Decapping and Chip-Documentation - Howto

[Noopy]

Hi all,


today I can show you a Howto regarding an easy way of decapping and documentating Chips:

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





I´ve structured the topic in different articles:

1. Decapping
https://www.richis-lab.de/Howto_Decap.htm

2. Optics
https://www.richis-lab.de/Howto_Optik.htm

3. Positioning and ligths
https://www.richis-lab.de/Howto_Licht.htm

4. Picture tuning
https://www.richis-lab.de/Howto_Tuning.htm


As usual text on my homepage is german but I will answer every question you have regarding the topic.


Have fun! 



Best regards,


Richard

[magic]

1. Decapping
https://www.richis-lab.de/Howto_Decap.htm

It seems that you got fire to work. I had problems with it and ruined many dies by overheating so I switched to a Chinese 936 iron



Normal soldering temperatures aren't sufficient, I drive the heater manually from 24V DC. Temperature is monitored by connecting the thermocouple sense pins to a TM-902C type thermometer.
About 600~650°C at the heater is sufficient. Aluminum melts at 660°C and the die is cooler than the heater so it should be safe; I have opened a dozen chips that way and never overheated a single one.
Plenty of magic smoke is emitted but there are no flames. Do it outside or under ventilation.

Alternatively, 65% nitric acid at boiling temperature (130°C or so IIRC) does a great job in a few minutes with zero risk to the die except for eating aluminum bonding pads. Disadvantages are fumes (do it outside) and lack of general availability in this EU shithole (find somebody who has access to it professionally).

I'm planning to try 95% or 98% sulfuric next time because I've seen reports that it doesn't attack aluminum. OTOH, sulfuric is said to require higher temperature, something like 200°C or more.

It is beneficial to "preprocess" the chip by cutting excess epoxy with tin snips, particularly for chemical methods.

2. Optics
https://www.richis-lab.de/Howto_Optik.htm

3. Positioning and ligths
https://www.richis-lab.de/Howto_Licht.htm

I haven't done any imaging yet myself.

I found this interesting blog, they show some decent pictures and talk a bit about their methodology. They used microscope lenses, though.
https://resnicklab.wordpress.com/2012/09/26/microscopes-and-imaging/

The general conclusion appears to be that best results are obtained with long focal length and long distance. Of course diffraction and aberrations still set a limit on practical magnification.
They also say that silicon is partly transparent and that this is the reason why you get nice colorful images with back illumination. I'm not sure if it's true.

It is 100% definitely not true and it was somebody else who said this nonsense. Light penetrates silicon to the depth of no more than a few microns. Nice colorful images are product of iridescence / thin film interference occurring in a layer of silicon dioxide covering the chip with varying thickness, partly left over from the fabrication process and partly added deliberately to protect the insides ("passivation").

[Noopy]

I didn´t have much problems with die damage due to the heating. Sometimes there was some kind of blistering but not bad.
I try to heat the package very quickly with a hot blue flame and only as long as necessary.
With my proceeding real damage comes only when you use the hot flame directly on last package parts on the die.

Interesting you use only "electric heat" to open the packages. Perhaps some time I´ll try that too.

Surely acids are the best way to get nice clean dies but I didn´t want to work with such dangerous chemicals.


Regarding the imaging I was proud having found a cheap method of taking photos with good magnification factors (as long as you already have a DSLR).
Surely there are special microscopes with special optics but you have to spend a lot of money for such a tool.

On the page you linked I couldn´t find a clue which resolution they really acchieve and how the final setup looks like. 
I think I have to study the page in more detail... 

[Noopy]

Hi all!


News on my page (yes, german...  ):
https://richis-lab.de/decap-ofen.htm


I have built an ofen for the decapping:




Works quite fine and reproducible! 




Something around 400°C for 2-5 minutes and you are inside the package. 


 

[magic]

That's seriously cool and much better than anything I have ever done

BTW, I actually like that smell

[Noopy]

That's seriously cool and much better than anything I have ever done

BTW, I actually like that smell

Thanks! 

Well, the smell reminds somehow of the youth but with time (and chips) it becomes too much... 
Thank god I have a porch! 

[GeorgeOfTheJungle]

  =>

[magic]

98% sulfuric, oleum or fuming nitric?

I tried  less concentrated acid once and bonding pads went away with the package

[GeorgeOfTheJungle]

That's seriously cool and much better than anything I have ever done

BTW, I actually like that smell

Thanks! 

Well, the smell reminds somehow of the youth but with time (and chips) it becomes too much... 
Thank god I have a porch! 

İt's the smell of magic smoke, isn't it?

[Noopy]

That's seriously cool and much better than anything I have ever done

BTW, I actually like that smell

Thanks! 

Well, the smell reminds somehow of the youth but with time (and chips) it becomes too much... 
Thank god I have a porch! 

İt's the smell of magic smoke, isn't it?

Quite right!
Thought about filling the smoke in tubes to refill the next failed IC. 

[magic]

I used saturated sodium nitrate dissolved in 98% hot (~100C) sulfuric acid. Don't use near boiling (337C) acid. It's gonna end up very bad if something happens.
Boiling H2SO4 burns basically anything organic instantly, while 100C ones give you a sec or two to wipe things off.

Did you try without the nitrate? I have heard of somebody using pure sulfuric acid to decap commercial parts for use in hybrid circuit prototypes, but I don't know if 98% is pure enough. And I think his temperature was higher.

There is certainly no point heating beyond 100~150°C if you add nitrate because all the nitric acid created in the reaction will evaporate or decompose. And yes, that's basically the nitrating mixture, except much hotter than usual, so it will burn/explode everything

[GeorgeOfTheJungle]

Silicon melts at 1414 ° C, gold at 1064, aluminium at 660, (what else can melt inside the chip?) if you can get the magic smoke at temps below that, why use nasty chemicals?

[GeorgeOfTheJungle]

But, do you expect the chip to work after decapping? What for? I mean, once decapped, the chip exposed to the air won't work for much longer, right?

[GeorgeOfTheJungle]

I've read it in the book "Microchip Fabrication" (Peter Van Zant) 

[magic]

I did an experiment once and heated a DIP8 opamp on the tip of soldering iron set to 450°C for ten minutes. It didn't work afterwards and the epoxy got a bit smelly but still too hard to break. So it seems you can't get working parts that way.

Plastics are permeable to air and moisture. If it's sold in plastic package it has to be chemically resistant. Chips are "passivated" with a layer of glass on top.

[MagicSmoker]

I used saturated sodium nitrate dissolved in 98% hot (~100C) sulfuric acid. Don't use near boiling (337C) acid. It's gonna end up very bad if something happens.
Boiling H2SO4 burns basically anything organic instantly, while 100C ones give you a sec or two to wipe things off.

Sounds like heaps of fun... and if I remember my inorganic chemistry correctly, this makes anhydrous nitric acid in situ, which is such a strong oxidizer it rapidly passivates most metals rather than dissolve them (I don't think aluminum bond wires will survive, however).

[magic]

Good enough that bare aluminum bonding pads survive

Pure nitric is the standard chemical in failure analysis labs. But it's expensive and rarely available to individuals.

[GeorgeOfTheJungle]

I've read it in the book "Microchip Fabrication" (Peter Van Zant) 

I've read the same in old soviet books. It doesn't hold anymore today.

Many modern chips are sold in solderable die form, called wlcsp or dsbga.

But are then put in a package of some sort before use, right? Even the cheapest COBs are.

[magic]

No. These are pretty much bare dies with solder blobs on them that you just put on your board upside down and reflow. Sometimes with interesting side effects.

https://www.raspberrypi.org/blog/xenon-death-flash-a-free-physics-lesson/

COBs are potted to protect their bonding wires from being torn. And maybe to stop light as well

[GeorgeOfTheJungle]

But it's been designed to be used like that, as the die of a CCD or a COG that can also be seen, which is not, I think, the same thing as if by removing the epoxy package it's been designed to be in and protected by, you had left it naked and exposed. Humidity and IIRC sodium are chip killers, and both can be in the air.

[iMo]

All chips are covered by few hundreds nm of glass passivation. The only openings are at the pads for the bond wires.
I wanted to see the die of the LT1021 Vref - I put the DIL8 epoxy into the direct flame of my kitchen stovetop, kept there for a couple of minutes (the package was yellow hot) and then I threw it into cold water. The epoxy package desintegrated while squeezed with fingers. No acids required.

[GeorgeOfTheJungle]

All chips are covered by few hundreds nm of glass passivation. The only openings are at the pads for the bond wires.
I wanted to see the die of the LT1021 Vref - I put the DIL8 epoxy into the direct flame of my kitchen stovetop, kept there for a couple of minutes (the package was yellow hot) and then I threw it into cold water. The epoxy package desintegrated while squeezed with fingers. No acids required.

Do me a favor: plug it in and let me know for how long it works

[GeorgeOfTheJungle]

Wow, this is a beautiful photo!


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

[Noopy]

Thanks! 

One of my favourites:



Soon I´ll have a MEMS-Sensor... Really cool pictures… 

New pictures will be bigger. 700px are outdated.
One day I´ll update the old ones… 

[iMo]

The chip itself is produced at pretty high temperatures. The wafers are yellow/white hot during various diffusion processes (around 1000 degC). Thus to put a chip die into a kitchen stovetop flame is for the die the same fun as when you sit yourself into a steam sauna. The aluminum or copper metalisation layers will not desintegrate as they are hold together by the glass passivation layer. Of course, nobody would guarantee you the chip may work after such an exercise.

[Noopy]

The chip itself is produced at pretty high temperatures. The wafers are yellow hot during various diffusion processes (around 1000 degC). Thus to put a chip die into a kitchen stovetop flame is for the die the same fun as when you sit yourself into a steam sauna. The aluminum or copper metalisation layers will not desintegrate as they are hold together by the glass passivation layer. Of course, nobody would guarantee you the chip may work after such an exercise.

You are right but a torch can ruin the metal layer:

[iMo]

It could be the metallization layers get ruined by the gases streaming off the epoxy package penetrating the glass passivation layer, when put into torch (the epoxy package contains up to 8% of water and 17% of epoxy resin).
Nice pictures!

[GeorgeOfTheJungle]

The chip itself is produced at pretty high temperatures. The wafers are yellow/white hot during various diffusion processes (around 1000 degC). Thus to put a chip die into a kitchen stovetop flame is for the die the same fun as when you sit yourself into a steam sauna. The aluminum or copper metalisation layers will not desintegrate as they are hold together by the glass passivation layer. Of course, nobody would guarantee you the chip may work after such an exercise.

I think the idea is not to go further than 600˚C or so maximum (aluminium melts at 660). 400˚C in noopy's oven seems to work fine: https://www.richis-lab.de/decap-ofen.htm

[GeorgeOfTheJungle]

Thanks! 

One of my favourites:

Yep, that's even better! 

What's a MEMS-Sensor, a solid state microphone?

[Noopy]

It could be the metallization layers get ruined by the gases streaming off the epoxy package penetrating the glass passivation layer, when put into torch (the epoxy package contains up to 8% of water and 17% of epoxy resin).
Nice pictures!

Sounds possible...

Thanks! 

I think the idea is not to go further than 600˚C or so maximum (aluminium melts at 660). 400˚C in noopy's oven seems to work fine: https://www.richis-lab.de/decap-ofen.htm

That´s the idea. Don´t risk more than neccesary. 

What's a MEMS-Sensor, a solid state microphone?

"Small, Low Power, 2-Axis ±3 g i MEMS® Accelerometer"
https://www.analog.com/media/en/technical-documentation/data-sheets/ADXL323.pdf
And it looks really nice! 

[iMo]

I think the idea is not to go further than 600˚C or so maximum (aluminium melts at 660). 400˚C in noopy's oven seems to work fine: https://www.richis-lab.de/decap-ofen.htm

Yea, I got the package in the same shape as on the last picture in the link
The biggest issue was to find the chip in the dirty mess
Sure, the gas oven or torch is not easy to regulate.

[magic]

But it's been designed to be used like that, as the die of a CCD or a COG that can also be seen, which is not, I think, the same thing as if by removing the epoxy package it's been designed to be in and protected by, you had left it naked and exposed. Humidity and IIRC sodium are chip killers, and both can be in the air.

Epoxy package doesn't protect from humidity. Quite the contrary, it soaks up like a sponge. Sensitivity to humidity is a well known problem with plastic-encapsulated voltage references, for example.

I'm pretty sure that oxygen diffuses right through it as well for that matter, and sodium hydroxide (the thing that sodium ultimately turns into in contact with humidity) probably too. Though I'm not sure if Na/Na₂O/NaOH seriously exists in the air in any serious quantities?

NaOH is a chip killer. It can dissolve every part except epoxy and copper. You could actually drop a chip in NaOH solution for a few years and see if epoxy will protect it

It could be the metallization layers get ruined by the gases streaming off the epoxy package penetrating the glass passivation layer, when put into torch (the epoxy package contains up to 8% of water and 17% of epoxy resin).
Nice pictures!

I have a simpler theory: the melting point of aluminium is only 660°C
I suppose it just melts, gathers in blobs and then solidifies again, displaced.

At any rate, as I said, 10 minutes at 450°C and the opamp I tested was totally dead and unreactive to anything.

[iMo]

Though I'm not sure if Na/Na₂O/NaOH seriously exists in the air in any serious quantities?

I was reading (or it was youtube?) some interesting stuff on history of chip making in Japan. They were coping with Na in air as they were located near the sea shore.

[GeorgeOfTheJungle]

Yep, it's salt, chlorine and sodium ions, both things are bad for the chips, near the sea there's lots of that in the moist air.

[GeorgeOfTheJungle]

Epoxy package doesn't protect from humidity. Quite the contrary, it soaks up like a sponge.

"Like a sponge"? Hyperbole for the win

[GeorgeOfTheJungle]

What's a MEMS-Sensor, a solid state microphone?

"Small, Low Power, 2-Axis ±3 g i MEMS® Accelerometer"
https://www.analog.com/media/en/technical-documentation/data-sheets/ADXL323.pdf
And it looks really nice! 

[TheUnnamedNewbie]

The chip itself is produced at pretty high temperatures. The wafers are yellow/white hot during various diffusion processes (around 1000 degC). Thus to put a chip die into a kitchen stovetop flame is for the die the same fun as when you sit yourself into a steam sauna. The aluminum or copper metalisation layers will not desintegrate as they are hold together by the glass passivation layer. Of course, nobody would guarantee you the chip may work after such an exercise.

All those high temperatures are done in the FEOL. Once you do metalization in the BEOL I don't think you see those temperatures anymore.

I believe that more modern technologies don't use glass passivation anymore but some polymers? I know colleagues at the lab complain about it a lot since it is a lot less robust and easy to accidentally probe through.

[magic]

Why would they do that? Isn't SiO₂ still the basic insulator used all over the die anyway? Something to do with high-k CMOS?

[Noopy]

Thanks for the link, looks very interesting! 

Why would they do that? Isn't SiO₂ still the basic insulator used all over the die anyway? Something to do with high-k CMOS?

Something with CTE mismatch. Having a low modulus layer like polyimide helps mitigate the stress. Even some power devices do this. Cree SiC MOSFETs all have this.

I know only one use of polyimide: It is used considerable time for memory as a additional layer. The reason is that in the mold compound there are always some radioactive particels which can change the logic level of small memory cells.
Interesting that it´s also used for SiC MOSFETs.

[GeorgeOfTheJungle]

Epoxy package doesn't protect from humidity. Quite the contrary, it soaks up like a sponge.

"Like a sponge"? Hyperbole for the win

"Microchip Fabrication, 3rd edition, p559"

Hermetic sealing results in a package that is impervious to the penetration of moisture and other gases. Hermetic seals are required for chips operating in harsh and demanding environments such as rockets or space satellites [..]
Nonhermetically sealed packages are adequate for most consumer applications [..] This sealing system provides good and adequate environmental protection of the chip, except in the most demanding situations. A better term for this type of enclosure sealing method would be "less hermetic". These packages are composed of epoxy resins or polymide materials and are generally referred to as "plastic packages".

[magic]

https://www.analog.com/en/analog-dialogue/articles/does-my-voltage-reference-design-hold-water.html

Does My Voltage Reference Design Hold Water? Methods of Managing Humidity and Performance in Precision Analog Systems

Seriously though, the question was if the plastic is critical for protecting the IC from "ambient" chemicals. To my limited knowledge, it's not.
Go ask blueskull how long his decapped transistors survived outside the package. I think his bewilderment suggests that it wasn't really a problem

[GeorgeOfTheJungle]

https://www.analog.com/en/analog-dialogue/articles/does-my-voltage-reference-design-hold-water.html

Does My Voltage Reference Design Hold Water? Methods of Managing Humidity and Performance in Precision Analog Systems

Meh, clickbait is everywhere these days 

Seriously though, the question was if the plastic is critical for protecting the IC from "ambient" chemicals. To my limited knowledge, it's not.

I've found this:

https://www.ti.com/lit/an/snoa300a/snoa300a.pdf

Current plastic packages use various molding compound formulations consisting of epoxy resin, silica
fillers, and other minor constituents. The epoxy resin is non-hermetic and absorbs a small percentage of
moisture through diffusion. Eventually, an equilibrium develops between the moisture content inside the
package and the ambient moisture and temperature conditions. If enough moisture is present in the
package during surface mounting, the intense reflow heat turns the moisture into saturated steam (1). This
extreme pressure, accompanied by a drop in flexural strength of the molding compound at temperatures
reaching up to 240°C, may cause the package to fracture allowing the steam to escape (2). Once the
fracture occurs and the steam has escaped, a greater threat now exists to the device. The fracture allows
moisture and ionic contaminants such as sodium, potassium, or chlorine to infiltrate the plastic package,
potentially causing corrosion, and eventually failure of the device. Therefore, moisture-induced cracking
must be prevented in order to maintain the long-term reliability of surface mount plastic packages.

Maybe Dave would like to investigate, sounds to me like a good theme for a cool for-nerds-only EEVBlog vidjeo: take some AVRs or something with the blinky sketch, decap them (cool die photos!) with noopy's oven system (Magic smoke!), let them run on a shelf 24/7 (boring), and see what happens: updates every Mailbag Monday.

[Noopy]

Maybe Dave would like to investigate, sounds to me like a good theme for a cool for-nerds-only EEVBlog vidjeo: take some AVRs or something with the blinky sketch, decap them (cool die photos!) with noopy's oven system (Magic smoke!), let them run on a shelf 24/7 (boring), and see what happens: updates every Mailbag Monday.

It would be a honor for me being mentioned in a Dave-Video! 
But I would have to bond new wires to a decapped die. That´s not possible for me... 
Opening a metal-can-package and let it run with high humidity would be possibility...

[magic]

Some people say that boiling chips in rosin gets the job done. Tried it today, here's the outcome:



There is some loss of volume, so it seems to work, but it certainly isn't fast. I boiled it for a good few minutes with no obvious effect so I cranked up temperature until it started to smoke. That was probably not a good idea

Or maybe it was? I don't even know if the attack occurred during those few minutes of slow boiling or due to the high heat at the end. Not sure if I want to do it again to find out.

I think I would rather simply drop the chip into an empty test tube and cook until it gives out the magic smoke.

[Noopy]

I have also heard from people trying rosin to decap chips.
In my view rosin doesn´t  really do the job. When people go to around 400°C they can destroy the epoxy but with this temperature you don´t Need rosin. 

[magic]

I actually wouldn't mind cooking it in some solvent which would cleanly strip all the decomposed epoxy in real time so that I don't need to do it manually. But it turns out that burning rosin is anything but clean

[iMo]

You would need a pretty strong acid for your cooking

[Noopy]

Heating a "wet" package sounds like a easy way to crack it. But my read is that there will only be one or more cracks. I´m afraid the epoxy will still be very hard to remove...

[cncjerry]

Imo,  loved your video.  That was something new for me.  Wondering what to do with them, maybe earrings if you have enough of them. ,pretty darn cool.

As far as  probing,  if you use your cnc mill with a probe in the chuck, it would make it much easier..

Thanks  for posting,

Jerry

[magic]

You would need a pretty strong acid for your cooking

I wouldn't; rosin is good enough, as I said. But it's slow as molasses and easily turns into a dirty mess.

Maybe there are other organic acids that could work. So far I found two possible candidates: benzoic and adipic. Both have high boiling points and are safe at room temperature. Their structure is also simpler than rosin acids, so hopefully pyrolysis products would be cleaner.

[magic]

Another silly experiment: synthetic candle wax

1 hour at 280°C: no effect whatsoever
5 minutes at 350°C: no effect whatsoever

Conclusions:
Chemically, synthetic wax does nothing to epoxy.
280°C isn't sufficient to cause thermal decomposition.
350°C isn't sufficient to cause quick thermal decomposition.

Now I wonder how hot I got that rosin the last time when it partly etched the surface of the package. I didn't monitor temperature

On the upside, wax doesn't stain the glassware

[Noopy]

Another silly experiment: synthetic candle wax
...

Well an attempt doesn't hurt! 

[magic]

A small update about my standard thermal decomposition process.

Shortly after my first post in this thread I bought a dedicated soldering iron heater for this duty. I chose the cheap and ubiquitous A1323 heater for Chinese 936 clones, which is similar to my iron's A1321 but with steel body and higher power.



Manually PWM-ing a 24V power supply gets boring pretty fast and I wanted better temperature repeatability too. With a bit of experimentation I found that constant 12V supply produces just about perfect temperature, which smokes the chips nicely but doesn't cause the metal layer to melt. The temperature reported by the thermocouple inside the heater is some 600~700°C, depending on ambient temperature, airflow etc.

So the complete setup is: a 12V / 1A power brick driving a cheapass auction site soldering iron heater

For DIP packages I use a two step process. First, the DIP is placed on the heater and tied with its leads to thoroughly bake the bottom. Then the bottom is crushed and the leads yanked out, leaving me with the upper half which contains the die. This is trimmed a bit and baked again, such that the die lies directly on the heater and all epoxy in contact with it gets roasted well. A few minutes suffices for each cycle. A bowl with water is placed below to catch anything falling from the heater and cool it down to safe temperature. A bit of thermal shock comes for free. Sometimes it causes spontaneous separation of the die after step 2.

[Noopy]

Looks good to me! 

[magic]

Another silly experiment: synthetic candle wax

1 hour at 280°C: no effect whatsoever
5 minutes at 350°C: no effect whatsoever

Funnily enough, this transistor still works. It has about 300 beta and over 100V Vceo (that's maximum I can test). Not bad for what the markings suggest should be a BC857.

I also performed additional testing with rosin. I boiled this no-name 78L05 for over two hours at about 270°C. To avoid messing up test tubes I used a thimble tied to my soldering iron heater with wires.

Nothing happened to the chip, only the surface got slightly matte. This chip also appeared to work afterwards, but I only tested quiescent current and output voltage under no load. Then I cracked it manually to see if it got any softer - nope, it didn't.


Conclusion: rosin really requires crazy high temperature to work and absolute maximum ratings for storage temperature are only advisory

[Mortymore]

Christopher Tarnovsky's, Flylogic Engineering website used to have the best chip photos, but they are no more. 
They were taken by IOActive

EDIT: Some info on the man: https://www.wired.com/2008/05/tarnovsky/ (watch the video)

[GeorgeOfTheJungle]

Christopher Tarnovsky's, Flylogic Engineering website used to have the best chip photos, but they are no more. 
They were taken by IOActive

EDIT: Some info on the man: https://www.wired.com/2008/05/tarnovsky/ (watch the video)

WOW ! 

[Mortymore]

I was once interested in the subject, so I'll leave some literature that I read at the time, in case someone is interested also.

From Cambridge University, Technical report 630: Semi-invasive attacks – A new approach to hardware security analysis

https://www.cl.cam.ac.uk/techreports/UCAM-CL-TR-630.html

[magic]

Some fun with 98% sulphuric acid.

I started safe, boiling only one drop of acid at a time An idea based on the well known nitric acid process for live analysis. However, sulphuric acid appears less suitable for that. In addition to needing higher temperature, it tends to create a dry and brittle crust on the attacked surface. This crust is still strong enough that it can't simply be washed away, but rotten enough that on the next round the acid sinks into it and creeps all over the place instead of staying where it was put and doing the work. Everything turns messy and etching rate slows down so I cheated by scratching the crust away. This of course damaged the bonding wires, but at least I got a clean die after some hour of work

The above process is a PITA and kinda pointless given the destruction of bonding wires, so I did the next chip the Zeptobars way: just dropped it in acid and boiled until everything turned into black sticky goo. This was a TO92 package and it only took a few minutes of cooking and tittle acid. I used the same thimble and soldering iron heater setup as in my previous experiments and heater power was also the same, which should yield about 270°C. I didn't measure exact temperature for lack of suitable sensor. One may think that a metal vessel would be destroyed by acid, but in practice corrosion is very low. I will see how long it lasts. At least it isn't going to crack and shatter from heat like glass could do. I placed a steel bowl below to catch spills and overflows. Overflows are possible because the reaction produces plenty of foam and bubbles.

One of the gold bonding wires survived the whole journey and somehow remained attached, but on the photo it looks like bonding pads may have been etched away by acid. I will never know for sure because I crushed the die trying to remove it from the metal carrier Admittedly I was being sloppy and used water instead of acetone for cleaning, but I'm not sure if the pads made it even to that point.

Acid also proved effective against stubborn epoxy residue on dice extracted by heating. I managed to clean up a few of those.

[Noopy]

Interesting. Thanks for sharing. 

For me that´s too messy and dangerous. 

[magic]

On the imaging front, I built a simple beam splitter to experiment with illumination through the lens. It's a small cardboard box painted black to absorb reflections with a 45° angled half-mirror inside and a hole in one wall.

The half-mirror is a piece of a "mirror" protection foil for smartphone displays. These foils are supposed to make the screen reflective when the LCD is blank but still be transparent enough for normal use. It probably isn't exactly 50% reflective and 50% transmissive, but whatever, good enough for us.



The box goes on top of the lens. Light is supplied through the hole in the side, bounces off the mirror downwards, passes through the lens back and forth and passes through the mirror to be observed from above. Half of the entering light passes through the mirror and is dumped on the wall and half of the light from the lens is reflected back towards the illumination source, but 25% efficiency is not the end of the world.

I tried it in combination with my favorite HD webcam lens, a lousy VGA webcam lens and a 10x biological microscope objective. A few sample images follow.

The webcam lenses have enough magnification that one can simply use them as a loupe and look directly into the lens. Magnification is good enough for low density analog ICs and with the HD lens and with some squinting I can even follow the super-die-shrunk Chinese LM358 chips. The first two images show how the two lenses perform on a TI NE5532 die. My camera couldn't capture the full field of view visible with naked eye and the image turned out a bit dark, but it gives a glimpse of the resolution available with the HD lens.

The third image is the same but with the 10x microscope objective. Magnification is very low. This lens could only possibly be usable for photography, i.e. projection onto a camera sensor at a sufficiently far distance, or as part of a compound microscope with an additional eyepiece lens. Go figure

[Noopy]

And again: Very interesting! 

[magic]

New version. This time with a larger box that holds a webcam sensor board to enable viewing on a computer and taking pictures.



An improvement was necessary in the way that the part of illumination stream which fails to reflect from the mirror is "wasted". The light is still projected on the rear wall, but the larger box allows this wall to be farther away and hidden from the sensor's view. The camera PCB has its original lens tube attached and the tube blocks the light scattered by the "waste wall" from reaching the sensor, except for the part which reflects back exactly towards the mirror.

Another improvement is the addition of Köhler illumination. I hoped it wouldn't be necessary but it does improve contrast considerably by reducing stray light. The lens on the illumination entrance is so-called collector lens, because it collects light from the illuminator, refracts it and projects a focused image of the illuminator straight into the aperture of the main lens. This way all of the light which reaches the entrance hole ends up where it belongs instead of bouncing around inside the box. The collector lens should be at the same distance from the mirror as the sensor, slightly larger than the sensor and its focal length must be shorter than the distance to the main lens (or it wouldn't focus properly). At least that's the theory, in practice the distance is only approximate and I adjust LED lamp position empirically for best contrast.

This setup takes some decent quality images already and is relatively reliable, main problems at this point being precision alignment of everything for uniformity of focus and illumination across the field.

By reducing the size of the light source (a 3mm LED) I got this image of the Chinese RC4558 which appears even a bit more colorful than normally. Not sure what happens here.

[GeorgeOfTheJungle]

Nice!

[Noopy]

That's a nice construction. I like getting good results out of low cost designs!
I think some 3D printed casing would improve the quality a lot.

[magic]

I think some 3D printed casing would improve the quality a lot.

For real photography time is probably better spent figuring out how to put this (or a better?) beamsplitter between a real microscope lens and a real camera.

Better construction would improve flatness of focus (I think the bottom right corner of my 5532 image is a bit soft, for example), hopefully uniformity of brightness, and maybe contrast a bit. But you are still left with a high crop, low resolution and noisy sensor, questionable demosaicing (regular dot artifacts appear in a few places on my shots) and auto white balance and sharpening that cannot be controlled. My webcam also lacks manual exposure mode so it can't even be used for panorama stitching.

I made this webcam mod to test the beamsplitter concept and for direct inspection and reverse engineering. If some corner is a bit out of focus, I will simply move the camera or tap my finger on the PCB to realign it. BTW, a big nuisance is the USB cable which pulls on the board. I tried stripping the last 10cm of external jacket and shield or using a piece of ribbon cable for the last 10cm but this breaks high speed USB.

The lens is close to its limits too. I estimate my resolution at about 0.5µm/px and an f/2 lens resolves maybe a line pair per 1.5µm so I'm already sampling at 150% Nyquist density. Maybe some minimal improvement could be gained by doubling the pixel density, but beyond that there is no hope without a better lens.

[Noopy]

I have updated the decapping part in the Howto section of my website:

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

I won´t translate everything. To tell the truth there is not much new information. But more coming soon... 

[Noopy]

I had done some tests with DMSO to decap epoxy packages and now here we have the pictures:

 


You get DMSO quite easy.
Don´t store DMSO too cold because it gets solid at 18°C.
Above 189°C DMSO tends to decompose explosively. This temperature can be lower if it is mixed with other chemicals.




A PTC heater guarantees that the temperature doesn´t go above 150°C.




The DSMO temperature seems to be a little above 110°C. A little bit more would be nice but too much is dangerous... 




I cooked two packages for two hours but didn´t see any change. The surface didn´t get brittle or soft or whatever. They looked like new parts. 
I won´t do more tests with DMSO (right now). Cooking packages for some hours in a solvent that tends to decompose is no good if you want to decap a lot of packages.


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

 

[magic]

I have always been skeptical, IMO professional labs wouldn't bother with acids if it really was so simple.

A possible alterative to those PTC heaters is a small 500W electric cooking stove and a triac dimmer. That way temperature is freely adjustable, but not actively regulated of course.

Safety of "interesting" chemicals can be improved by boiling a smaller quantity, just enough for a SO8 or similar chip.

edit
BTW, using the electric cooker I tested (again ) rosin and paraffin wax at temperatures which, IIRC, exceeded 400°C for many minutes. Really no effect besides lots of smoke. I still don't know how I managed to slightly erode that SOT23 package in boiling rosin last year, but it seems like waste of time to further experiment with it.

[Noopy]

I agree with you.
Probably DMSO would do the job with smaller packages, some more heat and a lot more time but that´s no good for me.
I would have done more tests if there had been some progress but I really didn´t see any change of the material.
By contrast with my decap furnance I can decap 20 parts in one hour. There are some losses/residues but that´s ok, it´s still a hobby.

[Noopy]

I won´t play with HF, but HCL is ok. So let´s see what we can do with HCL.






We had this BUX42 with an interesting metal layer:
https://www.eevblog.com/forum/projects/transistors-die-pictures/msg3545848/#msg3545848
https://www.richis-lab.de/Bipolar50.htm
The BUX42 uses the "wide-emitter narrow-contact" technique.




HCL can´t dissolve the SiO2 protection on the die but it can dissolve the Al metal layer. I used 18% HCL.






After two hours most of the metal is still ok since it is protected by the SiO2.
In the bond area the Al partly got dissolved. The emitter metal is thicker than the base metal. 




After three days and nine hours the HCL found its way under the SiO2.
On the emitter side the HCL is faster. That´s probably because of the bigger channel the metal layer leaves after the Al got dissolved.




After twelve days and seven hours most of the Al is dissolved.




In the dark grey area the metal layer had contact to the silicon. In the greenish area the metal layer was isolated from the silicon.






Near the bond area the SiO2 surface layer is broken down at some points.






Looking at the die on a shallower angle you can see the height of the gap the metal layer left.
You can also spot the lower SiO2 layer that partly covers the emitter area.


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

 

[Noopy]

I won´t play with HF, but HCL is ok. So let´s see what we can do with HCL.
...
https://www.richis-lab.de/Howto_Decap_HCL.htm

HCL works well to strip the metal layer. It´s quite a gentle process... ...but be careful if you have a aluminium package! With a lot of aluminium you get a lot of hydrogen and heat!

[magic]

I remember experimenting with attacking aluminium using kitchen chemistry years ago. I tried at least vinegar and citric acid, possibly with an addition of salt, and some of the combinations were effective. A piece of aluminium foil was chewed into small fragments in a few days.

Now, ceramic DIPs. A piece of clay holds everything together while the top is heated with gas flame and lifted up when the glass seal melts. I smeared some glass onto the bond wires and got a piece of clay to stick to it, but otherwise it's a job cleanly done. The dried clay is not very tough can be chipped off with a small screwdriver.

[Noopy]

Looking good!
But the die has seen a lot of temperature, right? 

[magic]

I will try to take pics later but I saw no obvious damage looking under a strong loupe.

I'm pretty sure they have to melt that glass in the factory during production, so hopefully it has reasonably low melting point. And it's the top cover which is exposed to flame and receives most of the heat. I removed it as soon as it was possible.

LOL, I may even test if it works because bond wires still appear intact. It is one of my objectives to preserve the pins and wires so I don't have to guess what was connected to where.

[Noopy]

I'm pretty sure they have to melt that glass in the factory during production, so hopefully it has reasonably low melting point.

Sounds reasonable. Never thought about that. 

LOL, I may even test if it works because bond wires still appear intact. It is one of my objectives to preserve the pins and wires so I don't have to guess what was connected to where.

 

[RoGeorge]

I've read ceramic cases when hit with a hammer sideways, or squeezed in a vice's jaws from sideways, sometimes split top-bottom relative to the die's plane. 

Recently I've read about a very similar technique to remove the metal cover from the top of i7 4790K processors cases (so to stick the radiator right on top if the main dies of the i7, for better cooling, overclocking, etc - i7 4790K was the model with an unlocked multiplier for the core frequency, so very overclock friendly).

Seems like the common idea is to apply sideways pressure and to let the up and down plates to fracture along a plane with maximum defects in the ceramic or other covering material, that plane being the one with the die(s).

Something similar with this https://youtu.be/Fo1KzqCQurk , but seen it with ceramic cases and smaller chips.  Can not find that link with ceramic case split now.  Never tried that myself, so no idea how often chipping ceramic packages with sideways pressure will damage the silicon die.

[Noopy]

I work with hammer and screwdriver to split ceramic packages. That works quite fine for me.

Removing the lid of i7 CPUs always look a little crazy. 

[magic]

It works, but pins go randomly with either half of the package. And there is some risk that something will go wrong and the whole package will break, taking the die with it.

Anyway, my die looks quite happy so far, no signs of thermal damage. Not sure if it's LF356 or LF357, it's one of those $1 AD797 from AliExpress.

By the way, if anyone ever wondered if those screw-on macro lenses are any good, this photo was taken with a 120mm equivalent lens with the following stack on it:
- B+W 52mm +4D macro filter
- Raynox DCR-250 +8D fancy Japanese achromatic macro filter
- RISE(UK) brand (China) 37mm +10D macro filter
It could be worse

[mazurov]

Here is another recipe, designed to strip epoxy paints:

https://patents.google.com/patent/CN101649068B/en

[Noopy]

By the way, if anyone ever wondered if those screw-on macro lenses are any good, this photo was taken with a 120mm equivalent lens with the following stack on it:
- B+W 52mm +4D macro filter
- Raynox DCR-250 +8D fancy Japanese achromatic macro filter
- RISE(UK) brand (China) 37mm +10D macro filter

By the way:
I have done quite some tests with the Canon MP-E 65mm 1-5x.
In the range 1-5x it makes really nice pictures and is a good addition between my Canon 100mm L Macro and my "extreme DSLR magnification setup".
But while getting more magnification (2x extender, bellow rings) the pictures quality is getting worse fast. So in the end I will stay with my 10-22mm construction.

Here is another recipe, designed to strip epoxy paints:

https://patents.google.com/patent/CN101649068B/en

"dichloromethane, formic acid, phenol and sulfuric acid"
I don´t like that... 

[mazurov]

"dichloromethane, formic acid, phenol and sulfuric acid"
I don´t like that... 

Still, much easier than nitric. DCM is kind of annoying to work with since it evaporates so quickly,  other ingredients are manageable.

[Noopy]

"dichloromethane, formic acid, phenol and sulfuric acid"
I don´t like that... 

Still, much easier than nitric. DCM is kind of annoying to work with since it evaporates so quickly,  other ingredients are manageable.

DCM sounds more than annoying (for me). 
I always was interested in chemistry but I still have great respect for unhealthy substances.

[magic]

This topic has been beaten to death

There is no such thing as "the" epoxy resin. It can be any polymer formed by reaction of epoxide groups on some monomer with any other stuff, and there is lots of such polymers with different properties.

Various substances like formic acid or even acetic acid (vinegar), DCM, DMF, DMSO, phenol are known to attack certain epoxies, sometimes at room temperature, sometimes hot. At the same time, they do nothing to ICs.

Myself, I'm tired of that stuff. If it hasn't been tested on ICs, I don't care

As for acids, nitric is way better than sulphuric. Less temperature, cleaner byproducts. But the fumes... And you need to be a terrorist to obtain it in some jurisdictions, or at least a commercial business.

[RoGeorge]

By the way:
I have done quite some tests with the Canon MP-E 65mm 1-5x.
...

I'm curious if you ever tried cameras with smaller sensor.  Usually DSLR's have big sensors, the more expensive are even full frame, as wide as the old 35 mm film.  This is a disadvantage for macro photography.  Bigger sensors implies less DoF (Depth of Field), this comes from the laws of optics.

The end results for closeups or macro shots are visibly better with a consumer camera (small sensor) than with a DSLR (wider sensor).  For example, when it's to take a PCB shot, my old Minolta DiMAGE Z1 (consumer camera, small sensor) makes nicer shots that looks better than the same shot taken with a Nikon D90 DSLR, because the smaller sensor has more DoF .

You said you use focus stacking to increase DoF, but have you ever tried focus stacking on photos made with a much smaller sensor, like a consumer camera, action camera or even a webcam?  I know for single shots there is a visible difference, but I'm curious if the difference in sensor size will still matter after applying focus stacking.

[magic]

The oracle says: stop down to f/22

I have done quite some tests with the Canon MP-E 65mm 1-5x.
In the range 1-5x it makes really nice pictures and is a good addition between my Canon 100mm L Macro and my "extreme DSLR magnification setup".
But while getting more magnification (2x extender, bellow rings) the pictures quality is getting worse fast. So in the end I will stay with my 10-22mm construction.

That one looks like a lens exactly for the kind of shots that I bodged yesterday.

If you get a chance, you could try the +5D Raynox DCR-150 in your reversed lens setup (screw it onto the reversed lens). It reduces the front side focus distance of a lens to less than 20cm while the lens "feels" like it is focused much further away, this could eliminate some of the problem with reversed lenses not working well at low enough magnification to be useful. DCR-250 is the same but for 125mm, probably too short.

The simple, single-element diopter filters are probably not worth the hassle if you are after high quality. They will create chromatic aberration around highlights, and possibly worse.

BTW, don't be lazy and use those snap-on adapters. They are plastic, and I have seen people selling used Raynox macros with broken adapters, which means that they can break, fall off and make all sort of mayhem. I bought a proper 43mm filter step down ring for mine.

[Noopy]

I'm curious if you ever tried cameras with smaller sensor.  Usually DSLR's have big sensors, the more expensive are even full frame, as wide as the old 35 mm film.  This is a disadvantage for macro photography.  Bigger sensors implies less DoF (Depth of Field), this comes from the laws of optics.

The end results for closeups or macro shots are visibly better with a consumer camera (small sensor) than with a DSLR (wider sensor).  For example, when it's to take a PCB shot, my old Minolta DiMAGE Z1 (consumer camera, small sensor) makes nicer shots that looks better than the same shot taken with a Nikon D90 DSLR, because the smaller sensor has more DoF .

You said you use focus stacking to increase DoF, but have you ever tried focus stacking on photos made with a much smaller sensor, like a consumer camera, action camera or even a webcam?  I know for single shots there is a visible difference, but I'm curious if the difference in sensor size will still matter after applying focus stacking.

That´s a very complex topic.
Up to now i haven´t tried smaller sensors because:
- I would have to built a robust mount for the combo (no easy task).
- The DoF is no bigger problem for me. For tilted pictures with my magnification I will always need focus stacking. Without tilting a die is flat enough to get everything in focus.
- The limit for the magnification is the lens, not the sensor. In theory my setup gives me 200nm / pixel but effectively the resolution is between 1 and 5µm.

One interesting point here: Reducing the aperture ("f/22") reduces the magnification! You need a 2.0 lens to get a lot of resolution in this case.
(I have tested a 35mm 2.0 but the image quality was worse...  )

If you get a chance, you could try the +5D Raynox DCR-150 in your reversed lens setup (screw it onto the reversed lens). It reduces the front side focus distance of a lens to less than 20cm while the lens "feels" like it is focused much further away, this could eliminate some of the problem with reversed lenses not working well at low enough magnification to be useful. DCR-250 is the same but for 125mm, probably too short.

Well it´s worth a try... 


Edit: There was a guy who calculated the resolution of my system (10mm, f/3.5). He came up with 1-2µm which fits quite well to my pictures.

[magic]

Rules of thumb in microscopy:

F-number to numerical aperture:
NA = 0.5 / F

Numerical aperture to resolution (Abbe criterion):
d = 0.3µ / NA (for 600nm light, or λ/2/NA in general)

The resulting d is the minimum distance between two identical points (or lines) that are "barely resolved", or smeared onto each other with a small dip in intensity between them.

For f/3.5, NA is 0.14 and d is ~2µ.

AFAIK those approximations are valid when one side of the lens is focused at infinity, which means that resolution can be slightly (up to twice?) worse at close to 1:1 magnification. That's not a big deal for most of the time.

[magic]

Of course, the above rules can be conveniently combined to:
d = λ · F

And I suppose F/2 could also work if you are seriously lazy.

My LF357 (compensation caps are much smaller than on Noopy's so I figure it must be the 7) really looks quite fine despite the temperature. But I still haven't checked if it works because I try to keep it dust free for testing of my new microscopic system.

Project objectives:
- capability to inspect ICs still partly in the package (CERDIP, TO220 - these are very hard to remove completely)
- better structural rigidity than offered by a matchbox
- better image quality if possible, please

Long story short, I bought a lot of M42 macro extension tubes and some thread adapters. Then with that stuff and some plastic sheets and paperboard I constructed an M42 epi-illuminator (45° half-mirror plus a hole in the wall) and - possibly the first in the world - M42 mount webcam

I experimented with various setups and this is the winner so far: 10x biological microscope objective (abused at infinity focus, seems not to mind it much), the illuminator, Raynox DCR-250 macro, 125mm of M42 tube, the webcam. The whole tower is supported vertically above the IC by another stack of M42 tubes, partly unscrewed so they work as focus adjustment.

Remaining problems:
- some chromatic aberration - likely from the objective (it's old junk and I have dropped it on a hard surface once, so maybe that's why)
- still imperfect illumination, you can see some areas are brighter and/or different shade than others

Not problems:
- detail and resolution - nice and clean
- contrast and color - it's just this damn LF357 looking dull, I may upload some sexier dice later to show what the system is capable of

The LF357 picture below is a mosaic of 7×5 images captured with the webcam, stitched in hugin, postprocessed and scaled 50%. IMHO it's not too bad, and the setup is pretty low cost (the Raynox being the worst offender), mostly DIY and highly tweakable.

[Noopy]

Interesting! 
Pictures please! (of the optical stack)

What kind of 45° half-mirror do you use?

How much distance is between the microscope objective and the die?

In my view the illumination is not bad! Hey that´s no Zeiss-20k€-darling! 

I would recommend some tonal correction / saturation:





Do you like Hugin? I tried it once and had some problems.  Helicon Focus is more foolproof but unfortunatelly sometimes does crazy things. 

[magic]

First goes the stack of M42 tubes with the M42-RMS adapter and the objective on top.
Onto that goes the illuminator. It's a simple construction using black paperboard and the good old smartphone mirror foil. It's not really a foil, it's a pretty stiff plastic sheet and it maintains shape well. Light enters through a 6mm hole in the wall (on the left) and there is no more hidden tricks in there.
Next is the Raynox with necessary thread adapters and ~125mm of M42 tubing.
The camera is bolted to a plastic disc and screwed onto the top of the tower with a short M42 tube.

Also shown, a webcam lens converted to fit the M42/RMS setup using another disc and some hot glue

How much distance is between the microscope objective and the die?

Common achromatic 10x 0.2NA objectives tend to have ~1cm working distance.
With 20x 0.4NA it drops dramatically to 1~2mm.
More NA tends to be <1mm and reportedly performs poorly without a cover glass (biological objectives compensate for it).
It should be listed on the spec sheet.

I would recommend some tonal correction / saturation

Sure you can send contrast and saturation into the stratosphere, but it also amplifies faults. If you go too hard on that LF357 picture, the brighter-reddish and darker-greenish vertical bands become more visible.

Do you like Hugin? I tried it once and had some problems.  Helicon Focus is more foolproof but unfortunatelly sometimes does crazy things. 

Well, it's the first thing I tried and it worked out with a bit of tweaking.

It slows down a lot with many images (several minutes to align the 35 parts of LF357) because it tries to match every one with each other. I don't know how to tell it what the correct layout is. Somewhere on Zeptobars there is a "behind the scenes" post with an example of terribly screwed up Hugin job because it found similar features on images which just aren't adjacent to each other and warped everything to join those images anyway.

Hugin detected that each of my images contains that brighter-darker gradient and decided to brighten up the left side of the die and darken the right side so that the gradient is consistent across the whole image. I had to tweak that correction semi-manually - that's why there is a version 2.

[Noopy]

Hugin detected that each of my images contains that brighter-darker gradient and decided to brighten up the left side of the die and darken the right side so that the gradient is consistent across the whole image. I had to tweak that correction semi-manually - that's why there is a version 2.

Well with Helicon Focus at least you can place the pictures in the right position.
If Helicon Focus merges the pictures shifted a little you can correct that by moving the unprocessed pictures. But it´s more or less pure luck if the "shifted" picture gets better. 

[magic]

The post about Hugin was on Ken Shirriff's blog, not Zeptobars.
http://www.righto.com/2015/12/creating-high-resolution-integrated.html

I measured the LF357 with caliper and it's pretty much bang on 2×1mm. So the scale of my "scan" is 3px/µ before scaling or 1.5px/µ as posted at 50% size.

Calculated field of view is 0.5mm in diameter out of 2mm the lens is supposed to produce (10x magnification and 20mm nominal output). A larger sensor could deliver more FOV (and expose more illumination defects ).

Anyway, I think I have beaten those popular "USB microscope" toys at least

A much simpler setup is also possible with finity focus, no Raynox and either a scope objective or a webcam lens. But I found it to have somewhat worse contrast. Maybe further work on illumination is needed. It's not perfectly aligned and I still look for suitable light sources. 5mm LED is a bit too small, a larger LED torch is quite uneven, additional optics a'la Kohler is bulky and inconvenient and still not perfect for some reason

[Noopy]

The post about Hugin was on Ken Shirriff's blog, not Zeptobars.
http://www.righto.com/2015/12/creating-high-resolution-integrated.html

"The next step is to fix the control points. Because Hugin optimizes globally, even a few bad control points can mess up the entire image."
Do you have to put/check control points at every picture pair? For me it looks like I have to do so and that can be quite an effort.

[magic]

Honestly, I just loaded my images and ran the "scanned images assistant" from the menu, figuring that this is the closest thing to what I am trying to achieve. This took a long time (maybe 15 minutes) but it got the job done, except for the exposure gradient which I had to override in expert mode.

I now tried again fully in expert mode and without those "assistants".

At first, I ran automatic control point detection because I definitely am too lazy to add points by hand. This took one minute and found spurious control points, joining the upper input JFET with the lower one. I ran geometry optimization to see what happens and it produced the nice layout attached below. It seems that Hugin tries to minimize average mismatch of control points and shrinks the input stage a little to bring the false control points closer together at the cost of slightly misaligning the correct ones.

With a bit of clicking around I found two useful options. Firstly, right clicking on the image list brings up a menu with an option called "manipulate image variables". Here there are choices like "linear panorama with 20% overlap" or "two row panorama with 20% overlap" - neither is ideal, but I "loaded" the linear panorama preset and clicked OK to see what happens. As expected, the images became arranged in a long train, one row after another.

The second useful option is the choice of "Hugin CPFind (prealigned)" in the feature matcher selection. This finds control points as usual, but it only considers pairs of images which are currently overlapping. It took only half minute and found useful control points horizontally and no wrong control points. After a run of geometric optimization my five rows became neatly aligned, but they were still placed in a few random places on the plane. (I don't know why it didn't put them all in the center on top of each other, but I don't complain.)

To combine the rows, I selected pairs of first images from consecutive rows (11+21, 21+31, ...) and ran control point search on each pair. Then another geometric optimization placed the rows in the correct order, but there were still no control points for most vertical and diagonal pairs. So yet another prealigned control point search and another round of optimization. The final result was good and stitched cleanly.

So there is a light in the tunnel
I think the next obvious step is to figure out how those "manipulate image variables" presets work and how to arrange everything in the correct number of rows from the beginning.

BTW, those PTO files are simple text files so it should also be possible to write a script which sets image coordinates to whatever we want.

[Noopy]

Thanks for the explanation!
Perhaps I start another try with Hugin.
Helicon Focus doesn´t work bad enough for a fast changeover. 

[magic]

Well, maybe not so fast. I encountered another problem: wrong CPs on a correct pair of images. For example, different fingers of the same transistor. But they have been rare so far and easy to find by opening the CP list and sorting it by "distance" (see Ken Shirriff).

It may get worse on ICs with a lot of repetitive features.

I solved the problem of pre-aligning multiple rows. For whatever reason, and at least on my system, the "manipulate image variables" window opens with just a few buttons. But if I enlarge it, suddenly a text box appears which hold the script to be executed. Modifying the "two row panorama" script for five rows is a matter of changing one number and then it works for me.

[magic]

What's the deal with the weird shadows on SDH112? One of those crazy things that Helicon Focus does?

Maybe we should just figure out how to correct brightness gradients with ordinary image editors before running those stitchers...

Well, maybe not so fast. I encountered another problem: wrong CPs on a correct pair of images. For example, different fingers of the same transistor. But they have been rare so far and easy to find by opening the CP list and sorting it by "distance" (see Ken Shirriff).

This has been resolved, at least for LF357. I plan to scan some more difficult IC and see how good it really is.

It turns out that CPFind is not entirely retarded and it doesn't just look for similar points on both images, but it looks for a set of similar points which can be aligned by transforming the images in some way. By default, it considers any homographic transformation, which is quite a lot of different distortions that can occur when shooting a large object from different places at different angles, but they simply don't occur here.

The solution is to change "ransacmode" to "rpy" and set the horizontal field of view absurdly narrow (1° seems OK) so that Hugin's pitch and yaw coordinates (p,y) are more or less equivalent to shifting the camera sideways (x,y). This requires going to preferences and defining a new control point detector. And then choosing to use it, of course.

http://hugin.sourceforge.net/docs/manual/Cpfind.html
http://hugin.sourceforge.net/docs/manual/Hugin_Parameters_for_Control_Point_Detectors_dialog.html

[Noopy]

What's the deal with the weird shadows on SDH112? One of those crazy things that Helicon Focus does?

 

Helicon Focus is innocent this time. Sometimes I have problems with the light. Working with a fixed exposure doesn't always work for me because the light from behind fluctuates at different locations of the die.
Often I can improve the picture quality with a second and a third try but sometimes I consider the quality as "good enough". 

[Noopy]

Maybe we should just figure out how to correct brightness gradients with ordinary image editors before running those stitchers...

That's not always possible. In my pictures often it's not only the brightness but also "the color" that differs.
I will try a "half mirror light solutions" similar to yours. Perhaps that gives more constant light quality.

[magic]

For large lenses you may need to use 28mm ring for that instead of 14mm as I did. Or maybe not, who knows, have fun
Mirror position and angle is quite important, otherwise the light goes somewhere to the side of the FOV. Infinity focus system seems to be somewhat less sensitive to that, so I am using it for now.

Further testing of my Hugin configuration revealed that Maxim definitely curses their ICs to thwart reverse engineering. Just when I thought I got a cleaner die than yours, I dropped an M42 tube on it

And there is more chromatic aberration due to brighter metal. I definitely need to fix illumination and get my webcam lens to work because it has virtually zero CA. I know it can be done because it was better in the matchbox than it is now. I may also try infinity focus with webcam, but I expect impractically high magnification.

Nevertheless, Hugin delivered. 18 frames with only 25% horizontal overlap, everything went smoothly and automatically. To recap, this is the flow that works:

0. Add a new control point finder in preferences; copy everything from "Hugin's CPFind (prealigned)" but change arguments to "--ransacmode rpy --prealigned blah blah blah". This only needs to be done once.
1. Load the images, specify HFOV as 1° or focal length as 9999 or something like that.
2. Run "manipulate image variables", load the "2 row panorama" preset, adjust for actual number of rows.
3. Select the customized control point finder and run it.
4. Manually add a pair of horizontal/vertical line control points on any image so the IC comes out rotated correctly.
5. Run the default geometric optimization.
6. Play with photometric optimizations if necessary (exposure, white balance).
7. Autosize, autocrop, stitch.
 

BTW, I think there may be some CMOS logic stuff on this reference, right under the output stage. The green rectangles further below also look like some bigger MOSFETs.

[mawyatt]

Been involved with chip imaging (mostly the chips we designed) since ~2000. We've done images as large a 30,000 by 20,000 pixels. There's lots to learn and here's some links that will help. Many world class macro/micro photographers are involved with these sites, especially the Photomacrography site hosted by the Rik the author of Zerene Stacker.

For a complete evaluation of potential lens be sure and visit Robert's site CloseUpPhotography.

Can post some of the setups we've developed over the years if folks are interested.

Best,

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

http://zerenesystems.com/cms/stacker

https://www.closeuphotography.com

http://extreme-macro.co.uk

[Noopy]

You have a really nice setup but unfortunately over all it's quite expensive.  I wish I had unlimited credit... 

[mawyatt]

We developed some setups around the very affordable WeMacro stand. This is an excellent setup which can be utilized both vertically and horizontally, and with some simple and inexpensive modifications becomes a quite respectable setup for serious macro work. This became our "Go To" setup for most of our chip imaging work, highly recommended!!

https://www.wemacro.com

Note the 1st image that comes up on the WeMacro site was requested by William at WeMacro, it used his affordable microscope objective, stand and is part of a non-proprietary test chip we developed ~15 years ago.

Here's some links about the stand and some modifications.


https://www.photomacrography.net/forum/viewtopic.php?f=25&t=34617&hilit=Modified+WeMacro+Stand
https://www.photomacrography.net/forum/viewtopic.php?f=25&t=38353&hilit=Modified+WeMacro+Stand
https://www.photomacrography.net/forum/viewtopic.php?t=38838&highlight=wemacro
https://www.photomacrography.net/forum/viewtopic.php?t=38511&highlight=wemacro
http://www.photomacrography.net/forum/viewtopic.php?t=38353&start=0&postdays=0&postorder=asc&highlight=stack+stitch

Best,

[mawyatt]

Also, if you spend some serious time on the sites mentioned, including CloseUpMacro, you'll find some really good lenses at bargain prices. Lenses like the old Soviet Lomo 3.7, or some repurposed scanner lenses. The Minolta 5400 scanner lens became an overnight hit because it was so good, and the price shot up after others found out, fortunately we got one before the price went too far

The cheap AmScope 4X isn't bad either, sure it's not in the Mitutoyo class, but for the price quite a bargain 

https://www.closeuphotography.com/seventeen-dollar-plan-4x-objective

The message is spend some time researching on the sites mentioned, it will pay off

You can build a very good setup with lenses, mechanicals, and linear stepper based focus rail stages for a modest investment. Of course this just opens your pocket book since later after you've mastered the setup and imaging techniques, in pursuit of the perfect image you'll want better optics, focus rails and so on.

So please don't discount everything as being too expensive, if you follow the advise above you'll end up with a quality lens and setup for a very modest investment, however there's no free lunch as you must do your homework

Best,

[Noopy]

Probably you are right.
Perhaps I should do some more research.
I´m quite happy with my equipment (resolution/€) but a little more pictures quality would be nice of course. 

I have seen a lot of your pictures with tilted chips.
- Did you have a solution for tilting the die to get it flat parallel to the lens? Often such precision tilting blocks are 100€ to 300€...
- Did you have a solution for light through the lens? Or did you just do tilted pictures?

[mawyatt]

Probably you are right.
Perhaps I should do some more research.
I´m quite happy with my equipment (resolution/€) but a little more pictures quality would be nice of course. 

I have seen a lot of your pictures with tilted chips.
- Did you have a solution for tilting the die to get it flat parallel to the lens? Often such precision tilting blocks are 100€ to 300€...
- Did you have a solution for light through the lens? Or did you just do tilted pictures?

Your pictures are quite nice with good detail, sharpness and contrast 

If you are looking for a little better quality, maybe when shown much larger then I highly recommend studying Robert O'Tools site, he has put in an enormous amount of effort in lens tests. Note that he also uses silicon wafers as the target subject. Photomacrography also has lots of information on various lenses and combinations, many of which are not expensive.

Precision tilting blocks can be expensive, watch eBay for some bargains, I got a couple last year for ~$75 each, however I haven't used them yet.

For the chip mounting, the chip backside was glued to a pin head which was held in place by an alligator clip on a photographic ball head. The head was either mounted to a XY or XYR micrometer stage or a XY linear rail, shown in the old images below when we 1st starting venturing into Stack and Stitch territory. The XYR micrometer stage is an inexpensive clone off eBay, ~$70, the ball heads were $10~15. The XY linear rails are surplus THK KR15s, also ~$75 with NEMA 11 motor included. The stand is a WeMacro ~$130.

Have never used thru the lens lighting, all my lighting was done with highly diffused light from multiple strobes or LEDs. Cheap eBay light tents work well as diffusers, have even used two, one small tent inside a larger tent. White foam cups also work very well.

Tilted images can be difficult since you need significantly more depth, but add a degree of 3D like to the image. One issue is that the perspective can change as the distance between the subject and lens changes when doing image stacking. This can become extremely difficult with stitching, much more so than stitching an insect because the chip has straight and right angle edges which must align between tiles. Here is where the special class of "Telecentric Lenses" comes into play, these lenses have a unique property that the magnification does not change with subject to lens distance. Do a search over at Photomacrography for details.

Best,

[magic]

Lighting through the lens (or reflected off the lens) is relevant in that it is a straightforward way to avoid focus stacking altogether and a requirement for serious NA/resolution.

Extreme macro technology may be applicable to Noopy's closeups of package internals, bond wires and stuff like that, but detailed die imaging is strict microphotograhpy territory and the path has been shown by guys like Svarichevsky and Shirriff IMO. The EM crowd has some useful information about optical components, though.

And I wouldn't expect bargains once they find out about something

edit
Purely for science, I ran my MAX6250 project through Hugin on default settings. I tested:
- run CP find, run geometry optimization
- switch to beginner mode and just click "align"
- the "scanned images" assistant

All of them failed hard and produced the familiar "cubist mess", albeit a different mess each time.

[mawyatt]

Lighting through the lens (or reflected off the lens) is relevant in that it is a straightforward way to avoid focus stacking altogether and a requirement for serious NA/resolution.

This is incorrect, lighting of any kind has absolutely nothing to do with focus stacking, NA or resolution!! They are all lens dependent.

DoF ~= Lambda/NA^2  for small NA

Resolution = Lambda/(2*NA)

NA = {(1/(2*F))*(M/(M+1)}

Effective Aperture = F(M+1)

Where Lambda is wavelength (~550nm for green), F is lens aperture, M is magnification

Examples:

The Mitutoyo 5X Inf objective lens has a M of 5, NA of 0.14, DoF of 28um, Resolution of 2um, and Effective Aperture of 18

The Mitutoyo 20X has M=20, NA=0.42, DoF=3.2uM, R=0.7um and EA=24

So if you are after higher Resolution then you want a larger NA, which causes less DoF. Focus Stacking is simply a means to get around the shallow DoF caused by higher M and Resolution needs, and has created some excellent high resolution images of subjects with significant effective depth (like chips tilted for example).

Lighting is a completely independent subject of itself, many techniques are available and used. Each has pros and cons, but we've never needed to use thru the lens for any of our work (we mostly image the chips we designed, some even at 50X).

The tradeoff between M, DoF and Resolution creates a barrier called Diffraction which is being addressed with post techniques called Deconvolution, but that's another highly involved subject.

Best,

[magic]

Well, it does. With proper lighting DoF can be rendered irrelevant, as it should be with a flat subject

As for resolution, the problem is that high NA objectives tend to have short working distance and not much space is left for lighting. Surely you can spend four figures on ELWD objectives and precision macro rails, but not everybody will go that far. And if one does, it is not entirely clear why the same money wouldn't be better spent on a metallographic scope and ordinary WD optics with more NA.

If you have infinite money, I would bet on the latter setup winning too. I don't quite see combining 1.4NA oil immersion with ping-pong balls.

All of the above applies to subjects that are indeed flat, of course.

[Noopy]

That´s a complex topic. You have to built a setup that does most of the things you have on your wish list with the amount of money you can spend.
My wish list is quite long and calls for different solutions:

• big parts / dies
  I take my Canon 100m 2.8L Macro and take pictures with a magnifications of 1x or lower.
  For "parts" I use diffused light by putting a paper over the part and moving a CFL desktop lamp above it. Works quite well.
  For big dies I use light from behind the die to get the nice colors. I like those pictures to get an overview since pictures stacking consumes more time.
  Adjustments are done by holding the part with a third hand and bending the arm. That´s good enough for 1x and lower.
• die - flat
  I take my reversed 10-22mm and add the tubes I need to get the right magnification: For example i take 10mm+0 for an overview and 10mm+2 for details or a big stacked picture.
  Light comes from behind and I´m quite fast with changing the magnification.
  I own a small micro manipulator for x, y and tilting over the two axes. That´s makes it easy to adjust the alignment.
  The sharpness is controlled with the focus ring.
• die - tilted
  Side views and "3D structures" need "tilted pictures" and focus stacking. But tilted dies often give some more sharpness and details too. I don´t know why.
  I use the setup described above and vary the focus. That´s worse than moving the object because the magnification changes but I´m quite fast and moving the object needs a though construction.
  Lightning is a CFL desktop lamp. I can vary the light angle to get the best possible pictures.
  
• something for inbetween 
  Sometimes I use my new Canon MP-E 65mm 1-5x. That´s the magnification at which my 100mm Macro isn´t good enough (with distance rings). The 10-22mm retro at it´s "lowest" configuration 22mm+0 has quite some magnification, sometimes too much. And the Canon MP-E 65mm 1-5x is really sharp (at 1-5x). 
  The Canon MP-E 65mm 1-5x has no focus so I have to use a macro rail. Since I have only a cheap one it´s not possible to mount the camera on the rail: The rail moves sideways a little while moving forward and the mirror movement makes the picture blurred. Mounting the lens on the rail is bad and mounting the camera is bad also. Because of that I place the die on the rail (with my third hand). That looks strange but guarantees sufficient image quality.

My whole setup is placed on a esd pad. That seems to be a perfect support. The parts stick a little on the surface. The pad is hard enough that the parts don´t sink into the pad but does a little damping.


A change in my setup has to improve the already good quality. It has to fit into the rest and it must not slow down the process. That´s no easy task. 
I´m afraid changing something makes me change a lot and would consume a lot of money. And for some problems I still have no solution. For example the light through/from the lens with a vertical setup. Of course I can built something like magic but I´m afraid such a modification makes the new setup worse than the old one...

[magic]

But tilted dies often give some more sharpness and details too. I don´t know why.

Shadows of all the vertical steps on the surface, perhaps?
Colors also change so sometimes it could be down to that.

I´m afraid changing something makes me change a lot and would consume a lot of money.

That's why I play with cheap junk to roughly figure out what matters and what does not

Light through the lens may help with shadows in the center of a big die. Otherwise, I'm not sure if it will be any better than what you have.

[mawyatt]

Well, it does. With proper lighting DoF can be rendered irrelevant, as it should be with a flat subject

Now you've restricted yourself to flat images because of a shallow DoF. Flat images tend to be less artistic than images with some depth, most folks I know prefer a tilted chip subject which shows some of the features better and gives a sense that a chip isn't just a 2 dimensional object.

As for resolution, the problem is that high NA objectives tend to have short working distance and not much space is left for lighting. Surely you can spend four figures on ELWD objectives and precision macro rails, but not everybody will go that far. And if one does, it is not entirely clear why the same money wouldn't be better spent on a metallographic scope and ordinary WD optics with more NA.

If you know what you are doing getting sufficient illumination even at 20~50X can be done without the need for ELWD objectives, or thru the lens lighting. As mentioned spend some serious time over at Photomacrography and learn from some of the world class macro photographers there (I'm not one).

If you have infinite money, I would bet on the latter setup winning too. I don't quite see combining 1.4NA oil immersion with ping-pong balls.

And why would anyone that knows anything about chips, lenses and macrophotography use an oil immersion lens on a chip, seems like a very messy recipe??

BTW don't discount a ping pong ball or foam cup used as a diffuser, if you look at the results over a Photomacrography you'll soon realize just how good they are as diffusers!! Some of the images below likely used a foam cup, but can't remember details since these are almost 10 years old, or older.

All of the above applies to subjects that are indeed flat, of course.

What about a die that has wire-bonds, how do you propose to show such a subject with a shallow DoF, yet the wire-bonds add depth?? The wire-bonds would come into the die image as a surrounding massive blurry mess, but utilizing focus stacking you can create an image like these very low resolution versions which shows the die and the wire bonds both with reasonable clarity.

Also, note the last image below used the $17 AmScope 4X we mentioned earlier, not an expensive lens at all 

Anyway, most will prefer the focus stacking rendered image IMO and why we went to all the effort to do such. Folks at IBM, Cornell, USC, MIT-LL, DARPA, Keysight, Harris, Excelis, ITT and a few others I can't mention thought so too

BTW our images have been printed or displayed in very large format (some 2 meters across), and inspected in detail, so they needed to be accurate and have very high resolution

Noopy's images are all very good and he sure does quite a few, my notes were if certain images of particular interest were to be enlarged then he might want to consider alternative techniques, including focus stacking. If he were here is the US and came by, I'd probably help him with some things, and even donate some equipment.

Best,

[mawyatt]

Noopy,

Your images and techniques are fine, producing nice images without much addition effort

Don't want to imply anything else, the equipment and technique I mentioned were just as information that you might find useful.

Your use of the stacked lenses (100mm and 10-22mm) is an excellent way to get around some of the diffraction limits imposed, see details over at PM, lots of  discussions on this stacked lens topic.

If you look at the last image I posed above, reposted here in slightly higher resolution, this is from an AmScope $17 4X lens than Robert O'Tool discovered a number of years ago. If you decide to acquire this lenses be very careful, lots of sub-par clones showed up after this was discovered.

Best,

[mawyatt]

One of the more difficult chip related things to image we discovered a number of years ago is new chip solder balls. These are like tiny (40~100um) bright spherical mirrors before they oxidize and reflect everything, especially the illumination sources. All sorts of illumination technique were tried and the highly diffused (usually 3 or 4 levels of diffusion) multiple strobes proved acceptable. This produces a very uniform illumination from all directions, without significant specular artifacts which detract from the final rendered image.

Best,

[Noopy]

magic, mawyatt I highly appreciate your knowledge and experiences! 
It´s a complex topic and the solution can be different for different requirements.
I like my setup but getting better is always a good thing. 

But tilted dies often give some more sharpness and details too. I don´t know why.

Shadows of all the vertical steps on the surface, perhaps?
Colors also change so sometimes it could be down to that.

I´m not sure... ...it´s different, sometimes better.





The 3D effect helps but it seems like it´s a little sharper too.

Your images and techniques are fine, producing nice images without much addition effort

Don't want to imply anything else, the equipment and technique I mentioned were just as information that you might find useful.

Your use of the stacked lenses (100mm and 10-22mm) is an excellent way to get around some of the diffraction limits imposed, see details over at PM, lots of  discussions on this stacked lens topic.

I appreciate your hints! 

But I don´t  use stacked lenses. I just use reversed lenses, mostly the 10-22mm.



Works fine! 

One of the more difficult chip related things to image we discovered a number of years ago is new chip solder balls.

Yeah, solder balls are wierd. 

[magic]

Flat images tend to be less artistic than images with some depth, most folks I know prefer a tilted chip subject which shows some of the features better and gives a sense that a chip isn't just a 2 dimensional object.

I repeatedly tried to make it clear that I am after transistors, not solder balls and bond wires. Those I would rather tear off so they don't obstruct the view

Your cheap 4x achromats would struggle to produce the LF357 image that I posted recently and this is 1980s technology. The Maxim is 1990s technology and my 10x achromat struggles to resolve some details. And there are chips from the 2000s that I would like to see. With the <2mm working distance of common high mag objectives I can only imagine external illumination as nothing but a massive PITA. Epi at least stays attached to the optics - mount it once and it keeps working, chip after chip.

And why would anyone that knows anything about chips, lenses and macrophotography use an oil immersion lens on a chip, seems like a very messy recipe??

Because the formula for resolution that you posted is incomplete and I'm sure you know it

The 3D effect helps but it seems like it´s a little sharper too.

It does look sharper indeed. It could be due to hard shadows, or I'm not sure what else. I presume it's the same optics, subject distance and magnification in each case?

At any rate, I applied some fake sharpness and contrast to it. I don't know if it's better, but it's different

But I don´t  use stacked lenses. I just use reversed lenses, mostly the 10-22mm.

You could. It really is as simple as getting a different kind of passive reverse mount adapter (<$5 delivered here) and replacing the extension tubes with a telephoto lens. You get the benefit of both lenses working near infinity focus, which most photographic lenses are optimized for. AF can be used for fine-tuning focus. And some gain in NA if working near unity magnification, but I don't know what magnification you use.

edit
Although it must be said that not all tele lenses are suitable for this job. Vignetting or all sorts of distortion around the edges is a possibility. But reverse mounting rings are cheap to try...

I'm sure I have seen a post explaining what sort of lens to look for on some internet forum. You could try to find it - it was photomacrography most likely. I think the first rule was that front element diameter is not supposed to be much larger than absolute aperture diameter (= focal length / F-number).

[RoGeorge]

The oil immersed objective is a trick used in microscopy.  It shows more details for the same optics.

At minute 4:35 there is an image with and without oil, side by side.

[magic]

It shows more detail with different optics. It's not just about sealing to keep oil from leaking into the objective as they say on the video, but the whole objective is designed for light coming at different angles and refracting differently.

In fact, immersion oils are produced to have similar refractive index as glass, so that the borders between the cover slip, the oil and the front surface of the objective almost disappear. Serious refraction starts to occur on internal surfaces of the objective, which can be ground precisely, aligned and then left alone undisturbed by external factors.

For that reason I wonder if biological oil objectives could actually be usable without covers slip, but I'm not rushing to buy one because I doubt I would be able to provide necessary focusing precision and illumination quality. To collect a wide cone of light from a reflective surface one has to first project an equally wide cone from above.

Once upon a time, Zeptobars posted photographs of some submicron CMOS stuff using oil immersion on a proper microscope.

[mawyatt]

Your cheap 4x achromats would struggle to produce the LF357 image that I posted recently and this is 1980s technology. The Maxim is 1990s technology and my 10x achromat struggles to resolve some details. And there are chips from the 2000s that I would like to see. With the <2mm working distance of common high mag objectives I can only imagine external illumination as nothing but a massive PITA. Epi at least stays attached to the optics - mount it once and it keeps working, chip after chip.

First off, these are not MY cheap achromats, ones I learned about from Robert O'Tool at Close-up Photography and Photomacrography, not a bad lenses for $17 back then. Second, we have no problems even at 20X with diffused illumination, or resolving detail

Because the formula for resolution that you posted is incorrect and I'm sure you know it

Seems Microscopy University at Nikon, Wiki (Optical Resolution), Leica, Zeiss must all be incorrect then, as they Refer to Resolution (r) = Lambda/(2NA) which is the generally accepted criteria, although some use Resolution (r) = 0.61*Lambda/NA based upon Rayleigh criteria. However, I'm sure you know much more about this than all the folks at Nikon Microscopy University, Wiki, Leica, and Zeiss 


Best,

[magic]

Upon some consideration I have already changed "incorrect" to "incomplete" while you were writing your post

Come on, we both know what monsters hide in that "NA" thing that you talk about and we both know that we know

[mawyatt]

Noopy,

Using stacked lenses rather than a revered lens, has some very useful benefits regarding diffusion, which becomes the ultimate limit in achieving quality images at higher magnification. Folks strive to keep the Effective Aperture, which is Lens Aperture*(1+Magnification), as low as possible to keep diffusion at bay, even at F10 diffusion starts to degrade things. Most of the higher end objectives are designed for a lower EF for this very reason. PM has some very useful information on stacked lenses.

For example, a typical 50mm lens stacked onto your 100mm usually produces a very good 2X combination. With your 22mm and the 100mm you might have a good ~5X and likely much better than using the 22mm alone, reversed.

I'll wager a Brewisky that if you give this lens stacking technique a try you'll be pleasantly surprised   

Hey, how come we don't have Brewisky icon?

Best,

[mawyatt]

Upon some consideration I have already changed "incorrect" to "incomplete" while you were writing your post

Come on, we both know what monsters hide in that "NA" thing that you talk about and we both know that we know

Yup, and a lot more "Monsters" hide in the forests of detail

Murphy is also on the same team as the "Monsters"

Best,

[mawyatt]

The oil immersed objective is a trick used in microscopy.  It shows more details for the same optics.

At minute 4:35 there is an image with and without oil, side by side.

A little familiar with this technique, but never wanted to get the chips covered in oil since keeping the surface clean is already a chore, and adding oil they'll become a dust magnet

The typical oil has an index of ~1.5 which is highly benefical. The semiconductor fabrication process has used this "trick" for a long time now, and one of the many means they use to defy diffraction physics

Best,

[Noopy]

The 3D effect helps but it seems like it´s a little sharper too.

It does look sharper indeed. It could be due to hard shadows, or I'm not sure what else. I presume it's the same optics, subject distance and magnification in each case?

At any rate, I applied some fake sharpness and contrast to it. I don't know if it's better, but it's different

Yes, it was the same optics, the same subject distance and magnification. I just tilted the die a bit and adjusted the light.
It´s annoying not to know what is happening but I´m happy to have a second possibility to get nice pictures. ...better than some fake sharpness and contrast. 



I have already bought an adapter to test stacked lenses. I wanted to try that earlier but somehow forgot about it.
I have quite some lenses here. First I will try 100mm + 10-22mm. Better or worse I will drink a beer afterwards. 
It´s a pity I´m no optics expert...  But I highly regards your advices! 

[magic]

Yes, it was the same optics, the same subject distance and magnification. I just tilted the die a bit and adjusted the light.
It´s annoying not to know what is happening but I´m happy to have a second possibility to get nice pictures.

Well, assuming that focus wasn't screwed up on the flat version...
I still think that shadows are responsible for more perceived sharpness.
I suspect that flare is responsible for the overall gray appearance and low contrast.

What's the size of this die and how far was it from diffraction limits? That grain on metal traces has single digit microns size, you may start to see diffraction effects at that level with some lenses at some magnification.

My old pictures of fake opamps are a perfect example of the contrast problem. They were shot using exactly your method but with even lower end gear.
I just looked at them again, they really are embarassingly ugly

...better than some fake sharpness and contrast. 

Wasn't it you who encouraged me to do more postprocessing just a few days ago?

[magic]

Folks strive to keep the Effective Aperture, which is Lens Aperture*(1+Magnification), as low as possible to keep diffusion at bay, even at F10 diffusion starts to degrade things.

Diffraction

That formula looks to me like it is supposed to ultimately determine the diameter of diffraction blur observed on the sensor.
Shouldn't those folks better be concerned with, ahem, Input Referred Effective Aperture, that is EA/mag?

[mawyatt]

Nikon DSLRs reports the EA with their lenses, most others do not. Generally this isn't big deal since what's the magnification of a portrait shot at 3 meters, or a landscape, something less than 0.1, so the lens aperture and EA are about the same. The Canon MPE for example reports the lens aperture but is used for close up macro work. Had a long discussion about this some time ago on DPR where they were using the MPE at 3~5X with a lens setting of F11 to get more DoF. The EA was actually 44~66 well into the image quality robbing diffraction territory, not sure they ever got this tho

The highly regarded Mitutoyo inf 5X and 10X objectives have an EA of ~18, so I've always considered this the transition region where diffraction begins to effect the image, others work around F9~13 and the image purists well below F9.

I confess I do have a purists type lens, the Printing Nikkor PN105 F2.8A. This was a reproduction lens from movie film days used to produce replicas of movie theater film, so it's deadly sharp from corner to corner and highly optimized for 1X, so the EA is 5.6. Was it expensive when I got it 5 years ago, you bet! However, it's worth twice that today, so exactly how expensive its that!! Same goes for the Red Porsche 911, 25 years old and worth twice what I paid for it new in 1996, and it's still appreciating!! Expensive yes, but also a good investment for a car

Anyway, EA is one of the many parameters that effect IQ.

Another benefit of image stacking I forgot to mention, many lenses get soft around the edges, some are sharp but at a different focus distances than the center. With these type lenses stacking can help with the edge sharpness and make the final rendering sharp across the entire frame, but at the expense of taking more time to capture the stacks, so one trades time for sharpness so to speak

Best,

[Noopy]

Well, assuming that focus wasn't screwed up on the flat version...

It wasn´t!  I have seen that quality difference in many pictures. Tilted was always better.

I suspect that flare is responsible for the overall gray appearance and low contrast.

Yes, perhaps my "against the lens light" deteriorates the quality.
I have bought a teleprompter mirror and will do some experiments with different light sources.

What's the size of this die and how far was it from diffraction limits? That grain on metal traces has single digit microns size, you may start to see diffraction effects at that level with some lenses at some magnification.

The die was 1,2mm x 1,2mm. The magnification should be around 19x. Diffraction limit... ...yes, seems to be the end of the line... 

...better than some fake sharpness and contrast. 

Wasn't it you who encouraged me to do more postprocessing just a few days ago?

Often that´s a good thing but getting more contrast out of the die is always better. 

[magic]

The relevance of EA at high magnification is that it tells you how much shutter speed you need for given light, which could be of interest to film photographers I suppose

If you want to know how it affects resolution, you go to the oracle (or do the math by hand ). You tell the oracle your camera model, or crop factor and megapixel count. You set "lens aperture" to the calculated EA number. The oracle shows you how aperture affects your pixels. You remember that only half of your pixels are green and only a quarter is red and that there is an anti-alias filter too. (The oracle helpfully reminds you not to worry if blur is about 2px because at that level you are screwed by your sensor anyway.) You also consider how much you intend to scale the image down for publication.

Without crop factor and without megapixel count of the final, published image those EA numbers are meaningless. It is not surprising to find an order of magnitude spread between different people, gear and purposes.

EA is very convenient in ordinary photography, because it is equal to the F-number at infinity focus, 10% worse at 0.1 magnification and 25% worse at 0.25 magnification. You do the blur calculation once for any combination of camera body and F-number and you are set for life, as long as you stay away from macro.

For the same reason, it's an ass-backwards way to deal with high-mag. Since the dawn of ages and long before digital imaging, microscopists talked about object-side aperture. You can think about it in two ways. One is to calculate "input referred" blur, which is the sensor side blur divided by magnification. It tells you how the blur compares to magnified features on the object. The object-side aperture formula is: F·(1+M)/M. Or you can imagine that your sensor is the object and your object is the sensor (easy to imagine if the lens is reverse-mounted) and trace the rays backwards and then calculate EA that way, assuming magnification 1/M. The object-side aperture formula is: F·(1+1/M). Observe that the two formulas really are the same and that magnification is almost irrelevant as long as it is far from unity, just like in ordinary photography.

So you go to the oracle again. You tell the oracle your object-side aperture. The oracle shows you how many microns of blur you get on your object. If your object is 1200µ in size and the calculated blur is 1.2µ then you know it's 0.1% of your object. You don't care about the sensor at all, unless pixels are too coarse to resolve features that aren't blurred by the optics - then it's time to increase magnification or get a better camera.

Only at close to unity magnification both the macroscopic and the microscopic approximations are equally wrong by a factor of 2x. Then you are screwed and you have to do the math precisely - such is the sad life of bug shooters and that's why they worry about EA.

Often that´s a good thing but getting more contrast out of the die is always better. 

I sometime use ISO10 trick. I take 10 identical shots at ISO100 and average them with GraphicsMagick. ISO noise is greatly reduced and more contrast enhancement can be applied. Lens flare remains

edit
When stacking lenses, diffraction surely occurs in both lenses so in theory you should do both calculations and combine the results. It could be interesting to see what comes out and whether the telephoto lens turns out to be insignificant in practice or whether (maybe) some of the combinations that people recommend aren't really as good as they think because of the aperture of the tele lens. Again, all of that is messy calculations involving pixel pitch and sensor size. Perhaps somebody has already figured it out.

It is generally recommended to use the tele wide open. In my experience with imaging the Chinese opamps (reversed webcam lens stacked onto a point and shoot) stopping down the camera's lens made no improvement and caused bad chromatic aberration

edit edit
Actually, it's trivial. The tele lens is operating normally and focused at infinity. So as long as it isn't some diffraction-bottlenecked megazoom piece of junk, it should be good.

[mawyatt]

The relevance of EA at high magnification is that it tells you how much shutter speed you need for given light, which could be of interest to film photographers I suppose

Also is a good indication on diffraction effects and why many use this, myself included.

If you want to know how it affects resolution, you go to the oracle (or do the math by hand ). You tell the oracle your camera model, or crop factor and megapixel count. You set "lens aperture" to the calculated EA number. The oracle shows you how it affects your pixels. You remember that only half of your pixels are green and only a quarter is red and that there is an anti-alias filter too. (The oracle helpfully reminds you not to worry if blur is about 2px because at that level you are screwed by your sensor anyway.) You also consider how much you intend to scale the image down for publication.

Agree, I was enlightened decades ago by Cambridge, it' a good resource, and confirms my experience with various lenses, diffraction and viewing details.

I sometime use ISO10 trick. I take 10 identical shots at ISO100 and average them with GraphicsMagick. ISO noise is greatly reduced and more contrast enhancement can be applied. Lens flare remains

ISO is a carry over from film days, and in digital cameras generally represents the amplifier gain preceding the ADC. The "ISO10 trick", isn't a "trick" at all, this is simple elementary signal processing 101, used everywhere signals are present and processed. This teaches the signal to noise ratio (SNR) improves as the square root of averages for uncorrelated signals (noise) and correlated signals (image), and since the flare is correlated from image to image just like the desired image details, the flare isn't attenuated, however the uncorrelated noise does attenuate. The net result is an improvement of SNR by square root of N, where N is the number of images, and under the conditions stated about correlated and uncorrelated "signals".

So this "trick" does not produce an ISO 10 for ten ISO 100 image averages, but because of the square root relationship more resembles an ISO 32 effect.

Edit, However this averaging does improve the final rendering and also contributes to why focus stacking shows improved image quality, since stacking is similar in some respects to just signal averaging. The main difference being stacking tends to ignore areas of image blurriness after image alignment, whereas simple averaging does not align or ignore image areas.

Anyway, hope this helps.

Best,

[mawyatt]

Just for clarification we began doing chip images before the turn of the century. At that time almost all images were from thru the microscope, however we decided to venture "outside" the microscope and employ different techniques in the quest for better chip images. These images were created to support our chip design efforts and presentations, and almost all images were of chips we designed.

Over the years the techniques and equipment evolved to what we have now, so lots of experimenting with what works and what doesn't over those years. Our work related to chip imaging techniques and equipment hasn't progressed in the past 3 years since the development of the piezo stages for sub-micron work, and previous fully Automated Stack & Stitch system, and since we haven't designed a new chip since retiring. However, this may change soon with another SOTA chip develoement, if things go as planned and we'll be back to doing some high resolution Stack & Stitch imaging of a very large and complex new type chip. Time will tell.

Best,

[magic]

The die was 1,2mm x 1,2mm. The magnification should be around 19x. Diffraction limit... ...yes, seems to be the end of the line... 

Yes, I think you may be seeing the limits of f/4 aperture.

I took my own f/2 image of the Chinese "659" type RC4558 and scaled it to the same size as yours here.
Die width is 1150px and about 700~750µ in the real world per my measurement. Scale is 1.6px/µ.
My sources give airy disk diameter at f/4 as 5µ, which is 8 pixels on our images.
I applied radius 4 blur to half of the image and then some sharpening to approximate what a camera or raw processor could do.
The result looks similar.

We may also take a close look at the f/2 original (~2px/µ scale).
Apparently, lines about 2µ apart are resolved, but those closer to 1.5µ apart (and less) are not.
Given that expected blur diameter is 2.5µ or 5px, in-camera processing has clearly worked very hard on this image

[Noopy]

Thanks for all the input! 

I definitely will try some stacked lenses.
I have interesting couples in stock:
- 100-400mm 4,5-5,6 AND 10-22mm 3,5-4,5 => huge magnification but due to diffraction probably not ideal
- 100mm 2,8 AND 10-22mm 3,5-4,5 => better quality than today?
- 100-400mm 4,5-5,6 OR 100mm 2,8 AND 35mm 1,8 => I wasn´t happy with the 35mm 1,8. Perhaps stacked it gets better.
- I have a nice 24-70mm 2,8 perhaps I will try that too. Probably as second lens because 70mm seems to be a little short for the first lens.
- Would it make any sense to stack the MP-E 65mm 1-5x? No, probably not...

[mawyatt]

magic,

We're on travel now and didn't have time to respond to your post regarding exposure time, sorry. Just noted that the post has been removed?? Maybe I still have it on my home Mac since EEVblog hasn't been updated on that computer for the past couple days.

Anyway have an idea why it was removed, since it contained some mis-information regarding exposure. Exposure time for optical sensors is integration time and the sensor converts incident photons to free electrons based upon the conversion efficiency. This creates a system where the signal (image) increases linearly with exposure with some assumptions),  the general noise (shot noise) increases with square root of exposure, so the net effect is as mentioned an improvement in SNR (IQ) as square root of N where N is exposure ratio. No different the standard procedure for other systems and followed elementary signal processing.

However, you can't just expose forever since "other" undesirable sources contrive to corrupt your image, dark current and so on, so no free lunch!! This is why long exposure sensors are generally cooled, to reduce additional long exposure noise for one source causing long exposure image degradation.

Anyway, now you know this about the noise effects on optical sensors, and thanks for removing the post with the mis-information regarding exposure as it might be confusing. Sorry for the late reply.

Hope this helps.

Best,

[Noopy]

However, you can't just expose forever since "other" undesirable sources contrive to corrupt your image, dark current and so on, so no free lunch!! This is why long exposure sensors are generally cooled, to reduce additional long exposure noise for one source causing long exposure image degradation.

OT:
I have done some pretty interesting shots with my camera in the fridge: Longtime exposure at 20°C vs. 4°C and such things. 

[mawyatt]

Clever, hopefully you were able to keep the lens and sensor from fogging over👍

Best,

[magic]

I have deleted one of my posts which focused mainly on the topic of image- vs object-side aperture metrics but it ended up being poorly written and I figured probably no one cares anyway

The ISO comment was there.

Exposure time for optical sensors is integration time and the sensor converts incident photons to free electrons based upon the conversion efficiency. This creates a system where the signal (image) increases linearly with exposure with some assumptions),  the general noise (shot noise) increases with square root of exposure, so the net effect is as mentioned an improvement in SNR (IQ) as square root of N where N is exposure ratio. No different the standard procedure for other systems and followed elementary signal processing.

No disagreement.

However, you can't just expose forever since "other" undesirable sources contrive to corrupt your image, dark current and so on, so no free lunch!! This is why long exposure sensors are generally cooled, to reduce additional long exposure noise for one source causing long exposure image degradation.

Most importantly, they seem to have a physical cap on how much charge they can capture and that may be not far above the exposure required for full scale response at native ISO. As I found out by overexposing a CCD sensor by a stop and trying to read it out at half the recommended ADC gain

thanks for removing the post with the mis-information regarding exposure

I still don't see what was supposed to be wrong with my comments regarding ISO. I claimed and I still maintain that averaging 10 exposures at ISO100 captures the same light as one exposure at ISO10 (which may not be supported by the sensor) and hence is expected to yield the same noise reduction as 1 exposure at ISO10, which is a ~3 times reduction indeed.

You seem to assume that 3 times reduction in noise wrt. ISO100 is achieved at ISO32; I find this idea highly dubious.

[RoGeorge]

OT:
I have done some pretty interesting shots with my camera in the fridge: Longtime exposure at 20°C vs. 4°C and such things. 

That's interesting you could do that.

Here's an idea    :   Would it make sense to have a freezer with a transparent window/lens monture in it, and a photo-camera kept inside the freezer at all times (at -30...-20*C), for less noise?

Instead of freezing the batteries, the camera can be powered by wires and operated by USB, externally.  The hole window/lens monture can be reached from outside, but has a transparent separator, so to be possible to change objectives without fogging the interior of the camera.

I know astronomers are using commercial CCD sensors from former consumer DSLR, CCD sensors that are cooled with Peltier elements and read with custom electronics.  Poking a hole in a freezer should be easier when compared to that, and the camera can stay untouched.  Old models of DSLR body only are very cheap considering their sensor's quality, should be just right for such a freezer setup.   

[Noopy]

Clever, hopefully you were able to keep the lens and sensor from fogging over👍

I left the camera in the fridge while taking the pictures. Additionally I wrapped the camera in a towel so the cold parts don´t see too much humidity while the door is open. With the towel it takes longer to cool the sensor down after a picture but the fridge is cooling 24/7. 
If you are interested you can download the pictures here:
http://www.richis-lab.de/temp/noise.zip
Names are self-explaining.
Interesting how different the picture size is depending on the noise.
The real noise comes with 300s @ISO1600.

Here's an idea    :   Would it make sense to have a freezer with a transparent window/lens monture in it, and a photo-camera kept inside the freezer at all times (at -30...-20*C), for less noise?

At temperatures below -10°C I would fear that some circuits stop working as designed (capacitors, mechanics,...).
The problem (water) is the junction between cold and room temperature. As far as I know the "professional sensor cooler" heat this junction somehow.


BUT:
The noise is not really a problem. My pictures are taken with 1/100s - 1s @ ISO100 - ISO1000.
Sometimes you see some colored pixels in dark pictures. That is not really noise. For example I often have a lonely red pixel in the lower right corner. I have bonded with this pixel.  It would be no bigger problem to subtract these pixel. My camera can take a dark picture and do the subtraction automatically. But that takes twice the time. 

[mawyatt]

Interesting stuff!! Many of the older asto imagers from Santa Barba Instrument Group were based on Kodac chips that were cooled with TE coolers. Those chips were fabricated by IBM, not Kodac tho.

Best

[RoGeorge]

I think there are more choices for the sensors now, the one I found out about recently (by serendipity) is ICX453AQ, apparently the sensor from Nikon D70 DSLR.  The sensor can be desoldered from second hand cameras, or bought as a spare part for about $20-50.

There is also a small kit board to read the sensor:  http://astroccd.org/2016/10/cam86/

[mawyatt]

Had a D70 long ago, superb DSLR. Took a number of years before CMOS could approach CCD noise performance. Don’t think CMOS ever surpassed CCDs in noise performance tho.

Best,

[magic]

Weekly status update from project "Beyond Matchbox"

Attached picture was produced using 100% webcam components plus the usual smartphone mirror foil and stuff to hold it all together. Scale is 2px/µ and resolution of some 500lp/mm (or one "meaningful" pixel per micron) appears to be achieved, although there isn't really a regular pattern anywhere on the die that would enable confirming it with certainty. That's close to the customary theoretical maximum, which is 700~1200lp/mm (depending on color) for a lens of typical for webcams f/2 aperture. Producing a usable image at the absolute limits would require heroic postprocessing effort anyway; this one is only sharpened a bit.

The lens is the same one which was used last year, but optical performance is somewhat better than the sample I posted yesterday. This is because the old system involved lens stacking on a compact camera, which contributed a bit of diffraction in its own lens and had relatively low magnification, barely sufficient for the pixel pitch of the sensor. With only the webcam lens, magnification is easily increased and thanks to Hugin I don't need to worry about field of view either.

The die is not very pretty because it came from my pile of ICs that didn't make it to the final cast of the "fake opamps" post last year. It shows signs of slight overheating, which is changed color in some places and a lot of pale areas on the metal. However, the tortoise pattern which looks like it could be surface cracking actually is not. It's just residue left by dried acetone. I have serious problems with getting those things clean: washing them leaves patterns like that, wiping them leaves dust Not sure how to deal with it, maybe some photographic sensor cleaning kits or lens paper? Such stuff is supposed to be dust free, right?

Unfortunately, M42 rings are barely suitable for setting focus with the webcam lens. Their threads are somewhat loose and they wobble. I had less problem using the 10x objective with DCR-250, not sure why. Not sure what to replace them with, they really are quite convenient in every other regard.

Sometimes you see some colored pixels in dark pictures. That is not really noise. For example I often have a lonely red pixel in the lower right corner. I have bonded with this pixel.  It would be no bigger problem to subtract these pixel. My camera can take a dark picture and do the subtraction automatically. But that takes twice the time. 

Some cameras support permanent bad pixel correction, which takes a dark shot and saves the list of stuck pixels for automatic removal from future photos.

BTW, maybe you should sell that "Canon" stuff and mount a webcam lens on the DSLR? I'm not sure if that has been tried yet

[Noopy]

Weekly status update from project "Beyond Matchbox"

It´s amazing what you are able to achieve with your webcam parts! 

I have serious problems with getting those things clean: washing them leaves patterns like that, wiping them leaves dust Not sure how to deal with it, maybe some photographic sensor cleaning kits or lens paper? Such stuff is supposed to be dust free, right?

I clean loose dirt with a normal kleenex and a good amount of IPA. If there are some fibers left I use canned compressed air to blow them away. Works fine for me.

Sometimes you see some colored pixels in dark pictures. That is not really noise. For example I often have a lonely red pixel in the lower right corner. I have bonded with this pixel.  It would be no bigger problem to subtract these pixel. My camera can take a dark picture and do the subtraction automatically. But that takes twice the time. 

Some cameras support permanent bad pixel correction, which takes a dark shot and saves the list of stuck pixels for automatic removal from future photos.

Yes, should be possible too.

BTW, maybe you should sell that "Canon" stuff and mount a webcam lens on the DSLR? I'm not sure if that has been tried yet

 

[magic]

People have been using webcam, smartphone and similar lenses for macro photography for many years. I think they work well because in their normal use they need to support very high pixel densities on the sensor side. The webcam I converted has ~3µ pixels and low resolution (1280x1024) so if the lens limited it further, the problems would be quite visible and one couldn't hide them by scaling down since even full resolution doesn't really fill a modern screen.

Or maybe I just got lucky with mine If anyone wants to know, it's Esperanza EC105, which appears to be a Polish company that sticks their logo on random Chinese stuff, so it may be available under different names elsewhere. I think any of those big, "cannon style" webcams with similar lens and HD or better resolution is likely to work.

The sensors on those things aren't great, though. One reason I picked the M42 system was to have a path for potential upgrade to better cameras, up to a MILC.

Another missing piece is a system to automatically move the die under the camera, so that even large dice could be "scanned" with little effort. Hugin is working reliably so far, so stitching a 1000 frame mossaic would hopefully only be a matter of waiting an hour for the result.

[Noopy]

- 100-400mm 4,5-5,6 AND 10-22mm 3,5-4,5 => huge magnification but due to diffraction probably not ideal
- 100mm 2,8 AND 10-22mm 3,5-4,5 => better quality than today?
- 100-400mm 4,5-5,6 OR 100mm 2,8 AND 35mm 1,8 => I wasn´t happy with the 35mm 1,8. Perhaps stacked it gets better.
- I have a nice 24-70mm 2,8 perhaps I will try that too. Probably as second lens because 70mm seems to be a little short for the first lens.
- Would it make any sense to stack the MP-E 65mm 1-5x? No, probably not...

I think I was quite lucky with my 10-22mm retro. I didn't get much better picture quality with the other combinations.

100mm stacked with 10-22mm was slightly better than 10-22mm retro without the 100mm but not very much. Besides that this combination gives me just a magnification of 10x and it is less handy. 

The stack 400mm - 24mm was huuuge.  It looked very funny and expensive. Sorry, no picture. Imagine a "Canon 100-400mm 4.5-5.6 l" zoomed to the maximum stacked with a "Canon 24-70 2.8 I" zoomed to the maximum (invers zoom). 
A magnification of 17x seemed to be good but the image quality was bad.

[magic]

10x magnification doesn't seem too bad. It's ~6px per minimum resolvable point distance at 550nm if I'm not mistaken*, probably near the reasonable minimum for comfortably avoiding sensor bottleneck.

The f/1.8 would be nice but even more magnification is needed to really take advantage of it on APS-C. It's not gonna be easy with f=35mm. That's a setup that would look PRO

*calculations specific to EOS 60D with f/4 lens, of course

[Noopy]

10x magnification is quite ok but I personally need more! 

[magic]

More magnification or more resolution?

Here's an interesting discovery: if 10x is needed to easily recover full resolution of f/4 on a typical APS-C sensor, then a typical 10x0.25 microscope objective (f/2 equivalent) may slightly outresolve such sensor. Same with 5x0.12, 20x0.4 and so on.

I'm sure it makes bug shooters happy, but it seems not ideal for high magnification closeups and getting the most fine detail out of the optics. Testing would be needed to establish the exact limits, including Bayer and anti-alias effects. I suppose blue is the worst case, assuming that you want to take advantage of the short wavelength and capture more detail in blue than in green/red. Red is also affected by 25% pixel density but at least it's blurred more so there is less detail to recover. Some analysis of the problem has been made at PM and linked by mawyatt last year, but it seems mostly theoretical.

Full frame is even worse. You have to buy the highest end to get a mere 45Mpx and the same 4.3µ pixel size.

Nikon 1?

edit
Alternative kludges are possible. One could try an infinity objective with +2D tube lens on 50cm extension (or 500mm telephoto) to double the magnification

These guys don't give a damn and just mount infinity objectives on a super long extension tube and focus them at a finite (but long) distance. However, I am not entirely sure if they know what they are doing and whether their techniques are optimal. The other day, they also identified a bipolar IC which could be some LM386 as a MEMS strain gauge

[Noopy]

10x should almost be enough for my 60D. But with a little more I can be sure that the magnification factor isn't the bottleneck.


Pictures with 100mm<=>10mm(r) show a similar "quality" than pictures with "distance"<=>10mm(r).

The resolution of the 100mm<=>10mm(r) seems to be a little better

BUT I have problems with the overall sharpness of the pictures. I have to do some more tests to check if it's a alignment problem or a lens problem or a handling problem.
The lower magnification makes it more difficult to find the focus.

[magic]

If one or two opposite sides are out of focus - that's alignment.
If all sides are soft but the center is sharp - that's the lenses not working very well together.

I have seen a mix of both effects on my PowerShot with webcam lens. Mostly the former, because small size and a rather amateur construction made it very difficult to align everything.

I used AF to fine-tune focus after it was roughly adjusted by tweaking distance.

[magic]

If anyone ever wondered what a SO8 package looks like inside, I found some old pictures from my experiments with blueskull's acid + saltpeter formula.

I'm not sure if I was supposed to make a saturated solution of saltpeter in acid or mix a saturated solution of saltpeter in water with acid; maybe the latter would have eliminated crystallization of salts on the chip and sped up the process, but dunno about corrosion OTOH.

Anyway, the junk covering the die on the first picture is just some salt from the process, nothing that came from the factory
The next two pictures were taken after cleanup and some mechanical separation.

[Noopy]

I have bought Armour Etch! 
I wanted to strip the metal layers of a ADR1000 and because of that I had to remove the SiO2. Unfortunately you need HF to remove SiO2 and HF is scary as shit. It´s corrosive, it´s toxic, in lower concentrations you don´t recognise it on your skin but it travels very fast in deeper regions of your body. Some gloves are no protection. => You don´t want to mess around with HF.
Armour Etch is a paste with ammonium bifluoride that generates HF. That´s also dangerous but you don´t have to handle large quantities of HF.






Here you see the die before etching. There are two metal layers.




Some Armour Etch on the die. 




After 6 minutes of HF you can spot some changes at the sides of the upper metal lanes.




5 minutes of 18% HCL dissolves the upper metal layer completely and removes a lot of the lower metal layer too. The SiO2 on top of the lower metal layer is still intact but the HCL takes it´s way through the SiO2 corridors. In the inner circle there is a small part of the metal left.
Where the upper metal layer is dissolved you can see the leftovers of the SiO2. Where the lower metal layer is dissolved you can see the SiO2 corridors.




6 more minutes of HF dissolves the SiO2 on top of the lower metal layer / the empty corridors. The metal leftover is still intact because the HF doesn´t dissolve the metal (Al).
The dark particles look like leftovers of the SiO2.




5 more minutes of HCL remove the metal completely.
There is still a non transparent yellowish layer where the lower metal was. HCL can´t dissolve the yellow layer. I assume that is a SiAl-complex.








More HF and you get rid of the yellow layer and the dark particles.
Unfortunately on this die the layers don´t show the nice colors we have seen with other dies. Nevertheless you can spot the different areas and conclude how the areas interact with each other.
Now you can do some analysis: https://www.eevblog.com/forum/metrology/lowest-drift-lowest-noise-voltage-reference/msg3733504/#msg3733504








before and after the treatment 


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

 

[magic]

Looks good.

At one point I tried NaOH solution (less toxic and attacks both glass and aluminium) but it was unusably slow. Boiling should help because everybody says that hot NaOH eats laboratory glassware, but I didn't try.
NaOH is fairly effective against aluminium spills from overheating (a matter of minutes).

Unfortunately on this die the layers don´t show the nice colors we have seen with other dies.

Yep, because it was the glass

There are some tricks to stain doped areas, but it looks like nasty stuff.
https://siliconpr0n.org/wiki/doku.php?id=delayer:wet#staining

[Noopy]

Unfortunately on this die the layers don´t show the nice colors we have seen with other dies.

Yep, because it was the glass

I had the hope there is still some ... ... resonance ... ... refraction ... ... whatever. 

[RoGeorge]

AFAIK the color is given by the nanostructures being (multiple of) half wavelength, and the incident and reflected wave interfering with each other, somewhere in the middle of this video it is explained with a few more details:



Assuming the refraction index is the same, and the metal or semiconductor layers are glossy (no nano "bumps" on the surface to turn them into a metamaterial), I guess it should be possible to calculate the glass thickness by measuring the reflected spectrum.

[Noopy]

I agree with you. 
Nevertheless I had the hope... 

[XantheFIN]

If anyone ever wondered what a SO8 package looks like inside, I found some old pictures from my experiments with blueskull's acid + saltpeter formula.

I'm not sure if I was supposed to make a saturated solution of saltpeter in acid or mix a saturated solution of saltpeter in water with acid; maybe the latter would have eliminated crystallization of salts on the chip and sped up the process, but dunno about corrosion OTOH.

Anyway, the junk covering the die on the first picture is just some salt from the process, nothing that came from the factory
The next two pictures were taken after cleanup and some mechanical separation.

Hi what is this formula you used in process? "blueskull's acid + saltpeter formula"?

[magic]

You can find it quoted here, the guy who posted it quit the forum and nuked most of his content.
https://www.eevblog.com/forum/projects/decapping-and-chip-documentation-howto/msg2875340/#msg2875340

It's a messy process and I'm not sure if I have done it the same way blueskull did or differently. At any rate:
- I now know that nitric acid evaporates from mixtures with concentrated H₂SO₄ quickly starting at ~80°C (this technique is used for distilling anhydrous HNO₃) so the whole process seems somewhat inefficient/wasteful
- I needed high temperature for anything to happen, it fumed quite a lot, I recirculated some vapor by putting a cap on top so it condensed (but don't pressurize the test tube)
- it took a few hours
- those HNO₃ and NO₂ fumes are toxic, just so you know
- maybe I was supposed to add water to the reaction (use a solution of KNO₃ instead of dropping dry salt into H₂SO₄) but not sure
- too much water means corrosion of bonding pads and/or bond wires

Generally, you should only bother with that if you want to preserve everything intact, including wires and pads.
For just a bare die, it's much easier to use:
- ordinary 65% nitric acid ~120°C in 5ml beaker with watch glass cover (don't sniff the fumes)
- 2ml sulphuric acid at 200~250°C - less toxic, more danger of burns, it bubbles, eye protection advised
- electric heater 400~600°C (needs a few tries to get it right)
- fire (Russian roulette)
- boiling rosin at who know what °C - some people say it works, I never succeeded

This source claims that pure concentrated sulphuric acid may work without corrosion, but it didn't work like that for me. Maybe I haven't used enough and the concentration dropped as water was produced in reactions.
http://www.stockly.com/forums/showthread.php?p=2239

See also
http://siliconpr0n.org/wiki/doku.php?id=decap:epoxy_acid

Have fun

[XantheFIN]

Thanks for answer  Sorry i did read full topic but missed the user name.

I can't have fun as hard to get any good stuff (sulfuric acid, hydroflouric acid, Nitric Acid) after all wanna be muham**** manufacturing explosions so very limited in EU. 

[Noopy]

Heat works quite well for me... 
1-5% failures is ok since I don´t have to work with nasty chemicals and I can do 25 parts per hour. 

[XantheFIN]

With only partial decapping and bond wires survives?...  I am looking to save chip in as whole and keep operational but just make it visible physically.

[Noopy]

Well no...

[magic]

Heh, I missed your flag. Yes, sulphuric acid has been banned by EUnuchs literally this year
I happen to have a "friendly" business which stocks some good stuff so I can occasionally play with it.

If you can't find anything like that or register your own company just for this purpose you are going to have problems.

For the record, it is possible to produce nitric acid from atmospheric nitrogen by electric discharge and concentrate it by distillation (even to fuming concentration with magnesium nitrate, if that thing isn't banned too). Information is easy to find, but both steps are PITA for a complete amateur and need some glassware. Efficiency isn't great, either.

You could try that rosin thing as alternative. This guy's results look somewhat encouraging:
https://www.eevblog.com/forum/projects/whats-inside-the-cheapest-and-fakest-jellybean-opamps/msg3545907/#msg3545907
My advice: use identical glass bottle and the same gas burner and the same colophony because I have tried similar things a few times and never got good results.

[Noopy]

I have an update for the Howto (https://www.richis-lab.de/Howto.htm). I will show you how I do panorama pictures.




To get good pictures for a panorama you have to move the die in X and Y while keeping it in the focus plane of the camera. The best way to do this is to use a micro positioning tool. I have such a tool which allows moving X and Y and tilting around the X- and Y-axis. The tilting is important to get the whole die into the focus plane.




I have a metal block on which I can screw the micro positioning tool. With the help of two locking screws I can move the tool up and down.

On the other side of the tool there is a dish with some double sided adhesive tape on which I put the die.

The horizontal arrangement is favourable because the camera does a good job in this position.




If you have a die in a package or on a ceramic plate or in the middle of a wafer you can get problems with shadowing. In this case the lower parts of the die are darker than the upper part.




To compensate the shadowing you can use the automatic exposure of your camera but if the die is very inhomogeneous that can cause other problems. Pictures with brighter areas get too dark and darker areas get too bright. It is better to change the exposure time manually.




A lot of people use Hugin. In my view Hugin demands a lot of picture adjustment, arrangement and marking by the user.
I use Helicon Focus.




If you are able to move the die quite accurate Helicon Focus is probably able to merge the panorama by itself. If the alignment is not perfect you have to do some work by yourself.




Most of the time moving the pictures to the right place is no bigger problem.






You can adjust in which area Helicon Focus is looking for the right spot to merge pictures. The initial value is 0,1%. For my arrangement that puts out bad results too often. 0,3% is a good value for me.






Edge smoothing 0% is good to find bad picture alignments.






Edge smoothing 100% doesn´t give you the best pictures. 100% moves the edges of the other picture into the foreground.






50% edge smoothing is best for a nice picture.




Done!




Rotating and cutting edges, contrast adjustment and ready to go! 


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

 

[mawyatt]

Nice 

Lenses for use in panorama type macros need to have very little distortion for chip imaging, whereas for typical insect and flowers this doesn't matter much since the "edges" are easier to blend that chips orthogonal features. When you get into depth type panoramas for that 3D effect, then the lens system needs to be somewhat telecentric as well as have little distortion. As usual no free lunch, and driving up cost and complexity!!

Anyway, you've done the community a great service with all this wonderful chip images! Few appreciate with what's involved in doing so, know we do!!

So hat's off to you 

BTW haven't seen a dual axis tilt stage like that, the manual ones we have are based upon 40 or 60mm square, this looks smaller.

Best

[Noopy]

It seems like I was very lucky with my Canon EF-S 10-22mm f/3,5-4,5 arrangement. 

The next few days I will update the HowTo with some focus stacking experience. Stay tuned! 

Thanks for the compliments!
I´m always happy to get new high quality pictures of things I wasn´t able to see some years ago. I like the mixture of old and new, sophisticated and simple, expensive and cheap, big and small, famous and unknown semiconductors. 

I know that´s a really small micro positioning stage. It was a gift. I have no idea where it came from. For me that´s the perfect tool. 

Greetings!

[Noopy]

A small update regarding the focus stacking.
I have built a new computer and now I get pictures a lot faster! 




Helicon Focus contains a benchmark. My old i7-4770K working at 4,3GHz on all cores with 16GB of DDR3 handles just 104 Megapixel per second.




My new i9-12900K rises the conversion speed up to 423 Megapixel per second.




My GeForce RTX 3070 Ti gives the system a real boost rising the conversion speed to 1.212 Megapixel per second. 




With my old system this picture took me more than one minute to render. My new system does the job in 12 seconds.


https://www.richis-lab.de/Howto_Focus.htm#Upgrade

 


...


I just realised that I have put the initial focus stacking post into the MEMS topic. 
If you are just reading this topic please jump to:

https://www.eevblog.com/forum/projects/mems-nice-die-pictures/msg4009918/#msg4009918

 

[magic]

Well, I suppose you could "edit" the post, copy its contents here and delete the old one.

BTW, I have a question about those ICs after SiO₂ removal.
Are the isolation areas depressed below the level of transistors or are they raised above?
Supposedly they should be depressed because oxidation, oxide stripping and re-oxidation consumes silicon from the surface, but it's hard to tell form the pictures.
Same goes for the "shadows" of buried diffusions.

[Noopy]

Well, I suppose you could "edit" the post, copy its contents here and delete the old one.

But then the "first" focus post would be behind the second one. I think I will leave it like it is right now.

BTW, I have a question about those ICs after SiO₂ removal.
Are the isolation areas depressed below the level of transistors or are they raised above?
Supposedly they should be depressed because oxidation, oxide stripping and re-oxidation consumes silicon from the surface, but it's hard to tell form the pictures.
Same goes for the "shadows" of buried diffusions.

That´s an interesting point. Depending on the light inclination the areas look like being depressed or raised. It´s absolutely puzzling how the optical effect changes with the light.

But I´m not sure about your conclusion. Oxidation consumes silicon, that´s right. But silicon oxide occupies more room than bare silicon, doesn´t it?

[magic]

If an image is confusing, you tend to assume that the light is coming from above or from the opposite side of you (rather than from you). That's why 3D user interface elements on computers are drawn as illuminated from top or top left.

Looking at the ADR1000 pictures you posted here, I see raised isolation at first but I can "force" myself to see it depressed with a bit of concentration. If I simply rotate the same image 180°, it works the opposite way.

I made a small experiment. This die is illuminated with a single LED from top left at a shallow angle. Another die is placed on top to confirm the direction of light. It looks to me like isolations, metal contact windows and likely P/N+ diffusions are all depressed. There are also four test structures ("B" letters) at the bottom. The outer ones (including metal) seem raised, the inner two appear depressed.

The reason for such structure is because at first silicon is consumed to make oxide, then the oxide is dissolved to expose silicon to doping, then more silicon is consumed to make new oxide for further masks. Thickness of silicon keeps decreasing with each processing step applied to given area and the oxide also isn't as thick as on areas where no processing has been done.

That's how I see it.

[Noopy]

That´s a good explanation for the look of the pictures. 

And you are right with the height of the areas. Of course every oxide layer that is removed due to processing deepens the area. 





Here a quick comparison how different the structures can look like.

[magic]

Perhaps nothing new for regular readers of this thread, but I will post it for completeness

https://zeptobars.com/en/read/National-LH0042CH-hybrid-opamp-FET-SEM
https://zeptobars.com/en/read/SEM-deep-dive-1
https://zeptobars.com/en/read/RCA-CA3019-BJT-transistor-array

These are SEM micrographs of ICs at Zeptobars which show what the 3D structure of the surface looks like.

[magic]

co lo pho ny


Well, yes. I found an article written by the CCC guys who popularized this technique a few years ago. Perhaps the most important part:

In "Präparationstechniken für die Fehleranalyse an integrierten Halbleiterschaltungen" von Friedrich Beck aus dem Jahr 1988 habe ich einen Hinweis auf die Entkapselung mittels Kolophonium gefunden. Kolophonium enthält Abietinsäure (etwa 40%). Beck schreibt: "Zum Öffnen wird das Bauteil im Drahtkörbchen in das auf 320 - 360 °C erwärmte Kolophonium getaucht, bis der Chip völlig freiliegt (5 - 10 Minuten); anschließend läßt sich mit trockenem Aceton das anhaftende Kolophonium entfernen." (S. 19, 20) Beck schreibt ferner, dass die Kolophoniumrückstände den Abzug versiffen und das schwer zu entfernen ist. Dies sei der Grund, warum das Verfahren selten angewandt wird.

As far as I understand, they say that the whole idea came from an old German book about semiconductor failure analysis, and according to the book, temperature of 320~360°C is required and it must be maintained for 5~10 minutes because that's how long the reaction takes. Also something about this technique rarely being used in practice because colophony messed up fume hoods

The rest of the article describes author's own experiments, and they were less happy because the process worked, but took an hour or two per chip. However, the author says nothing about any attempts at precise temperature control, he just used a spirit burner.


Well, I thought that I will be clever and perhaps manage to achieve reasonable speed by regulating temperature precisely in the recommended range. I loaded my old thimble cooker with colophony and two different SO8 chips, connected it to a variable DC power supply, shoved a thermocouple in it and adjusted power to reach and maintain progressively increasing temperatures above 300°C. (I tried 270°C for two hours previously, so it surely made no sense to bother with less than 300°C).

15 minutes at 320~330°C. There were small bubbles and some smoke coming, nothing terrible.
15 minutes at 330~340°C. Similar to above. Some colophony lost, but not a lot.
Then I refilled and tried to hit 350~360°C, but it ended up more like 360~380°C. Now, bubbling and smoking got much worse and colophony was disappearing faster.
After another 10~15 minutes, there was almost nothing left so I turned it off.


And the outcome? Almost nothing. Maybe the chips are a little smaller than initially, but surely not dissolved fully or even significantly. I have no patience to wait a few hours and see what happens.

This guy says it should be a few minutes. At this point I don't understand what's going on. At least I know that it isn't about temperature - less than 300°C is completely ineffective, and 400°C or more is lots of smoke and colophony gone.

Maybe there is something wrong with my colophony. I use some "activated" colophony for soldering. I will look if I can find unprocessed natural colophony somewhere. Other than that, no idea

[Noopy]

Thanks for trying and sharing the results!

It seems like there are a lot of urban legends around.

Sometime I have to test DMSO again with more temperature...

[T3sl4co1l]

Possible the method worked on older resins but not today's. Like the various colored ICs of the 70s, or blob-tops and other somewhat gummy / not-hard transistor bodies.

That doesn't explain how the other guy had success, unless the SOIC(s) were older or just different.  Possible there are still differences today, such as green vs. halogenated resins, etc.

Tim

[Noopy]

I bought a new tool from Thorlabs.
With these pliers you can cut through metal cans (not to big, TO-3 is not possible). A screw makes you stop before you are to deep in the package (a little longer screw would be better).




Nice! 
...and I´m a lot faster now. 


https://www.richis-lab.de/Howto_Decap_Metall.htm#Update

 

[magic]

LOL, never realized that somebody makes this kind of can openers

A screw makes you stop before you are to deep in the package (a little longer screw would be better).

I'm sure it's some off the shelf screw using some standard thread.

If you are lucky it's just M3 or M4, if you are unlucky it some weird-ass "imperial" thing but maybe somebody could ID it if you take outer diameter and thread pitch measurements.

[Noopy]

A screw makes you stop before you are to deep in the package (a little longer screw would be better).

I'm sure it's some off the shelf screw using some standard thread.

If you are lucky it's just M3 or M4, if you are unlucky it some weird-ass "imperial" thing but maybe somebody could ID it if you take outer diameter and thread pitch measurements.

Yeah, should be no bigger problem. 

[Noopy]

Did I say you can´t cut open bigger packages like TO3?
Haha, you can unscrew the two rollers and screw them into the lower holes! Now you can cut open TO3 packages! 


In addition I have added a small advices to the drying agent: I have heard in old devices sometimes barium oxide was used. Barium oxide is poisonous so you better don´t cut transistors open on top of your lunch.


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

 

[Noopy]

For all of you doing "dirty decapping" with heat and are struggling with sticky remainings:

I did some experiments with a fiberglass brush. Sounds rude but did the job quite well. 
It seems that the passivation doesn't get scratched. Bigger metal areas without passivation get scratched but most ICs doesn't have that. I'm talking about metal layers like these:




Next time I will try to remove polyimide (after some mild heat treatment that usually is not enough to decompose it).

[RoGeorge]

Read today about this unusual decapping method that uses high voltage/plasma from a Tesla coil, instead of chemicals.  Linking it here just for the docs:
https://hackaday.com/2023/07/12/no-acid-open-ics-with-a-tesla-coil/
https://hackaday.io/project/191416-integrated-circuit-decapping-without-acid

[Noopy]

I have updated the optical section of my HowTo:






First of all the Canon 90D with the Canon EF 100mm f/2,8L Macro. That´s a surprising performance for a "normal" Macro objective. You can identify the 10µm lines! 

The Canon 90D has an APS-C sensor and therefore multiplies by a factor of 1,6. A magnification factor of 1,6 seems very small, but you have to take the high resolution of the digital camera into account. The Canon 90D offers a resolution of 32,5 megapixels. If you are satisfied with 2 megapixels, you already achieve a magnification factor of 26:1. A high pixel density is therefore an advantage here.




We will take a closer look at this area...




Taking pictures of these small things don´t close the aperture. You need magnification and resolution capacity. Closing the aperture reduces the resolution capacity and your magnification gets useless.




Distance rings can increase the magnification but the pictures is just a little better. With more distance it gets worse.




The Canon MP-E 65mm f/2,8 1-5x macro is a magnifying lens that provides image scales between 1:1 and 5:1. On the Canon 90D, this corresponds to an image scale of up to 8:1. If you are satisfied with a 2 megapixel section, this is a magnification factor of 130:1.

One disadvantage of the magnifying lens is the lack of a focus setting. You therefore have to move the camera or the object to focus the image. This is not an easy task with such high magnification factors.




Compared to the Canon EF 100mm f/2,8L Macro, the image quality of the magnifying lens is significantly better.




Another option for achieving high magnification factors is the use of so-called retro adapters. This allows the lens to be attached to the camera with the lens normally on the outside. The adapters required for this are available for various camera types and filter threads. The camera can then no longer adjust the aperture and focussing. However, the aperture setting is not usually required and focussing is done manually anyway.

Not every lens is suitable for this application. A small focal length produces a high magnification factor in this configuration. The best results were achieved with the Canon EF-S 10-22mm f/3,5-4,5 shown here. The Canon EF-S 18-135mm f/3,5-5,6 kit lens is also well suited, but this only has a sluggish micromotor focus adjustment, not an ultrasonic motor. The otherwise almost flawless Canon EF 24-70mm f/2,8L produces surprisingly poor images. Looking at the specifications, the Canon EF 28MM f/1,8 USM also seems very suitable. It has a small focal length for a large magnification factor, it has a large aperture, which promises a high resolving power and ultrasonic focussing. In fact, this lens also delivers noticeably poorer results than the Canon EF-S 10-22mm f/3,5-4,5.




This image shows just how important a small focal length is. The 1:1 image section was taken at the 22mm setting of the Canon EF-S 10-22mm f/3,5-4,5. The limiting factor here is the resolution, but the magnification factor does not offer any major reserves too.




With the 10mm setting, I already had to reduce the image size slightly. At first glance, you might think that the image has deteriorated somewhat, but in fact the opposite is the case. In the top left-hand corner there is a test structure with smaller and larger elements. In the case of the larger elements, you can already see the contacts to the metal layer. Up to this point, however, the magnifying lens still offers better results.




If the retro system is supplemented with distance rings, the magnification factor increases. The images above already show that the resolution is the limiting factor. At the same time, the spacer rings reduce the amount of light reaching the sensor. However, this measure still has a positive effect on the image quality. This can be explained by the fact that lenses are optimised for certain minimum distances. With the Canon EF-S 10-22mm f/3,5-4,5, the close-up limit is 24cm. This is the minimum distance that must be maintained between the object and the camera sensor. Most cameras have a symbol on them that indicates the exact location at which the sensor is located. The distance between the object and the lens is only a few centimetres. With the distance rings, the total length of the optical path is approximately in the range of the close-up limit.




The image quality is significantly better with the 65mm distance rings.






With a second set of spacer rings between camera and lens, the distance increases to 130mm and the image quality improves slightly. The magnification factor increases further.

A third set of spacer rings, which increases the distance to 195mm, does not improve the image quality any further.




With the Canon EF-S 10-22mm f/3,5-4,5 and 130mm distance rings, one millimetre is imaged with 6.030 pixels. This means that there are six pixels in the distance of one micrometre. The sensor of the Canon 90D is 22,3mm wide and resolves this width with 6.960 pixels. That is 312 pixels per millimetre. This results in a magnification factor of 19,3:1. In relation to a 2 megapixel image section, the factor is 314:1. Here you can see a 1:1 image section. Depending on which structures are involved, elements as small as 1 µm can be recognised.




This 1:1 image section shows the lines on the scale with their distance of 10µm.




If you know the magnification factor of the structure, you can carry out measurements in the images. The freeware ImageJ is very well suited for this.

If the image is focussed with the focus ring of the lens, the magnification factor also changes. The focus setting is not saved, resulting in an additional measurement error when calculating the actual size from the number of pixels. In the case of the Canon EF-S 10-22mm f/3.5-4.5, however, the additional measurement error is a reasonable 1%.




Here you can see the image sections that result from the different configurations. A large magnification factor is desirable in order to make small details visible. At the same time, however, the image section becomes smaller and smaller. You then have to either create panoramic images or take individual detailed images with the appropriate magnification factors.

The high resolution of digital cameras is a great advantage here, as a tolerably large image section remains available even with large magnification factors.




 

Instead of the spacer rings, you can install a lens in a normal configuration between the camera and the retro lens. This requires an adapter with two matching filter threads. The additional lens adapts the light path of the retro lens to the camera. The magnification factor then corresponds to the ratio of the focal lengths. Here, the Canon EF 100-400mm f/4,5-5,6L IS USM with the Canon EF 24-70mm f/2,8L produces a magnification factor of 16,6:1, in relation to 2 megapixels this would be a magnification factor of 271:1.

Configurations with different lenses were tested. No combination delivers good image quality. The resolution is poorer, especially in the peripheral areas and chromatic aberration is clearly visible.


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

 

[magic]

What about your microscope images, though? You can't hide it from me

Taking pictures of these small things don´t close the aperture. You need magnification and resolution capacity. Closing the aperture reduces the resolution capacity and your magnification gets useless.

Actually, your f/2.8 sample is not much sharper than f/8, so it's limited by aberrations rather than diffraction.
It is possible that stopping down slightly (maybe f/3.5 or f/4) would bring small improvement. This appears to be common in photographic lenses.

If the retro system is supplemented with distance rings, the magnification factor increases. The images above already show that the resolution is the limiting factor. At the same time, the spacer rings reduce the amount of light reaching the sensor. However, this measure still has a positive effect on the image quality. This can be explained by the fact that lenses are optimised for certain minimum distances. With the Canon EF-S 10-22mm f/3,5-4,5, the close-up limit is 24cm. This is the minimum distance that must be maintained between the object and the camera sensor.

It may be possible to reduce minimum focus distance with a closeup lens in order to reduce magnification and gain field of view. 6 pixels per micron is overkill.

[Noopy]

What about your microscope images, though? You can't hide it from me

  You are right. Unfortunately ther is not much to show. I got a few pictures from other people, I payed for a few pictures and somtimes I can put my hands on a real microscope.

But I should add an comment that not every picture is taken with the DSLR. I don´t want to make poeple sad trying the DSLR way.

Taking pictures of these small things don´t close the aperture. You need magnification and resolution capacity. Closing the aperture reduces the resolution capacity and your magnification gets useless.

Actually, your f/2.8 sample is not much sharper than f/8, so it's limited by aberrations rather than diffraction.
It is possible that stopping down slightly (maybe f/3.5 or f/4) would bring small improvement. This appears to be common in photographic lenses.

I agree with you, the sharpness is very similar but the f/8 is a little worse I would say. I once tried smaller steps but didn´t find a configuration that show better quality pictures. The real increase in quality comes with the distance between camera and lens.

If the retro system is supplemented with distance rings, the magnification factor increases. The images above already show that the resolution is the limiting factor. At the same time, the spacer rings reduce the amount of light reaching the sensor. However, this measure still has a positive effect on the image quality. This can be explained by the fact that lenses are optimised for certain minimum distances. With the Canon EF-S 10-22mm f/3,5-4,5, the close-up limit is 24cm. This is the minimum distance that must be maintained between the object and the camera sensor.

It may be possible to reduce minimum focus distance with a closeup lens in order to reduce magnification and gain field of view. 6 pixels per micron is overkill.

Here I agree with you too. But I once tried a closeup lens and wasn´t happy with the pictures. Perhaps the quality of the lens wasn´t good enough. 
6 pixels per micron is more than we need but not way to much. As I said sometimes I can recognise 1µm structures. (Unfortunately I have no µm scale to double check.)

[Noopy]

An update to the light topic. For a lot of you guys there is not much new information but it was necessary to tune some things and add some pictures.




With very high magnification factors, the mechanical structure of the optical system must be as stable as possible. The slightest movements or even vibrations can lead to blurred images. There are tripods that fix the camera with its lens vertically. The object is then positioned underneath. These tripods are also available motorised so that they can be used for focusing. An alternative is to align the optical axis horizontally. In this case, the camera rests on an ESD mat, the surface of which provides a certain degree of adhesion and at the same time absorbs vibrations.

The camera is triggered with a remote control so that the camera is not moved. Mirrorless cameras are an advantage at this point. The movement of the mirror already generates enough vibrations that the image quality sometimes deteriorates. With SLR cameras, you can use the mirror lock-up function. The camera first moves the mirror and exposes the sensor only after a short pause. The Canon 90D offers a particularly advantageous function in this respect. The electronic shutter makes it possible to take pictures without activating the mechanical shutter, so that nothing moves in the camera at all.

At high magnification factors, the area that is displayed in focus is very narrow. The system must be focussed accordingly. Once you have roughly positioned the object to be imaged, focussing is finally carried out using the focus ring. An ultrasonic focussing system is advantageous here, as its focus ring is very smooth-running. The image quality is monitored via the camera's monitor, which is usually sufficient. However, the camera could of course also be connected to an external monitor.

The object is fixed and positioned using a classic third hand and a more modern version of this aid. A micropositioning system would enable easier and more precise positioning, but is less flexible.The third hand allows the object to be tilted quickly. It can also be used to quickly switch from individual dies to larger packages. At high magnification factors, it is extremely important to keep the arm for aligning the object as short as possible. The length seen in the image on the right increases the effect of small vibrations.

As you can see here, the distance between the lens and the object is relatively large. This makes it easier to document elements with taller structures in the immediate vicinity.




One crocodile clip of the third hand is used for larger elements. The other crocodile clip is wrapped with double-sided adhesive tape. This makes it easy to attach and detach the elements.The third hand is very stable, but can still be moved easily if the picture section is not the right one.




If you want to create panoramic images, you need an XY micromanipulator. This is the only way to take series of images of sufficient quality for larger panoramas. The micromanipulator shown here is very slim, which makes it easier to bring in light from behind the object. There is a round surface on the front which is covered with double-sided adhesive tape. In addition to X and Y displacement, the micromanipulator also allows swivelling around the X and Y axes. Panning is very important in order to bring the object completely into the focal plane.

The coarse alignment in the X and Z directions can be achieved by moving the micromanipulator. A small holder was constructed for the coarse alignment in the Y direction. Two grub screws attach a metal block to a threaded rod. The micromanipulator has a union nut at the rear end with which it is screwed to the metal block.




Different light sources and lighting angles create very different optical effects. This can be used to emphasise certain structures.




The surface of integrated circuits appears black. The metal layer consists of aluminium or copper and shows the typical colours of a metal surface. The silicon oxide layer on top, which protects the integrated circuit, is transparent. It is almost impossible to recognise details with such an image. Apart from the metal layer, only a few edges are faintly visible. The edges are the result of unevenness that occurs when the individual elements of the integrated circuit are integrated.

If coaxial lighting is used, the various elements integrated in the silicon are clearly visible. For an analysis, it is important to understand that the areas do not actually bear these colours. The colours are caused by resonances in the thin silicon oxide layer on the surface of the integrated circuit. When building up the different areas, the silicon oxide layer is repeatedly opened locally and then a new layer is applied over the entire surface. This results in silicon oxide layers of different thicknesses on the various areas. Destructive and constructive interferences occur for some wavelengths and different colours can be observed accordingly.

When interpreting the colours, you must always bear in mind that these are not real colours, but light resonances.
This means that different colours can occur for the same component with different types of illumination. As the colours depend on the thickness of the silicon oxide layer, it can happen that different areas show the same colours. Bevels also form at the edges between two areas, which can show different colours even though there is no third area in between.

In principle, any light source can be used. However, a fluorescent tube produced the most colours, which is very helpful when analysing circuits.




Microscopes that offer coaxial light couple the light for illuminating of the object with a semi-transparent mirror into the optical axis of the objective. This is difficult to realise with the setup described here. However, it has been shown that a similar effect occurs when the light source is placed behind the object.

If the light source is fixed directly behind the object, the image quality deteriorates. Placing it a little way off the optical axis proves to be optimal. The die is positioned vertically on the optical axis of the lens. It appears that some of the light is reflected by the lens and then acts like coaxial light.




Interestingly, the same effect occurs if you position the die at an angle and shine the light on the object from a similar angle. Due to the tilted position, focus stacking is absolutely necessary here.




It can be seen that the tilted images (right) reveal a little more details.




The images taken at an angle have another advantage. They show the surface structure of the IC. However, you have to be careful when interpreting the structures. The image on the left was taken at an angle and with illumination behind the IC. Even if it looks partially different in this example, the image quality is usually somewhat poorer in this configuration. However, the surface structure can be seen as it actually is on the chip. The transistors are located in recesses.

The above arrangement usually results in the centre image. At first glance, the surface structure appears inverted. The transistors now appear to be located on sockets. How this effect arises is unclear. If you want to understand and interpret the surface structure, it helps to assume that the light source is located below the image. In this way, the shadows correspond to the actual geometries. This becomes clearer if you rotate the image by 180° (right). Here the transistors appear to be located in recesses, as is the case in reality.




If you are documenting higher and more complex structures, good light distribution is very important. The fluorescent tube with its reflector is significantly larger than the objects to be depicted, but it still produces disturbing reflections and shadows. A folded sheet of paper is a simple remedy. It reduces the light intensity more than a diffuser, which is designed for this purpose. However, this is not a major disadvantage here because the exposure time is not critical.


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

 

[DavidAlfa]

Image this!
https://www.servethehome.com/stmicroelectronics-makes-a-18k-big-sky-sensor-so-large-only-four-on-a-300mm-wafer/

[Noopy]

As soon as I get one of these I promise to take some pictures. 

[magic]

Once you have roughly positioned the object to be imaged, focussing is finally carried out using the focus ring.

This wouldn't work with lots of mirrorless system lenses as they focus by means of motors moving lens groups inside.
Even manual focusing requires power (and perhaps data connection with the camera too).

Microscopes that offer coaxial light couple the light for illuminating of the object with a semi-transparent mirror into the optical axis of the objective. This is difficult to realise with the setup described here.

Well, you could drill a large hole in the side of an extension tube and put a thin glass plate inside at 45°. That's what I do, but with smaller lenses and smaller sensors. (Forget smartphone mirror foil, it's crap. I use microscope slide or cover glasses now.)

Or put a glass between the reversed lens and the die, plenty of space for that with SLR lenses thanks to long flange focal distance. This introduces some astigmatism if lens aperture is too wide and IMO sharpness is barely passable near f/4 even with 0.17mm glass, but it's very easy to experiment with.

With these techniques the chip can be 90° to the camera to avoid focus stacking, but light may still be angled to produce some shadows.

The above arrangement usually results in the centre image. At first glance, the surface structure appears inverted. The transistors now appear to be located on sockets. How this effect arises is unclear. If you want to understand and interpret the surface structure, it helps to assume that the light source is located below the image. In this way, the shadows correspond to the actual geometries. This becomes clearer if you rotate the image by 180° (right). Here the transistors appear to be located in recesses, as is the case in reality.

Did you try simply moving the lamp further away from the camera, above the chip?
If you have problems with flare, make a suitable lens hood from black paper.

[Noopy]

Once you have roughly positioned the object to be imaged, focussing is finally carried out using the focus ring.

This wouldn't work with lots of mirrorless system lenses as they focus by means of motors moving lens groups inside.
Even manual focusing requires power (and perhaps data connection with the camera too).

You need a "big" mirrorless camer like the EOS R. 
These system cameras are probably not the right tool.

Microscopes that offer coaxial light couple the light for illuminating of the object with a semi-transparent mirror into the optical axis of the objective. This is difficult to realise with the setup described here.

Well, you could drill a large hole in the side of an extension tube and put a thin glass plate inside at 45°. That's what I do, but with smaller lenses and smaller sensors. (Forget smartphone mirror foil, it's crap. I use microscope slide or cover glasses now.)

Or put a glass between the reversed lens and the die, plenty of space for that with SLR lenses thanks to long flange focal distance. This introduces some astigmatism if lens aperture is too wide and IMO sharpness is barely passable near f/4 even with 0.17mm glass, but it's very easy to experiment with.

With these techniques the chip can be 90° to the camera to avoid focus stacking, but light may still be angled to produce some shadows.

I once tried a semi transparent mirror between the lens and the die. The image quality deteriorated badly. Perhaps the problem was the thickness of the mirror. Well it was the one thing I was able to put my hands on...

Would just thin glass do the job?

The above arrangement usually results in the centre image. At first glance, the surface structure appears inverted. The transistors now appear to be located on sockets. How this effect arises is unclear. If you want to understand and interpret the surface structure, it helps to assume that the light source is located below the image. In this way, the shadows correspond to the actual geometries. This becomes clearer if you rotate the image by 180° (right). Here the transistors appear to be located in recesses, as is the case in reality.

Did you try simply moving the lamp further away from the camera, above the chip?
If you have problems with flare, make a suitable lens hood from black paper.

If you move the lamp further away but remain the angle it doesn´t change anything. You just have less light.
If you move the lamp away from the camera changing the angle then at first you get "normal" black&white pictures and then you get pictures like the left of the three I showed here.

Flare is not really a problem as long as you don´t put the light directly in the optical path of the lens. 

[magic]

Sorry, I remembered wrong - the astigmatism really sucks. I tried it on OP07 with a 5x microscope objective (same chip and same objective as here). I believe it has about similar aperture as f/4 photographic lens in reverse so the results may be similar too. Test images attached: one focused on horizontal lines, one on vertical lines, one in the middle, one with the mirror behind the lens (between the lens and the camera). Differences are quite obvious. Ignore bad white balance.

Astigmatism dramatically increases with aperture (10x objective results were no better than 5x, maybe worse) and equally dramatically decreases when stopping down. Effective aperture (on the camera side) of basic microscope objectives is ~f/20 and I can't see any difference with a 1mm glass plate being there or not. Placing the glass in front may work well at lower magnifications, where wide (object-space) aperture is not so critical for resolution. I remember reading that some coin photographer was happy with it.

Yes, ordinary glass works well, but it needs to be high quality and flat to avoid spots of blur and distortion. I found that "$1 for a pack of 100" cover glasses (24x24x0.17mm) are not satisfactory, and this is visible right away - if I look through such glass and move it a little, the scene behind shifts and distorts. Professional glasses from reputable manufacturers should be better. OTOH, my "$2 for a pack of 50" slide glasses (75x25x1mm) are perfect, YMMV. The coin photographer used photographic UV filters - these should be good too, but I don't know their thickness.

A glass plate at 45° angle passes about 80% of light and reflects 20% sideways (10% from each air to glass surface, these two reflections are offset from one another by the thickness of glass but it's not a big deal in lighting). A potentially bigger issue is ghost image - 10% of passing light reflects internally instead of exiting the glass, and 10% of it reflects again and exists in parallel with the original light. However, 1% of the original image is over 6 stops darker and rarely visible. Did anyone notice ghost image on my LM4562? See below with contrast tweaked for maximum visibility.

[Noopy]

With the glass between the camera and the lens the picture looks very good. 

My problem is that I want a mechanical robust solution which is still good to handle. I should start working with 3D printers...