แผ่นกรองเคมีแลกเปลี่ยนไอออนกำจัด AMC ที่เป็นกรด

Acid targets move up while every non-target drops to the floor — by design, not by defect. Activated carbon pores take in everything, so capacity is shared with organics and nitrogen oxides; on ion exchange resin only the anionic functional groups react, leaving the whole capacity for acids. The trade-off is that ammonia, chlorine, ozone and organophosphate dopants are not addressed at all, so sites that must control those still need carbon alongside.

Efficiency falls only 8.5 percentage points over five years, and the reason is what the medium refuses to take in. Over the same period an activated carbon bed has already given most of its capacity to organics and nitrogen oxides, leaving little for acids. In the field this translates to hydrogen fluoride held below a third of the cleanroom target, with boron from the BF₃ by-product at half the target value.

On the carbon filter 87% of what was captured has nothing to do with acids — organics and nitrogen oxides fill the pores first, leaving just 26 g of capacity actually spent on acids. Ion exchange resin has no adsorption pores; only the anionic functional groups react, so total load is a tenth as large and three quarters of it is acid. That is where the service-life gap comes from at identical size and airflow: not from capturing more, but from not spending capacity on things it was never meant to capture.

Spikes are the real test for a chemical filter — a process leak, a neighbouring stack, a seasonal shift in outdoor air can raise concentration by an order of magnitude within hours. Activated carbon loses 18.8 percentage points because stronger acids displace what the pores already held, so efficiency bleeds from both ends. Ion exchange binds chemically and has no displacement mechanism, losing only 13.5 points — its post-spike 66.3% is almost level with where the carbon filter stood before the spike, at 70.2%.

Activated carbon draws the process solvent PGMEA into its pores, where prolonged heat and humidity hydrolyse it, splitting off acetic acid that is then released downstream — the filter turns from a barrier into a source, measured at a net emission of 4.14 ppbv that attacks the chiller heat exchanger directly. Five hours of clean-air purging only brings the release down to 0.51 ppbv, which means the acetic acid is not surface residue but is still being generated. Ion exchange resin has no adsorption pores and does not take up esters, so there is nothing stored to cleave; after five years it is still removing.

A bare reticle has no pod to protect it, so any ammonium sulphate or acetate crystal forming on the surface affects imaging directly — the control limit sits at the 200 pptv level. Downstream of the filter the reading falls below the limit of quantification, meaning the headroom is not "just inside spec" but unmeasurable. A second stocker filter at six years still delivers 88.1% on sulphur dioxide and 78.2% on acetic acid: weak acids decaying faster than strong ones is expected, because their bond to the functional group is weaker and later-arriving strong acids displace them. Change-out intervals should therefore be set by the weak-acid figure.

The ion exchange filter is at 49.4% after eight years, which looks worse than 90.2% for carbon at six months — except that is sixteen times the duration. The top row explains why: 43–46% of the carbon load is organics, so capacity was consumed by construction-phase diesel fumes and ambient volatile compounds long before winter sulphur dioxide arrived. Ion exchange takes up no organics at all; almost the entire load is acid salts, so capacity went into acids from day one through year eight. The selection question is whether efficiency holds across your change-out interval, not who is higher when new. Conversely, a site that must also control organics and nitrogen oxides cannot rely on ion exchange alone.