I guess that's me!
Most points have been covered, but a few points to clarify.
In these sorts of discussions it is easiest to think of indium as tin's lazy
cousin.
Tin dissolves gold rapidly. So in a tin/lead solder you tend to have hard
tin/gold intermetallics adjacent to a lead rich layer. This will crack when
stressed. The stress levels required can be very low. A gentle tap can be
sufficient in extreme instances. The problem is reduced slightly and
counter-intuitively with high tin alloys such as SAC. The more concentrated
tin dissolves the gold even faster but as there is 50% more per joint the
gold is diluted further.
While tin dissolves gold rapidly, indium being the last cousin does so
slowly. The difference is about 12:1
For the same sort of reasons Indium makes solder joints more slowly than
tin. This needs to be kept in mind when selecting /formulating fluxes for
indium alloy soldering.
So indium can make solder joints to gold surfaces. Note that indium does
dissolve gold even if only slowly, things speed up/change when they're
hotter, so for most high indium based alloys a max operating temperature of
125C is recommended.
Tin makes brittle /hard intermetallics with gold. Indium has a similar
relationship with copper. Pure indium will solid state diffuse into copper,
even at room temperature, it will do so indefinitely even at room
temperature till all the indium has gone. The resultant intermetallic is
hard and brittle. [The interface can end up looking as though the mythical
tin worm has eaten it]. Pure indium is rarely used of course and bare copper
is easily avoided.
So when looking at soldering thick gold with tin alloys or copper with high
indium am alloys to me the answer is: Get rid of the Sn/In or the Au/Cu.
IF possible plate with 4-5 um Nickel and you can use any alloy you like
IF you can't change the gold surface, then you have to change solder
Getting rid of tin is easy enough, just use a Pb based alloy instead. The
technical problem then is high melting point. (See below on Au/Sn)
IF you need a low melting point without tin then you are usually looking at
indium instead. There are any number of Indium lead combinations.
Personally I prefer 60In/40Pb. This has a similar Melting range to tin/lead
so represents a minimum change solution. Joints formed by indium/lead alloys
have excellent thermal cycling properties which is another reason for using
them.
Gold tin is an alternative. This makes a uniform very high strength bond
witch is difficult to break.
The MP of 80Au/Sn is 280C. This puts reflow over 300C and most soft solder
fluxes will thermally degrade during reflow, giving a hard to remove
"caramelised" residue. You'll get a solder joint OK but it might be hard to
see. Flux removal will be difficult. For these reasons nitrogen inerting is
good, it's not necessary to get ultra low PPM as we just want to protect the
flux not the joint formation. A low oxygen atmosphere will simplify
cleaning. The high MP also gives long profile times and cooling down can
also be an issue. At high temps most organics - substrates etc - are pretty
delicate.
Costs
Tin is sold by the ton[ne]
Indium by the pound/kilo
Gold by the ounce.
Enough said!
Availability
In spite of the high MP 80Au/Sn is processible by some solder companies
pretty much as Sn/Pbs. So fine powder paste is possible and quite common.
Solid wire is possible from a few companies, flux core is not possible.
No one can afford bar solder in this alloy
.
When looking at availability of Indium alloys as solder paste we need to
recall another interesting property of indium: perfectly clean indium will
cold weld to itself faster than you can say superglue. This clearly has
implications when trying to make oxide free indium (alloy) powder.....
So very fine powder in high indium alloys is.... just lets say "difficult".
It will almost certainly be made to order in small quantities from a very
limited supplier base, making for relatively high prices. (But still nowhere
near Gold Tin.) Check lead times also as these will vary according to the
time of ordering. Solid wires are common to order, flux core no.
When using indium alloys especially as pastes, all the rules about handling,
shelf life etc, processing conditions/control are more true and really do
need to be observed.
Regards
Mike Fenner
Bonding Services & Products
M: +44 [0] 7810 526 317
T: +44 [0] 1865 522 663
-----Original Message-----
From: TechNet [mailto:[log in to unmask]] On Behalf Of Inge Hernefjord
Sent: Friday, March 29, 2013 3:50 PM
To: [log in to unmask]
Subject: Re: [TN] solder which does ot scavenge gold
Hi Guy,
Uve already got guru's advice, so maybe I don't add much of interest. Only
tell you that we had a nightmare with Indium soldering to gold. The result
was rhat we banned Indium for the next coming 1,000 years. Why? I'll brief
you and if you want deeper insight, you should contact the old english
Indium man. Nobody beats him, natural after half life at Indium.
So what...
Independent of what Indium alloy you use, the first phase when soldering
with SAC or other high-Tin solders, is the formation of a AuIn2 layer. This
takes only seconds. The process eats at least 1 micrometer of gold. If the
gold is thinner, there will not be a correct IMC and the component will
loosen when thermomechanically exposed. So, remember: the gold that you
solder to, must be thicker than 1 um!
Furhermore, if your total solder joint is thin, let's say 15 um or thinner,
AND you have a thick gold, there is a risk of a second IMC, namely Au9In2.
It's not ternary, just one single phase and is known for its brittlement
and having overall lower data than AuIn2. If the soldering top temperature
is too long, you may even get Kirkendall voids,which makes Au9In2 even more
brittle.
I hope I remember the above correctly. I worked with tat problem about 30
years ago!
Inge
PS there is a an old paper in my Dropbox under 'Soldering....' The
headline is " Aspects on Indium Soldering...' I have better reports, but
don't know where they are..
On 29 March 2013 06:27, Guy Ramsey <[log in to unmask]> wrote:
> Background info:
> We were asked to populate and assembly with a Ma-Com part, MA46H120, a
GaAs
> Constant Gamma Flip-Chip Varactor Diode. The data sheet says:
>
> Mounting Techniques - These chips were designed to be inserted onto hard
or
> soft substrates with the junction side down. They can be mounted with
> conductive epoxy or with a low temperature solder preform. The die can
also
> be assembled with the junction side up, and wire or ribbon bonds made to
> the
> pads.
>
> Solder Die Attachment - Solder which does not scavenge gold, such as
> Indalloy #2 (80In-15Pb-5Ag) is recommended. Sn-Pb based solders are not
> recommended due to solder Embrittlement. Do not expose die to a
> temperature
> greater than 235C, or greater than 200C for longer than 10 seconds.
>
> The Indalloy #2 cost $2,222.00 for 100gm or type five solder paste.
>
> We now have another customer asking for us to solder to thick gold. They
> don't have a low temp requirement and asked for SnAu solder (Indalloy
#182)
> 80Au-20Sn. . . I shudder to think what that will cost, and I don't think
> the
> part, an inductor, will survive the 300C reflow.
>
> Question:
> Does SAC 305 "scavenge gold"? Would it form a brittle solder connection on
> a
> thick soft gold pad?
>
> Guy
>
>
>
>
>
>
> ______________________________________________________________________
> This email has been scanned by the Symantec Email Security.cloud service.
> For more information please contact helpdesk at x2960 or [log in to unmask]
> ______________________________________________________________________
>
______________________________________________________________________
This email has been scanned by the Symantec Email Security.cloud service.
For more information please contact helpdesk at x2960 or [log in to unmask]
______________________________________________________________________
______________________________________________________________________
This email has been scanned by the Symantec Email Security.cloud service.
For more information please contact helpdesk at x2960 or [log in to unmask]
______________________________________________________________________
|
