Wolfgang,
As Albert says, many of the problems associated with wire
bonding arise from the quality of the materials, set up of the
machines, the design of the product and the bonding jigs being used,
with out knowing much about the returns you are getting (are they not
bonding or just having low strength etc?) here are a bunch of
comments.
PCB pads,
Need to be elementally "clean" on the surface, most
contaminants that lie ontop of a plating will either act as a
lubricant which will reduce bonding energy, but more importantly will
also run the risk of being held within the bond which can precipitate
voids and lead to early failure. Auger electron microscopy, is great
for this analysis, but expect to see loads of C and O anyway
Pad metals themselves need to be "pure" there are loads of
papers on this requirement, and reading between the lines they suggest
that the total impurities within a plated layer should not exceed 1%,
with each individual impurity not to exceed 0.1%, this is a massive
level of junk in a plating, and most good plating houses and board
suppliers should be able to maintain levels well below this. One word
of warning, dont allow H or Tl to get into the plated layers,
especially into Au, as they can lead to embrittled layers, and
outgassing during bonding or subsequent heat treatments.
Bonding metallurgy, care should be taken when plating a
single board that will have both Au and Al wires bonded to it, while
Au bonding likes nice soft thick Au, Al should only have enough Au to
protect the underlying bondable layer, this requirement for Al is due
to the famous Kirkendall voiding (previously known as purple plague)
which is related to the diffusion couples that operate between Au and
Al.
Total pad hardness, ideally this should be matched as close
as possible to the hardness of the wire material being used, if one is
harder, it will scrub away at the other and give a poor bond.
Machine set up,
This is probably out of your hands, but a well run DOE can
often take a lot of ambiguity out of the bonding process.
Make use of both the pull and shear tests. Pull will let
you know where the weakest point of the bond is, it will probably be
the heel of the bonds in Au, but the wire itself for the Al. If the
feet lift, there is a major problem, either with the bond parameters
or the plating quality. Shear testing lets you measure the actual
strength of the layers holding the bond together, and the failure
point should be in the foot of the wire itself, and never within the
pad or intermetallic.
Design.
Substrates are often not fully supported across their area,
this can lead to ulatrasonic attenuation and energy loss during wire
bonding. basically the pads vibrate, and the ulatrasonics cant do
their job, check that the pads are supported.
Wire geometry, and type
Wire shape greatly effects the reliability of bonds, too
long, too short, too tight, too high, its a mine field, but most
decent manufacturers of modules know what they are doing.
Wire materials vary, the Au will be doped, but does it have
a high elongation? if not, it may be work hardening too much during
step back and embrittling the wire. for Al, at about 10mil, there are
debates as to what is best 99.99 (4N) or 99.999 (5N) purity. I would
stick with 5N at this size, its more ductile, which gives better
flexibility during bonding, and it can support its own weight during
vibration etc.
Failure
There are really two known problems with wire bonds
1) they dont stick when bonded
- check pull and shear strength
Depending on point of failure review parameters
- listen during bonding, if there is a high spitch buzz,
there is vibration in the system somewhere, "seek and
destroy"
- getting the plating analysed, and look for high surface
contaminants
2) They fall off, or break during service
- Check the Metallurgy, are they compatible? not Al onto
thick Au
- Check pull and shear strength when bonded, and after
various heat treatments (identify what switches it on and
off and examine the joints closely
- If bond quality is good, and then drops off rapidly, then
its probably something to do with contamination within
the system (check the plating etc)
- Its not Kirkendall voiding is it?
- Are the wires breaking in the span? check wire geometry
wire type, bond quality, overbonding?
On the hole, FR4 should be easily bondable with both Au and Al
wires, providing that the pads are clean, the metalllurgy is
compatible, and the bonding process is under control.
I hope this helps in some way, if not read through Harmans, famous
book on bonding "Wire Bonding in Microelectronics" its got everything
covered, and should be the bible for most Bonding Engineers anyway.
Another good contact is Bob Clements at TWI in Cambridge, he knows his
stuff, and has an equiped lab to do some trials with.
Good luck
Roger
Roger Massey
Materials Technologist
Motorola AIEG
Stotfold
England
______________________________ Reply Separator _________________________________
Subject: Re: [TN] wirebonding
Author: "TechNet E-Mail Forum." <[log in to unmask]> at #email
Date: 13/07/99 09:02
Dear Mr. Wolfgang,
I think thatyou have to consider the parameters from the bonder and the
quality of PCB itself/ the die.
For the bonder parameter, you know exactly the time, pressure for the right
bonding performance and of course, the wear and tear of the aluminum wire
pointer will affect the bondability as well.
For the PCB, you have to specify to your board maker they must supply to
you boards which are ultrasonic bondable. The important factors for good
bondability PCBs are nickel thickness, nickel hardness, the gold layer and
the cleanliness of the surface going to be wire bonded. For the die, you
usually do not have to acknowledge since they are ready for bonding.
In my experience, poor bonding come from the PCB quality and the improper
adjustment of bonder.
Hope this help.
____________________________________________________________________________
_____________
At 03:10 PM 1999/7/12 -0400, Erat, Wolfgang wrote:
>Good Day
>
>Would appreciate any and all input on reliability data / experience /
>potential failuremodes when wirebonding Alu wire to electroless nickel /
>immersion Gold boards.
>
>HiRel Automotive application
>wedgebonding
>ultrasonic wire bonding
>10 mil AL wire
>Au 3 to 5 microns
>
>appreciate your response
>
>[log in to unmask]
>
>
Albert Mok
|