Challenges For Achieving Automotive Grade 1/0 Reliability In FCBGA and fcCSP Packages

As the quantity, complexity, and functions of electronic devices in automobiles increase, understanding and characterizing package reliability is of significant concern and importance. The Automotive Electronics Council (AEC) Q-100 specification for Grade 1 and 0 reliability introduces unique challenges as thermal cycling (TC) and high temperature storage (HTS) requirements increase. Additionally, high operating temperatures and the need for 15-year reliability with zero-defect quality often require evaluating package reliability to as much as twice the conventional AEC Q-100 grade requirements. Meeting these challenges requires solutions in the areas of materials characterization, simulation, and design of experiments that satisfy the needs for both having a comprehensive understanding of device reliability and a competitive development timeline.

Automotive packages increasingly require greater integration of device functionality for sensors, advanced driver-assistance systems (ADAS), infotainment, in-vehicle and between-vehicle networking. These requirements are driving demand for larger package sizes with advanced design rules not previously utilized in automotive packages. Simultaneously, the mission profiles for automotive packages are expanding with the development of self-driving capabilities, electric charging, and other technologies. This requires certain components to effectively be always on and operating for extended periods at high temperatures.

Automotive reliability requirements

Table 1: AEC Q-100 Temperature Cycle and High Temperature Storage Requirements

Table 1 lists the AEC Q-100 reliability requirements for grades 0-3 along with the device operating temperature range used as a guideline to determine the correct grade for a given device [1]. Both Grades 0 and 1 extend the temperature cycle range from -55 °C up to 150 °C. This large temperature range introduces high thermomechanical stresses, especially in large body size Flip Chip Ball Grid Array (FCBGA) packages. High temperature storage requirements up to 1000 hours at 175°C increase concerns over failure modes such as the growth of intermetallic compounds (IMC) in the die to substrate interconnects along with degradation of organic materials.

FCBGA evaluations

Multiple simulations were performed modeling a range of properties for FCBGA materials and evaluating the resulting stresses for several failure modes including metal pad stress, stress in the Low-k dielectric, the interfacial stress between the solder mask and underfill, and room temperature warpage.

Sensitivity plots created from the simulation results indicate the relative impact of material property variations for each material on the stress for each failure location. Figure 1 shows the stress sensitivity plot for metal pad stress indicating that the buildup and underfill modulus and coefficient of thermal expansion (CTE2) of the underfill have the largest impact on this stress.

Fig. 1: Stress sensitivity plot for metal pad stress.

The results of the simulations indicated that the buildup, solder resist, and underfill were critical materials that required characterization to understand how their material properties changed after aging during temperature cycling. Samples were created for these materials and subjected to temperature cycling betting -55 and 150°C. Modulus, CTE and the glass transition temperature were measured before cycling and after 1000 and 2000 temperature cycles.

Figure 2 displays the results of modulus measurements from the temperature cycling property characterization. In this case, the solder resist showed a significant increase in modulus below the glass transition temperature relative to the other materials.

Fig. 2: Modulus measurements before and after temperature cycling.

Based on the results of the simulation and aged material property measurements, a test vehicle was designed to evaluate selected substrate and underfill materials expected to be suitable for Grade 1 and 0 package reliability. The test vehicle included a 35×35-mm substrate and 15×15-mm die with 150-µm pitch copper pillar. Figure 3 shows the results of 175°C HTS on the underfill fillet. The high temperature exposure leads to oxidation and cracking in the underfill fillet. Figure 4 (a) shows that the high temperature exposure leads to IMC growth and solder voiding. Figure 4 (b) shows the results of HTS in a substrate with electroless nickel electroless palladium immersion gold (ENEPIG) surface finish which eliminates this issue.

Fig. 3: Underfill fillet after 1000 hours 175°C HTS.

Fig. 4: 175°C HTS results: (a) Substrate with Solder on Pad (SOP) finish and (b) Substrate with ENEPIG finish.

Figure 5 shows the results from a previous test on a similar 35-mm substrate to Grade 0 conditions using the 150°C HTS condition. Neither of the failures observed for Grade 0 packages in 175°C were present at 150°C with no underfill cracking or solder voiding being detected [2]. This suggests that the 1000-hour exposure at 175°C is much harsher than the alternative Grade 0 requirement of 2000 hours at 150°C.

Fig. 5: Results from 150°C HTS: (a) Underfill after HTS with no cracking and (b) FC bump after HTS with no voiding.

fcCSP evaluations

In flip chip Chip Scale Packaging (fcCSP) packages, the primarily problem encountered when testing to Grade 0 and 2xGrade0 conditions relates to the oxidation of the epoxy molding compound (EMC) molded underfill which can result in cracking after extended HTS. Figure 6 shows an example of this failure with a crack extending from the surface of the EMC down to the substrate after 2xG0 HTS at 150°C.

Fig. 6: EMC cracking after 2xG0 HTS.

To screen alternative EMC materials before requiring sample builds and reliability testing, samples of multiple materials were created and subjected to HTS at 150, 175, and 200°C. Aged properties were measured to identify which EMC materials had the highest likelihood of resisting cracking in HTS. Modulus, CTE, shrinkage, flexural strength, and the thickness of the oxidized EMC layer were measured at intervals up to the 2x Grade 0 HTS requirement.

Results from the oxidation measurements are shown in figure 7 indicating different EMC materials have different oxidation rates and characteristics. In EMC A, there is a thick fully oxidized layer of EMC after 150°C HTS while EMC B shows a thick layer of fully oxidized material with a large reaction layer. After 200° HTS, EMC B has a much thinner layer of fully oxidized material.

Fig. 7: Oxidation measurements after 2000 hours HTS.

Modulus measurements taken from the aged EMC samples show that the transition from glassy to rubbery increases over a wider range of temperatures. Dynamic mechanical analysis (DMA) modulus measurements from one EMC material are shown in Figure 8.

Fig. 8: Modulus measurements from aged EMC.

The large variation in modulus between the aged and non-aged EMC material at the Grade 0 HTS temperatures of 150°C and 175°C is a suspected contributor to the EMC cracking observed.

A 12×12-mm test vehicle was designed for testing multiple EMCs selected for Grade 0 based on the aged material property measurements. EMCs were selected for their low shrinkage and oxidation behavior along with one high glass transition temperature EMC. Three EMC candidates were successfully tested to 2x Grade 0 conditions at both 150°C and 175°C with no EMC cracking and no failures detected in temperature cycling.

Conclusion

Simulation and material characterization have been shown to be important steps in developing and evaluating new material sets for automotive packages. This allows the identification of critical parameters and potential materials before building test samples for reliability testing. This process reduces the time required to solve problems that will arise for future automotive packages as they increase further in size and complexity.

Reliability testing has shown that the Grade 0 high temperature storage conditions of 1000 hours at 175°C and 150°C for 2000 hours are not equivalent for all failure modes, with the 175°C requirement being much harsher and resulting in material failures that do not occur at 150°C.

For the FCBGA packages, a robust G1 BOM has been developed. The same BOM meets Grade 0 temperature cycling requirements but eliminating the underfill cracking that occurs at 175°C HTS remains a challenge.

The fcCSP package has passed qualifications to 2xG0 with multiple EMC materials thanks to early material characterization used as a tool to identify suitable EMC candidates.

References

1. AEC – Q100 Rev – H: Failure Mechanism Based Stress Test Qualification For Integrated Circuits (base document).
2. Dias, M. Kelly, D. Balaraman, H. Shoji, T. Shiraiwa, K.S. Oh, J.Y. Park, “Challenges and approaches to developing automotive grade 1/0 FCBGA package capability,” 2019 IEEE 69th Electronic Components and Technology Conference (ECTC).