10 Fundamentals of Bonding Technology and Process Materials …
319
regions due to their temperature-related activation. This causes a larger number of
atoms diffusing from hot to cold areas than atoms in the opposite direction. The
result is net diffusion (mass transport) in the direction of the negative temperature
gradients. Solder joints processed by TCB lacks IMCs at their interfaces and are not
able to handle the local Joule Heating efficiently compared to IMC rich solder joints
processed by reflow bonding. Therefore, Joule Heating and TM could have much
expedited the in situ diffusion kinetics of Cu and Ni atoms that eventually generates
voids, depletion of Cu bond pads, and defects (Fig. 10.58), resulting in the premature
failure of TCB-fabricated flip-chip packages.
However, it is unlikely that the IMC morphology is a direct cause of the EM failure
as there is a huge volume of β-Sn in the middle of the solder joint, and this Cu 6 Sn 5
IMCs are located mainly at the interfaces. Rather, a crystallographic orientation of
β-Sn should be taken into an account for a holistic diagnosis of a premature failure
of solder joints fabricated by the TCB process. Understanding the crystallographic
orientation of anisotropic β-Sn is important to fathom the bulk diffusion of Cu and Ni
atoms through the Sn crystals. Because of its anisotropy, diffusivity of Cu atoms are
500 times faster along the c-axis than along the other two orthogonal directions (aand b-axis) at 25 °C [61]. In addition diffusivity of Ni atoms are ~7 × 10
4 times faster
along the c-axis than along the other two orthogonal directions at 120 °C [62]. Note
that the temperature of the two-processed flip-chip packages during EM-aging test
is higher than both reported temperatures, 25 and 120 °C, hence the actual diffusion
rate will be much higher, especially for Cu atoms.
Effects of Crystallographic Orientation of β-Sn on EM Degradation. Inverse
pole figure (IPF) maps for all reflow- and TCB-joints are illustrated in Fig. 10.55.
The crystallographic orientation corresponding to each pixel in these IPF maps is
the orientation which corresponds to the vertical direction (direction of an electron flow). For example, a pixel corresponding to [001] configuration has its c-axis
pointing directly vertically. Hence, in this [001] configuration, the misorientation
between the vertical direction and the c-axis is 0°. This statement is confirmed by
visually comparing against the misorientation maps presented in Fig. 10.56. Crystals
with orientation close to the [001] configuration correspondingly have low misorientation angle values. Difference in misorientation values (spatial distribution of
c-axis) between the two types of joints is important as there is a faster diffusion of
Cu and Ni atoms when misorientation value is 90°. However, it is visually difficult
to discriminate between reflow and TCB solder joints from Fig. 10.60. Quantitative measures of the c-axis orientation are shown in Fig. 10.61. Although there are
some local differences in the angle probability density, there appears to be no significant differences. In addition, the mean value statistics are nearly identical across
the two process conditions. Therefore, it may be concluded from these results that
both bonding processes produce similar solder joint microstructures with respect to
c-axis orientation.
The c-axis orientation is only one measure of crystallographic orientation. Texture
information, represented in the IPF maps shown in Fig. 10.65, also includes additional
information about the secondary crystal orientations.
319
regions due to their temperature-related activation. This causes a larger number of
atoms diffusing from hot to cold areas than atoms in the opposite direction. The
result is net diffusion (mass transport) in the direction of the negative temperature
gradients. Solder joints processed by TCB lacks IMCs at their interfaces and are not
able to handle the local Joule Heating efficiently compared to IMC rich solder joints
processed by reflow bonding. Therefore, Joule Heating and TM could have much
expedited the in situ diffusion kinetics of Cu and Ni atoms that eventually generates
voids, depletion of Cu bond pads, and defects (Fig. 10.58), resulting in the premature
failure of TCB-fabricated flip-chip packages.
However, it is unlikely that the IMC morphology is a direct cause of the EM failure
as there is a huge volume of β-Sn in the middle of the solder joint, and this Cu 6 Sn 5
IMCs are located mainly at the interfaces. Rather, a crystallographic orientation of
β-Sn should be taken into an account for a holistic diagnosis of a premature failure
of solder joints fabricated by the TCB process. Understanding the crystallographic
orientation of anisotropic β-Sn is important to fathom the bulk diffusion of Cu and Ni
atoms through the Sn crystals. Because of its anisotropy, diffusivity of Cu atoms are
500 times faster along the c-axis than along the other two orthogonal directions (aand b-axis) at 25 °C [61]. In addition diffusivity of Ni atoms are ~7 × 10
4 times faster
along the c-axis than along the other two orthogonal directions at 120 °C [62]. Note
that the temperature of the two-processed flip-chip packages during EM-aging test
is higher than both reported temperatures, 25 and 120 °C, hence the actual diffusion
rate will be much higher, especially for Cu atoms.
Effects of Crystallographic Orientation of β-Sn on EM Degradation. Inverse
pole figure (IPF) maps for all reflow- and TCB-joints are illustrated in Fig. 10.55.
The crystallographic orientation corresponding to each pixel in these IPF maps is
the orientation which corresponds to the vertical direction (direction of an electron flow). For example, a pixel corresponding to [001] configuration has its c-axis
pointing directly vertically. Hence, in this [001] configuration, the misorientation
between the vertical direction and the c-axis is 0°. This statement is confirmed by
visually comparing against the misorientation maps presented in Fig. 10.56. Crystals
with orientation close to the [001] configuration correspondingly have low misorientation angle values. Difference in misorientation values (spatial distribution of
c-axis) between the two types of joints is important as there is a faster diffusion of
Cu and Ni atoms when misorientation value is 90°. However, it is visually difficult
to discriminate between reflow and TCB solder joints from Fig. 10.60. Quantitative measures of the c-axis orientation are shown in Fig. 10.61. Although there are
some local differences in the angle probability density, there appears to be no significant differences. In addition, the mean value statistics are nearly identical across
the two process conditions. Therefore, it may be concluded from these results that
both bonding processes produce similar solder joint microstructures with respect to
c-axis orientation.
The c-axis orientation is only one measure of crystallographic orientation. Texture
information, represented in the IPF maps shown in Fig. 10.65, also includes additional
information about the secondary crystal orientations.