Hard Mask Open
The hard-mask open — HM Open or ACL Open depending on the maker — etches a pattern into a sacrificial amorphous carbon, polysilicon, or silicon nitride layer so that layer can then stand in for the photoresist during every demanding etch that follows it, because ordinary photoresist simply cannot survive the long, aggressive etches a DRAM cell module actually needs. This single step repeats at nearly every patterned level across a DRAM cell array — active area, buried wordline, bitline, storage node hole — which is exactly why it's listed as a step "used in every cell module" rather than tied to any one feature.
The entire reason this extra layer exists is that photoresist alone would simply disappear before the real etch finished. A DRAM cell module's deep, long-duration etches — the kind that cut a buried wordline trench or an active area line — expose the masking material to far more plasma energy and far longer exposure time than resist is built to survive, so the pattern gets transferred one level down into amorphous carbon (ACL) first; amorphous carbon can take that punishment and still hold its shape, trading one extra etch step up front for a masking material that will actually still be there, with its edges intact, by the time the real pattern-defining etch finishes.
The specific gas chemistry used to open an amorphous carbon hard mask has to balance etch rate against selectivity to everything around it. Oxygen-based chemistries, sometimes blended with SO2 or COS, strip carbon efficiently by converting it to volatile gases, but that same reactivity has to be tuned carefully so the etch opens the hard mask cleanly without also attacking the oxide or nitride layers beneath it or the photoresist pattern still sitting above it during the transfer — getting this selectivity wrong either leaves hard-mask residue that blocks the next etch or erodes supporting layers the process depends on staying intact.
This step is also where SADP and SAQP multi-patterning schemes begin, since the hard mask it opens often becomes the mandrel the rest of the pattern gets built around. In a self-aligned double or quadruple patterning flow used to print features finer than a single lithography exposure can resolve on its own, the hard-mask-open step defines exactly the core lines that later spacer and mandrel-pull etches will use as their reference — a hard-mask open etch tuned even slightly off-target here propagates its error through every subsequent multi-patterning step that depends on it.
Read the hard-mask open through a *borrowed-durability* lens rather than a *just-another-pattern-transfer* lens — the pattern itself doesn't change at this step, what changes is which material has to survive everything that comes next, and amorphous carbon, poly, or nitride simply outlasts photoresist at the punishing etch depths and durations a DRAM cell module actually requires.