Packaging · Topic 9 of 14
0%
Wafer-level packaging
Building the package while the dies are still on the wafer — smaller, thinner, and often without a separate substrate.
What you’ll learn
- What wafer-level packaging (WLP) is
- How it differs from traditional packaging
- Fan-out WLP and its trade-offs
Why it matters
WLP enables the very small, thin packages phones and wearables need, and underpins fan-out and much of advanced packaging.
Explanation
In traditional packaging, dies are cut from the wafer first and then packaged one by one. In wafer-level packaging (WLP), the packaging steps are done while the dies are still on (or reconstituted onto) a wafer, and the wafer is diced at the very end — so many packages are built in parallel.
In the simplest “fan-in” WLP, connections stay within the die's own footprint, giving a package barely larger than the die (chip-scale). Fan-out WLP re-embeds dies in a moulded wafer and adds redistribution layers (RDL) that spread connections beyond the die edge — more connections without a separate substrate.
Advantages: smaller, thinner packages and, for fan-out, no substrate cost and shorter connections. Limitations: warpage and yield are harder to control, and the connection count is limited by the redistribution layers rather than a full substrate.
Visual explanation
Key terminology
- RDL (redistribution layer)
- A thin metal layer that re-routes a die's connections to new positions.
- Fan-in / fan-out
- Keeping connections within the die footprint, versus spreading them beyond it.
- Chip-scale package (CSP)
- A package barely larger than the die itself.
Example
Common mistakes
Watch out for:
- Assuming WLP is just “a smaller package” — fan-out WLP is a different process that packages many dies in parallel on a wafer.
Real-world application
Fan-out WLP became mainstream in smartphones and is a stepping stone toward chiplet-style integration.
Explore related concepts
Continue learning
Wafer-level packaging
Up next
2.5D integration
Packaging
Placing multiple dies side-by-side on a shared interposer.
Continue learning →