- NVR (Non-Volatile Residue) is measured by "weighing" — evaporate 100 mL of extract dry, and whatever remains on the dish is the result in mg/L
- TOC (Total Organic Carbon) is measured by "burning" — oxidize all organics into CO₂ and measure the carbon content; sensitivity reaches the ppb level
- The two methods catch different things: NVR catches non-volatile heavy components that TOC can't see; TOC catches small volatile molecules that NVR can't detect
- Semiconductor UPW spec: TOC < 1 ppb (target); pharmaceutical WFI spec: TOC < 500 ppb (USP <643>); advanced process filter cartridges: NVR < 1 mg/L
- This article delivers a complete test workflow, a comparison across four major materials from public data, JIUNYUAN lab measurements, and a debunking of common marketing claims
- Why does the semiconductor industry measure NVR and TOC every single day?
- NVR: weigh-based — "how much non-volatile residue is left"
- TOC: burn-based — measuring carbon content
- NVR vs TOC: they catch completely different things (with comparison table)
- Detailed test workflow
- NVR / TOC comparison across four major materials (public industry ranges)
- JIUNYUAN lab data: NVR and long-chain hydrocarbons from a 10 nm PTFE filter
- Spec quick-reference: semiconductor UPW, pharmaceutical WFI, semiconductor photoresist
- Common errors and marketing claims
- Frequently Asked Questions
- References
Why does the semiconductor industry measure NVR and TOC every single day?
Cleaning baths, photoresist coaters, and UPW systems in wafer fabs are wrapped in a "filtration maze" of cartridges in series. These cartridges are themselves polymers, and the question is: do they leach a tiny bit of themselves into the fluid during operation, contaminating downstream wafers? — That's exactly what "extractables" testing is meant to answer.
Extractables fall into two camps: organic leachables (polymer oligomers, plasticizers, residual monomers, cleaning agent residues) and inorganic leachables (metal ions, TDS). This article focuses on the organic side, where the industry's two standard yardsticks are NVR and TOC.
The difference between NVR and TOC is often blurred. But in practice, they are two yardsticks with completely different scales: NVR's smallest division is about 0.1 mg/L (100 ppb level), while TOC can measure down to 0.05 ppb — a difference of three orders of magnitude. But TOC misses the "uncombustible" ash, while NVR misses "volatile small molecules." One is heavy, the other is sensitive — each covers the other's blind spot.
NVR: weigh-based — "how much non-volatile residue is left"
NVR (Non-Volatile Residue) is the oldest and most brute-force method — take a fixed volume of extract, evaporate it dry, and weigh the residual solids.
Standard method (ASTM E1235)
- Sampling: typically 100 mL of extract (more sample = more sensitive, but takes longer)
- Container: platinum or stainless-steel evaporation dish, dried at 105 °C to constant weight beforehand and recorded as W₀
- Evaporation: completely evaporate the solvent in a 105 °C oven or vacuum rotary evaporator (~4–6 hours for aqueous samples)
- Constant weight: return to 105 °C oven for 30 min, cool in a desiccator, weigh as W₁
- Calculation: NVR (mg/L) = (W₁ − W₀) × 10³ ÷ V (sample volume in L)
What does NVR catch?
- Polymer oligomers (low-MW residual oligomers from PE, PP processes)
- Plasticizers and antioxidants (DOP, Irganox 1010, and other additives)
- Lubricants and mold release agents (silicone oils, PEG-based)
- Cartridge cleaning agent residues (surfactants, mid-boiling-point IPA impurities)
- Inorganic salts (also weighed in — a limitation of NVR)
What does NVR miss?
Anything with a boiling point below 105 °C — methanol, ethanol, acetone, IPA, dichloromethane, low-MW monomers — all evaporate during drying. So NVR cannot measure "volatile organic compounds (VOCs)". This is exactly where TOC steps in.
TOC: burn-based — measuring carbon content
TOC (Total Organic Carbon) takes the opposite approach — it doesn't matter whether you're volatile or non-volatile, molecule or ion: as long as you contain carbon, I'll burn you into CO₂, measure the CO₂ concentration with infrared (NDIR), and back-calculate the organic carbon content.
Three mainstream oxidation methods
Relationship between TC, IC, and TOC
TOC is not measured directly in one shot, but rather computed by subtraction: TOC = TC − IC:
- TC (Total Carbon): all forms of carbon, including organic + inorganic (carbonates, dissolved CO₂)
- IC (Inorganic Carbon): the CO₂ portion driven off after acidifying the sample
- TOC: what remains after subtracting IC — the organic carbon
What is USP <643>?
USP <643> is the legally required method for TOC in pharmaceutical Water for Injection (WFI) and Purified Water (PW). Key requirements:
- TOC limit: 500 ppb (same standard for PW and WFI)
- System suitability test: 1.19 ppm sucrose + 0.75 ppm 1,4-benzoquinone as standard solutions, with 85–115% recovery
- Both on-line and off-line modes are supported, but continuous monitoring is best practice
NVR vs TOC: they catch completely different things
| Item | NVR (Non-Volatile Residue) | TOC (Total Organic Carbon) |
|---|---|---|
| Principle | Evaporation + gravimetric | Oxidation to CO₂ + NDIR infrared absorption |
| Units | mg/L, ppm | ppb, µg C/L |
| Detection limit | ~0.1 mg/L (100 ppb) | 0.05 ppb (online instruments) |
| Detects | Non-volatile organics + inorganic salts | All organic carbon (volatile + non-volatile) |
| Misses | Volatiles with bp < 105 °C | Carbon-free impurities (metals, silicates) |
| Standard methods | ASTM E1235, SEMI C72 | USP <643>, SEMI C79 |
| Instrument cost | Oven + microbalance (< NT$500K) | Dedicated TOC analyzer (NT$800K–2.5M) |
| Analysis time | 4–8 hours / sample | 3–10 min / sample |
| Main applications | Cartridge cleanliness, solvent purity | UPW, WFI, cleaning validation |
| Online monitoring | None | Standard practice (24/7 real-time) |
Detailed test workflow
Whether NVR or TOC, 80% of the cartridge testing essence is in sample preparation, and only 20% in the instrument. For typical semiconductor / pharmaceutical cartridge extractables testing, the standard steps are:
1. Pre-flushing (Flushing / Conditioning)
New cartridges always have an initial peak that won't wash out — manufacturing, packaging, and shipping cannot be perfectly clean. SEMI C72 spec: pre-flush with 18 MΩ·cm UPW for at least 30 min or 10 cartridge volumes. For pharmaceutical applications, follow each plant's SOP — typically IPA → WFI → repeated rinsing until TOC stabilizes.
2. Extract collection
Use a standard cleaning fluid (semiconductor: UPW; pharma: WFI; chemical cartridges: corresponding solvent in use) to collect extract via "static soaking" or "dynamic recirculation":
- Static soaking: cartridge submerged in solvent, sampled at fixed time (24–72 hr) + fixed temperature (25–70 °C)
- Dynamic recirculation: recirculate at a constant flow rate through the cartridge, taking outlet samples at intervals for analysis
3. Analysis
- NVR: 100 mL sample → evaporate at 105 °C → weigh on microbalance
- TOC: 5–10 mL sample (typically auto-injected) → oxidation → NDIR → calculate ppb
4. Identification (optional)
When NVR or TOC is abnormally elevated and contamination needs to be traced, further analysis is performed using GC-MS (volatile species) or LC-MS (polar / non-volatile species) to fingerprint specific molecules in the dried residue or concentrated extract. Advanced semiconductor manufacturers also run ICP-MS for metal extractables to build a complete extractables profile.
NVR / TOC comparison across four major materials (public industry ranges)
Putting four cartridge materials — PP, Nylon, UPE, PTFE — through identical extractables testing (SEMI C72 conditions, UPW 25 °C dynamic recirculation) yields large differences. Below are typical industry-disclosed ranges:
| Material | NVR (mg/L) | TOC (ppb) | Notes |
|---|---|---|---|
| PP (polypropylene) | 2.0–8.0 | 200–1500 | Contains antioxidants + oligomers |
| Nylon-6,6 | 1.5–5.0 | 150–800 | Adipic acid / hexamethylenediamine monomer residue |
| UPE (ultra-high-molecular-weight PE) | 0.05–0.3 | 5–50 | No additives, extremely low leaching |
| PTFE | 0.05–0.2 | 3–30 | Chemically inert, low TOC |
The table reveals two things:
- PP / Nylon are inherently "high background" materials — it's not that the manufacturer has poor quality, but that the materials themselves contain process additives. So PP prefilters are well suited for coarse filtration — cheap and durable — but should never sit in the last filter position of a UPW system.
- UPE and PTFE are inherently "low-leaching" materials — their molecular chains are inert, with no antioxidants, plasticizers, or residual monomers. This is why semiconductor advanced-process final filters (point-of-use) are almost exclusively UPE or PTFE.
JIUNYUAN lab data: NVR and long-chain hydrocarbons from a 10 nm PTFE filter
The previous section shows ranges from public industry data. This section replaces them with one set of real measurements from the JIUNYUAN laboratory: the same 10 nm PTFE filter model, tested once as a new cartridge and once after reclaim, for NVR, with GC-MS used to check C12–C30 long-chain hydrocarbons (LHC).
| Test item | Unit | Spec limit (10 nm grade) | New cartridge | Reclaimed cartridge |
|---|---|---|---|---|
| Non-volatile residue (NVR) | mg/device | 18 | 3.5 | 1.8 |
| NVR as share of spec limit | — | — | 19.4% | 10.0% |
| C12–C30 long-chain hydrocarbons | — | — | < 0.5 ppm | < 0.5 ppm |
Three things to read in these numbers
- The unit is per cartridge, not per liter of extract. mg/device is the total non-volatile mass released by the whole cartridge; the mg/L values in the previous table are concentrations in the extraction liquid. Converting between them requires the extract volume, so 3.5 cannot be compared directly with the 0.05–0.2 mg/L range above. When comparing CoAs from different suppliers, first confirm that the units and extraction conditions match.
- The new cartridge shows higher NVR than the reclaimed one. The new cartridge measured 3.5 mg/device, 19.4% of the spec limit; the reclaimed cartridge measured 1.8 mg/device, 10.0%. New cartridges carry a small amount of organic residue from manufacturing and packaging, which is one reason they need a pre-flush before going on line; the IPA pre-clean and 1% SPM oxidation steps in the reclaim process remove this residue together with organics adsorbed during use.
- NVR tells you how much, not what. Identifying which organics make up the residue takes qualitative GC-MS analysis. The JIUNYUAN GC-MS has a detection limit of 1–100 ppb, can use IPA, Hexane or OK73 as the extraction solvent, and targets LHC long-chain hydrocarbons and phthalate plasticizers. In this data set, C12–C30 LHC was below 0.5 ppm for both the new and the reclaimed cartridge. To run an extraction test with your own process solvent, see Filter Cleanliness Analysis Service.
Spec quick-reference: semiconductor UPW, pharmaceutical WFI, semiconductor photoresist
| Application | Key parameter | Limit (target) | Source |
|---|---|---|---|
| Semiconductor UPW (advanced process) | TOC | < 1 ppb | SEMI F63, ITRS |
| Semiconductor UPW (general process) | TOC | < 5 ppb | ASTM D5127 Type E-1 |
| Pharmaceutical WFI | TOC | < 500 ppb | USP <643>, EP 2.2.44 |
| Pharmaceutical Purified Water (PW) | TOC | < 500 ppb | USP <643> |
| Semiconductor cartridge (point-of-use) | NVR | < 1 mg/L | SEMI C72, plant-specific spec |
| Semiconductor cartridge (UPE high-end) | TOC extractables | < 50 ppb (after 30 min flush) | Entegris / Pall specifications |
| Photoresist / developer cartridges | Metal extractables | < 0.5 ppt (single element) | SEMI C79, fab spec |
| Injectable filters | Non-volatile extractables | Validation + chemical identification required | USP <1663>, <1664> |
Common errors and marketing claims
Frequently Asked Questions
Doing both NVR and TOC — isn't that redundant?
It's not redundant — they're complementary. NVR catches heavy components (plasticizers, oligomers), TOC catches all carbon-containing species (including small volatile molecules). High-end semiconductor cartridge manufacturers typically report both; pharma focuses on TOC, with NVR used as manufacturing QC. If budget forces a choice, look at the target fluid: pure-water applications prioritize TOC; solvent / chemical applications prioritize NVR.
Why does TOC spike high then drop after a new cartridge is installed?
That's called extractables peak. During manufacturing, shipping, and aseptic packaging, the cartridge surface and pores adsorb low-MW species, which are flushed out in the first wave once on-line. So the SOP must specify "flush to TOC stable" — typically 30 min – 2 hr to settle into spec. Advanced fabs do "conditioning recipes" after cartridge replacement, only releasing to production once TOC drops back to baseline.
Which has lower NVR — UPE or PTFE?
Public data puts both extremely low (< 0.3 mg/L) — the gap is small and easily masked by instrument noise. The real difference is chemical resistance: PTFE maintains low leaching even in strong acids/bases and strong oxidizers; UPE is best in neutral / weakly alkaline conditions and degrades under strong oxidizers (O₃, persulfate). So choose UPE for photoresist and developer, PTFE for HF / SC1.
Are portable TOC analyzers (under NT$1M) sufficient?
It depends on the use case. Pharmaceutical WFI (500 ppb spec): portable models often carry compliance certification, sufficient for routine QC. Semiconductor UPW (< 1 ppb spec): requires UV-persulfate ultra-low TOC online instruments (Sievers 500 RL, M9, Anatel A1000 class) — portable units don't have the detection limit.
Must extractables testing always use UPW as the solvent?
It depends on the actual fluid the cartridge will contact. Principle: use the actual operating fluid for extraction. Photoresist cartridges should be extracted with PGMEA; HF cartridges with diluted HF; pharmaceutical aqueous cartridges with WFI. Using UPW as a "universal test" only reflects leaching behavior under aqueous conditions and cannot represent solvent scenarios.
How do I convert mg/L on an NVR report to g/cm²?
You can't convert directly — you need to know the cartridge area and extraction volume. Area loading (mg/m²) = NVR (mg/L) × extraction volume (L) ÷ effective cartridge area (m²). For a 0.6 m² cartridge with 1 L extract at NVR 0.5 mg/L, that's 0.83 mg/m² area loading. International comparisons typically use area loading or mg/cm³ media volume loading.
References
- ASTM E1235 — Standard Test Method for Gravimetric Determination of Nonvolatile Residue (NVR) in Environmentally Controlled Areas
- USP <643> Total Organic Carbon — TOC specification for purified water and WFI
- SEMI C72 — Test Method for Determination of Trace Non-Volatile Residue in UHP Chemicals
- SEMI F63 — Guide for Ultrapure Water Used in Semiconductor Processing
- Veolia / Sievers 500 RL Online TOC Analyzer — UPW online TOC monitoring (0.05 ppb detection limit)
- Entegris — UPE membrane technical notes (semiconductor UPW final filter NVR / TOC data)
- Pall — Ultrapure Water Filtration (cartridge extractables and TOC specifications)
- ASTM D5127 — Standard Guide for Ultra-Pure Water Used in the Electronics and Semiconductor Industries
- USP <1663> / <1664> — Extractables and Leachables Assessment for Pharmaceutical Packaging Systems