รายละเอียดบล็อก

How can a wafer be thinned to ultra-thin levels?
What does “ultra-thin wafer” mean?

Typical thickness definitions (8"/12" wafers)

• Standard wafer: 600–775 μm

• Thin wafer: 150–200 μm

• Ultra-thin wafer: < 100 μm

• Extremely thin wafer: 50 μm, 30 μm, or even 10–20 μm

Why thin wafers?

• Lower total stack thickness, shorten TSVs, and reduce RC delay

• Lower electrical resistance and improve thermal dissipation

• Satisfy ultra-slim product requirements (mobile, wearables, advanced packaging)

Main risks with ultra-thin wafers

1. Dramatically reduced mechanical strength

2. Increased warpage (stress-induced bow/warp)

3. Challenging handling (pickup, transport, chucking, alignment)

4. High vulnerability of front-side structures, leading to cracks and breakage

Common approaches to achieve ultra-thin wafers

1. DBG (Dicing Before Grinding)
   The wafer is partially diced (scribes are cut deep but not fully through), so each die outline is defined while the wafer still behaves as a single piece. The wafer is then back-grinded to the target thickness, progressively removing the remaining silicon until the residual layer is ground through, enabling clean die separation with improved control.

2. Taiko process (rim-retained thinning)
   Only the central area is thinned, while the outer rim is kept thick. The retained rim acts as a reinforcement ring, improving stiffness, reducing warpage risk, and making handling more stable during downstream processing.

3. Temporary wafer bonding (carrier support)
   The wafer is temporarily bonded to a carrier (a “temporary backbone”), transforming a glass-paper-like fragile wafer into a manageable, processable assembly. The carrier provides mechanical support, protects front-side features, and buffers thermal/mechanical stress—allowing thinning to tens of microns while still enabling demanding steps such as TSV processing, electroplating, and bonding. This is a foundational enabler for modern 3D packaging.