Darc Process

The DARC process deposits a dielectric anti-reflective coating — typically SiON or silicon-rich nitride — by CVD rather than spinning it on, and it suppresses reflections by precise destructive interference rather than by the absorption mechanism the BARC process covers. Both films solve the same underlying swing-curve problem, but DARC's domain is the opposite of BARC's: it wins specifically where substrate reflectivity is well-controlled and uniform enough that one precisely tuned thickness can actually hit its target everywhere it's deposited.

SiON/SiRN target film
Refractive index tuned by composition
CVD deposit at quarter-wave thickness
Thickness = λ / (4 × n)
Destructive interference cancels the reflectiontwo reflected waves, half a wavelength out of phase
Reflection off DARC top surface
Partial reflection at the air/DARC interface
Reflection off DARC/substrate interface
Travels an extra λ/2 round trip — cancels the first
Net reflectance near zero
At the design wavelength, for this one thickness
Often doubles as hard mask
Dielectric, dry-etchable, high thermal budget

The physics behind DARC is ordinary thin-film destructive interference, the same mechanism behind anti-glare coatings on eyeglasses, tuned to one specific thickness rather than left to vary the way an uncontrolled resist film's thickness does. Light hits the top of the DARC film and partially reflects immediately; the rest transmits through, reflects off the substrate underneath, and travels back up through the DARC film a second time. If the DARC film's thickness is set to one-quarter of the exposure wavelength divided by the film's refractive index, that second reflected wave ends up exactly half a wavelength out of phase with the first — and two waves of equal amplitude, half a wavelength apart, cancel each other out almost completely. Get that thickness right, for a known refractive index and a known exposure wavelength, and reflectance at that specific wavelength drops close to zero.

DARC: Two Reflections, Cancelled by Phase quarter-wave thickness puts the two reflected waves exactly half a wavelength out of phase Substrate DARC — thickness = λ / 4n incident light reflection ① — off DARC top reflection ② — off substrate, extra round trip adds λ/2 phase shift ① and ② are equal amplitude, opposite phase — they cancel, net reflectance ≈ 0 this only works precisely at one thickness and one wavelength — exactly the opposite trade-off from BARC's broad absorption tolerance

That precision is exactly why DARC needs a well-controlled, uniform substrate to work as intended, and exactly why BARC remains the better choice wherever substrate reflectivity varies significantly across the chip. A quarter-wave cancellation is tuned against one assumed substrate reflectivity and phase relationship; deposit the same DARC thickness over a region with meaningfully different underlying reflectivity, and the destructive interference condition shifts away from ideal, letting more light reflect back into the resist than the design intended. BARC's absorption mechanism tolerates that kind of substrate variation reasonably well because it doesn't depend on a precise phase relationship at all; DARC's interference mechanism performs best — often better than BARC — specifically where the substrate underneath is reliably uniform, which makes DARC the natural choice over a blanket metal or hard-mask layer rather than over a chip's full range of varying pattern density.

DARC's dielectric, inorganic composition also lets it take on a second job BARC's organic chemistry can't: serving as part of the hard mask stack itself, rather than being stripped away before the real pattern-transfer etch even starts. An organic BARC is consumed or removed early in the flow, the same way the top photoresist layer in a trilayer stack eventually is; a dielectric DARC film, being chemically similar to the hard mask materials covered in the APF and spin-on carbon processes, can survive — and actively participate in — a subsequent dry etch, meaning the same layer that suppressed reflections during exposure can go on to help transfer the pattern into whatever lies beneath it. That dual-purpose capability, combined with DARC's higher thermal budget tolerance compared to an organic film, is why a CVD dielectric ARC is often the right choice specifically when the layer underneath also needs a mask material that won't decompose under later process heat.

Read the DARC process as the interference-based mirror image of BARC's absorption-based approach: the same swing-curve, reflectance-control problem, solved with the opposite physical mechanism, and each one's ideal operating domain is defined by exactly the condition the other one struggles with — DARC thrives on controlled, uniform reflectivity and tolerates thermal budget BARC can't; BARC thrives on the reflectivity variation that would push DARC's precisely tuned interference condition out of range.

Take DARC process further

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