Transistors - die pictures

[Noopy]

The Siliconix VN88 is an old Power-MOSFET.






The die is 1,8mm x 1,0mm and was protected with a silicone like potting.




That´s an interesting die marking.




There is a small resistor between the gate bondpad and the gates. The surrounding area is connected quite massive to the source. Probably that is a zener protecting the gate-source-isolation. 


https://www.richis-lab.de/FET14.htm

 

[T3sl4co1l]

Interesting, they don't show the zener, at least in the -AFD (TO-220) version, at least in the Vishay/Siliconix datasheet from 2001.

Not sure if that's V-groove (VMOS innit?) or trench (VDMOS?), definitely not lateral in any case!

Tim

[Noopy]

Nothing in the datassheet. I found that interesting too! 

Some people state it is a VMOS...

[vishkas]

Looks Interesting

[Noopy]

The П217Б (P217B) is an old germanium power transistor built by Pluton, a company you can still find in moscow. (Thanks to serg-el for the hint!)
45V / 7,5A / 30W




That is a package you don´t see every day.




In the package we find a lot of dehumidifying powder.




Well meanwhile we know these transistor stackups.


https://www.richis-lab.de/Bipolar47.htm

 

[Noopy]

The Ferranti ZTX312 is a transistor which switches quite fast: ton=20ns, toff=15ns
That makes it interesting to compare it with the general purpose ZTX108C (https://www.richis-lab.de/Bipolar37.htm)






Nice, there is an E to recognise the emitter. 
The die is 0,44mm x 0,44mm. The active area is just 0,11mm x 0,09mm. With two emitter areas and three base contacts the transistor gets quite fast.
Interesting point: The ZTX312 allows 500mA while the ZTX108C is specified with just 100mA. A drawback of the ZTX312 is the lower amplification (less than a tenth of the ZTX108C).


https://www.richis-lab.de/Bipolar49.htm

 

[Gyro]

I think you may find that the 'E' stands for E-line.  This family dates back to the Ferranti E-Line series.

They use a Silicone rather than Epoxy packaging material, allowing higher junction temperatures and dissipations than TO92. They were used a lot in Mil spec equipment.

[Noopy]

Sounds plausible! 

I didn´t realize the higher operation temperature in the datasheet but I realized that the mold was a little different. 

[Gyro]

Yes, I did wonder if you had found the packaging different / harder to remove. 

[Noopy]

Yes, I did wonder if you had found the packaging different / harder to remove. 

Epoxy is a little more brittle after the baking process.
The silicone is more like a little too hard rubber but you can still "crumble" the die out of the residues. 

[David Hess]

I think you may find that the 'E' stands for E-line.  This family dates back to the Ferranti E-Line series.

Is that the same as the E-line transistors from Zetex, now Diodes Incorporated?

They use a Silicone rather than Epoxy packaging material, allowing higher junction temperatures and dissipations than TO92. They were used a lot in Mil spec equipment.

The E-line transistors from Zetex were known for higher current gain and lower saturation voltage at higher current but not particularly higher power.  I thought Zetex was repurposing old IC production equipment for transistor production which gave them a much smaller feature size to work with.  On Semiconductor and Fairchild and I assume others eventually produced their own versions.

I always hoped that they would extend these designs to the TO-225 and TO-220 packages but this never really happened.

[T3sl4co1l]

Yes, Ferranti --> Zetex --> Diodes.

Low-Vce(sat) parts are available in power SMDs up to DPAKs, not sure about D2PAK or leaded (aside from IPAK which almost doesn't count ), and up to 8A or so I think.  I don't have any part numbers offhand but I've seen a number of Japanese type numbers that perform very well.  They are (were?) common in CCFL backlights (Baxandall push-pull oscillator).

They also have very good inverted hFE, despite the low Vebo (i.e., asymmetrical junctions).  This is no coincidence, it's a necessary part of getting low Vce(sat).

One thing that's never been made, as far as I know, is a high fT model.  Would be nice to have much faster parts for gate drives I guess; but then, it's hard to make a gate drive that powerful (multiple amperes peak), go much faster than 10ns anyway, and you'll always be limited by stray inductance in that case.  So, shrug.

Ed: oh, I do remember one, https://fscdn.rohm.com/en/products/databook/datasheet-nrnd/discrete/transistor/bipolar/2sc5001.pdf which is since obsolete so YMMV, but you get the idea.

Tim

[Noopy]

...

I had to correct some misspelling. 


The voltage rating of the ZTX312 is interesting:
Vceo = 12V
Vcbo = 35V
That´s quite a difference!
I assume that is because of a very high doping of the emitter (and the base?) which they had to do to get fast switching. What is your opinion?

[T3sl4co1l]

Hmm, wonder if that was gold doped.  Would explain the higher leakage (arguably; leakage at 25°C is low enough it might be a testing limit, but it's also not a very round number?), and the relatively low voltage and switching time.  hFE curves are not provided (steeply reduced hFE at low Ic is a hallmark of such types).

Higher C-B leakage I think is a factor in Vcbo vs. Vceo breakdown, since the leakage is effectively multiplied; I'm not sure what all physics is really at work, as the ratio can be quite significant (as in this case); other times it's much more modest.

The boundary between the two conditions is a kind of avalanche breakdown that generates very high noise, indeed the avalanche multiplication can extend so far that the device saturates in a fraction of a nanosecond, while carrying several amperes in the process.  (Evidently this phenomenon involves two mechanisms: a high-level injection mechanism which crowds out the base doping concentration, effectively causing local punch-through (effectively a C-E short); and subsequent rapid heating, causing formation of a current filament (at high temperatures, intrinsic carriers outnumber dopants, maintaining punch-through).  That perhaps explains why recovery from this mechanism is so slow -- several microseconds at least.)

FWIW, I'd love to see microscopy of a bare die under such operation -- but I'm also afraid that there would be so little light emitted (due to the very low duty cycle, < 0.1% is typical) that it might not show at all.

(The usual setup is: a large pullup resistor to charge the collector, typically 10-100k to +100V or thereabouts; a B-E resistor, typically 4.7k or thereabouts, depending on type; and some kind of load, at least a few pF from C-E but also including an output coupling network like a 50 ohm transmission line or whatever.  100-120V is enough to cause 2N3904 to break down in this way; 2N2369 needs less, 60-80V I think; it even works for high voltage types, but because breakdown occurs as a narrow filament, power transistors are essentially dead weight for this -- a 1500V 10A transistor fails with only a few 100s of pF load, hardly more than a 300V 100mA type can handle.)

Tim

[exe]

I have a question. Current western semiconductor companies such as onsemi, vishay, diodes, infineon, nexperia and rohm, do they still produce bjts by themselves, or they just relabel Chinese parts? Asking because bjts are so cheap, how they can make money from them...

[Noopy]

Higher C-B leakage I think is a factor in Vcbo vs. Vceo breakdown, since the leakage is effectively multiplied; I'm not sure what all physics is really at work, as the ratio can be quite significant (as in this case); other times it's much more modest.

That sound plausible.
Surprisingly the hfe of the ZTX312 is quite low. I would have suspected more similar breakdown voltages than for example the ZTX108C has to offer: hfe=450-900 and 45V/30V.

FWIW, I'd love to see microscopy of a bare die under such operation -- but I'm also afraid that there would be so little light emitted (due to the very low duty cycle, < 0.1% is typical) that it might not show at all.

(The usual setup is: a large pullup resistor to charge the collector, typically 10-100k to +100V or thereabouts; a B-E resistor, typically 4.7k or thereabouts, depending on type; and some kind of load, at least a few pF from C-E but also including an output coupling network like a 50 ohm transmission line or whatever.  100-120V is enough to cause 2N3904 to break down in this way; 2N2369 needs less, 60-80V I think; it even works for high voltage types, but because breakdown occurs as a narrow filament, power transistors are essentially dead weight for this -- a 1500V 10A transistor fails with only a few 100s of pF load, hardly more than a 300V 100mA type can handle.)

Wouldn´t it be possible to test the CE-breakdown in a constant manner? If we use a low voltage transistor the power dissipation is perhaps low enough to take pictures for some seconds.

I have a question. Current western semiconductor companies such as onsemi, vishay, diodes, infineon, nexperia and rohm, do they still produce bjts by themselves, or they just relabel Chinese parts? Asking because bjts are so cheap, how they can make money from them...

I don´t know. Perhaps they are so cheap because you get 10000000000000000 transistors out of a 300mm wafer. 

[magic]

Rohm is very much an eastern company

Given that reputable manufacturers are able to provide detailed detailed datasheets, and that occasionally there are differences in typical characteristics like hFE vs Ic, it seems they still design and control the manufacture of their products. Which exact Asian country the trannies are fabbed and packaged in you could ask your supplier; the country of origin is often given.

[T3sl4co1l]

Wouldn´t it be possible to test the CE-breakdown in a constant manner? If we use a low voltage transistor the power dissipation is perhaps low enough to take pictures for some seconds.

Yes, it can be done at some mA -- but I wonder if the result is different in "simmering" versus "switching" operation.

Might be even more interesting in infrared, or far IR for that matter -- good luck getting a lens tight enough though, or enough framerate to see more than a blob of spread-out heat anyway?

(I know semi mfgs have tools to do this sort of thing -- IR microscopes to inspect fabbed dies, from the back side even, as silicon is transparent in longer wavelengths.  Not enough resolution to see individual transistors of course, but enough to have an idea where a problem might be.  Those would probably be able to resolve something... the question then is, can anything cheaper than a million dollars do the same? )

Tim

[David Hess]

Wouldn´t it be possible to test the CE-breakdown in a constant manner? If we use a low voltage transistor the power dissipation is perhaps low enough to take pictures for some seconds.

You can, but at low current density, non-uniformity in the junction means that only a small part of the entire junction is likely to break down and be visible.  Breakdown at a much higher current density will show the entire junction, but will quickly overheat the junction unless the time is limited.

Parts intended for avalanche operation, like avalanche rectifiers, have special processing to make the junction more uniform preventing hot spots which would destroy them at lower current.

[Noopy]

Well highspeed IR microscopy is a little to much for my equipment! 

I will try some more breakdown pictures as soon as I have a transistor which fits to this experiments. 

[Noopy]

KT808, a russian power transistor built by Iskra: 130V, 10A, 60W, 2MHz




There is some potting on the die.




Here we see the MESA trench.




There is a scratch probably caused by a testing probe.
It seems  like the die is protected with polyimid.




Hm, there are metal flakes under the potting.




The avalanche breakdown light of the base emitter junction is quite uneven. (-6,5V / 0,1A)




1A, still uneven.





And one more KT808. It´s an older one.




There are no grooves in the heatspreader that can drain tin.




The bondpads are welded on the side of the connector pin.




There is no yellow protection layer on this transistor.
Quite extreme welding on the die...




MESA




Not really clean... 




Lights on! 






Even more uneven at -7V/0,1A




2A


https://www.richis-lab.de/Bipolar48.htm

 

[Noopy]

BUX42
250V, 12A / 15Ap, 8MHz, 120W






A nice transistor. 
Perhaps the C is kind of a binning?
You can barely see the tin of the die attach.




It seems like the tin was quite runny. It flowed a long way inside the trenches on the heatspreader.
The tin layer between the die and the heatspreader is as thin as possible to get best heat conduction.






There is some distance between the emitter contact and the base-emitter-junction. It is a wide-emitter narrow-contact design which ensures equal current distribution.
Interesting: The base metal doesn´t contact the base area on the left edge of the die.




118, perhaps the name of the design.






Lights on!  ...at -15V.




20mA




100mA




200mA
Only the junction above the base contacts is glowing! 




500mA




1A
At higher currents the the other side of the base comes into play.
But you can clearly see that there is no base contact on the left side. The base resistance is higher at this edge and the current crowds in the rest of the die.


https://www.richis-lab.de/Bipolar50.htm

 

[SilverSolder]

Some amazing pictures there, @Noopy, have you got some new equipment or techniques you are playing with?  Or do you just keep getting better?

[T3sl4co1l]

Oh wow weird, is that... two, three thicknesses of metallization?

I've always wondered if they do things with metallization thickness but never saw much evidence for it (also, it's extra masks, why bother..?), sure looks like they did that here though.  Would the thinner first layer (particularly for the emitter) serve as ballasting?

Tim

[Noopy]

Some amazing pictures there, @Noopy, have you got some new equipment or techniques you are playing with?  Or do you just keep getting better?

Actually I´m doing experiments with the Canon MP-E 65mm f/2.8 1-5x but most of the improvement is due to experience and patience. 
And sometimes I´m more lucky with my pictures. 

Oh wow weird, is that... two, three thicknesses of metallization?

I've always wondered if they do things with metallization thickness but never saw much evidence for it (also, it's extra masks, why bother..?), sure looks like they did that here though.  Would the thinner first layer (particularly for the emitter) serve as ballasting?

I don´t think there is more than one metal layer. The metal layer looks quite different but I´m pretty sure that is just due to the contact areas and the different heights of the underlaying layer.
The emitter contact for example is the wide-emitter narrow-contact design.