TL;DR
This study investigates the mechanical "warpage" of lidded flip-chip chip-scale packages (fcCSP), a critical factor in manufacturing yield. By comparing experimental data with finite element analysis (FEA), the researchers found that substrate core properties are the most influential factor in controlling package flatness, and that traditional simulations often struggle to predict the behavior of non-molded packages due to complex residual stresses.
What Problem Does This Paper Address?
In the world of high-performance electronics, flip-chip technology is king because it allows for more connections in a smaller footprint. However, as these packages get larger (17mm and above), they face a major mechanical hurdle: warpage.
Because a package is a "sandwich" of materials with different Coefficients of Thermal Expansion (CTE)—like silicon, organic substrates, and plastic molding—it bends and twists as it heats up and cools down during manufacturing. If a package warps too much, it won't sit flat on a circuit board (poor coplanarity), leading to soldering defects like "non-wet opens" or "solder bridging." While engineers use computer simulations (FEA) to predict this, these models often oversimplify the manufacturing process, leading to a gap between theory and reality.
Methods
The authors took a dual approach, combining rigorous experimental testing with simplified FEA modeling:
- Experimental Test Vehicles: They built two sizes of lidded fcCSP (19x19 mm and 21x21 mm) and varied several parameters: die thickness, substrate core materials, and molding compounds (MC).
- Measurement Techniques:
- Optical Inspection: Used to measure "coplanarity" (how flat the solder balls are) at room temperature (25 °C).
- Shadow Moiré: A laser-based technique used to measure warpage at the peak reflow temperature (245 °C).
- C-SAM: Acoustic microscopy used to check for internal defects like voids in the underfill.
- Simulation: They used Ansys to create a quarter-symmetry model. Notably, they employed an "element birth and death" technique to simulate the assembly sequence, attempting to capture how residual stresses build up as each component (die, underfill, mold, lid) is added.
Datasets
This paper does not use public machine learning datasets. Instead, it generates a primary experimental dataset:
- Sample Size: Coplanarity data was collected from 100 samples per group, while warpage data at reflow was collected from 10 samples per group.
- Configurations: 14 distinct experimental groups (7 for each package size) were tested to isolate the effects of specific material changes.
Key Insights and Ideas
- Substrate Core is the Primary Lever: The most effective way to control room-temperature coplanarity is by modifying the substrate core material. Switching to a core with a specific CTE (Material C2 in the study) cut warpage nearly in half compared to the baseline.
- The "Lid Dominance" Effect: While thicker dies and molding compounds generally help keep a package flat at room temperature, the trend can reverse at high temperatures. Once the organic materials exceed their glass transition temperature ($T_g$), they soften, and the stiff metal lid begins to dominate the package's shape.
- The MUF Challenge: Molded Underfill (MUF) is efficient but prone to "voids" (air bubbles) if the pressure isn't perfect. The study found that these voids cause massive variations in warpage, making MUF groups much less predictable than those using Capillary Underfill (CUF).
- Simulation's "Blind Spot": FEA models were reasonably accurate for fully molded packages (within 35–50 µm) but failed significantly for bare-die packages. This is because the simulation's assumption of a "stress-free temperature" doesn't account for the complex relaxation that happens in a substrate before it is "locked in" by molding material.
Key Results
- 21x21 mm Package: This larger size proved challenging; only one experimental group (Group 3, featuring a specific substrate core) successfully met the JEP95 industry standards for coplanarity.
- 19x19 mm Package: Because this package uses larger solder balls, the industry requirements are more relaxed. All experimental groups for this size easily met JEDEC standards.
- Warpage Trends: The study confirmed that for lidded packages, warpage at 245 °C is generally "positive" (crying face shape), while room temperature coplanarity is more sensitive to the specific balance of the substrate and mold properties.
Limitations and Future Work
The authors acknowledge that their simulation still has gaps. It excludes cure shrinkage (the way plastic shrinks as it chemically hardens) and spatial non-uniformity (how filler particles in the plastic might settle unevenly).
Future work should focus on:
- Integrating chemical shrinkage models into the FEA workflow.
- Better characterizing the "actual" stress-free state of the substrate after lamination but before assembly.
- Optimizing the Film Assisted Molding (FAM) process to eliminate the voids found in MUF configurations.
Summary updated