While a tremendous amount of work has been done in leadless solder joint reliability, the detailed thermal fatigue failure mechanisms and their relations with key parameters, e.g., solder fillet, inner end geometry, and microstructure, have, however, not yet been fully understood. And life prediction methodologies that account for the physics of failure still remain to be developed. In this presentation the failure mechanisms of 2512 LCR solder joints will be discussed based on thermal fatigue test results, and reliability models for fatigue crack initiation and propagation and the effect of solder height will be elaborated. During thermal cycling, leadless solder joints experience a complex stress and strain history. As a result, crack initiation and propagation occur and lead to the failure of the joint. The crack pattern and path depend on the location of crack initiation, fatigue resistance of solder, strength of interfaces, e.g, solder/termination or solder/pad, etc. Moreover, multiple cracks can also occur at various high stress/strain locations. Typical thermal fatigue failure mechanisms include: 1. The crack starts at the solder/termination junction at the inner end between LCR and PWB and propagates along the metallization interface of alumina/Pd-Ag or Pd-Ag/Ni. This generally results from a poor adhesion of these metallization layers. 2. The crack starts at the solder/nickel Junction at the inner end and curves into the chip resistor, leading to a early brittle fracture of the component. While not frequently seen, any pre-existing surface flaws on the component surface can lead to such an early failure. 3. The crack starts at the solder/nickel or solder/Cu-pad junction at the inner end and propagates through the joint under the component and into the fillet at 45 degrees or vertically. This is probably the most common failure mechanism seen in the test and has been widely documented in the literature. However, the present study shows that for this failure mechanism there can exist a three-dimensional effect in crack propagation and two different sub-mechanisms leading to similar fatigue life. The results presented here are a part of a series of studies conducted on the reliability of LCR solder joints under harsh automotive environments. Test boards on which 2512 LCRs were mounted with 62Sn-36Pb-2Ag solder were subjected to thermal cycling between -40 to 100 degrees C with different hold times and ramp rates. In the failure analysis, the crack length at different numbers of thermal cycles was measured and used to develop a life prediction model. The thermal cyclic stresses and strains in the solder joint were simulated by 2D and 3D nonlinear finite element modeling. A few observations can be made as follows: 1. For some joints there exists a 3D effect in crack propagation, i.e., the crack would tend to propagate in a direction not parallel to the long axis of the resistor, as shown in Figure 1. The results of three dimensional finite element modeling suggest that the crack initiation is associated with not only the shear and normal strains in the vertical plane, but with the transverse shear in the horizontal plane. 2. More than one failure mechanism have been observed. The dominant one corresponds to crack initiation at the solder-termination corner at the inner end and propagation through the solder joint. This mechanism can be further divided into two sub-mechanisms. One is associated with a short crack initiation life followed by a long propagation life, and the other, on the contrary, corresponds to a relatively long crack initiation life and a short propagation life. While these two mechanisms may be associated with different crack dominant parameters, the current test results st-tow that their fatigue lives are similar, i.e., crack initiation and propagation somewhat have compensated each other. 3. It was found that a linear relation exists between the number of thermal cycles and the crack length multiplied by the solder height. This gives a better estimate of the crack initiation life. A Coffin-Manson type of life prediction equation for crack initiation was developed based on the equivalent strain range at the inner end of the joint. It is noted that in order to accommodate the observed two failure mechanisms, the life prediction model for crack propagation is coupled with that for crack initiation. 4. The slopes of the linear relation between the number of thermal cycles and crack length multiplied by solder height are useful in developing life prediction models for crack propagation. The assumption of average shear strain range being independent of crack growth has led to a modified Coffin-Manson type equation with crack length explicitly shown in the equation, The total fatigue life is the sum of the crack initiation and propagation lives. 5. While the average shear strain range decreases with solder height, the local peel and shear strain ranges at the corner of the inner end increase with solder height (Figures 2 to fi), which indicate an opposite effect on the fatigue life as compared to the average shear strain range, This is attributed to the increase of the difference of bending curvature between the LCR and the PCB as solder height increases. A new design of resistor termination has been proposed to minimize these local stresses and strains. An optimal solder standoff in terms of reliability was also obtained based on the consideration of both crack initiation and propagation.
