Decapping and Chip-Documentation - Howto

[magic]

 

I think I already mentioned it a few times in other threads, but maybe not here yet: I did (approximately) the same thing, two years ago. I bought a biological microscope stand plus several various RMS objectives and I mounted my DIY illuminator and USB camera stack on that. (I would have bought a complete metallographic scope, but they are not that cheap and common in this post-commie hole here). For "flat" closeups of silicon dice you simply can't beat the convenience of a real scope, with its mechanical stability, precision focus, X-Y movements and switchable objectives.

The opamp pictures I posted last year all came from this setup. They don't look too bad and each took no more than a few minutes to produce, I like working like that

Meanwhile I'm experimenting with focus stacking on the M.Zuiko+Raynox setup. A nice thing about using a non-reversed lens is that it works normally, including focus (which is electronic on Olympus lenses). I can produce a stack simply by rotating the focus ring a little between each shot, without any extra gear like focus rails or whatnot. And it seems to be working. The electronic focus ring moves so lightly that operating it doesn't even move the camera too much.

[Noopy]

Meanwhile I'm experimenting with focus stacking on the M.Zuiko+Raynox setup. A nice thing about using a non-reversed lens is that it works normally, including focus (which is electronic on Olympus lenses). I can produce a stack simply by rotating the focus ring a little between each shot, without any extra gear like focus rails or whatnot. And it seems to be working. The electronic focus ring moves so lightly that operating it doesn't even move the camera too much.

But manual focusing by rotating the focus ring a little is also possible with reversed-lens setups. That is what I usually do.


...a microscope is so cool! 

[magic]

I don't know, there is something about the size, weight and noise of SLRs that I just don't like very much
On mirrorless systems, electronic manual focus is the norm. No camera connection, no focus. AFAIK it's same thing with Canon and Nikon too.

[Noopy]

Ah, ok, of course you need a mechanical focus ring. 

[mawyatt]

The various Raynox Close Up Lens types work quite well with many different lens types including microscope objectives. We used the Raynox DCR-250 and DCR-150 with various Mitutoyo objectives as "tube lenses" with superb results for chip imaging. These have been used for decades for this purpose by many folks, you can find lots of detailed info over here:

https://www.photomacrography.net/index.htm

Recall using lens focusing (lens focus rings) for stacking around 1 to 1, however as magnification increases beyond 1 to 1 changing to physically moving the lens/subject (focus rails) usually improves results, you can find details above at PM.

There are many lenses that can be utilized for chip imaging, many are inexpensive and/or repurposed. These range from the old film types, to generic objectives, old Russian Lomo using special glass (illegal to produce today), copier/scanner lenses (Minolta DImage) and so on. Here's some great resources on such.   

https://www.closeuphotography.com/

http://extreme-macro.co.uk

Noopy, please keep those images coming, you have produced an incredible amount of quality chip/die images

Best,

[Noopy]

Noopy, please keep those images coming, you have produced an incredible amount of quality chip/die images

I will do so! 
The pictures will just get better in quality. 

Best regards!

[Noopy]

After learning quite a lot about microscopes I finaly bought one! For professional images of integrated circuits you usually use reflected light microscopes. Among other things, these microscopes can be used to examine the crystal structures of metals. Therefore they are often referred to as metallurgical microscopes. As the name reflected light microscope suggests, the object is illuminated from above. This type of illumination is also known as coaxial illumination. With very high magnification factors and correspondingly small working distances, it is often not possible to illuminate the object sufficiently in any other way. The incidence of light from above also ensures that the interference in the semiconductor structures correlates strongly with the layer thicknesses. This results in the familiar, very colorful images that clearly show the different structures of the circuits.

Unfortunately, reflected light microscopes are expensive. It should also be noted that, as with cameras, you can also invest a lot of money in the objectives. In the hobby sector, the relatively old and correspondingly inexpensive Olympus BH2 is often used. The microscope, suitable lenses and spare parts are readily available. The successor to the BH2 is the BX family. Good offers can also be found from the BX family from time to time. Here you can see the Olympus BX51.

At high magnification factors, reflected light microscopes deliver significantly better image quality than can be achieved with an SLR camera. However, they also have disadvantages. If you want to image objects with a certain depth, you are often struggeling with a short working distance and a shallow depth of field.




The microscopes in the BX family have a modular design and offer many different configuration options. There are models with transmitted light, reflected light and with both light options. Only the reflected light option is integrated here. The light source is a halogen lamp (yellow). It is located in a housing behind the microscope. The brightness can be adjusted using a dimmer integrated in the microscope.

The light source is connected to an elongated module. The elements required for so-called Köhler illumination are integrated in this module (red). More on this later. There are two slots where filters can be inserted (cyan). However, the microscope also offers options for filters in other places. The light is ultimately directed into the light path via a semi-transparent mirror (blue). The light with which the object is illuminated therefore also passes through the objective (pink).

The light reflected from the object (white) passes through the objective, the semi-transparent mirror and finally reaches the triocular head. There, the light is guided to the eyepieces (green) via prisms (orange). The triocular head also has a connection for a camera. A slider allows you to switch between eyepiece and camera. The light can also be split in a ratio of 30:70.

 


With the older Olympus BH2, the lenses do not yet provide a perfect image of the object. Special eyepieces compensate for the weaknesses of the lenses. For this reason, you also need a special eyepiece in the connection for the camera. The BH2 is also so old that the camera connection was still designed for analog cameras, especially for cameras with large film formats. The type of eyepiece that is suitable for today's common sensor sizes is very rare and correspondingly expensive.

The lenses of the BX family have the great advantage that they are fully corrected, that means they deliver an error-free image within the limits of their capabilities. There are various old and new adapters for the camera mount, which are more or less expensive. They usually contain lenses that reduce the projected image of the object. However, if the projected image circle matches the camera sensor to some extent, a camera can also be connected directly to the microscope without optical elements. All you need is a mechanical adapter with a so-called dovetail for connection to the microscope and a bayonet for the camera. Such adapters can be purchased cheaply from Aliexpress or made yourself using a 3D printer. Without additional lenses there is no risk of introducing additional optical errors into the system.

As with macro photography, it is also an advantage here if the camera can take pictures without moving the mirror and curtain. The BX51 is very stable and heavy. At high magnification factors, vibrations nevertheless lead to a deterioration in image quality.




The length of the camera adapter should correspond to the specifications. Olympus developed the UIS2 objectives for the BX family. In a “Microscope Components Guide” Olympus shows with which distances the optical system was designed. The relevant length here is 102mm from the dovetail above the trinoculars to the sensor. Please note the flange focal distance, which is 44mm for Canon cameras with EF lenses. This means that the adapter itself should be 58mm long.

If you want to adapt a camera to a microscope, you have to make sure that the sensor size and pixel size match the microscope and the objectives. Many of the UIS2 objectives offer an image circle with a diameter of 26,5mm. The dimensions of the APS-C sensor in the Canon 90D are 22,3mm x 14,8mm. The sensor is therefore very well illuminated. With full-frame cameras (sensor width 36mm) the illumination would no longer be sufficient. Cameras for microscopes often have much smaller sensors, which means that a lot of the projected image cannot be used. To adjust the image area, you can increase or decrease the distance to the microscope. This simply enlarges or reduces the image circle. At the same time, however, you risk a deterioration in image quality. Alternatively, as already described, you can use adapters with lenses that provide the desired magnification factor. However, these lenses must be of high quality, otherwise the image quality will deteriorate too.

When selecting a camera, you must also pay attention to the pixel size. The pixels should be small enough so that they are not the limiting factor in the resolution of the overall system. If the pixels are significantly smaller than the resolution, this is not a disadvantage optically, but you will produce very large images in which many pixels contain the same information as the neighboring pixels. This is the case with the Canon 90D used here.




The Olympus BX51 has an XY stage, which is extremely helpful for positioning and moving the object. When documenting integrated circuits, it is important that the die is positioned as ideally as possible in the narrow focal plane of the microscope. A tilting stage is used to achieve this. Two micrometer screws allow the object to be tilted around the X and Y axes.

The entire stage moves in the Y direction. In the X direction, only the two black screws, which can be seen on the right, move. These screws normally fix a sample holder, which in turn fixes a carrier glass with the object to be examined. Since in this case the relatively large tilting platform has to be moved, a PCB was cut to fit and fixed to the X-rail with the two screws. The tilting platform is not too heavy, which means that this mechanically less than ideal solution delivers very useful results.




The vertical element shown here can be used to move the XY table. The lower cylinder controls the X-axis. The upper cylinder controls the Y-axis. The rotary knob next to it is present on both sides and controls the Z-axis. The cylindrical body of this rotary knob represents the coarse drive. The cover provides a fine drive, which is very useful for large magnification factors. A smaller knob can be screwed onto the lid, which in this picture is just on the other side of the microscope.




Köhler illumination is a special type of illumination that shapes the light in such a way that it illuminates the object as optimally as possible. To do this, for every objective two diaphragms must be adjusted, the field diaphragm “FS” and the aperture diaphragm “AS”.




The field diaphragm makes it possible to limit the illuminated area. In this image, the diaphragm has been closed so far that it is visible in the image. Normally, the diaphragm is opened from here until it is just no longer visible. It makes sense to restrict the illuminated area as much as possible, as otherwise there is a risk of light being refracted and reflected by lenses or the walls of the light path. If it then finds its way to the camera, this stray light degrades the image quality.






To set the aperture diaphragm optimally, you have to remove one eyepiece and observe the image that appears in it. If the aperture diaphragm is closed, it becomes visible in the image. The best possible image quality is achieved when the aperture diaphragm restricts the image to 2/3 to 80%. A wider aperture increases the resolution, but reduces the contrast and depth of field. A diaphragm that is closed further reduces the resolution slightly, but increases the contrast and depth of field.




The images shown here show how the aperture diaphragm setting affects the image quality. The objective used had a magnification factor of 20x and a numerical aperture of 0,4. The completely closed aperture (right) provides the worst image quality. The difference between the fully open aperture (left) and the approximately 2/3 closed aperture is not too great with a 20x lens. However, you can see a little more detail in some places with the 2/3 aperture.




With a 100x lens, the differences are somewhat more obvious.




The image on the left is the best that could be achieved with a retro objective. The image on the right, which was taken just with a 20x objective, clearly shows the better quality of the microscope images.






If coaxial illumination is used, this is also referred to as brightfield microscopy. The BX51 presented here also offers the option of darkfield illumination. Special objectives are required for this. With dark-field illumination, a light channel directs the light downwards around the actual objective. At the lower end of the objective, the light is then reflected towards the object.




The darkfield illumination primarily emphasizes structures such as edges, irregularities and soiling, which appear particularly bright with the side illumination.


https://www.richis-lab.de/Howto_Microscope.htm

 

[RoGeorge]

Congrats for your new microscope! 

[iMo]

Fantastic!!! Waiting on new REF80 shots!!
 

[Noopy]

Thank you! I'm really happy! 

[mawyatt]

Nice "Scope" Noopy

Best

[magic]

That's some good stuff
I see you are focus stacking already

[Noopy]

In my next posting I will show you what objectives I bought and why. Also an interesting topic!

Future plans in this area are: automatisation and RGB illumination

I see you are focus stacking already

How did you figure it out?  Yes, the 100x objective needs focus stacking.

Fantastic!!! Waiting on new REF80 shots!!
 

Waiting for the REF80.  But should arrive soon.

[iMo]

What you may need next is a software for an automatic translation of the die pictures into a schematics. Perhaps "AI" could do it..?? 

[magic]

BTW, this explains pretty well how these reflected light microscopes work.

My matchbox was a slightly simplified version of the same basic idea.

As for REF80, I'm afraid there is a lot of top layer metal there.

[Noopy]

What you may need next is a software for an automatic translation of the die pictures into a schematics. Perhaps "AI" could do it..?? 

I think AI could help a lot in this field. But I´m afraid there are not very much customers for such an AI product and so the price to pay would be high. Who knows, perhaps there are already some companies doing reverse engineering with AI...

As for REF80, I'm afraid there is a lot of top layer metal there.

We will see. in such special analog circuits there are often just two metal layers. We will see. As you know I can at least strip all layers. 

[Noopy]

In its “Microscope Components Guide”, Olympus shows the most important parameters that are usually shown on objectives. The first line shows the name of the objective. The second line shows the magnification factor and the numerical aperture. Letters at the end of the second line indicate special variants.

The infinity sign in the third line shows that the lens is infinity corrected. This means that the light rays leave the lens in parallel. Most modern lenses are infinity corrected. The following number stands for the cover glass thickness, which is included in the calculation of the optics. In biology, a cover glass is usually placed on the object. Here the number 0 indicates that you should work without a cover glass in order to obtain an optimal image. The last number is the field number. It indicates the diameter of the image in the eyepiece.

If a lens is designed for use with an immersion liquid, this is usually printed on the lens. Sometimes the working distance is also printed. A colored ring makes it easy to see the magnification factor at a glance. Certain colors are assigned to common magnification factors.




The magnification factor often seems to be the most important parameter of an objective. However, as long as the camera sensor offers sufficient resolution, the numerical aperture is much more important. The NA value defines the resolving power of the lens. The “Microscope Components Guide” shows how to calculate which structures can be resolved. For an objective with an NA of 0,9 the minimal structure width is 0,37µm. The resolving power also depends on the wavelength of the light. Green light was used in this calculation. With red light, the resolution drops to 0,47µm.




In the “Microscope Components Guide”, Olympus also explains the meaning of the designations of its objectives. The M as the first letter indicates that it is a metallurgical objective, i.e. designed for coaxial illumination. LM and SLM lenses offer increased working distances. LC stands for special variants that can be used for reflected light microscopy through glass layers.

PL or plan means that the objective projects an image without curvature. If the image is only viewed directly, without a camera, a certain amount of image curvature is not a problem. However, when using cameras and if you want to create panoramic images, the images must be as distortion-free as possible. The following letters show how well the chromatic aberration has been corrected. This is followed by the magnification factor. The last letters stand for special properties.




The following images are from the standard cell ASIC U1525FC007, which was manufactured with a minimum structure width of 1.5µm (https://www.richis-lab.de/logic27.htm).




The UPlanFLN 4x 0,13 objective offers the smallest magnification factor of the objectives examined here. It can image an area of 5,6mm x 3,7mm. The die of the U1525FC007 is very large with an edge length of 7,5mm. ICs that are not so large can be completely imaged with the UPlanFLN 4x 0,13. For larger objects, a panorama must be created. In this case, 6 images would be necessary for the U1525FC007. Objectives with such small magnification factors usually also have a small numerical aperture. Mathematically, the value 0.13 can only resolve structure widths of 2,6µm (wavelength 550nm).

The UPlanFLN 4x 0,13 is actually a biological objective, i.e. intended for transmitted light microscopy. However, no optical weaknesses can be seen in reflected light microscopy. Low magnification factors usually offer a large working distance. Here it is 17mm. The depth of field is also relatively large with 16,27µm. The preceding U in the designation shows that this is a revised version of the objective.




The UPlanFLN 4x 0,13 provides good overview images. However, if you want to view a single standard cell, the resolution is not sufficient. At full resolution, there is one pixel for 800nm on the die.




The LMPlanFl 10x 0,25 BD is an objective with a large working distance of 21mm. It is optimized for use in metallurgical microscopy and also allows darkfield illumination. With a magnification factor of 10, the area that is imaged shrinks to 2,3mm x 1,5mm. If you want to capture the entire die, 24 images are required. At the same time, the focal plane shrinks to 18µm. The NA value of 0,25 is not particularly high for a lens with a magnification factor of 10. This is due to the optimization for a large working distance. The objective allows structure sizes of 1,3 µm to be resolved.




The standard cells are now more clearly recognizable. The minimum structure width of the process is 1,5 µm. It is already possible to analyze the circuit. However, in areas where several structures overlap, the image is still unclear.




The LMPlanFlN 20x 0,40 BD is also a lens with a long working distance, which in this case is 12mm. The magnification factor is 20 and the NA is specified as 0,4. With an imaged area of 1,1mm x 0,8mm, at least 70 images are required to fully document the die. The resolution drops to 0,84µm. However, the focal plane also shrinks and is now just 6,1µm.




With the depth of field of the LMPlanFlN 20x 0,40 BD, it is difficult to align the die sufficiently flat. With this process, you can just take pictures without focus stacking, although the image quality is already borderline in this respect. The top left corner shows a certain amount of blurring. This could be due to imperfect alignment. However, it is also possible that the objective has a weakness there. If you want to create a panoramic image with this image quality, the blurring can already lead to artifacts and make circuit analysis more difficult. Instead of performing focus stacking, you can also crop the image and only use the sharp area. However, this reduces the area that can be imaged at once even further.

With the available depth of field of 6,1µm, you have to bear in mind that the structures of the integrated circuit have a certain height. A metal layer may well be 1µm thick. The oxide layers between the wiring levels can also be 1µm thick. This quickly leads to total thicknesses that require focus stacking even with perfect alignment. Overall, however, the image quality is now sufficient to analyze the circuit. The two polysilicon layers are clearly visible. Assigning overlapping structures to their functions is also possible.




The magnification factor of the UMPlanFl 50x 0,80 BD is just 0,45mm x 0,30mm. This means that 442 images would be required to fully image the die. At the same time, however, a very high numerical aperture of 0,8 is achieved, allowing a resolution of 0,42µm. However, the significantly increased numerical aperture comes at the price of a small working distance of just 1,0mm.




The image section is now very small. However, the structure of the standard cells can be seen very clearly.




With the shallow depth of field of 1,3 µm, focus stacking must be carried out. If the limits are defined optimally, 14 individual images are sufficient for this circuit. Focus stacking with 14 images is not a major effort. However, you must bear in mind that the effort is multiplied. This means that 6.188 images are required to fully map the ASIC.




The working distance of 1,0 mm can quickly become a problem. If the circuit is located in a ceramic package which has side walls, there are usually collisions. Larger superstructures are out of the question anyway.




There is also a objective with a large working distance in this range. The LMPlanFl 100x 0,80 BD also offers an NA of 0,8 but allows a working distance of 3,3mm. These specifications come at the price of an increased magnification factor of 100x and a further reduced depth of field of 0,87µm. The LMPlanFl 100x 0,80 BD images just 0,23mm x 0,15mm of the circuit. This means that 1.650 images are required for a panorama. With the 17 images required for focus stacking, this means that 28.050 images would have to be taken in order to optimally image the ASIC.




The image quality is just as good as with the UMPlanFl 50x 0,80 BD, only the imaged area has been reduced.




The diagram above is taken from the article “Systematic design of microscope objectives. Part I: System review and analysis” by Yueqian Zhang and Herbert Gross and shows two important relationships. Firstly, the numerical aperture and thus the resolving power of ordinary light microscopy cannot be increased arbitrarily. The limit is 0,95. For green light, this corresponds to a resolving power of 0,35 µm. If you want to increase the NA value further, you have to replace the air between the objective and the object with another medium and thus optimize the refraction of the light. This process is called immersion. Water or various oils can be used for this purpose. The lenses must of course be designed for this. With oil immersion, NA values of 1,5 can be achieved, which corresponds to a resolving power of 0,22 µm.

The diagram also shows how critical the thickness of the cover glass is if the lens has been optimized for use with a cover glass. There are many biological lenses on the second-hand market for which this applies. With low NA values, the thickness of the cover glass is less critical. This goes so far that objectives such as the UPlanFLN 4x 0,13 do not even specify a number for the cover glass. There is just a “-”. It simply does not matter whether and which cover glass is used. At higher NA values, however, the image quality deteriorates increasingly if the correct cover glass is not used. Lenses for oil immersion are less critical in this respect, as the refractive index of oil is relatively similar to the refractive index of glass.




The Zeiss Primo Plan-Achromat 100x 1,25 shown here is a typical biological objective designed for oil immersion. Next to the NA value is the note “Oil”. This objective has some disadvantages. Although Zeiss objectives are infinity corrected, they usually still have minor optical weaknesses that are corrected within a Zeiss microscope. This results in somewhat poorer image quality in an Olympus microscope. In addition, the field of view is smaller. This means that the camera sensor is no longer illuminated in the best possible way. The working distance is 0,14mm. If you work without a cover glass, you can add its thickness of 0,17mm.




Such a lens is used as follows: You place a drop of immersion oil on the object and then move the lens into this drop. There are different oils. You do not necessarily have to use a special oil from one manufacturer. However, it should of course be an oil that is intended for microscopy. Care must be taken to ensure that dry objectives do not come into contact with the oil, as it can penetrate the interior and impair the image quality.




Even without the cover glass, the result is a very interesting image. The image quality deteriorates considerably at the edges of the image. This is not surprising given the small field of view.




In direct comparison, the Zeiss 100x 1,25 does not give you much more detail than the LMPlanFl 100x 0,80 BD. It even appears blurrier in some areas. However, it is noticeable that the vias in the metal layer are clearly different. They are shown larger and so structures can be seen in them that were not visible before.




With the darkfield illumination option, the Olympus BX51 uses BD lenses with an M26 thread. If you want to use other lenses, you usually need adapters, but these can be obtained cheaply from Aliexpress.




Here you can see a more recent integrated circuit with several metal layers. The LMPlanFl 100x 0,80 BD lens was used and focus stacking was performed. However, the shallow depth of field of the lens can also be used to advantage with such circuits, as can be seen in the following images.








The different focus levels sometimes make it easier to trace the lines or to work out special structures of one level more clearly.


https://www.richis-lab.de/Howto_Microscope_Objectives.htm

 

[RoGeorge]

Wow, impressive! 

[magic]

This is turning into Optical Equipment Anonymous

Your Nikon looks interesting, but that 0,17 is going to suck a lot, as you have perhaps found out already. Chips can be glued to cover glasses, which definitely helps with such objectives, at the cost of some glare (nobody seems to be doing cover glasses with AR coating on top side ). I'm not yet entirely sure what would be the best glue for this application. I tried a common water-based paper glue, but it leaves bubbles after drying. Molten colophony works and is very quick to apply, but it's prone to cracking. I don't like permanent glues which can't be easily removed if necessary.

The Zeiss is definitely held back by its uncorrected CA. The edges become visibly sharper with just a simple software correction and they would surely be sharper still if the CA weren't there in the first place (software corrects for different RGB channels seeing different magnification, but not for different magnification of different wavelengths seen by the same channel). The center is certainly better than on Olympus, as it should be. The small dots on poly are smaller, for example. You would likely see a more obvious improvement if using a newer and denser CMOS chip. But it doesn't matter much because of the CA.

Those DoF numbers are slightly suspect. I certainly find it hard to believe that 10x/0.25 has more DoF than 4x/0.13 And the two /0.80 should theoretically be equal if taken to their resolution limit, but perhaps the 50x spec assumes that your sensor (or eye) doesn't have enough resolution to notice that some details are slightly out of focus.

Another thing is that those numbers assume dry air, while chip layers are "immersed" in the passivation oxide with refractive index significantly higher than unity. NA of the ray pencils remains unchanged as they enter a different medium, but their angle changes and DoF is higher than in air. This also means that you cannot gain DoF by applying an immersion medium above the chip, which idea has once been suggested by a certain blogger...

[Noopy]

Your Nikon looks interesting, but that 0,17 is going to suck a lot, as you have perhaps found out already. Chips can be glued to cover glasses, which definitely helps with such objectives, at the cost of some glare (nobody seems to be doing cover glasses with AR coating on top side ). I'm not yet entirely sure what would be the best glue for this application. I tried a common water-based paper glue, but it leaves bubbles after drying. Molten colophony works and is very quick to apply, but it's prone to cracking. I don't like permanent glues which can't be easily removed if necessary.

The Nikon was just an experiment. There is a reason why you don´t see pictures of it. 
In my view using a coverglass is not an option in our business. Either the picture quality is good enough without a coverglass or you have to find a different objective.

The Zeiss is definitely held back by its uncorrected CA. The edges become visibly sharper with just a simple software correction and they would surely be sharper still if the CA weren't there in the first place (software corrects for different RGB channels seeing different magnification, but not for different magnification of different wavelengths seen by the same channel). The center is certainly better than on Olympus, as it should be. The small dots on poly are smaller, for example. You would likely see a more obvious improvement if using a newer and denser CMOS chip. But it doesn't matter much because of the CA.

I got the Zeiss very cheap so it was ok to start with immersion microscopy. But it´s not ideal of course.
A good immersion objective would be nice. But you always have to ask yourself if you really need it. I´m not sure if it makes sense for me to invest in such an objective. A NA of 0,8 is already a lot.

Those DoF numbers are slightly suspect. I certainly find it hard to believe that 10x/0.25 has more DoF than 4x/0.13 And the two /0.80 should theoretically be equal if taken to their resolution limit, but perhaps the 50x spec assumes that your sensor (or eye) doesn't have enough resolution to notice that some details are slightly out of focus.

Hm, you are right. I didn´t realize that these numbers are a little strange.
The DoF numbers of the MPLAN are take from the Olympus catalog so they have to be right. For the 4x 0,13 I found no numbers from Olympus directly. I took the numbers from Edmund Optics:
https://www.edmundoptics.de/p/olympus-uplfln-4x-objective/29226/
The numbers of the two 0,8 objectives are both taken from the Olympus UIS2 catalog so the difference should be correct:
https://exhibitors.productronica.com/download/1121_1_4_2240_4_1_223/productronica_m1703e_microscope_components_guide_enpdf.pdf
And for me looking back to "normal camera objectives" it makes sense. A longer lens has less DoF. But I´m no expert regarding optics...

Another thing is that those numbers assume dry air, while chip layers are "immersed" in the passivation oxide with refractive index significantly higher than unity. NA of the ray pencils remains unchanged as they enter a different medium, but their angle changes and DoF is higher than in air. This also means that you cannot gain DoF by applying an immersion medium above the chip, which idea has once been suggested by a certain blogger...

That´s a point. The passivation increase the DoF by 1,5.

[magic]

The Nikon was just an experiment. There is a reason why you don´t see pictures of it. 
In my view using a coverglass is not an option in our business. Either the picture quality is good enough without a coverglass or you have to find a different objective.

Well, either that, or you simply shoot through a cover glass (example)
For a quick job like here there is no need for glue, just a drop of water.
But the chip must be flat: no package, no bond wires, no other junk.

I got the Zeiss very cheap so it was ok to start with immersion microscopy. But it´s not ideal of course.
A good immersion objective would be nice. But you always have to ask yourself if you really need it. I´m not sure if it makes sense for me to invest in such an objective. A NA of 0,8 is already a lot.

Certainly overkill for old chips, but for every objective there is a CMOS chip which the objective cannot fully resolve

This bicycle speed meter chip was done with a /0.75 and it could use some more resolution. OTOH, each 2x increase in NA and resolution is a 4x decrease in DoF

The DoF numbers of the MPLAN are take from the Olympus catalog so they have to be right. For the 4x 0,13 I found no numbers from Olympus directly. I took the numbers from Edmund Optics:
https://www.edmundoptics.de/p/olympus-uplfln-4x-objective/29226/
The numbers of the two 0,8 objectives are both taken from the Olympus UIS2 catalog so the difference should be correct:
https://exhibitors.productronica.com/download/1121_1_4_2240_4_1_223/productronica_m1703e_microscope_components_guide_enpdf.pdf
And for me looking back to "normal camera objectives" it makes sense. A longer lens has less DoF. But I´m no expert regarding optics...

This website maintained by Nikon explains roughly what goes into DoF. There is a component which is supposed to represent how much defocus blur you can tolerate because it's still smaller than diffraction, and there is a second component which represents how much further blur you can tolerate due to limited output image resolution. The former only depends on NA, the latter also on magnification. With an infinite resolution sensor, magnification theoretically shouldn't matter.

The 4x spec of only 16μm may be wrong, or maybe it's calculated with some other formula based on other criteria of what it means to be "not significantly worse than diffraction alone".

[Noopy]

The Nikon was just an experiment. There is a reason why you don´t see pictures of it. 
In my view using a coverglass is not an option in our business. Either the picture quality is good enough without a coverglass or you have to find a different objective.

Well, either that, or you simply shoot through a cover glass (example)
For a quick job like here there is no need for glue, just a drop of water.
But the chip must be flat: no package, no bond wires, no other junk.

OK, let´s say it is an option but not a very comfortable one. Alone the handling of the coverslips. Then you need the right thickness and it has to be tight on the die. Which is also a problem with a drop of water it seems...
No, not my solution. 

I got the Zeiss very cheap so it was ok to start with immersion microscopy. But it´s not ideal of course.
A good immersion objective would be nice. But you always have to ask yourself if you really need it. I´m not sure if it makes sense for me to invest in such an objective. A NA of 0,8 is already a lot.

Certainly overkill for old chips, but for every objective there is a CMOS chip which the objective cannot fully resolve

I agree with you. I have a objective in mind that would be a really cool addition but I don´t have the money (at the moment)...  :'(

This website maintained by Nikon explains roughly what goes into DoF. There is a component which is supposed to represent how much defocus blur you can tolerate because it's still smaller than diffraction, and there is a second component which represents how much further blur you can tolerate due to limited output image resolution. The former only depends on NA, the latter also on magnification. With an infinite resolution sensor, magnification theoretically shouldn't matter.

The 4x spec of only 16μm may be wrong, or maybe it's calculated with some other formula based on other criteria of what it means to be "not significantly worse than diffraction alone".

I have already seen this Nikon calculator but didn´t take a closer look.
I have a very high resolution with my Canon 90D, the resolution is actually a lot too high. But it´s not a infinite resolution sensor so the magnification should still matter.
Nevertheless it shows that the DoF is not really a hard number. 

[Noopy]

Dirt on microscope objectives is often difficult to detect. The front lens of the UMPlanFl 50x 0.80 BD is so small that it is impossible to assess its cleanliness with the naked eye. The close-up with a macro objective gives a hint, but also is not ideal.




In this case, the dirt was noticed when adjusting the Köhler illumination. After removing an eyepiece, artifacts could be seen in the image that appeared there.




For a clear assessment of cleanliness, a certain magnification is required, as here by a factor of 4. In the area on the left, a formerly liquid substance appears to have solidified. Many particles are distributed across the lenses.




Surprisingly, the image quality of this lens is still very good.




The Zeiss brochure “Das saubere Mikroskop” provides valuable information on how to clean a microscope objective correctly. In principle, dirt can accumulate at many points in the optical path. However, the front lens of the objective is the most critical. According to Zeiss, this is also the only area of reflected light microscopes that the user should clean himself. Loose dust can be removed with a blower. A cleaning agent is required for firmly adhering dirt.

Choosing the right cleaning agent is extremely important to avoid damaging the lens. Lenses can be made of glass, quartz or polymers and often have a coating that reduces stray light. The lenses are glued into the lens. The surrounding area is usually painted black. All these elements can be damaged by overly aggressive cleaning agents. The first step should therefore be to add a little washing-up liquid to distilled water and try to loosen any stubborn dirt. If this is not sufficient, you will need to use a solvent.




Zeiss recommends high-purity gasoline with a boiling point below 44°C as a solvent. The gasonline shown here boils between 40°C and 60°C. Special wiping sticks can be used to apply the gasoline and wipe off the dirt. These are plastic sticks with a polyester cover. However, if high quality is important, these wiping sticks are very expensive. A cheaper option, which Zeiss also recommends, is the use of so-called eye cotton. This is a very clean, pure cotton. Cotton buds are not usually suitable. You can wrap the cotton on a wooden stick. Bamboo would be well suited here. The cotton should not be touched with the fingers as it can absorb the oil from the skin. The cotton is then moistened with the gasoline. It should not be too moist so that no gasoline flows into the lens. Do not apply too much pressure when cleaning. Larger areas should be wiped in a spiral from the inside to the outside.




The front element of the objective is now generally clean.






Although the dirty lens already produced a good image (top), the image quality could still be noticeably improved by cleaning (bottom).




The illumination is set as close as possible to the same for both images. The focus stacking and the adjustment of the tonal value correction were also carried out in the same way for both images. Nevertheless, artifacts sometimes occur during processing (yellow arrow).

The difference in quality between the images is most noticeable at the edges of the metal layer. These areas are shown in much greater detail with the cleaned lens. In some places, the structure of the edge is now visible for the first time (green arrow). In some cases, the surface structures of certain areas are now also more clearly visible. The image shows better contrast overall.


https://www.richis-lab.de/Howto_Microscope_Objectives_Cleaning.htm

 

[Noopy]

Let´s once again talk about oil immersion microscopy. As already described it is not possible to achieve a numerical aperture higher than 0,95 with conventional light microscopy. For green light, this corresponds to a resolving power of 0,35 µm. If the oil immersion technique is used, a numerical aperture of 1,5 can be achieved. This allows a resolving power of 0,22µm. In oil immersion microscopy, a special oil is introduced between the objective and the object. The objective must of course be designed for this. The higher refractive index of the oil makes it possible to construct objectives with a larger numerical aperture.




We already had the Zeiss Primo Plan-Achromat 100x 1,25. It is a typical biological objective designed for oil immersion. Next to the NA value is the note “Oil”. This objective has some disadvantages. Although Zeiss objectives are infinity corrected, they usually still have minor optical weaknesses that are corrected within a Zeiss microscope. This results in slightly poorer image quality in an Olympus microscope. In addition, the field of view is smaller. This means that the camera sensor is no longer illuminated in the best possible way. The working distance is 0,14mm. If you do without a cover glass, you can add its thickness of 0,17mm.




Such a lens is used as follows: You place a drop of immersion oil on the object and then move the lens into this drop. The surface tension ensures that the oil remains between the lens and the object, even if you move to other parts of the object. Care must be taken to ensure that dry lenses do not come into contact with the oil, as it can penetrate the interior and impair the image quality.




Even without the cover glass, the result is a very interesting image. The image quality deteriorates significantly at the edges of the image. This is not surprising due to the small field of view.




In direct comparison with the LMPlanFl 100x 0,80 BD, the Zeiss 100x 1,25 does not deliver much more detail. It even appears blurrier in places. However, it is noticeable that the vias in the metal layer are clearly different. They are reproduced larger and so structures can be seen in them that were not visible before. With irregular surfaces, immersion microscopy often provides better images, as the oil fills gaps and its refractive index is very similar to that of glass.




With the MPlanApo 100x 1,40, Olympus offers an oil immersion objective that was developed for reflected light microscopy. Consequently, no cover glass is required for an optimal image and the optics are designed to reflect as little of the light introduced from above as possible. With 0,24 µm, the resolution of this objective is very close to the resolution limit of light microscopy. The working distance is just 0,1mm.




Special oils are used for oil immersion microscopy. A cheaper oil is shown on the left. For the MPlanApo 100x 1,40, the IMMOIL-F30CC sold by Olympus is recommended. This oil is considerably more expensive, has a shelf life of only three years and should be stored in the refrigerator. Before use, it must first be brought to room temperature. With 1,518, the refractive index of IMMOIL-F30CC is slightly higher than that of the cheaper oil.




Especially if you want to examine a small die with oil immersion, you have to fix it in some way. An ordinary, water-soluble glue stick is very suitable for this. A cover glass can be used as a carrier. Isopropanol and acetone do usually not affect such glue from a glue stick, so that the die can still be cleaned without any problems. However, the bond can be removed quickly and easily with water.




The cover glass can then be fixed to the positioning table with adhesive tape.




The images above show a section of the LIN transceiver ATA6663 (https://www.richis-lab.de/transceiver02.htm). As the depth of field is very shallow, both images consist of 70 individual images each. It is difficult to find the optimum settings for focus stacking. This is why artifacts can be found in some places. The image on the left was taken with the cheaper immersion oil, the one on the right with the IMMOIL-F30CC. The overview already clearly shows that the image on the right offers more detail.

Both images are available in full resolution:
https://www.richis-lab.de/images/howto/15x05aXL.jpg (28MB)
https://www.richis-lab.de/images/howto/15x05bXL.jpg (27MB)






The sections shown here represent an area of 83µm x 57µm. The vias, which are visible in the metal layer, have a diameter of approximately 0,7µm. The cheaper immersion oil (above) produces a much blurrier image. With the IMMOIL-F30CC (bottom), on the other hand, even the structures on the metal layers stand out clearly.

 
The description of the IMMOIL-F30CC states that the lens should be cleaned after use with a cloth soaked in alcohol.


https://www.richis-lab.de/Howto_Microscope_OilObjectives.htm

 

[iMo]

I wonder how this guy has stripped off the top metal layer of the rp2350 so nicely (around 12:30 to 16:00):