Cover slip / coverslip thickness / correction / effect

Have questions about the equipment used for macro- or micro- photography? Post those questions in this forum.

Moderators: rjlittlefield, ChrisR, Chris S., Pau

Lou Jost
Posts: 5933
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

See for example Eq 4 in this link:
https://www.edmundoptics.de/resources/a ... on-plates/

Likewise this site ( https://www.olympus-lifescience.com/en/ ... spherical/ ) says:

"Complicating these conditions is the fact that for almost all materials, refractive index is a function of both wavelength and temperature"

and

"Until recent years, achromats were corrected spherically only for green light, although they were corrected chromatically for two wavelengths. Also, apochromats were corrected spherically for two wavelengths, blue and green, but were corrected chromatically for three wavelengths. The highest-quality modern microscope objectives address spherical aberrations in a number of ways including special lens-grinding techniques, improved glass formulations, and better control of optical pathways through use of multiple-lens elements. Currently, the highest quality objectives, planapochromats, are spherically corrected for four wavelengths, as are planfluorites (but not to quite as close a tolerance)."

Lou Jost
Posts: 5933
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

The tolerances based on this graph seem surprisingly large to me.

The correction collar on my Oly 20x 0.45 objective has quite a large throw between 0 and 2mm. I wonder why it has such a delicate sensitive mechanism, if the media thickness tolerance is so large (at worst, 0.5mm for violet light). Has anyone noticed whether tiny movements of a correction collar really make a difference in the image for a 0.45 objective?

Edit: They WERE too large, because I misplaced a decimal point (see Rik's comment below). The numbers in my original comment are now fixed. I leave the erroneous numbers in this comment so readers can see why I was suspicious of my figures.
Last edited by Lou Jost on Thu Mar 07, 2019 3:23 pm, edited 1 time in total.

rjlittlefield
Site Admin
Posts: 23543
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Lou Jost wrote:See for example Eq 4 in this link:
https://www.edmundoptics.de/resources/a ... on-plates/

Likewise this site (https://www.olympus-lifescience.com/en/ ... spherical/) says:

"Complicating these conditions is the fact that for almost all materials, refractive index is a function of both wavelength and temperature"

and

"Until recent years, achromats were corrected spherically only for green light, although they were corrected chromatically for two wavelengths. Also, apochromats were corrected spherically for two wavelengths, blue and green, but were corrected chromatically for three wavelengths. The highest-quality modern microscope objectives address spherical aberrations in a number of ways including special lens-grinding techniques, improved glass formulations, and better control of optical pathways through use of multiple-lens elements. Currently, the highest quality objectives, planapochromats, are spherically corrected for four wavelengths, as are planfluorites (but not to quite as close a tolerance)."
Great, thanks for the references. It's very helpful to see what you were looking at.

So then, hopefully to clarify, I see two different situations.

The first is where an optical system is optimized to give small or zero SA at one wavelength, and then you change the wavelength, and see significant SA. In this case the additional SA will come largely from change in refractive index with wavelength. That is, the optical path lengths changed when you changed the wavelength, and the resulting SA depends on difference errors in the new path lengths, divided by the new wavelength.

The second is where an optical system is optimized for small or zero SA with no added medium, and then you add medium so as to get significant SA. In this case the amount of additional SA that you get will depend strongly on wavelength, but now it's due mostly to the pathlength/lambda effect and will persist even between two wavelengths for which the path length differences are identical.

The context of this thread strikes me as the second situation, hence my quibbling about "because the medium's index of refraction varies with wavelength".
The tolerances based on this graph seem surprisingly large to me.
I agree, and I'm wondering if your computation has gone awry.

I have a spreadsheet that computes SA wavefront errors based on first principles of geometry, not on summary formulas like Edmund's equation 4.

I've checked the numbers produced by the spreadsheet against Thorlabs' graph, and they agree closely. For example at NA 0.4 and cover glass 0.19, I compute a wavefront error of 0.075 waves, where the graph shows something around 0.070. Similarly at NA 0.5 and cover glass 0.4, I compute 0.35 waves where the graph shows around 0.38 .

But now if I plug in some of your numbers, like NA 0.45 and 1.13 mm, I get an SA of 0.68 waves, where Thorlabs' "diffraction limit" line is drawn around 0.07 waves. Similarly if I plug NA 0.40 into your formula, then your formula computes a value of 1.82, where in Thorlabs' graph, the orange NA 0.40 line crosses the dashed "diffraction limit" line at around 0.19.

Is one of your coefficients perhaps off by a factor of 10 or so?

--Rik

Lou Jost
Posts: 5933
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

Yes that's it, I made a decimal error, 0.7 instead of 0.07, when deriving the equation from the graph! That solves everything. I've edited my earlier comment and put the new numbers in. This provides a much harsher limitation and makes much more sense.

Have you had any more insights on why the 0.07*lambda condition is harsher than other commonly used conditions?

rjlittlefield
Site Admin
Posts: 23543
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Lou Jost wrote:Have you had any more insights on why the 0.07*lambda condition is harsher than other commonly used conditions?
No, no new insights. My best guess, which I'm not very confident about, is what I posted ABOVE.
It may have to do with the interaction between SA and defocus, or between SA and some other aberrations. While I very much doubt that anybody can see 1/16 lambda SA in an otherwise perfect image of a perfectly planar subject, it is very simple to see the difference between 1/4 lambda defocus and 5/16 lambda defocus. So it could be that a standard for SA around 1/16 lambda reflects some threshold of further degradation, added to an image that is already degraded by other aberrations.
--Rik

Lou Jost
Posts: 5933
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

Then for stacking, with small enough step size, can we reduce the defocus error and then accept more SA?

This SA business is critical to my current application. I was hoping to make stack-and-stitch mosaics of some of my newly-discovered micro-orchids, but the process takes too long in air, and the very fragile cloud forest flowers begin to wilt before the stack is done, or even before they get to my house from the forest. So I have been storing and photographing them under water. When laying flat they are about 2-4mm tall (z axis) so they have to be covered with slightly more water than that. I wanted to use my telecentric 10x Mitutoyo on full frame for this project but the SA from 5mm of water seems to ruin this plan. Even my 7.5x will be marginal. But maybe doing finer step sizes could earn me an extra millimeter or so of water, and that might enable me to use at least the 7.5x.

I don't know if the 5x Mitu is sufficiently close to telecentric for this stitching. But I would have liked to use slightly higher m in case these pictures get used in a museum exhibit and blown up large.

rjlittlefield
Site Admin
Posts: 23543
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Lou Jost wrote:Then for stacking, with small enough step size, can we reduce the defocus error and then accept more SA?
I don't know. The math suggests that this may work, but I am not confident that the model is right. It is time to experiment, I think.

--Rik

Lou Jost
Posts: 5933
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

I am now pretty confident that my equation (and the Thorlabs criterion on which it is based) is now realistic and accurate. We've discussed SA in relation to glass sensor packs elsewhere on the forum. In that discussion, we mentioned resolution measurements by Roger Ciccalo (of LensRentals) and Brian Caldwell which showed that the 4.2mm thick MFT filter pack made a significant impact, but only for lenses faster than f/2.8. My equation based on the Thorlabs graph predicts that lenses would notice the absence of a 4.2mm thick medium beginning at about f/2.75, which is exactly what happens.

From here I am going to start a new thread on this subject for my upcoming experiments, since I have to find a way to defeat this limit if I am going to be able to make high-resolution stack-and-stitch images of my micro-orchids (which are about 2-4mm thick) under water.

My idea is to find lenses that are designed to shoot through thick glass, like reversed MFT lenses, or Mitu G Plan objectives, or some "x-ray" lenses. Then I can effectively double the SA tolerance of the lens by adding water above the midpoint depth of the orchid until it matches the thickness of the medium that the lens expects. I can then go both up and down by the tolerance amount, giving me twice the tolerance that would be found using only an air gap between the lens and medium. This might be enough to solve my problem. On to experiments!

ChrisR
Site Admin
Posts: 8662
Joined: Sat Mar 14, 2009 3:58 am
Location: Near London, UK

Post by ChrisR »

There's a simpler formula from Leitz, in a document I came across in relation to a different topic, at http://www.science-info.net/docs/leitz/ ... -15-64.pdf.

Meanwhile an experienced microscopist at another forum relates that the effect of CG thickness errors depends in his experience on the the individual objective or application, suggesting that the method of SA correction used may make a difference to the Thorlabs equation. Lou's terminating phrase is apposite!
Chris R

Lou Jost
Posts: 5933
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

Chris, that's an excellent resource! But the formulas there seem to to answer a different question than the formula I extracted from the Thorlabs graph. The Leitz formulas relate the amount of spherical aberration to the amount of tube length needed to correct it, in a finite objective. My formula gives the impact of SA on the image. Your document does seem to reference a formula like mine in their "Technical Information Bulletin Vol 1 #4, but they don't quote the formula explicitly, juts one numerical result.

But the Leitz document reminds us that we can partly correct SA in finite systems by changing tube length. This means that we can, counter-intuitively, combat SA due to the medium by ADDING medium and moving the point of best lens correction below the surface of the medium, setting it to the midpoint depth of the subject. Then, as with my suggestion of using MFT lenses, we double the tolerable correction error, taking advantage of the tolerable deviations in both the positive and negative depth directions from the point of best correction. That will be the key to doing high res work.

A similar thing may be possible with infinity objectives by focusing the tube lens.

I'll start a new thread for experiments which I began yesterday..

Lou Jost
Posts: 5933
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

Returning to the question of the wavelength dependence of SA, I finally found an equation for coverslip tolerance that explicitly contains wavelength. It confirms my suggestion that tolerance is proportional to wavelength. Doubling the wavelength doubles the tolerance; halving the wavelength halves the tolerance.

https://www.researchgate.net/publicatio ... er_Glasses

abednego1995
Posts: 84
Joined: Tue Nov 01, 2016 11:53 pm

Post by abednego1995 »

Absolutely. As Lou points out in the ref. paper, SA induced by coverslip thickness and index variations will differ with wavelength.

However, how and how far it is compensated for depends on the objective.
As already pointed out in previous messages, normal achromats are usually SA corrected for only one wavelength, and I doubt that it would be flat across the aperture. Better corrected objectives will have better correction across multiple wavelengths, but even with the best it will be around 4 wavelengths and will be progressively worse with increasing aperture.

If you can get patent documents for a certain objective, It helps to see and understand those curves. As a possibility, if the residual SA at a certain wavelength and certain NA is the opposite of induced SA at a specific suboptimal parameter, I imagine monochromatic imaging at that wavelength would improve resolution for a specific periodic object feature. Possibility in getting some joy in using otherwise "unspectacular" objectives.

Cheers,
John
Last edited by abednego1995 on Mon Mar 11, 2019 7:38 pm, edited 2 times in total.

Lou Jost
Posts: 5933
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

Yes, using monochromatic light does turn some poor lenses into great ones!

I think these equations are assuming the lens is perfectly corrected, and are telling us how sensitive a perfect lens would be to the addition of a thick medium.

Medwar
Posts: 66
Joined: Mon Apr 16, 2012 1:46 am

Post by Medwar »

The topic of correcting spherical abberrations is very interesting for me.
I found information about a solution that is pretending to correct SA in tube lens (TL) for infinite objectives.
https://www.infinity-usa.com/infocus/
And I found an article on the same topic, though the article is written by the producer if the lens, and doesn't reveal important information on the design, except advertising the same product:
http://sharedlab.bme.wisc.edu/research/ ... ection.pdf

My goal is to make photos not through coverglass, but through a thick medium with high RI. E.g. 2mm of medium with RI = 1,7. So enourmous 2 mm should be corrected for RI =1.7

Does anyone has an idea on how to correct SA in TL, and what will be the LIMITS of such correction? For specific NA, e.g. 0.42 or 0.55 - what depth and with what RI can theoretically be corrected at maximum?

I wrote a letter to infinity-usa, but I have imprudently written "physical person" in the company field, and got no answer. I think maybe I could try to write them again, now filling in the name of any very rich buisiness to pretend for better results. Maybe someone of you guys already works for some big lab and can write them from your corporate email?

Lou Jost
Posts: 5933
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

I had looked at that device earlier....assumed it was way too expensive for me. But it does suggest that there is a way to correct SA between the objective and tube lens...I've played with simply refocusing but that didn't seem to work. Changing the tube length while using a Raynox or similar tube lens might be worth trying.

Post Reply Previous topicNext topic