Modeling and Experiments on an Isothermal Fatigue Test for Solder Joints

1Stephen Ridout, 2Milos Dusek, 1Chris Bailey, 2Chris Hunt

1Centre of Numerical Modelling and Process Analysis
University of Greenwich
Email: s.w.ridout@gre.ac.uk

2NPL Materials Centre
National Physical Laboratory
Queens Road, Teddington
Middlesex, TW11 0LW
UK

Abstract
This paper investigates an isothermal fatigue test for
solder joints developed at the NPL. The test specimen is a
lap joint between two copper arms. During the test the
displacement at the ends of the copper are controlled and
the force measured.
The modeling results in the paper show that the
displacement across the solder joint is not equal to the
displacement applied at the end of the specimen. This is
due to deformation within the copper arms. A method is
described to compensate for this difference.
The strain distribution in the solder was determined by
finite element analysis and compared to the distribution
generated by a theoretical ‘ideal’ test which generates an
almost pure shear mode in the solder. By using a damage-
based constitutive law the shape of the crack generated in
the specimen has been predicted for both the actual test
and the ideal pure shear test.
Results from the simulations are also compared with
experimental data using SnAgCu solder.
1. Description of the fatigue test
This test is intended to quantify the fatigue resistance
of solder joints under conditions similar to those
encountered in real-world use.
The test specimen consists of a SnAgCu lap joint
between two copper arms (figure 1). The machine which
the specimen is placed in is capable of accurately
controlling the displacement at the ends of the copper
arms and measuring the applied force.
In the test a periodic displacement profile is applied to
the sample causing the solder to deform and become
damaged due to ductile fatigue. As the solder becomes
damaged, its ability to resist further deformation is
reduced and so the force measured will decrease. When
the maximum force per cycle has dropped to 50% (or
some other arbitrary percentage) of the original maximum
force per cycle, then the test is considered complete and
the joint is assumed to have failed. In this way, the fatigue
resistance (cycles to failure) of the solder joint can be
quantified for different magnitudes of applied strain.

Fig 1: The test specimen. A SAC joint between two
copper arms
2. Modeling the test
The purpose of the modelling work presented is
threefold:
1. It is necessary to establish that the displacement
being generated across the solder joint is as
prescribed.
2. To determine the nature of the solder deformation for
three different notch depths (see figure 2) and for the
ideal test. This will show us how close the
deformation mode is to pure shear.
3. The load-drop curve and crack shape in the solder
will be predicted for an actual test specimen and for
the ideal test. Simulation results will be compared
with experimental data for a specimen with no notch.



Fig 2: The Finite Element mesh

All the modeling work has been conducted using a
Finite Element code. Figure 2 illustrates the mesh density
in the region of the solder joint. The copper has been