 |
| |
 |
| |
 |
| |
 |
| |
 |
| |
 |
| |
 |
| |
 |
| |
 |
| |
 |
| EPSI INC. - Electronics Packaging Solutions International Inc. Contact us at: P.O. Box 1522, Montclair, NJ 07042 USA, tel.: +1 (973)746-3796, fax: +1 (973)655-0815 e-mail: jpclech@aol.com |
AUGUST 2004
"An obstacle-controlled creep model for SnPb and Sn-based lead-free solders"
by J-P. Clech, to be presented at SMTA International 2004, Chicago, IL, Sept 26-30, 2004. For further
info on the conference, visit: http://www.smta.org
ABSTRACT
This paper presents the application of physically-based, obstacle-controlled creep models to the analysis of
steady-state creep rates for eutectic SnPb and seven lead-free solders: Sn58Bi, Sn0.7Cu, Sn3.5Ag, Sn4Ag,
Sn3.8Ag0.7Cu, Sn3.5Ag0.75Cu and Sn2.5Ag0.8Cu0.5Sb. Lead-free, Sn-based solders are amenable to
such models since dispersed intermetallics, grain boundaries, the lattice of the tin matrix itself and other
dislocations, are all obstacles that impede dislocation motions. The basic single-cell, obstacle-controlled
creep model, after Frost and Ashby (1982), features a stress-dependent activation energy and an athermal
flow strength parameter which represents the maximum flow strength of the material at the absolute zero (0
Kelvin). Depending on how many creep mechanisms have been identified, the solder creep model is
formulated as a single- or double-cell model. In the latter case, creep rates for the two mechanisms are
additive. For each alloy, one set of steady-state creep measurements is used to determine the scaling
constants and the physical parameters of the model (4 or 8 constants in total for the single- or two-cell
models, respectively). Each model is then tested against independent test results (as many as nine test cases
for the Sn3.5Ag alloy).
The obstacle-controlled, solder creep models allow for the bridging of tension, compression and shear test
results as well as that of creep, strength and stress relaxation data, often without, and sometimes with the use
of a simple, multiplicative calibration factor. The models also allow for the prediction of solder joint
stress/strain measurements during thermal cycling of soldered assemblies. The need for calibration factors
suggests that creep models derived from a given mechanical test, and specimen type or size, should not be
used without justification in the stress/strain analysis of soldered assemblies. Model calibration and validation
is a critical step in the application of creep models to stress analysis or reliability models.
The use of obstacle-controlled creep models resolves many anomalies observed in the classical analysis of
lead-free solder creep data, including activation energies and power-law exponents that were found to be
stress- and/or temperature-dependent. In accord with experts who warned against the use of power-law
and hyperbolic sine creep models for engineering metals, we recommend caution when using such models for
stress or reliability analysis of lead-free assemblies. As demonstrated by bouncing the models in this paper
against many independent datasets, obstacle-controlled creep models offer a promising alternative.
Using the two-cell creep models, creep contour charts were generated to quantify the contribution of
competing creep mechanisms to the total creep rates. The patterns of creep contour lines are quite different
for Sn37Pb and Sn3.8Ag0.7Cu, a reflection of vastly different creep mechanisms. The Sn3.Ag0.7Cu creep
contour chart suggests a transition from one mechanism to another that is highly temperature related. The
transition occurs at about 75C, in agreement with microstructural and creep rate analysis conducted by
Vianco et al. (2004) on Sn3.9Ag0.6Cu solder.
"An obstacle-controlled creep model for SnPb and Sn-based lead-free solders"
by J-P. Clech, to be presented at SMTA International 2004, Chicago, IL, Sept 26-30, 2004. For further
info on the conference, visit: http://www.smta.org
ABSTRACT
This paper presents the application of physically-based, obstacle-controlled creep models to the analysis of
steady-state creep rates for eutectic SnPb and seven lead-free solders: Sn58Bi, Sn0.7Cu, Sn3.5Ag, Sn4Ag,
Sn3.8Ag0.7Cu, Sn3.5Ag0.75Cu and Sn2.5Ag0.8Cu0.5Sb. Lead-free, Sn-based solders are amenable to
such models since dispersed intermetallics, grain boundaries, the lattice of the tin matrix itself and other
dislocations, are all obstacles that impede dislocation motions. The basic single-cell, obstacle-controlled
creep model, after Frost and Ashby (1982), features a stress-dependent activation energy and an athermal
flow strength parameter which represents the maximum flow strength of the material at the absolute zero (0
Kelvin). Depending on how many creep mechanisms have been identified, the solder creep model is
formulated as a single- or double-cell model. In the latter case, creep rates for the two mechanisms are
additive. For each alloy, one set of steady-state creep measurements is used to determine the scaling
constants and the physical parameters of the model (4 or 8 constants in total for the single- or two-cell
models, respectively). Each model is then tested against independent test results (as many as nine test cases
for the Sn3.5Ag alloy).
The obstacle-controlled, solder creep models allow for the bridging of tension, compression and shear test
results as well as that of creep, strength and stress relaxation data, often without, and sometimes with the use
of a simple, multiplicative calibration factor. The models also allow for the prediction of solder joint
stress/strain measurements during thermal cycling of soldered assemblies. The need for calibration factors
suggests that creep models derived from a given mechanical test, and specimen type or size, should not be
used without justification in the stress/strain analysis of soldered assemblies. Model calibration and validation
is a critical step in the application of creep models to stress analysis or reliability models.
The use of obstacle-controlled creep models resolves many anomalies observed in the classical analysis of
lead-free solder creep data, including activation energies and power-law exponents that were found to be
stress- and/or temperature-dependent. In accord with experts who warned against the use of power-law
and hyperbolic sine creep models for engineering metals, we recommend caution when using such models for
stress or reliability analysis of lead-free assemblies. As demonstrated by bouncing the models in this paper
against many independent datasets, obstacle-controlled creep models offer a promising alternative.
Using the two-cell creep models, creep contour charts were generated to quantify the contribution of
competing creep mechanisms to the total creep rates. The patterns of creep contour lines are quite different
for Sn37Pb and Sn3.8Ag0.7Cu, a reflection of vastly different creep mechanisms. The Sn3.Ag0.7Cu creep
contour chart suggests a transition from one mechanism to another that is highly temperature related. The
transition occurs at about 75C, in agreement with microstructural and creep rate analysis conducted by
Vianco et al. (2004) on Sn3.9Ag0.6Cu solder.