Metalisation creep?

[N_Cook]

Why would a SMPS high speed Unitrode rectifier stud diode 1N5812 , 20 amp,
fail sitting around in a normal human occupation heated environment for 10
years unused?
Currently reads a stable 6.2 ohms. Replaced with a new one but I intend
passing ever increasing current through the old one to see what happens.
Any other ideas concerning cause or analysis ?

 

[cj_elec_tech]

I would suspect ESD (electro-static discharge) damage is a possibility - particularly if it's been sitting in a plastic parts drawer for years.
Static build up will punch-through almost any PN junction causing devices to read short circuit or exhibit leakage (resistance values from low to high). In some cases the damage will only show up when the device nears it's operating limits, in other cases they fail for no apparent reason at low load levels.

I've seen many 'new' components which were faulty straight from the parts drawer - it's astounding how much static can build up when a person is sitting on a chair or walks across a floor (let-alone synthetic carpet!). Poor ESD handling precautions are almost always to blame.

I shudder every time I see someone 'putting together' a PC or other electronic equipment without using even elementary ESD precautions - and people wonder why a part of their PC suddenly fails a few years down the track. Many times I've heard "my hard drive died the other day and I lost all my ......" followed by " ... dad put a new one in and now I have to download all my nnnnn files again......" - Yup you bet! And don't be surprised when the new one fails a year later either - good old Dad (or next-door neighbour) who knows *all about* computers or electronics.....

Sorry for ranting; I've learned from very hard and *expensive* experiences about ESD damage.

Of course I could be wrong.....

Maybe....

 

[N_Cook]

A Motorola 2N2904A outside the ps had gone B-E ohmic of 77 ohms in the same
timespan of un-use. Would these faults have occured if regularly used , ie
the m/c would have needed repairing twice in 10 years ?

 

[Mark Zacharias]

Why would a SMPS high speed Unitrode rectifier stud diode 1N5812 , 20 amp,
fail sitting around in a normal human occupation heated environment for 10
years unused?
Currently reads a stable 6.2 ohms. Replaced with a new one but I intend
passing ever increasing current through the old one to see what happens.
Any other ideas concerning cause or analysis ?

I suspect tarnish of interior metals, especially if they contain any silver.
I see this sort of thing in Euro capacitors and metal connectors,
fuseholders, and the like. I also have to scrape it off the leads of old
semi's that have been in a parts drawer for years.

Mark Z.

 

[N_Cook]

I'm not an expert on packaging, but it's possible that package leakage
allowed moisture to enter the device. Moisture then rots the aluminum
(or copper?) wire bonds and possibly the surface layers. If you're
curious, a helium leakage test will show if there are any leaks.


Currently *what* is reading 6.2 ohms? Forward resistance? What does
a good 1N5812 read? What's the reverse resistance? Got a curve
tracer?


I don't have any info on what's inside the diode, but if it's a bunch
of diodes in parallel, and some of them have opened thanks to internal
corrosion, my guess(tm) is that you'll see something resembling a
good, but much less than 20A rated rectifier.


Yeah, try non-destructive testing before you destroy it. Otherwise,
break it open and look inside before you melt the guts into a blob of
silicon.

--
# Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
# 831-336-2558
# http://802.11junk.com [email protected]
# http://www.LearnByDestroying.com AE6KS

DVM resistance either way round is 6.2
Ant chance of Beryllium oxide or whatever the serious nasty is , likely to
be inside?
Which plane to go in with a dremmel cutting disc?

 

[N_Cook]

I suspect tarnish of interior metals, especially if they contain any silver.
I see this sort of thing in Euro capacitors and metal connectors,
fuseholders, and the like. I also have to scrape it off the leads of old
semi's that have been in a parts drawer for years.

Mark Z.

Most oxides, sulphides, carbonates of copper silver lead etc seem to be
insulators except silver oxide. I 've not found AgO conductivity , could it
give 6 ohms over the geometries in a stud diode?

 

[N_Cook]

It's blown. No way was that caused by package leakage or corrosion
which would result in an open circuit. What you have is an amorphous
blob of silicon. It's also a single junction, so my theory about
parallel diodes is all wrong. Sorry(tm).


Berylium oxide is white and is used as an insulator. It looks like a
hard ceramic. I looked at the data sheet but no insulator material
was specified. It's a DO-4 case, which methinks is just molded
plastic. If it's white ceramic, don't grind.


Circumscribe through the plastic near the metal base. It's a "cup"
which should come off easily. You may need to cut the solder eye off
the anode lead to remove the "cup". Sometimes, the guts is filled
with silicon grease.

--
Jeff Liebermann [email protected]
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

I've never done that before. Dremmelled around the top of the can exposing
plug of ceramic or glass which easily came away. As did the central copper
pin , presumably the joint to the die failed with the Dremmel vibration.
The casing is presumably steel , but no corrossion on the outsde. But on the
inside of the can is about 90 percent covered in brown deposit, presumably
rust. No grease or anything like that inside. Before dremmelling, the
reading had changed to 10 ohm and it still reads 10 ohm between stud and
weld blob on the die. I wil dig out my microscope but I doubt I'd see any
sign of crept silver, a few atoms thick, especially as it is presumably on
the edge of the die. I'd have to grind off the rest of the cap to view the
side of the die.
Certainly no obvious burn marks on the shiny sliver and no smoke/sputter
trails. As this failed in storage, or theoretically at the point of last
switch off , then no such damage expected.
Is ferric/ferrous oxide conductive? if some off that could magnetically
shift and migrate to the die.

Any other suggestions what/how to explore further ? eg fine needle point and
DVM going over the die surface plotting ohmage

 

[N_Cook]

I've now removed the remaining cap cylinder and unravelled it and the patchy
rust film, or whatever, stops abruptly at where the glass seal was.

Much more interestingly viewing the die bond area to the base with a 30x
microscope. This appears to be a gold metalisation that has been eaten by
snails , if you've ever seen paper that has been in a leaking shed and so
grazed. Roughly linear tracks and then blobs of missing gold that seem to
have shamfered or grazed edges rather than the plating lifting and then
leaving sharp edges. One side of the die seems to have some of this gold as
a sort of powder in a few lines like iron filings over a magnet, rather than
solid metallic tin whisker needles, collecting there. Something seems to
have etched away the gold into a dust and then electrostatically or
magnetically drawn it to the die edge. It still measures 10 ohm so not a
will'o'whisp, frail accumulation . Perhaps I will take some microphotos
before probing the accretion with a needle. These grazing marks are roughly
midway between die bond edge and the edge of the support pillar.

So not an atomic level process of metal migration , so what is the process
that went on here presumably mainly when unpowered and very slow .

 

[N_Cook]

Some pics of all this, 1mm graph paper background

The sides of the cap unfurled, A-A is the distinct cutoff line of the rusted
part, the roughness at the ends is my butchery, also clean originally
http://home.graffiti.net/diverse:graffiti.net/1N5812_a.jpg

The green glass filled seal around the anode pin
http://home.graffiti.net/diverse:graffiti.net/1N5812_b.jpg

Its unaffected partner, as far as not being ohmic anyway
http://home.graffiti.net/diverse:graffiti.net/1N5812_c.jpg

View of the gold metalisation over the base metal stub, Os are below some of
the largest areas of erosion. I don't know where the missing gold went to ,
no flakes or obvious dust lying around in the diode when first broken into.
The weld for the anode pin is off-centre as can be seen in the pic, is that
relevant ?
http://home.graffiti.net/diverse:graffiti.net/1N5812_d.jpg

Edgewise view of the die , showing gold metalisation up to the die but much
is eroded away , eg base metal showing between C-C and the most obvious
accretion bridging the edge of the die marked with the Vs
http://home.graffiti.net/diverse:graffiti.net/1N5812_e.jpg

looking down on the large gold metalised areas with Bs showing the most
obvious bits of base metal, the bright line is the die edge
http://home.graffiti.net/diverse:graffiti.net/1N5812_f.jpg
the wavy gold edge ,near it, is probably metalisation over some underlying
pad so presumably at manufacture rather than a later formation

These diodes were stored anode upwards in kit , untouched for 13 years, so
something made the dust aggregate and migrate vertically , against gravity

it is now reading 13 ohms

 

[N_Cook]

I forgot to say these were made in 1977 and the image of the whole die on
the hex base is placed on an Al ring for the purpose of photoing

 

[N_Cook]

The first image of the can unfurled looks like the seal of the can allowed
moisture to enter and cause rusting of the interior/inside surface of the
can.

If not moisture infiltration, then I would suspect that the metal (or
plating/coating material) was incompatible with other materials inside the
can.

Another possibility could be that the metal used for the can was specified
to be extremely reliable, but the metal alloy/finish was flawed or otherwise
contaminated before assembly of the diode.

The design construction technique is flawed IMO if the can isn't more secure
than just a pressed-together fit.
It might be interesting to see if the military/areospace version of these
devices were fabricated by the same method.

Rust is conductive, and I would suspect that the diode failure was going to
happen eventually.

So much for long term reliability of heavy duty semiconductors.

Another possible ingress route is via the anode pin. When I cut through it,
it was in 2 parts and loose, a pin surrounded by a tube, both of copper and
somehow combined somewhere near the solder eyelet end. Maybe cap is
assembled after central pin welded to the die, then the cylinder placed
over, then glass seal and both copper sections fixed across at the eyelet.