99.95 % and 99.995 % sound like they differ by only 0.045 %. But — 10× more particles slip through.
"One More 9" Costs an Order of Magnitude
Per EN 1822 / ISO 29463, high-efficiency filters come in five grades:
Biểu đồ 1: Thang hiệu suất HEPA / ULPA (EN 1822 / ISO 29463)
Mỗi cấp tăng làm giảm số hạt lọt qua 10 lần — nghiêm ngặt hơn 10 lần cấp trước
| Cấp | Hiệu suất (@MPPS) | Số hạt lọt / triệu |
|---|---|---|
| H13 | 99.95 % | ≤ 500 hạt / triệu |
| H14 | 99.995 % | ≤ 50 hạt / triệu |
| U15 | 99.9995 % | ≤ 5 hạt / triệu |
| U16 | 99.99995 % | ≤ 0.5 hạt / triệu |
| U17 | 99.999995 % | ≤ 0.05 hạt / triệu |
Hiệu suất đo tại MPPS (kích thước hạt xuyên thấu nhất, ~0.1–0.3 μm) theo EN 1822 / ISO 29463. "Lọt qua" = số hạt xuyên qua lưới trên mỗi 1.000.000 hạt thử.
Every rung drops the penetration count by one decade:
- ▸H13 (≥99.95 %) — up to 500 particles slip through per million challenge particles
- ▸H14 (≥99.995 %) — up to 50
- ▸U15 (≥99.9995 %) — up to 5
- ▸U16 (≥99.99995 %) — up to 0.5
- ▸U17 (≥99.999995 %) — up to 0.05
From H13 to U17 the efficiency gap is 10,000×. That's what "one more 9" actually buys.
MPPS: The Filter's Weakest Point
Intuition says a filter is worse at catching smaller particles. Wrong.
Filters capture via three overlapping mechanisms: inertial impaction (large particles), interception (mid), and Brownian diffusion (small). In the 0.1–0.3 μm window none of the three dominates — that's the MPPS (Most Penetrating Particle Size).
EN 1822 / ISO 29463 measure every efficiency rating at the MPPS. Not at a convenient size where the filter looks good — at the single worst point of the efficiency curve.
"H14 ≥ 99.995 %" really means: even at the hardest-to-catch particle size, no more than 50 ppm slip through.
What Does 27 % More Pressure Drop Buy You?
Same size (Baisheng 610 × 610 × 292 mm), same airflow (1,000 CMH):
Biểu đồ 2: So sánh tổn thất áp suất ban đầu — H14 HEPA vs U15 ULPA
Lọc Baisheng 610 × 610 × 292 mm, đo tại lưu lượng định mức 1.000 CMH
Mỗi 60 Pa tổn thất áp tăng làm công suất quạt tăng ~10–15%, tiêu thụ thêm 200–400 kWh mỗi FFU mỗi năm. Nâng cấp ULPA quá mức làm tăng cả chi phí đầu tư lẫn chi phí vận hành.
U15 runs 60 Pa higher than H14 — ~27 % more pressure drop.
That 60 Pa is not free. It means:
- ▸Fans pull 10–15 % more power to push air through
- ▸Each FFU burns an extra 200–400 kWh per year
- ▸A fab with 1,000 FFUs adds hundreds of thousands to its annual power bill
Every extra 9 costs pressure drop and energy. Pay where it matters, skip where it doesn't.
HEPA or ULPA? Start From Your ISO Class
Filter selection isn't "higher is better" — it's driven by the ISO 14644-1 cleanliness requirement:
Biểu đồ 3: Ánh xạ ISO Class ↔ cấp lọc
Cấp lọc và kiểu dòng khí đề xuất cho mỗi cấp sạch ISO 14644-1
| ISO Class | Lọc đề xuất | Ứng dụng điển hình | Dòng khí |
|---|---|---|---|
| Class 3 (↓) | U15 / U16 ULPA | Bán dẫn tiên tiến, quang học EUV | Lớp |
| Class 4–5 | H14 HEPA | Bán dẫn, OLED, dược vô trùng | Lớp |
| Class 6–7 | H13 HEPA | Lắp ráp điện tử, kiểm quang, dược C/D | Rối được |
| Class 8–9 | Trung / cao hiệu | Đóng gói thực phẩm, ngoại vi phòng mổ | Rối được |
Hướng dẫn kỹ thuật tổng quát. ISO Class 1–2 ngoài phạm vi bảng này.
Quick mapping:
- ▸Advanced semi lithography / EUV (ISO Class 3 and below) — must use U15 or U16 ULPA
- ▸General semi, OLED, aseptic pharma fill (ISO Class 4–5) — H14 HEPA is enough
- ▸Electronic assembly, optics inspection, pharma grade C/D (ISO Class 6–7) — H13 HEPA suffices
- ▸Food packaging, OR perimeter (ISO Class 8–9) — medium or high-efficiency filters are fine
Forcing U16 into an ISO Class 5 aseptic fill line just burns cash for no engineering benefit. Conversely, saving money with H14 in an EUV optics bay can scrap a single million-dollar reticle.
Three Real-World Selection Traps
Trap 1: Blindly Chasing the Highest Grade
Over-spec drives up both upfront cost and long-term operating cost (higher ΔP → more electricity). Pin the ISO Class first, then pick the filter — not the other way round.
Trap 2: Looking Only at Average Efficiency, Not Per-Unit Scan Test
"This filter averages H14 efficiency" does not mean no local pinhole leaks. EN 1822-4 requires a per-unit scan test for H14 / U15 and above: a mobile probe scans the entire filter face, confirming no local point exceeds the penetration limit.
Always demand the per-unit Scan Test report on delivery. Without it, an average-efficiency certificate can't catch local leaks.
Trap 3: Forgetting to Set a Terminal ΔP = Replacement Trigger
Initial pressure drop is not forever. As dust accumulates, ΔP climbs. The engineering rule of thumb: set terminal ΔP at 2–2.5× initial ΔP and replace at that point.
Example for H14 HEPA: 220 Pa initial → terminal at 440–550 Pa. Hit that, swap it out. Running fans against perpetually high ΔP burns energy and shortens motor life.
Bottom Line: Higher Grade Isn't Better — The Right Grade Is
When picking HEPA / ULPA, the real questions are not "which is strongest" but:
- 1What ISO Class does my process need?
- 2Has MPPS efficiency been verified? (Scan test)
- 3Does the pressure/energy profile match my fan curve?
- 4When does terminal ΔP arrive? Budget for how many replacements?
The right grade matters more than the expensive grade.
