Organic nanoparticles with tunable size and rigidity by hyperbranching and cross-linking using microemulsion ATRP

The oNPs based on hyperbranching and cross-linking were prepared using a ME ATRP method (SI Appendix, Fig. S1). The aqueous phase contained hydrophilic Cu catalyst complex, [Br−Cu^II(TPMA)]⁺ [TPMA = tris(2-pyridylmethyl)amine], together with surfactants in water. Brij 98, i.e., polyoxyethylene (20) oleyl ether, served as the primary surfactant at the oil–water interface. Its nonionic nature ensured the formation of stable ME droplets with sub-100 nm hydrodynamic diameters (65). Additionally, SDS, an anionic surfactant, was selected as a cosurfactant (SDS: Brij 98 = 10: 90% mol/mol) to enhance electrostatic interactions for a better-stabilized ME. The presence of dodecyl sulfate anion (DS⁻) also facilitated the transport of catalyst through interfaces via ion pair formation with Cu²⁺, thereby improving polymerization efficiency (68, 69). Sodium bromide (NaBr) was added to mitigate the dissociation of the Cu catalyst (70). The oil phase, composed of the premixed inimer BiBEM (SI Appendix, Fig. S2) and cross-linker EGDMA, was introduced dropwise into the aqueous phase without any organic solvent. This oil–water mixture was homogenized using a vortex mixer to form ME, setting the stage for copolymerization within ME compartments upon feeding ascorbic acid (AsAc) as the reducing agent.

The reaction reached high comonomer conversion after 2 h, as evidenced by the absence of vinyl proton peaks in proton NMR (¹H NMR) spectra. The hydrodynamic sizes of ME droplets were analyzed using dynamic light scattering (DLS) at three distinct stages of the same reaction (SI Appendix, Fig. S3): immediately after vortex mixing (t = 0 min), at t = 3 min, and just before product precipitation (t = 2 h). Across these stages, the number distribution mode revealed consistent monomodal peaks, though with the progressively reduced droplet size. The volume distribution accentuates the presence of larger nanoparticles (SI Appendix, Eqs. S1–S3) and the ME showed bimodal size distributions immediately after vortex mixing, despite having peak positions similar to those at the subsequent stages. This observation suggested that the initial mixing process resulted in a diverse range of droplet sizes, which may converge toward a more uniform size distribution as the reaction progressed. The purification of the final oNP products involved precipitation in methanol, dissolution in tetrahydrofuran (THF), centrifugation, and repeated dialysis. This process yielded oNPs that could be either maintained in THF as a stable solution or dried under a vacuum to form solids for further studies and subsequent SI-ATRP reactions. The stability of oNPs in acidic conditions was evaluated through a 24-h exposure to trifluoroacetic acid (TFA, 10 mL) with oNPs (B60E40, 41.6 mg) dissolved in dichloromethane (DCM, anhydrous, 50 mL). The impact of TFA on the size of the oNPs over this period was negligible (SI Appendix, Fig. S4), leading us to conclude that the oNPs are stable under acidic conditions.

The preparation of oNPs with varying compositional ratios of BiBEM to EGDMA (from 90:10 to 60:40 mol/mol) was carried out with either Brij 98 as the sole surfactant or Brij 98/SDS as cosurfactants (Table 1). These systems were denoted as B_xE_100-x and SB_xE_100-x, respectively, where x indicated the BiBEM molar ratio. Following purification, the oNP samples were characterized by size exclusion chromatography (SEC) with THF eluent. The apparent molecular weights (M_{n, RI}) and their distributions (i.e., M_w/M_n) were determined using the refractive index (RI) detector and PMMA standards. The absolute molecular weights (M_{n, MALS}) were measured by the multiangle light scattering (MALS) detector and calibrated using sample concentrations. Additionally, the hydrodynamic diameters of the ME droplets (after 2 h, prior to precipitation) and of the oNPs after redispersion in THF were analyzed by DLS using the volume-averaged distribution.

Given the absence of solvents other than the comonomers within the reaction compartments, the original size of oNP was presumed to closely match the hydrodynamic diameter of the ME before product precipitation. However, the increasing cross-linking density contributed to significant size alterations upon redispersion in THF (Fig. 2A). For oNP with a lower cross-linker to inimer ratio, the redispersion led to the expansion of their hyperbranched structures. This resulted in a “swollen” effect, evidenced by the increased hydrodynamic diameters. Conversely, oNPs with a higher proportion of EGDMA demonstrated a “shrunk” effect upon redispersion. This diminished penetration by solvent molecules led to a decreased hydrodynamic size. Additionally, the size reduction could also be attributed to the loss of unreacted comonomers, which could have been incorporated to a cross-linked and vitrified matrix at high conversion. Therefore, despite the general trend of size reduction in both ME and redispersed states, oNPs with increased EGDMA content showed a gradual “size change (% v/v)” transition from “swollen” to “shrunk” effect (Fig. 2B). This dynamic interplay between the oNP compositions and its dissolution behavior highlighted the importance of macromolecular structure design of oNP. Hydrodynamic diameters of redispersed oNP products varied from 20.9 to 36.4 nm with Brij 98 as the sole surfactant and from 17.7 to 26.5 nm with inclusion of SDS as a cosurfactant (SI Appendix, Fig. S5). The smaller size of the latter were attributed to the improved electrostatic stability of ME provided by SDS. Using the same formulations, anionic surfactants produce smaller droplet sizes than nonionic analogs (71). The negatively charged ME droplets (due to anionic SDS) prevented coalescence; their electrostatic repulsions were stronger than the steric repulsion from the nonionic surfactant Brij 98. Notably, in both the “Brij 98 only” and “cosurfactant” systems, the “size changes” varied in a similar trend from approximately +35 to −20% v/v (Fig. 2C). This demonstrated that the compositional ratios of oNPs played a crucial role in regulating the internal architecture thereby influencing their dissolution behavior.

Since each oNP could be considered as a single macromolecule, the size distributions of oNPs were also characterized as the molecular weight distributions, as indicated by the THF SEC RI detector (Fig. 2D). The oNPs synthesized exclusively with Brij 98 exhibited a higher dispersity (M_w/M_n = 1.58) starting from B90E10 but demonstrated improved size uniformity with increasing amounts of cross-linker (Fig. 2E). A more uniform distribution with a lower dispersity (M_w/M_n = 1.23) was observed for B60E40, attributed to the overall denser architectures regulated by EGDMA. Moreover, with the introduction of a cosurfactant system, oNPs started with low dispersity at lower EGDMA compositions (M_w/M_n = 1.23 for SB90E10). The uniformity further improved as the EGDMA content increased, achieving a dispersity of 1.07 for SB60E40 (Fig. 2F). The enhancement in oNP uniformity was achievable either by elevating the cross-linker concentration or by adding SDS as a cosurfactant. The latter's effectiveness was twofold: First, the anionic nature of SDS promoted electrostatic stabilization of the ME, leading to more uniform size distributions. Second, SDS facilitated the transport of the copper catalyst across oil–water interfaces, enabling more controlled copolymerization and, consequently, more uniform chain growth.

The morphology of oNPs was analyzed using atomic force microscopy (AFM) for three specific BiBEM/EGDMA compositions: 90:10, 75:25, and 60:40. Compared with transmission electron microscopy, AFM provides height information and three-dimensional cross-sectional views, illustrating the morphology of oNPs more effectively. The oNP sample solutions (0.05 mg/mL in THF) were drop-cast onto a silicon wafer and placed in a vacuum oven at 45 °C for 24 h to facilitate drying and the removal of solvent residue. Samples synthesized with Brij 98 only exhibited larger size contrasts and were used for bulk film imaging (Fig. 3 A–C), while the more uniform oNPs, synthesized using cosurfactants, were highly diluted (0.05 mg/mL) for dispersed sample imaging (Fig. 3 D–F). Images of bulk films showed spherical oNPs with some particles partially merging due to the entanglement of interfacial hyperbranched structures. With increasing EGDMA compositions, higher numbers of oNPs were observed within the same area (800 × 800 nm), corroborating the size reduction observed in DLS. Furthermore, a change in uniformity was noted, with dispersed oNPs evenly distributed and showing minimal agglomeration. Despite a relatively low dispersity (M_w/M_n = 1.23) measured for SB90E10, a broad size distribution was evident, with both large and small oNPs visible. Enhanced size uniformity was observed with increased cross-linker content, as seen in the SB60E40 samples, which displayed sharp cone-like structures in 3D cross-sectional height images. The rigidity of oNPs was initially estimated by analyzing the height-to-width ratio, with the expected 1:1 ratio for “inorganic” NPs. For each uniformly dispersed SB_xE_100-x sample, 12 individual oNPs were selected for analysis (SI Appendix, Figs. S6–S12). After excluding the maximal and minimal values, the average height-to-width ratio was calculated, starting from 0.22 for SB90E10 with low EGDMA composition and increasing to a plateau around 0.45 for SB75E25 (SI Appendix, Fig. S13 and Table S2). This rough calculation suggested that a higher cross-linking density led to an increased oNP rigidity.

The density of solid oNPs was measured using a helium pycnometer. This method involves displacing the skeletal volume of the oNPs with helium gas, with the density calculated based on the volume-pressure relationship dictated by Boyle’s Law. The solid densities recorded were (SI Appendix, Table S3) 1.4452 ± 0.0061 g/cm³ for B90E10, 1.4249 ± 0.0073 g/cm³ for B75E25, and 1.4368 ± 0.0027 g/cm³ for B60E40. These values, closely aligned with each other, are all significantly higher than densities of the respective comonomers (1.303 g/cm³ for BiBEM and 1.051 g/cm³ for EGDMA), indicating the dense architecture of the oNPs. The highest density, observed in B90E10, is attributed to a larger content of BiBEM. Meanwhile, B60E40 had the second-highest density due to a higher cross-linking content. This dense yet relatively low-density structure of the oNPs makes them advantageous for the fabrication of lightweight composite materials in comparison to their inorganic counterparts.

To investigate the intraparticle elastic properties, Brillouin light spectroscopy (BLS) was employed to further elucidate the adjustable internal rigidity of oNPs. This technique examines the inelastic scattering between light and acoustic phonons within a material. The latter cause a frequency shift in the scattered light (f_B, GHz) from which the elastic modulus of a material can be deduced (72–74). The study analyzed three oNP bulk films with variable cross-linker contents: B90E10, B75E25, and B60E40 (SI Appendix, Fig. S14). The frequency shift f_B(q) of the scattered light, as a function of the scattering wavevector (q, nm⁻¹), was plotted accordingly for these samples (Fig. 4). The observed linear acoustic relation yields the longitudinal sound velocity c_L via $f_B=c_{L}q/2\pi$ (where for the propagation parallel to the film, $q=(4\pi /\lambda )\sin (\theta /2)$ with θ denoting the scattering angle and λ the wavelength of light) (75, 76). The calculated sound velocities increased with the EGDMA content in oNPs, as 2,412 m s⁻¹ for B90E10, 2,520 m s⁻¹ for B75E25, and 2,550 m s⁻¹ for B60E40, respectively; the experimental error amounts to less than 1%. The increase in sound velocity corresponded to about 12% enhancement in the estimated longitudinal modulus (M_oNP) of the oNP from B90E10 to B60E40 assuming the same density ($M_{\mathit{oNP}}=\rho c_L^2$, where ρ is the density). The bulk-like modulus revealed by BLS suggested that oNPs featured near bulk density. This indicated the effectiveness of the dual cross-link and hyperbranching mechanism underlying oNP formation.

The thermal behavior of the oNPs was evaluated using differential scanning calorimetry (DSC) (SI Appendix, Fig. S15). This analysis involved comparing the thermal properties of PBiBEM and PEGDMA homopolymers, which were synthesized using a similar ME ATRP method, to those of the oNPs. The glass transition temperature (T_g) for each sample was identified by the peak position of the differential heat flow derivative (DDSC). The findings revealed that PBiBEM displayed a distinct and relatively narrow glass transition with a T_g around 34 °C. In contrast, the PEGDMA homopolymer did not exhibit a noticeable glass transition even when the temperature range was extended up to 195 °C (SI Appendix, Fig. S16), although previous studies estimated PEGDMA T_g to be around 144 °C using oxygen permeability data obtained from electron spin resonance (77). The oNP samples, in comparison to PBiBEM, all showed broader glass transitions which became less discernible with increasing EGDMA content. Additionally, their T_gs shifted from 37 °C for SB90E10 to 43 °C for SB60E40. This trend suggested that as the hyperbranched structure became more densely cross-linked, the internal macromolecular mobility was significantly restricted. Despite the expectation that highly cross-linked oNPs would exhibit their predicted T_gs at higher temperatures, the observed weak glass transitions and T_gs around 40 °C were attributed to the presence of remaining dangling chains. Further thermostability testing through thermogravimetric analysis (TGA) indicated that the oNPs were thermally stable up to 250 °C (SI Appendix, Fig. S17).

ATRP provides macromolecules with high end-group fidelity for chain extensions (78, 79). The large fraction of bromide from BiBEM enabled oNPs to graft polymeric brushes directly from the surface (i.e., oNP-Br), bypassing additional surface modifications needed, for example, for silica particles. In a model SI-ATRP reaction, ethyl α-bromoisobutyrate (EBiB) acted as a “sacrificial initiator” to estimate the molecular weight of the grafted chains (Fig. 5A) (80, 81). Methyl methacrylate (MMA) was chosen as the monomer to determine the absolute molecular weight. Surface-initiated activators regenerated by electron transfer (SI-ARGET) ATRP method was employed, with tin(II) 2-ethylhexanoate [Sn(Oct)₂] serving as the reducing agent (82–84) The grafting process was quenched at relatively low viscosity to mitigate the radical coupling effects and cross-linking between brush particles. The mixtures of PMMA-grafted oNPs (oNP-g-PMMA) and linear PMMA (EBiB-PMMA) were purified through precipitation in methanol and redispersion in THF for further analysis. The THF SEC revealed that the elution time for particle brushes B75E25-g-PMMA shifted toward a higher molecular weight compared to the original oNP (Fig. 5B), confirming the retention of end-group functionality in the oNP-Br macroinitiator. Given the distinct separation between the peak positions, the molecular weight indicated by the peak of isolated EBiB-PMMA was attributed to the grafted PMMA brush layers. (SI Appendix, Figs. S18 and S19).

Grafting density (σ, chains nm⁻²), a pivotal parameter for nanoparticle brushes, defines the number of polymer chains grafted per unit surface area (85) To accurately determine the grafting density of oNP-g-PMMA, extensive reaction metrics were exploited, including the hydrodynamic diameter (D_oNP) and the calibrated molecular weight (M_n,MALS,oNP) of oNPs, monomer conversion of MMA (conv.), and the absolute molecular weight of grafted PMMA brushes (M_{n, brush}). The comprehensive calculation methodology was detailed in SI Appendix, Eqs. S4–S12 and Table S4), with the consolidated findings presented in Table 2.

The efficacy of oNP-Br macroinitiator with multiple initiating sites for SI-ATRP was evidenced by the average number of chains grafted or the number of “active bromides” (N_{active -Br}) from the oNP surface. The term “active” was introduced to highlight the difference to “buried -Br” that were not able to initiate chains due to the steric hindrance and the lack of accessibility to activators and monomers in the interior or even at the peripheral areas of oNP-Br. Our findings revealed that the N_{active -Br} for oNP-g-PMMA consistently decreased with an increase in EGDMA composition. Despite the reduction in hydrodynamic sizes, the grafting density (N_{active -Br} per surface area) remained similar, ranging from 0.48 to 0.67 chains nm⁻². These values were similar to the grafting densities for densely grafted inorganic nanoparticle brushes (86) For inorganic nanoparticles, polymer grafts can be initiated from a single layer from the solid surface. However, initiation from the oNP-Br surface could involve several Br-layers due to the hyperbranched structure at the fringe (SI Appendix, Fig. S20). Consequently, the proportion of active -Br (f_{active -Br}) was approximately 11% based on the average total number of BiBEM molecules per oNP. The thickness of this multilayer (Δh) pertaining to active -Br was estimated, based on several simplified assumptions (SI Appendix, Eqs. S13 and S14 and Table S5): 1) the oNPs were considered to be perfectly spherical; 2) BiBEM and EGDMA, both being methacrylates, were assumed to have comparable reactivity ratios; and 3) there was no sterically hindered -Br at the oNP surface (87). For B90E10-g-PMMA, Δh was estimated at 0.71 nm. Considering the statistical segment length of a vinyl monomer unit as approximately 0.25 nm, the 0.71 nm thickness corresponded to about 2 to 3 layers of BiBEM containing active -Br. With an increase in cross-linking density, Δh gradually decreased to 0.38 nm, influenced by the reduced size and enhanced internal rigidity, yet still indicative of nearly a bilayer of active -Br.

The oNP-g-PMMA particle brushes were efficiently purified via ultrahigh-speed centrifugation at 16,000 × g for 1 h to separate linear polymers (EBiB-PMMA). Subsequent DLS measurement determined the intensity-averaged hydrodynamic sizes of the purified particle brushes (SI Appendix, Fig. S21). For instance, B75E25-g-PMMA dispersed in THF had an average diameter of 144 nm without any signs of aggregation. This measurement closely aligned with the approximated size expected for brush particles in a good solvent, assuming a degree of polymerization of 416 (SI Appendix, Eqs. S15–S17 and Table S6). A representative morphology study using AFM was conducted on the additionally synthesized polymer-grafted oNPs (Fig. 5C), with poly(methyl acrylate) (PMA) brushes from oNP B60E40 (B60E40-g-PMA). This combination of oNP and polymer was selected to highlight the contrast between the most rigid oNP and the relatively soft PMA brushes (with T_g of approximately 10 °C). The linear PMA polymers, originating from sacrificial initiators (EBiB-PMA) within the product mixtures, were separated through ultrahigh-speed centrifugation. AFM height images depicted the nanoparticle brushes as evenly distributed, with high uniformity in size and no evidence of aggregation. Moreover, a distinct height contrast was observed suggesting a “core and shell” structure for each brush-tethered oNP, indicative of the interfaces between the oNP cores and the PMA brush layers.

A.K., M.R.B., and K.M. designed research; R.Y., J.T., A.G., J.J., X.H., Y.Z., H.W., Q.L., and G.F. performed research; G.F. and A.K. analyzed data; and R.Y., M.R.B., and K.M. wrote the paper.

The authors declare no competing interest.