Krf (krypton Fluoride)
KrF (krypton fluoride) excimer lasers produce 248nm deep-UV light, and were the industry's workhorse exposure source from roughly 1996 to 2005 — carrying the transition from i-line mercury-lamp lithography into the deep-UV era, and driving the 250nm, 180nm, 150nm, 130nm, and 110nm technology nodes in turn. KrF works on the same excited-dimer physics as ArF (see our ArF article): an electrically excited krypton-fluorine gas mixture forms an unstable KrF* molecule that has no stable bonded ground state, forcing it to emit a 248.327nm photon and fly apart the instant it drops out of its excited state.
No modern chip is exposed entirely on one scanner generation — and KrF's continued production role is the clearest evidence of that. A leading-edge logic chip reserves its newest, most expensive EUV and ArF immersion tool time for the layers whose pitch genuinely demands it — typically gate and fin-level patterning. Contacts and vias, needing fine resolution but not always the absolute tightest pitch, often run on ArF immersion with multiple patterning. But implant layers, isolation structures, and relaxed-pitch backend metal don't need anywhere near that resolution — and running them on a $150M+ EUV scanner anyway would be pure waste. Those layers run on KrF tools instead, many of them fully depreciated decades-old equipment that costs a fraction as much to purchase and operate.
This is also exactly why KrF still anchors high-volume production at mature nodes. Many foundries run 28nm and 40nm production almost entirely on KrF lithography, because those node generations' design rules were defined with KrF's resolution in mind from the start. MEMS devices, compound semiconductor processes, and specialty analog/RF chips frequently don't need sub-100nm features at all, making KrF's lower cost per wafer a straightforward economic win rather than a compromise.
KrF's own resolution ceiling comes from the same Rayleigh relationship governing every lithography generation. With NA up to roughly 0.85 and a practical k₁ around 0.35, KrF alone resolves down to about 100–110nm in a single exposure — resolution enhancement techniques (OPC, phase-shift masks, off-axis illumination; see our articles on each) pushed select layers into sub-100nm territory, but KrF was never going to carry leading-edge nodes indefinitely the way ArF and EUV eventually did.
KrF resist chemistry was a genuine first, not just a smaller version of what came before. Polyhydroxystyrene (PHS)-based chemically amplified resist, developed specifically for KrF, was the industry's first generation of chemically amplified resist chemistry — the same acid-catalyzed deprotection mechanism (see our photoresist and post-exposure-bake articles) that every DUV and EUV resist since has built on. That amplification was essential precisely because early excimer sources weren't bright enough to expose non-amplified resist at production-viable doses; without PHS chemically amplified resist, KrF lithography itself wouldn't have been commercially practical.
Production specs reflect a mature, highly reliable platform: pulse energy of 10–40 mJ, repetition rates up to 4 kHz, bandwidth under 0.6 pm FWHM after line-narrowing, and reliability measured in billions of pulses between gas refills — numbers that explain why, nearly three decades after its introduction, KrF scanners are still running production wafers every day.