Laser industry news, technology and market intelligence
LasersNewsPost
Ultrafast & Precision Micromachining

Stealth dicing wins on kerf, ablation wins on tolerance for the unexpected

Modifying material inside the wafer wastes no width and produces no debris. It also demands a substrate that behaves exactly as modelled.

By LasersNews Desk··1 min read
Close-up of a scientist examining samples under a microscope in a laboratory setting.
Photo by Tima Miroshnichenko on Pexels

Separating semiconductor die has two laser approaches. Ablation removes material along the street, cutting a groove. Stealth dicing focuses pulses inside the wafer to create a modified layer, after which tape expansion cracks the wafer along it.

The case for stealth

Kerf width is zero. Nothing is removed, so no street width is consumed, which on a wafer of small die directly increases die per wafer.

There is no debris and no ablation plume, so no particulate contamination and no need for protective coatings or post-dice cleaning. For MEMS, image sensors and devices with exposed structures, that is decisive.

The process is fast, since the beam scans without removing material, and it generates little heat at the surface.

The case for ablation

Stealth relies on controlled crack propagation from the modified layer. That requires predictable material behaviour: consistent thickness, known crystal orientation, and a wafer free of stress that would divert the crack.

Wafers with heavy metallisation, thick dielectric stacks, test structures in the street or bonded layers complicate that. The crack must pass through everything, and materials that do not cleave cooperatively cause deviation or incomplete separation.

Ablation does not care. It removes what is in front of it, whatever that is. For heterogeneous stacks, ablation — often combined with mechanical sawing — remains more robust.

The hybrid

Many production flows use both: ablation to remove metallisation and dielectric in the street, then stealth through the remaining silicon. That captures most of the kerf benefit while handling the difficult layers directly.

What decides

The choice follows the wafer, not preference. Thin, clean, uniform silicon with narrow streets favours stealth strongly. Thick, heterogeneous, metallised wafers favour ablation or hybrid approaches.

As packaging pushes toward thinner wafers and narrower streets, stealth's share has grown — but the exceptions are structural rather than transitional, and both processes will keep their places.

This article was produced by the LasersNews AI desk and reviewed by our editors.

Related reading