LTZ1000 - nice die pictures

[Noopy]

Hi all,


today I have some nice LTZ1000 die pictures for you.




The package...




You know the structure but the picture is very nice, isn´t it?




Zoom in...




More zoom...




Does anybody know what kind of logo that is?


Some more pictures here:

https://www.richis-lab.de/REF03.htm


Greetings,


Richard

[doktor pyta]

Regarding logo: Super Zener

[TiN]

Very nice. This should be in metrology

[Noopy]

Regarding logo: Super Zener

Aha! 

Very nice. This should be in metrology

I thought about it... But since I have only pictures and less technical information I decided to put it here.
If a moderator thinks it should be in metrology please shift it! 

[iMo]

Nice!
There are 4 heaters left unconnected. Interesting..

[Noopy]

Nice!
There are 4 heaters left unconnected. Interesting..

A few minutes ago I learned that the other heaters are used in the RMS-converter LT1088.

Because of the use in the LT1088 the heaters are connected different so that the current is flowing in the one direction in the one heater and in the other direction in the other heater. Low inductance is important for the LT1088.

...I think I have to check a LT1088 in detail... 

[iMo]

The round structure is the Q1, the zener is inside the Q1 - almost perfect thermal coupling.
The Q2 (the sensor) is made of 4 transistors in parallel - in form of X around the Q1.

[magic]

Hi Noopy, I don't think I agree with your comments about substrate grounding

The substrate has to be the lowest potential in the circuit, with few exceptions*. Otherwise unlimited current would flow from the P type substrate to any of the random N type things (like NPN collectors) made in the N type epitaxial layer.

I think the brown areas are plain, unmodifed N epitaxial layer. So the "thing" connected to pin 1 is simply the whole reminder of the epitaxial layer which hasn't been separated into transistor tubs by the violet isolation diffusions.

Isolation diffusions are P type and they go all the way down to the P substrate, so we see at once that the substrate is electrically connected to pin 4 by the ring of isolation diffusion separating Q1 from Q2 clones. There appears to be yet another parasitic diode not included on the schematic, from pin 4 (substrate/isolation) to pin 8 (Q2 collector). You could detect it with a DMM.

Nominally, pin 4 is Q1 base. Of course the substrate can't be the base, the base has to be located under the emitter. There are actually two concentric "emitters", the outer one is the zener cathode. Interestingly, the base has no metal connection on the surface. Instead, it appears that there is a hidden column of P type material extending from the base all the way down to the substrate. This is the reason why pin 4 is connected to the substrate. Not sure what's the point of that, perhaps voltage at the bottom of the base was more stable than at the top

Q1 collector tub is another part of the circuit with no surface connection. Instead, it's connected from the bottom with four deep "jumper links", which might be made in the N+ burried layer, embedded in the substrate below the reach of the isolation ring. It seems somebody worked very hard to make Q1 collector current flow "upwards" through the base and stay away from the surface for some reason.

*exceptions: some emitters are one diode drop lower potential than the substrate, because they are located on bases whose potential is equal to the substrate.

edit
Now is time to power that puppy up and see if it still works
(BTW, if anyone actually wants to try that, cover the IC from light. Light will cause unwanted conduction through reverse biased junctions, perhaps even damage).

[iMo]

They do test the 4 unconnected heaters (see the pads) - why?
Perhaps quick thermal cycling before the actual selection process?

[Andreas]

Nominally, pin 4 is Q1 base. Of course the substrate can't be the base, the base has to be located under the emitter.

Pin 4 has high ohmic connection to the package.
I measured 65 Ohms on a non-A Version.

https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg2848044/#msg2848044

with best regards

Andreas

[magic]

Found this
https://www.electronicdesign.com/technologies/analog/article/21804689/whats-all-this-ltz1000-stuff-anyway

One useful thing to understand (it took me 15 years) about the LTZ1000 is that pin 4 is the p-substrate. All the other pins operate above that (which is why you want the heater a diode-drop up) ... except for the mysterious case of pins 6 and 7, which operate 0.6V below the substrate.” Carl Nelson, the LTZ1000 designer, notes he did this to allow for a true subsurface Kelvin connection to the bottom of the Zener, reducing its resistance.

Those mysterious pins are the trap
There are other ICs with below-the-substrate emitters, like LM317.

[Dr. Frank]

Btw: Carl Nelson designed this part, therefore CN are his initials, obviously.
MG I don't know.

SZKP could also mean Super Zener , KP maybe Kelvin Probing.
Frank

[Noopy]

Thanks for all your input!
You are right regarding the substrate! 
I did also some measurements: I could meassure something a little above 100 package to pin 4.
That´s much more consistent! 


I´m pretty sure the LTZ1000 would still work. Bonds are ok. I thought about searching for glowing in the zenerregion. 
Worked good with a 2N3055:



https://www.richis-lab.de/2N3055.htm

But I don´t think that´s possible with a small buried zener.


Why testing the heater?
A good question! The dies for the LT1088 have a different metal layer, so you don´t really have to check the heaters.* Perhaps they can check some quality points. Or they do some early aging?


Well out of the schematic pins 6 and 7 have to be 0,6V below the substrate because of the base-emitter of the connected transistors.


Thanks for the hint to Carl Nelson!


*Not sure about that. I will have to check that! 

[SilverSolder]

Fantastic pictures (love the light emitting 2N3055).   

[Noopy]

I thought about searching for glowing in the zenerregion. 
Worked good with a 2N3055:



https://www.richis-lab.de/2N3055.htm

But I don´t think that´s possible with a small buried zener.

Is it possible to put a light in the LTZ1000?
Well yes! 

I proudly present a glowing LTZ1000-Zener:



You can see that the pn-junction is in the middle of the cathode-ring of the zener. That means the junction is under the cathode-ring. That was to be expected. Nevertheless nice to see.


I can also put a light in the pn-junction of the transistor Q1. 

Whole story here:

https://www.richis-lab.de/REF03.htm#Breakdown

 

[magic]

Kinda surprised by the Q1 result. Why isn't it glowing all around the emitter?


BTW, look for "noise reduction" in the camera's settings
If you are lucky it may enable fully automatic dark frame subtraction, I know that Canon has it on their compact cameras.

[Noopy]

Kinda surprised by the Q1 result. Why isn't it glowing all around the emitter?

It seems that the luminescence of such a second breakdown is not always perfectly uniform.
I have seen such effects in the older 2N3055.

I asume that there are areas where the pn-structure breaches and then these areas emit light. When there is a breached area the current has a path to flow through and there doesn´t have to be a second breach or uniform breach over the whole pn-structure.
My assumption would match with the noise effects you have with z-diodes conducting only very little current. Literature tells us that there are different areas conducting or not.

In the LTZ1000 it´s perhaps sufficient for the current to flow only the shortest way under the metal conducts through the pn-structure. Pherhaps for the other half of the pn-circle the resistance is already too high to breach.


I have seen that the glowing points of the z-diode are always at the same locations.
It seems that there are impurities or imperfect structures that cause breach at special points.
Other explanation: In this special points there is more ionisation leading to visible light.

BTW, look for "noise reduction" in the camera's settings
If you are lucky it may enable fully automatic dark frame subtraction, I know that Canon has it on their compact cameras.

You are right. My 60D has such a setting too. But you have to wait twice the time and it takes some time to take a picture of this small glimpse of light. Even at high ISO. 
But I will try it next time. 

[Noopy]

Nice!
There are 4 heaters left unconnected. Interesting..

A few minutes ago I learned that the other heaters are used in the RMS-converter LT1088.

Because of the use in the LT1088 the heaters are connected different so that the current is flowing in the one direction in the one heater and in the other direction in the other heater. Low inductance is important for the LT1088.

...I think I have to check a LT1088 in detail... 

Of course I had to check what´s inside the LT1088:

https://www.richis-lab.de/LT1088.htm





Surprise:
The LT1088 doesn´t use the LTZ1000-die!
It´s a similar die but they changed more than only the metal layer.

So again the question is why the LTZ1000-heater was built this way.
- special testing after die production?
- (quick) aging after die production?
- a version with a more powerful heater?
Probably we will never know...


Edit: Adding teaser-picture... 

[iMo]

They bond the 1st two rings for Civilian apps, then they add - the 3rd ring for Navy, the 4th ring for Arctic Forces, 5th ring for Air Force, and the 6th for Space Defense

[Noopy]

They bond the 1st two rings for Civilian apps, then they add - the 3rd ring for Navy, the 4th ring for Arctic Forces, 5th ring for Air Force, and the 6th for Space Defense

Well at least possible... 

[aheid]

 Works of art, dies and photos both. Thanks again!

[magic]

Why does the artwork in the center remind me of a competitor's logo?

[macboy]

I believe that the die itself (the doped silicon) is probably identical between the two. Clearly, the polysilicon layer (resistive elements) is different in addition to the metal layer(s). There is no reason to believe that the die itself is not common, especially with so many small details being identical between them.

[Ranayna]

These pictures, and the others you are posting here, are absolutely amazing 
The first picture in this thread reminds me of my own curiosity causing me to file open various metal canned components when I was young

Back then I was mostly disappointed about how empty they were...

[Noopy]

Thanks for everybody giving me positive feedback! 
I think I have to go on with decapping! 

Why does the artwork in the center remind me of a competitor's logo?

Analog Devices? It´s similar but not the same...
Today NS and Analog are merged. Perhaps there was some collaboration before...

I believe that the die itself (the doped silicon) is probably identical between the two. Clearly, the polysilicon layer (resistive elements) is different in addition to the metal layer(s). There is no reason to believe that the die itself is not common, especially with so many small details being identical between them.

There is some similarity between LTZ1000 and LT1088. The isolations in the middle are the same, also the bondpads, the four transistors to measure the temperature and some metal structures. But the heaters are different, the buried zener and it´s transistor are missing completely...
It´s not completely different but they changed more than only the metal layer.

[macboy]

There is some similarity between LTZ1000 and LT1088. The isolations in the middle are the same, also the bondpads, the four transistors to measure the temperature and some metal structures. But the heaters are different, the buried zener and it´s transistor are missing completely...
It´s not completely different but they changed more than only the metal layer.

The heaters are made from the polysilicon layer, which is a deposited layer, similar to metal layers, not part of the crystalline silicon die. Also, there are clearly unused structures around the middle of the LT1088 which don't need to be there, but which are visually identical to useful structures on the LTZ1000. Is there any other explanation for their presence?

[Noopy]

There is some similarity between LTZ1000 and LT1088. The isolations in the middle are the same, also the bondpads, the four transistors to measure the temperature and some metal structures. But the heaters are different, the buried zener and it´s transistor are missing completely...
It´s not completely different but they changed more than only the metal layer.

The heaters are made from the polysilicon layer, which is a deposited layer, similar to metal layers, not part of the crystalline silicon die. Also, there are clearly unused structures around the middle of the LT1088 which don't need to be there, but which are visually identical to useful structures on the LTZ1000. Is there any other explanation for their presence?

Are you sure the heaters are made from the polysilicon layer? Couldn´t it be possible that they are made with base- or emitter-doping?

The unused structures in the middle of the LT1088 are the isolation trenches (don´t know whether that´s really the right name for it) and you are right that are the same as in the LTZ1000.

I only wanted to say that they changed more than only the metal layer…

[iMo]

You may doublecheck that based on the sheet resistance of the rings/heaters.
In 1088 the outer heater is 250ohm, made of 2x500ohm half_rings in parallel (low inductance).
Inner heater is 50ohm, made of 2x100ohm half_rings in parallel.

The LTZ outer heater is 300ohm = 2x600ohm full_rings in parallel, afaik.

[Noopy]

Let me explain why the difference is important for me anyway:

I found it very interesting that the LTZ1000 has two (four) unused heater rings beside the outer one (two) connected heater rings.
From that point on I gambled what they did with the unused heaters.
People say that the unused heaters are important for the LT1088 were they are used for the 50Ohm/250Ohm heaters. That would explain why there are two unconnected heaters in the LTZ1000.
Now that we have good pictures of the LT1088 we can say that the unused LTZ1000-heaters are no spare parts for the LT1088 and we come back to the interesting question why they have integrated These heaters. 

Addon:
Perhaps we have a misunderstanding regarding my sentence that the LT1088 doesn´t use the LTZ1000-die.
Of course they used the same ""base-material"" but they changed more than only the metal layer.

[iMo]

The 1088 and LTZ are absolutely different devices.
The 6 rings in LTZ have nothing to do with 1088, imho.
Based on the 1088 operation principle - a heater changing an input current into the heat (=rms), while sensing the heat with two diodes - they simply reused the design of LTZ as you "cannot do it better" (at that time they messed with it).
The 2 diodes in 1088 are made of those 4 symmetric transistors around the center (those 4 are in parallel).

PS: also the wiring of those heaters is different - see my above post. With 1088 you do need a low inductance of the heaters (you want to measure a high frequency AC rms), with LTZ you do not care.

[Noopy]

The 1088 and LTZ are absolutely different devices.
The 6 rings in LTZ have nothing to do with 1088, imho.
Based on the 1088 operation principle - a heater changing an input current into the heat (=rms), while sensing the heat with two diodes - they simply reused the design of LTZ as you "cannot do it better" (at that time they messed with it).
The 2 diodes in 1088 are made of those 4 symmetric transistors around the center (those 4 are in parallel).

Thanks!
That´s exactly what I thought. 

[Kleinstein]

I see two possible reasons for the spare heaters:
One is that they had the LT1088 use in mind when they made the LTZ1000 masks and only later decided to make a separate LT1088.

Another point could be the control loop for the heater. The heater closer to the center could result in faster response and thus possibly a little simpler (smaller caps) control circuit. To this might have been an option to get easier control if really needed.  There is not much cost involved in having the extra heaters inside - the point that can add to the cost is the die size and thus the outer heater. So the inner rings could also have be there for a test if it may be possible to get away with a smaller (e.g. half the area) die using only the inner heaters. The LM399 run pretty hot for some reason. To run the LTZ1000 in the same temperature range would need quite some power so that the outer heater may need more than some 12 V. So a lower resistance heater (possibly using both) would help to operate at higher temperature without needed a higher voltage.  It turned out that high temperature operation is not desirable.

It is funny to also see some glowing around the buried zener.  AFFAIK the glowing is coming from hot electrons hitting the oxide and exiting higher energy states there - a little like writing to the EPROM.  This mechanism is supposed to be part of the high noise in conventional zeners. So the glowing should be highly undesirable as it likely is also causing noise and may cause aging too. My understanding of the buried zener was that it avoids the hot electrons at the surface.

Of cause it could be that the open LTZ1000 is one of the rare bad ones with high noise.

[Noopy]

I see two possible reasons for the spare heaters:

Absolutely possible... 

It is funny to also see some glowing around the buried zener.  AFFAIK the glowing is coming from hot electrons hitting the oxide and exiting higher energy states there - a little like writing to the EPROM.  This mechanism is supposed to be part of the high noise in conventional zeners. So the glowing should be highly undesirable as it likely is also causing noise and may cause aging too. My understanding of the buried zener was that it avoids the hot electrons at the surface.

Of cause it could be that the open LTZ1000 is one of the rare bad ones with high noise.

You mean the glowing in the LTZ1000-zener while there is flowing a current through it?
I only see glowing under the ring, where the buried zener border should be.
My explanation is second breakdown ionisation. I assume you mean the same with hot electrons. But that should not happen at the surface.

[Noopy]

Hi all!

I wasn´t happy with the quality of my overview-picture of the LTZ1000.
So I have taken a new one:



Thanks to hotlinking the picture is also in the first post. 

By the way a picture of the die attachment:




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

[TiN]

Noopy, beauty.

Now you need to get LTZ1000A and photograph it's die attach with bubbles too.

[SilverSolder]

Noopy, beauty.

Now you need to get LTZ1000A and photograph it's die attach with bubbles too.

Perhaps the bubbles are the cause of popcorn noise!   

[Noopy]

Noopy, beauty.

Now you need to get LTZ1000A and photograph it's die attach with bubbles too.

Thanks! 

Definitely i need a LTZ1000A!  I wonder if I´m able to capture the bubbles on a picture...

[Noopy]

Good news:

I updated some pictures of the LTZ1000 

https://www.richis-lab.de/REF03.htm

Due to hotlinking the pictures in this thread are up to date.


Bad news:

I tried some more avalanche breakdown pictures. And... Well… The current limit was set to high... 



R.I.P.

[SilverSolder]

Well, that's one way to get it to emit light! 

[Noopy]

Well, that's one way to get it to emit light! 

A bright light for a very short time. 

[Noopy]

Now you need to get LTZ1000A and photograph it's die attach with bubbles too.

I still have no LTZ1000A but I have realized that my LT1088 has the same "air impregnated polymer"! So I took some more pictures of the LT1088:






The distance between the die and the package seems to be a little bit more than 100µm.
I meassured 28 bubbles. The average diameter was ~70µm. The diameter of the biggest bubbles is round about 100µm.

LT1088:
https://richis-lab.de/LT1088.htm
LTZ1000:
https://richis-lab.de/REF03.htm

 

[doktor pyta]

I'm answering question:

[Noopy]

I'm answering question:
[...]

 

[coromonadalix]

One thing i'm curious about :   for parts like this, or put in a metal can,   do they vaccum the air out ?!

[Noopy]

One thing i'm curious about :   for parts like this, or put in a metal can,   do they vaccum the air out ?!

I don´t know. 
High voltage says UV-EPROMs are placed in a vacuum (or at least low pressure):





https://richis-lab.de/hv-mixed_2.htm

 

[bson]

I wonder if those unused ring traces could potentially be used for guard currents, to reduce noise pickup from the heater traces?

[RoGeorge]

High voltage says UV-EPROMs are placed in a vacuum (or at least low pressure):



https://richis-lab.de/hv-mixed_2.htm

 

Wow, very interesting!
I didn't know that.   

From your blog, you were using 20kV, but I assume most of the voltage drop will be across the length of the two sparks, therefore the voltage between the Vcc and the address pin 4 of the EPROM must have been much lower than 20kV.


 
Have you tried doing that with direct contact and much lower voltage?  I wonder what would be the minimum voltage that can still make the wire glow.  Also, was the negative pole of the 20kV on the left side (pin 4)?

Does the whole chip lights up instead of just the bonding wire if the polarity is reversed?

[Noopy]

I wonder if those unused ring traces could potentially be used for guard currents, to reduce noise pickup from the heater traces?

That´s at least possible... 

I didn't know that.   

Well that´s not a standard testcase. 

From your blog, you were using 20kV, but I assume most of the voltage drop will be across the length of the two sparks, therefore the voltage between the Vcc and the address pin 4 of the EPROM must have been much lower than 20kV.

Yes, the voltage across the EPROM was lower than 20kV.
How much... Don´t know... 

Have you tried doing that with direct contact and much lower voltage?  I wonder what would be the minimum voltage that can still make the wire glow.  Also, was the negative pole of the 20kV on the left side (pin 4)?

No, didn´t try that. The Heinzinger supply can do lower voltages. I have never tried that but should be possible. Taking such pictures isn´t easy because of the timing but I can try to take some more with different voltages.
I´m not sure where the negative pole was, sorry. 

Does the whole chip lights up instead of just the bonding wire if the polarity is reversed?

I didn´t zap special pins. I just took the pins near to my electrode.

I uploaded two more pictures:







With 20kV the ESD-pad is enough to conduct the high voltage... 
Here the positive high voltage has to be on the right side.