[volt-nuts] PCBs with ceramic substrates

cheater00 cheater00 cheater00 at gmail.com
Sun Apr 16 11:27:06 EDT 2017


>
>
> > On Sun, 16 Apr 2017 13:17 Poul-Henning Kamp, <phk at phk.freebsd.dk> wrote:

> In message <87inm44nl4.fsf at devereux.me.uk>, John Devereux writes:
>
> > There is also the question of exactly what properties of FR4 are
> > limiting for "metrology" use.
>
> FR4 are certainly not without its own problems, in particular
> with respect to mechanical/electrical phenomena (bending, tempco etc.)
>
> And saying "Ceramic" isn't really narrowing down what we mean
> anyway, pretty much anything in the Rogers catalog is in game:
>
> https://www.rogerscorp.com/acs/products.aspx
>
> I suspect only experiments would be able to tell which of many
> possible desirable properties are most beneficial ?

On Mon, 10 Apr 2017 20:06 Poul-Henning Kamp, <phk at phk.freebsd.dk> wrote:
> Chuck Harris writes:
>
> > Google is your friend. Do some research on ceramics.
>
> We can add that the toughest materials, on pretty much all parameters,
> is what is becoming known as "metalic glass", which is just a fancy
> way of saying "ceramics made of metals".
>
>
> --
> Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
> phk at FreeBSD.ORG | TCP/IP since RFC 956
> FreeBSD committer | BSD since 4.3-tahoe
> Never attribute to malice what can adequately be explained by
incompetence.

Indeed your friend would have told you that the materials to be had are
Al2O3, AlN, BeO, Si3N4, or ZrO2.

The first is commonly found on old cooking pots or bicycle parts, and you'd
know that the whiteish film, while resistant to water, is easile malleable.
It has low fracture resistance, sometimes even if sintered. Its
monocrystalline version, sapphire, is not commonly used in manufacturing.
While it has excellent electrical properties, any experimental chemist will
tell you that sapphire windows, tiny bits of (synthetic) sapphire, are
prohibitively expensive and have to be handled with care; the largest ones
affordable to a well funded university are a couple inches across. Sintered
Al2O3 is used in spark plugs. I think it needs to be sintered into form and
cannot be milled or otherwise machined. Spark plugs also have huge gaps in
the metal casing due to different thermal expansion between ceramics and
metals such as say copper traces placed on a pcb. So even if you can solder
the board without fracturing either the substrate or the traces, just
keeping it in variable weather will fracture either one due to them working
all the time.

AlN is commonly used as a structural support in cheap heating elements, eg.
the dreaded "drop your iron and it's dead" soldering tip heating elements.
It's the reason everyone, including cheap Chinese manufacturers, moved to
including the heater directly in the tip as one exchangeable part. I've had
a tile stove with AlN heating elements inside and after 5 years the beads
housing the heating wire started falling apart on their own, without having
been touched once, while placed on a solid concrete base. This is likely
due to thermal shock, and goes to show objects out of AlN will have a
mechanically fragile structure on the interface of separate crystals that
make them up. But since ceramics are all porous they will all eventually
include other substances that will break the crystalline structure.

BeO is highly piezoelectric /and/ pyroelectric, which is just fuuun for
standards. This is due to its Wurtzlite crystalline structure, which it
shares with AlN. What it doesn't share with aluminum nitride is its ability
to give you cancer. It's so common it has its own name, berylliosis.

Si3N4 is ok. Except it covers in SiO2, silica, which is a thermal
insulator, so your chips can be nice and cozy. It's also nicely
piezoelectrical, which is why it's used in quartz oscillators. Quartz is
another name for silica. Si3N4 is also very hard, therefore fragile. So if
you shock mount it, it might be ok, otherwise just the thermal expansion of
whatever it's mounted to rigidly might stress it. Such as the copper
traces, which you have to balance out on both sides, or you're risking
fracture. It is commonly used in ball bearings. It can't really be milled.
So you only get the pre-made sizes. If you keep the board in SF6 or mineral
oil all the way from manufacture through use, it might be good, but don't
quote me on that.

Anyone who's ever owned one of those cheap tomato knives out of white
ceramic knows how easily ZrO2 chips when it is in a thin structure.

To know most of this stuff you'd have to have cooked for yourself or been
outside on a bike a few times, and googled for the other stuff, but I guess
making padding for a mail sig takes its toll, so I won't attribute this one
to malice.

A big issue with ceramics of all kinds is their thermal expansion
coefficient. It is incompatible with copper. Here's a company extolling
their revolutionary copper cladding process:

"Unlike the fragility of thick and thin film circuits, the DBCu circuits
allow rough handling and can endure repeated eutectic bonding temperatures.
The selection of the ratio of copper thickness to ceramic thickness is an
important consideration in the initial design of the circuit. One sided
bonding on thin substrates can cause severe camber conditions that can
degrade the bond with repeated temperature cycling due to shear and
bimetalic type bending forces or lead to ceramic fracturing. These effects
can be reduced by double sided bonding to equalize the forces between
opposite faces of the ceramic. Leaving copper in non-electrical areas will
not only improve lateral heat .spreading but also minimize bending stresses
during temperature cycling."

This is a new thing, so apparently neither wide spread nor available back
them fragile ceramic pcbs encountered by people on the list were being made.

Another issue is that they are porous. So anything floating in the air will
find its way into the inside of the crystal lattice, and because that is
ridgid it will wedge the ceramic apart, like an axe. This is my guess as to
why old ceramic boards fail. Note all ceramic pots are glazed. Glazing
wasn't always known. If you go to a museum and look at artefacts you will
only find glazed ceramics and no unglazed ceramics from that same period.
It's not because people suddenly stopped making unglazed ceramics - glazing
was expensive - but because the unglazed ones disintegrated over time,
unlike glazed ones. It's really lucky when you find unglazed ceramics, and
it's usually due to them sitting for a few centuries in some sort of mud
that was chemically compatible with them. Usually the same local mud they
were made out of. This should tell you a lot about the lovely long term
robustness of porous ceramics. They have nothing on dense organic layers
such as used in the manufacture of "normal" pcbs.

Watch this video to see a guy who's handy at repairing stuff find a ceramic
pcb isn't the "toughest material, pretty much on all parameters":
https://youtu.be/_x8ii58-W_Y (it's around 13:20). This is just cheap HP
junk which lost nothing in a metrology lab, but still. I guess a
manufacturer who actually cares about quality would have made the board out
of sapphire.

Al2O3, AlN, BeO, and Si3N4 substrates can be had in sub $50 MOQ from
alibaba, and you can get Zr2O3 bricks, so go for the experimental route
Paul. As far as experience goes hands on is loads better than arm chair.

I've investigated both ceramic guns and engines that were brought up
earlier.

I quickly remembered the only ceramic gun is the Glock 7. I live near the
Glock factory. The Glock 7 is a fully ceramic gun used to avoid detection
by metal detectors in Die Hard 2. It doesn't exist outside of Die Hard 2.

The thermal expansion coefficient, britlleness due to lack of plasticity,
and thermal insulative character are all reasons why we have no ceramic
engines. The only search result talking about fully ceramic engines as a
practical idea is a crackpot site called "why ceramic rotary engines" that
lists different kinds of ceramic materials and brings up a pish tosh of
random nice facts about them, many of which don't even apply to building
engines, such as thermal shock resistance. Unless you expect your engines
to be jet engines that is, in which case I guess that is right, because
NASA once made a ceramic jet nozzle that was a few inches big.

Google is indeed a good friend, but I guess he's shy to bring stuff up with
people who use a royal "we".

>


More information about the volt-nuts mailing list