Heavy Metal Index
Ask the IndexAboutIndex Pro

Release of TiO2 and ZnO nanoparticles from sunscreens into natural waters: detection and discrimination from natural particles using SP ICP-ToF-MS

Cite this pageVersion history

Overview

This open-access Environmental Science: Nano paper from the Wilkinson group (Université de Montréal) characterizes titanium dioxide and zinc oxide nanoparticles released from two commercial mineral sunscreens (designated SS1 and SS2) into three aqueous matrices (ultrapure water, a soft natural water from Lac-Croche, and a hard natural water from the St-Lawrence River) after two short-term exposures (~15 min and ~60 min), using inductively coupled plasma time-of-flight mass spectrometry (ICP-ToF-MS) operated in single-particle mode. The work is primarily methodological: the authors evaluate which features of single-particle SP-ICP-ToF-MS signatures (elemental ratios, isotopic ratios, particle purity, polydispersity, Spearman rank correlation analysis) best distinguish engineered nanoparticles from naturally occurring nanoparticles in environmentally realistic mixtures. Substantive product-chemistry findings include confirmation of manufacturer-declared ZnO and TiO2 loadings by acid digestion, observation that nearly all ZnO released into natural waters dissolves to ionic Zn²⁺ (which adsorbs onto background colloids and aluminosilicate/iron-oxide natural nanoparticles) within 60 minutes while TiO2 remains predominantly in nanoparticulate form, and quantitation of dissolved-Zn concentrations in the natural waters that reach the mg/L range after one hour of contact with the sunscreen surrogate.

Key numbers

Page citations refer to the published PDF (Environ. Sci.: Nano, 2025, 12, 4994-5007). All values are reported as the authors reported them; no unit conversions applied.

Sunscreen composition (manufacturer specifications and acid-digestion verification)

  • SS1: 24.08% (w/w) ZnO per manufacturer’s specification; acid-digestion measurement 28 ± 2% (m/m) ZnO. No TiO2 declared or detected (p. 4998).
  • SS2: 9% (w/w) ZnO and 7% (w/w) TiO2 per manufacturer’s specification; acid-digestion measurements (9.8 ± 0.4)% (m/m) ZnO and (6.3 ± 0.2)% (m/m) TiO2 (p. 4998).
  • Acid digestion: ~300 mg sunscreen in 4 mL 2:1 H₂SO₄:HNO₃ mixture, microwave-digested in a 20 mL Teflon tube on an Anton Paar 20SVT50 with a ramped temperature profile (150 °C / 220 °C / 250 °C); triplicate digestions with standard deviations reported (p. 4996).

Sunscreen application geometry and water volumes

  • Mass applied to a 50 mL polypropylene tube via cotton-tipped applicator: SS1 28.0-32.0 mg; SS2 600.0-650.0 mg (p. 4995). Coverage area: SS1 ~19 cm²; SS2 ~31 cm². The sunscreen layer thickness was less than 1 mm.
  • Aqueous volume per exposure: ~50 g of ultrapure / soft / hard water added to each tube (p. 4995).
  • Three independent exposure dates per sunscreen, two exposure times per date (~10-20 min and 65-72 min, referred to in text as ~15 min and ~60 min); supernatant collected, diluted 20-25× with Milli-Q, vortexed for 4 min immediately prior to analysis (p. 4995-4996).

Natural-water composition and background-particle counts

  • Soft water (Lac-Croche): pH ~6.0, DOC ~3.4 mg C/L. Hard water (St-Lawrence River): pH ~7.5, DOC ~2.0 mg C/L. The hard water had greater ionic content than the soft water (Table S1).
  • Background particle number concentrations (without added sunscreen, after the standard dilution): 2.6 × 10⁵ particles/mL in the soft water; 1.8 × 10⁶ particles/mL in the hard water (p. 4997). The hard water yielded ~9× more particles than the soft water at the same dilution factor, attributed to the watershed difference (small forest lake versus large urban river).
  • Major detected colloidal elements in the two natural waters (per Fig. 1 pie charts based on mass proportions): soft water Si 71.0%, Fe 21.3%, Al 6.7%, others 1.0%; hard water Si 47.2%, Fe 35.6%, Al 14.6%, Mn 1.4%, others 1.1%. Si–Fe oxyhydroxides and aluminosilicates dominate both matrices.

ZnO nanoparticle release and dissolution (Table S5, S7, S8; p. 4998-5000)

  • SP-ICP-ToF-MS size detection limit improvement with Chelex-100 cation-exchange resin: ~102-130 nm (without resin) to ~53-75 nm (with resin). Reduction of dissolved-Zn background from ~2.3 mg/L (soft water; Table S8) to below instrumental detection limits enabled detection of small ZnO particles.
  • For SS1 in ultrapure and soft water, ~46-56% of the released Zn-containing particles were single-element Zn particles; in hard water this fraction dropped to ~13%.
  • For SS2, single-element Zn particles represented ~2-9% of detected particles in the natural waters and ~50% in the ultrapure water.
  • Particles containing both Si and Zn were detected at ~5% in hard water, decreasing to ~10% in soft water, ~4% in ultrapure water for SS1. These accounted for ~70% of total detected mass in SS1 (per Fig. 2 SS1 results).
  • For SS2, Ti-containing particles ranged from ~31% to 57% in the natural waters and 41% in ultrapure water.
  • Dissolved Zn after one hour of exposure (in the absence of the Chelex-100 resin, from Table S5; cited p. 5001): SS2 reached 1928 ± 490 µg/L in ultrapure water, 2351 ± 489 µg/L in soft water, and 1325 ± 357 µg/L in hard water. Dissolved Zn from SS2 was 10-20× higher than from SS1.
  • Zn nanoparticle mass release for SS2 at 60 min (Table S8; cited p. 5000): 312 ± 180 ng/g sunscreen in ultrapure water, 162 ± 51 ng/g in soft water, 171 ± 72 ng/g in hard water.

TiO2 nanoparticle release (SS2 only; p. 5001)

  • TiO2 release as a function of exposure time for SS2: 153 ng titanium per gram of sunscreen at 9 min increasing to 865 ng/g at 60 min (cited p. 4998).
  • Ti nanoparticle mass release for SS2 (Table S8; cited p. 5001): 301 ± 161 ng/g sunscreen in ultrapure water; 865 ± 182 ng/g in soft water; 865 ± 463 ng/g in hard water.
  • Unlike Zn, the TiO2 particles were poorly soluble; good mass balances were obtained between the particle-number-derived totals in the natural waters and the values from independent acid digestion of SS2.
  • Ti–Zn heteroaggregates accounted for 5-10% of particle numbers in SS2 exposures across all three matrices.

Particle mass distributions (Fig. 4; p. 5001)

  • Peak particle mass: ~6 fg for the ZnO nanoparticles and ~7 fg for the TiO2 nanoparticles. Mass distributions were similar across exposure times for each formulation.
  • Under the assumption of spherical ZnO with density 5.6 g/cm³, a 6 fg particle corresponds to a particle radius of ~68 nm.
  • Under the assumption of spherical TiO2 with density 4.26 g/cm³, a 7 fg particle corresponds to a particle radius of ~87 nm.

Elemental and isotopic ratios as discriminators (Fig. 5, 6, 7; p. 5002-5003)

  • Si/Zn ratios for SS1 in ultrapure water averaged 19.5 (n=509 particles); in soft water with SS1, 21.3 (n=2616); in hard water with SS1, 24.5 (n=1268). In the soft and hard waters without added SS1, no particles containing both Si and Zn were found (only Si-only particles were observed).
  • Fe/Zn ratios for SS1 in ultrapure water averaged 0.9 (n=2); in soft water without SS1, 7.4 (n=2); in soft water with SS1, 0.2 (n=179); in hard water with SS1, 0.4 (n=625). The ultrapure-water n=2 statistic confirms that Fe–Zn particles in the SS1-exposed natural waters were predominantly formed in the natural water rather than released from the sunscreen.
  • For SS2, the Ti/Al ratio for most detected particles was <2; this was proposed (refs 18, 27 in the source) as a discriminator for engineered TiO2 from natural Ti-bearing particles, but the high natural-Ti background in the local waters limited Ti/Al utility for SS2 in this study.
  • Isotopic ratios ⁶⁶Zn/⁶⁸Zn (SS1 and SS2) and ⁴⁷Ti/⁴⁹Ti (SS2): for the largest particle masses, isotopic ratios matched the natural-abundance values in both the sunscreen-derived particles and the natural-water colloids; the authors conclude that isotopic ratios were not a useful discriminator between ENPs and NNPs for these elements.

Spearman rank correlation analysis (Fig. 8; p. 5004)

  • Sample-sample correlation coefficients (sunscreen-natural-water mixtures versus reference matrices):
  • Soft water with SS1 vs hard water with SS1: r = 0.76 (SS1); r = 0.91 (SS2).
  • Soft water without sunscreen (SW) vs hard water without sunscreen (HW): r = 0.82.
  • Soft water with SS1 (SS1-SW) vs soft water alone (SW): r = 0.72.
  • Hard water with SS1 (SS1-HW) vs hard water alone (HW): r = 0.61.
  • Soft water with SS2 (SS2-SW) vs soft water alone (SW): r = 0.91 (driven by overlap of Zn, Si, Ti single-element particles between SS2 and natural-water particle populations).
  • Hard water with SS2 (SS2-HW) vs hard water alone (HW): r = 0.76.
  • Most-different pair: hard water alone vs sunscreen-in-ultrapure-water for SS1 (r = 0.22); SS2 (r = 0.31). Sunscreen-derived particle profiles were most distinguishable from natural-water-only profiles when sunscreen exposure had not yet altered the natural-water particle population.

Methods

Sunscreens and water matrices. Two unbranded commercial mineral sunscreens were used. SS1 manufacturer specification: 24.08% (w/w) ZnO with no TiO2. SS2 manufacturer specification: 9% (w/w) ZnO + 7% (w/w) TiO2. Both manufacturer ingredient lists are reproduced in Table S2 of the source (paper does not name the brands). Three aqueous matrices: ultrapure water (Milli-Q, 18.2 MΩ·cm at 25 °C, TOC < 2 µg C/L; Millipore), soft water from Lac-Croche (45.989519, −73.564012; pH ~6.0, DOC ~3.4 mg C/L), hard water from the St-Lawrence River (Cartesian 45.454009, −73.564012; pH ~7.5, DOC ~2.0 mg C/L). Water samples were refrigerated at 4 °C ≤9 days before analysis.

Sunscreen application. Each sunscreen was applied with a cotton-tipped applicator (Innovatek, Canada) to the inner walls of a 50 mL polypropylene tube (Greiner Bio-One, Austria) at 28.0-32.0 mg (SS1; coverage ~19 cm²) or 600.0-650.0 mg (SS2; coverage ~31 cm²). ~50 g of ultrapure / soft / hard water was added per tube. Tubes were placed on a circular rotator (Fisher Scientific) at 25 rpm at 21 °C under controlled artificial light, then sampled at ~10-20 min and 65-72 min (referred to as ~15 min and ~60 min). Supernatants were diluted 20-25× with ultrapure water, vortexed, and analyzed within one day. Three independent experiments per sunscreen, with two exposure times per experiment.

Acid digestion of sunscreens (for total Zn and Ti). ~300 mg of sunscreen was digested in 4 mL of a 2:1 v/v H₂SO₄:HNO₃ mixture in a 20 mL Teflon tube on an Anton Paar 20SVT50 microwave digester with a ramped temperature profile (150 °C 5 min hold / 150 °C 15 min hold / 220 °C 10 min / 220 °C 15 min hold / 250 °C 5 min / 250 °C 15 min hold). Triplicate digestions per sunscreen.

Total elemental analysis by ICP-MS. Digested solutions were diluted 2-3 × 10⁴× into 2% v/v 2:1 H₂SO₄:HNO₃ and analyzed on a PerkinElmer NexION 5000 ICP-MS. Multi-element calibration standards (71A Inorganic Ventures; Ti standard from Analytichem; both 2-40 µg/L). Six calibration concentrations from 2.0 to 40.0 µg/L from most elements. Quality-control standards (QC4, Plasma CAL; QC21, PerkinElmer Mississauga) were analyzed after calibration and every 12 samples. ⁴⁷Ti analyzed in oxygen DRC mode; ⁶⁸Zn in He KED mode. ⁴⁵Sc as internal standard. Other natural-water elements (²³Na, ²⁴Mg, ²⁷Al, ²⁸Si, ³¹P, ³⁹K, ⁴³Ca, ⁵⁶Fe) measured on the same NexION 5000 with 2% v/v HNO₃ matrix. Relative standard deviations of replicate measurements generally below 2.0%.

Single-particle ICP-ToF-MS (SP-ICP-ToF-MS). Vitesse SP-ICP-ToF-MS instrument (Nu Instruments, UK) equipped with a micro-flow concentric glass nebulizer (0.4 mL/min), quartz cyclonic spray chamber (Peltier-cooled to 4 °C), and quartz injector (1.5 mm internal diameter) at transport rate 0.027 mL/min. Mass-to-charge range 26-210 amu. Samples analyzed sequentially: ultrapure water first, then soft water, then hard water; carryover monitored and eliminated by Milli-Q rinsing between samples.

Transport efficiency calibration. 100 ng/L suspensions of PEG carboxyl-functionalized ultra-uniform gold nanospheres (AUXU50 and AUXU100, NanoComposix, USA; 49 and 102.4 nm) for the particle size method; suspensions of 200 ng/L citrate-functionalized silver nanospheres (AGCN60, NanoXact, USA; nominal diameter 60 nm) for waste-collection / particle-number transport efficiency. ICP ionic gold standards (Analytichem, Canada) for instrument sensitivity. Average particle transport efficiency 0.4564 µL/s.

Ion-exchange resin pretreatment. A PFA column (12.9 mm × 64.5 mm; Elemental Scientific) packed with sodium-form Chelex-100 (50-100 mesh; Sigma Life Science, Canada) was placed ahead of the ICP-ToF-MS sample introduction. Resin conditioning: 1.5 M HNO₃ for 20 min at 0.5 mL/min; ultrapure water for 20 min at 4 mL/min; 0.1 M NaOH for 20 min at 0.5 mL/min; ultrapure water for 20 min at 4 mL/min. Resin rinsed for 4 min between samples; replaced after 10 samples; rinsed with 1.5 M HNO₃ for 20 min then with ultrapure water for 20 min before sealing for storage. The resin reduced dissolved Zn from ~2.3 mg/L (soft water) to below instrumental detection limits without significantly reducing particle numbers, improving the ZnO nanoparticle size detection limit from 102-130 nm to 53-75 nm.

Data treatment. Nu Quant Vitesse software for acquisition; TOFVision (Wilkinson research group; github.com/Houssame-EA/TOFVision) for figure generation and data treatment. Detection frequency thresholds for particle counting: >0.2% of total particles for the hard water; >1% for ultrapure / soft waters. ⁴⁷Ti chosen over ⁴⁹Ti because of fewer interferences and higher abundance; ⁴⁸Ti avoided due to ⁴⁸Ca interference. Spearman rank correlation analysis applied as a non-parametric, ranking-based alternative to hierarchical cluster analysis, with sample-count weighting to adjust for statistical reliability differences across samples.

Method limitations stated by the authors. SP-ICP-ToF-MS as deployed measures NP but also a substantial fraction of natural colloidal particles. Even with the Chelex-100 resin and the multi-element / isotopic / particle-purity discriminators tested, several similarities remained between engineered and natural nanoparticles, especially in chemically heterogeneous and polydisperse waters. The Ti/Al ratio discriminator (≤2 for engineered, higher for natural) was useful for SS1 but degraded for SS2 because of the high natural Ti background in the local waters. Cluster-analysis approaches (machine learning, hierarchical clustering) were tested in the supplementary information but were highly sensitive to input parameters (number of clusters, dendrogram height), leading to inconsistent identification of important particle groups in these databases; the Spearman correlation approach was preferred for direct pairwise comparison of sample-level elemental profiles.

Implications

Certification (HMTc). The paper’s primary product-chemistry contribution is acid-digestion verification of two commercial mineral-sunscreen formulations: SS1 was measured at 28 ± 2% (m/m) ZnO against the manufacturer’s declared 24.08% (w/w); SS2 was measured at 9.8 ± 0.4% ZnO and 6.3 ± 0.2% TiO2 against the manufacturer’s declared 9% and 7% (w/w). These are total-Zn and total-Ti measurements of the active mineral UV-filter content (ZnO and TiO2 are intentional formulation actives in mineral sunscreens, not contaminants). The paper does not report heavy-metal contamination data (Pb, Cd, As, Hg, Ni, Al, Cr, Sn) in the sunscreen products themselves; the trace metals it measures (Al, Fe, Mn, Si, Ce, La, Cr) are reported as natural-water background or as alumina-coating discriminators for the engineered TiO2, not as contamination of the sunscreen. The paper therefore does not contribute directly to HMTc threshold-setting for the baby-sunscreen-mineral or sun-suntan-products rows. It contributes occurrence data for the analytical chemistry of mineral-sunscreen actives in environmentally realistic matrices, which is upstream of any HMTc threshold work on those rows.

Courses. Useful for any course covering single-particle ICP-MS / ICP-ToF-MS analytical chemistry, nanoparticle characterization in complex matrices, or environmental fate of mineral sunscreens. The methodology section is a worked example of how Chelex-100 cation-exchange pretreatment lowers dissolved-metal background sufficiently to expand the accessible particle-size detection range. The Spearman correlation approach as an alternative to hierarchical clustering for sample-level discrimination is also instructive.

App. Limited direct relevance. The paper does not characterize finished baby-sunscreen contamination by HMTc analytes and so does not feed the consumer-app ingredient-level or product-level contamination score for sunscreens.

Microbiome. Not addressed. The paper studies environmental fate in lake and river water; biological compartments (sediment, organisms, microbial communities) are out of scope.

Update history

No substantive edit history is available in this build. The full commit record is available in git.

Referenced by