Curtaining Artifact

Curtaining is a milling artifact that shows up in a focused ion beam cross-section as a series of vertical streaks running straight through the image, caused by the ion beam milling different materials at different rates as it cuts down through a chip's heterogeneous layer stack rather than by any real defect in the sample. A failure-analysis engineer trusts a FIB cross-section to show exactly what is physically inside a chip, so an artifact that looks like a defect but isn't is far more dangerous than one that is obviously fake — curtaining is the single most common way a perfectly good chip gets misread as having a structural problem it never had.

It is worse exactly where the engineer most needs a clean image: at material boundaries. A cross-section that cuts through alternating copper lines and dielectric, or through a grainy polycrystalline film next to a smooth single-crystal region, mills unevenly precisely because those materials remove at different sputter yields under the same ion dose — the streaks follow the beam's scan direction and tend to get worse the deeper the mill goes, because small unevenness at the top compounds as the beam continues cutting downward through an already-uneven surface.

Fixing it is entirely about how the mill is done, never about the sample itself. Depositing a protective cap layer of a single uniform material over the region of interest before milling gives the beam something mechanically consistent to cut through at the very top of the cross-section, rotating the sample or the beam's milling direction periodically breaks up the streak pattern before it can compound, and a final low-angle broad ion-beam polish after the main FIB cut smooths away residual streaking that the fine beam alone left behind — all three techniques attack the same root cause, uneven sputter yield across a heterogeneous material stack, rather than trying to compensate for it afterward.

Tool choice also matters, which is why FA labs increasingly mill with xenon plasma instead of gallium. Gallium-ion FIB systems from makers like Thermo Fisher Scientific and Zeiss are the long-standing workhorse for failure analysis, but their relatively narrow beam and limited current are part of why curtaining shows up so readily on large cross-sections; newer xenon-plasma FIB systems mill at much higher current and tend to produce flatter, less-curtained surfaces over a bigger area, which is part of why they have become the preferred tool for the large, full-die cross-sections modern multi-layer chips increasingly need for failure analysis.

Read curtaining through an *artifact-of-the-tool* lens rather than a *feature-of-the-sample* lens — every vertical streak in a FIB image is a question an analyst has to ask before trusting what they're looking at, because the chip underneath never had those lines in it, the ion beam put them there on the way to showing everything else.