Date: August 12, 2005
To: LeadFree Forum
From: Jennie S. Hwang
RE: 96.5Sn3.5Ag-Properties & Commercial End-use Performance vs. SnAgCu
Recently, multiple queries have been received and some been posted in this
Forum regarding the subject topic: 96.5Sn3.5Ag--Properties & End-use Performance
vs. SnAgCu.
I am submitting the following input, outlining the basic properties,
mechanisms and product services in various industry sectors over the last 25-year of
SMT manufacturing. Within what I have been directly involved in term of
hands-on manufacturing and problem-solving, the specific end-uses of 96.5Sn3.5Ag have
fallen in power IC diode applications (attaching diode to heat sink with no
voids allowed), communication switch cards (SMT double-sided PCB assembly), and
passive component assembly, such as resistor network (packaging SIP and DIP
network on hybrid thick film ceramic substrates). Various production-related
and reliability concerns have been previously raised and resolved over the
years, including leaching and solder joint voids.
The alloy worked well on the production floor, as well as during the
products' service life under the <defined> process conditions and service
environments.
Overall, SnAg (E) has demonstrated great success in the above-mentioned
commercial applications. However, note that the production processes were not
constrained by the process temperature (< 230 oC for reflow) in those cases, due
to either the nature of the products or to the selected PCB material and the
whole system. The actual processes were conducted at a peak temperature in the
range of 250--260 oC, depending on a specific application.
The following outlines some general observations related to applications:
Because of its melting temperature at 221oC, SnAg(E) has been used in the
first-step of two-step reflow processes back in 1980s. Although SnAg (E) imparts
a higher wetting angle than 63Sn37Pb, the alloy's intrinsic wetting ability is
considered adequate under the compatible soldering conditions. Its surface
tension is very close to 63Sn37Pb.
If considering accelerated testing conditions, the results vary with a set of
testing parameters, as well recognized. However, with prolonged imposition of
cyclic strains (such as up to 5000 cycles, 0-100 oC with a reasonable range
of dwell and rate of temperature change, micro-cavities were noted along grain
boundaries, which is not a surprise. This micro-cavities along grain
boundaries should not be confused with solder joint voids that are normally observed in
the as-reflowed solder joints. In the latter case, voids are the result of
different causes (not to illustrate here). Voiding during the reflow process is
perhaps among the most intricate phenomena in SMT manufacturing, which however
can be minimized as shown in many demanding application, such as diode
attachment. In other word, the observed voids are not inherent from the alloy per
se.
Comparing the tensile strength at room temperature, 96.5Sn3.5Ag is still
lower than 63Sn37Pb. Nonetheless, SnAg (E) solder joints on PCB have performed to
the acceptable level in those applications.
As to leaching of the thick film conductors, the issue was well under control
by the selection of thick film composition and solder paste make-up in
conjunction with the proper process setting.
On SnAg's properties, characteristics and strengthening mechanism in creep
and fatigue, below is a snapshot:
During the solidification, two phases--Sn solid phase and e intermediate
intermetallic phase (Ag3Sn)-- form cooperatively from the homogeneous liquid (L),
i.e.,
L Sn + e(Ag3Sn)
Sn crystals are normally of white b tin with a body-centered tetragonal (BCT)
lattice structure. The e phase is homogeneous phase consisting of between
23.7 at.% and 25.0 at % Sn. The e phase at 25.0 at % Sn corresponds to the
stochiometric compound Ag3Sn, and has a slightly rhombically deformed hexagonal
close-packed (HCP) super-lattice with four atoms per unit cell.
The solid solubility of Ag in Sn is rather restricted due to the more stable
e intermediate phase, and is somewhere between 0.1 wt.% and 0.02 wt.%,
depending on how much the temperature is below the eutectic melting temperature. When
cooling temperature reaches below the eutectic tie-line, both Sn phase and e
phase (Ag3Sn) will be heterogeneously nucleated as the result of instantaneous
composition fluctuation in the liquid. When Sn crystals grow, excess Ag atoms
in liquid ahead of the Sn phase will be rejected. Ag atoms diffuse a distance
laterally and are then incorporated in e phase (Ag3Sn). At the same time, the
excess Sn atoms that are rejected ahead of the e phase (Ag3Sn) diffuse to the
tips of the adjacent Sn phase.
Generally, the Sn phase in the eutectic composition of SnAg (E) predominates
over the e phase (Ag3Sn). The volume fraction of e phase (Ag3Sn) in eutectic
microstructure can be estimated to be only 3.8 vol.%. The Ag solute rejected
from the Sn phase during Sn phase solidification must diffuse a much longer
distance to reach e phase (Ag3Sn) in comparison with the eutectic solidification
of Pb-Sn eutectic solder. Consequently, the Sn phase and the e phase (Ag3Sn)
have difficulty to grow cooperatively in a planar solid/liquid interface to
reach a lamellar or rod eutectic structure as frequently observed in 63Sn37Pb
solder. Instead, the enrichment of Ag in liquid ahead of the Sn phase may cause
an effective constitutive super-cooling. The constitutive supper-cooling can
then promote the growth of Sn phase in dendrites, and the e phase can finally
fill in the valleys of the dendrite arms. This is often termed as a <divorced
eutectic> where the e phase appears as isolated islands/particles (dark) and
the Sn phase appears as a dendritic matrix (light), under an
electron-microscope. The factors that control the formation of the divorced eutectic is likely
associated with entropy factor of melting.
Strengthening Mechanisms
Under load, the monotonic mechanical flow of solder alloys typically
consisted of an elastic region, a strain hardening region, a stress-recovery region,
and a cracking region.. Strain hardening continues until necking occurs at the
maximum load or the ultimate tensile strength (sTS). Necking is normally
caused by an inhomogeneous plastic deformation somewhere in the gauge length, and
it is associated with strain localization. Stress-recovery mechanisms are
believed to be dominant in the region after necking and before abrupt fracture for
high temperature deformation.
From the mechanical point of view, solder alloying is a strengthening process
in Sn-matrix. Like 63Sn37Pb, through established second phase strengthening
mechanisms, binary alloying always leads to a higher strength than pure tin
with a reduction in plasticity. However, necking or inhomogeneous plastic
deformation in both 96.5Sn3.5Ag and 63Sn37Pb occurred much earlier than that in pure
Sn due to the presence of heterogeneous solute atoms and second phases in the
alloy structure. Nevertheless, it is these strengthening "defects" that result
in a higher lattice friction stress and a higher strain-hardening rate,
therefore the higher strength.
To understand the strengthening mechanisms of SnAg (E), consider that the
yield strength (sy) of two-phase solder alloys generally follow the linear rule
of mixture in the volume fraction of the second phase (Ag3Sn) and the matrix.
Since there is essentially no solid solubility of Ag in Sn-matrix, the data for
pure Sn can be taken as a value at the volume fraction of zero. The linear
rule of mixture predicts that the yield strength (sy) of Ag3Sn intermetallic
compound should be around 756 MPa. Therefore, the Ag3Sn particles are viewed to
deform only in elastic region during the tensile flow of 96.5Sn3.5Ag. The sizes
of Ag3Sn particles are of the order of grains. The particle spacing is
significantly larger than the typical equilibrium radius of dislocation curvature.
Thus, the dislocations in Sn-matrix can freely pass by the Ag3Sn particles. The
strengthening effect of Ag3Sn particles is interpreted as the result of the
long-range internal stress built by the elastic modulus and volume differences
between Ag3Sn and Sn-matrix.
The shear flow strength of the 96.5Sn3.5Ag solder joint was lower than that
of the 63Sn37Pb solder joint, but the shear plasticity of the 96.5Sn3.5Ag
solder joints was much higher.
Creep Mechanisms
Creep is a plastic deformation process. The creep or plastic deformation of
solders at the typical application temperatures (-50oC--150oC) of electronic
packages falls in the regime of high temperatures, i.e., the operating
temperatures are above half the melting temperatures of solders. The creep mechanisms
can be identified by relating steady-state creep rate (ep) and stress (s) to
temperature (T) and structure, where the internal structure is dynamically
constant at a given stress and temperature. The phenomenological, physics-based
relationship for high- temperature plastic deformation mechanisms is a general
and dimensionless Dorn equation
(unable to include the equation due to multiple
symbols)
where D is the appropriate diffusion coefficient, E the Young's modulus, b
the dislocation Burgers vector, d the grain size (or interphase spacing), k the
universal Boltzmann constant, and A, n and p the characteristic constants. The
magnitude of the constants A, n, p and that of D reflect the particular
deformation mechanism that is controlling. The diffusion coefficient D is given by
D = Do exp(-DH/kT)
where the pre-exponential Do and the activation energy DH refer to a specific
diffusion mechanism, e.g., lattice self-diffusion DHl, grain boundary
diffusion DHg.b., solute diffusion DHs and dislocation diffusion DHdisl..
96.5Sn3.5Ag solder joints exhibited a power-law creep with the stress
exponent n=3.3 at the relative lower stresses between around 2 MPa and 10 MPa. The
power-law creep broke down at around 10 MPa towards the value (20 MPa) of the
shear strength for the solder joint, as indicated by the high stress exponent
n=11.2. In contrast, 63Sn37Pb solder joints demonstrated a power-law creep with
the stress exponent n=5.0 in the intermediate stresses between around 8 MPa
and 30 Mpa and the power-law creep broken down at the highest stresses with the
stress exponent n=19.4.
The creep rates of 96.5Sn3.5Ag solder joints were higher or their creep
resistance was lower than those of 63Sn37Pb solder joints.
The power-law creep indicates a plastic deformation mechanism with
dislocation climb or lattice diffusion as a rate-controlling process. The dislocation
glide such as cross slip is generally believed to participate in the
rate-controlling processes of power-law broken down creep. The creep data at different
temperatures and corresponding microstructures are needed in order to fully
specify the detailed operating mechanisms.
Fatigue Mechanisms
The fatigue life of 96.5Sn3.5Ag eutectic solder was quite comparable to that
of 63Sn37Pb, but much higher than that of pure Sn. For the second phase
strengthening of solder alloys, the logarithm of fatigue life (Nf) is expected to
follow the rule of mixture with the volume fraction of the second phase (Ag3Sn).
The Ag3Sn particles are generally considered to be among the most effective
blocks for fatigue crack propagation in Sn-matrix. In the meantime, finer
grains in Sn-matrix partitioned by the formation of Ag3Sn particles should also
extend the fatigue life by enhancing grain boundary gliding mechanisms. On the
other hand, the rule of mixture predicts that 20.5 vol.% Ag3Sn will be required
in the Sn-matrix in order to prevent Sn-Ag binary solders from the failure
under the fatigue conditions. At such a high content of Ag, however, the solder
will have a too high melting temperature to be useful for electronic
interconnections.
The overall fatigue life is controlled by the fatigue fracture capacity of
either Sn-matrix or inter-phase bonding. The fatigue lives of 96.5Sn3.5Ag solder
joints at about 10% shear strain ranges were equivalent to those of 63Sn37Pb
solder joints. When the shear fatigue strain amplitudes were approaching 20%
or over, the fatigue lives of 96.5Sn3.5Ag became much higher than those of
63Sn/37Pb. Below 10% shear strain range, however, the fatigue lives of 96.5S3.5Ag
were shorter than those of 63Sn37Pb.
At the large strain amplitudes, fatigue crack propagation is a dominating
event throughout the fatigue lifetime. In terms of the mechanisms identified,
Ag3Sn particles in 96.5Sn3.5Ag is a much more effective block for the fatigue
crack propagation than the Pb-rich second phase in 63Sn37Pb. Therefore, the
blockage of the Ag3Sn particles is considered the underlining factor for the higher
fatigue resistance at the large strain amplitudes.
At the small strain amplitudes, fatigue crack initiation is a dominating
event throughout the fatigue lifetime. Since the cyclic deformation in the process
of fatigue crack initiation almost entirely takes place in the soft Sn-matrix
in 96.5Sn3.5Ag, the Ag3Sn particles will play little role in retarding the
cyclic deformation damage or fatigue crack initiation, resulting in the lower
fatigue resistance at the small strain amplitudes.
The occurrence of crossover in the measurement of fatigue resistance suggests
that 96.5Sn3.5Ag solder joints are expected outperform 63Sn37Pb solder joints
in resisting the fatigue failure for electronics packaging designs when
requiring a larger expected cyclic strain range. However, 96.5Sn3.5Ag solder joints
are expected fail earlier than 63Sn37Pb solder joints at smaller in-service
cyclic strain ranges.
Comparative performance SnAg vs. SnAgCu vs. SnPb
For detailed comparison in materila properties including stress vs. strain
behavior and other mechanical performance, textbook: <Environment-Friendly
electronics--Lead-free Technology> (www.LeadFreeService.com):
Chapter 6--Binary alloys (Section 6.1 SnAg)
Chapter 8-SnAgCu
(Some portion of the abvoe has been posted on February 11, 2004)
Cordially
Jennie Hwang
Dr. Jennie S. Hwang
H-Technologies Group, Inc.
Tel: 216-839-1000
Fax: 216-896-0405 or 216-464-5728
Personal e-mail: [log in to unmask]
-----------------------
-------------------------------------------------------------------------------Leadfee Mail List provided as a service by IPC using LISTSERV 1.8d
To unsubscribe, send a message to [log in to unmask] with following text in
the BODY (NOT the subject field): SIGNOFF Leadfree
To temporarily stop/(start) delivery of Leadree for vacation breaks send: SET Leadfree NOMAIL/(MAIL)
Search previous postings at: http://listserv.ipc.org/archives
Please visit IPC web site http://www.ipc.org/contentpage.asp?Pageid=4.3.16 for additional information, or contact Keach Sasamori at [log in to unmask] or 847-615-7100 ext.2815
-------------------------------------------------------------------------------
|
