRC M398, and sharpened to 17 DPS with resin bond diamond stones down to 10 µm. Afterwards, the tested stone is used, first in a back and forth movement until the surface becomes homogenous, and then alternating strokes (5-5-3-2) on each side, for a total of 20 strokes from the spine towards the apex (edge trailing) per side. No pressure is applied but the weight of the apparatus.
The edge is then analysed in the electron microscope for breakouts and morphological appearance.
The stone has a rough, high amount of feedback during sharpening. A homogenous, matte surface is the result. After a couple of strokes, the edge has a detectable burr or prow formed. Let’s take a look at this under the SEM:
SEM micrographs of the edge finished with the PDT silver 650 grit stone. Instrument: Thermo Fischer PhenomXL SEM.
We can see quite a few signs of burnishing and plastic deformation here: not only is the surface showing lots of small micro burrs/prow formations, but also the cutting edge shows a very visible (bent towards the viewing direction) burr/prow. It is at least a couple of microns wide. I would believe that this is overall a sign that the stone is not cutting very freely, but creating a lot of pressure through it’s dense, hard matrix. This helps with the quick formation of something that feels like a burr – but is more plastic deformed material. With some stropping, this will likely be refined and raised to be sharper, but it does not look like a well formed apex.
Optical micrographs of the edge created by the PDT Silver stone. Instrument: Leica Emspira
Overall, I think this is a very hard, durable stone. The results are homogeneous with few deeper scratches. A well formed edge with a very regular, low waviness bevel is formed. I did not really like the plastic deformation happening to the bevel, but can now understand the “hype” on the internet about this stone – after all, it creates something very much like a burr with just a few strokes! I would expect this stone to require regular rework to renew it’s cutting capabilities, for example by etching the bond to release new, sharp grains.
Disclaimer: This post is probably going to trigger a lot of strong opinions. I don’t mind that. If you disagree with something I state, I’d love to have you show me where I went wrong! Contact details are in the impressum. If you also just want to shout at me because you disagree, that is fine as well.
In part 1 of this, my good friend Roman Kasé and I started to look into the effect of stropping on a finely sharpened blade. Let me pull up two important pictures from that to start up our discussion for part 2!
Very high magnification SEM images of the apex (top view) in a Vanadis 8 (66 HRC) blade, sharpened without (First/left pictures) and with stropping. Instrument: Zeiss GeminiSEM560
I’d like to make a point about what we are seeing here. This is a very high magnification. The entire width of the image is 1.1 µm. Compare that to the theoretical limit of an optical microscope: The Abbe diffraction limit is defined as d = λ / 2 NA – where NA is the numerical aperture, a value that can be used to describe the resoluting capabilities of a lens. At best, this value can become 1. So the theoretical resolution of any optical system is half the wavelength. The wavelength of green light is around 530 nm, so a fancy, 3 digit k$ laser scanning microscope could at best have a resolution of 0.25 µm – just shy of 5 pixels would fill the whole image here! The burr you see on the unstropped blade is about 30 nm. Compare this to the lattice constant of martensitic steel (the distance between two atoms) which is 0.286 nm. So this burr is somewhere in the region of 100 atomic layers – this is unbelievably tiny, and would appear to any easily accessible test a “no burr” result. Surprisingly enough, with stropping, the burr got actually larger and more inhomogeneous.
For the 2nd part of this series, we are looking at a different steel, Boehler M398 at 65 HRC. This is also the steel used in the vast majority of the stone reviews in my blog. It’s a popular “super steel” for knives.
As a quick reminder, for the stropping, we used fresh leather strops loaded with different emulsions with the following approaches:
1.) No Stropping – pure ground edge! this is our base truth to what we compare the stropping
2.) “best practice”: 5 strokes with 1, 0.5 and 0.25 µm diamond emulsion on leather
3.) “overstropping” – 50 strokes with the 1, 0.5 and 0.25 µm diamond emulsion.
First, we had to establish the new “baseline”, aka: Images of the apex after just sharpening, no stropping:
SEM images of the unstropped M398 blade. (Approach #1). Instrument: Zeiss GeminiSEM560
One thing of note is – M398 doesn’t sharpen as nicely as the Vanadis 8 does! Nevertheless, the “no strop” approach gave us a very nicely defined apex in the range of 100 nm, with very low burr formation. The burrs are in the 50 nm range.
Next, we look at the “best practice” of stropping, so a few passes with increasingly smaller diamond emulsions (1, 0.5 and 0.25 µm diamond emulsion).
SEM images of the “best practice” stropped M398 blade (Approach #2). Instrument: Zeiss GeminiSEM560
Once again, we get a more pronounced burr, which is sticking out further in direction towards the cutting action. The burr has increased in width as well.
Next, we are taking a look at an overstropped result. For this, poor Roman had to strop 50 passes each with the 1 – 0.5 – 0.25 micrometre progression!
SEM images of the “overstropped” M398 blade (Approach #3). Instrument: Zeiss GeminiSEM560
I find the images here very interesting. The blade bevels are much more polished – details are harder to pick out, with a smoother surface. Moreover, the apex as gotten a lot finer – but also formed large, very fine foil type burrs. While we didn’t record higher magnification images on this, the pixel resolution allows us to measure the apex width in decent accuracy, giving us an apex width of about 50 nm. Weirdly enough, this blade was the slowest to grip and split a hair, even though the apex is defined the best of all the M398 blades analysed here. I once again have to compliment Roman on his stropping technique – there is no apparent round of the apex visible! But one deduction I can take from this is: with overstropping, a relevant amount of material is removed. If you don’t keep the pressure and angle under very tight control, you will actually be able to remove enough material to round of the apex.
Last, we’re going to look at the “coarse grit” stropped cutting edge.
SEM images of the 6 µm stropped M398 blade (Approach #4). Instrument: Zeiss GeminiSEM560
One can see that the surface of the bevel becomes more scratched, and the apex rougher. I think the often cited theory, that coarse grit stropping gives an edge more “bite” in terms of micro serrations is true! Nevertheless, it’s the widest apex of the 4 blades, with the worst BESS score.
Let’s compare the high magnification pictures of the first 3 blades:
Comparison of (from left/first to right/last) the unstropped, “best practice stropped” and “overstropped” blade. Instrument: Zeiss GeminiSEM560.
It is clearly visible, that the unstropped (aka: just ground) blade shows a very defined apex, with a very low amount of burr formation. The “best practice” burr formation gives a larger, forward facing burr. With overstropping, in this steel, the flanks become much nicer and more polished – but also refines the apex to a smaller width.
What key takeaways are to be deduced from this?
One point that should be made clear: The analysis of the apex is shown here for briefness sake at one location, but was homogeneous and comparable at several locations along the edge. The magnification used to show the apex here are very high, and the blade sharpness is unlike anything I’ve seen before in terms of SEM images on the internet.
I think we see a bit of a repeat performance from the previous part in Vanadis 8. The ground edge shows a clearly defined edge, homogenoues and relatively fine, with an apex just over 100 nm.
With stropping in what I would call the widely accepted best practice, a larger burr of slightly higher (120 nm) width is raised, facing outward of the blade. The perceived sharpness of this blade is very high – we were able to record an easy hair whittling video under the microscope on this blade:
Overstropping actually removed “a lot” of material, and refined both the apex as well as the surface finish. Nevertheless, this is a double edged sword: unless you are an OG at sharpening like Roman is, you will likely round over your apex with this approach.
So, is stropping finally dead, this time?
I don’t think so. This second data point adds a lot to the theory I am starting to form. Meaning, stropping raises a forward facing burr, and that burr performs better during the usual “proofs of sharpness”, such as whittling a hair or performing a BESS test. We’ve seen in part 1, that the stropped edge didn’t withstand the very limited cutting test as well. So while I still hesitate to call stropping dead, I am becoming cautiously convinced that maybe it’s not the best approach for a functional edge. More research is needed and I am willing to go down this rabbit hole.
A future part on this series, in a couple of weeks, will look at the effect of stropping vs no stropping in a cheaper, more simple steel.
Disclaimer: This post is probably going to trigger a lot of strong opinions. I don’t mind that. If you disagree with something I state, I’d love to have you show me where I went wrong! Contact details are in the impressum. If you also just want to shout at me because you disagree, that is fine as well.
Sharpening is a wonderfully complex and interesting process. The technique, steel and abrasive all influence the result to a major extent. Degrees of freedom and influencing factors are nearly uncountable. Moreover, the results we achieve are nearly impossible to visualise – the apex width on a sharp cutting edge is impossible to analyse with easily accessible means such as an optical microscope. For this reason, the standard approach towards quantifying sharpness, but also the methods used to sharpen a cutting edge are based on proofs of sharpness – for example, slicing through paper, splitting a hair or recording the cutting pressure on standardised tests such as a BESS test, where a test media (a nylon fibre) is cut through.
The common approach to sharpening is to use a coarse stone to set the bevel geometry, and then use progressively finer sharpening stones to refine that apex, until as a last step, a process called “stropping” is undertaken, where the blade is dragged across a (often pliable) media, sometimes coated in loose abrasive. Now, this process can be adapted infinitely, but generalised, this is what a sharpening process entails.
I have long been wondering, what happens during stropping. Why does it increase sharpness so dramatically, even if one strops just on the palm of ones hand or on a blank piece of leather? What is happening, at a microscopic level, to the apex?
My good friend Roman Kasé visited me, and together we set out to discover the secrets of sharpening and stropping. If you do not know Roman, you should check out his homepage – he is a masterful artisan making wonderful knives, but also an absolute beast and OG at sharpening.
What We Did (Experimental Setup)
Roman and I looked at two different steels – Vanadis 8 (66 HRC) and M398 (65 HRC), both wonderful high tech powder metallurgical steels. Every steel blank was prepared in an identical way:
Rough grinding was undertaken with an ATOMA F400 EP stone. The bevel was ground to an angle of 17 DPS, and the initial grinding was undertaken until the whole bevel was homogenous, with a clearly formed burr and no distinguishable carbide breakouts at 50x optical magnification.
Afterwards, a progression of resin bond stones is used to refine the apex and achieve a high degree of sharpness. For this, we used my own design stones – Dr. Marv’s Scientific Sharpening stones, going through a grit progression of 80, 60, 40, 20, 10, 5, 2.5 and 1 micrometre. Each stone was employed for 2 minutes, with the exception of the 1 micrometre stone which was used for a couple of passes only.
For the stropping, we used fresh leather strops loaded with different emulsions (Manufacturer: Stroppy Stuff) with the following approaches:
1.) No Stropping – pure ground edge! this is our base truth to what we compare the stropping
2.) “best practice”: 5 strokes with 1, 0.5 and 0.25 µm diamond emulsion on leather
3.) “overstropping” – 50 strokes with the 1, 0.5 and 0.25 µm diamond emulsion.
Approaches 1 & 2 were applied to both the Vanadis 8 and M398, whereas 3 & 4 were applied only to the M398. For M398, check it out in part 2.
The Vanadis 8 was then used to cut 5x through a 16 mm manila rope, and wear was compared between the stropped and not-stropped blade.
How we analysed the results:
After sharpening, every blade was cleaned via a non contact process. For this, a steam cleaner was employed. There, heated water steam is ejected at around 3.5 bar towards the cutting edge. This can be considered a very gentle cleaning, as the force from the gas jet is very low. Afterwards, the blades are rinsed with ultra pure, analytical grade ethanol and blow dried with compressed (3.2 bar), ultra pure nitrogen gas. They are then inserted into a ultra high resolution SEM – a Zeiss GeminiSEM560 and plasma cleaned (2 min / 30W forward power). The SEM has a sub-nanometre resolution at all accelerating voltages. Images are recorded via the SE2 detector (sub 25kx magnification) and the InLens ultra high resolution SE1 detector above that magnification.
Let’s look at the results:
Let’s start of with the Vanadis 8 blade. This was first sharpened via my stones, and then analysed. Afterwards, we took it out, stropped it with approach #2 (see above), and analysed it again.
Sharpened cutting edge in Vanadis8, without stropping. Note the very nicely formed apex with low burr. Instrument: Zeiss GeminiSEM560
This is a wonderful look at the apex. A couple of things can be noted about the blade: We do have a couple of scratches, which are in the width of < 1 micrometre. The carbides are easily detectable as some very slight “bumps” across the flanks of the picture. The apex is homogeneous and very straight. Please also note the magnification of the pictures you are looking at – to get a sense of the scale, a got parameter in the databar is “width”, which shows you the width of the picture you are seeing. The maximum magnification shot you can see here features a magnification of 100kx (polaroid standard, look further down for those shots), so the picture shows you a 1.14 µm wide excerpt. This is out of this world in terms of magnification.
Next, let’s take a look how stropping changes the edge.
Sharpened cutting edge in vanadis 8 after stropping. Note the forward facing raised, nanometric burr. Instrument: Zeiss GeminiSEM560
Immediately visible, even at low magnifications is the improved surface finish of the cutting bevel. The diamond definitely abraded some material!
If we compare the two results with some measurements:
Comparison between the two apex: Left side/first picture: no stropping. Right side/second picture: stropped apex
It can be noted, that after stropping, a prowl shaped burr exists on the blade, that is facing straight outward at the cutting edge. This burr is significantly wider and more inhomogeneous than on the sharpened blade. The direction of the burr faces straight outward – which makes for a secondary, nanometric apex that is of a much sharper angle than the apex behind it.
Sharpness wise, both edges easily shaved and sliced through a piece of paper. The stropped edge showed a reduced BESS reading compared to the unstropped edge.
I’d like to also note that Romans stropping technique appears to be flawless – there is not discernible rounding of the apex here! He uses very light pressure and also lowers the angle by 0.5 DPS to compensate for the deflection of the leather.
Let’s take a look at the blade after 5 cuts through 16 mm manila rope. For this, we first tested the stropped blade, and then resharpened with identical stone progression, checked via SEM and then cut again. Pictures of the apex were taken at the location where the cut happened::
Vanadis 8 blade, unstropped, after 5 cuts through the 16 mm manila rope.
Even a low amount of 5 cuts already dulls the apex noticeably. Instead of a nanometric width, there is now a roughly 1 micrometre wide apex. At some positions, some brittle breakouts of carbides are visible.
Comparing this with the stropped edge:
Vanadis 8 blade, stropped after taking 5 cuts through a manila rope.
The apex is in a similar condition. It looks a bit more irregular, and folded/deformed compared to the unstropped blade.
Comparing the two apex side by side:
Comparison between the two used apex: Left/first picture: no stropping, right/second picture: stropped edge
The stropped edge actually became noticeably duller/wider than the unstropped edge.
What key takeaways are to be deduced from this?
One point that should be made clear: The analysis of the apex is shown here for briefness sake at one location, but was homogeneous and comparable at several locations along the edge. The magnification used to show the apex here is absurdly high – very few images such as these have been openly shown on the internet, if at all.
On this steel, and with our approach, stropping actually raised a nanometric, forward facing burr. It is undetectable by optical or (human) tactile means, as the size (both of the apex and the burr) is in the double digit nanometre range.
The burr improved the sharpness of the blade in the usual proofs of sharpness, such as whittling a hair or slicing paper.
After slicing just 5 cuts through manila rope, the apex has rounded over significantly. This would still be nearly undetectable with optical means, and it still felt like a very sharp knife. The stropped knife has a higher amount of apex wear than the unstropped one.
Does this mean you should stop stropping?
I don’t think so. This is a single data point in a single steel. Especially on steels where deburring is more difficult, stropping might behave differently. Moreover, the blade was sharpened with very specialised stones that are designed around a superior cutting action, thus further reducing burr formation.
I think this raises more questions – especially ones such as: where does the material for this burr come from? Is this burr responsible for the perceived increase in sharpness on soft and thin materials such as paper, hair or a bess test?
For me personally, I have stopped stropping when I started using my own stones. They produce a devilish sharp edge, with a very clearly defined apex. The increase in sharpness which can be undertaken from stropping doesn’t seem to translate into a longer lasting sharp edge. More data points for this are needed – which we will be giving in PART 2 (TBC).
This is part of an ongoing blog series about metrology. It explains physics, principles and use cases of modern metrology devices.
TL;DR: Explains how a SEM works. Deep dive into the electron-beam interaction, showing how every sensor gives a different picture and what they could be used for. Lot’s of solid state physics, but in the fun “look at how amazing nature is” way, not in the “equations of horror and despair” way.This should be a fun read and easily understandable, even if you haven’t thought about physics since school.
From the amount of scanning electron microscope (SEM) pictures in this blog, you can guess that I’m a huge fan and heavy user of these wonderful devices.
Brief historical overview& resolution limit
The basic idea behind a SEM is the Abbe diffraction limit of resolution. Ernst Abbe was a pretty cool dude – he lived in Germany during the late 19th century. He defined the foundations of modern optics, had a very impressive beard and is credited with owning Carl Zeiss for some time and the creation of Schott AG. In precision engineering, he has had a lasting impact, mostly for his definition of Abbe Error Compliance (Measurement device in axis of movement is more precise than parallel to axis), but also the Abbe diffraction limit.
It basically states, that the minimum resolvable feature size d is a function of the wavelength of your radiation λ divided by 2 times the index of refraction of the immersed medium n (for example air) times the half-angle subtended by the objective lens θ. The numerical aperture NA describes the resolving power of a objective lens, and is the product of n * sin θ. Thus we have for our system resolution:
d = λ / 2 NA
If you have air between your objective lens and sample, NA can only ever be below 1. Very high quality, large magnification objective lenses can for example have 100x/NA 0.95, coming very close to this theoretical limit. This means, at absolute best, our smallest, resolvable feature is about half the wavelength of the radiation. If you have a nice, confocal microscope, your system might use a green LED at 532 nm, thus your systems resolving resolution is in the range of 0.25 µm. There’s a couple of techniques to get around this limitation, but with visible radiation, you are not going to get massive improvements in lateral resolution. But: the wavelength of radiation is inversely proportional to the energy of the radiation:
E = h*f and: λ = c/f
Visible light typically has an energy of 0.5 – 3 eV, a WLAN signal about 5 µeV. X-Rays start somewhere around 1 keV, and most SEM have their resolution sweet spot at 15-30 keV. Modern tunneling electron microscopes TEM are in the range of 200-300 keV. Sadly, we do not have a TEM at Kern Microtechnik GmbH. *chicken scratches one onto the “Dr.Marv purchase wishlist”*
Now, the actual resolution of a SEM isn’t as close to the theoretical limit as optical microscopy has, because it is surprisingly difficult to compensate all beam and lens (magnetic field) errors. Aberration error correction is something that is only now really hitting the market.
Nevertheless, even a small, entry level desktop SEM like our Thermo Fischer PhenomXL spots a datasheet resolution of smaller than 10 nanometre. Typically, this is achieved as the distance between gold nanoparticles on carbon. Very conductive, maximum elemental contrast and clearly defined boundary edges. It goes without saying, that this is the easiest possible image for a SEM!
SEM micrograph of a very dirty, hydrocarbon contaminated resolution test object. What you see is gold nanoparticles on carbon, at a very high magnification (500kX) and very low accelerating voltage (1 keV). Analysis has shown that our instrument is within specification, even here: 0.7 nm @1keV. Taken with the magnificent Zeiss GeminiSEM560.
At lower energy, the electrons are also much slower, thus experiencing more extraneous influences such as magnetic stray fields or vibrations from body or acoustic noise. A high resolution SEM will have sub 1 nanometre resolution over the entire energy range.
General working principle of a SEM
We have defined that instead of using a beam of visible light, an electron microscope uses a beam of electrons to look at matter. At minimum, an electron microscope consists out of an electron source, some condenser, scanning and focusing “lenses” (which are actually coils with a magnetic field), an aperture, a vacuum chamber with the sample as well as a sensor to detect the signal.
Below is a schematic view of the column design of our ultra high resolution, Zeiss GeminiSEM560.
Schematic cross section of a high resolution SEM column. Pictured detector is a SE2 Everhart Thornley type.
At the electron source, electrons are generated. There’s two typical ways: thermionic emitters, where either a tungsten or a LaB6 filament is heated until free electrons are emitted. The second option is a field emission gun (FEG), where a small filament is heated, but the electrons are removed via a strong electric field. FEG are typically more stable, have less noise and a narrower energy spread. They are more expensive and set higher requirements to the vacuum system.
The electron beam is then accelerated via an electric potential, and then shaped and focused via condenser lenses. The beam current is regulated via an aperture orifice. The beam is then focused and scanned across the sample in a regular pattern via the objective lens. This scanning is not a continuous process, but instead the beam dwells for a short amount of time at every “pixel” position. A detector simply counts the signal emission at every point, thus creating a black and white picture from the sample – electron beam interaction.
It is a very basic principle, but the interaction of the beam and the sample is very complex, and many sensors exist to detect different types of signals. What is really nice about this scanning and way of detecting a picture compared to having a high resolution sensor with many pixels is that all sensors have the same focus point – so you can typically seamlessly switch between sensors and don’t have to refocus.
Electron – Matter – Interaction
In order to understand the different pictures and data created from a SEM, we need to take a quick detour towards high-energy electron beam interaction with matter. When matter is hit with fast electrons (the primary electrons, PE), a couple of possible interactions can happen. The below schematic shows the 4 dominant types, mainly back-scatter electrons (BSE), secondary electrons (SE, type 1 and 2) and x-ray emission (hv). The interaction volume depends on beam energy, but is typically in the range of < 10 nm for SE1, 1-50 nm for SE2, 50-1000 nm for BSE and 1-10 µm for hv. Because of scattering, the interaction volume is shaped a bit like a pear.
Schematic beam interaction with matter. The 4 main emitted signals are shown: BSE, SE1, SE2 and hv.
The interaction volume depends on beam energy, but is typically in the range of < 10 nm for SE1, 1-50 nm for SE2, 50-1000 nm for BSE and 1-10 µm for hv. Because of scattering, the interaction volume is shaped a bit like a pear. To show this interaction, I’ve prepared a small Monte-Carlo scattering simulation highlight this interaction volume, and how deep the different species might reach. This is for a high energy beam in a light material.
Interaction volume of high energy electrons in a light material. SE are highlighted in green.
We will have to dig a bit deeper into the creation of each of these, but also how they change the picture and what data and conclusions we can draw from them. For this, I’ve put a very used, nearly broken carbide end mill into the SEM.
First/left picture: The used endmill, before being inserted into the SEM chamber. Second/Right picture: the inside of the chamber, with the visible polshoe, SE2 detector and illuminated chamberscope. A couple more complex sensors are visible in the background.
After pumping the chamber empty of air, activating the SE2 sensor and focusing, we can generate an overview image of the tool. Because the SEM flares the field at the objective lens, we get a much larger FOV, but heavy distortions. This is mostly useful for navigating and finding a feature (or even: where the heck am I currently!).
SE2 overview picture of the inserted endmill. Instrument: Zeiss GeminiSEM560
Secondary Electrons
Sometimes, when the incident electrons interact with an atom, they do so through inelastic scattering with the shell electrons. This ionises the electron, via ejecting a shell electron, the so called secondary electron. If it’s the primary electron, these SE are called SE1, and are very surface sensitive and detected via a SE detector inside the electron column. If it’s created by backscatter electrons ionising the atoms, they are called SE2 and are detected via an in-chamber detector. These are very sensitive to topography, so the resulting picture is a good representation of the shape and surface of the sample. Because they are created by BSE, the interaction volume is a bit deeper, and the signal can’t resolve very fine surface detail. This sensor is very susceptible to static charge up on the sample.
Schematic depiction of the SE creation process. Note that the incident species can also be BSE, and not only SE.
The SE2 sensor is very fast in it’s readout speed, and typically, especially at longer working distances (distance between the pole piece and the sample) exhibits a strong signal. If the sample is non-conductive, this is my first choice in focusing the picture and for navigating. Because it is at an angle inside the chamber, the sensor gives a very good depth representation of the sample. Pictures look plastic and 3 dimensional.
SEM micrograph of the cutting edge. Signal A = sensor used, in this case the chamber SE2 type. The picture has depth, and nicely shows the morphology of the grinding marks, the coating and particles on the tool.
Switching to the InLens SE1 detector, the picture changes in it’s appearance:
SEM micrograph of the cutting edge. Signal A = sensor used, in this case the InLens SE1 type. The picture has lost some depth, but gained some detail on the surface structure. Besides the grinding marks, the micro-roughness of the coating is now visible.
Because this sensor has a very small interaction volume, it shows fine surface details. Whereas the SE2 sensor mostly showed the grinding path along the tool cutting edge, this sensor shows the micro roughness of the coating, and highlights different sections of the build up edge through finer detail. At the same time, some depth perception is lost, resulting in a flatter picture.
Back Scatter Electrons
Back scatter electrons are created from elastic scattering (reflection) with the nucleus of the atoms. Because of this, the electrons have a lot of energy. The chance for elastic scattering depends on the mass of the nucleus, thus heavier elements give you a brighter signal. Therefore, the BSE signal gives you a material contrast.
Schematic depiction of the SE creation process. Note that the incident species is either the PE, or a lower energy already back scattered BSE.
The same cutting tool we looked at in the SEM can also be visualised with back scatter electrons. For this, our GeminiSEM560 is equipped with two different one: the ESB detector, that sits very high up in the column, and a retractable diode type 4 sector BSD sensor that can be fitted exactly below the pole piece.
Because we can always activate it, let’s start with the ESB detector picture. We can see that the image is flattened a lot – this sensor is not really picking up any topography.
In column ESB detector SEM micrograph of the cutting edge. Note the lighter coloured structures – these are heavier elements than the darker coloured structures.
This sensor is quite “slow”, in the sense of it not getting a lot of signal. The above picture took a bit over 4 minutes to record.
The SEM is fitted with a diode type, 4 sector BSE detector, that can be retracted and inserted via a pneumatic cylinder. Because it is sitting below the pole piece, it is much quicker, and still shows some surface structure.
Chamber BSD SEM micrograph of the cutting edge. The picture has very little depth, showing only a minimal amount of surface structures. The BUE material is clearly distinguishable, showing 2 different materials via the inherent BSD material contrast.
This sensor is much quicker, and shows some topographic details. A bit more depth perception than on the ESB sensor is given.
X-Ray creation (EDS – Energy dispersive x-ray spectroscopy)
The PE are able to create x-rays. I find this absolutely fascinating, and one of my favourite tools inside the SEM. When the PE interacts with the atomic shell, sometimes an electron is ejected (the SE). If this happens at a lower shell, a hole (missing electron spot) is created. Because most systems strive to lower their potential energy, a higher shell electron will then drop down. Because the new orbit has a lower radius, there is now some excess energy. Through this energy, just like Einstein foretold, a particle is created, specifically a x-ray photon. Because the distance between shells is dependent on the weight of the atomic core and it’s configuration (proton number), the energy difference is unique for every element.
Schematic description of the x-ray creation process. An incident electron creates a hole in an inner shell through inelastic scattering (a SE is ejected). A higher shell electron drops down to fill the hole (blue arrow), because of energy conversion the lower orbit energy results in the creation of a x-ray photon.
By recording lots of x-ray photons, and sorting them by their inherent energy, we can identify and map an elemental distribution over our picture. This process is called energy dispersive x-ray spectroscopy. It doesn’t tell you “This is REX 121 steel”, but instead after several minutes you can say “oh, we have about 12 % chromium in our sample. Maybe. I hope. Pinky promise!?”. This is the part where TV shows have ruined science for us scientists. But if done right, at correct readout sampling rates, high beam energy, you get fantastic results.
What can we see on our tool edge? First and foremost the coating material – Ti, Al, N, Cr, resulting from a ceramic high tech multi layer coating applied to most modern tools. Build up edge and particles from aluminum, some stuck carbon, places where the coating has failed and tungsten carbide is visible through the coating. Pretty nifty, eh?
EDS analysis of the tool edge.
The real expertise in EDS is looking at the data and then making judgement calls and drawing from experience and know how. We once found a particle we wanted to analyse, and it contained bromine. People were stumped, because which part of the machine contains bromine? In the end it was the base of our powder coating, and the particle we found at a place it shouldn’t be was the powder coating from the machine enclosure that had flaked off. We managed to nail that down, because the only reasonable expectation of where bromine could be used was exactly that. Comparing a fresh piece revealed identical chemical composition.
Manage Consent
To provide the best experiences, we use technologies like cookies to store and/or access device information. Consenting to these technologies will allow us to process data such as browsing behavior or unique IDs on this site. Not consenting or withdrawing consent, may adversely affect certain features and functions.
Functional
Always active
The technical storage or access is strictly necessary for the legitimate purpose of enabling the use of a specific service explicitly requested by the subscriber or user, or for the sole purpose of carrying out the transmission of a communication over an electronic communications network.
Preferences
The technical storage or access is necessary for the legitimate purpose of storing preferences that are not requested by the subscriber or user.
Statistics
The technical storage or access that is used exclusively for statistical purposes.The technical storage or access that is used exclusively for anonymous statistical purposes. Without a subpoena, voluntary compliance on the part of your Internet Service Provider, or additional records from a third party, information stored or retrieved for this purpose alone cannot usually be used to identify you.
Marketing
The technical storage or access is required to create user profiles to send advertising, or to track the user on a website or across several websites for similar marketing purposes.
