- For the same cartridge, "99.9%" versus "99.99%" can hide a 100× difference — what matters is the test method and the size of the challenge particle
- General-purpose liquids use β-ratio (ISO 16889 multipass), pharma uses ASTM F838 bacterial challenge, and semiconductors use single-pass gold-nanoparticle challenge — three frameworks that don't translate to each other
- β1000 = 99.9%, β5000 = 99.98%, β20000 = 99.995% — the numbers grow but the efficiency gain is tiny
- This article uses comparison tables and real bubble-point data so you can see through datasheet marketing
- Why does "filtration efficiency" vary so widely? The same cartridge can be marketed as 99.9% or 99.99%
- Nominal vs absolute: where the two numbers part ways
- β-ratio and LRV: the industry "report card"
- From general (µm) to nanoscale test methods
- Pharma vs semiconductor: two worlds with different requirements
- Integrity testing (Bubble Point / Diffusion Flow) at a glance
- Common pitfalls / marketing tricks
- Frequently asked questions
- References
Why does "filtration efficiency" vary so widely? The same cartridge can be marketed as 99.9% or 99.99%
The most misleading thing on a cartridge datasheet is that the number "99.9%" by itself means nothing. It only becomes comparable when paired with three preconditions: challenge particle size, test method, and single-pass vs recirculating. Change any one of these variables and the same cartridge can swing from 99% to 99.999% — three orders of magnitude apart.
An extreme example: for a 5 µm nominal nonwoven bag filter, the supplier can choose to "test it with 50 µm particles" and casually claim "99.99% filtration efficiency." But challenge it with actual 5 µm glass microspheres and you might only see 60% retention. The supplier isn't necessarily lying — it's that "test conditions" are not mandatorily disclosed.
What this article aims to do is to break down the vague concept of "filtration efficiency" into five verifiable axes: rating type (nominal / absolute), efficiency metric (β / LRV / %), challenge particle (glass bead / bacteria / gold nanoparticle), test workflow (single-pass / multipass), and industry validation (ISO 16889 / ASTM F838 / proprietary). Once you've read it, you'll know exactly what questions to ask the supplier next time you pick up a datasheet.
Nominal vs absolute: where the two numbers part ways
The oldest classification of cartridge ratings is nominal versus absolute. Both translate to "micron rating" colloquially, but their substantive meaning is very different.
Nominal rating
A nominal rating is defined as the "capability of retaining 50% of particles at that size". So a 5 µm nominal cartridge means: when 5 µm particles are introduced, roughly half are retained and half pass through. It sounds bad, but this is the industry standard.
Nominal ratings have no unified test method — Vendor A may use glass beads, Vendor B uses AC test dust, Vendor C may simply estimate from in-house process data. Two cartridges both labelled "5 µm nominal" can therefore differ by more than 3× in actual performance. Common in PP meltblown, string-wound, and nonwoven depth-type cartridges.
Absolute rating
An absolute rating is much stricter, defined as "the smallest pore size at which ≥ 99.9% of particles at that size are retained". So a 0.22 µm absolute membrane retains at least 99.9% of 0.22 µm particles. This corresponds to β ≥ 1000.
An absolute rating is meaningful only when paired with a standard test method — commonly ISO 16889 multipass (hydraulic oil), ASTM F838 bacterial challenge (pharma), or proprietary latex bead / gold sol challenge (semiconductor).
Nominal vs absolute side-by-side
| Item | Nominal | Absolute |
|---|---|---|
| Typical retention efficiency | 50% (no enforced standard) | ≥ 99.9% |
| Equivalent β-ratio | β ≈ 2 | β ≥ 1000 |
| Equivalent LRV | 0.3 | ≥ 3 |
| Common cartridge types | PP meltblown, string-wound, bag filter | Membrane filter, pleated PES/PTFE |
| Test method | Inconsistent across vendors | ISO 16889 / ASTM F838 / standard beads |
| Unit price | Low | Medium to high |
| Suitable for | Pre-filtration, downstream protection | Final sterile filtration, nanoscale retention |
| Datasheet marketing risk | Very high | Low |
β-ratio and LRV: the industry "report card"
Nominal vs absolute is just a coarse split. The metrics that allow real precision when comparing cartridge efficiency are β-ratio and LRV (log reduction value). These two are the industry's report card — what looks like "99.9% filtration" can be parsed in much greater detail.
What β-ratio is
The β-ratio definition is straightforward: upstream particle count ÷ downstream particle count, measured at a specified particle size. For example, β10(c) = 1000 means that for every 1000 particles of 10 µm upstream, only 1 remains downstream. Converted to efficiency: (β−1)/β × 100% = 99.9%.
The "(c)" is not decoration — it's the ISO 16889:2008 marker indicating ISO 11171 NIST-traceable standard test dust. Old-style β values without (c) (the ISO 4572-era β10 = 1000) do not equal current β10(c) = 1000 — the new standard is stricter.
β-ratio vs efficiency
| β value | Efficiency | Downstream fraction | Equivalent LRV | Application tier |
|---|---|---|---|---|
| β = 2 | 50% | 1/2 | 0.3 | Nominal-grade baseline |
| β = 10 | 90% | 1/10 | 1 | Entry-level industrial pre-filtration |
| β = 75 | 98.7% | 1/75 | 1.9 | Hydraulic-system minimum |
| β = 100 | 99% | 1/100 | 2 | General industrial |
| β = 200 | 99.5% | 1/200 | 2.3 | Precision hydraulics |
| β = 1000 | 99.9% | 1/1000 | 3 | Absolute-grade threshold |
| β = 5000 | 99.98% | 1/5000 | 3.7 | High-cleanliness processes |
| β = 20000 | 99.995% | 1/20000 | 4.3 | Nanoscale |
LRV: the metric pharma prefers
The pharmaceutical and biotech world tends to use LRV (Log Reduction Value) instead of β:
LRV = log₁₀(upstream / downstream)
So LRV = 3 means 99.9%, LRV = 6 means 99.9999%, LRV = 9 means 99.9999999%. Each additional log adds another zero of cleanliness downstream.
The minimum regulatory bar for pharmaceutical sterile filtration is LRV ≥ 7 — out of one billion bacteria upstream, no more than 100 may pass. A 0.22 µm sterilizing-grade cartridge subjected to ASTM F838 (10⁷ B. diminuta/cm²) must show no growth (0 CFU) downstream, corresponding to LRV ≥ 7.
From general (µm) to nanoscale test methods
Different particle sizes require different challenges, because no single "standard particle" covers the size range from 25 µm down to 1 nm. Practically, four tiers exist:
General (µm tier): hydraulic / food / industrial water
Hydraulic oil and industrial water primarily use ISO 16889:2022 multipass. ISO Medium Test Dust is continuously injected into a recirculating loop while a laser particle counter monitors upstream and downstream size distributions, and the β7(c)–β25(c) retention efficiencies are calculated at each size. This is the industry's most credible method because it simulates real service conditions (accumulating challenge over time) rather than a static, one-shot test.
Food and industrial water more often use single-pass latex bead challenge: a single particle size (5, 10, or 20 µm) of latex standard beads is challenged once, and β is calculated. Simple and inexpensive — but blind to long-term performance decay.
Sterile grade (µm tier): pharma
The pharma industry recognizes only one method — ASTM F838 — using Brevundimonas diminuta (ATCC 19146), the smallest bacterium accepted by industry. Cells measure 0.3–0.4 × 0.6–1.0 µm, capable of passing through slightly enlarged 0.22 µm pores.
The challenge protocol: 10⁷ B. diminuta cells per cm² of effective filter area are injected, downstream effluent is collected on a 0.45 µm membrane and cultured, and the result must be 0 CFU to pass. LRV ≥ 7 is the minimum requirement. The FDA writes this method into the very definition of "sterilizing-grade filter cartridge."
Nanoscale (nm tier): semiconductor
Below 50 nm, B. diminuta is too large, so the challenge must shift to gold nanoparticles (gold sol) or fluorescent nanobeads. Entegris published a method in 2017 using 3 nm, 5 nm, and 10 nm UPE membranes with gold-nanoparticle challenge to measure true sub-10 nm efficiency.
Filtration of EUV chemically amplified resists (CAR) operates at this tier. Pall, Entegris, and Cobetter all offer proprietary 1 nm / 3 nm UPE point-of-use filters, with challenge particles ranging from gold nanoparticles to 1.8 nm polystyrene latex. Note that there is no widely published unified standard at this tier akin to ASTM F838 — every supplier uses a slightly different method.
Four test methods at a glance
| Test method | Challenge agent | Applicable size | Industry | Standard |
|---|---|---|---|---|
| Multipass | ISO MTD test dust | 4–25 µm | Hydraulic oil | ISO 16889:2022 |
| Single-pass latex challenge | PSL latex standard beads | 0.5–20 µm | Food / beverage / chemical | Vendor methods |
| Bacterial challenge | B. diminuta 10⁷/cm² | 0.2–0.45 µm | Pharmaceutical | ASTM F838-20 |
| Virus challenge | PP7 phage / Φ-X174 | 20–60 nm | Biotech / plasma | ASTM E2720 |
| Gold nanoparticle challenge | Gold sol 3–30 nm | 3–30 nm | Semiconductor UPE | Vendor methods / SEMI |
Pharma vs semiconductor: two worlds with different requirements
For two cartridges both labelled 0.22 µm, pharma and semiconductor do not actually want the same cartridge. The literal phrase "filtration efficiency" is identical, but the scoring rubrics behind it are completely different.
Integrity testing (Bubble Point / Diffusion Flow) at a glance
Filtration efficiency is a pre-shipment lab metric. Integrity testing is field validation — confirming the cartridge has no holes, pinholes, or assembly defects before and after use.
Bubble Point
Principle: when a fully wetted membrane's pores are sealed by surface tension, pressure is gradually applied until gas finally pushes the liquid out of the largest pore — that pressure is the bubble point. The smaller the pore, the higher the BP.
| Pore size | Membrane | Wetting medium | Typical Bubble Point |
|---|---|---|---|
| 0.45 µm | Hydrophilic PES | Water | 2.0–2.8 bar (29–41 psi) |
| 0.22 µm | Hydrophilic PES | Water | 3.4–4.4 bar (50–64 psi) |
| 0.22 µm | Hydrophilic PVDF | Water | ~3.5 bar |
| 0.22 µm | Hydrophilic PTFE | Water | 4.4 bar (64 psi) |
| 0.22 µm | Hydrophobic PTFE | 60% IPA / water | 0.7–0.85 bar |
| 0.1 µm | Hydrophilic PES | Water | 5.5–6.5 bar |
| 0.05 µm | UPE | IPA | 1.5–2.0 bar (special low-BP design) |
Diffusion Flow
BP is "destructive" in the sense that you have to reach it to know it. Diffusion Flow is non-destructive: at 75–80% of BP, you measure the diffusion-driven gas flow through the wetted membrane. If diffusion flow exceeds the supplier spec, a pinhole or assembly defect is present, but the membrane has not yet been pushed past failure.
BP is not equivalent to filtration efficiency
Many junior engineers mistake BP for an efficiency indicator — "higher BP = better cartridge." This is wrong. BP only reflects the largest pore, while efficiency depends on pore-size distribution, total area, and depth structure. A cartridge that passes BP still requires F838 / β-ratio data from the supplier — BP only confirms it isn't broken before shipment.
Common pitfalls / marketing tricks
Frequently asked questions
If β1000 = 99.9%, is β5000 = 99.98% really "much stronger"?
Numerically it looks 5× stronger, but efficiency only rises by 0.08%. For ordinary industrial fluids that's irrelevant. But for EUV photoresist and injectables, where a single downstream contaminant is catastrophic, that 0.08% is the line between yield and scrap. That's why high-end semi and biotech plants pay more for β5000-class absolute-grade cartridges.
Is absolute 0.22 µm guaranteed to be sterilizing?
Not necessarily. "0.22 µm absolute" is just a physical-pore-size descriptor, while "sterilizing-grade" is the regulatory certification of ASTM F838 LRV ≥ 7. A 0.22 µm absolute membrane sold without F838 validation may not be used for pharmaceutical sterile filtration. When buying, verify the supplier provides a bacterial-challenge report — pore size alone is not enough.
Why doesn't the semiconductor industry use ASTM F838?
Because B. diminuta is 0.3–0.4 µm, far too large for sub-50 nm UPE membranes — challenge it and 100% gets retained, returning infinite LRV with no resolving power. Semiconductors must instead use gold nanoparticles (3–30 nm) or polystyrene latex (5–20 nm) as challenge to reveal differences between membranes.
Which is more accurate, diffusion flow or bubble point?
It depends on membrane area. Small membranes (< 0.5 m²) favour BP — pinholes shift BP visibly. Large membranes (> 1 m²) use diffusion flow — small pinholes barely change BP on large membranes but clearly raise diffusion flow. Both FDA and EU GMP accept either method.
If a datasheet doesn't state the test method, can I trust the numbers?
No. Ask three questions: (1) Which standard? ASTM F838 / ISO 16889 / proprietary? (2) What is the challenge agent? Standard test dust / B. diminuta / latex bead? (3) Single pass or multipass?. If the supplier won't answer any of these, the credibility of the number drops to zero automatically.
Why does the same cartridge show LRV = 6 but β = 1000?
Because they reflect different challenge agents. LRV = 6 likely uses B. diminuta (0.3–0.4 µm) for a 10⁶× reduction; β1000 likely uses 0.5 µm latex bead in a single-pass test. The same cartridge naturally has different retention efficiencies for different particles. When reading datasheets, always pair the metric with its challenge agent.
Is there really a difference between UPE 1 nm and 3 nm ratings?
Yes. Live data from Entegris and Pall on EUV CAR photoresist filtration show that 1 nm UPE delivers defect density on average 10–20% lower than 3 nm UPE after EUV exposure. The yield impact at 3 nm nodes is significant, but cost is 1.5–2× higher. Whether to upgrade depends on the fab's yield economics.
Should nominal-grade cartridges be retired entirely?
No. Nominal cartridges remain very important as pre-filtration to protect expensive downstream absolute-grade membranes — a 5 µm PP meltblown can stop 80–90% of large particles, extending the life of a downstream 0.22 µm absolute membrane by 5–10×. From a total-cost-of-ownership standpoint, a nominal + absolute "two-stage" design is often cheaper than a single absolute-only stage.
References
- ISO 16889:2022 — Hydraulic fluid power Filters Multi-pass method (β-ratio standard)
- ASTM F838-20 — Bacterial Retention of Membrane Filters (pharmaceutical sterile-filtration standard)
- Lösungsfabrik — Bacterial Retention Test according to ASTM F838-20 (B. diminuta details)
- Pall — Understanding Liquid Particle Removal Performance (β-ratio and nominal / absolute comparison)
- Entegris — Sub-10 nm Filter Retention Measurement Methodology
- Entegris — UPE Membrane Technology for POU EUV CAR Photoresist Filtration
- SPIE — Updated UPE Filter Design for POU EUV CAR Photoresist Filtration
- PMC — Defect Detection Sensitivity of Bubble-Point-Type Tests for Sterilizing-Grade Filters
- Critical Process — Integrity Testing of Hydrophilic Membrane Filters: Diffusive Flow
- Machinery Lubrication — Understanding Filter Beta Ratios
- ScienceDirect — Gold Nanoparticle Mixture Retention Test for UF Membrane Functional Pore Sizes