The Application of Anisotropically Collapsing Gels, Deep Learning, and Optical Microscopy for Chemical Characterization of Nanoparticles and Nanoplastics

Abstract

The surface chemistry of nanomaterials, particularly the density of functional groups, governs their behavior in applications such as bioanalysis, bioimaging, and environmental impact studies. Here, we report a precise method to quantify carboxyl groups per nanoparticle by combining anisotropically collapsing agarose gels for nanoparticle immobilization with fluorescence microscopy and acid-base titration. We applied this approach to photon-upconversion nanoparticles (UCNPs) coated with poly(acrylic acid) (PAA) and fluorescence-labeled polystyrene nanoparticles (PNs), which serve as models for bioimaging and environmental pollutants, respectively. UCNPs exhibited 152 ± 14 thousand carboxyl groups per particle (∼11 groups/nm²), while PNs were characterized with 38 ± 3.6 thousand groups (∼1.7 groups/nm²). The limit of detection was 6.4 and 1.9 thousand carboxyl groups per nanoparticle, and the limit of quantification was determined at 21 and 6.2 thousand carboxyl groups per nanoparticle for UCNP-PAAs and PNs, respectively. High intrinsic luminescence enabled direct imaging of UCNPs, while PNs required fluorescence staining with Nile Red to overcome low signal-to-noise ratios. The study also discussed the critical influence of nanoparticle concentration and titration conditions on the assay performance. This method advances the precise characterization of surface chemistry, offering insights into nanoparticle structure that extend beyond the resolution of electron microscopy. Our findings establish a robust platform for investigating the interplay of surface chemistry with nanoparticle function and fate in technological and environmental contexts, with broad applicability across nanomaterials.

1

Scheme of studied nanoparticles. (a) UCNP-PAAs provide intrinsic luminescence (blue coloration, excitation 976 nm, emission 802 nm), (b) Intrinsically nonluminescent PNs are soaked with Nile red to become fluorescent (pink coloration, excitation 520 nm, emission 560 nm). In panels (a) and (b), the ribbons represent PAA chains on the nanoparticle surfaces.

2

Nanoparticle counting assay based on anisotropic agarose gel collapse and acid–base titration. (a) Nanoparticles in aqueous dispersion are mixed with melted agarose and cast into a thin microlayer. (b) Upon air drying, the agarose collapses anisotropically into a submicrometer layer, imaged by optical microscopy. (c) Artificial intelligence algorithms are employed to count the nanoparticles. (d) Photograph of the agarose casting setup between two glass slides, secured with metal clips (spacer not visible). (e) Acid–base titration of nanoparticle samples using phenolphthalein as an indicator: before equivalence (left) and at the equivalence point (right).

3

Characterization of photon-upconversion nanoparticles (UCNP). (a) Size distribution of nanoparticles with different surface modifications. (b) UCNP-PAAs emit visible blue light under a 976 nm laser beam. (c) Emission spectrum of UCNP-PAAs upon 976 nm excitation. (d) Photon-upconversion micrograph of UCNP-PAAs in collapsed agarose gel (976 nm excitation, 800 ± 25 nm emission, 3000 ms exposure). (e) Annotated version of panel (d): red dots indicate nanoparticle localizations identified by U-net; green circles represent measured spot intensities; blue circles mark spots excluded from intensity analysis; magenta circles indicate background signal placed sufficiently far from nanoparticles to avoid interference. (f) Spot intensity histogram fitted to a linear combination of three Gaussian curves, corresponding to monomers, dimers, and trimers of UCNP-PAAs.

4

Characterization of polystyrene nanoparticles (PNs). (a) Visual appearance under white light: Nile-PNs (left) and PNs (right). (b) Size distributions of Nile-PNs (blue) and PNs (orange) measured by dynamic light scattering. (c) Fluorescence properties: Nile-PNs emit red fluorescence under 520 nm excitation, whereas PNs are nonfluorescent. (d) Spectroscopic data (nanoparticle concentration 1 nmol L^–1): excitation spectra at 650 nm emission for Nile-PNs (spectrum 1) and PNs (spectrum 2); fluorescence spectra at 520 nm excitation for Nile-PNs (spectrum 3) and PNs (spectrum 4). (e) Fluorescence micrograph of Nile-PNs in collapsed agarose gel (520-nm excitation, 650 ± 25 nm emission, 3000 ms exposure). (f) Annotated image of panel (e): red dots indicate nanoparticle localization by U-net; green circles mark measured spot intensities; blue circles are excluded from intensity measurement; magenta circles indicate the reading of a local background. (g) Spot intensity histogram fitted with a Gaussian curve. (h) Agarose gel electrophoresis of PNs (left) and Nile-PNs (right) in 0.2% (w/w) agarose (100 V, 60 min). The gel was imaged using light transmission mode to detect light scattering by both PNs and Nile-PNs. Imaging details are provided in Note S2 and Scheme S2 in the Supporting Information).