240,000 Particles per Liter: Reading the Nanoplastics Data Before You Design a Filter

240,000 Particles per Liter: Reading the Nanoplastics Data Before You Design a Filter In January 2024 a paper in PNAS put a number on something people had been worrying about without measuring: roughly 240,000 detectable plastic particles in a liter of bottled water, about 90% of them below one micrometer. It is a striking figure, and it has been repeated ever since as though it settled the engineering question. It did not. Read alongside the toxicology, the certification standards and the measurement methods, that number defines a gap rather than a specification — and the gap is exactly where a hardware team can either waste a year or find a real product. For a hardware engineer. The question is not whether nanoplastics are present in the water; it is which claim can be printed and defended. That boundary is drawn in the certification section, and wording that survives a retailer's technical review is in the spec sheet section. For a quality lead. What NSF/ANSI 42, 53, 58, 61 and 401 actually certify — and the particle size band at which each one stops — is tabulated under the standards map. For a sourcing manager. What to require from a component supplier when no standard test method exists at all is set out in the spec sheet questions. For an NPI manager. Which subsystems decide whether a unit survives its warranty period, and how each one routes through manufacturing, is in the subsystem section. What the PNAS study actually measured The paper is Qian and colleagues, Rapid single-particle chemical imaging of nanoplastics by SRS microscopy, PNAS 121(3), published January 2024. The technique matters: it is the reason the number is so much larger than earlier ones. The team used hyperspectral stimulated Raman scattering microscopy, which probes a sample with two tuned lasers so that specific molecular bonds resonate, plus a data-driven spectral matching algorithm to identify polymer type particle by particle. What that buys is sensitivity and chemical specificity at the same time — the combination earlier methods could not deliver. Raman and FTIR microscopy could identify plastic but lacked nano-scale sensitivity. Electron microscopy and atomic force microscopy could see nano-scale objects but could not tell you what they were made of. Applied to bottled water, the platform returned these figures: About 2.4 ± 1.3 × 105 particles per liter — roughly 240,000, with a wide spread across samples. Roughly 90% of those are nanoplastics, defined here as particles below 1 µm. Seven common polymer types were identified, including PET, polyamide and polypropylene. The technique detects particles down to about 100 nm. Much of the plastic appears to originate from the bottle itself and from the reverse osmosis membrane filter used to treat the water before bottling. Two design-relevant facts sit inside that list. The first is that the packaging and the process are themselves sources: squeeze the bottle, heat it, or open and close the cap repeatedly and more particles come off. The second is that the instrument's floor is around 100 nm. Anything smaller is not counted, which means 240,000 is a lower bound on what is there, not a measurement of all of it. What the paper did not establish The paper counts and identifies particles. It does not measure harm, and the authors are careful about that. The stated motivation is that nanoplastics are believed to be more toxic than microplastics because their smaller size makes them more amenable to entering the body. The evidence cited for crossing biological barriers comes largely from studies using engineered particles with fluorescent or metal labels, not from this cohort of real-world samples. The toxicology baseline is set by the WHO's 2019 report Microplastics in drinking-water, which remains the reference assessment. Its conclusions, in its own terms: Microplastics in drinking water appear to pose a low concern to human health at current levels reported in the scientific literature, across all three hazard routes the report examined: the physical particle, the chemicals associated with it, and microbial biofilm. Particles larger than 150 µm are not likely to be absorbed by the human body, and uptake of smaller particles is expected to be limited. Absorption and distribution of very small particles including nano-sized ones may be higher — but the data is extremely limited. There is not enough information to draw a conclusion on the toxicity of nanoplastic particles, and no reliable information showing that they are a concern. WHO does not recommend routine monitoring of microplastics in drinking water. It recommends that suppliers and regulators prioritize microbial pathogens and chemicals of known risk, on the grounds that treatment optimized for those also removes microplastics. That last point is easy to misread as indifference. It is not. It is a resource-allocation position from a public health body, and it is accompanied by an explicit call for more research and for reducing plastic pollution. For a product team, though, the operational consequence is concrete: there is no regulatory driver forcing anyone to measure nanoplastics in drinking water, which means nobody has built the standardized method you would need to make a verified claim. The honest framingMeasured: particle counts and polymer identification in bottled water, from a small pilot. Not established: that these particles harm people at the levels found. Both halves of that sentence matter, and a product built on only the first half is built on a claim it cannot support. The certification boundary sits above the particles everyone is worried about Here is the part that surprises most teams, and it is the most actionable thing in this article. Microplastic reduction can be certified in the United States. The mechanism is NSF/ANSI 401, Drinking Water Treatment Units — Emerging Compounds/Incidental Contaminants, and NSF's official listings show "Microplastics Reduction" as a live certification claim on products from multiple manufacturers. The catch is the size definition. NSF's Joint Committee on Drinking Water Treatment Units reviewed the available research and set its microplastics definition at particles from 1 µm to 5,000 µm. Separately, the NSF/ANSI 42 particulate Class I reduction claim is based on particles in the 0.5 µm to 1 µm range, and NSF determined that the Class I claim could serve as an appropriate measure of microplastics reduction, with identical testing requirements under 401 and 42. Now put those two boundaries next to the PNAS finding: SourceSize rangeWhat it means for a product claim NSF/ANSI 401 microplastics definition1 µm – 5,000 µmThe certifiable band NSF/ANSI 42 particulate Class I0.5 µm – 1 µmThe test method used to support the claim PNAS 2024 finding in bottled water~90% of detected particles below 1 µm; instrument floor ~100 nmThe band that carries the headline number WHO 2019 absorption assessment>150 µm unlikely to be absorbed; nano-range data extremely limitedThe band where the toxicology question is open Read that table once more. A filter can be certified for microplastics reduction, and that certification covers particles above roughly half a micrometer. The particles that produced the 240,000 figure are overwhelmingly below one micrometer, and the particles whose health significance is most open are smaller still. The certificate and the headline number describe different populations. This is not a criticism of NSF, whose committee worked from the research available to it, and it is not an argument that certified filters do nothing. It is an argument about what a claim can be made to mean. If your spec sheet says "removes microplastics" and your certification is NSF/ANSI 401, you are making a true statement about a defined size band. If your marketing implies the 240,000 number, you have stepped outside the band your evidence covers. The measurement problem underneath all of it Even the counting is unsettled. In the briefing for the WHO report, Bruce Gordon, the report's coordinator for water and sanitation, was unusually blunt. One study may report a thousand particles per liter in a water source and another may report one, he said. The difference, he explained, comes down to the pore size of the filter used to prepare the sample. His summary was that the research methods are, at that stage, quite weak. WHO's own list of research needs begins with developing standard methods for measuring microplastic particles in water. That is the first item, before any question of health effects. For a hardware team, this has a hard consequence. You cannot write a verifiable incoming or outgoing specification for a quantity that has no agreed measurement method. You can specify a test protocol of your own and apply it consistently, which is better than nothing, but it is not the same thing as compliance, and it will not survive contact with a retailer's technical reviewer or a regulator's letter. The reverse-osmosis membrane is the counter-argument, and it is a real one. RO membranes are typically described with a pore scale around 0.0001 µm, which is three to four orders of magnitude below the nanoplastic band and below the instrument floor of the PNAS method. On physical grounds, a particle larger than the pore does not pass. What RO cannot give you is a certified claim for the size band in question, because the certification does not test it. The boiling result, and what it tells an engineer In February 2024, a group at Guangzhou Medical University and Jinan University published a result in Environmental Science & Technology Letters that reframes the whole category. Yu, Wang, Liu, Li and Zeng showed that nano/microplastics of polystyrene, polyethylene and polypropylene co-precipitate with calcium carbonate scale when tap water is boiled. The numbers, from the paper: Boiling hard water (above 120 mg/L CaCO3) removes at least 80% of those polymers across a size range of 0.1 µm to 150 µm. Removal climbs with temperature: from about 2% at 25 °C to 28% at 90 °C, then sharply to 84% once the water reaches 100 °C. Concentration in the test fell from 30 particles per microliter to 4.8. Efficiency tracks hardness: about 34% at 80 mg/L CaCO3, 84% at 180 mg/L, and 90% at 300 mg/L. Even soft water below 60 mg/L saw roughly 25%. The mechanism is encapsulation: calcium carbonate nucleates on the particles, encrusts them, and the aggregate settles into the scale you scrape off the kettle. Two things in that list should stop a product team in its tracks. First, the tested size range starts at 0.1 µm — which is the instrument floor of the PNAS method and well below the NSF certification band. A five-minute boil with a coffee filter addresses, on these measurements, a population that the certification system does not. Second, the particles are not destroyed. They move from the water into the scale. That is a materials-handling statement, and it is the kind of statement engineers are supposed to be precise about. What this does and does not meanIt does not mean filtration is pointless. The study used spiked samples of one city's tap water, it does not address pathogens, dissolved contaminants or taste, and boiling is not a product. What it means is that a claim of "removes nanoplastics" has a low-cost competitor on the very size band it would target, and that competitor's evidence is published in a peer-reviewed ACS journal. What a point-of-use device actually has to do If the nanoplastic claim is a dead end as a headline, the product category is not. Households buy treatment devices for reasons that are measurable and certifiable, and those reasons are where the engineering should go. A typical under-sink or countertop unit breaks down as follows, and every line is work we do: SubsystemComponentsManufacturing routeWhat actually gets specified Pressure boundaryHousings, manifolds, port fittingsInjection molding for volume; CNC for low volume, metal manifolds and prototypesBurst and cyclic pressure, creep, material compatibility with potable water Separation mediaSediment pre-filter, carbon block, RO membraneBought-in media, assembled in-houseCertified reduction claims per contaminant, service life in gallons, flow rate HydraulicsPump, solenoid valve, flow restrictor, check valvesSourced, integratedFlow rate, pressure drop, acoustics, duty cycle DisinfectionUV-C LED assemblyPCBA plus optical and thermal designDose, LED lifetime, thermal path, optical safety interlock SensingTDS / conductivity probe, flow meter, leak sensorPCBA plus potted or isolated probesAccuracy over temperature, fouling behavior, calibration drift Control and connectivityMCU board, filter-life logic, display, wireless moduleEMS: SMT, conformal coating, box buildStandby power, EMC, firmware update path, data handling Sealing and assemblyO-rings, potting, ultrasonic or spin weldsProcess engineeringLeak rate at production, not just at qualification Two failure modes dominate the field, and neither has anything to do with nanoplastics. The first is leakage: a slow seep under a sink causes more customer damage and more brand damage than any contaminant claim recovers. The second is filter-life indication that drifts, because a replace-on-time reminder that ignores actual throughput is wrong in both directions. Both are manufacturing and firmware problems. See our PCBA testing guide for how in-circuit and functional coverage catches the electrical half of this, and conformal coating for what protects a board living in a damp cabinet. On the machining side, housings and manifolds are where CNC-machined housings and enclosures earn their place in low and mid volume: the tooling for a molded housing is a commitment you make once, and a machined manifold lets you change port geometry between pilot runs. Our plastics machining guide covers the material behavior that matters when the housing itself is machined rather than molded. The standards that apply, and the ones that don't Part of the value of mapping this is knowing which claims have a testing apparatus behind them. This is the short version. StandardScopeStatus for a nanoplastics claim NSF/ANSI 42Aesthetic effects: chlorine taste and odor, particulate Class I (0.5–1 µm)Usable. Class I particulate is the test route for microplastics reduction NSF/ANSI 53Health-related contaminant reduction: lead, cysts, VOCs, PFASUsable and valuable, per contaminant. Microplastics not a health claim here NSF/ANSI 58Reverse osmosis systems: TDS reduction mandatory, many optional contaminantsUsable. The broadest performance certification for an RO unit NSF/ANSI 401Emerging compounds, microplastics defined 1–5,000 µmUsable within the 1 µm floor. Does not cover the sub-micron band NSF/ANSI 61Health effects of drinking water system components (material safety)Usable. Covers what the device leaches, not what it removes IEC 60335-1Safety of household appliances, with applicable part 2 sectionsMandatory in practice for the electrical and thermal safety case EU 10/2011Plastic materials intended to come into contact with foodApplies to wetted plastic parts for the EU market EU 2020/2184Drinking water quality directive, with a watch list mechanism for substances of emerging concernSets the EU frame. It does not give you a nanoplastics claim WHO 2019 assessmentRisk characterization, not a product standardThe reason no regulator is forcing routine measurement The pattern across that table is the actual finding of this article. There is a mature, well-instrumented certification system for water treatment, and it stops just above the particle size that produced the headline. Engineering around that boundary is not cynicism; it is the difference between a claim that survives a challenge and one that does not. A spec sheet that survives contact with reality If you are specifying a point-of-use device, or evaluating a supplier's proposal, these are the questions that produce answers you can defend. They are grouped by who needs them, because a procurement lead, an NPI engineer and a quality engineer are buying different evidence from the same meeting. Name the contaminant and the standard number. "Certified" without a standard number and a contaminant list is not a claim. "Certified to NSF/ANSI 53 for lead and cyst reduction" is. Ask for the size band behind any particle claim. If a supplier says microplastics, ask whether the claim rests on NSF/ANSI 401 or 42 Class I, and note that both have a floor around one micrometer. Specify leakage as a production metric, not a qualification one. A pressure-hold or helium test on every unit costs less than one warranty claim on a flooded kitchen. This is the single highest-value line in the whole specification. Define filter-life indication. Time-based, volume-based, or sensor-based, and what happens at end of life. Drift in a conductivity probe is a calibration problem with a customer-facing consequence. Separate wetted-material compliance from performance. NSF/ANSI 61 and EU 10/2011 govern what the device adds to the water. NSF/ANSI 42/53/58/401 govern what it takes out. A proposal that conflates them is incomplete. Require a hygiene story for the wet side. Stagnant water, warm cabinets, and carbon media are a microbial growth problem long before they are a nanoplastic one, and WHO's position is that microbial risk is the priority. Put acoustic and standby power numbers in the contract. Both are noticed daily, and neither appears on a contaminant reduction sheet. Our EVT, DVT and PVT gates article sets out where each of these belongs in a development program, and low-volume EMS covers how certification lots and pilot builds are handled when the first order is in the hundreds. What we would tell you if you brought us this brief We build the metal, plastic and electronic parts of devices like this in Dongguan: molded and machined housings and manifolds, the control PCBA, the sensor and UV-C assemblies, and the final box build. If you came to us with a brief whose headline was "removes nanoplastics from drinking water," this is what we would say. We would ask what you can verify, and how. There is no standard method for counting nanoplastics in water, which WHO lists as its first research need. Without one, any performance number you print is a protocol of your own devising, and it will be challenged the first time a retailer's technical reviewer reads it. We would point at the certification boundary. NSF/ANSI 401 and 42 Class I are real, respected, and bounded at roughly half to one micrometer. That is a defensible claim with a defined meaning, and it is not the claim the viral number implies. We would push the engineering toward what is measurable. Lead, cysts, PFAS, chlorine taste and odor, TDS, arsenic, nitrate: every one of these has a standard, a test method, and a certification route. So does leak rate, acoustics, flow rate, service life and standby power. Those are the fields where a device can actually be better than the one next to it, and where better is demonstrable. We would build the parts that decide whether it survives the warranty period. A housing that does not creep, a manifold that does not weep at the port, a board that is clean and coated for a damp cabinet, a conductivity probe that still reads correctly in year two. None of that is glamorous and all of it is what the returns data will show. Our electronics DFM review starts exactly there, and injection molding covers the tooling commitment you make once and live with. FAQ Is bottled water dangerous because of nanoplastics? The evidence does not support that conclusion. The PNAS study measured and identified particles; it did not measure harm. WHO's 2019 assessment concluded that microplastics in drinking water appear to pose a low concern to human health at current reported levels, while noting that data on the nano size range is extremely limited and that more research is needed. The honest answer is that the exposure is measured and the risk is not established. Can a water filter be certified to remove microplastics? Yes. NSF/ANSI 401 covers emerging compounds and includes a microplastics reduction claim, and NSF's official listings show it in use. The definition applied is particles from 1 µm to 5,000 µm, and the test route is the same particulate Class I procedure used under NSF/ANSI 42, which is based on the 0.5 to 1 µm range. Particles below that band — which is where about 90% of the particles in the PNAS study sit — are outside what the certification tests. Does reverse osmosis remove nanoplastics? On physical grounds, a membrane with a pore scale around 0.0001 µm will reject particles far larger than that. The problem is not physics, it is evidence: the certification system does not test the sub-micron band, so you cannot convert the physical argument into a certified claim for that size range. RO is nonetheless the strongest performance certification available for a point-of-use unit, through NSF/ANSI 58. Why does WHO not recommend routine monitoring for microplastics in drinking water? Because it recommends prioritizing known risks. WHO's position is that microbial pathogens and chemicals such as arsenic and lead are established causes of harm, and that treatment optimized for those also removes microplastics. It pairs that with a call for more research and an explicit research need for standardized measurement methods. Does boiling water actually reduce nanoplastics? Published results say yes, with conditions. In Environmental Science & Technology Letters (2024), boiling hard water above 120 mg/L CaCO3 removed at least 80% of polystyrene, polyethylene and polypropylene particles between 0.1 and 150 µm, rising to about 84% at 100 °C and up to 90% at 300 mg/L hardness. Efficiency falls with hardness — soft water saw around 25% — and the particles are transferred into the scale rather than destroyed. What should a water treatment device actually compete on? Contaminants with a standard behind them: lead and cysts under NSF/ANSI 53, TDS and a long list of contaminants under NSF/ANSI 58, chlorine taste and odor under NSF/ANSI 42. And on the things customers notice every day but no contaminant sheet lists: leak rate, acoustics, flow rate, filter-life accuracy, standby power and service life. Is there a standard method for measuring nanoplastics in water? No, not an agreed one. WHO lists developing standard methods for measuring microplastic particles as its first research need. Reporting in the field varies by orders of magnitude between studies, substantially because sample preparation filters differ in pore size. That is the reason a nanoplastics performance claim is difficult to defend regardless of how the product performs. Sources Qian N, Gao X, Lang X, et al. Rapid single-particle chemical imaging of nanoplastics by SRS microscopy. Proceedings of the National Academy of Sciences. 2024;121(3):e2300582121. doi:10.1073/pnas.2300582121 World Health Organization. WHO calls for more research into microplastics and a crackdown on plastic pollution (22 August 2019) — conclusions and research needs of the report Microplastics in drinking-water. Yu Z, Wang J-J, Liu L-Y, Li Z, Zeng EY. Drinking Boiled Tap Water Reduces Human Intake of Nanoplastics and Microplastics. Environmental Science & Technology Letters. 2024;11(3):273–279. doi:10.1021/acs.estlett.4c00081 NSF International. Certified Drinking Water Treatment Units listings — NSF/ANSI 401 emerging compounds, including microplastics reduction claims, and NSF/ANSI 42 particulate Class I. NSF/ANSI 42, 53, 58, 61 and 401 — Drinking Water Treatment Units, aesthetic effects; health effects; reverse osmosis systems; drinking water system components; emerging compounds. IEC 60335-1, Safety of household and similar electrical appliances, with applicable part 2 sections. Commission Regulation (EU) No 10/2011 on plastic materials and articles intended to come into contact with food; Directive (EU) 2020/2184 on the quality of water intended for human consumption. This article reflects the state of publicly available research, standards and certification listings as of September 5, 2026. Standard editions, certification scopes and regulatory positions change; verify current versions with NSF, IEC and the relevant authority before relying on them for a compliance decision. Nothing here is medical advice, and nothing here is a claim about the health effects of nanoplastics, which remain an open research question. Photographs of the Nex-G facility are not yet available for this article; product imagery is being sourced. Building a water treatment device?The claims that survive are the ones with a standard behind them. Nex-G machines and molds housings and manifolds, builds the control PCBA, and does the box build — in Dongguan, from prototype through volume.Request a quoteMore articles Related articlesCNC Machined Housings and EnclosuresInjection Molding in China: A Buyer's GuidePCBA Testing GuideLow-Volume & Prototype EMS in China