Razor blade edge with reflected DIC

[SiBurning]

These were taken on an Olympus BH using an Mplan-N reflected DIC 100x objective (NA=0.90) with a 1.3M pixel USB camera in the eyepiece. Please forgive the abberations from not using the correcting eyepieces, as well as the inaccuracy this causes in my measurements.

This is focused right off the edge (at least in some spots)


Detuned things a bit to get more in focus.


Detuned a lot. The way that edge sticks out seems to be an illusion caused by the DIC shadows.


Checking with a 1mm micrometer with 10 micron divisions, the camera shows about 107 x 85 microns. The pictures were taken at 1280x1024. They're cropped (not resized) here to 600x600 in order to meet the 300kb limit, and otherwise untouched. They show about 50 microns on a side (one's a tad smaller). On my 100dpi monitor, the magnification is 3000. My understanding is that the theoretical limit of resolution should be about 300 nm. That requires about 167 pixels in these photographs. So these pictures are taken at about 4 times the maximum resolution. What I don't understand is how there appears to be more detail than that, and it's lost if I shrink the picture.

Much of what's there to see eludes me, although most of it doesn't show in the pictures. Previously I've looked mostly with a stereo microscope. Diffraction is an issue with razor blade edges at any magnification. High power and DIC really make things interesting. The small depth of field makes it difficult to tell the relationship between points that appear below or near each other. Maybe my scope needs a tuneup (I know it does. Both irises are off center). Fragments move around when I change the focus, and some of them look like mere ilusions--impossible geometries--like something from a Lovecraft cyclopean nightmare. It appears that DIC and diffraction artifacts make things move around, but I don't have a compatible non-DIC 100X objective to compare against. I'm trying to satisfy myself that this will go away when I tune the scope, but the visual errors are on the order of half a micron, so I'm not holding onto much hope.

Too many questions. Too little knowledge. Most of what I've read about metallurgical microscope work says you need something polished flat. At this point, I have to agree.

[rjlittlefield]

SiBurning, welcome aboard!

I have no experience with DIC, so other people will have to address most of your questions.

But I may be able to help some with the resolution issue.

300 nm is the spacing between two peaks that are just barely resolved. Projected onto a digital sensor, that requires minimally 2 pixels per 300 nm (one for a peak, one for a trough), not 1 pixel as your calculation seems to use. The problem is, at that pixel density most of the contrast will be lost and some of the features will disappear entirely. To preserve most of the contrast of the finest detail requires 3-4 pixels per 300 nm, which is pretty close to the 600 pixels you're showing here.

If you check the calculator at MicroscopyU, you'll find that they give a number based on 2 pixels per 310 nm, giving 322 pixels in 50 microns. That number is sort of a compromise. It's based on minimal sampling, which as mentioned will significantly degrade the finest detail. In some sense it's the pixel density that minimally samples a spatial frequency for which diffraction cuts contrast almost to zero anyway. When I say it that way, the number sounds crazy. On the other hand, that same density puts at least 4 pixels per cycle at a spatial frequency for which diffraction would cut contrast to at best 50%. In that sense the number gives good sampling of all the detail that is well resolved by the optics.

So...

The pictures you're showing here do appear to contain finer detail than I would expect, but only by something under 2X, certainly not 4X. I agree that some visible features are lost when I downsample to 50% and then restore the original size, and that surprises me. But I notice that some of the "features" occur in severely OOF areas, so I am inclined to think that those are sensor noise, possibly amplified in digital processing. This makes me wonder about the finest features that appear in focused areas. What if anything happened to the images after the USB eyepiece was done with them?

--Rik

[SiBurning]

That's very interesting. Looking again, it does appear that most of the loss can be explained by lack of contrast. There's also some color differences, which I originally blamed wholesale on the camera and the missing color correcting eyepieces. Your comment that "features occur in severely OOF areas" leads to me to an idea. Some of the most serious areas of loss show up as blobs of brown, which makes me wonder if it's reaching the limit in part of the color spectrum, giving a blurry red and black look to the area without the other colors to balance it. For example, in the third picture about 3.25-3.5" (of 6") from the bottom there's an indent that lost a lot of detail going into the peak--what's left appears as a dark & blurry brown blob. I'll have to compare against the eyepieces, and also take a look at the unnatural greens.

I hadn't thought about diffraction in terms of its effects on contrast and the camera sensor, and certainly not in terms of losing part of the spectrum. At least not this early on. These objectives were a sort of sudden impulse purchase, and are clearly better than my Neo set. On the stereo scopes, and even here, diffraction shows up as light bending around the edge--either a sharp band of color, some flecks deflected around an otherwise invisible rough spot, or a clear and somewhat abrupt distortion wave that appears just beyond the edge.

I'll try to find a way to get better color fidelity out of this camera or find a way to mount my Nikon. Also play around with different eyepieces to see what correction is applied in the proper eyepieces.

Your comment about how "features occur in severely OOF areas" also indirectly resolved another question. The prisms are marked 1.1, which I now believe refers to how the prism lengthens the tube length giving an empty magnification factor of 1.1.

Thanks for the insights and the welcome.

[rjlittlefield]

To help get some additional common viewpoint...

The effects of diffraction can be modeled by isolating its effect on a point source, treating the subject as an array of point sources, and adding up the corresponding array of effects. The diffracted image of a point source is an Airy disk with rings. For an isolated edge, it's a softened edge with ringing. And for a extended sinusoidal grating, it's a simple loss of contrast. Depending on what you're trying to think about, any one of those effects may be most convenient. For more about the contrast viewpoint, see THIS post and the surrounding thread. For discussion of what happens as you change pixels per cycle, see HERE.

Still trying to make sense of your images...

You've raised some additional points that may help explain what's going on.

It's especially hard to think about the relationship between the image and the sensor. The discussion above and the calculator at MicroscopyU casually assume a single-color model with a single pixel spacing. But I presume your camera actually has a Bayer filter in front of the sensor. If so, then out of every 2x2 block of pixels, there is only one "red" pixel and one "blue" pixel. In front of those, there may be an antialiasing filter. Depending on exactly how the filters and the de-Bayering algorithm work, the system resolution may be only half as good for red and blue as it is for green. I can vaguely imagine it may also be vulnerable to "spurious resolution" in which rapid changes of hue would turn into rapid changes of intensity. This last possibility is especially hard to think about because rapid changes of hue are exactly what you get from diffracting at a bunch of different wavelengths at once.

Oversampling by spreading your image over more pixels would help to eliminate any Bayer effects that may be confusing the issue. It might also help to narrow your chromatic bandwidth by sticking a green filter in the system.

Just rambling...

--Rik

[SiBurning]

Still trying to make sense of your images...
--Rik

Apologies for not thinking ahead and warning about this. The short answer is that most of what you actually see in these pictures are blobs of coatings.

I'm not sure about the rules here regarding external links, but I'll take a chance... This is part of a personal project of mine (more like a quest) that started about a year ago on a shaving-related forum, to take pictures of hair and razors blades under (initially) my stereo microscope, and which eventually got me interested enough to get this reflecting compound scope. Point is, the pictures here are part of a larger set I posted there last week. The others might shed just a tiny bit more light on the subject. http://badgerandblade.com/vb/showthread.php?t=95526. Scroll down to post #12 for this series. Fair warning, there's about 30 large pictures total in that thread. For the record, this is a Feather (brand) double edge razor blade.

Again, most of what you see is coatings. Looking at a used blade would be helpful, but I have no pictures yet. A mildly used blade (3-4 shaves and left to sit for a week) shows hardly any wear on the stainless steel blade itself, but the coating breaks off in chunks. There may be several layers of coatings--it appears as if one has been applied before a final sharpening, but it needs more study of different batches of blades to be sure. The last layer(s) are sloppy blobs, and those blobs is what shows lumpy in the pictures posted here. They're responsible for all of those irregular shapes. I'll eventually look for a way to remove the coating chemically or by heat and expose the actual metal od the blade. The lumps also hide the lower layers of coating as well as the sharpening marks. I do see the sharpening marks in the eyepiece under the blobs, though they get lost in most of the pictures.

[rjlittlefield]

External links are encouraged whenever they help the discussion.

You can even hot-link images from other sites (using the [img] tag), as long as the other sites allow.

Thanks for the explanation about the subject. You answered several questions I had not even thought to ask.

When I wrote "making sense", I was still thinking of ways that your images could appear to have more detail that physics seems to allow. That was the intent of my ramblings about the Bayer filter etc.

But your explanation shifts my attention back to the subject. Usually when I see a ground edge under a microscope, it shows clear grinding marks clear to the edge. See for example Charles Krebs' razor blade HERE. In your images, I can see what appear to be coarse grinding marks far away from the edge, but no such marks at the edge itself. I had just about resigned myself to the idea that this particular blade had been polished so finely that the remaining scratches could not be resolved optically. But if there is a coating involved, then perhaps the coating is simply obscuring the finest grinding marks. Or perhaps not, and the grinding really is as fine as I had thought.

I will go read your blog. Thanks for the link.

--Rik

[Charles Krebs]

SiBurning,

Welcome.

, but I don't have a compatible non-DIC 100X objective to compare against.

This confuses me a little. Usually the only difference between a "DIC" objective and a non-DIC is that the former has been certified by the manufacturer to be particularly strain-free, which makes them well suited for lighting methods that relies on polarization. Frequently the optics are otherwise identical.

If you remove the Nomarski prism, DIC/polarizer, and the analyzer from the light path in your microscope you would have "normal" brightfield with an objective marked DIC.

Can you describe the equipment you are using in a little more detail?

[SiBurning]

Sorry. I knew that was wrong, and should have gone back and corrected myself. It's just not clear to me how to safely remove the prims and whether the objective would fit the nosepiece afterward. I just got these and haven't had the time to look into it. May as well add I got called out of town suddenly and will be out for another week, so I can't do anything for now except read and learn.

It's a BHM fitted with transmitted light using the base and condenser from a BHC. The nosepiece on the table has Neo objectives 5X through 40X, the ones on the DIC prisms are the Mplan-N. Both are on page 10 of the Olympus High Quality Optics brochure online. I believe the eyepieces are WF10X on page 13, but can't verify that, and anyway they're irrelevant (for now) to the pictures. A polarizer and analyser are installed.



One interesting thing I've read is that prisms have a better theoretical resolution than lenses (100nm vs. 200nm). It's also tempting to speculate on whether two beams are better than one, so to speak, because the light is split and rejoined. But a little knowledge is a dangerous thing, so I'll stop there.

It seems important enough to note again that the light and irises are somewhat off center, and that the compensation eyepieces aren't used here. Both of these should add additional errors.

[SiBurning]

I got a new (trinocular) head that's aligned properly, so the light's much better. After spending a few hours with it, I'm convinced that the main problem in the previous pictures is that the light (the head) was misaligned. The new pictures make a lot more sense. Here's another shot showing about 100 microns of blade length, roughly the same as the others in terms of blade length. It's not meant to correct the previous images, but just to illustrate overall what a used blade looks like. OnceI get the rest sorted out, I'll attempt stacking. Further details on this picture (and many more older shots) are on the B&B shaving forum where this project got started.

[SiBurning]

Thanks to the very cool Zerene Stacker (which I licensed immediately)...

This is a DMap stack of about 20 images with 20% contrast threshold. They're taken with the 100X objective using the polarizers and DIC prisms. The camera is an amscope MD600 1.3Mp USB. The big white thing is soap <edit: maybe not, see a few posts down>, along with some particles of the coating. It's obviously deflecting the image of the surface in places.


Small adjustments to the focus produce large changes in the DIC images. For example, with very little movement of the focus, I watched a crevice move, change color, and become a hill. The movement was small enough that the surrounding area was still largely in focus, although the hill was now in that surrounding area. This obviously complicates making a stack, so I also took a stack with the polarizer & analyzer removed, effectively eliminating DIC. Without the polarization, the overall color is fixed, and adjusting the prisms has only a subtle and insignificant effect, aside from some minor (non color) changes to contrast in places.

For these, I used about 18 images at 50% contrast after removing the polarizer & analyzer, but keeping the prisms in place. On the bottom right, can see a chunk of coating hanging off the edge of the blade. seemingly stuck in the soap.


For some perspective (with apologies for the poor photograph) I switched to the 10X objective and mounted a Nikon D40 using the adapter for a Meiji stereo microscope. The picture is 580 microns across. The (dark) soap patch is front and center with the piece of metal coating hanging off of it. The primary bevel is about 0.5 mm long, and the secondary seems about 1/5 of that. The other pictures all stay within that secondary bevel.


The first thing I learned about a year ago using the stereo scope to photograph blades is how important it is to use appropriate lighting to caprture specific details. I seem to have come full circle. Seems like the best way to proceed is to play around with different subjects. For the blades, I'm going to back off on the magnification and try to improve on a larger scale first. The blade depth is awfully thin, making it harder than necessary to learn how to use the microscope, take pictures, and stack them.

[rjlittlefield]

It's fun to see the progress here! Also interesting to hear your impressions of DIC. I've never had hands-on with a DIC scope, and it's pretty much impossible to guess how that might feel, from looking at the gorgeous images posted by the gurus around here.

One question: how do we know the chunk of clear stuff is soap and not a bit of skin?

--Rik

[SiBurning]

This is great! I'm so fascinated and clueless, I feel just like a little kid again.

I just shaved a bit with plain water and a clean blade and took a look using darkfield. What a junkyard! There's a sort of thick plasticy coating over and around the hairs and metal fragments. I'll have to stain it to get a look at any cells inside, but one step at a time. Short answer... That's probably hair and skin cells that sort of blended into the water along with any residue from my face, the soap, my tap water, and whatever's on the razor, such as coatings and machining oils. They all sort of emulsify or melt into the water leaving a film when the water evaporates.

With DIC, the best detail seems to be when things look closest to full extinction of the cross polars. It requires setting the prism to a neutral point. This puts the the image in grayscale, perhaps only because of the limited amount of light. Adjusting it slightly from there gives the best differential contrast, while setting it further out gives pretty colors, but loses detail. It really works like a contrast function, but with distances converted into color and shadow.

The only way I can describe the DIC abberations I mentioned earlier is to relate them to general (spherical and axial?) abberations. While adjusting the focus on a normal lens, object points at different distances get displaced radially (I think). With DIC, as the image points displace either radially or axially, the shadows and colors change as well, but they don't quite move together with the image.

[SiBurning]

Quick follow up on what's in a typical blob after shaving. Nice mix of clear, brown, and red hairs, and some kind of goop. No soap was used, just water and a clean blade, some tap water, and whatever contaminants.


With this setup (darkfield and the NeoPak 10X lens) there's about 1/2" of working distance, which means plenty of lighting options. Just don't have the time today.

Thanks for the suggestion.

[SiBurning]

I've been busy learning more, so here's an update. Rather than show the razor to scale, it makes more sense to show a 1mm micrometer with 100 divisions. The etched lines are about 1 micron wide +/-10%. The regularity of the micrometer adds distortion (glare and colored lines) above and beyond what shows up with the razor blades, particularly in the camera.

These use a BHMJL set up for 160mm and a 1.3Mp camera direct mounted in the tricocular tube without a photo eyepiece. The glare in the center of all the pictures is due to not having a really good illuminator (I think). One's off center, and the other only has a single adjustable aperture.







The Olympus Resource Center does mention how DIC causes things to look like there's moving around. You can sort of make that out by comparing the little bits, especially on the vertical sides of the etching. They grow, change shape, hills morph into valleys, and things just generally move around.

It also appear that the best 3D imaging and detail show up with the smallest prism offset. Moving the prism further adds pretty colors, and changes the shadow location, but those extra colors come at the expense of detail, and at the extremes become glaring colored lights. Of course, it also helps to bring things to attention that aren't otherwise so visible.

The last thing I'm working out is that this really needs a rotating stage, like a polarized microscope. Rotating the polarizer/analyzer helps, but these blades have a grinding direction, and it helps to align the light separately from the grind marks.

On to the razor blades. This is a new Iridium Super. I didn't measure this set, but the little balls are comparable in size to those in the previous pictures. It's the same setup, but I can't swear they're to the same scale.

This is a stack of 9 images using DIC with an MPlan 100N objective. Detail would definitely be improved by moving to 20 or 30 images, but I'd guess that half of the details I expect to see never show up. The details are just too small.


A single DIC image. I don't remember the lens used.


Darkfield with a NeoPlan 40.


I'm still trying to come to terms with resolution. A lot of times I'm frustrated that the finest detail seems impossible to focus on, or it might seem like a smooth blob with some indications that finer details exist. (I know it's defraction--now a point, now a blob, now an airy disc.) At other times, I sit back, maybe do the math, and realize these aren't half bad. The 100X lens is pretty frustrating because even moving the mouse (on a separate table) to take a picture shakes the entire setup, changing the focus, moving the stage, and causing the camera software to rebalance the lighting. The 40X is much better behaved.

There's more razor blade pictures in my Micrsoscope photography thread at Badger & Blade.

This confuses me a little. Usually the only difference between a "DIC" objective and a non-DIC is that the former has been certified by the manufacturer to be particularly strain-free, which makes them well suited for lighting methods that relies on polarization. Frequently the optics are otherwise identical.

If you remove the Nomarski prism, DIC/polarizer, and the analyzer from the light path in your microscope you would have "normal" brightfield with an objective marked DIC.

Can you describe the equipment you are using in a little more detail?

From what I've read, this is true, but the opposite isn't true for all setups. It has something to do with the image being formed inside or outside the objective. Anyway, I have two sets now, and also upgraded to NeoPlans.

[wantok]

These use a BHMJL set up for 160mm and a 1.3Mp camera direct mounted in the tricocular tube without a photo eyepiece. The glare in the center of all the pictures is due to not having a really good illuminator (I think)....

How do you like this set up? A friend has a BHMJL for his metallurgical work, and I was thinking of playing around with setting up a camera, just for fun.

BTW, very interesting photos (and I like the thread at badger and blade).