Second Tin Whisker Workshop
The iNEMI held the second annual Tin Whisker Workshop the day
before ECTC started in Orlando May 31, 2005. Chaired by Ron Gedney and moderated by Henning
Leidecker from NASA about 100
engineers wrestled with the consequences of pure tin coatings
on electronic component leads. Pure tin has become one of the
common options now that the long successful Pb-Sn coatings are
discouraged for environmental reasons. Results of more than 9,000
hours of testing under various stress tests were discussed. Leidecker
started the session by saying this is a real problem that was
responsible for destroying a space craft 8 years ago. A whisker
caused an arc that became a 100 Amp event. He pointed out that
the thousands of fuses in the Hubble can arc uncontrollably with
only 25 V once the air leaks out of their package.
Valeska Schroeder of HP described the incubation time before
the whisker growth begins. The several accelerated test methods
turned out not to be all that accelerating however humidity, temperature
cycling, and sustained high temperature seemed to be an accelerant.
The humidity above 85% would cause condensation in a non-uniform
chamber and thus corrosion which was directly correlated to whisker
growth.
William J. Boettinger
of NIST described his study of the correlation of compressive
stress in the tin plated film with whisker growth. He also mentioned
that the columnar grain structure encouraged whisker growth whereas
the vertically multi-grain structure as seen in SnPb films suppresses
whisker growth. For example he found tin on copper leads has x10
the compressive stress of a plated Tin/Lead finish. Adding Cu
to Sn caused compression whereas Pb additions put it in tension.
He recommended pulse plating of tin to attempt to get horizontal
grain boundaries that slow down whisker growth.
George Galyon of IBM made a grand attempt to summarize
all the results of the study groups. First he stressed that compression
in the tin film is the driving force even in those few cases where
whiskers appear to grow in tension. He noted that at the tin-copper
interface, the copper diffused so fast that Kirkendall void developed.
This diffusion could result in build up of compression in the
tin film. However, if this tin film is kept thinner than 2 microns
the whisker problem is usually controlled. In contrast, tin on
nickel sees voiding in the tin and hence tends to put tin in tension.
Chen Xu of Cookson used X-ray data to confirm the state
of stress in the tin films. He reported that the tin in Sn/Ni
on copper leads are tensile after aging if not before. Also the
tin directly on copper leads get more compressive during aging
up to 20 MPa. The flexure beam work of S. Lal of FCI added confirmation
to these results although everyone was cautioned that the stress
state of the complete beam system was not the same as the stress
state within the surface tin.
John Lau mentioned that when performing the high temperature
surface soldering of the assembly line the resulting tension supplied
by the board may wipe-out any initial compression in the isolated
tin lead. The moral may be to assemble soon after tin plating.
Luu Nguyen mentioned that most high volume manufacturers
for consumer products prefer to keep using standard matte tin
that they have already qualified at great expense despite the
growing understanding from the iNEMI groups. It was also suggested
that since there is no grant money chasing the whisker problem
there have been few academic champions of this complex phenomena.
Asa Frye of IBM
used EBSD to obtain the find the local details of grain crystal
structure near the whisker base. This approach allowed one to
know the stress state of the grains in question and to confirm
the zonal structure analysis which tried to determine the stress
of each layer in a multilayer / multiprocessed structure. This
data added to the suspicion several group members have developed
that unintended copper contaminants in tin plating are often the
cause of unwanted stress.
Marc Dittes described the humidity tests performed. He
confirmed that condensation induced corrosion also was advantageous
for whisker growth. The difficulty of making chambers uniform
throughout means that high humidity tests result in condensation
that causes corrosion. He also confirmed that board mounting of
the lead stops whisker growth. Thus, he suggested that the component
should not be the level tested by the manufacturers, but rather
the final system configuration.
Peng Su of Freescale stated that the potential between
exposed Cu and Sn was the main driver for corrosion in condensation
and that this appeared a driver for much whisker growth. With
board mounted devices he had seen no whisker growth even with
pure tin remaining exposed above the solder wetted region. He
noticed that the bare tin was always coated with carbon compounds
coming from the flux residue, thus protecting the tin. It was
also noted that mechanical trimming of the leads caused local
compression regions even in otherwise tensile tin and corrosion
and whiskers soon followed. Sharp cutting tools caused less problem
than dull.
J W Osenbach of Agere mentioned philosophically that every
lead will eventually corrode since tin wants to become oxide.
He stressed that coupon tests miss the important edge effects
of leads. He showed how easy it was to see Kirkendall where there
are none. He also mentioned that once condensed water occurred
it took several weeks for it to evaporate in typical test chamber
conditions. Mobile devices could be seeing lots of condensation
in normal use.
Joe Smetana of
Alcatel presented an engaging visual image of how the whiskers
grow out of tin grains. Andre Egli of Rohm & Haas saw
whisker growth as a means of reducing compressive stress when
general extrusion is not allowed by the system constraints. The
SnO on the surface inhibits normal surface diffusion from relieving
stress. He showed how the surface morphology changed enormously
from the first 30 seconds after plating to 6 days later at room
temperature. In particular, all the initial screw dislocations
have disappeared. He showed a sequence of under the surface changes
that determined if whiskers would be encouraged or discouraged.
He saw the dynamics as a race between surface "fiberous"
structure formation that blended the grains versus whisker formation.