PFAS are not a single chemical but a sprawling family of more than 10,000 synthetic fluorinated compounds, defined under the framework adopted by the Organisation for Economic Co-operation and Development and the United Nations Environment Programme by the presence of carbon-fluorine bonds. That bond, one of the strongest in organic chemistry, makes these substances extraordinarily resistant to degradation, earning them the nickname forever chemicals. Their persistence, tendency to bioaccumulate and documented associations with adverse health effects have placed them at the centre of public health concern worldwide. In cosmetics, PFAS have historically served useful functions: they improve spreadability, create water-resistant films, and help pigments adhere to skin for hours. It is precisely these performance benefits that made them attractive formulators’ tools, and precisely these properties that now make them regulatory targets.

The new analysis, conducted by Margaux Vincent-Graindorge and Lionel Ripoll, took a deliberately simple approach: keyword screening of the International Nomenclature of Cosmetic Ingredients, or INCI, lists that regulations require on every product label. The dataset covered 1,233 products spanning three categories: 500 makeup items, 111 oral care products, and 622 other personal care products such as skin creams and cleansers. The results were immediately intriguing. Just over 19 percent of the products declared at least one fluorine-containing ingredient on their labels. But when the researchers applied the strict OECD and UNEP definition of a PFAS, requiring an actual carbon-fluorine bond, the number collapsed to a mere 0.9 percent of products. In other words, fewer than one in a hundred products declared an ingredient that legally qualifies as a PFAS.

The ingredient-level classification reveals why the gap exists, and it is not evidence of concealed chemicals. Almost all of the declared fluorine in these products belongs to compounds with no carbon-fluorine bond at all. The dominant contributor is synthetic fluorphlogopite, a synthetic mica prized in makeup for its shimmer and purity, which appeared in 134 products. The second major contributor is a category of dental fluoride salts, such as those used in toothpastes to prevent cavities, found in 88 products. These inorganic fluorides are chemically worlds apart from PFAS: they lack the persistent organofluorine backbone that defines the forever-chemical family, yet a casual reading of a label mentioning fluoride or fluorphlogopite could easily lead a consumer, or even a careless screener, to lump them together with PFAS.

Perhaps the most visually striking finding is how strongly fluorine declaration is structured by product type. In oral care products, 73 percent declared at least one fluorine-containing ingredient, an unsurprising figure given the near-universal use of cavity-preventing fluoride salts in toothpaste. In makeup, the figure was 28.2 percent, driven overwhelmingly by fluorphlogopite and related pearlescent agents. Across all other personal care products, the figure plummeted to just 1.9 percent. The authors report that this distribution is highly statistically significant, with a chi-squared statistic of 355.7 across two degrees of freedom and a p-value below 0.001. The pattern tells a coherent story: when fluorine appears on a cosmetic label, it almost always has an innocent, well-understood explanation rooted in the product’s function, not in the stealth use of forever chemicals.

So why have analytical studies repeatedly found fluorinated compounds in cosmetics at levels that seem to dwarf what labels declare? The answer, the researchers argue, lies in the difference between two families of measurement techniques. Targeted methods, typically using liquid chromatography coupled with tandem mass spectrometry, identify and quantify specific PFAS compounds against reference standards. They are precise but blind to anything outside their target list, which is a serious limitation in a chemical family of 10,000 members. Total-fluorine methods, by contrast, such as combustion ion chromatography and the extractable organic fluorine approach, capture every fluorine atom in a sample regardless of compound identity. These methods have powered many of the alarming headlines, because they routinely report fluorine masses far exceeding what any declared PFAS ingredient could explain.

The critical insight from the new study is that total fluorine is not synonymous with intentionally added PFAS. When the label data and the analytical surveys are placed side by side, the deliberate use of PFAS as functional ingredients in finished cosmetic formulations appears to be uncommon. The organofluorine that analytical surveys detect, the authors contend, most likely enters products through routes that no ingredient list could ever capture: process water used in manufacturing, residues on shared production equipment, or fluorinated contamination present in treated raw materials before they ever reach the formulator. This distinction matters enormously for policy, because the regulatory frameworks now proliferating around the world are almost universally keyed to intentional addition, targeting PFAS that manufacturers deliberately put into their products through defined substance lists.

The regulatory landscape is indeed tightening rapidly. The European Union operates restrictions under the REACH regulation and the Persistent Organic Pollutants regulation, while the European Chemicals Agency has broader restrictions under consideration. France has enacted a dedicated law on products containing PFAS, adopted in February 2025. Canada has proposed measures targeting forever chemicals, including action on firefighting foams, and has published a state-of-PFAS report assessing the whole class. Yet if a meaningful share of the fluorine in cosmetics arrives as unintended contamination rather than deliberate formulation, these frameworks, however well intentioned, may miss the very exposure pathways they aim to control. A regulation that bans intentionally added PFAS cannot, by construction, address fluorine that rides in on process water or a treated raw material.

The study also carries a cautionary tale for how science communicates risk. Several widely shared claims about PFAS in cosmetics rest on total-fluorine measurements interpreted through the lens of intentional addition, a combination that can inflate perceived prevalence. A sunscreen study that demonstrated transdermal absorption of a radiolabelled PFAS, and biomonitoring work linking personal care product use to serum PFAS levels, underscore that dermal exposure is a genuine pathway worth studying. But the new INCI analysis suggests that the magnitude of that pathway from finished cosmetics may be smaller than feared, and that the field needs better tools to distinguish listed ingredients from residual contamination. The authors point toward suspect screening workflows using high-resolution mass spectrometry, and toward fluorine mass-balance approaches that reconcile total fluorine with identified compounds, as the way forward.

For consumers, the practical takeaway is one of measured reassurance rather than complacency. The vast majority of fluorine appearing on cosmetic labels is synthetic mica in a shimmering eyeshadow or a cavity-fighting fluoride salt in toothpaste, neither of which is a PFAS in the regulatory or toxicological sense. For policymakers, the message is more demanding: if forever chemicals are sneaking into cosmetics through water, equipment and raw materials, then protecting the public will require controlling contamination at the supply-chain level, not merely policing ingredient lists. And for the scientific community, the study is a reminder that a keyword on a label and a fluorine atom in a chromatogram are two very different kinds of evidence, and that conflating them risks distorting both public understanding and the regulations meant to keep people safe. The full INCI dataset and the Python screening script behind the analysis are available from the corresponding authors, offering other researchers a transparent starting point for replicating and extending the work.

Subject of Research: Declared versus actual presence of PFAS in cosmetic products based on INCI ingredient list analysis

Article Title: Declared versus actual presence of PFAS in cosmetic products: an INCI list analysis and critical appraisal of current regulatory frameworks

Article References: Vincent-Graindorge, M., & Ripoll, L. (2026). Declared versus actual presence of PFAS in cosmetic products: an INCI list analysis and critical appraisal of current regulatory frameworks. Environmental Science and Pollution Research, 33(30), 15883-15896. https://doi.org/10.1007/s11356-026-38214-9

Image Credits: AI Generated

DOI: 10.1007/s11356-026-38214-9

Keywords: PFAS, cosmetics, INCI, fluorine, fluorphlogopite, regulation, dermal exposure, total fluorine, extractable organic fluorine, OECD, forever chemicals, personal care products

Proton exchange membrane water electrolyzers, or PEMWE systems, split water into hydrogen and oxygen using an acidic polymer membrane sandwiched between two electrodes. They are prized for their high efficiency and rapid dynamic response, which makes them natural partners for intermittent renewable electricity. But achieving the target lifetimes of more than 80,000 hours remains a critical challenge. Despite intensive research into degradation, the transferability and comparability of insights across different systems remain limited, and the field has struggled to generate reproducible data that meaningfully compare one laboratory’s results with another’s.

The core difficulty is manufacturing variability. Even when two catalyst coated membranes nominally contain the same membrane material, differences in production—solvent content, catalyst layer structure, interfacial contact quality, and microscopic defects such as pinholes—can produce markedly different operating behavior. Because manufacturers rarely disclose their production processes, these influences are difficult to assess directly. Harmonized testing protocols and interlaboratory round-robin initiatives have tried to close the gap, but they cannot fully account for variations baked into nominally identical materials before a cell is ever assembled.

The Graz team, led by Özge Kiziltan with colleagues including Viktor Hacker and Merit Bodner, focused on a metric that is almost trivially easy to measure: the open circuit voltage, or OCV, recorded for 120 seconds immediately after the conditioning step. In fuel cells, OCV is a well-established indicator of membrane health, because a degraded, gas-permeable membrane produces mixed potentials at the electrodes that drag the measured voltage down. In water electrolyzers, by contrast, only a handful of studies have examined OCV behavior, even though deviations from the ideal Nernst potential are known to be linked mainly to gas crossover through the membrane, with secondary contributions from internal leakages, catalyst oxidation states, and hydration differences.

The researchers tested two commercially available catalyst coated membranes that both used the same Nafion 115 membrane material, one from Quintech and one from IonPower’s Hydrion line. Each sample underwent a 12-hour hydration period followed by electrochemical activation, after which the team recorded the 120-second OCV response, polarization curves, and electrochemical impedance spectra. The cells were 4 square centimeter single-cell fixtures operated at 80 degrees Celsius with ultrapure water circulated through the anode at 40 milliliters per minute. To probe robustness, the experiments were run in two independent configurations using different flow plates—one titanium, one platinized titanium—and two different activation protocols ending at either 2 or 3 amperes per square centimeter.

The results were striking. One group of membranes, designated CCM.A, exhibited a rapid decline in open circuit voltage, plunging to a minimum voltage of roughly 0.32 volts within ten seconds of current interruption. The other group, CCM.B, showed a far more gradual response, with the voltage continuing to decrease throughout the entire 120-second hold and stabilizing near 1.21 volts. In a direct comparison between two samples measured with identical flow plates and identical activation protocols, the difference in minimum voltage reached 0.357 volts—an enormous gap for devices built around the same membrane material. Notably, switching the flow plate material barely changed the OCV decay of CCM.A, with the coefficient of variation rising only from 4.9 to 5.1 percent, suggesting the effect is intrinsic to the membrane assembly rather than an artifact of the cell hardware.

The physics behind the signal is subtle. Open circuit conditions are not purely passive; capacitive relaxation, changes in electrode oxidation state, and gas crossover all overlap within the measurement window. When current stops flowing, the overpotentials built up during operation relax quickly, which explains an initial voltage drop. But fast relaxation alone cannot account for the extremely low minimum voltages observed for CCM.A. Instead, the researchers argue, parasitic electrochemical reactions driven by gas crossover—hydrogen oxidation or oxygen reduction at the opposing electrode—establish mixed potentials that collapse the measured cell voltage. Prior work cited in the study shows that even crossover fluxes as small as about one nanomole per square centimeter per second can shift electrode potentials by several millivolts per second under open circuit conditions, so tiny leaks can dominate the response.

Crucially, the OCV signature predicted real operational trouble. During a 100-hour constant current hold, the sample with the unstable OCV response showed a pronounced but fully reversible voltage transient: interruptions for characterization triggered a voltage increase peaking around six hours after returning to constant current, which then recovered over roughly twenty hours. The reversible nature and long duration of this effect point to crossover-induced mixed-potential phenomena rather than irreversible catalyst or membrane degradation. The membrane with the stable OCV response, operated at an even lower current density, showed no comparable reversible voltage loss at all. Reapplying current gradually restores the catalyst oxidation states, which explains why the elevated overpotentials fade with time.

The team also examined whether conventional membrane degradation metrics could explain the OCV differences, and here the answer was a clear no. Using combustion ion chromatography to measure total fluorine and ion-selective electrodes to measure fluoride ions in the effluent water, they found that only the more stable membrane group produced quantifiable concentrations, while samples from the rapidly decaying group fell below detection limits—likely because the large volume of circulating anode water dilutes emitted fluorine species below what instruments can detect. Total fluorine values exceeded fluoride values by about 9.9 percent on average, confirming the presence of non-ionic fluorinated fragments that conventional fluoride analysis misses. Most importantly, no clear correlation emerged between these degradation metrics and OCV behavior, underscoring that fluoride emission and total fluorine are poorly suited as universal early-stage indicators, particularly given detection limits and sampling dilution.

The study’s conclusions are deliberately two-sided. On one hand, the early OCV response proved repeatable within a controlled material set and offers a practical, observation-based screening tool: two minutes of measurement, requiring no specialized instrumentation, can flag membranes susceptible to transient instability before more obvious degradation indicators appear and before operational safety is compromised. On the other hand, the researchers caution that OCV behavior is not a global metric. It reflects the combined influence of gas crossover, electrode-state effects, and conditioning history—the final activation current density demonstrably shaped the response—and nominally identical materials from different manufacturers produced irreconcilable differences. The authors call for reproducibility-aware diagnostic frameworks, comprehensive reporting of material properties, and healthy skepticism when comparing metrics across systems based solely on nominal material equivalence. For a hydrogen economy racing toward durability targets measured in decades of operation, the message is that sometimes the most informative diagnostic is also the simplest one: watch what the voltage does when you let go.

Subject of Research: Open circuit voltage diagnostics for proton exchange membrane water electrolysis

Article Title: Initial open circuit voltage response as a practical diagnostic approach in proton exchange membrane water electrolysis

Article References: Kiziltan, Ö., Preihs, M. A., Raab, A., Feldmann, J., Hacker, V., & Bodner, M. (2026). Initial open circuit voltage response as a practical diagnostic approach in proton exchange membrane water electrolysis. Discover Electrochemistry, 3(1), Article 58. https://doi.org/10.1007/s44373-026-00147-w

Image Credits: AI Generated

DOI: 10.1007/s44373-026-00147-w

Keywords: proton exchange membrane water electrolysis, open circuit voltage, green hydrogen, gas crossover, catalyst coated membrane, membrane degradation, fluoride emission, total fluorine, electrochemical diagnostics, Nafion 115, durability testing, mixed potential