Knowledge base · Technical article

Yellow Light Room Lighting: What Wavelength Actually Protects the Resist?

A yellow room does not block UV; it removes the band that exposes the resist. DNQ-novolak photoresists absorb from roughly 320 to 450 nm, so filters are normally specified to cut everything below about 500–520 nm, leaving amber light around 570–590 nm. The exact cut-off belongs to the resist and the dose budget, not to a lighting standard.

Updated 2026-11-09·8 min read·BOWI Lighting

Why does ordinary white light expose a resist that is not being exposed?

Photoresist exposure is a dose, not an event. The film accumulates actinic energy as D = I × t, and once the accumulated dose passes the material’s critical dose, the resist develops whether or not a mask was in place. A room that looks yellow to the eye can still deliver that dose, because the eye is weighted at 555 nm and the resist is weighted at 365–436 nm. This is the reason a lux reading says nothing about photolithography safety: a lux meter is blind to the only part of the spectrum that matters.

Published measurements make the scale concrete. In one recent controlled comparison, a white-lit room fully developed a standard positive resist in 0.67 hours, and a legacy “yellow” room did it in 1.0 hour — visually compliant, functionally useless for a wafer held under that light for a shift. A properly filtered system pushed the same test beyond 223 hours. Resist handling guidance states the same thing from the other direction: white fluorescent tubes deliver a critical dose in minutes, and a filter that leaks as little as 0.1 % in the short-wave visible is enough for direct sunlight to expose a film in under an hour.

What the process sees afterwards is not “a bit of haze”. Partial exposure changes dissolution behaviour: critical dimension drift, rounded or T-topped profiles, scum in areas that should have cleared, and develop times that no longer reproduce. Because the effect accumulates across a shift, it usually surfaces as a batch problem blamed on the coater, the developer or the resist batch.

What wavelength does a yellow room actually need?

No ISO, EN or SEMI clause fixes a number. The threshold is where the specific resist stops absorbing, and it moves with chemistry, film thickness and the dose budget. What can be stated is the range and the direction of the trade-off.

Actinic band and practical cut-off by resist type
Resist / processActinic bandPractical cut-offSource that meets it
Broadband DNQ-novolak, g/h/i-line, 1–3 µm film320–450 nm, with residual sensitivity into blue and weak greenBelow 500–520 nmNarrow-band amber 570–590 nm LED, or a fluorescent tube behind film rated below 520 nm
i-line only, 365 nm, thin film365 nm, plus a measurable blue response in many productsBelow 500 nmAmber LED with a measured spectrum; the risk here is 365–436 nm leakage, not UV-C
Thick-film PCB / MEMS resist, 20–60 µmBroadband to about 450 nm, with a larger dose budgetBelow 520 nm, sometimes lowerAmber or red; verify the actual 400–500 nm transmission of the delivered luminaire
Chemically amplified DUV resist, 248 / 193 nmVery high sensitivity; blue and even green can matter at high doseBelow 560 nm is common for handling areasRed-orange source, or filtered amber; do not hold unexposed film in a standard yellow room
Image-reversal and cross-linking negative resistBlue-sensitive; cross-linking is cumulative and irreversibleBelow 520–560 nmNarrow-band amber or red, with spectral data rather than a colour description

Note: These cut-offs are resist- and process-specific engineering values, not clauses of a lighting standard. Measure the installed luminaire between 350 and 550 nm, and set the limit from the resist supplier’s sensitivity curve plus the maximum time a coated wafer spends under ambient light.

The second half of the answer is the source spectrum, not its colour. Amber light at 590 nm is safe because there is nothing below 500 nm left to leak; the room looks yellow because the eye has been given one narrow band.

Does an amber LED, a filtered tube or a coated lamp block more?

Four constructions are sold as yellow-room lighting, and only two of them are spectrally trustworthy by construction rather than by appearance.

Yellow-room light sources compared by spectrum
ConstructionSpectrumLeakage riskWhat to check
True amber LED (AlInGaP, about 590 nm)Narrow band, no blue pumpLow, if the die bin and spectrum are verifiedInstant start, dimmable, long life; ask for the 350–550 nm scan rather than the colour temperature
Phosphor-converted “yellow” LED (blue chip plus YAG phosphor)Strong 450–460 nm pump peak under a broad yellow humpHigh — the pump sits inside the resist’s absorption bandCommon in low-cost lamps; a yellow diffuser hides the blue peak completely
Fluorescent tube with external yellow film or sleeveFiltered mercury lines; residual 405 and 436 nm depends on the filmMedium — film ages, cracks and heat-cyclesCheap retrofit; replace the film on a schedule and re-verify transmission
Coated or dipped “gold” lampCoating absorbs short-wave and degrades with heatMedium to high over lifeVerify transmission at end of life, not on day one

Note: “Yellow” is a colour description, not a spectral specification. The only acceptable evidence is a transmission or irradiance scan covering 350–550 nm, taken from the delivered product at the delivered mounting height.

Does mounting height change the risk?

Yes, more than most specifications assume. Irradiance from a linear tube falls roughly as 1/h close to the source and as 1/h² further away. Put a number on it: treating a 120 cm tube as a cosine-weighted line source, the modelled irradiance at 53 cm comes out at roughly 25–30 times the value at 180 cm, and a hand-held lamp can be brought closer than 53 cm. A portable inspection lamp held half a metre above a coated wafer is therefore not a small addition to the room; it can be the dominant dose of the shift.

The design consequence is simple: keep resist handling at bench height with general lighting above 2 m, avoid unfiltered task lamps inside the yellow area, and check the lights inside fume hoods and pass boxes, which sit closest to the work. A filtered LED at short distance behaves very differently from an unfiltered one at the same distance, and neither behaves like the ceiling luminaire.

What illuminance does a yellow room need?

The lighting design is still a task design. BS EN 12464-1 puts general production and laboratory work at about 500 lx maintained, and detailed inspection at 750 lx or more; GB 50073-2013 §9.2.3 recommends 200–500 lx for production rooms without daylight windows, and §9.2.4 requires general lighting uniformity of at least 0.7. Two caveats apply in an amber room: colour rendering is poor by construction, because blue and red are missing, so resist-to-substrate contrast is weaker and the illuminance that feels adequate in white light is often not adequate here; and none of these figures says anything about actinic dose.

Where an operator has to judge colour — resist thickness interference colour, for instance — that judgement belongs at a filtered white inspection station outside the yellow area, or in a measurement, not in a brighter yellow luminaire.

Is a yellow room still a cleanroom?

Usually yes: most yellow areas are ISO 5–8, and the wavelength changes nothing about particles. The luminaire still needs a sealed room face, a rear rating where it opens into a ceiling void, glass rather than a yellowing plastic diffuser, and a gasket that survives the cleaning agents. A fixture that meets the spectrum and fails the seal has solved the wrong problem; the selection logic is the same as for any classified room, and only the photometric layer changes.

How do you verify a yellow room that is already running?

  • —Scan the spectrum, not the colour: 350–550 nm transmission or irradiance from the installed fixture, at the height it is actually mounted.
  • —Run the resist test: coat a wafer, soft-bake it, mask half of it, leave it at a working position for one shift, then develop. A difference between the halves is a failing room.
  • —Use the blue-object check as a fast indicator: if blue objects look blue rather than grey or black, the filter is passing short-wave light.
  • —Age the filter deliberately. Film and sleeves crack and bleach; put replacement on a calendar rather than on a defect report.
  • —Audit the lights nobody specified: exit signs, emergency luminaires, indicator LEDs, screen backlights, microscope lamps, door vision panels, skylights and windows.
  • —Check portable and task lamps inside the area, and the lighting inside hoods and pass boxes, which works closest to the resist.
  • —Confirm that replacement lamps are the same spectral product, not “a yellow tube from the store”.
  • —Record the measurement with date and lamp serial so drift is visible at the next audit.

Three misjudgements worth correcting early

  • —“Yellow light is UV-free.” Amber light is UV-free only if the source is narrow-band amber. A phosphor-converted yellow LED contains a blue pump at 450 nm, inside the resist’s absorption band, and the diffuser makes it look correct.
  • —“500 nm is the rule.” 500 nm is common practice; resist suppliers often ask for absorption below about 520 nm, and DUV handling areas may need more. Take the number from the resist data sheet and the dose budget.
  • —“If it looks dim, it is safe.” Safety is spectral, not visual. A room can be uncomfortably dim and still expose a film; a bright room with a clean cut-off can be safe. Measure the band, not the brightness.
Selection Q&A

Questions engineers ask

What wavelength is used in a yellow light room?

Amber sources around 570–590 nm are the common choice, with the filter specified to remove everything below roughly 500–520 nm. The cut-off is set by the resist’s absorption — DNQ-novolak resists absorb from about 320 to 450 nm — not by a lighting standard.

Is 500 nm or 520 nm correct?

Both appear in practice. 500 nm is the common shorthand for g/h/i-line resists; resist handling guidance often recommends filters with strong absorption below about 520 nm because sensitivity extends into blue and weakly into green. Use the higher figure where film thickness or the dose budget is tight.

Can I use yellow LED tubes in an existing cleanroom fixture?

Only if the product is narrow-band amber with a measured spectrum and the fixture keeps its seal, rear rating and cleanable face. A phosphor-converted “yellow” lamp with a 450 nm pump can be worse than the fluorescent tube it replaces.

Does a yellow room need a different illuminance?

The task still sets it: about 500 lx for general work and 750 lx or more for detailed inspection, with uniformity of 0.7 or better in Chinese practice. Amber light renders colour poorly, so contrast-dependent work often needs more than the white-light figure, and lux never indicates UV safety.

How do I test whether my yellow room is safe?

Expose a masked, resist-coated wafer at a working position for a full shift and develop it; a difference between the masked and unmasked halves means the room is leaking. A spectral scan of the installed luminaire between 350 and 550 nm is the faster route when the equipment is available.