Noncovalent synthesis of higher-order super-heterostructures from structurally distinct molecular π-systems via secondary nucleation

Abstract

Noncovalently connecting supramolecular assemblies of structurally distinct π-systems to result in higher-order heterostructures is a grand challenge due to the problem of phase segregation into individual components. Although a few supramolecular heterostructures were recently reported, they are based on molecular π-systems with minimal structural differences. In this study, we demonstrate a rational approach to construct fully supramolecular bicomponent higher-order super-heterostructures such as fibers-reinforced two-dimensional nanoplatelet clusters using structurally distinct π-systems. We have achieved this using near-infrared absorbing tetraimide dye and perylene diimide as the building blocks. Here, perylene diimide acts as a source of dormant monomers, and the self-assembled structures of the tetraimide dye serve as the seeds. Interestingly, tetraimide dye self-assemble into two-dimensional nanoplatelet clusters with switchable hierarchy in dimethyl sulfoxide. When the dormant monomers of perylene diimide are added to the tetraimide seeds, nucleation occurs on the surface of the two-dimensional nanoplatelets of the clusters, and their switchable hierarchy allowed us to synthesize two types of corresponding super-heterostructures selectively. Our investigations further reveal that secondary nucleation and dispersive interactions guide the formation of these super-heterostructures.

Non-covalently connecting supramolecular assemblies of structurally distinct π-systems which yield higher-order heterostructures are challenging due to phase segregation into individual components. Here, the authors demonstrate a rational approach to construct fully supramolecular bicomponent higher-order super-heterostructures using structurally distinct π-systems.

Introduction

Multicomponent hierarchical organization is the key to the function of biological systems¹. Even the smallest bacteria have hierarchical organization that help their survival, such as flagellum for propagation^2,3 etc. One such case of a functional material formed via bicomponent hierarchical organization is the bone, which is constructed from rigid hydroxyapatite crystals interwoven with flexible collagen fibrils^4,5. In organic solar cells, hierarchical supramolecular organization of donor and acceptor components is crucial to achieve the best device performance⁶. On the other hand, the synthesis methods for various inorganic heterostructures have been well established. It was also proved that these heterostructures offer higher performance and programmable functionality than the single-component homostructures^7,8. To synthesize organic heterostructures, supramolecular self-assembly is an attractive approach. Supramolecular self-assembly has been well explored in the field of covalent polymers to create various heterostructures using covalent polymers as building blocks^9–11. However, it is a grand challenge to construct supramolecular heterostructures entirely from small molecular components having substantial structural differences, relying on noncovalent interactions due to the problem of self-sorting or phase separation.

In the last decade, living supramolecular assembly based on primary nucleation was explored to synthesize homo and block supramolecular assemblies with controlled dimensions and sequence, mainly using hydrogen-bonded π-systems^12–21. In these examples, π-systems with minimal structural differences, such as a change in the side chains, were employed to result in block supramolecular assemblies without self-sorting/phase segregation. However, so far, one-dimensional (1D)²² and 2D¹⁷ supramolecular assemblies could be synthesized via living supramolecular assembly based on primary nucleation, because here the topology of the seed dictates the topology of the resulting assemblies. On the other hand, studies on peptide self-assembly reveal mechanisms that generate undesirable neurotoxic oligomeric species during aggregation of Aβ peptide into amyloid fibrils^23–25, via a surface-catalyzed secondary nucleation mechanism. Hence, recently, living supramolecular assembly based on secondary nucleation^26–30 has been explored to construct supramolecular assemblies beyond 1D and 2D topologies^31–33. Some recently achieved higher-order topologies using hydrogen-bonded π-systems are nanopolycatenanes³⁴ and archimedian spirals^35,36. In these examples, hydrogen-bonded π-systems are used to achieve cooperativity during self-assembly. However, utilization of hydrogen-bonded π-systems with significantly different molecular structures to synthesize higher-order heterostructures by overcoming phase segregation is yet to be explored. Even in amyloid peptide fibers, dissimilarity of morphology prevents nucleation on the surface, leading to phase segregation^37,38. These reports suggest that meticulous molecular designs must be explored for synthesizing higher-order supramolecular heterostructures via secondary nucleation.

We have recently shown that cooperative supramolecular assembly can be achieved using π-systems such as alkyl perylene diimides (PDIs), solely based on dispersive interactions³⁹. We have extended this approach for synthesizing higher-order supramolecular structures such as 3D spherulites and scarf-like heterostructures via secondary nucleation³². Here, the heterostructures were synthesized using two PDIs with a variation only in their side chains. Che et al. have also shown interesting topologies such as scarf-like structures using different side chains appended PDIs^40,41. Interestingly, unlike hydrogen-bonded π-systems, alkyl PDIs require only dispersive interactions (π-π stacking and van der Waals interactions) to nucleate and elongate³⁹. Hence, we anticipated that the kinetically trapped monomers of alkyl PDIs can be nucleated and elongated even on the supramolecular assemblies (seeds) of other π-systems with significantly different molecular structures via secondary nucleation. If successful, this would allow us to synthesize exciting higher-order functional supramolecular heterostructures with diverse topologies beyond 1D and 2D.

In the present study, we demonstrate a significant advancement in the field of supramolecular self-assembly, which will pave the way for the synthesis of higher-order, functional supramolecular heterostructures entirely from small molecular components, even with substantial structural differences. We have demonstrated this using a combination of two different types of π-systems, namely, near-infrared (NIR) absorbing blue colored tetraimide (TTI) and a perylene diimide (PDI) (Fig. 1a). Here, the self-assembled nanostructures of TTI act as seeds and the dormant monomers of PDI nucleate and elongate into 1D nanofibers on these seeds via secondary nucleation. TTI with 2-ethylhexyl side chains (Eh-B) forms 2D nanoplatelet clusters in dimethyl sulfoxide (DMSO) (Fig. 1b). Interestingly, the 2D nanoplatelet clusters of Eh-B can be switched between smaller (4 to 5 2D platelets) and larger (>100 2D platelets) clusters in DMSO. The PDI we have used in this study is an alkyl PDI with 2-ethylhexyl side chains at the imide position (Eh-PDI)^32,39. When 2D platelet clusters are used as seeds, nucleation happens on the surface of 2D platelets, resulting in two types (smaller and larger) of fiber-reinforced 2D platelet clusters (SHet-1 and SHet-2) (Fig. 1b). We have further found that these super-heterostructures show better photothermal properties than individual assemblies.

Fig. 1. Molecular structures and a summary of super-heterostructures synthesis.

a Molecular structures of TTI (Eh-B) and perylene diimide (Eh-PDI). b Schematic representation of self-assembled structures (two-dimensional (2D) nanoplatelet clusters) of Eh-B with switchable hierarchy and the formation of two types of super-heterostructures (SHet-1 and SHet-2) via secondary nucleation. Here, Eh-PDI molecules act as dormant monomers, and self-assembled structures of Eh-B act as seeds. Schematic representations were created using Blender 4.4.

Results and discussion

Eh-B is synthesized according to the literature methods (Fig. 1a, Supplementary Section 2 and Supplementary Fig. 1-3)⁴². The presence of phenyl triazoline diones at the bay positions yields NIR absorption and improves the stacking efficiency of TTIs compared to corresponding PDIs³¹. First, we investigated the solution state supramolecular self-assembly of Eh-B alone. A 25 μM solution of Eh-B exists as a monomer in dichloroethane (DCE) and self-assembles in DMSO as indicated by a decrease in absorbance and broadening of the absorption spectra (Fig. 2a and Supplementary Fig. 4a), accompanied by emission quenching (Supplementary Fig. 4b). Field emission scanning electron microscopy (FE-SEM) analysis of self-assembled structures of Eh-B formed in DMSO showed the formation of clusters of 2D nanoplatelets (Fig. 2c). Interestingly, by heating the Eh-B clusters in DMSO at 363 K for 2 min. and cooling to 298 K, smaller 2D nanoplatelet clusters composed of 4 to 5 nanoplatelets were formed (See Methods and Fig. 2b,d). Time-dependent absorption and FE-SEM imaging indicated that these smaller clusters convert into larger clusters having more than 100 2D nanoplatelets over a period of 60 min. (Fig. 2). These observations indicate that the self-assembled structures of Eh-B exhibit intrinsic switchable hierarchical organization. Several sharp peaks in the powder X-ray diffraction pattern of 2D nanoplatelet clusters of Eh-B indicate a crystalline arrangement. The appearance of peaks with d-spacings at 2.06 nm (d) and 1.03 nm (d/2) indicates a lamellar-like arrangement of Eh-B molecules in the 2D nanoplatelets (Supplementary Fig. 5).

Fig. 2. Self-assembly of Eh-B.

a Absorption spectra of Eh-B in DCE (c = 25 µM, l = 1 cm) and its assembly in DMSO, showing spectral changes of transition from small two-dimensional (2D) nanoplatelet clusters to large 2D nanoplatelet clusters. b Photographs of the cuvettes containing small (right) and large (left) 2D nanoplatelet clusters of Eh-B and their interconversion in DMSO. Field emission scanning electron microscopy (FE-SEM) images and schematic illustration of self-assembled structures of Eh-B (c = 25 µM), showing c large and d small 2D nanoplatelet clusters. Schematic representations were created using Blender 4.4.

To understand the differences between the two types of 2D nanoplatelet clusters formed by Eh-B, we have probed their absorption spectra at various temperatures and also by denaturation experiments. Firstly, the DMSO solutions (25 µM) having large and small 2D nanoplatelet clusters were heated at 0.5 K/min, with absorbance monitored at 320 nm (Supplementary Fig. 6). The increase in the absorbance at 320 nm indicates the disassembly. For small nanoplatelet clusters, absorbance at 320 nm increased sharply above 345 K. However, in the case of large nanoplatelet clusters, this change is observed above 338 K. This distinction is due to the initial disintegration of larger 2D nanoplatelet clusters into smaller ones, after which the heating curve closely matches that of the smaller 2D nanoplatelet clusters (Supplementary Fig. 6c). Next, denaturation studies were performed on these assemblies by systematically varying the ratio of good (Dichloroethane, DCE) and poor solvent (DMSO). In both cases, the denaturation curves shift toward higher fractions of good solvent (DCE) with increasing concentration, indicating concentration-dependent stabilization of the assemblies (Supplementary Figs. 7 and 8). At a given concentration, assemblies originating from larger nanoplatelet clusters require a slightly higher fraction of DCE to undergo disassembly, whereas those derived from smaller nanoplatelet clusters destabilize at comparatively lower DCE fractions (Supplementary Figs. 7 and 8). This behavior can be rationalized considering that the smaller nanoplatelet clusters are generated through fragmentation of the larger nanoplatelet clusters during denaturation. This indicates that the denaturation of large nanoplatelet clusters involves both disassembly into small nanoplatelet clusters and monomers. By fitting the denaturation data into the Matlab isqnonlin solver, we have obtained ΔG⁰, m and σ (Supplementary Section 3.2). The difference in the ΔG⁰ values for the two types of nanoplatelet clusters is <5 kJ/mol (Supplementary Table 1). The cooperative factor (σ) is slightly less than 1, indicating the weekly cooperative nature of the self-assembly process³⁹.

The formation of interesting self-assembled structures, such as 2D nanoplatelet clusters with switchable hierarchy by Eh-B encouraged us to use them as seeds for the synthesis of supramolecular heterostructures. We have previously reported that Eh-PDI forms supramolecular polymers in polar solvents such as DMSO, following a nucleation-elongation mechanism driven by dispersive interactions³⁹ Moreover, Eh-PDI also forms dormant monomers, which nucleate and elongate on the seeds of PDI with different side chains³². Since Eh-PDIs require only dispersive interactions for growth, we envisioned that they could be nucleated on the supramolecular assemblies (seeds) of other types of molecules, such as TTIs, despite having significant differences in their molecular structures and self-assembled morphology. At 15 µM, Eh-PDI exists as a monomer in DCE and forms supramolecular polymers in DMSO, as evidenced by absorption and emission spectra (Supplementary Fig. 9). At this concentration in DMSO, the monomeric state of Eh-PDI is obtained by heating the DMSO solution to 363 K (Supplementary Fig. 10a). We monitored the absorbance changes at 525 nm wavelength when the Eh-PDI solution was heated at a rate of 1 K/min. to 363 K and cooled back to 293 K. We observed that the heating curve and cooling curve do not exactly match, indicating the presence of a large hysteresis even at a slow cooling rate of 1 K/min. (Supplementary Fig. 10b). We have explored this broad thermal hysteresis to create the dormant monomeric state of Eh-PDI in DMSO. By rapidly cooling (~20 K/min) the monomeric state of Eh-PDI from 363 K to 298 K, we have obtained its dormant monomeric state, which is stable for more than an hour at 298 K. Further, the dormant monomers self-assemble into long fibers after ageing overnight at room temperature (Supplementary Fig. 11). Seeds are prepared by sonicating these 1D supramolecular assemblies. The UV-Vis absorption data of these prefabricated seeds match to that of supramolecular assemblies, indicating that sonication does not alter the aggregates (Supplementary Fig. 12a).

Homo-seeding was attained by adding prefabricated seeds to dormant monomers of Eh-PDI at various mol%, such as 10 mol%, 25 mol% and 50 mol% and their aggregation kinetics were monitored at 525 nm wavelength (Fig. 3 and Supplementary Fig. 12b). The average size of Eh-PDI seeds is around 3 to 4 µm (Fig. 3c). The decrease in the absorbance of Eh-PDI at 525 nm after addition of the seeds indicates a seed-induced living supramolecular assembly. The average length of fibers obtained is 5 µm, 20 µm and 50 µm for seed percentages of 50%, 25% and 10%, respectively (Fig. 3d,e,f). To further understand the type of nucleation involved during homo-seeding of Eh-PDI, we carried out living supramolecular polymerization at various monomer concentrations of 15 µM, 20 µM, 25 µM and 30 µM at a constant seed concentration of 10 µM⁴³. The kinetics of self-assembly were monitored at 525 nm wavelength after the addition of prefabricated seed. Here, α represents the extent of supramolecular assembly. A double logarithmic plot between concentration of dormant monomer and aggregation half-time (t₅₀) gives a scaling exponent value (γ value) of −0.918, (Fig. 3a), which corresponds to reaction order (n₁) of ~2. The self-assembly kinetics fit well into a seed-induced primary nucleation elongation model in Amylofit software with an insignificant mean residual error (Fig. 3b, Supplementary Section 3.1 and Supplementary Table 2)⁴³. Further, the aggregation kinetics show a poor fit into the seed-induced secondary nucleation elongation model, with a comparatively higher mean residual error, thus proving that the growth of Eh-PDI fibers via homo-seeding happens through a primary nucleation elongation process (Supplementary Fig. 13 and Supplementary Table 2).

Fig. 3. Homo-seeding of Eh-PDI.

a Log–log plot of half-times of self-assembly kinetics of homo-seeding of Eh-PDI versus initial concentration of Eh-PDI. The plot has a slope of −0.918, which is also known as the scaling exponent (γ). b Kinetic profiles (monitored at 525 nm) obtained by varying the concentration of dormant monomers of Eh-PDI. Here, α is the degree of assembly. Field emission scanning electron microscopy (FE-SEM) images of Eh-PDI (15 µM) obtained after adding different seed molar ratios, c 100% seed, d 50% seed, e 25% seed, and f 10% seed to the dormant monomers.

We attempted to use 2D platelet clusters of Eh-B as seeds to nucleate and elongate Eh-PDI dormant monomers (Fig. 4a). As discussed above, Eh-B self-assembles in DMSO into large and small 2D nanoplatelet clusters. Among these, first we have used small clusters as seeds (Fig. 2d). This hetero-seeding process is also monitored by probing the absorbance at 525 nm. The decrease in absorbance of Eh-PDI dormant monomers at 525 nm after seed addition is indicative of supramolecular polymerization of Eh-PDI. This is further confirmed by absorption spectra taken over 45 min. where the Eh-PDI absorption spectra resemble its self-assembled form (Supplementary Fig. 14). The appearance of large self-assembled structures in the solution after hetero-seeding indicates the formation of higher-order supramolecular assemblies (Fig. 4b). To understand the hetero-seeding mechanism more comprehensively, concentration-dependent kinetic experiments were performed. With increasing the concentration of Eh-PDI at the constant concentration of Eh-B seed (25 µM), the half-time decreased from 1698 sec. at 8.5 µM to 335 sec. at 18.5 µM. Log-log analysis of half-times versus monomer concentration yielded a scaling exponent (γ) of −2.177, consistent with a secondary nucleation growth mechanism (Fig. 4c and Supplementary Section 3.1)⁴³. The self-assembly kinetics showed excellent fit to the seeded secondary nucleation model in Amylofit online software, with negligible mean residual error (Fig. 4d and Supplementary Table 3). In contrast, primary nucleation models produced poor fits with high residual errors (Supplementary Fig. 15 and Supplementary Table 3).

Fig. 4. Synthesis and characterization of SHet-1.

a Schematic representation of the growth of Eh-PDI fibers on the surface of two-dimensional (2D) nanoplatelets of the small cluster seeds of Eh-B via secondary nucleation elongation process. b Photographs of solutions of dormant monomers of Eh-PDI (15 µM) before and after the addition of small 2D nanoplatelet clusters of Eh-B seeds (25 µM) and waiting for 45 min. c Log–log plot of half-times of self-assembly kinetics of hetero-seeding of Eh-PDI and Eh-B versus initial concentration of Eh-PDI. The plot has a slope of −2.177, which is also known as the scaling exponent (γ). d Kinetic profiles (monitored at 525 nm) obtained by varying the concentration of dormant monomers of Eh-PDI at a fixed concentration of small 2D nanoplatelets cluster seeds of Eh-B (25 µM) and their fit into the seed-induced secondary nucleation model of the amylofit online software. Here, α is the degree of assembly. Field emission scanning electron microscopy (FE-SEM) images and schematic illustrations showing the morphological evolution of fiber reinforced small 2D platelet clusters (SHet-1) obtained by introducing Eh-B seeds to the dormant monomers of Eh-PDI and spin coating the solution on a silicon wafer after e <1 min., f 10 min., g 30 min, and h 45 min. The white arrow in ‘e’ indicates the Eh-PDI fibers. Schematic representations were created using Blender 4.4.

Although this hetero-seeding follows a secondary nucleation-elongation process, there is no guarantee for the heterostructures' formation because the newly grown structures can detach from the seeds^28,37,38. Moreover, if heterostructures are formed, it is difficult to predict what the morphology of the resultant heterostructures would be and how they are formed. Hence, to answer these questions, we have performed time-dependent FE-SEM imaging of the heterostructures formed by 25 µM Eh-B seed and 15 µM of dormant monomers of Eh-PDI. After the addition of Eh-B seeds to the dormant monomers of Eh-PDI, in less than a min. thin fibers of Eh-PDI started appearing on the surface 2D nanoplatelets of Eh-B (Fig. 4e and Supplementary Fig. 16a). This clearly proves that growth starts on the surface of Eh-B seeds. After 1 min, we can see nascent fibers of Eh-PDI growing on 2D nanoplatelets of the small clusters of Eh-B (Supplementary Fig. 16b,c). These fibers grow on the surface of 2D nanoplatelets of Eh-B seeds, and the overall size of the heterostructure is nearly 3 µm. After nearly 10 min. of addition of seed, the heterostructures further develop in size to nearly 6 µm (Fig. 4f). After 30 min. of seed addition, we observe that the heterostructures are more evolved than earlier architectures, with more elongated and networked Eh-PDI fibers (Fig. 4g). The fibers originated from these heterostructure extend radially along all directions. Also, fibers do not originate from the ends of the seed but are spread across the surface of the seeds. After 45 min., the morphology demonstrates a more advanced interconnected network (SHet-1) morphologically somewhat similar to a biological neural network with multiple nodes⁴⁴ (Fig. 4h). In these systems, individual nodal centers give rise to highly distributed and networked fibers. This progression from simple radial to advanced, interconnected morphologies demonstrates the increasing complexity and hierarchical organization achievable through controlled supramolecular assembly. The above observation unambiguously proves that there is no detachment of newly formed fibers from the seeds. Since the fibers of Eh-PDI are nucleated and elongated on the surface of 2D platelets of the cluster, the resultant super-heterostructures also resemble fibers reinforced 2D platelet clusters. Further, we investigated morphological changes by reducing the ratio of Eh-PDI to 8.5 µM and 10 µM, at a constant seed concentration (25 µM) (Supplementary Fig. 17a,b). The reinforcement density of Eh-PDI fibers in the SHet-1 is reduced as its concentration is reduced.

Building on our earlier description that Eh-B smaller seeds evolve into larger 2D nanoplatelet clusters upon ageing, we hypothesized that these larger clusters could also serve as seeds to result in higher-order heterostructures via secondary nucleation (Fig. 5a). Accordingly, the aged seeds were introduced into the Eh-PDI dormant monomers, and the self-assembly process was monitored through changes in absorbance at 525 nm. The observed decrease in absorbance is very slight and did not saturate as rapidly as with smaller Eh-B seeds. This could be due to reduced surface availability and solution heterogeneity of larger 2D nanoplatelets cluster seeds (Fig. 2b,c). Hence, we could not obtain good kinetic profiles to fit the secondary nucleation model (Supplementary Fig. 18a). Despite this, approximately after 6 h, the solution resembles that of SHet-1, indicating the formation of heterostructures (Supplementary Fig. 18b), which was further investigated by FE-SEM imaging at different time intervals. Morphological analysis revealed that, within 0.5 h of seed addition, small thin Eh-PDI fibers began to interconnect individual 2D nanoplatelets across the large Eh-B cluster (Fig. 5b). After 3 h, the fibers became more pronounced and extended radially (Fig. 5c). At this stage, the fibers had elongated significantly and began to envelop the seed structures, leading to the formation of an interwoven network. After 6 h, the Eh-B seeds were fully reinforced and enveloped by Eh-PDI fibers to result in fibers reinforced large 2D platelet clusters (SHet-2) (Fig. 5d). The side view of FE-SEM image further shows the uniformity of growth of these Eh-PDI fibers over Eh-B surface (Fig. 5d). We also analyzed the FE-SEM images obtained by reducing the amount of Eh-PDI in the hetero-seeding. We observe that the coverage or reinforcement density of Eh-PDI fibers on Eh-B seeds reduces with a reduction in the concentration of Eh-PDI (Supplementary Fig. 17c,d).

Fig. 5. Synthesis and characterization of SHet-2.

a Schematic illustration of the growth of Eh-PDI fibers on the surface of two-dimensional (2D) nanoplatelets of large cluster seeds of Eh-B via secondary nucleation. Field emission scanning electron microscopy (FE-SEM) images and schematic illustrations showing the morphological evolution of fiber reinforced large 2D platelet clusters (SHet-2) obtained by introducing Eh-B large 2D platelet clusters seeds to the dormant monomers of Eh-PDI and spin coating the solution on a silicon wafer after (b) 0.5 h, (c) 3 h, and (d) 6 h. The white arrow in ‘b’ indicates the Eh-PDI fiber. Schematic representations were created using Blender 4.4.

The above results indicate a two-step process for the noncovalent synthesis of SHet-1 and SHet-2. Here, the Eh-B seeds and the dormant monomers of Eh-PDI are prepared separately and mixed together to result in SHet-1 and SHet-2. To understand the necessity of this protocol, we have performed a few control experiments. Firstly, we have mixed the pre-assembled Eh-PDI fibers and Eh-B 2D nanoplatelet clusters (small and large) in DMSO and analyzed their morphology using FE-SEM. FE-SEM images showed the presence of individually segregated assemblies of Eh-PDI fibers and Eh-B nanoplatelet clusters, and heterostructures were not observed (Supplementary Fig. 19). Next, we mixed Eh-PDI and Eh-B in their monomeric state in DCE and introduced them into DMSO to induce self-assembly. Here, the final solvent composition is 1 vol% of DCE in DMSO. In this case, FE-SEM images revealed predominantly individually segregated assemblies of Eh-PDI fibers and Eh-B nanoplatelet clusters, with a few heterostructures (Supplementary Fig. 20). These control experiments clearly demonstrate that mixing of Eh-B seeds and the dormant monomers of Eh-PDI is the best way to synthesize of SHet-1 and SHet-2.

Our results indicate that SHet-1 and SHet-2 show differences in their size and formation kinetics. To probe these structural differences on their physicochemical properties, we have evaluated the rheological properties to see how these differences will translate further. Under lower shear rates (<10 s^-1), SHet-2 showed significantly higher viscosity than SHet-1, whereas the viscosity of SHet-1 is slightly higher than that of solvent (DMSO) alone (Supplementary Fig. 21). At higher shear rates, both heterostructures thin out and match the DMSO viscosity, proving the networks fully break down to simple solvent flow, a signature of supramolecular assemblies⁴⁵. This study indicates that in addition to size and assembly kinetics, SHet-1 and SHet-2 also significantly differ in their rheological properties.

We further aimed to study the packing of molecules in the super-heterostructures formed by the assembly of Eh-B with Eh-PDI. For this purpose, we carried out PXRD (powder X-ray diffraction) of the thin films containing super-heterostructures and individual assemblies. The X-ray diffraction pattern of Eh-PDI fibers alone shows peaks with d-spacing at 2.14 nm, 1.07 nm, 0.94 nm, 0.92 nm and 0.31 nm. The appearance of peaks at d (2.14 nm) and d/2 (1.07 nm) indicates lamellar packing of Eh-PDI (Supplementary Fig. 22). In the case of the PXRD pattern of heterostructures formed from Eh-B and Eh-PDI mixture, the peaks corresponding to the lamellar packing of Eh-B are preserved (Supplementary Fig. 22). However, no separate peaks are observed for Eh-PDI. Moreover, a broad, low-intensity peak at 1.56 nm appeared in the heterostructures. Hence, the peak at 1.56 nm in SHet-1 could be due to a change in the molecular organization of Eh-PDI, as Eh-B diffraction peaks are retained (Supplementary Fig. 22). This is further evidenced by FE-SEM images, which show that Eh-PDI alone forms long, belt-like nanofibers (Supplementary Fig. 11, 19 and 20). But in the heterostructures, Eh-PDI exists as thin and short nanofibers, whereas the morphology of Eh-B is retained (Fig. 4e–h and Supplementary Figs. 16 and 17).

It is already established that individual Eh-PDI fibers melt and depolymerize into monomers above 333 K (Supplementary Fig. 10). However, Eh-B larger seeds do not fully depolymerize into monomers under these conditions, rather transform into smaller clusters (Fig. 2c,d and Supplementary Fig. 6). This observation led us to hypothesize that heating could be used to selectively dissociate Eh-PDI fibers in the SHet-2 and to study whether SHet-2 regenerates upon cooling back or transforms into SHet-1 due to the presence of smaller nanoplatelets (Fig. 6a). When the fully developed SHet-2 was heated to 363 K, the solution containing suspended particles now became noticeably more transparent, and the absorbance of Eh-PDI resembled its monomeric form (Supplementary Fig. 23). By spin-coating this hot solution on a silicon wafer, we have investigated the morphology using FE-SEM. The microscopy images confirmed that the Eh-PDI fibers had vanished entirely from the heterostructures (Fig. 6b). Furthermore, at this elevated temperature, the larger 2D platelet cluster seeds began to break apart into smaller units, as previously described. Rapid cooling of the solution to 298 K trapped the Eh-PDI in its dormant monomeric state. At the same time, the larger Eh-B seeds were converted into smaller and more dispersed 2D nanoplatelet clusters. The rapid formation of the heterostructures with morphology similar to SHet-1 further indicates that the presence of these smaller Eh-B seeds allowed the Eh-PDI dormant monomers to quickly nucleate on their surfaces (Fig. 6c). This transformation was also evident in the absorption spectrum of the rapidly cooled mixture (Supplementary Fig. 23). Hence, by taking advantage of the temperature-controlled switchable hierarchy of the Eh-B seeds and formation of Eh-PDI dormant monomers, we could successfully transform super-heterostructures, SHet-2 to SHet-1.

Fig. 6. Conversion of SHet-2 to SHet-1.

a Schematic illustration of the disintegration of SHet-2 at higher temperature into smaller two-dimensional (2D) nanoplatelets and the rapid nucleation of Eh-PDI on it, to generate SHet-1. b Field emission scanning electron microscopy (FE-SEM) image obtained by heating SHet-2 to 363 K, shows only Eh-B seeds and absence of any Eh-PDI fibers (inset: larger Eh-B seeds disintegrate into smaller Eh-B nanoplatelets upon heating to 363 K). c After rapidly cooling the heated solution and aging for 15 min., the obtained morphology is similar to SHet-1 due to the formation of small 2D nanoplatelet clusters of Eh-B seeds and rapid nucleation of Eh-PDI on them in situ. Schematic representations were created using Blender 4.4.

The formation of the heterostructures from distinct molecular π-systems encouraged us further to investigate their photophysical properties. We found that nanofibers of Eh-PDI in solid-state show fluorescence from 600 to 800 nm with an emission maximum of 630 nm (Supplementary Fig. 24). On the other hand, the self-assembled structures of Eh-B do not show any fluorescence. Recently, we showed that these types of NIR imides have good photothermal properties^42,46. Since light-to-heat conversion is a competing process for photoluminescence, molecules or materials with low fluorescence quantum yields, such as TTIs, can efficiently convert light into heat via non-radiative decay. The photothermal energy generated is further used for various applications such as photothermal therapy and steam generation, to name a few^42,46–48. Interestingly, we observed a substantial spectral overlap between the emission spectra of Eh-PDI nanofibers and the absorption spectra of Eh-B (Supplementary Figs. 24 and 25). This overlap suggests the potential for Förster resonance energy transfer (FRET) from Eh-PDI to Eh-B in the heterostructures. Such energy transfer is expected to further quench the emission intensity of Eh-PDI. In that case, the heterostructures will show negligible emission due to the non-emissive nature of the acceptor (Eh-B) assemblies, which is advantageous for generating photothermal energy. Indeed, we observed a reduced emission from Eh-PDI in heterostructure (SHet-1) compared to its individual self-assembled form (Fig. 7a). Although energy transfer typically results in enhanced emission from the acceptor, in this case, no such increase is observed due to the intrinsically non-emissive nature of the self-assembled structures of acceptors (Eh-B)⁴⁹. This quenching of Eh-PDI emission is also evident in the solid-state. Thin films comprising the heterostructures and individual self-assembled components were fabricated, and spectral analyses revealed a similar degree of spectral overlap between Eh-PDI emission and the absorption of Eh-B (Supplementary Fig. 24a). Notably, in the solid state, the emission quenching of Eh-PDI within the heterostructures is even more pronounced compared to that observed in solution, reinforcing the presence of efficient energy transfer processes in both solution and solid-state (Supplementary Fig. 24b).

Fig. 7. Photophysical and photothermal properties.

a Emission spectra (corrected for absorption) of self-assembled structures of Eh-PDI, Eh-B and SHet-1 in DMSO (λ_ex = 580 nm). b Photothermal heat generation of blank (quartz substrate) and thin films of self-assembled structures of Eh-PDI, Eh-B, and SHet-1, after illuminating with 1 Sun of simulated sunlight for 10 min. c Bar graph showing heat generation by individual self-assembled structures of Eh-PDI, Eh-B, and their heterostructure (SHet-1).

The quenching of Eh-PDI solid-state emission in the heterostructures encouraged us to investigate the photothermal properties of individual molecular assemblies and their corresponding heterostructure (SHet-1) using simulated sunlight irradiation. For this purpose, Eh-PDI, Eh-B and SHet-1 were drop-casted onto a quartz substrate having nearly equal photon absorption (Supplementary Fig. 26) for photothermal measurements, and the heat generated under 1 Sun simulated sunlight was monitored using an infrared thermal camera (See Methods). The initial photothermal temperature increase of blank quartz was measured to be 37.3 °C, after 10 min. of 1 Sun irradiation (1 kW m^-2) (Fig. 7b, c). Under similar conditions, Eh-PDI exhibited heat generation of 51.8 °C due to its lack of NIR absorption and its partial fluorescence emission in the solid state, as discussed earlier. Eh-B alone produced heat up to 58.6 °C under similar conditions. SHet-1 exhibited temperature reaching values around 60.8°C, which is higher than Eh-B. Here, Eh-B is a good candidate for light-to-heat conversion owing to its strong NIR absorption and negligible solid-state quantum yield. Interestingly, the SHet-1 exhibited more photothermal heat than pure Eh-B, despite having a ~ 37% lower molar quantity of Eh-B in the SHet-1 (Fig. 7). This is due to improved non-radiative decay of Eh-PDI in the heterostructures. As a result, heterostructures generated more photothermal heat than individual molecular assemblies.

In conclusion, we have shown a rational approach for the non-covalent synthesis of fully supramolecular higher-order super-heterostructures using structurally distinct molecular π-systems. Our results indicate that the dormant monomers of Eh-PDI efficiently nucleate and elongate on the supramolecular assemblies of Eh-B to result in super-heterostructures, namely, fibers reinforced by small and large clusters of 2D nanoplatelets (SHet-1 and SHet-2). This process is guided by secondary nucleation and dispersive interactions. Moreover, using a temperature-dependent switchable hierarchy of the seeds, we have achieved in situ conversion of SHet-2 into SHet-1. Due to the significant overlap of emission of Eh-PDI with absorbance of Eh-B, the emission of Eh-PDI assemblies is further quenched in the heterostructures, contributing to the enhanced light-to-heat conversion. Our results further indicate that by creating dormant monomers from molecules where the dispersive interactions drive the self-assembly, it is possible to synthesize diverse heterostructures with topologies beyond 1D and 2D. Notably, despite significant differences in the molecular structures and self-assembled morphologies of Eh-B and Eh-PDI, the formation of heterostructures highlights the robustness of our approach. We believe that the design principles and supramolecular heterostructures described here can expand the toolbox for noncovalent synthesis and provide guidelines for synthesizing new super-heterostructures with tailored functions.

Methods

Eh-B self-assembly and formation of 2D nanoplatelet clusters

The 1 mM stock solution of Eh-B is prepared in DCE by taking the required amount of Eh-B and heating the solution to 60 °C, and cooled back to 25 °C. From the stock solution of Eh-B in DCE (1 mM), 75 µL was taken in a vial, and DCE was evaporated. To this, 0.5 mL DMSO was added and sonicated; later, this solution was heated to 90 °C for 2 min. and allowed to cool for 5 min. at 25 °C (at this stage the concentration is 300 µM). The obtained solution is diluted in DMSO, such that the final concentration is 25 µM. This 25 µM solution was immediately spin-coated on a silicon wafer and dried under vacuum at room temperature. FE-SEM analysis revealed the formation of smaller 2D nanoplatelet clusters.

Further, upon aging the 300 µM DMSO solution for 1 h, larger aggregates are seen. These are further diluted by adding DMSO such that the final concentration is 25 µM. When this solution was drop-casted on a silicon wafer and dried under vacuum at room temperature, FE-SEM analysis revealed the formation of larger 2D nanoplatelet clusters.

Denaturation studies

The smaller and larger 2D nanoplatelet clusters were prepared. For this purpose, from the stock solution of 1 mM in DCE, 75 µL was taken in a vial, and DCE was evaporated. Subsequently, 0.5 mL of DMSO was added, and the mixture was sonicated. The resulting solution was heated to 90 °C for 2 min. and then cooled to 25 °C for 5 min. (concentration=300 µM). When this DMSO solution was aged for 1 h prior to dilution, larger aggregates were obtained. For the denaturation studies, the required amount of DMSO was added to the solutions, followed by the addition of DCE to make the final volume 3 mL and final concentration 25 µM. Absorbance spectra were measured for each of the solutions, and aggregation was monitored at 784 nm. The denaturation experiments were carried out for three different concentrations of 10 µM, 25 µM and 40 µM and the data were fit into MATLAB isqnonlin solver to obtain thermodynamic parameters.

Preparation of the dormant monomer state of Eh-PDI

From the stock solution of Eh-PDI in DCE (1 mM), 45 µL was transferred to a vial. The solution containing DCE was evaporated, and 3 mL of DMSO was added and sonicated well. This 15 µM solution of Eh-PDI DMSO solution was heated up to 363 K and cooled rapidly (~ 20 K/min) to 298 K to achieve the dormant monomer state. This state was stable for more than 1 h.

Homo-seeding of Eh-PDI

For homo-seeding experiments, prefabricated seeds from the SPs of Eh-PDI were prepared by allowing the dormant monomer state overnight at 25 °C to form long fibers. This solution was sonicated for 30 min. at 303 K to attain small seeds. To attain length control of Eh-PDI, these prefabricated seeds were added in different mol% to the dormant monomer state of Eh-PDI. For kinetic analysis of homo-seeding, seed concentration was kept constant (10 µM), and Eh-PDI dormant monomer concentration was varied (15 µM, 20 µM, 25 µM and 30 µM). The kinetics were monitored at 525 nm.

Formation of SHet-1 and SHet-2

To a dormant monomer solution of Eh-PDI, small 2D nanoplatelet clusters of Eh-B seed solution were added such that the final concentration of Eh-PDI is 15 µM and Eh-B is 25 µM. After the addition of the seed, the samples were kept at 25 °C to facilitate the hetero-seeding. FE-SEM imaging was carried out on these samples at different time intervals. For kinetic analysis, concentration of Eh-B seed was kept constant (25 µM), and Eh-PDI was varied (8.5 µM, 10 µM, 12.5 µM, 15 µM and 18.5 µM).

For synthesizing SHet-2, large 2D nanoplatelet clusters of Eh-B were added to the dormant monomers of Eh-PDI, such that in the final solution, the concentration of Eh-B is 25 µM and Eh-PDI is 15 µM. At different time intervals, samples were spin-coated on a silicon wafer and dried under vacuum at room temperature. Further, different concentrations of dormant monomer (10 µM, 15 µM, 20 µM and 25 µM) state were used while keeping Eh-B seed concentration constant (25 µM) to analyze the kinetics.

Photothermal measurements

Solutions of Eh-PDI, Eh-B, and SHet-1 in DMSO were drop-casted onto a quartz substrate and dried under vacuum at room temperature to ensure complete solvent evaporation. The films were prepared such that their photon absorption in the 400–1000 nm range was nearly the same in all samples. Each of these films was illuminated by simulated sunlight of 1 Sun, and their photothermal heat generation was monitored using an infrared thermal camera.

Author contributions

K.V.R. conceived the idea. S.R.P. performed all the experiments. G.N.K. performed FE-SEM analysis. All the authors discussed the results and contributed to the manuscript writing. The manuscript was written through the contributions of all authors. All authors have given approval to the final version of the manuscript.

Peer review

Peer review information

Nature Communications thanks Yanke Che and the other, anonymous, reviewers for their contribution to the peer review of this work. A peer review file is available.

Data availability

All data are available from the corresponding author upon request. The UV-Vis absorbance and photothermal heat generation data generated in this study are provided in the Supplementary Information. Source data [10.5281/zenodo.19109629] are provided in this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-72700-5.