Multilayer mesoporous vesosomes of block copolymer

Abstract

Multilayer mesoporous vesosomes uniquely combine hollow multilayer shells with ordered mesopores, providing a basis for multilevel gating, controlled release, and programmed reaction cascades, yet robust construction remains challenging. Herein, we report a molecular engineering strategy based on 3D-confined self-assembly to produce multilayer vesosomes with nearly hexagonally ordered shell perforations. Selective association of pentadecylphenol (PDP) and 1,5-dibromopentane (DBP) with the poly(4-vinylpyridine) block precisely tunes block-copolymer volume fraction to stabilize an unconventional perforated lamellar morphology, while reversible hydrogen bonding enables region-selective pore generation. Experiments and simulations reveal cooperative roles of confinement and hydrogen bonding in directing morphological evolution, with DBP further refining assembly through electrostatic and cross-linking interactions. The resulting multilayer mesoporous vesosomes feature a pyridine-rich corona and show iodine capture (1.26 g g⁻¹), good recyclability, and faster uptake than nonporous multilayer analogues. This work offers a versatile route to complex hollow architectures and clarifies key determinants governing their formation.

The construction of multilayer mesoporous vesosomes that combine hollow multilayer shells with ordered mesopores remains a significant challenge. Here the authors report a molecular engineering strategy based on 3D-confined self-assembly to produce multilayer vesosomes with nearly hexagonally ordered shell perforations.

Introduction

Vesicles are closed aqueous micro-compartments formed by the self-assembly of amphiphilic molecules^1–5. To surpass the functional limitations of conventional simple vesicles, the concept of the vesosome was introduced in the mid-1990s, inspired by biological multivesicular bodies⁶. These hierarchical assemblies are primarily categorized into two topologies: (i) nested vesosomes, where an outer lumen encapsulates multiple independent smaller vesicles derived from trapped aqueous droplets, forming a disordered, multicompartment architecture; and (ii) multilayer vesosomes, characterized by onion-like shells separated by liquid spacers^7,8. While early research prioritized lipid-based nested systems to minimize leakage, multilayer architectures have gained prominence^9,10. Their stacked interfacial barriers and tunable interlayer spacing not only maximize cargo retention and stability but also enable compartmentalized loading and programmed release^11–13. This combination of structural robustness and functional versatility establishes them as superior platforms for advanced drug delivery and nanoreactors^14,15.

Self-assembly of amphiphilic block copolymers (BCPs) serves as a versatile platform for constructing multilayer vesosomes^16–18. In many systems, these hierarchical architectures emerge under kinetically controlled, non-equilibrium conditions: variations in selective-solvent composition and exchange protocols drive morphological transitions, eventually trapping the shells as chain mobility becomes progressively restricted^19,20. Representative examples include onion-like multilayer aggregates formed by poly(tert-butyl acrylate)-block-poly(2-vinylpyridine) (PtBA-b-P2VP) in acidic aqueous media, which demonstrate the critical role of ionization state in regulating curvature and stacking²¹. Similarly, phase-diagram and pathway studies on model systems such as polystyrene-block-poly(acrylic acid) (PS-b-PAA) reveal that the accessibility of multilayer morphologies is strictly governed by solvent quality, block architecture, and solvent-exchange kinetics^22–24. More recently, other advanced fabrication strategies have been developed²⁵. For instance, polymerization-induced self-assembly (PISA) provides a direct route to onion-like multilayer structures^26–28, while liquid-solid interfacial fusion assembly (LSIFA) enables the precise fabrication of uniform and tunable uni- and multilayer vesosomes²⁹. Despite these advances, the walls of multilayer vesosomes in reported systems are typically non-mesoporous. A key next objective is to engineer multilayer vesosomes in which each shell incorporates stable, ordered, through-going mesopores. Such an architecture would preserve the structural benefits of stacked barriers while facilitating rapid mass transport and improving interfacial accessibility. Conceptually, this target structure aligns with the perforated-lamella morphology, effectively bridging a multilayer framework with permeable transport channels. Realizing this specific phase motif within a multishelled architecture represents a critical frontier in vesicular design.

In this context, emulsion-based three-dimensional confined self-assembly (3D-CSA) offers a promising platform by directing BCP organization inside droplets under geometric confinement coupled to solvent evaporation/solidification, a setting that naturally favors particles with well-defined nanostructures^30–37. In some cases, these droplet-derived microparticles can be converted into porous structures via self-emulsification or post-processing methods, such as selective swelling^38–47. Previous studies using 3D-CSA have successfully yielded closed multilayer vesosomes without mesopores, including hydrogen-bonding-assisted multilayer particles, BCP capsules with structure-dependent release and stimuli-responsive photonic microparticles^48–51. Unlike conventional solution assembly, the phase-separation behavior in droplets parallels that in bulk systems, which theoretically renders the perforated-lamella accessible^52–57. However, accessing the target perforated-lamella phase in vesosomes remains elusive due to its intrinsic instability. Theory and simulation suggest that the perforated-lamella state occupies an extremely narrow thermodynamic window^58–61, so its experimental realization typically demands stringent control over the assembly pathway. In practice, the hollow framework and shell curvature must be maintained while the nascent through-pore network is stabilized against the thermodynamic driving forces that otherwise promote relaxation into simpler lamellar or cylindrical states. As a result, experimental demonstrations of multilayer vesosomes featuring ordered, through-going mesopores have remained scarce over the past two decades.

Here, we demonstrate the fabrication of multilayer vesosomes with ordered mesopores via an emulsion-based 3D-CSA strategy, empowered by synergistic molecular engineering (Fig. 1). We combine reversible hydrogen bonding from 3-pentadecylphenol (PDP) and cross-linking from 1,5-dibromopentane (DBP) to program the volume fraction and the effective packing parameter (p_eff) of polystyrene-block-poly(4-vinylpyridine) (PS-b-P4VP), thereby steering its assembly into the elusive perforated-lamella morphology. Subsequent selective removal of PDP generates near-hexagonal mesopores across every shell while exposing pyridinic interfaces, yielding a unique architecture visualized by 3D electron tomography. Experiments and SCFT modeling together elucidate the cooperative roles of molecular packing control and spatial confinement in stabilizing the perforations. The proposed mechanism is further evidenced by the controlled synthesis of AB_n monolayer mesoporous vesicles (n = 1–8) and the capture of evolutionary intermediates. The resulting multilayer mesoporous vesosomes, featuring a reactive P4VP corona, exhibit an iodine uptake capacity of 1.26 g g⁻¹, 81.6% capacity retention over repeated cycles, and faster uptake kinetics than onion-like multilayer vesosomes without mesopores.

Fig. 1.

Schematic diagram of the molecular-engineering-enabled 3D confined self-assembly (3D-CSA) for constructing multilayer vesosomes with ordered mesopores. The pathway contrasts the conventional nested vesosomes and multilayer vesosomes without mesopores.

Results

Synthesis and structure of multilayer mesoporous vesosomes

Ordered multilayer mesoporous vesosomes were prepared by emulsion-solvent evaporation driven 3D-CSA. In order to precisely tune the packing parameter required for the unconventional morphology, perforated-lamella, a molecular engineering strategy by combining side-group grafting and cross-linking was adopted. First, the BCP PS₄₀₉-b-P4VP₁₇₁ (73.9 wt% PS) was dissolved in a good solvent (e.g., chloroform), and PDP was added to form comb-like supramolecules, denoted as PS-b-P4VP(PDP)_1.2, via hydrogen bonding to the pyridine units. DBP was then introduced as a crosslinker to yield PS-b-P4VP(PDP)_1.2/DBP_0.4, where 1.2 and 0.4 denote the molar ratios of PDP and Br equivalents (from DBP) to pyridine units, respectively. Fourier transform infrared spectroscopy (FT-IR) of the PS-b-P4VP(PDP)_1.2/DBP_0.4 complex (Supplementary Fig. 1, see Supplementary Information) revealed critical interactions: the characteristic pyridine ring vibration of non-bonded P4VP at 1598.9 cm⁻¹ underwent distinctive peak splitting into doublet signals at 1600.2 cm⁻¹ and 1589.2 cm⁻¹ upon PDP binding, confirming hydrogen-bond formation. Concurrently, the emergence of a new absorption band at 1662.5 cm⁻¹ provided direct evidence for quaternization between DBP’s halogens and P4VP’s nitrogen atoms. The PS-b-P4VP(PDP)_1.2/DBP_0.4 solution was emulsified into an aqueous polyvinyl alcohol (PVA) solution to generate emulsion droplets, which produced BCP microparticles⁴⁸ with microphase structures through solvent (chloroform) evaporation.

A key step is the reversibility of PDP-P4VP hydrogen-bonding in ethanol/water, which enables selective swelling and removal of PDP without dissolving the PS-rich framework⁴⁸. The mobility of the PS block at room temperature is limited because of its poor solubility (in ethanol) and high glass transition temperature (T_g), which is important for the generation of a stable frame for the mesoporous vesosomes. Thus, the ethanol-induced structural evolution mainly arises from selective swelling of the P4VP-domain and PDP extraction within the preorganized multilayer framework. In parallel, partial cross-linking by DBP furnishes sufficient fixation of the P4VP domains to retain curved lamellae and the multishell framework during swelling. As PDP is released, the polar domains selectively disassemble/reorganize, transforming perforated lamellae into mesoporous lamellae. Thus, the synergy of reversible supramolecular association and mild cross-linking under 3D confinement yields multilayer mesoporous vesosomes. Their accessible pore/shell interfaces are enriched in P4VP/pyridyl groups, while the inner domains are mainly PS (Fig. 2a).

Fig. 2. Morphology and porosity of multilayer mesoporous vesosomes.

a Schematic diagram of ordered multilayer mesoporous vesosomes formed by 3D-CSA via molecular engineering. b, c TEM and SEM images of multilayer mesoporous vesosomes. d A dark-field high-resolution TEM (HRTEM) image of a multilayer mesoporous vesosome and corresponding element mapping images. e Comparison of Brunauer-Emmett-Teller (BET) specific surface areas of the multilayer mesoporous vesosomes (blue) and the multilayer vesosomes without mesopores (green).

The TEM image (Fig. 2b) reveals a multishelled architecture in which every shell presents a periodic in-plane fine structure, with shell thickness h₁ = 37.1 ± 3.2 nm and in-layer period d₁ = 71.9 ± 2.4 nm. In agreement, the SEM image (Fig. 2c) resolves hexagonally arranged surface pores with pore diameter l₀= 22.5 ± 1.7 nm and in-layer period d₂ = 72.5 ± 3.6 nm; the close match d₁ ≈ d₂ indicates that the TEM contrast reflects the pore lattice observed by SEM, and that the periodic motif is intrinsic to each shell rather than an imaging artifact. Furthermore, elemental mapping (Fig. 2d) demonstrates uniform distributions of C, N and Br across individual particles; accordingly, the pyridinic nitrogen lining the mesopores provides accessible sites for post-functionalization. In addition, nitrogen adsorption-desorption (Supplementary Fig. 2) exhibits a type-IV isotherm, confirming mesoporosity. The multilayer perforated vesosomes exhibit a much higher Brunauer-Emmett-Teller (BET) specific surface area (125.89 m² g⁻¹) than the multilayer vesosomes without mesopores (11.27 m² g⁻¹), consistent with the open, interconnected through-pores of perforated lamellae versus the more closed multilamellae in onion structures (Fig. 2e). The sorption-derived pore size is ≈ 17 nm, smaller than l₀, which is consistent with differences between gas-phase models and real pore geometry, potential drying-induced shrinkage/partial closure, and the 3D interconnection of through-pores in perforated lamellae that broadens the apparent distribution. The pore size can also be tuned by varying the molecular weight of the BCP. When PS₄₀₉-b-P4VP₁₇₁ was replaced with PS₅₈₆-b-P4VP₂₀₀ under otherwise identical conditions, multilayer mesoporous vesosomes with a larger pore diameter of 24.4 ± 3.2 nm were obtained (Supplementary Fig. 3). The in-layer periods measured from SEM and TEM images also increased to 73.1 ± 4.5 nm and 74.0 ± 2.5 nm, respectively.

A TEM tilt series was acquired from −60° to +56° (Movie S1), and representative raw frames are shown in Fig. 3a. The aligned series was reconstructed tomographically (Movie S2), yielding a 3D volume in which xy-orthoslices extracted at successive z-positions (Fig. 3b) reveal a shell-by-shell repetition of the same ordered porous motif through the particle thickness. Thus, the contrast modulation that defines the in-plane lattice persists along the z-axis, indicating that the pores traverse the shell rather than terminating internally. Orthogonal projections of the full volume (top, bottom, and side views; Supplementary Fig. 4) further corroborate the radial continuity of the multilayer framework. For spatial registry, we separately reconstructed the outermost shell and the interior domains and then merged the volumes (Fig. 3c), showing that the shells are nested concentrically and share a common pore-lattice orientation over a large angular sector. To interrogate connectivity, serial reslices were generated through the 3D volume (Movies S3–S5), which provide a more faithful interpretation than single orthoplanes; complementary orthoslice views along the x, y, and z axes (Movies S6–S8) confirm continuous, percolating mesoporosity within each shell. Taken together, these observations show that each particle consists of a nested, multilayer skeleton in which every shell contains hexagonally arranged through-pores, consistent with an ordered perforated-lamella architecture. In further support, an SEM image of a partially fractured (incomplete) vesosome directly exposes the internal perforated lamellar shells (Supplementary Fig. 5), providing complementary real-space evidence for the multilayer, through-pore framework.

Fig. 3. 3D electron-tomographic reconstruction of multilayer mesoporous vesosomes.

a Raw TEM images of multilayer mesoporous vesosomes taken at various angles from −60° to +56°. b Representative xy-slices and corresponding binary segmentations from the 3D tomographic reconstruction of multilayer mesoporous vesosomes. The numbers above the images indicate the slice index. The intervals in the z-direction between successive slices are 39.5 nm, and the voxel size along the z-axis is 0.789 nm. c A 3D volume rendering of multilayer mesoporous vesosomes, generated by separately reconstructing the outermost shell and interior domains and then merging the volumes, showing top, bottom, and side views.

Formation mechanism of multilayer mesoporous vesosomes

Control of packing-parameter by hydrogen-bonded side groups and cross-linking

To elucidate the formation mechanism of multilayer mesoporous vesosomes, the effect of introducing DBP was first assessed. Notably, the presence of two halogen atoms in a single DBP molecule allows it to quaternize two pyridine groups, serving as a crosslinker within the P4VP microdomain. Thus, DBP plays a dual role: quaternization increases the polarity of P4VP-rich domains and contributes to packing-parameter regulation, while cross-linking stabilizes the perforated lamellar framework during ethanol treatment and PDP extraction. Although earlier studies identify DBP as a cross-linker that enhances the stability of P4VP domains⁶², its specific role in adjusting the packing parameter remained unclear. Accordingly, microparticle morphologies with and without DBP were compared under otherwise identical 3D-CSA conditions, together with their morphologies after ethanol treatment (Fig. 4a–f). In the absence of DBP, the supramolecular PS-b-P4VP(PDP)_1.2 self-assembles under identical 3D-CSA conditions into onion-like microparticles (Fig. 4b), which can yield multilayer vesosomes without mesopores (Fig. 4c). By contrast, in the PS-b-P4VP(PDP)_1.2/DBP_0.4 system, centrifugation in different solvents gives clearly distinguishable outcomes: in ethanol, the particles readily form a visible pellet, whereas in water they remain largely suspended and no pellet is observed (Supplementary Fig. 6). UV-vis of the ethanol supernatant confirms PDP removal after centrifugation (Supplementary Fig. 7), supporting supramolecular swelling/removal as the origin of pore space. TEM images of the water-treated particles reveal that the originally onion-like lamellae become locally perforated, with P4VP-rich domains penetrating into the PS-rich shells (Fig. 4e). Subsequent ethanol treatment further enlarges these perforations and promotes the coalescence of internal voids into 3D-interconnected mesopores and hollow interiors, ultimately yielding multilayer mesoporous vesosomes (Fig. 4f). Notably, light intermolecular cross-linking between P4VP segments constrains interfacial chain splay and suppresses local area fluctuations, effectively lowering the interfacial area per chain (a₀↓)⁶³. At the same time, quaternization increases the polarity of P4VP and its affinity toward ethanol, so that a larger P4VP-rich volume is retained during selective swelling and PDP removal (v↑). Within the qualitative framework of the packing parameter, we therefore consider an effective packing parameter p_eff = v/(a₀ l_c): the coupled v↑ and a₀↓ shifts move the system beyond the symmetric-lamella window (p_eff ≈ 1) into the perforated-lamella regime under droplet confinement, enabling multilayer mesoporous vesosomes rather than onion-like stacks⁶³.

Fig. 4. Molecular engineering of the perforated-lamella morphology under spherical confinement.

a Schematic representation of PS-b-P4VP(PDP)_1.2 microparticles without DBP and b, c corresponding TEM images. d Schematic representation of PS-b-P4VP(PDP)_1.2/DBP_0.4 microparticles and e, f corresponding TEM images, showing the role of DBP in the morphology of multilayer mesoporous vesosomes. g FT-IR spectra of PS-b-P4VP(PDP)_x/DBP_0.4 microparticles with different PDP content. h Graph of the relationship between the relative content of PDP (x) and the theoretical volumetric fractions of P4VP(PDP) (f_P4VP(PDP)), the effective packing parameter and the morphology of microparticles in the PS-b-P4VP(PDP)_x/DBP_0.4 systems. i The simulated diagrams of PS-b-P4VP(PDP)_x/DBP system under spherical 3D confinement highlight the evolution with increasing supramolecular loading x. The outer blue envelope denotes the droplet boundary.

Additionally, supramolecular interactions of bottlebrush polymers with hydrogen-bonding side groups frequently give rise to distinctive self-assembly behaviors⁶⁴, and open the possibility of tuning the effective packing parameter (p_eff). FT-IR first quantifies the extent of supramolecular association. As the PDP/P4VP molar ratio x increases, the pyridine ring vibration at ca. 1598.8 cm⁻¹ shifts progressively (Fig. 4g) and approaches a plateau at x ≈ 2.0. From x = 0.6 to x = 3.0, the position of this band moves monotonically, reflecting increased hydrogen-bond loading. At x = 1.2, the shift remains below the saturation level, whereas the spectra at x = 2.0 and x = 3.0 are nearly identical, indicating that hydrogen-bond saturation is essentially reached by x ≈ 2.0. This FT-IR trend thus provides an experimental handle on the hydrogen-bond loading for tuning domain volumes under 3D-CSA. In parallel, the particle morphologies evolve coherently with x: starting from spherical and cylindrical states at low x, passing through lamellae and semi-perforated intermediates, and reaching ordered perforated-lamella shells near x ≈ 1.2; at higher x, inverse-type structures (e.g., inverse cylinders/hexosomes) emerge as p_eff > 1 becomes dominant. These coupled trends are summarized in Fig. 4h as a conceptual mapping linking x (hydrogen-bond loading) to an increasing P4VP(PDP)-rich effective volume fraction (f_P4VP(PDP)) and a corresponding upward shift in p_eff, highlighting the narrow composition window for forming multilayer mesoporous vesosomes with perforated lamellae. The basic structural units and how x steers the effective packing are summarized in the schematic of PS-b-P4VP(PDP)_x/DBP_0.4, while representative experimental outcomes are consistent with the trends already established in the TEM survey (Supplementary Fig. 8).

A narrow window near the perforation regime also yields hierarchical byproducts featuring perforated outer shells and helical inner columns; their yield and 3D confirmation are documented in Supplementary Figs. 9, 10 and Movies S9 and S10. Occasional incomplete particles further reveal the nested (Supplementary Fig. 11), inside-out stacking of the hierarchical framework. The emergence of typical hexosome structures (Supplementary Fig. 12) further confirms the effectiveness of packing-parameter control in this system. Unlike the multilayer mesoporous vesosomes with ordered surface mesopores (Fig. 1c), these hexosome structures show clear cylindrical domains in SEM images and no continuous perforated lamellar shell. To synthesize these observations, Table S1 presents a comprehensive composition-morphology summary that integrates experimental data with the calculated volume fractions of P4VP(PDP) and PS over the range x = 0–3.0. Notably, the multilayer perforated region at x = 1.2, with a PS volume fraction (f_PS) of ≈ 0.38, falls within the perforation-favoring window extracted from theory. Finally, the block ratio r = M_n(PS)/M_n(P4VP) imposes an additional constraint: when r approaches or falls below unity, systems at x > 1 yield inverse spherical, non-porous particles rather than perforated lamellae (Supplementary Fig. 13 and Table S2), underscoring that supramolecular loading (x) and molecular architecture (r) must be co-tuned.

The incorporation of PDP triggers supramolecular assembly with P4VP via hydrogen bonding, leading to significant volumetric expansion of the P4VP-PDP complex. This results in a phase volume inversion where the P4VP-PDP phase surpasses PS in volume fraction, despite the higher molecular weight of PS. To validate these assignments, we employed SCFT based on the Gaussian-chain model and masking scheme to predict the equilibrium morphologies under spherical confinement⁶¹. The mesoporous cavity wall was represented by a smoothed density profile ϕ_W(r) that regularizes the solid-void interface:

$${\varphi }_{\mathrm{W}}\left(r\right)=\frac{1}{2}\left\{1-tanh\left[\left(R-r-\sigma /2\right)/\lambda \right]\right\},\mathrm{for}R-\sigma \le r\le R$$ 1

where r is the radial coordinate, R is the cavity radius, and σ and λ denote the interface width and shape (softness) parameters, respectively. In this study, we set σ = 0.5 R_g and λ = 0.3 R_g (R_g, polymer radius of gyration). The full model formulation, boundary conditions, and numerical scheme are described in the SI. At fixed DBP_0.4, the simulated pathway (Fig. 4i) reproduces the experimental progression and places the perforated-lamella window near x ≈ 1.2, supporting the conclusion that H-bonded supramolecular loading (x) and light cross-linking (DBP) act in concert to stabilize the polar domain under droplet confinement.

Effect of confinement degree

Here, we therefore investigate the confinement effect as a governing parameter and use the droplet size (or equivalently R/R_g) to program pore multiplicity within a single perforated shell, thereby bridging the monolayer mesoporous vesicle family to the multilayer mesoporous vesosomes formed at larger R. These AB_n-type monolayer mesoporous vesicles are mainly observed in smaller particles under stronger confinement, where only one perforated shell can be accommodated. They represent the strong-confinement limit of perforated-lamella formation and provide a simplified structural model for understanding how curvature and packing frustration regulate pore formation in confined vesosomal architectures. Figure 5a schematizes the confinement pathway during solvent evaporation: varying the initial emulsion droplet radius R tunes the effective confinement strength (R/R_g), which in turn dictates how many perforations can be accommodated within a monolayer shell. Representative TEM images (Fig. 5b) reveal a systematic series of AB_n-type monolayer mesoporous vesicles spanning single-pore, nanoring, double-, triple-, four-, five-pore, seven-pore and eight-pore states (denoted S₁, S₂, S₂', S₃, S₄, S₅, S₇, and S₈). Additional examples demonstrating the prevalence and diversity of these structures are provided in Supplementary Fig. 14.

Fig. 5. Confinement-guided formation of AB_n-type monolayer mesoporous vesicles.

a A schematic illustration of the confinement effect during solvent evaporation. b Representative TEM images of AB_n-type monolayer mesoporous vesicles with different pore numbers. The inset is an SEM image with a 50 nm scale bar. c Schematic particle-size distributions under relatively weak and strong confinement. d Graph of the relationship between the radius of AB_n-type monolayer mesoporous vesicles and the number of their pores. Error bars represent s.d. from measurements of 20 individual particles. e The relationship between pore number n and the predicted inter-pore angle(s) θ, together with the corresponding molecular models for each AB_n topology. f Helmholtz free-energy differences for a series of AB_n-type monolayer mesoporous vesicles with increasing pore number at different confinement levels R/R_g. g Schematic diagrams of simulated self-assembly pathways of structural progression in AB_n-type monolayer mesoporous vesicles.

Compared with the multilayer mesoporous vesosomes in Fig. 2 (weaker confinement), the AB_n monolayer mesoporous vesicles form under stronger confinement (Fig. 5c). Consistently, the vesicle radius increases monotonically from 50.8 ± 2.6 nm to 93.1 ± 2.8 nm as the pore number grows from 1 to 8 (Fig. 5d), indicating that larger cavities allow the shell to host more perforations. The pore diameter is 22.3 ± 2.5 nm, and the pore-wall thickness is 38.3 ± 3.1 nm. In addition, several particles have noncircular (anisotropic) boundaries, as indicated by the yellow dashed circles in Supplementary Fig. 15. The outer edge deviates from a circle (slightly elongated, faceted, or undulated), which is consistent with a competition between interfacial area minimization and in-plane stretching or packing penalties. To rationalize these patterns, Fig. 5e correlates the pore number n with the predicted inter-pore angle(s) θ and presents the corresponding molecular models for each AB_n topology, which indicate the underlying point arrangement and nearest-neighbor connectivity.

Complementing the experiment, we performed SCFT-based calculations under spherical confinement using the framework employed elsewhere in this work, with the confinement size (cavity radius R) explicitly fixed. By scanning x and R/R_g, the simulations yield twelve distinct structures spanning the experimentally observed sequence from single-pore to multipore states (up to nine pores). The resulting free-energy landscape (Fig. 5f) and the simulated structural progression (Fig. 5g) jointly rationalize confinement-selected pore multiplicity. Specifically, Fig. 5f presents the Helmholtz free-energy difference for vesicles with different pore numbers at various confinement levels R/R_g. ΔF/(nk_BT) exhibits a non-monotonic dependence on pore number: introducing pores initially lowers the free energy by releasing packing frustration and better accommodating curvature, but beyond an intermediate pore multiplicity, the free energy increases again due to the rapidly growing interfacial area and in-plane stretching penalties. Consequently, each confinement level features an optimal pore-number window, consistent with the experimentally observed discretized pore multiplicities. Consistent with this energetic picture, Fig. 5g summarizes the simulated self-assembly pathway under increasing confinement, showing how the system evolves among competing AB_n topologies as R/R_g varies. The simulated structures and pore multiplicities closely match the TEM/SEM observations in Fig. 5b, while a few additional simulated states (e.g., S₂′, S₅′) appear only rarely or transiently in experiments, consistent with a narrow kinetic window at high confinement. Taken together, the theory provides direct evidence that confinement selects pore multiplicity and geometry: as R decreases, the system cannot build many shells and instead minimizes free energy by multiplying perforations within a single shell. This AB_n monolayer family represents the strong-confinement limit of the same mechanism that, at larger R, yields multilayer perforated lamellae. Therefore, the monolayer mesoporous vesicles provide a simplified structural model for understanding how perforated lamellae nucleate, accommodate curvature, and further develop into multilayer mesoporous vesosomes when the confinement R increases.

Beyond morphology, the multilayer perforated system exhibits simple, predictive geometric-scaling relations that link confinement (droplet size), shell number, and pore multiplicity. These relations arise from pore-packing geometry under spherical curvature and provide compact formulas that anticipate how structure evolves as size changes. Experimentally, multilayer mesoporous vesosomes with different sizes and layer numbers are readily observed. Figure 6a schematically illustrates vesosomes with varying shell numbers and defines the key geometric parameters. Figure 6b shows representative TEM and SEM images of multilayer mesoporous vesosomes with 2–6 shells. A broader survey (Fig. 6c) indicates that the shell number increases with the vesosome radius, from 109.8 ± 6.7 nm to 340.5 ± 14.2 nm. As established above, the shell number in multilayer mesoporous vesosomes is tunable. Although the manual membrane-emulsification method used here cannot yet achieve uniform size control, this limitation could be addressed in the future by optimizing SPG membrane emulsification.

Fig. 6. Confinement-controlled multilayer mesoporous vesosomes.

a 3D cartoon representation. b TEM and SEM images of multilayer mesoporous vesosomes with 2–6 layers. c Graph of the relationship between the radius of multilayer mesoporous vesosomes and the number of their layers. Error bars represent s.d. from measurements of 20 individual particles. d Graph of a linear correlation between the number of pores per layer (m) and the radius (R/R_g) in double-shell mesoporous vesosomes by theoretical calculations. Helmholtz free-energy differences for a series of (e, f) double-layer mesoporous vesosomes with increasing pore number at different confinement levels R/R_g.

It can be observed that as the radius of the emulsion droplets (R₀) increases, both the R of the multilayer mesoporous vesosomes and the position of the shell layer exhibit a positive shift. When an additional layer can be formed, the inner layer moves outward. Using the in-layer period d ≈ 72.5 nm measured from Fig. 2, and assuming an approximately hexagonal pore lattice, the areal pore density is estimated as:

$${\rho }_0\approx \frac{2}{\sqrt{3}{d}^2}\approx \left(2.2\pm 0.2\right)\times {10}^{-4}pores{nm}^{-2}$$ 2

which corresponds to a surface coverage of ~5–10% by pore openings, consistent with the SEM images.

For multilayer particles, the radius of the n^th shell is well described by a linear relation: R_n = α + sn with fitted parameters α ≈ −10.2 nm and s ≈ 55.3 nm. Here, s is the effective radial pitch between successive perforated shells, i.e., the average distance between adjacent pore lattices along the radial direction, and is therefore comparable to the sum of the shell thickness and the inter-shell spacing.

Because pores occupy the shell surfaces with density ρ₀, the number of pores on the n^th shell and the cumulative pore number up to shell n follow directly from geometry:

$$m\left(n\right)={4\pi {\rho }_0\left(\alpha +sn\right)}^2$$ 3

$$y\left(n\right)=4\pi {\rho }_0\left[n{\alpha }^2+\alpha sn\left(n+1\right)+\frac{s^2}{6}n\left(n+1\right)\left(2n+1\right)\right]$$ 4

implying m(n)∝n² and y(n)∝n³.

Eliminating n via n = (R − α)/s yields an explicit pore-radius relation:

$$y\left(R\right)={\alpha R}^3+b{R}^2+cR+d$$ 5

with coefficients determined solely by ρ₀, s and α:

$$a=\frac{4\pi {\rho }_0}{3s}=1.7\times {10}^{-5}{nm}^{-3},b=2\pi {\rho }_0=1.4\times {10}^{-3}{nm}^{-2}$$

$$c=\frac{2\pi {\rho }_{0}s}{3}=2.5\times {10}^{-2}{nm}^{-1},d=-\frac{2\pi {\rho }_0\alpha }{3s}\left(2{\alpha }^2+3\alpha s+s^2\right)=0.13$$

Here R is in nm and y is dimensionless. The leading cubic term aR³ captures the dominant scaling of total pore number with particle size. Using these relations with the fitted α s, and ρ₀, the model predicts, for example:

$$n=5:m\left(5\right)\approx 196,y\left(5\right)\approx 420$$

These relations enable inference of both the per-layer pore number m and the total pore count y from either the layer index n or the particle radius R. Theoretical calculations further revealed a linear correlation between the number of pores (m) per layer and the radius (R) in double-layer mesoporous vesosomes (Fig. 6d). SCFT simulations demonstrated that increasing the effective radius induced a proportional increase in the total pore number (n) from 10 to 38 pores per shell. This scaling law arises from curvature energy minimization, that is, larger radii accommodate higher pore densities while maintaining constant pore spacing. These findings establish pore number as a tunable structural parameter governed by confinement dimensions. To clarify how pore multiplicity influences the stability of double-layer perforated vesosomes, we calculated the Helmholtz free-energy difference ΔF/(nk_BT) for a series of double-layer vesosomes with different pore numbers at various confinement levels R/R_g. In Fig. 6e, f, curves labeled m_i represent double-layer perforated vesosomes containing i pores in total. The confinement-dependent free-energy trends are analogous to those observed for the monolayer AB_n mesoporous vesicle series in Fig. 5, indicating that, at each R/R_g, only a narrow window of pore multiplicities is thermodynamically favored.

Adsorption performance of multilayer mesoporous vesosomes

In addition, the multilayer mesoporous vesosomes expose abundant pyridinic groups continuously along both the external surfaces and internal pore walls, providing chemically accessible binding sites for functional species⁶⁵. In addition to this chemical functionality, their perforated-lamella architecture offers a pronounced structural advantage: as shown above, the multilayer mesoporous vesosomes possess a much higher specific surface area than multilayer vesosomes without mesopores, implying more efficient transport through abundant mesoporous pathways. Taking iodine adsorption as an example, Fig. 7a schematically illustrates the fundamental difference in mass-transfer behavior. This structural distinction is directly reflected in the adsorption behavior. Upon exposure to iodine vapor at 25 °C, a thin film of multilayer mesoporous vesosomes rapidly darkens (Fig. 7b), indicating fast interfacial uptake. The uptake kinetics further quantify this advantage (Fig. 7c): at identical sorbent mass, the multilayer vesosomes reach an adsorption capacity of 0.67 g g⁻¹ within 30 min, whereas the control group achieves only 0.41 g g⁻¹ over the same period. Importantly, the final capacity is up to 1.26 g g⁻¹, and the multilayer mesoporous vesosomes also retain 81.6% of their initial capacity after five adsorption-desorption cycles (Fig. 7d), confirming good structural stability and regenerability. Aqueous adsorption experiments using KI₃ solutions further highlight the kinetic benefit of the perforated multilayer architecture (Fig. 7e). At equal sorbent mass, both materials reduce the KI₃ concentration from 100 mg L⁻¹ to approximately 2.45 mg L⁻¹ (≈97.6% removal). This terminal concentration (~2.45 mg L⁻¹) is near the UV-vis quantification limit (Supplementary Fig. 16). However, the multilayer mesoporous vesosomes reach near-complete depletion within ~4 h, whereas the control group requires ~8 h to reach a comparable residual concentration. Overall, these results demonstrate that embedding ordered mesopores within multilayer shells enables faster iodine uptake without sacrificing ultimate capacity. The combination of high surface accessibility and efficient transport pathways establishes multilayer mesoporous vesosomes as a structurally optimized platform for rapid iodine capture, and suggests broader potential for designing hierarchical sorbents where kinetics is the critical performance metric.

Fig. 7. Iodine capture by multilayer mesoporous vesosomes.

a Conceptual comparison of iodine mass transfer in multilayer mesoporous vesosomes versus the multilayer vesosomes without mesopores. b Photograph of multilayer mesoporous vesosomes in the α-phase after exposure to iodine vapor for 20 h in a sealed vessel at 25 °C. c Iodine uptake kinetics at 25 °C; inset highlights the rapid uptake within the first 30 min. d Cycling stability of iodine capture by multilayer vesosomes over five adsorption-desorption cycles. e Time-dependent KI₃ concentration monitored by UV-vis; inset shows corresponding photographs of supernatants collected at 0 and 8 h.

Discussion

In summary, we have successfully realized the synthesis of multilayer vesosomes with ordered mesopores by innovatively combining 3D-CSA with a molecular engineering strategy. Our systematic investigation, supported by theoretical simulations, reveals that the cooperative effect of hydrogen bonding and cross-linking is the key determinant for the formation of the unconventional perforated-lamella morphology. The utility of these multilayer mesoporous vesosomes was demonstrated through their application as high-performance adsorbents. With a rich, accessible P4VP corona, the multilayer mesoporous vesosomes deliver 1.26 g g⁻¹ iodine uptake and 81.6% retention over five cycles, while exhibiting faster uptake kinetics than non-mesoporous multilayer vesosomes. This work not only provides a practical and versatile pathway to complex ordered hollow multishelled architectures but also clarifies the critical factors governing their formation. Improved size uniformity can be achieved in the future by optimizing the membrane emulsification technique. We anticipate that the resulting architecture will uniquely integrate two highly desirable structural features. Such a hierarchical framework is expected to enable multilevel gating, hierarchical controlled release, and programmed reaction cascades.

Methods

Materials

PS₄₀₉-b-P4VP₁₇₁ (M_n = 69 kg/mol, M_w/M_n = 1.15), PS₅₈₆-b-P4VP₂₀₀ (M_n = 82 kg/mol, M_w/M_n = 1.04), PS₉₄-b-P4VP₉₅ (M_n = 19.8 kg/mol, M_w/M_n = 1.08) and PS₁₆₃-b-P4VP₄₆₆ (M_n = 66 kg/mol, M_w/M_n = 1.15) were purchased from Polymer Source, Inc. Their molecular weights and compositions were determined by supplier-provided GPC/SEC and ¹H NMR data, respectively. For PS₄₀₉-b-P4VP₁₇₁, the PS weight fraction was 73.9 wt%. Poly(vinyl alcohol) (PVA, average M_w = 13-23 kg mol⁻¹, 87–89% hydrolyzed) was purchased from Aldrich. 3-Pentadecylphenol (PDP, purity >90%), 1,5-dibromopentane (DBP, purity 98%) was purchased from Aladdin Biochemical Technology Co., Ltd. Chloroform (analytical grade), iodine (I₂, purity 99.7%), and potassium iodide (KI, purity 99.0%) were purchased from Sinopharm Chemical Reagent Co., Ltd. Absolute ethanol (EtOH, purity 99%) was purchased from Shanghai Titan Scientific Co., Ltd. Deionized water (Millipore Milli-Q grade) with a resistivity of 18.0 MΩ was used in all the experiments. Materials were used without further purification except for PDP. PDP was purified by recrystallization in n-hexane.

Preparation of initial BCP Microparticles by 3D-CSA

PS-b-P4VP and PDP were separately dissolved in chloroform to prepare stock solutions with concentrations of 10.0 mg mL⁻¹. For a typical preparation of PS-b-P4VP(PDP)_1.2/DBP_0.4 microparticles, 0.5 mL of the PS-b-P4VP solution, containing 5.0 mg of PS₄₀₉-b-P4VP₁₇₁, was mixed with 0.45 mL of the PDP solution, containing 4.53 mg of PDP. The mixture was stirred at room temperature for 24 h to form PS-b-P4VP(PDP)_1.2 supramolecules, where the subscript represents the molar ratio of PDP to pyridine units of P4VP. DBP was prediluted by chloroform to 10 vol% before addition. The amount of DBP was controlled to provide bromine atoms at 0.4 equivalents relative to the pyridine units of P4VP, corresponding to 0.60 mg of DBP. The mixture was further stirred at room temperature for at least 4 h to obtain the PS-b-P4VP(PDP)_1.2/DBP_0.4 solution. Then, 0.1 mL of the mixed solution was emulsified with 1.0 mL of PVA aqueous solution (5.0 mg mL⁻¹) by using a membrane-extrusion emulsification technique. The emulsion was collected in an open vial (10 mL), and microparticles were generated after natural evaporation of chloroform for at least 48 h at 30 °C.

Preparation of multilayer mesoporous vesosomes

Approximately 100 μL of the above-obtained microparticle dispersion (containing approximately 1.0 mg of microparticles) was mixed with 900 μL of ethanol. The mixture was shaken for approximately 30 s to ensure homogeneous mixing and then centrifuged (14000 r min⁻¹ for 10 min). Afterwards, the precipitate was dispersed in ethanol by brief ultrasonication for approximately 30 s and centrifuged to remove the supernatant. This ethanol washing step was repeated twice. The whole ethanol treatment process was performed at room temperature. Finally, both PVA and PDP were removed, yielding multilayer mesoporous vesosomes.

Preparation of monolayer mesoporous vesicles

The preparation of monolayer mesoporous vesicles is similar to that of multilayer mesoporous vesosomes, except that the membrane-extrusion emulsification step for forming the initial BCP microparticles requires an increased number of passes, at least 30. Here, one pass is defined as one single passage of the emulsion through the membrane in the membrane-extrusion emulsification device.

Preparation of multilayer vesosomes without mesopores

The procedure was identical to that used for preparing the multilayer mesoporous vesosomes, except that DBP was not added and the ethanol-treatment step was extended. Specifically, 100 μL of the obtained initial BCP microparticle dispersion was centrifuged to remove the supernatant, and the precipitate was then immersed in 1 mL of ethanol for 24 h at room temperature.

Iodine adsorption test

For iodine vapor adsorption, 5 mg of microspheres was placed in a sealed container with excess iodine crystals without direct contact and kept at 25 °C. At selected times, the sample was weighed after removing surface iodine crystals to determine the iodine uptake. For KI₃ solution adsorption, 5 mg of microspheres was added to 0.2 g L⁻¹ KI₃ aqueous solution and kept at 25 °C. At selected times, the supernatant was collected after centrifugation and analyzed by UV-vis spectroscopy to determine the residual KI₃ concentration, adsorption capacity, and removal efficiency.

Characterization

The internal structure of the particles was observed by using a Hitachi HT7700 transmission electron microscope (TEM) operated at an accelerating voltage of 100 kV. Element mapping was measured by an energy dispersive X-ray spectroscopy (EDX) by using a Talos F200X high-resolution TEM (HRTEM) operated at an accelerating voltage of 200 kV. The TEM models for 3D reconstruction of microparticles were JEM-1400/JEM-2100 (JEOL Ltd., Tokyo, Japan), and refinement of the initial alignment and tomographic reconstruction of the tilt series were performed in IMOD. 3D visualization of the final volumes was carried out in UCSF Chimera and Amira-Avizo Software (Thermo Fisher Scientific, MA, USA). The surface morphology of the particles was observed by using a Hitachi SU8010 scanning electron microscope (SEM) operated at an acceleration voltage of 1.0 kV. Before TEM and SEM characterization, the samples were stained with iodine vapor for 1 h. UV-vis spectra were detected by spectrophotometers (TU-1810 and UV-2700). The specific surface area and pore size were analyzed by using an ASAP 2420-4MP automatic specific surface area and porosity analyser. Hydrogen-bonding and quaternization interactions were analyzed by Fourier transform infrared spectroscopy (FT-IR, Bruker VERTEX 70).

Theoretical modeling via SCFT

Free energy functional and wall density profile

To predict the equilibrium morphologies of the supramolecular BCP system in spherical confinement, we employed self-consistent field theory (SCFT) under the Gaussian chain approximation. The experimental system was coarse-grained as an AB diblock copolymer with C blocks grafted onto the B block, where the A, B, and C blocks represent PS, P4VP, and PDP/DBP-associated side groups, respectively. The volume fraction of the A block was fixed at f_A = 0.74, corresponding to the PS fraction of PS₄₀₉-b-P4VP₁₇₁ used in the experiment. The segregation strengths were set as χ_ABN = χ_ACN = 25 and χ_BCN = 0, where χ is the Flory-Huggins interaction parameter. The volume fraction of the C block (f_C) was varied to represent the PDP/P4VP molar ratio x. The spherical confinement was used to mimic the emulsion droplet environment during 3D confined self-assembly. The wall interface of the mesoporous cavity was modeled by a smoothed density profile:

$${\varphi }_{\mathrm{W}}\left(r\right)=\frac{1}{2}\left\{1-tanh\left[\left(R-r-\sigma /2\right)/\lambda \right]\right\},\mathrm{for}R-\sigma \le r\le R$$ 6

where σ and λ denote the width and shape of the wall interface, respectively. In this study, we choose σ = 0.5R_g and λ = 0.3R_g.

SCFT equations and numerical solution

The free energy functional of SCFT on the basis of Gaussian chain model at a given temperature T is written as:

$$\frac{F}{n{k}_{\mathrm{B}}T}=-\ln Q+\frac{1}{V}{\int }_{∣\mathbf{r}∣<R}d\mathbf{r}\left\{\right){\chi }_{\mathrm{AB}}N{\varphi }_{\mathrm{A}}\left(\mathbf{r}\right){\varphi }_{\mathrm{B}}\left(\mathbf{r}\right)+{\chi }_{\mathrm{AC}}N{\varphi }_{\mathrm{A}}\left(\mathbf{r}\right){\varphi }_{\mathrm{C}}\left(\mathbf{r}\right)+{\chi }_{\mathrm{BC}}N{\varphi }_{\mathrm{B}}\left(\mathbf{r}\right){\varphi }_{\mathrm{C}}\left(\mathbf{r}\right)$$

$$+{\chi }_{\mathrm{AW}}N{\varphi }_{\mathrm{A}}\left(\mathbf{r}\right){\varphi }_{\mathrm{W}}\left(\mathbf{r}\right)+{\chi }_{\mathrm{BW}}N{\varphi }_{\mathrm{B}}\left(\mathbf{r}\right){\varphi }_{\mathrm{W}}\left(\mathbf{r}\right)\mathit{+}{\chi }_{\mathrm{CW}}N{\varphi }_{\mathrm{C}}\left(\mathbf{r}\right){\varphi }_{\mathrm{W}}\left(\mathbf{r}\right)$$

$${-{\omega }_{\mathrm{A}}\left(\mathbf{r}\right)\varphi }_{\mathrm{A}}\left(\mathbf{r}\right){-{\omega }_{\mathrm{B}}\left(\mathbf{r}\right)\varphi }_{\mathrm{B}}\left(\mathbf{r}\right){-{\omega }_{\mathrm{C}}\left(\mathbf{r}\right)\varphi }_{\mathrm{C}}\left(\mathbf{r}\right)$$

$$-\eta \left(\mathbf{r}\right)\left[1-{\varphi }_{\mathrm{A}}\left(\mathbf{r}\right)-{\varphi }_{\mathrm{B}}\left(\mathbf{r}\right)-{\varphi }_{\mathrm{C}}\left(\mathbf{r}\right)-{\varphi }_{\mathrm{W}}\left(\mathbf{r}\right)\right]\mathit{\}}$$ 7

where k_B is the Boltzmann constant. ϕ_A(r), ϕ_B(r) and ϕ_C(r) are the local volume fractions of A-, B- and C-blocks at position r, respectively, while ω_A(r), ω_B(r) and ω_C(r) are their potential fields. The quantity Q is the partition function of a single copolymer chain interacting with the mean fields ω_A(r), ω_B(r) and ω_C(r). η(r) is a Lagrange multiplier used to enforce the incompressibility condition, ϕ_A(r) + ϕ_B(r) + ϕ_C(r) + ϕ_W(r) = 1. We fix χ_AWN > χ_BWN = χ_CWN to make the wall surface prefer the B-blocks (P4VP), reflecting the preferential location of P4VP-rich domains near the PVA-stabilized emulsion interface.

Minimization of the free energy with respect to ϕ_A(r), ϕ_B(r), ϕ_C(r), ω_A(r), ω_B(r) and ω_C(r) leads to the following SCFT equations:

$${\omega }_{\mathrm{A}}\left(\mathbf{r}\right)={\chi }_{\mathrm{AB}}N{\varphi }_{\mathrm{B}}\left(\mathbf{r}\right)\mathit{+}{\chi }_{\mathrm{AC}}N{\varphi }_{\mathrm{C}}\left(\mathbf{r}\right)\mathit{+}{\chi }_{\mathrm{AW}}N{\varphi }_{\mathrm{W}}\left(\mathbf{r}\right)\mathit{+}\eta \left(\mathbf{r}\right)$$ 8

$${\omega }_{\mathrm{B}}\left(\mathbf{r}\right)={\chi }_{\mathrm{AB}}N{\varphi }_{\mathrm{A}}\left(\mathbf{r}\right)\mathit{+}{\chi }_{\mathrm{BC}}N{\varphi }_{\mathrm{C}}\left(\mathbf{r}\right)\mathit{+}{\chi }_{\mathrm{BW}}N{\varphi }_{\mathrm{W}}\left(\mathbf{r}\right)\mathit{+}\eta \left(\mathbf{r}\right)$$ 9

$${\omega }_{\mathrm{C}}\left(\mathbf{r}\right)={\chi }_{\mathrm{AC}}N{\varphi }_{\mathrm{A}}\left(\mathbf{r}\right)\mathit{+}{\chi }_{\mathrm{BC}}N{\varphi }_{\mathrm{B}}\left(\mathbf{r}\right)\mathit{+}{\chi }_{\mathrm{CW}}N{\varphi }_{\mathrm{W}}\left(\mathbf{r}\right)\mathit{+}\eta \left(\mathbf{r}\right)$$ 10

$${\varphi }_{\mathrm{A}}\left(\mathbf{r}\right)=\frac{1}{Q}{\int }_0^{f_{\mathrm{A}}}ds{q}_{\mathrm{A}}\left(\mathbf{r},s\right)q_{\mathrm{A}}^{†}\left(\mathbf{r},s\right)$$ 11

$${\varphi }_{\mathrm{B}}\left(\mathbf{r}\right)=\frac{1}{Q}{\int }_{f_{\mathrm{A}}}^{f_{\mathrm{A}}+f_{\mathrm{B}}}ds{q}_{\mathrm{B}}\left(\mathbf{r},s\right)q_{\mathrm{B}}^{†}\left(\mathbf{r},s\right)$$ 12

$${\varphi }_{\mathrm{C}}\left(\mathit{r}\right)=\frac{1}{Q}{\int }_{f_{\mathrm{A}}+f_{\mathrm{B}}}^{1}ds{q}_{\mathrm{C}}\left(\mathbf{r},s\right)q_{\mathrm{C}}^{†}\left(\mathbf{r},s\right)$$ 13

$$Q=\frac{1}{V}\int d\mathbf{r}{q_{\mathrm{K}}\left(\mathbf{r},s\right)q}_{\mathrm{K}}^{†}\left(\mathbf{r},s\right)$$ 14

In the above equations, q_K(r, s) and q_K^†(r, s) (K = A, B and C) are the propagator functions of the K-block starting from two distinguishable ends at s = 0 and s = 1. Satisfying the following modified diffusion equations:

$$\frac{\partial q_{\mathrm{K}}\left(\mathbf{r},s\right)}{\partial s}={\nabla }^2{q}_{\mathrm{K}}\left(\mathbf{r},s\right)-{\omega }_{\mathrm{K}}\left(\mathbf{r},s\right)q_{\mathrm{K}}\left(\mathbf{r},s\right)$$ 15

$$-\frac{\partial q_{\mathrm{K}}^{†}\left(\mathbf{r},s\right)}{\partial s}={\nabla }^2{q}_{\mathrm{K}}^{†}\left(\mathbf{r},s\right)-{\omega }_{\mathrm{K}}\left(\mathbf{r},s\right)q_{\mathrm{K}}^{†}\left(\mathbf{r},s\right)$$ 16

In our implementation, the modified diffusion equations are solved using the pseudospectral method. The system is put into a cubic box with a size slightly larger than the diameter 2 R of the spherical cavity, and thus, the periodic boundary conditions can be imposed in all three directions. The computational box is discretized into a grid of N_x × N_y × N_z = 128 × 128 × 128. The contour length of the copolymer is evenly divided into 100 pieces, that is, Δs = 0.01. The Anderson mixing scheme is applied to accelerate the convergence process of the solution of the SCFT equations.

The above theoretical predictions successfully identified eighteen distinct morphological families arising from spherical confinement self-assembly. Six additional phases emerge in the PS-b-P4VP(PDP)_x/DBP_0.4 system as x varies: spherical phase microspheres, two cylindrical phase variants, lamellar structures exhibiting either budding or onion-like layering, and multilayer mesoporous vesosomes (perforated lamellar microparticles). Twelve configurations correspond to AB_n-type monolayer mesoporous vesicles exhibiting progressive pore multiplication: from monoporous (single pore) and nanoring architectures through biporous and triporous intermediates, culminating in structures with nine symmetrically arranged pores. This evolutionary sequence reflects confinement-driven entropy minimization as cavity dimensions contract. Quantitative agreement with experimental characterization was observed across all predicted architectures. This concordance validates the SCFT parameterization and establishes confinement geometry and copolymer composition as orthogonal control axes for hierarchical microstructure engineering.

Author contributions

M.W., J.H.H. and H.S. contributed equally to this work. M.W. conceived the study, designed and performed the experiments, analyzed the experimental data, and wrote the original manuscript. J.H.H. performed the 3D electron tomography reconstruction, analyzed the reconstruction results, and wrote the related sections of the manuscript. H.S. conducted the SCFT simulations and contributed to the writing of the related sections. W.L. supervised the simulation study. E.L. supervised the 3D electron tomography reconstruction and data analysis. X.Z created the 3D illustrations of the multilayer mesoporous vesosomes. S.T., S.C. and H.J.K. reviewed the manuscript and provided critical comments and revisions. R.D. and J.Z. conceived and supervised the overall project, contributed to the experimental design and interpretation of the results, and revised the manuscript. All of the authors contributed to the writing.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. [A peer review file is available].

Funding

This work was supported by the National Natural Science Foundation of China (NSFC) (52322314, 52273010, 52293474 and 52511540066), the Interdisciplinary Research Program of Huazhong University of Science and Technology (2025JCYJ003) and the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (RS−2024-00322052).

Data availability

All data supporting the findings of this study are included in the manuscript or the Supplementary Information, or available from the corresponding authors upon request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-77259-9.