[tinwhiskers] Re: Response to numerous postings
- From: "Whittaker, Dewey (EHCOE)" <Dewey.Whittaker@xxxxxxxxxxxxx>
- To: <tinwhiskers@xxxxxxxxxxxxx>
- Date: Tue, 18 Mar 2008 08:59:04 -0700
Gordon,
It was nice to hear from you again. Good points and feedback.
Dewey
________________________________
From: tinwhiskers-bounce@xxxxxxxxxxxxx
[mailto:tinwhiskers-bounce@xxxxxxxxxxxxx] On Behalf Of
gordondavy@xxxxxxxxxxx
Sent: Monday, March 17, 2008 5:55 PM
To: tinwhiskers@xxxxxxxxxxxxx
Subject: [tinwhiskers] Response to numerous postings
This is a response to an amazing number of postings on March 13 and 14.
Apparently there has been a pent-up demand for the new tin whiskers
forum. I list below some comments from a number of those postings, and
my response. I will start with the general comment that it helps to keep
in mind that the real goal is not to prevent tin whiskers but to prevent
short circuits caused by tin whiskers.
I await someone who can do the math on how statistically significantly
conformal coatings "mitigate" tin whiskers. (Bob Landman)
Bob, youre going to have to wait longer than you can afford to get a
statistically significant answer to your mitigation question. First,
youre going to have to live with the fact that conformal coating will
not entirely prevent short circuits caused by tin whiskers. But that
really doesnt matter. The requirement is not that all such short
circuits be prevented, but that their frequency of occurrence is
acceptably low no higher than the frequency of occurrence of other
kinds of circuit failure. I believe, for reasons elaborated on below,
that this is a reasonable expectation.
If a whisker can grow from one pin on an IC package, then certainly, it
can also grow from adjacent pins as well and then don't we have the
perfect opportunity for shorts? (Bob Landman)
Two whiskers can grow through conformal coating and meet. The presence
of an electrical potential between the whiskers increases the
probability of their meeting, but as Henning has already commented,
theres still a question of whether they will actually meet, and if they
do, whether (because of their oxide coatings) they will cause a short
circuit. Keeping in mind my comment above, I believe that the
probability of two whiskers penetrating a conformal coating and meeting
is low enough for most high-rel applications.
any trace of silicon on the surface Raymond Bennett
Silicone on a surface will prevent the proper adhesion of a conformal
coating. However, the primary concern is how this lack of adhesion
prevents the coating from performing its primary function, which is to
adhere to the surface so as to prevent the opportunity for a continuous
layer of liquid water from forming. I dont think anyone knows how much
a poorly adhering conformal coating affects the probability of
penetration of a whisker from below, but in preventing penetration from
above (i.e., from the outside) it will probably work as well as a
well-adhered coating (see Joe Kanes point 5). For typical voltages (<50
V) the thickness required to prevent a short circuit is probably on the
order of tens of nanometers.
I seem to remember somewhere that the tips of the whiskers are so
thin, that they can bore right through a conformal coating...no matter
how well applied. Phil Shewmaker
Theres been no such evidence that I know of. Certainly a thin whisker
is sharper than a thick one, but it also buckles more easily.
there has not been adequate research into the use of ternary or
quaternary lead-free solder alloys for whisker mitigation John Wayt
The research on what elements reduce the tendency of tin to form
whiskers was done fifty years ago. The only element that has a
well-established ability is lead. Bismuth may have a small reducing
effect. Thats the complete list.
The tin crystallizes and grows Steve Smith
Tin is crystalline when it is electroplated, and it is crystalline when
it freezes. Theres no way to get amorphous tin.
Remember that we don't have an intact layer of tin oxide on the
surface. The thermal cycling continually shatters it, just as colliding
sheets of ice on the ocean break and reform creating a jagged
ice-sheet. Steve Smith
The time required for an oxide coating to start forming on a freshly
exposed layer of metallic tin is measured in microseconds. The layer
quickly grows to a few nanometers and then effectively stops growing.
Moisture has been shown to be a major driver of whiskering, so any idea
that any conformal coating can prevent whiskering I would hold up in the
air between thumb and forefinger and at arms length, and drop into a
high-humidity environment for a few thousand hours, and then come back
and gingerly poke it with a long stick. Steve Smith
Moisture does seem to accelerate whisker growth, but it is not
necessary. The greater concern for moisture should be what it does to
the properties of conformal coatings. They all absorb moisture to some
extent and therefore become softer. That still might not make any
difference on how easy it is for a whisker to penetrate from above.
As far as I am aware the promotion of use of conformal coatings to aid
RoHS issues were not to help tin whiskers, but stop oxidisation of
crystalline tin joints. Using a conformal coating stops the oxidation
process and improves short term reliability. Mark Vaughn
Conformal coatings are applied to prevent dendritic growth between
adjacent metalizations at different electrical potentials. It does not
stop oxidation, nor does it need to.
Tin loves to form a crystal structure, plumbers tell the difference
between tin and solder bar, by bending the bar and listening to it, the
crystalline structure of tin will crackle. Mark Vaughn
Pure tin makes a noise called tin cry when it is bent. That has no
relevance to anything being discussed in this forum.
With Pb joints the eutectic point, that is the potentially crystalline
phase during cooling lasts for about a 2 degree temperature change, with
a Tin based lead free solder that is around 39 to 42 degrees. Mark
Vaughn
I have no idea what Mark is intending to say.
This makes it very difficult to ensure a manually soldered joint is a
good one, especially when the joint looks so damn awful in lead free.
Mark Vaughn
Manually soldered solder connections have remained satisfactory for many
decades. If theres enough of it and the wetting to the surfaces to be
connected is satisfactory, their appearance (i.e., texture and luster)
is unimportant.
So there is a high probability that manual joints with contain
crystalline fishures that over time usually a few days to weeks will
oxidise and form a bad/dry joint. Mark Vaughn
Manually soldered connections do not contain fissures, and their
oxidation is virtually complete by the time the solder has cooled to
room temperature.
I expect it's more related to the electrical current flowing and
voltage charges across the tin lattice structure. Some crystalline
structures are not complete and are said to have acute voids, in which
nucleation sites become seeded, brought on by electrical charge to form
crystals that due to differences in angles of the crystal and void
become ejected, sounds like a whisker to me. Mark Vaughn
The relationship between potential or current and whisker growth is
tenuous at best. The tin plating does not ordinarily contain acute
voids (whatever that term might be intended to convey), and while
nucleation of whiskers is still not well understood, since metals are
electrical conductors, electrical charges are unlikely to have anything
to do with it.
Sadly this is not my field, trying to remember some college stuff 20
years old, and I don't remember the differences with between pure and
solvent crystalline grow, so I may be talking bunkum, but I don't think
I'm far off. Mark Vaughn
Solvent crystalline growth is irrelevant to tin plating.
I was also wondering if anyone has evaluated the effects of Hydrogen
Sulfide on these connections. Jacob Brodsky
I dont think that hydrogen sulfide has a measured effect on the growth
of tin whiskers (tin sulfide is not a very stable compound), but it has
a major effect on the growth of silver whiskers, as discussed at length
on the NASA whisker web site.
Are there conformal coatings that we could put in a purchase
specification that might help us here? Is it something we could do
ourselves after purchase? Jacob Brodsky
As discussed above and below, conformal coating is the only practical
mitigation technique available to the OEM other than solder dipping or
(in the case of chip components) replating.
<While the solder alloys are not pure tin, the soldering process does
not wet the entire surface of the parts that have become pure tin
plated.>
Wetting the entire surface is rare. Henning Leidecker
Small terminations are completely coated with a layer of solder and
hence if the solder contains lead they have very little risk of whisker
formation. Larger terminations are at risk of whisker growth from any
tin-plated area not reached by the solder. Even areas under but near the
edge of the solder can grow whiskers since the solder froze before it
could completely dissolve the tin.
If the solder has lead in it, and is applied as a hot liquid, then the
lead alloys nicely into the tin and no long whiskers grow. But
evaporating or electroplating some lead onto the pure tin is not so
effective at getting the lead into the pure tin. Henning Leidecker
Neither evaporating nor electroplating lead onto tin is a commercially
available process. There is a proprietary process for replacing existing
tin plating with electroplated tin-lead, and that is only practical for
chip components. That process does not leave any lead-free tin.
If the tin surface has already grown some whiskers, and then a leaded
alloy is flowed on, the whiskers can sometimes remain: they melt at 232C
(and the tough oxide on their surface can maintain them as needles
encased in a shell, at up to 260C and sometimes a bit more), while
eutectic lead-tin solder melts at 183C and might be applied at less than
250C. Henning Leidecker
While it is true that because of its oxide sheath (thin, but very
tough!) a tin whisker can survive heating well above the melting
temperature of tin, because of its very tiny cross section it will
dissolve instantly upon contact with molten solder.
Whiskers were able to push through THIN areas of our coating.
Warning: Conformal coating processes often suffer from incomplete
coating. For example, shadowing effects during spray application,
thinning of the coating from surface tension and other forces during
curing can leave you with less coverage than the process documentation
tells you that you have. Good idea to INSPECT and verify YOUR coating
processes ability to coat all areas of interest to you. Jay Brusse
Any conformally coated assembly will have thin areas (anywhere there is
a small radius of curvature as well as any shadowed area). Therefore it
is only practical to assume that some whiskers will grow on a
conformally coated assembly. I see no point of inspecting for thickness
since there remains no practical way to make the coating thick in the
small-radius-of-curvature regions. See my comments above about how much
of a concern thin coating is.
large grained 100% matte tin plating on copper-lead frames resists
whiskering even when exposed to compressive forces and held at elevated
temperatures in humid environments. Bright tin, however, had whisker
growth reaching 700 5m even at ambient temperatures. Extensive cyclic
compression tests with select matte tin chemistries, considered a
worst-case evaluation, yielded whisker growth of less than 20 5m. A
nickel barrier under tin helps prevent the formation of compressive
stresses caused by the growth of CuSn intermetallics and hence reduces
the one driving force for tin whisker growth. If a nickel barrier layer
is not possible, tin plating is annealed after plating for 1 hour at
1500C to promote controlled intermetallic compound formation and reduced
tin whisker risks. Patrick Lavery, quoted by Bob Landman
The problem any OEM has with the type of tin plating being received is
that it cant be inspected to find out if it is good (matte, large
grain) or not so good (bright, fine grain). Any assembly will contain
components from dozens if not hundreds of component manufacturers, and
even one manufacturer can have unnoticed changes in the tin plating
process from day to day. Similarly, there is no practical way to
determine whether a nickel underplate is present or whether the plating
had been reflowed within one hour. (What would you do if you knew?)
Of these finishes, ENIG has been shown to be the most robust for lead
free soldering. Patrick Lavery, quoted by Bob Landman
ENIG wont grow tin whiskers, but it can have failures due to black pad
and brittle fracture. Consider OSP or immersion silver.
A thick conformal coating (silicone-based polymer with 90% to 92%
filled ceramic particles) used in power brick products as part of a
thermal management strategyadds resistance to whisker growth. Patrick
Lavery, quoted by Bob Landman
The power brick approach seems to me to be a very effective technique to
prevent short circuits due to tin whiskers, provided the coating doesnt
have voids that span the space between tin-plated surfaces at different
electrical potentials.
Gordon Davy
Other related posts:
