Access Official Technical PDF

Share Summary:LinkedInEmail

Electromigration reliability is an important design consideration in wafer-level chip-scale packages (WL-CSPs), where miniaturization and higher current density can increase the risk of solder-joint failure. In these packages, current crowding near the interface between a solder bump and its under bump metallurgy (UBM) can promote void formation. Three-dimensional finite element analysis and design of experiments (DOE) offer a way to assess how joint geometry affects predicted time to failure (TTF).

A simulation study of WL-CSP solder joints examined the combined effects of UBM rim angle, diameter and thickness, as well as solder bump shape. Its results point to geometry as a practical design variable: different configurations produced different predicted void growth behavior and TTF. The findings are specific to the modeled package and conditions, but illustrate how simulation can help compare candidate designs before fabrication.

Why electromigration threatens solder joints

Electromigration is the movement of atoms under the influence of electrical current. As atoms redistribute, some regions can become depleted. If depletion reaches a critical level, a void can form and grow. In a solder joint, the region next to the UBM is particularly important because the geometry of the electrical contact can concentrate current there.

That concentration, known as current crowding, creates a local condition that can encourage void formation. As a void expands across the solder/UBM interface, it can compromise the joint. The challenge for package designers is therefore not just to select materials, but also to understand how the UBM and bump geometry shape local electrical, thermal and mechanical conditions.

How the electromigration reliability model works

The modeled approach tracks changes in normalized atomic density over time. Rather than considering only the force associated with electron flow, its formulation accounts for several contributions to atomic transport:

  • Electron wind force, associated with the momentum transfer from moving electrons.
  • Stress gradients, which influence atom movement through differences in local stress.
  • Temperature gradients, which affect transport across regions at different temperatures.
  • Atomic density gradients, which drive redistribution as density varies within the material.

The analysis couples thermal, electrical and structural calculations with a diffusion model. First, a three-dimensional finite element model calculates current density and temperature distributions. The atomic density redistribution is then evaluated over time. When the modeled density falls below a defined threshold, elements are deactivated to represent void growth. The study uses a 15% increase in bump electrical resistance as the package failure criterion for calculating TTF.

This distinction between void initiation and package-level failure matters. A local void indicates damage has begun, while the resistance criterion provides a defined endpoint for comparing predicted service life across configurations. The model’s simulated failure times were reported to agree with test mean time to failure results overall, although a difference at one current condition was noted. That qualification underscores the importance of validating material parameters and modeling assumptions.

What the geometry study found

The DOE varied UBM rim angle across seven settings from 0° to 90°, UBM diameter across three values, UBM thickness across three values, and bump shape across four configurations. The simulations were run at 1.7 A. Across the evaluated cases, the voids appeared in the solder bump adjacent to the UBM, consistent with the modeled current-crowding location.

UBM angle, diameter and thickness

UBM rim angle had a substantial effect on predicted void growth and TTF. In the modeled layout, a 75° rim angle produced the longest TTF. The 0° and 90° configurations performed least well against electromigration in the comparison. This result should be treated as a geometry-specific optimum, not a universal angle for every package design.

The simulations also showed a longer predicted electromigration life as UBM diameter increased. For the thickness comparison, conducted with a 30° rim angle, thicker UBM layers corresponded to longer TTF. These trends provide useful directions for design exploration, while actual decisions still need to account for the complete package architecture and applicable manufacturing constraints.

Solder bump shape

Bump shape also changed predicted TTF. The comparison included spheroid, cylinder, cube and double-ball geometries. The cylinder produced the longest reported TTF at 20,458.6 hours, followed by the cube at 15,941.7 hours and the spheroid at 1,947.2 hours. The double-ball configuration had the shortest result, at 945.3 hours.

The study describes the cylinder, cube and single-ball shapes as having the same volume and height, while the double ball had twice the volume of a single ball. These results indicate that shape can influence modeled electromigration life; they do not establish that a particular shape will be best under all operating and design conditions.

Applying the findings to package design

For engineers and system integrators, the central implication is that solder-joint geometry should be assessed as part of the electrical and reliability design—not treated as a purely mechanical detail. A useful evaluation can compare UBM dimensions and rim angles alongside bump shapes, then examine where current density, temperature and predicted atomic depletion concentrate.

Simulation can help narrow design options and identify potential weak locations, but its conclusions depend on the geometry, material properties, boundary conditions and failure criteria used. Comparing model predictions with test data is therefore important, particularly when results will inform a production design.

Conclusion

Electromigration reliability in WL-CSP solder joints is closely tied to local conditions near the UBM interface, where current crowding can contribute to void formation. The reported simulations show that UBM rim angle, diameter, thickness and bump shape can affect predicted time to failure. For practical design work, these findings support using geometry-focused simulation and validation to compare package options and reduce the risk of relying on a single assumed joint configuration.

Access Official Technical PDF

About The Author