DIC with condenser NA less than objective NA

[Beatsy]

I've been looking at the effect on resolution when condenser NA is less than objective NA. I couldn't find much written about this except for a brief mention in "Photography Through The Microscope" which states "...effective NA is very approximately the arithmetic mean (average) of the two".

My motivation to explore this is partly curiosity and partly a need to use a 0.63 condenser for the best DIC contrast in certain situations. DIC on my Zeiss ICM 405 comprises two Nomarski prisms in the condenser ("I" for objective N.A. <=0.4 and "II" for N.A. >0.6) with specific, matched sliders behind the objectives. I often mix and match sliders as I only have three but it usually works OK. For instance, a 100/1.3 PlanApo used with an N.A. 1.4 condenser oiled to the slide works really well with prism II and the slider for a 63/1.4 PlanApo. I have the 63/1.4 objective too, but it has a spot of delamination at the edge which wipes out most of the DIC contrast at full aperture.

Anyway, in short, most of my DIC slider/objective combinations work best with condenser NA=0.63 so I wanted to discover the resolution I could get from a high-NA objective while using DIC in this configuration. To date, I always worked with condenser NA close to or higher than objective NA ('cos I thought that's what you should do).

My quick test compared Neofluar 40/0.75 + 0.63 condenser front lens with Plan Neofluar 40/0.9 multi-immersion (water) + 0.63 and 0.9 front lenses. Images are unstacked single-shots though I took many of each to ensure similar focus across the set. I changed DIC sliders and/or prisms for each image too. This accounts for differences in OOF flare and DOF. Other experiments show mis-matched sliders can reduce resolution so it could skew these result a little. I used a late victorian strew containing Surirella Gemma frustules as the test subject. Surirella striae spacing is ~500nm, punctae ~400nm (which I expected to be a borderline case for NA0.9 objective + NA0.63 condenser).

First - Neofluar 40/0.75 + NA0.63 condenser. Avg NA0.69 - min resolved 480nm. As expected, striae resolved but not punctae.


Second - Plan Neofluar 40/0.90 imm + dry NA1.4 condenser (=NA1.0 condenser). Avg NA0.95 - min resolved 350nm. As expected punctae resolved.


Last - Plan Neofluar 40/0.90 imm + NA0.63 condenser. Avg NA0.765 - min resolved 440nm. Punctae resolved pretty much the same as above!?? I think there are more in focus due to the reduced condenser NA giving more DOF.


That last result was pretty surprising - I didn't expect the dots to be resolved as well as they were.

Conclusions? Well, nothing definite as there were too many variables and every test used a mismatched slider/objective pair. But I've satisfied myself that objective NA affects resolution far more than condenser NA (for an un-oiled condenser) and I can get better resolution with a higher NA objective, even if condenser NA is relatively low. Oh, and trying it out is always better than relying on equations or theory

Incidentally, I tried the correct slider for the 40/0.9 with 0.63 condenser. Although it produced the best contrast by far, it only resolved the dots when the DIC gradient was shifted so the darkest part was on the specimen. Any shift either way smeared the dots together again. The result shown above for 40/0.9 at 0.63 used the slider designed for a 63/1.4 PlanApo - go figure!

So I guess I need to experiment a bit more - but interested to hear any insight or anecdotes others may have about this.

[JohnyM]

Check M.Pluta "Advanced light microscopy" Vol 1 -3.
Condenser aperture is also determining a objective aperture. Ie if condenser has Na 0,4 and lens is 0,65 then effective aperture is only 0,4.
Also if condenser aperture is higher than maximum objective Na it is lowering the microscope resolving ability (it can be easly seen when using a diffraction mesh or maybe grating? ). So it should be matching or lower (condenser Na) for maximum preformance.
I've found that point resolving formula (very close to linear with small diatom) is best for calculating the realistic resolving ability. Also keep in mind that when using light other than monochrome and incoherent, the standard error (or devotion?) is higher (also the formula is slightly different).

[Pau]

Beatsy, interesting observations.

Matching DIC components is a kind of roulette, sometimes you're lucky and things work better than expected.

About condenser aperture, the classic formula is:

system NA= (objective NA X condenser NA) /2, so both can limit NA and resolution but not to the extent that Johny says.

But theorical resolution is only part of the story. To form the visible image you need both resolution and contrast and more contrast often limits resolution but can make visible details sometimes even under the theorical resolving power. For exemple, in BF when you close the condenser aperture to get contrast or in Phase Contrast whose condenser annuly are limiting the aperture but can make visible very fine details like fagella.
DIC is more complex to understand (at least for me ) but the resolution vs contrast tradeof also applies, even when you're using the full condenser aperture (some makers like Olympus now make series of DIC sliders diferencied for high res or high contrast)

In my limited experience the classic formula is not very accurate, at least when the difference between condenser and objective aperture is very big. Like you I usually find the objective NA more critical.

[rjlittlefield]

In my limited experience the classic formula is not very accurate, at least when the difference between condenser and objective aperture is very big. Like you I usually find the objective NA more critical.

A useful intuition about this is that resolution depends on the light that actually enters the objective. Some subjects cause the light to spread over wide angles, so that even though light arriving from the condenser may have a small NA, light leaving the subject can fill the entire aperture of the objective. In the extreme case the NA of the condenser becomes irrelevant. The classic averaging formula reflects some assumption about how light gets spread by the subject, so it will be more or less accurate in different cases.

--Rik

[Pau]

Good point, Rik. In darkfield microscopy the direct condenser light actually does not enter into the objective, just the light scattered by the sample, being a good example of your idea, but with any illumination technique, even with BF you really don't want the direct condenser light but the one reflected, retarded, diffracted or emitted by the subject.

[JohnyM]

Indeed, formula is most aplicable to phase, regular objects, like Rik said - and many other subject can dramatically change true resolving power, but it can never be better than the theoretical - always worse. Also in darkfield and phase contrast image forming is on absolutely different laws, so considered formula isn't applicable.
Resolving power is strictly theoretical and is rarely obtained, mostly in confocal and super resolution microscopes which are fine tuned for specific airy disc. There are 23 different formula for point resolution in M.Pluta (Pluta and Nomarski are considered fathers of useable DIC) book, and many more for grid and linear, and all of them are different but correct in specific situation.
There is different formula for coherent and for incoherent light, also abberations cause limited resolution. DIC optics is basing on small shearing, which also cause lower resolutin (and also this is the cause of directional contrast in DIC and other interference methods). The higher the shear distance - the higher contrast (eventually image is splitted in two), lower shear - higher resolution but lower contrast.
Also resolving power is different in X, Y and Z axes.
Making things visible and resolving it's two different stories. Ie you can detect a single molecule in fluorescence microscopy, but you cannot tell the difference if there is two or two thousand stacked in one pile.

Dry abbe condenser is giving about 0,8 sometimes higher, sometimes lower. Better corrected lenses are 0,9 Na and Dry maximum is 0,95 but no higher.

Pau, You're wrong. Objective Na is the limit or resolution of the system (with exception in some specific systems like 4Pi and other super-resolution). Condenser Na should be matching or lower (higher aperture than objective lens can limit the resolution, and iris is somehow simmilar to field diaphragm).
Ie after H.Hopkins basic grid resolution (suitable for most brightfield observations): if your objective Na is 0,65 and condenser Na is 0,40 you get formula d=L/0,65+0,4
If your objective Na is 0,40 and condenser Na is 0,65 you get formula
d=L/0,40+0,40 (in reality, higher the condenser Na, lower the resolution).

Where: d - microscope theoretical grid resolution, L - wavelenght of used light.

As you said, in practice contrast with full Na is very low and it's commonly considered as correct if condenser Na is stopped to obtain about 2/3 Na (that's why 0,9 condensers are commonly chosen over the 1,4 Oil).

Summary - You cannot make your resolution higher with only better condenser, if you already matched objective Na.

I can post references if you're interested. Sorry for my bad english, I'm not a native speaker.


Edit: About DIC combinations, i know what you mean It's all about matching condenser and objective aperture conjugate plans. Magnification and aperture has little meaning on this. Thats why i liked my PZO DIC so much. It's much more flexible than any other DIC system. Now i'll be testing DIC sliders and different objectives on Microphot.

[rjlittlefield]

Objective Na is the limit or resolution of the system

I agree. That is why I disagree with your claim that

"if condenser has Na 0,4 and lens is 0,65 then effective aperture is only 0,4. "

In fact the effective aperture can be as high as 0,65 in this case, if the subject spreads light well.

Condenser Na should be matching or lower (higher aperture than objective lens can limit the resolution, and iris is somehow simmilar to field diaphragm).

I think that what you wrote is not what you intended to mean.

There is no problem with resolution if the condenser has larger NA than the subject. The excess light will just get wasted by not entering the objective. There may however be a big loss of contrast, especially if stray light from outside the field starts bouncing around inside the objective. (One of my objectives has quite shiny inside walls. It's actually a bit scary to look into!)

Summary - You cannot make your resolution higher with only better condenser, if you already matched objective Na.

Agreed. But resolution does not necessarily drop as low as the condenser NA, either.

--Rik

[Ichthyophthirius]

Hi all,

The calculation the the resolution of the condenser/objective system is the subject of a number of papers. Here is the result from one of them:

"For instance, if an objective of N.A. 1.4 is used with a condenser-aperture of 1.4, and the iris diaphragm is then closed to give N.A. 0.7, p only changes from 240 to 285 nm.; or, to express the same facts in another way, the resolution falls from 417 to 351 lines per 100 p, a drop of only 16 per cent." For N.A. 0.6, the autor calculated 293 nm for green light.

Baker (1952): Remarks on the Effect of the Aperture of the Condenser on Resolution by the Microscope. Quarterly Journal of Microscopical Science 93 (4): 375-377
http://jcs.biologists.org/content/s3-93/24/375.full.pdf

The N.A. of the objective has the biggest effect on the resolution of the image, since, as Rik said, the scattered/diffracted light from the object is still collected over the whole objective aperture. Within reason, the condenser aperture is only relevant of your object has features within the 240-290 nm range (as in Suriella), so only really for diatomists.

This is why Zeiss does allow/recommend/list the use of the Planapo 64/1.4 together with the condenser N.A. 0.63. It's normal practice.

[Perl]

If the Condenser are Na 0.4 and Objective 0.65 the Na of the system can not be more than 0.4 - the lowest Na in system will limit what to get out - do the messure and se

[Ichthyophthirius]

Beatsy, for the newer DIC system, Zeiss produced two INKO condensers,
46 52 85 with front lens N.A. 1.4 and
46 52 73 with front lens N.A. 0.63 (long working distance condenser)

The function of the condenser is to image the interference fringes of the compensator prism (condensor prism) in the back focal plane of the objective. If you exchange the front lenses (1.4 and 0.63, or even 0.9), they will not do that accurately any more and in any case, the distance between the fringes in the image will change, meaning that they don't fit the objective prism any more. You can, however, find chance combinations that do work, as you said.

Zeiss did list the use of condensor 46 52 73 (front lens N.A. 0.63) with the Planapo 64/1.4 (and its own slider) as normal use. The condensor 46 52 73 will only work with front lens 0.63 and its own factory-fitted prism II in that case, but not with front lens 1.4 or prism II from condensor 46 52 85 (it might work if you exchanged both; haven't tested it). As stated above, there is a minimal loss of resolution when using the NA 0.63 condensor. However, there are advantages. As Rik said, there is better contrast because of less stray light. Also, a low NA condensor has less spherical aberation than a high NA condensor. This means that it is easier for the condensor to image the interference fringes through thick specimens and/or layers of water. So this condensor is "more forgiving" when it comes to sample preparation and easier to use.

There has been a trend for the microscope manufacturers to use DIC condensors with N.A. 0.9 and lower as standard. They are sufficent for most uses.

[Ichthyophthirius]

Beatsy, You obviously have experience with diatoms and know what methods work best for you. But two remarks that might be interesting.

First, using DIC slightly reduces resolution of your image (DIC trades contrast for resolution, as JohnyM wrote above; specifically in x- and y-axes; z-axis resolution is dramatically improved in DIC). For the very finest details, brightfield or oblique illumination will give you better resolution. The manufacturers recognised the loss of resolution in DIC, and are now offering lower-contrast high-resolution DIC sets (i.e. they offer high-contrast as well as high-resolution sets, depending on what you want to image). The Zeiss system you have would probably be classed as universal, i.e. somewhere in the middle.

You used 350nm light for one of the images. One thing to keep in mind is that the polars (polarisation filters) used in regular microscopes are not designed for that wavelength. They already have a violet/blue gap (they are quite transparent for light around 400 nm). They will be even more transparent at 350nm. They will also be damaged by UV, although I don't know how badly by this relatively long wavelength.

Regards, Ichthy

[Ichthyophthirius]

If the Condenser are Na 0.4 and Objective 0.65 the Na of the system can not be more than 0.4 - the lowest Na in system will limit what to get out - do the messure and se

Hi Pau, The effect on resolution is less than one might think. Have a look at the published literature, for example Pluta (1989) Advanced Light Microscopy or the paper I linked above, if you can access it.

[JohnyM]

Interesting discussion we have here On Polish forum, members do great photos, but lack of technical expertise

Objective Na is the limit or resolution of the system

I agree. That is why I disagree with your claim that

"if condenser has Na 0,4 and lens is 0,65 then effective aperture is only 0,4. "

In fact the effective aperture can be as high as 0,65 in this case, if the subject spreads light well.
...
--Rik

Exactly, thats why i mentioned, that there is many formulas with different situations. Maybe i did not make it very clear, thanks for making it so .

But in case when condenser aperture is higher than objective lens, the resolution is dropping slightly due to many factore, ie "darkfield effect" of scattered light, which is not strong enough to form an image but can make immage somehow blurrish.
I post an photo of M. Pluta book, Polish edition, which illustrates discussed problems (It's the same book as "Advanced light microscopy", but all three volumes in one and in Polish).

I do lack of exipment to make a quality scan, but essentials are visible. I'll try to translate:
"Impact of condenser numerical aperture on image contrast and microscope resolution: a) numerical aperture Ak of condenser HIGHER than numerical aperture Aob of objective, b) Ak = Aob, c) Ak/Aob = 0,65 , d)Ak/Aob = 0,50 , e) Ak/Aob = 0,30 , f) Ak/Aob = 0,20. Objective lens planachromatic power 40x and numerical aperture Aob = 0,65. Specimen: amplitude diffraction grid p= 1,1 um. "

Example is perfect, like Rik said it's amplitude, so has ability to fill empty aperture with some light and system is quite effective with Ak/Aob as low as 0,5. But you can also see that image is slightly worse, when condenser Na is exceeding the objective Na (ie bright spots are slightly larger). "b)" has perfect resolution and "c)" has perfect compromise of resolution and contrast. Im not sure if picture show it as well as original.

...For the very finest details, brightfield or oblique illumination will give you better resolution.

Exactly, but bear in mind, that obligue light is so similar to DIC due to its uneven resolution in two dimensions, so it have same (lower, but depending on how much aperture is closed) directionality as DIC.

[Beatsy]

Wow - this forum is an incredibly useful resource! Thanks everyone for your comments. An extremely interesting thread.

Special thanks to Ichthyophthirius for the link to the Baker paper and the info on Zeiss DIC condensers. The Zeiss info explained a couple of things that had been puzzling me. I have the 0.63 LWD DIC condenser, which accounts for the "less than optimal" DIC performance with 1.4 front lens added instead. I've been looking for a 0.9 (pol) front lens for 9 years (in case that might work better), but rumour has it that Zeiss never released it. There may be "sales samples" around though - somewhere!?

I should clear up a couple of things that were unclear in my first post. All images were taken in white light. Each is 100% crop from the centre of a RAW camera image that was downsampled 50% (my 5d MkII on the 3.2x mag factor of the Zeiss front photo port oversamples too much at 40x and above).

Also, "min resolved" figures above each image are theoretical minimums calculated using the average of condenser and objective NA in the classic formula (assuming 550nm light wavelength).

Finally, I checked (measured) the feature spacings. Striae are 500nm and punctae are 415nm on this specimen. The 40/0.9 with NA0.63 condenser is clearly outperforming the (simplified) prediction of the classic equation!

Thanks again for all your comments.

[Ichthyophthirius]

Hi, Yeah sorry, I misread that your calculations as you using a 350nm ultraviolet illumination

The front lens 0.63 is the only one that will work as intended on this condenser. Their exchange for 0.9 or 1.4 was not intended by Zeiss and will only give good results by accident. As far as I am aware, the phase rings will also not work as intended - an indication, that the magnification of the front lenses is different.