Nencioni et al. 2026 — In vitro Franz-cell adsorption of Cd/Co/Cr/Pb/Ni by a prototype zeolite-containing anti-pollution cream
This Materials paper evaluates an experimental zeolite-containing cosmetic cream (3 g zeolite per 100 g cream) as an anti-pollution topical formulation, using a Quality by Design (QbD) workflow and the Franz diffusion cell model to quantify in vitro adsorption and retention of five heavy metals (Cd, Co, Cr, Pb, Ni) from a 0.04 ppm spiked acidic metal solution. A 2 × 2 × 2 full factorial design (membrane type Strat-M® vs silicone; dosage 10 vs 20 mg; dosage regimen infinite vs finite) followed by a 12-cell physiological-conditions permeation study (Strat-M® + PBS pH 7.4; 32 ± 1 °C; 12 h) was applied. The zeolite formulation adsorbed Cr, Co, and Cd statistically significantly more than the placebo (one-way ANOVA p = 0.026, p < 0.001, p = 0.004 in Experiment 1; F(11) = 26.88, 127.4, 35.59 with p < 0.001 for Cr, Co, Ni in Experiment 2), while Pb and Ni patterns were less consistent (Pb p = 0.147 in Exp 1; F(11) = 4.6, p = 0.58 in Exp 2; Cd F(11) = 11.03, p = 0.08 in Exp 2). The optimal QbD operating window (composite desirability D = 0.89, ≈ 0.8851 in the run summary) was Strat-M® membrane + finite dosage regimen + 20 mg dosage. The authors estimate ≈ 23% of the observed adsorption is specifically attributable to zeolite (the remainder reflecting cream-matrix retention). This is in vitro proof-of-concept mitigation evidence, not contamination occurrence data; the “metal concentrations” reported throughout are residuals from a deliberately spiked donor solution.
Key numbers
- Sample basis (Section 2.1, p. 2; Section 2.2.1-2.2.2, p. 3): single prototype cream formulation, 3 g zeolite per 100 g cream, 17.3% lipophilic phase, herbal-derived ingredients including plant waxes and Rosmarinus officinalis leaf extract (Supplementary Table S1). Spiked donor solution: mixed aqueous Cd + Co + Cr + Pb + Ni at 0.04 ppm each. Receptor compartment in Experiment 1: diluted acidic metal solution mimicking interstitial-skin-fluid acidity. Receptor compartment in Experiment 2: PBS pH 7.4 (Na₂HPO₄ 5.98 g/L, KH₂PO₄ 0.19 g/L, NaCl 8.8 g/L, adjusted with H₃PO₄). Both experiments held at 32 ± 1 °C with continuous magnetic stirring. Endpoint timing differs by experiment: Experiment 1 (adsorption screening) measured residuals after 24 h of incubation (Section 3.3, p. 7); Experiment 2 (physiological permeation) ran 12 h (Section 2.2.2, p. 3; Section 3.4, p. 9).
- Donor-spike concentration was 0.04 ppm per metal (Section 2.1, p. 2). For context, the printed residual donor + membrane concentrations in Table 4 (Experiment 2) are in the 0.3-2.4 ppm range, an order of magnitude above the donor-spike; the authors attribute the higher absolute levels in part to background metal content in the placebo cream’s herbal-derived ingredients (Discussion p. 10: “Pb levels detected in the placebo may be related to trace amounts naturally present in some herbal-derived raw materials … <10 ppm for Pb and Ni” per supplier specs).
- Experimental factorial domain (Table 2, p. 6): Membrane (Silicone vs Strat-M®); Dosage (10 mg vs 20 mg); Dosage regimen (0 infinite vs 3 finite). NOTE: Table 2 prints the “Dosage” row values as 0 / 3 and the “Dosage regimen” row values as 10 / 20, which is the inverse of the body-text definitions (Section 3.1, p. 5: “dosage … 10 mg and 20 mg” and “dosage regimen … finite dosing conditions”). Table 3 column headers carry the same swap. The numerical values themselves are unaffected; only the column/row labels are transposed. The reading used throughout this page follows the body-text definitions.
- Run-level residuals in the donor + membrane compartment (Table 3, p. 6; metal concentrations in ppm; CCD configuration with 12 runs, α = 2, 8 factorial points + 4 axial points). Columns rendered as (Run | Membrane | Dosage mg | Regimen | Cr | Co | Cd | Ni | Pb):
- Run 1: Strat-M® | 10 | 3 | Cr 0.325 | Co 0.385 | Cd 0.415 | Ni 0.367 | Pb 0.385.
- Run 2: Silicone | 20 | 3 | Cr 0.398 | Co 0.387 | Cd 0.399 | Ni 0.340 | Pb 0.358.
- Run 3: Strat-M® | 20 | 3 | Cr 0.169 | Co 0.214 | Cd 0.227 | Ni 0.237 | Pb 0.289.
- Run 4: Strat-M® | 10 | 0 | Cr 0.353 | Co 0.451 | Cd 0.441 | Ni 0.404 | Pb 0.396.
- Run 5: Silicone | 10 | 3 | Cr 0.336 | Co 0.428 | Cd 0.431 | Ni 0.403 | Pb 0.404.
- Run 6: Silicone | 10 | 0 | Cr 0.338 | Co 0.424 | Cd 0.436 | Ni 0.371 | Pb 0.393.
- Run 7: Silicone | 20 | 0 | Cr 0.405 | Co 0.411 | Cd 0.397 | Ni 0.373 | Pb 0.313.
- Run 8: Strat-M® | 20 | 0 | Cr 0.396 | Co 0.403 | Cd 0.406 | Ni 0.395 | Pb 0.345.
- Run 9: Silicone | 10 | 3 | Cr 0.347 | Co 0.417 | Cd 0.425 | Ni 0.408 | Pb 0.412.
- Run 10: Silicone | 20 | 3 | Cr 0.405 | Co 0.385 | Cd 0.389 | Ni 0.414 | Pb 0.397.
- Run 11: Strat-M® | 20 | 3 | Cr 0.232 | Co 0.248 | Cd 0.285 | Ni 0.299 | Pb 0.302.
- Run 12: Strat-M® | 10 | 3 | Cr 0.329 | Co 0.388 | Cd 0.402 | Ni 0.389 | Pb 0.393.
- Lowest-permeation runs (greatest adsorption): Run 3 and Run 11 (Strat-M®, 20 mg, finite regimen), with Cr 0.169-0.232 ppm, Co 0.214-0.248 ppm, Cd 0.227-0.285 ppm (Section 3, p. 6).
- Optimisation by desirability function (Section 3.2, p. 7): maximum overall D = 0.89 (composite 0.8851) at Strat-M® + finite regimen + 20 mg dosage. Individual analyte desirability values for Pb, Ni, Cd, Co, and Cr “consistently above 0.82 and reaching up to 0.95 for Pb.” The body text mentions a “30 mg” optimal in one sentence; this appears to be a typographic inconsistency with the 20 mg upper bound of the factorial domain and is documented in Verification notes.
- Experiment 2 (Section 3.4, p. 9; physiological permeation; 12 Franz cells with Strat-M®; receptor PBS pH 7.4; no receptor sampling — donor + membrane residuals only). Donor + membrane residual concentrations after 12 h (Table 4, p. 9; ppm):
- Cell 1A (placebo + memb): Pb 2.423, Cd 0.561, Co 0.589, Cr 0.810, Ni 0.515.
- Cell 2A (zeolite + memb): Pb 1.270, Cd 0.35, Co 0.377, Cr 0.582, Ni 0.405.
- Cell 3A (placebo + memb): Pb 1.308, Cd 0.816, Co 0.569, Cr 0.653, Ni 0.546.
- Cell 4A (zeolite + memb): Pb 1.137, Cd 0.465, Co 0.416, Cr 0.526, Ni 0.454.
- Cell 5A (placebo + memb): Pb 1.830, Cd 0.517, Co 0.523, Cr 0.666, Ni 0.548.
- Cell 6A (zeolite + memb): Pb 1.039, Cd 0.477, Co 0.381, Cr 0.541, Ni 0.473.
- Cell 1B (zeolite + memb): Pb 0.906, Cd 0.462, Co 0.329, Cr 0.498, Ni 0.453.
- Cell 2B (placebo + memb): Pb 1.078, Cd 0.466, Co 0.570, Cr 0.626, Ni 0.535.
- Cell 3B (zeolite + memb): Pb 0.908, Cd 0.333, Co 0.384, Cr 0.534, Ni 0.497.
- Cell 4B (placebo + memb): Pb 1.242, Cd 0.605, Co 0.547, Cr 0.777, Ni 0.576.
- Cell 5B (zeolite + memb): Pb 0.896, Cd 0.375, Co 0.411, Cr 0.532, Ni 0.441.
- Cell 6B (placebo + memb): Pb 1.374, Cd 0.613, Co 0.564, Cr 0.812, Ni 0.579.
- Experiment 2 ANOVA (Section 3.4, p. 9): Cr F(11) = 26.88, p < 0.001 (significant); Co F(11) = 127.4, p < 0.001 (significant); Ni F(11) = 35.59, p < 0.001 (significant); Cd F(11) = 11.03, p = 0.08 (trend, not significant at α = 0.05); Pb F(11) = 4.6, p = 0.58 (not significant).
- Experiment 1 statistics (Section 3.3.2, p. 8): one-way ANOVA between zeolite-treated and placebo formulations: Cr p = 0.026, Co p < 0.001, Cd p = 0.004 (all significant); Pb p = 0.147 and Ni p = 0.58 (neither significant). Coefficients of determination (R²) exceeded 0.9 for Cr, Co, and Cd. Tukey HSD post-hoc confirmed consistency across replicates.
- Adsorption attribution (Section 4 Discussion, p. 10): “approximately 23% of the observed adsorption is specifically attributable to the presence of zeolite” relative to placebo. The remainder reflects matrix retention by the cream base itself.
- Background metal levels in the placebo cream from herbal-derived ingredients (Section 4 Discussion, p. 10): supplier specifications report < 10 ppm for both Pb and Ni in the herbal raw materials used. These background levels affected absolute residuals but not the relative zeolite-vs-placebo comparison.
- Selectivity ranking (Section 4 Discussion, p. 10): zeolite preferentially binds smaller-hydrated-radius, higher-charge-density cations: Cr³⁺ > Co²⁺ > Cd²⁺ ≫ Ni²⁺ ≈ Pb²⁺. Pb’s weaker binding attributed to lower hydration energy and formation of weakly soluble Pb complexes at the acidic pH used; Ni’s weaker binding attributed to competing affinity for organic functional groups in the cream base.
Methods (brief)
A prototype zeolite-containing cream (3% zeolite w/w, 17.3% lipophilic phase, herbal-derived plant-wax and Rosmarinus officinalis leaf-extract components per Supplementary Table S1) and its matched placebo were compared in vitro using vertical Franz diffusion cells. Donor compartment held cream + spiked acidic metal solution (Cd + Co + Cr + Pb + Ni at 0.04 ppm each). Two synthetic membranes were tested — Strat-M® and medical-grade silicone, 15 cm × 20 cm × 0.025 cm (Bioplexus, Boston, PA, USA; Strat-M® supplied by Merck Millipore, Burlington, MA, USA). Reagents: PBS buffer salts (Na₂HPO₄, KH₂PO₄, NaCl, H₃PO₄) from Supelco (Bellefonte, PA, USA). Experiment 1 (adsorption screening) used a diluted acidic metal solution in the receptor compartment to mimic interstitial-skin-fluid acidity for preliminary zeolite-vs-placebo screening; residuals were measured after 24 h of incubation (Section 3.3, p. 7). Experiment 2 (physiological permeation) used PBS pH 7.4 in the receptor compartment, 12 Franz cells (6 zeolite, 6 placebo, all Strat-M®), 20 mg dose, 0.4 mL acidic metal solution applied to the donor, 12 h duration (Section 2.2.2, p. 3; Section 3.4, p. 9), 32 ± 1 °C, continuous magnetic stirring; no receptor sampling — the analysis focused on residuals in the donor + membrane compartments. Quantification by inductively coupled plasma-optical emission spectroscopy (ICP-OES) after ultra-pure nitric-acid digestion (2-5% v/v), 0.22 µm filtration, and dilution-corrected calibration against certified multi-element standards. Statistical analysis: one-way ANOVA with Tukey HSD post-hoc; coefficient of determination R² for response-surface model fit; desirability function for global optimization. Software: MINITAB® 19.2020.1 (Section 2.4, p. 3-4). aQbD workflow: Analytical Target Profile (ATP), Critical Analytical Attributes (CAA), Critical Method Variables (CMV), Ishikawa cause-and-effect mapping (Figure 1, p. 4), 2 × 2 × 2 full factorial DoE, central composite design (12 runs), response-surface analysis, and desirability-function optimization. Stated limitations: in vitro only (Franz cells, synthetic membranes — not human skin); no receptor sampling in Experiment 2 (cannot quantify what crossed the membrane, only what was retained donor-side); single prototype formulation (n = 1 product, not a sample frame of marketed creams); 0.04 ppm donor spike is a single-concentration screen, not a dose-response curve; only five metals tested; no inorganic-vs-organic speciation of Cr (Cr-VI not measured separately from total Cr); 12 h endpoint only (no time-course beyond that); no in vivo validation; placebo cream itself contained trace metal content from herbal-derived raw materials (< 10 ppm Pb and Ni per supplier specifications), inflating absolute residuals.
Implications
- Certification (HMTc): This paper is in vitro mitigation / efficacy evidence for an active cosmetic ingredient (zeolite) tested in a prototype anti-pollution skin care cream, not contamination occurrence on a marketed product. It does not contribute to occurrence pools for any HMTc product-category row. Its relevance to HMTc is on the mitigation side: it offers mechanistic and statistical support for zeolite (and similar negatively-charged aluminosilicate adsorbents) as a candidate active for anti-pollution leave-on cosmetic formulations, with selectivity strongest for Cr, Co, and Cd and weakest for Pb and Ni. If a future HMTc cosmetics ratchet program incentivises anti-pollution actives via lower thresholds or premium tiers, the data here are usable as one of several efficacy anchors. The “23% of adsorption attributable to zeolite” figure also flags that herbal-derived cosmetic matrices carry meaningful intrinsic metal background even before any active is added — a relevant input for raw-material sourcing standards on cosmetic ingredient lines.
- Courses: Useful teaching case for (a) Franz diffusion cell methodology and the choice between Strat-M® and silicone synthetic membranes, (b) the QbD / aQbD framework (ATP → CAA → CMV → DoE → response surface → desirability) applied to a topical-product efficacy question, (c) the distinction between mitigation-efficacy studies and contamination-occurrence studies, (d) the bias risk introduced when a placebo contains intrinsic metal background from herbal raw materials, and (e) Hofmeister/hydration-energy reasoning for why zeolites preferentially bind Cr³⁺ and Co²⁺ over Pb²⁺ and Ni²⁺.
- App: Out of scope. This is in vitro adsorption-method data on a prototype, not a measured contamination level on a consumer product. The app does not consume Franz-cell residuals.
- Microbiome (if applicable): Not directly addressed. Zeolite’s effect on skin microbiome is mentioned only by implication (the formulation forms a “continuous film” and “transient metal scavenger” layer, Section 2.1, p. 3; Section 4, p. 10) but no microbiome endpoints are measured.
Wiki pages this source may touch
Verification notes
- 2026-05-17 fresh ingest (Claude Opus 4.7, autonomous v2.0 manual-fetch skill, daemon tick): NEW path. Three identity checks against
wiki/sources/returned no hits: DOI10.3390/ma19040685not present; raw_handleMFK_in-vitro-analysis-of-heavy-metal-adsorption-by-zeonot present; cite-key stemnencioni2026not present. PDF SHA-2560ca37d97aaf0516ea45c98b9eaa9f23cf84479c5bbf41da07fe92325fc1544d3. - This paper is mitigation / efficacy evidence, not contamination occurrence. The metals quantified (Cd, Co, Cr, Pb, Ni) come from a deliberately spiked 0.04 ppm donor solution applied to a Franz diffusion cell, plus background trace metals (< 10 ppm Pb, Ni per supplier specs) in the herbal-derived placebo cream matrix. No consumer product was sampled. The “residual concentrations” in Tables 3 and 4 are donor-side and membrane-retained residuals, not contamination levels of a marketed item.
metals: [Cd, Co, Cr, Pb, Ni]uses total-Cr (not Cr-VI) per conventions Part 14 speciation discipline: ICP-OES quantifies total chromium without inorganic-vs-organic Cr-VI speciation (Section 2.3, p. 3). No Hg or As analytes were measured.ingredients: []— the paper’s “active ingredient” is zeolite (a microporous aluminosilicate), which is not in the controlledingredients/taxonomy. The paper does not measure an ingredient-level food or commodity contamination. No ingredient routing destination applies.products: []— the formulation is a development-stage prototype, not a marketed consumer product. No anti-pollution leave-on cosmetic row currently exists in the HMTc product taxonomy. The PDF is filed underraw/Manual Fetch Kimi /Children Personal Care Papers/babycare_01_Exposure_Pathways/for Karen’s exposure-pathway indexing intent (anti-pollution cosmetic actives as a dermal-mitigation literature thread), but the prototype is not children-specific. Per the daemon tick’s hard constraints this session does not create new product pages; if Karen later wants this and similar mitigation/efficacy sources routed to an “anti-pollution-leave-on” or “cosmetic-actives-mitigation-evidence” row, that requires a Step 0 Lock decision.matrices: [cosmetic-personal-care, in-vitro-method]—cosmetic-personal-careis the established bare-string matrix for cosmetic-formulation sources (precedent set bybashir2025-heavy-metals-cosmetics-kashmir);in-vitro-methodflags that the measured signal is Franz-cell adsorption / retention residuals from a spiked solution rather than direct sampling of a finished consumer product. The latter term is added here as a proposed new bare-string vocabulary entry parallel toexposure-modelingused elsewhere, awaiting formal vocabulary review.jurisdictions: []— the paper has no jurisdictional framing. Author affiliations span Switzerland (IBSA Institut Biochimique SA, Pambio Noranco) and Italy (University of Udine; Biofarma Group, Mereto di Tomba), but no national survey scope or regulatory-comparison framework anchors the work to a specific jurisdiction.evidence_tier: A— peer-reviewed in Materials (MDPI; Open Access CC BY); n = 12 Franz cells on a single prototype formulation, which is a methodological constraint rather than a quality issue. Statistical methodology (factorial DoE, ANOVA + Tukey HSD, desirability optimisation) is fully documented. Downstream synthesis treating this as one mitigation-evidence anchor among several is appropriate; treating it as primary efficacy proof for a regulatory or certification claim would over-extend a single-prototype, in-vitro, single-spike-concentration design.- Brand firewall (Part 12, strict reading locked 2026-05-17): No consumer cosmetic brand is named in the paper. Methods-section scientific-reproducibility vendor names are retained per Exception 2: Strat-M® membrane (Merck Millipore, Burlington, MA, USA), medical-grade silicone membrane (Bioplexus, Boston, PA, USA), PBS reagent salts (Supelco, Bellefonte, PA, USA), MINITAB® statistical software (19.2020.1). The cream formulation itself is anonymised as “zeolite-based cream” / “placebo cream” throughout the paper; no brand attribution to a contamination value exists. The author-affiliated firms (IBSA Institut Biochimique SA, Biofarma Group Srl) are affiliations on the byline, not brand-level cosmetic identifiers attached to a measured value.
- Wiki/HMTc firewall (Part 2): the paper’s Conclusion does not propose HMTc-style thresholds; it argues for zeolite as a “multifunctional active ingredient for anti-pollution cosmetic formulations” but stops at efficacy, not at policy or certification limits. The Implications section above interprets the paper as one of several possible mitigation anchors without proposing certification values.
- Paper-internal labelling issue (Tables 2 and 3): Table 2 prints row labels in inverted order — the row showing values 0/3 is labelled “Dosage” but body text (Section 3.1, p. 5) defines “Dosage regimen … finite dosing conditions” as the 0/3 variable, while the row showing values 10/20 is labelled “Dosage regimen” but body text defines “Dosage … 10 mg and 20 mg” as the mg variable. Table 3 column headers carry the same inversion. The Key numbers section above presents the data using the body-text definitions (Dosage = mg; Regimen = 0/3) rather than the column labels printed in Tables 2-3. This is a paper-internal labelling inconsistency, not a numerical error.
- Body-text “30 mg” inconsistency (Section 3.2, p. 7): the optimisation paragraph states “a formulation dosage of 30 mg” as the optimal level, but the factorial domain (Table 2) is bounded at 10-20 mg; the same paragraph two sentences later confirms “Strat-M® membrane, a finite dosage regimen, and a 20 mg dosage.” The 30 mg phrase is a typographic inconsistency with the experimental domain; the 20 mg value matches the run-level data in Table 3 (Run 3 and Run 11 are Strat-M® + 20 mg + finite regimen, the lowest-permeation runs). Treated as 20 mg in the Key numbers.
- Table 4 cell-label sanity check: zeolite vs placebo cell allocation in Experiment 2 mixes A-series and B-series labels (1A placebo, 2A zeolite, 3A placebo, 4A zeolite, 5A placebo, 6A zeolite; 1B zeolite, 2B placebo, 3B zeolite, 4B placebo, 5B zeolite, 6B placebo). The count is 6 placebo + 6 zeolite as stated in the body text (Section 3.4, p. 9). The printed Table 4 shows each cell’s measured concentration only in the column corresponding to its treatment (e.g., placebo cells have values in the “Placebo” columns only); the empty columns are not zero values but the inapplicable treatment for that cell. Reproduced as printed.
- Cell 2A Co value (Table 4): the printed cell at the intersection of “Cell 2A zeol + memb” and the “Co (ppm) Zeolite” column reads 0.377 with a grey-shaded background; the row also has values in the placebo Co column for the A-series, but Co’s A-series placebo values appear to align with the rows below. Cross-checked against the body-text statistical summary (F(11) = 127.4, p < 0.001 for Co between zeolite vs placebo) which is consistent with the zeolite Co residuals (0.329-0.416 ppm) being lower than placebo Co residuals (0.523-0.589 ppm) across the cells. Reproduced verbatim above.
- Donor-spike-vs-residual mass-balance note: donor concentration was 0.04 ppm per metal but Table 3 residuals are 0.17-0.45 ppm and Table 4 residuals are 0.30-2.42 ppm. The order-of-magnitude gap is explained by (a) the residual being measured in the donor + membrane after evaporation / sorption concentrates the residual solid + sorbed phase, (b) trace metal content in the cream matrix from herbal-derived raw materials (Section 4 p. 10: < 10 ppm Pb and Ni per supplier spec; Section 3.3.1 p. 8 notes Pb’s “elevated concentrations relative to the initial diluted acidic solution likely reflect the unexpectedly high Pb content already present in the placebo”). The relevant signal is zeolite-vs-placebo difference, not absolute residual concentration; this is correctly framed in the paper.
- No Supplementary Table S1 access in this ingest — composition details for the prototype cream are referenced (Section 2.1, p. 3) but not extracted from the supplementary file at https://www.mdpi.com/article/10.3390/ma19040685/s1. If the future synthesis pass needs the full ingredient list, fetch S1 separately.
- Conflicts of interest (p. 11): author Alessandro Nencioni was employed by IBSA Institut Biochimique SA; author Emanuele Nencioni was employed by Biofarma Group Srl. The remaining author (Michela Bulfoni) declares no conflict. Industry employment of two of three authors is relevant context for evidence weighting in a future synthesis pass; the work itself is funded as “no external funding.”
- Audit subagent (2026-05-17, fresh-context general-purpose, autonomous v2.0 manual-fetch skill): REVISE verdict; 2 findings applied, 0 rejected as false positives. Checks 2-5 all clean (✅).
- CHECK 1 ⚠️ Experiment-1 vs Experiment-2 incubation-time conflation: wiki initially stated “12 h duration” for both experiments in sample_population frontmatter and in Key numbers. Verified against PDF Section 3.3, p. 7 (“the residual metal content was measured after 24 h of incubation” for Experiment 1) and Section 2.2.2, p. 3 / Section 3.4, p. 9 (“12 h” for Experiment 2). Corrected: sample_population now reads “Experiment 1 incubation 24 h; Experiment 2 permeation 12 h”; Key numbers now distinguishes the two endpoints with section citations; Methods (brief) now explicitly states the 24 h endpoint for Experiment 1 and 12 h for Experiment 2.
- CHECK 1 ⚠️ Cell 2A Cd typographic padding: wiki had “Cd 0.350”; PDF Table 4 prints “0.35” (no trailing zero). Verified — same value, only typographic fidelity. Corrected to “0.35” to match the printed cell exactly.
Page history
The five most recent substantive edits to this page. The full version history lives in git; when DOI minting comes online (see schema docs), each entry below will also link to a version-pinned DataCite DOI.