The Invisible Killer: How Semiconductor Fabs Fight AMC (Airborne Molecular Contamination)

1. The Invisible Killer

Magnify a single human hair by 1,000,000×, and you see a pillar about 0.1 cm across. Magnify the hardest-to-capture particle for a HEPA filter by the same factor, and you see a ball about 3 cm across. Magnify an "AMC gaseous molecule" by the same factor, and you see a tiny grain of sand about 1 mm across — small enough that HEPA filters cannot catch it at all.

This is one of the most frustrating problems in semiconductor, display, and OLED manufacturing: there is something in the air you cannot see, smell, or easily detect, yet it can ruin an entire wafer.

Its name: AMC (Airborne Molecular Contamination).

AMC molecules are 0.3–1.5 nm — roughly 100 to 1,000 times smaller than HEPA's most-penetrating size (MPPS, 0.1–0.3 μm). Even the world's most expensive ULPA filter cannot block them.

2. What Exactly Is AMC?

AMC is simply "chemical compounds floating in the air as individual molecules." Not dust, not bacteria — molecules of ammonia, hydrogen sulfide, hydrochloric acid, NMP, organic solvents, dopant gases. Names you heard in high-school chemistry. Silent enemies in a fab.

The semiconductor industry classifies AMC into four groups (SEMI F21):

Each class attacks the process differently:

• ▸Acids (MA) corrode metal lines, tarnish copper wiring, cause shorts or opens.
• ▸Bases (MB) — the most famous damage is "T-top": the top layer of DUV photoresist degrades, producing T-shaped deformation in the patterned structure.
• ▸Condensables (MC) — high-boiling-point molecules condense on wafer surfaces or optics, creating haze.
• ▸Dopants (MD) — trace amounts can shift the semiconductor's electrical characteristics and drift device parameters.

Grades follow "MX-N" notation: X = class, N = concentration in ppt (parts per trillion). "MA-10" means "acids at ≤10 ppt."

What does 1 ppt mean? Dissolve 1 gram of salt in 100,000 tons of water, mix, take 1 mL — that's roughly 1 ppt.

3. What Is the Actual Cost?

In sub-14 nm processes, a batch of wafers can cost tens of millions of NTD. Typical AMC disasters include:

1. 1T-top on photoresist — a small ammonia spike (maybe someone used a cleaner containing ammonia in the next building) ruins an entire DUV exposure run.
2. 2Wafer hazing — organic outgassing inside a FOUP during transfer makes the wafer useless by the next station.
3. 3Copper tarnishing — trace SO₂ on copper lines changes resistance, causing leakage.
4. 4Dopant drift — faint traces of B₂H₆ or AsH₃ altering transistor characteristics.

These losses are rarely caught immediately. Often they surface only during final test ("Why did yield drop 5%?") — and the root cause is invisible, possibly not even on the production line but in the return-air duct.

4. Where Does AMC Come From?

The real difficulty: AMC has multiple, distributed sources.

1. 1Outdoor air — nearby petrochemical plants, electroplating shops, farms, or traffic bring HCl, SO₂, NOx, NH₃ into the make-up air system.
2. 2Process leakage / exhaust — your own tools generate NMP, IPA, PGME; local exhaust helps but cannot be perfect.
3. 3Material outgassing — the most insidious source. FOUP plastics, wall paint, floor adhesive, even operator gloves slowly release trace chemicals.

AMC control must therefore address BOTH make-up air and recirculation:

• ▸Make-up air intake: large chemical filters
• ▸Inside the cleanroom: chemical filter layer above FFU
• ▸Process tools: mini-environment or FOUP-level filters
• ▸Critical areas: point-of-use gas purification

5. How Do You Even Measure It?

AMC is measured in ppt. Your home CO detector runs at ppm — a million times coarser.

Two categories of measurement:

High-precision, long sampling (ppt-grade):

• ▸Adsorption tube + ATD-GC/MS
• ▸Impinger + ion chromatography
• ▸Measures down to ppt but each sample takes 1–3 days — not suitable for real-time alarming

Real-time (<3 minutes):

• ▸Ion Mobility Spectrometry (IMS): NH₃, amines, 0.1 ppb limit
• ▸UV Fluorescence (API): SO₂, H₂S, 0.4 ppb
• ▸Chemiluminescence (CL): NO, NO₂, NH₃, 1 ppb
• ▸Photo-Ionization Detector (PID): total VOCs, 20 ppb

Advanced fabs run both: real-time for alarming, GC/MS for forensic root-cause analysis.

6. How a Chemical Filter "Captures" Molecules

This is the technical core. Chemical filters use three distinct mechanisms:

Physisorption is the baseline: pitted activated carbon traps molecules via van der Waals forces. Binding energy is low (<10 kcal/mol), good news is it handles most VOCs, bad news is it's reversible — molecules can desorb at higher temperatures.

Chemisorption is more aggressive: the molecule reacts with the media to form a new, stable compound, permanently locked in. Binding energy 10–100 kcal/mol. Achieved by "impregnating" activated carbon with reactive chemicals:

• ▸H₃PO₄-impregnated carbon captures NH₃ (forms ammonium phosphate)
• ▸KOH or K₂CO₃-impregnated carbon captures H₂S, SO₂ (forms potassium sulfide/sulfate)
• ▸KMnO₄-impregnated carbon oxidizes reducing gases

Ion exchange uses resin-based media where surface ions swap with AMC ions — highly effective for trace polar molecules.

A real-world chemical filter combines multiple mechanisms: pre-filter for particles, middle layer for VOCs (physisorption), inner layer for acids/bases (chemisorption).

7. What Affects Chemical Filter Performance

Installing a chemical filter is not a set-and-forget proposition. Four variables dramatically affect efficiency:

1. Humidity — the largest and most counter-intuitive factor:

ITRI test data: under 80 ppm toluene challenge, 75% RH capacity is 114% higher than 35% RH. Why? Water film helps non-polar VOCs condense in carbon pores.

But for polar gases like NH₃ or SO₂, high humidity hydrolyzes the impregnated carbon and reduces efficiency. The "dry or humid" question depends entirely on what you're trying to block.

2. Temperature — higher temperature weakens physisorption (easier desorption) but accelerates chemisorption reaction rates.

3. Flow velocity — faster air = shorter contact time = lower efficiency. Typical face velocity 0.3–2.5 m/s.

4. Concentration — high concentration exhausts the filter quickly. Low concentration reduces capture probability per encounter.

This is why chemical filter selection cannot be done from a catalog alone — it must be tailored to the facility's real temperature, humidity, target species, and concentration profile.

8. Taiwan's Chemical Filter Testing Capability

A chemical filter is only useful if it can be verified. One of Taiwan's most important infrastructure investments over the past decade has been building local chemical-filter testing capability:

1. 1Internationally-compliant test platforms — JIS B9901, JIS B8330, ISO/TS-11155-2, NT VVS 109; filter sizes 592–610 mm, depth 30–300 mm; T 10–35°C, RH 30–95%, face velocity 0.3–2.5 m/s; challenge gases NH₃, H₂S, SO₂, DMS, toluene, IPA (10 ppb – 10 ppm).
2. 2Alignment with international certification — ASHRAE 145.1 (media), ASHRAE 145.2 (assembly), ISO 10121 — so Taiwan-made filters can be internationally certified.
3. 3FOUP outgassing methodology — measurement protocols for trace AMC inside wafer carriers, supporting material selection and cleaning process optimization.

9. Conclusion: AMC Control Is Never a Single-Point Fix

Why doesn't HEPA catch AMC? Because they live in different worlds. HEPA targets particles (solids, droplets). AMC is molecular (gaseous). Two entirely different battles.

A comprehensive AMC control strategy requires:

1. 1Source reduction — low-outgassing materials, plastics, gloves
2. 2Make-up air purification — large chemical filters at intake
3. 3Recirculation purification — filter layer above FFU
4. 4Local purification — mini-environments, FOUP filters, point-of-use gas
5. 5Real-time monitoring — IMS, UV fluorescence, PID
6. 6Periodic forensics — ATD-GC/MS for root-cause investigation
7. 7Scheduled replacement — chemical filters have a service life

Baisheng Tech's role: as the Taiwan distributor of NIPPON MUKI chemical filters and a long-term partner of ITRI, we provide end-to-end selection, testing, and deployment support for AMC control.

Related Standards

• ▸SEMI F21 — AMC classification
• ▸ASHRAE 145.1 / 145.2 — chemical filter efficiency test methods
• ▸ISO 10121 — gas-phase pollutant removal test standard
• ▸JIS B9901 / B8330 — gas-adsorption filter test methods