Line Cut Etch

The line cut — also called chop etch or trim — takes the continuous parallel lines that self-aligned spacer patterning just produced and chops them into the separate, electrically isolated islands a DRAM cell actually needs, because a line by itself is never the final feature: active areas and landing pads both need to be discrete, bounded shapes, not an unbroken stripe running the full length of the array. Everything before this step has been about getting a line's width and pitch exactly right; this step is where that line stops being a line at all and becomes a sequence of separate, individually addressable pieces.

Continuous spacer-defined lines
correct width and pitch, but unbroken end to end
Cut mask patterned perpendicular to the lines
defines where each line gets severed
Chop etch removes material at the cut locations
turns lines into isolated islands
Discrete active-area or landing-pad islands
ready for the next module

The cut mask that defines where the line gets severed runs in the opposite direction from the lines themselves, which means this etch has to respect two different overlay budgets at once. The lines were already placed by the mandrel-and-spacer sequence, tuned for pitch along one axis; the cut pattern comes from an entirely separate lithography step oriented perpendicular to that axis, and the cut's own placement tolerance along the line's length has to be tight enough that the resulting island neither comes out too short, losing contact area, nor leaves behind a sliver bridge that fails to fully separate two cells that were supposed to be isolated from each other.

From Continuous Line to Isolated Island a perpendicular cut mask severs each line at defined points before the cut unbroken lines, full array length after the cut cut band discrete, isolated islands too short a cut leaves a bridging sliver; too long shrinks the island's own contact area

Which feature this cut ultimately forms, active area or landing pad, depends entirely on where in the flow it's used, even though the etch mechanics are essentially the same. An active-area cut severs the diagonal active lines that self-aligned quadruple patterning produced into the bar-shaped silicon islands each transistor sits on, while a landing-pad cut separates adjacent pads built later in the flow from accidentally touching each other — both uses share the same underlying chop-etch logic, which is why the list of DRAM etch applications treats "cut" and "chop" as the same family of step even though they appear at different points in the process and cut through different materials.

Because this etch has to clear material completely at the cut band without disturbing anything on either side of it, it's tuned for sharp, well-controlled sidewall profile in both directions it runs across. High-resolution etch platforms from Lam Research and Tokyo Electron, used for the same mandrel and spacer steps feeding into this one, run the cut etch with chemistry selected to clear quickly through the cut band while stopping cleanly, since any residual material left in the gap defeats the isolation this step exists to create, and any overshoot into the island on either side costs real contact area the next module depends on.

Read the line cut through a *where-the-line-stops-being-a-line* lens rather than a *finishing-touch* lens — every prior step in this patterning sequence was about getting a continuous line's dimensions right, and this is the one step that decides the line was never actually the goal, the discrete pieces it gets chopped into are.

Take line cut etch further

Ask the copilot about this term, or have our engineers assess it against your process.