Non-macrocyclic molecular design enables cavity-adaptive cocrystals with high elasticity and low-threshold lasing

Abstract

The construction of cavity-adaptive cocrystals (CACs) based on linear non-macrocyclic π-conjugated hosts remains challenging due to their inherently weak and less predictable binding capabilities compared to conventional macrocyclic hosts. Here we overcome this limitation through the rational design and synthesis of CACs employing a structurally engineered non-macrocyclic host. This host combines a linear π-conjugated backbone with sterically hindered groups, allowing dynamic cavity modulation while maintaining its luminescent characteristics upon solvent and guest encapsulation. Heteroatom-modified guest molecules of similar size form molecular cavities with comparable dimensions yet exhibit different luminescence characteristics. Further tailoring the host’s π-backbone yields color-tailorable CACs. These systems exhibit relatively strong intralayer C − H···π interactions coupled with weak interlayer forces, leading to high elasticity and low-threshold lasing action in single crystals and microcrystals. This work establishes a non-macrocyclic host/guest platform with adaptive cavities, demonstrating its potential as flexible laser emitters.

Cavity-adaptive crystals have useful phototunable properties, but can be challenging to prepare. Here, the authors report the development of such cocrystals by use of a non-macrocyclic host, with elasticity and lasing action.

Introduction

Supramolecular host-guest complexes have garnered significant interest due to their diverse applications in drug carriers, gas separation, molecular machines, and optics^1–5. The field has evolved significantly since Pedersen’s pioneering discovery of crown ethers in the 1960s⁶, with subsequent development of diverse macrocyclic hosts including cyclodextrins⁷, cucurbiturils⁸, and pillararenes^9–11. These systems demonstrate inherent molecular recognition capabilities through their preorganized molecular cavities. Among these, cyclophanes represent a particularly important class of macrocyclic receptors^12,13. Notable examples include Stoddart’s positively charged ExBox⁴⁺ (Fig. 1a), which exhibits exceptional binding affinity for polycyclic aromatic hydrocarbons (PAHs)¹⁴, and Würthner’s neutral perylene diimide-based cyclophanes¹⁵. The latter was further extended to larger terrylene bisimide (TBI) systems with enhanced PAH binding capabilities¹⁶. While traditional host-guest chemistry has primarily focused on solution-phase guest binding and cavity modulation, with relatively few studies investigating solid-state photofunctional properties. Particularly, pillararene- and cyclophane-based complexes have demonstrated unusual phenomena such as solid-state lasing¹⁷ and thermally activated delayed fluorescence (TADF)^18,19, due to the profound interplay between hosts and aromatic guests. The confined host cavity environment enables fluorescence enhancement and/or stabilization of triplet excitons, allowing precise modulation of excited-state optical properties.

Fig. 1. Representative macrocyclic cyclophane hosts and sterically engineered non-macrocyclic hosts, as well as PAH guests.

a, b Schematic showing para-xylylene spaced viologen- and perylene diimide-based cyclophane hosts and a non-macrocyclic host modified with a planar C₆₄ core and four imide substituents (a) as well as the chemical structure of a potential non-macrocyclic host (1) modified with a linear DSB backbone and two terminal steric groups (b). Planar and linear guests are also involved.

Beyond these well-established structural frameworks in host-guest chemistry, recent studies have revealed exceptions where sterically hindered aromatic compounds featuring a central planar π-core and symmetric bulky terminal groups/ions act as non-macrocyclic hosts^20,21. Through mediation by charge-transfer (CT) and/or hydrogen bonding interactions, these non-macrocyclic compounds can effectively encapsulate solvent molecules or small π-conjugated guests, leading to the formation of cavity-adaptive cocrystals (CACs). This approach has enabled advanced solid-state photofunctional properties, including tunable emissions²⁰ and near-infrared room-temperature phosphorescence (RTP)²¹. Notably, Würthner’s group employed a C₆₄ nanographene derivative with four imide substituents to construct CACs with PAHs and metal-organic compounds (e.g., coronene and platinum porphyrin), yielding multilayer nanographene assemblies^22,23. Inspired by these discoveries, we propose a sterically hindered distyrylbenzene (DSB) derivative^24–28 as an effective host for linear and planar PAHs. Our strategy leverages the weak C−H···π interactions to form photofunctional CACs, capitalizing on DSB’s large molecular dimensions and high fluorescence efficiency. This system holds promise as a versatile platform for the rational construction of CACs toward high-efficiency solid-state lasing and the unlocking of additional physicochemical properties.

Herein, we developed a linear DSB derivative decorated with terminal steric groups as a potential host for constructing CACs. Such a non-macrocyclic DSB-based compound dynamically encapsulates both solvent molecules and PAHs, forming host/guest microcrystals and single crystals with tunable molecular cavity sizes, i.e., CACs. The host binds smaller and larger guests in distinct 1:2 and 1:1 stoichiometric ratios through C−H···π and/or C−H···O interactions, while adjacent host molecules are stabilized by weak van der Waals forces between terminal groups. The resulting host/guest microcrystals maintain high-efficiency photoluminescence with only slight spectral shifts compared to the pristine host. Introducing heteroatoms into guests would not affect the cavity structures but would result in different luminescence behaviors. The emission colors of these CACs can be tailored by modifying the host’s central π-backbones. The synergistic intralayer and interlayer interactions endow the CAC single crystals with high elasticity, while high crystallinity and photoluminescence quantum yields (PLQYs) allow low-threshold lasering in microcrystals, even in elastic single crystals. This research provides deep insight into the development of CACs with flexible photonic properties.

Results and discussion

Structural design of DSB-derived host 1 and host/guest microcrystal characterization

We chose linear 4,4’-bis[4-(di-p-tolylamino)styryl]biphenyl (1, Fig. 1b) as a host for functional CACs based on two key advantages: (1) Its biphenyl-bridged DSB backbone with bulky p-tolyl terminals may potentially allow the accommodation of diverse PAH guests through extended π-conjugation length and confinement effect. (2) Such a type of π-backbones generally exhibits high PL efficiencies and crystallinity due to weak intermolecular C−H···π forces^29,30. We initially investigated the crystalline assemblies of 1 obtained from different solvent mixtures (Fig. 2a, see Supplementary Table 1 for details). Slow evaporation of 1 from a tetrahydrofuran/ethanol mixture (THF/EtOH; v/v = 1:1, C₁ = 2.5 mM) yielded well-defined cyan-emitting microribbons (MRs), as verified by fluorescence microscopy and SEM results (FM/SEM, Fig. 2b, c). These MRs have hundreds of micrometers in length and tens of micrometers in width. Instead of MRs, slow evaporation of 1 from 1,4-dioxane/n-propanol (Dio/PA) and toluene/isopropanol (Tol/IPA) gave green-emitting rectangle and parallelogram microplates (MPs), respectively. After examining more solvent mixtures, we found that the luminescence color of the microcrystals is governed by the type of good solvent used in the preparation. The use of non-aromatic solvents generally yields cyan-emitting microcrystals, with the sole exception of Dio, which enables green-emitting microcrystals. Conversely, when aromatic solvents (e.g., chlorobenzene (CB) and p-xylene (PX)) were used, green-emitting MPs were consistently produced (Supplementary Figs. 1−3). These distinctly different assembled morphologies indicate the effect of solvents on the crystal structures of 1.

Fig. 2. 1-Derived assemblies formed from diverse solvent mixtures and corresponding molecular packing.

a Schematic diagram showing the solvent-mediated assembly of host 1 (THF/EtOH, Dio/PA, and Tol/IPA) and corresponding packing arrangements. b, c FM (b) and SEM(c) images of 1, 1·Dio, and 1·Tol microcrystals. All samples were excited by UV light. Scale bar, 20 μm.

Centimeter-sized crystals of 1 suitable for single-crystal X-ray diffraction (SCXRD) analysis were grown from THF/EtOH, Dio/PA, and Tol/IPA, respectively. Crystallographic analysis revealed that both Dio and Tol solvent molecules are encapsulated within the flexible linear cavities of 1 in a 2:1 stoichiometry (denoted as 1·Dio and 1·Tol), whereas THF molecules are excluded (Fig. 2a and Supplementary Table 2 and Fig. 4). In its solvent-free form, the linear π-skeleton of 1 adopts a twisted configuration, stabilized by multiple C − H···π interactions (Supplementary Fig. 5). In contrast, the π-skeleton of host 1 in 1·Tol and 1·Dio undergoes planarization to facilitate solvent encapsulation, clearly confirming the formation of CACs. Although THF is structurally and dimensionally similar to both Dio and Tol, it features only a single ether oxygen atom. In contrast, Dio contains two such oxygen atoms, and Tol possesses a conjugated benzene ring. Crystallographic analysis reveals that the 1·Tol is stabilized primarily by weak C–H···π interactions between host 1 and Tol. In comparison, 1·Dio exhibits additional dual and cooperative C–H···O interactions, which further enhance its structural stability. Unlike Dio or Tol, THF lacks the capacity to provide sufficient interaction sites or binding driving forces, leading to its significantly weaker binding affinity for host 1. Powder X-ray diffraction (PXRD) patterns of these three assemblies also verify the presence or absence of solvent inclusion (Supplementary Fig. 6). These results imply that solvent inclusion within the host framework is a prerequisite for green emission, a process that occurs efficiently with aromatic solvents but is largely impeded with non-aromatic ones. Notably, host 1 exhibits no specific selectivity for the inclusion of non-aromatic Dio molecules, in stark contrast to its preferential selectivity for benzene-derived solvents. Morphology predictions (Supplementary Fig. 7) reveal that the top/bottom planes of 1, 1·Dio, and 1·Tol microcrystals are predominantly stabilized by van der Waals forces between terminal steric groups, with preferential growth along the [100] directions. The single-crystalline nature of these assemblies was further corroborated by polarized optical microscopy (POM), TEM image, and selected-area electron diffraction (SAED) pattern (Supplementary Figs. 8 and 9).

The absence of face-to-face π-π interactions in compound 1 results in a modest PL red-shift (~15 nm) of its microcrystals (1 MRs) relative to the monomeric solution, along with a solid-state photoluminescence quantum yield (PLQY) of 36.30% (Supplementary Fig. 6 and Table 3)³¹. Apparently, aromatic solvent Tol has a larger molecular size compared to Dio, thus leading to a more expansive cavity space in 1·Tol. While both 1·Tol and 1·Dio microcrystals involve solvent encapsulation, the extent of red shift in their PL peaks differs slightly relative to pure host MRs. Moreover, PLQYs of 1·Tol and 1·Dio MPs are both significantly enhanced, accompanied by reduced PL decay lifetimes compared to the host 1 (Supplementary Fig. 10 and Table 3). The enhancement in luminescence efficiency is presumably due to the solvent inclusion-induced planarization of the π-backbone in host 1, which suppresses molecular vibrations and rotations, thus minimizing non-radiative decay pathways³². Specifically, 1·Tol MPs achieve an exceptionally high PLQY of 58.10% with a slightly short lifetime of 1.02 ns. Unfortunately, however, 1·Tol and 1·Dio MPs readily decompose within less than a day (Supplementary Fig. 11), due to solvent desorption. Driven by the phenomenon of solvent encapsulation, particularly the selective inclusion effect of aromatic solvents, we conceived that large linear π-conjugated compounds could serve as alternative guests to construct robust CACs.

Synthesis of cavity-expandable host/guest microcrystals via complexation of host 1 and PAH guests

Building upon the above strategy, we chose four π-conjugated molecules with variable linear dimensions involving styrene (ST), trans-stilbene (SB), DSB, and trans, trans-distyrylbiphenyl (BPDSB) as guest components (Fig. 3a and Supplementary Fig. 12). This selection enables a comprehensive evaluation of the correlation between cavity dimensions and guest molecular length. For comparative analysis, we initially synthesized ribbon- or plate-like microcrystals of the four guests, which exhibited emissions ranging from blue-purple to cyan (Supplementary Tables 4, 5 and Fig. 13). Subsequently, we performed equimolar cocrystallization of host 1 with each linear guest molecule from precursor solutions (Supplementary Table 6). FM/SEM observations (Fig. 3b, c) reveal that the resulting co-assemblies appear as similar ribbon- or plate-like microcrystals, but with green or cyan emission instead.

Fig. 3. Molecular design, structural analysis, and spectroscopic characterization of CACs formed between 1 and diverse PAHs.

a Chemical structures of representative linear guests (ST, SB, DSB, and BPDSB) and schematic illustration of 1-derived CAC formation. b, c FM/SEM images of the resulting binary microcrystals. All samples were excited by UV light. Scale bar, 50 μm. d Diffuse reflectance absorption (dashed curves) and PL spectra (solid curves) of 1 (cyan curves), 1·ST (orange curves), 1·SB (purple curves), 1·DSB^a (pink curves), and 1·BPDSB (blue curves) microcrystals. e Schematic illustration showing the inclusion of SB and BPDSB molecules within 1. f, g Packing motifs of 1·SB, 1·ST, 1·DSB^a, and 1·BPDSB from front view (f) and side view (g).

Crystallographic information demonstrates that host 1 maintains its dynamic binding capacity for all four guest molecules, forming the cocrystals termed as 1·ST, 1·SB, 1·DSB^a, and 1·BPDSB (Fig. 3f, g and Supplementary Tables 7, 8). Notably, structural analysis uncovers distinct binding stoichiometries: each cavity of host 1 incorporates two ST molecules arranged with inversion symmetry, while SB, DSB, and BPDSB molecules exhibit 1:1 host/guest binding. In the 1·ST system, ST serves as both the solvent and the guest. The resulting cavity sizes follow the sequence: V_1·BPDSB ˃ V_1·DSB^a ˃ V_1·ST ˃ V_1·SB. Although longer guest molecules typically yield larger cavities, the larger cavity size in 1·ST compared to 1·SB arises from the accommodation of two shorter ST molecules per molecular cavity. Similar to 1·Dio and 1·Tol, the linear π-core of host 1 in all four cocrystals undergoes a conformational change and becomes planar upon complexation (Supplementary Fig. 14), thereby creating optimally sized cavities for guest accommodation. Molecular packing analysis reveals that these CACs are mainly stabilized by C−H···π interactions between 1 and each guest. Notably, the cavity size of 1 reaches its maximum when complexed with BPDSB, matching the fully extended π-core of host 1 (Fig. 3e). Unlike macrocyclic hosts with fixed intrinsic cavities^8,33 or porous frameworks with predetermined voids^34–36, host 1 presents exceptional cavity adaptability when complexation with guests of varying lengths, enabling the formation of guest-induced CACs.

Among the four host/guest microcrystals (1·ST, 1·SB, 1·DSB^a, and 1·BPDSB) examined, all except 1·BPDSB utilized pure Dio or a THF/Dio mixture (v/v = 3:2) during liquid-phase preparation. To exclude any contribution of Dio to the cavity dimensions of these CACs, we intentionally replaced this mixed solvent system. Notably, the use of pure THF as the good solvent for preparing microcrystals of 1·SB and 1·DSB^a still affords similar assembled morphologies and luminescence colors (Supplementary Fig. 15). PXRD analysis confirmed that the host/guest microcrystals, whether prepared from pure THF or a THF/Dio mixture, possess identical crystal structures. These findings imply a competitive interplay between linear guest molecules (e.g., SB and DSB) and good solvents (e.g., Dio and Tol) during cocrystal growth. The stronger intermolecular interactions of the linear guests with host 1, however, dominate and effectively preclude the incorporation of solvent molecules into the host framework.

PXRD patterns of the four host/guest assemblies match well with their SCXRD patterns and the top/bottom crystal planes of all microcrystals are enclosed by (001) planes (Supplementary Fig. 16), as confirmed by predicted growth morphologies (Supplementary Fig. 17). POM images reveal their single-crystalline nature (Supplementary Fig. 18). In comparison to host 1 MRs, the PL peaks of 1·Tol, 1·SB, and 1·DSB^a microcrystals show a slight red shift, while 1·Dio and 1·ST microcrystals exhibit a more pronounced red shift (Fig. 3d; Supplementary Figs. 6 and 19). Based on the distinct stacking modes of host molecules within these CAC systems, we infer that host–host interactions are stronger in 1·Dio and 1·ST, whereas they are relatively weaker in 1·Tol, 1·SB, and 1·DSB^a (Fig. 3f and Supplementary Fig. 4). Absorption spectra of these four microcrystals (1·ST, 1·SB, 1·DSB^a, and 1·BPDSB) closely resemble those of 1 MRs, indicating that host 1 dominates their optical properties. Compared to the pure 1 MRs, the PL lifetimes of the four microcrystals (Supplementary Fig. 20) either increase (1·ST and 1·SB) or decrease (1·DSB^a and 1·BPDSB). While all four host/guest assemblies exhibit markedly enhanced PLQYs in the range of 62.85–95.36%, significantly exceeding the value of 36.30% observed for the pristine host 1 (Supplementary Table 9 and Fig. 21). By analogy with the solvated crystals of 1, we attribute this enhancement to the rigidification of the host’s π-system following PAH encapsulation, which suppresses non-radiative transition pathways and consequently boosts PL efficiency³². Notably, these CACs maintain structural integrity for six months in dark conditions, possibly due to their extended π-interaction surfaces compared to 1·Dio and 1·Tol.

Further decreasing the molar ratio of 1 to DSB in a Dio/EtOH mixture (e.g., 1:2) led to green-emitting MPs instead of cyan-emitting MRs, indicating the formation of another cocrystal (Supplementary Fig. 22). Crystallographic information confirms a cocrystal comprising 1 and DSB in a stoichiometric ratio of 2:3 (denoted as 1·DSB^b), in which DSB molecules are selectively trapped within two distinct cavities of host 1. One molecular cavity accommodates only a single DSB molecule, while the other can simultaneously encapsulate two DSB molecules. These two types of cavities are arranged in an alternating fashion along the b axis, while the DSB molecules in both cavities adopt edge-to-face stacking along the a axis (Supplementary Fig. 23). The entire structure is stabilized by C − H···π interactions between the host framework and the enclosed DSB molecules. Morphological characterization reveals that 1·DSB^b MPs preferentially grow along the [100] direction (Supplementary Figs. 24 and 25). The green emission at 505 nm for 1·DSB^b MPs exhibits a slight red-shift compared to 1·DSB^a MRs (Supplementary Fig. 26), probably due to enhanced intermolecular interactions.

To further investigate the effects of heteroatoms (e.g., N and S) on the cavities of host 1, we examined three linear PAH guests with comparable lengths: p-quaterphenyl (4P), 4,4-di(4-pyridyl)biphenyl (4PPy), and α-quarterthiophene (4S) (Fig. 4a and Supplementary Fig. 27). Host 1 still exhibits guest-dependent cavity adaptability, as evidenced by equimolar guest binding and nearly identical cavity sizes (Supplementary Fig. 28). Likewise, intermolecular C − H···π interactions act as a dominant force (Supplementary Fig. 29), which may endow these cocrystals with high PLQYs. Comparatively, 1·4P MRs exhibit cyan emission while both 1·4PPy and 1·4S MRs display yellow-green emissions (Fig. 4b and Supplementary Table 10, Fig. 30). Notably, the main PL peak of 1·4PPy MRs shows a 40 nm red-shift compared to 1·4P (Fig. 4c and Supplementary Fig. 31), probably due to additional CT interactions between the aniline groups of 1 and the pyridyl moieties of 4PPy. To verify this conclusion, we further conducted absorption, PL lifetime, and electron paramagnetic resonance (EPR) measurements of 1·4PPy MRs. The diffuse reflectance absorption spectrum displays a new long-wavelength absorption band that is absent in either pure host 1 or guest 4PPy. Compared to host 1 (τ = 1.31 ns), the 1·4PPy MRs shows a prolonged PL lifetime of 2.26 ns. Additionally, an EPR signal with a g-value of 2.00526 (close to free electron) unambiguously indicates the presence of unpaired electrons and radical species in 1·4PPy, providing direct proof of a CT process³⁷. The above results undoubtedly confirm the presence of weak CT interactions between host 1 and the guest 4PPy (Supplementary Fig. 32). While the yellow-green emission of 1·4S MRs likely arises from sulfur-rich 4S. POM data and morphology simulations indicate that the microcrystals exhibit single-crystalline characteristics with preferential growth along the [100] direction (Supplementary Figs. 33 and 34). While PXRD patterns and crystallographic information verify their nearly identical crystal structures despite heteroatom modifications (Fig. 4d; Supplementary Table 11 and Fig. 31). In contrast to 1·4PPy MRs, both 1·4P and 1·4S MRs exhibit similarly short fluorescence lifetimes (~1.2 ns, Supplementary Fig. 35).

Fig. 4. Structural, morphological, and spectroscopic characterization of 1·4P, 1·4PPy, and 1·4S microcrystals.

a Chemical structures of 4P, 4PPy, and 4S as well as packing motifs of 1·4P, 1·4PPy, and 1·4S. b FM images of single MRs of 1·4P, 1·4PPy, and 1·4S. Scale bar, 50 μm. c, d Solid-state (c) diffuse reflectance absorption (dashed curves) and PL spectra (solid curves) as well as (d) PXRD patterns of 1·4P (purple curves), 1·4PPy (blue curves), and 1·4S (pink curves).

Notably, PL lifetimes of 1-derived host/guest microcrystals either increase or decrease relative to that of pure host 1, with most values falling within the range of 1–2 ns (Supplementary Fig. 36). We attribute this variation to the synergistic interaction between the host and guest molecules, which collectively modulates the competition between radiative and non‑radiative transition pathways in the CACs³⁸. Specifically, guest encapsulation promotes planarization of the host molecular skeleton, enhancing host-guest interactions and thereby facilitating the radiative decay process. Concurrently, this synergistic effect may also suppress non‑radiative pathways present in the pure host crystal by reducing lattice defects and restricting molecular vibrations and rotations. The short, nanosecond-scale PL lifetime promotes rapid radiative recombination, minimizing energy loss due to triplet accumulation and thereby establishing a favorable photophysical basis for low-threshold lasing. Furthermore, the host framework demonstrates versatile guest-binding capability, accommodating not only linear molecules but also planar π-conjugated systems such as anthracene (An), tetracene (Tc) and their halogenated derivatives (Supplementary Figs. 37 and 38).

Theoretical calculations of 1-derived CACs

The excited-state properties of the CAC systems were investigated using time-dependent density functional theory (TD-DFT). An analysis of the transition orbitals corresponding to the maximum oscillator strength in the absorption spectrum (Supplementary Table 12) shows that most cocrystals (e.g., 1·DSB^a, and 1·4P) exhibit electron and hole densities primarily localized on host 1 (Supplementary Figs. 39 and 40). This indicates that the initial photoexcitation takes place within the host material, dominating the subsequent electronic transitions and emission. However, in rare cases like 1·4S, the PL spectrum exhibits the characteristic emission of the 4S guest molecule, which appears to contradict the host-dominated excitation. We propose that this is due to a highly efficient Förster resonance energy transfer (FRET) from the host 1 to the 4S guest, facilitated by the significant spectral overlap between the host’s emission and the guest’s absorption. Under periodic boundary conditions, TD-DFT calculations reveal that the key electronic transition associated with the low-energy absorption band in 1·4PPy is characterized by a shift of electron density from the π-conjugated backbone of host 1 to the pyridine ring of guest 4PPy. The spatial separation of this CT pathway is visually confirmed by the plotted hole and electron density distributions (Supplementary Fig. 40b). Such distinct separation of the frontier orbitals is a hallmark of CT excitation. Additionally, energy decomposition analysis (EDA) using the sobEDA method revealed that both electrostatic and dispersion interactions play crucial roles in stabilizing the CAC systems (Supplementary Fig. 41). To identify the dominant structural origins of the dispersion contributions, we performed a systematic sobEDA analysis on a representative 1:1 host/guest structure, strategically fragmenting host 1 into two components: terminal substituents and π-backbone (Supplementary Figs. 42 and 43). The results demonstrate that dispersion interactions primarily occur between the host’s π-backbone and guest molecule (π-surface interactions), with negligible contributions from terminal substituents. This conclusively establishes adjacent π-surface interactions as the key stabilizing factor in these CACs.

Synthesis of CACs derived from variable host π-backbones

The linear π-backbone of host 1 plays a pivotal role in determining the cavity structures and luminescent performances of the resulting CACs. To systematically explore the influence of π-backbones, we strategically devised and synthesized additional host molecules: N⁴,N⁴,N^4”‘,N^4”‘-tetra-p-tolyl-[1,1^’:4^’,1^”:4^”,1^”‘-quaterphenyl]−4,4^”‘-diamine and 4,4’-(1,4-phenylenedivinyl)bis[N,N-di(p-tolyl)aniline] (2 and 3 for short, respectively; Fig. 5a), both possessing linear π-backbones. Similar to their solution-state behavior, the 1D crystalline assemblies of 2 and 3 display blue and cyan emissions, respectively (Fig. 5c; Supplementary Table 14 and Figs. 44−46), resulting from the differences in π-backbones. Crystallographic information reveals that molecular geometries and packing of 2 and 3 are analogous to those of 1 (Fig. 5b; Supplementary Table 15 and Figs. 47, 48), suggesting they share similar guest encapsulation capabilities.

Fig. 5. Packing, assembled morphologies, and luminescence colors of hosts 2, 3, and 4 as well as their CACs.

a Chemical structures of (a) hosts 2, 3, and 4. b Packing motifs of hosts 2, 3, and 4 when viewed down the c axis, along with PL spectra of their assemblies (2, blue curve; 3, green curve; 4, orange curve). c FM images of 2, 2·4P, 3, 3·DSB, 4, and 4·DPA microcrystals. All samples were excited by UV light. Scale bar, 10 μm. d Packing motifs of 2·4P, 3·DSB, and 4·DPA as well as their respective PL spectra (2·4P, bule curve; 3·DSB, green curve; 4·DPA, red curve).

As anticipated, host 2 forms well-defined molecular cavities that enable 1:1 stoichiometric binding with 4P during cocrystallization (Fig. 5c, d; Supplementary Table 16 and Figs. 49, 50). Intriguingly, however, the π-backbone of host 2 exhibits reduced planarity upon guest complexation compared to host 1, presumedly due to weakened C − H···π interactions in 2·4P. Within these cavities, the 4P guest molecules adopt a nearly parallel arrangement, creating 1D molecular channels. Similarly, host 3 bearing a DSB-based π-backbone gave analogous cavity and crystal structures when complexed with DSB (3·DSB), while retaining significant backbone distortion (Fig. 5a, b, d). 2·4P and 3·DSB were assembled into blue- and green-emitting MPs, respectively, exhibiting prolonged PL lifetimes compared to pure host microcrystals (Fig. 5c; Supplementary Figs. 51−55 and Table 17). Perversely, 4,4’-(anthracene-9,10-diylbis(ethyne-2,1-diyl))bis(N,N-di-p-tolylaniline) (4) featuring an extended π-backbone (Fig. 5a, b) allows the inclusion of 9,10-diphenylanthracene (DPA) but only forms small cavity structures near its terminal groups (Fig. 5d and Supplementary Fig. 49). The resulting 4·DPA MPs exhibit red emission with a relatively high PLQY of 34.21%, caused by the slippage of adjacent 4 molecules along their long molecular axes in 4·DPA (Fig. 5c, d and Supplementary Table 17).

Elastic and lasing behaviors of CACs

In the 1-derived CAC systems, C−H···π interactions serve as the dominant force between 1 and guests, while additional weak interlayer C − H···π interactions resulting from 1’s bulky substituents occur orthogonally. This combination of intermolecular forces creates favorable conditions for the formation of flexible crystalline materials³⁹. As demonstrated, a centimeter-sized 1·ST single crystal was initially examined, demonstrating pronounced elastic properties. When subjected to axial mechanical stress, the ribbon-shaped crystal, mounted on a glass coverslip, underwent giant macroscopic bending (a maximum bending angle of ~180°) while maintaining both high-efficiency green emission and crystallinity. This crystal shows excellent reversibility through multiple deformation cycles (Fig. 6b and Supplementary Movie 1). Significantly, five additional ribbon-shaped single crystals (1·SB, 1·DSB^a, 1·4P, 1·4PPy, and 1·4S) also exhibit macroscopic elasticity with each capable of being bent into stable U-shaped configurations without fracture (Supplementary Movies 2–6).

Fig. 6. Elastic and lasing performances of CACs.

a Local interlayer molecular packing schematic diagram of CACs when viewed down the b-axis, and its schematic diagram of the corresponding single crystal elastic response. b Elastic bending behavior of six ribbon-like single crystals: 1·SB, 1·ST, 1·DSB^a, 1·4P, 1·4PPy, 1·4S. Scale bar, 5 mm. c PL images of 1 and nine host/guest microcrystals under 400 nm fs-laser excitation. Scale bar, 25 μm. d Schematic illustration of fs-laser vertical excitation at the maximum curvature points of the (001) facet of 1·DSB^a single crystals in both straight and bent states. The insets show the corresponding side-view photographs excited by UV light. Scale bar, 3 mm. e Low- and high-resolution PL spectra of the straight and bent 1·DSB^a crystals recorded above the lasing thresholds. The insets show the corresponding top-view PL images. Scale bar, 50 μm.

To quantitatively assess the mechanical properties of CACs, we performed uniaxial tensile tests on their single crystals using a laboratory-built instrument (Supplementary Fig. 56)⁴⁰. The average tensile modulus was determined as follows: 1·ST (0.152 GPa), 1·SB (0.277 GPa), 1·DSB^a (0.514 GPa), and 1·4PPy (0.662 GPa) (Supplementary Figs. 57 and 58). Control experiments further confirmed the effect of crystals thickness; for instance, thinner 1·ST crystals (23 μm) displayed greater bending flexibility than thicker ones (49 μm), consistent with classical mechanical expectations (Supplementary Fig. 59 and Movies 7–9). Molecular packing motifs of these six cocrystals when viewed down the b axis (Fig. 6a and Supplementary Figs. 60 and 61) reveal that multiple terminal substituents form interlocked hinge structures avoiding long-range interlayer slippage. Within the layers, the host and guest molecules form an extensive C–H···π interaction network, which tightly locks adjacent molecules and allows for immediate structural recovery upon the removal of external stress. As the guest size increases, the layer‑confined host/guest interactions strengthen, leading to a corresponding rise in tensile modulus. Notably, 1·ST exhibits a comparatively low modulus, which can be attributed to its distinctive herringbone‑like packing. This molecular arrangement enables the crystal to accommodate and sustain substantial deformation over a wide strain range under applied stress. Similar to single-component molecular crystals^40–43, these CACs all display great elasticity, showcasing their superior potential for designing flexible crystalline materials.

Noticeably, the combination of high PLQYs and short fluorescence lifetime is highly favorable for achieving amplified spontaneous emission (ASE) and lasing^44–47, suggesting that these well-crystallized host/guest microcrystals may function effectively as both microscale gain media and optical cavities. Comparative studies of the optically pumped characteristics of 1, 1·Dio, and 1·Tol microcrystals using a 400 nm femtosecond pulsed laser (Supplementary Fig. 62) revealed intense green emissions from their tips and lateral facets at higher pumping densities, accompanied by significantly enhanced PL intensity and markedly narrowed PL linewidth (Fig. 6c; Supplementary Figs. 63 and 64). The characteristic S-shaped nonlinear dependence of both PL intensity and linewidth on pump fluence clearly indicates the transition from spontaneous emission to ASE or lasing. Moreover, this confirms that microcrystals of 1 and 1·Tol exhibit ASE, while 1·Dio MPs show clear lasing action. The ASE/lasing thresholds (E_th) for 1, 1·Dio, and 1·Tol microcrystals were estimated to be 30.7, 8.04, and 7.01 μJ·cm^-2, respectively. The solvated microcrystals of 1 exhibit significantly reduced lasing thresholds compared to the pristine host MRs, which may be attributed to their extrinsic cavity structures. However, both pristine host MRs and solvated microcrystals of 1 suffer from rapid degradation in luminescence and lasing properties upon exposure to air, thereby limiting their practical application in photonic devices.

Notably, all seven host/guest microcrystals based on host 1 demonstrated wavelength-tunable lasing actions under identical testing conditions and their lasing thresholds were 2 to 15 times lower than that of pure host 1 MRs (Table 1 and Supplementary Figs. 65–72). Among these, the best-performing system, 1·BPDSB MPs, achieves an ultralow threshold of 2.61 μJ/cm². The CAC system outperforms both pure host 1 and previously reported DSB-based crystals by achieving a significantly lower lasing threshold (Supplementary Table 18), confirming the efficacy of this strategy for enhanced lasing performance. The effective reduction in the lasing threshold of this cocrystal system is primarily attributed to a significant increase in the radiative decay rate (k_r), which is influenced by cavity structure, size, and the degree of guest conjugation. As described above, structural analysis reveals that the cavity size of 1-derived CACs (1·SB, 1·ST, 1·DSB^a, 1·BPDSB) is effectively modulated by the linear length of the styryl-functionalized guest. Evidently, a clear negative correlation between the lasing threshold and the cavity size is observed: The lasing threshold shows a progressive decrease with increasing cavity size (Supplementary Fig. 73). This trend is consistent with the observation that 1·BPDSB, which has the largest cavity size, also has the lowest lasing threshold. For similar cavity sizes, the lasing threshold is governed by the guest conjugation degree. Despite the structural similarity, 1·DSB^a exhibited a significantly lower lasing threshold than 1·4P and 1·4PPy (E_{th, 1·DSBa} = 4.14  μJ/cm²; E_{th, 1·4P} = 7.55 μJ/cm²; E_{th, 1·4PPy} = 13.90 μJ/cm²). We attribute this to the extended π-conjugation length in DSB, which features a double-bond linkage versus the single bond in 4P and 4PPy (Supplementary Fig. 73). This trend holds for styryl-modified guests, where an increased conjugation degree consistently leads to a lower lasing threshold in their cocrystals. We propose that a larger size and greater conjugation in the guest strengthen host/guest intermolecular interactions. This, in turn, facilitates superior electronic coupling and accelerates k_r, thereby improving lasing performance.

Table 1.

Photophysical parameters of 1, 1·SB, 1·ST, 1·DSB^a, 1·DSB^b, 1·BPDSB, 1·4 P 1·4 PPy, 2·4 P, and 3·DSB microcrystals

	1	1 ∙ SB	1 ∙ ST	1 ∙ DSBa	1 ∙ DSBb	1 ∙ BPDSB	1 ∙ 4 P	1 ∙ 4 PPy	1 ∙ 4S	2 ∙ 4 P	3 ∙ DSB
τ (ns)	1.31	1.63	2.00	0.85	1.38	1.05	1.20	2.26	1.27	1.7	1.67
PLQYs (%)	36.30	62.85	79.35	74.44	85.60	95.36	63.99	75.94	21.03	72.28	90.92
λem (nm)	492	508	509	493	507	502	493	533	549	443	506
Eth (μJ·cm-2)	30.70	11.80	10.95	4.14	5.53	2.61	7.55	13.90	/	9.26	7.28
kr (ns-1)	0.26	0.39	0.40	0.87	0.62	0.91	0.53	0.34	0.17	0.43	0.54

Fundamentally, the formation of the CACs substantially alters the molecular packing, and the resulting confined cavity structure results in a significant PLQY enhancement of CACs. Notably, despite some variation, the PL lifetime (τ) of the host/guest microcrystals remains within the 1–2 ns range, which is consistent with that of pristine host 1 (Supplementary Fig. 36). According to the relation $K_{\mathrm{r}}=\frac{{\Phi }_f}{{\mathrm{\tau }}_{\mathrm{f}}}$, the significant improvement in Φ contributes to an increase in the k_r value of the CACs (Table 1 and Supplementary Figs. 21 and 74)²⁸. For instance, 1·BPDSB MPs exhibit a lower lasing threshold due to their higher kᵣ, whereas 1·4S MRs, with a lower kᵣ, show no lasing activity. This difference is attributed to the larger conjugation degree of the BPDSB guest. Although SB has a longer conjugation length than ST, the molecular cavity of 1·SB is smaller than that of 1·ST, which leads to a reduced kᵣ and an increased lasing threshold in its microcrystals. Additionally, the extra CT interaction between 1 and 4PPy results in an extended fluorescence lifetime, thereby reducing kᵣ and leading to an increased threshold in 1·4PPy MRs relative to that of 1·4P MRs. We also found that pristine microcrystals of hosts 2, 3, and 4 exhibited no ASE or lasing even at high pump energies (Supplementary Fig. 75). Upon guest encapsulation, however, the resulting 2·4P and 3·DSB microcrystals showed low-threshold lasing with thresholds of 9.26 and 7.28 μJ/cm², respectively (Supplementary Fig. 76). This demonstrates that guest encapsulation enhances the system’s gain, activating lasing in otherwise inactive host materials and thereby expanding the material toolbox for flexible lasers. However, no laser performance was observed in the 1·4S and 4·DPA microcrystals (Supplementary Fig. 77), which indicates their unsuitability as optical gain media.

Specifically for the 1·DSB^b MRs, the laser mode spacing follows the Fabry-Perot (FP) resonance relationship, $\mathrm{\Delta }\mathrm{\lambda }=\frac{{\mathrm{\lambda }}^2}{{2Ln}_g}$, where λ is the wavelength, L is the optical path length, and n_g is the group refractive index (Supplementary Fig. 78). The mode spacing decreases with increasing cavity length and exhibits a linear dependence on the reciprocal cavity length (1/L). Similar FP-type lasing characteristics are observed in 1·BPDSB MPs (Supplementary Fig. 79). Based on the flexible nature of CAC single crystals and their microcrystals’ lasing ability, we studied how bending affects their lasing. When a 400 nm fs laser was pointed at the surfaces of three ribbon-like 1·DSB^a single crystals in both straight and bent states, each showed bright PL spots (Fig. 6d, e). The straight crystal only produced ASE, with no clear laser peaks. However, the bent crystals exhibited high-resolution, fine PL peaks, confirming significant laser action (Supplementary Fig. 80). The more the crystal was bent, the lower the energy needed to start lasing. This is likely because the straight crystal cannot guide light effectively for waveguide feedback, while bending helps create distributed feedback and curved resonant cavities. This initial application case demonstrates the significant advantages of CACs in the field of flexible photonics. The inclusion of large π-conjugated guests significantly improves the stability of the host crystals, demonstrating promising application potential. After four months of air exposure at room temperature, six of the seven microcrystals maintain stable lasing thresholds (E_th′), showing only a minor increase (Supplementary Table 19 and Fig. 81). This enhanced durability further undercovers the additional advantage of CACs in boosting material stability.

In summary, we have developed a class of CACs where linear π-backbones functionalized with terminal bulky substituents serve as non-macrocyclic host frameworks. While solvent inclusion initially yielded 1:2 host/guest configurations, further exploration reveals that most cocrystals adopt a 1:1 stoichiometry, with cavity dimensions primarily dictated by the guest molecule size. Similar cavity architectures can be realized through subtle modifications of guest structures while retaining their consistent dimensions. The resulting host/guest microcrystals exhibit well-faceted 1D and 2D morphologies with high PLQYs. Intriguingly, modifying the host’s π-backbone enables tunable emission colors without altering the cavity structure. The improved cocrystal stability compared to pure host 1 arises from strong C − H···π interactions between a host and a guest, supplemented by weaker interactions among the host’s terminal substituents. Photophysical studies reveal host-dominated emission characteristics. The interlayer interlocking architecture of these systems endows their ribbon-like single crystals with excellent elasticity. Combining high crystallinity, high luminescent efficiency, short fluorescence lifetime, and enhanced stability, these binary microcrystals, even single crystals, show great promise as solid-state laser emitters.

This work establishes a linear non‑macrocyclic CAC system as a valuable structural framework, effectively overcoming key limitations of traditional macrocyclic host–guest chemistry, such as restricted cavity size, synthetic challenges, and constrained photophysical properties. By decoupling the cavity‑forming element (terminal steric bulk) from the emissive π‑backbone, our design not only accommodates a wide range of PAH guests but also allows independent tuning of cavity structure and emission color. The integration of high-performance luminescence, lasing capability, and notable elastic mechanical properties within a single organic cocrystal platform is rarely achieved. We have successfully constructed a diverse library of more than ten distinct cocrystals, though this only begins to explore the vast accessible structural space. By systematically tuning the structures of the host and guest molecules, solid‑state emission properties can be tailored on demand. Beyond luminescence, the guest‑dependent modulation of the molecular cavities also creates opportunities for emerging functions such as elastic mechanical response. Overall, this study significantly advances both fundamental understanding and application potential of flexible organic photonic materials.

Methods

Chemicals

Host compounds involving 4,4-bis[4-(di-p-tolylamino)styryl]biphenyl (1, 98%) and 4,4’-(1,4-phenylenedivinyl)bis[N,N-di(p-tolyl)aniline] (3, 99%) were purchased from Meryer and used without further purification. N4,N4,N4”^‘,N4”^‘-tetra-p-tolyl-[1,1^’:4^’,1^”:4^”,1^”‘-quaterphenyl]−4,4^”‘-diamine (2, 98%) and 4,4’-(anthracene-9,10-diylbis(ethyne-2,1-diyl))bis(N,N-di-p-tolylaniline) (4, 98%) were synthesized following the reported methods.

Guest compounds involving styrene (ST, 98%), trans-stilbene (SB, 98%), trans, trans-1,4-distryrylbenzene (DSB, 98%), p-quaterphenyl (4P, 98%), 4,4-di(4-pyridyl)biphenyl (4PPy, 98%), and α-quaterthiophene (4S, 98%), 9,10-diphenylanthracene (DPA, 98%) were purchased from TCI, and used without further purification. And 4,4’-di((E)-styryl)−1,1’-biphenyl (BPDSB, 98%) was synthesized following the reported methods.

Organic solvents involving tetrahydrofuran (THF, A.R.), 1,4-dioxane (Dio, A.R.), toluene (Tol, A.R.), dichloromethane (DCM, A.R.), acetone (AC, A.R.), N,N-dimethylformamide (DMF, A.R.), dimethyl sulfoxide (DMSO, A.R.), limonene (Lim, A.R.), chlorobenzene (CB, A.R.), α-methylstyrene (α-MST, A.R.), p-xylene (PX, A.R.), cyclohexane (CHX, A.R.), hexane (A.R.), n-propanol (PA, A.R.), isopropanol (IPA, A.R.), ethanol (EtOH, A.R.), methanol (MeOH, A.R.) were purchased from Beijing Chemical Agent Ltd., China. All solvents were directly used without further purification.

Synthesis of host and guest compounds

The detailed synthetic procedures of 2, 4, and BPDSB are described in Supplementary Section 1.

Solution assembly of four host materials

As a representative example, pure 1 MRs was performed by slow evaporation of a drop of 1 in a THF/EtOH mixture (v/v = 1:1, C₁ = 2.5 mM) on a silicon or a quartz substrate at room temperature. Similarly, 2 MRs were prepared by slow evaporation of its monomer solution in a THF/EtOH mixture (v/v = 1:1, C₂ = 0.5 mM) on a silicon or a quartz substrate at room temperature. And 2 MRs were obtained by slow evaporation of a drop of 3 in a Dio/EtOH mixture (v/v = 1:1, C₃ = 5 mM) on a silicon or a quartz substrate at room temperature. 4 MRs were prepared by slow evaporation from THF/EtOH (v/v = 1:2, C₄ = 0.8 mM) on a silicon or a quartz substrate at room temperature, as shown in Supplementary Table 13.

Preparation of large-sized single crystals of four host materials

Single crystals of 1 or 4 were synthesized by slow evaporation of their monomer solution in THF/EtOH (v/v = 1:1, C₁ = 1 mM or C₄ = 0.5 mM) in a sealed glass bottle. Similarly, a single crystal of 3 was formed by slow evaporation of a stock solution of 3 in Dio/EtOH (v/v = 1:1, C₃ = 1 mM). A single crystal of 2 was obtained via an anti-solvent diffusion method combined with solvent-evaporation induced crystallization. Typically, 2 mL of a stock solution of 2 in THF (C₂ = 0.5 mM) was sealed in a 50 mL weighing bottle that contained an unsealed weighing bottle loaded with 6 mL of ethanol. EtOH vapor acted as an anti-solvent and was crucial to the growth of large-sized crystals of 2.

Preparation of 1·Dio and 1·Tol MPs

Typically, 1·Dio MPs were achieved by slow evaporation of a drop of a precursor solution of 1 in Dio/PA (v/v = 1:1, C₁ = 1 mM) on a silicon or a quartz substrate. Similarly, 1·Tol MPs were prepared by slow evaporation of 1 from Tol/IPA (v/v = 1:1, C₁ = 2.5 mM).

Preparation of large-sized single crystals of 1·Dio and 1·Tol

Single crystals of 1·Dio were synthesized by direct evaporation of a monomer solution of 1 in Dio/PA (v/v = 1:1, C₁ = 2.5 mM) in a sealed glass bottle. Similarly, single crystals of 1·Tol were formed by slow evaporation of a stock solution of 1 in Tol/IPA (v/v = 1:1, C₁ = 2.5 mM).

Preparation of the microcrystals of pure guests

Typically, SB MPs were achieved by slow evaporation of a drop of a precursor solution of SB in DCM/EtOH (v/v = 1:1, C_SB = 50 mM) on a silicon or a quartz substrate. Similarly, DSB MRs were prepared by slow evaporation from a Dio/CHX mixture (v/v = 1:1, C_DSB = 2.67 mM). 4P MPs were synthesized through slow evaporation from Dio (C_4P = 2 mM), while 4S MPs were acquired by slow evaporation from a Dio/EtOH mixture (v/v = 1:1, C_4S = 1.25 mM). DPA MRs were prepared from a DCM/EtOH mixture (v/v = 1:1, C_DPA = 1.5 mM), as detailed in Supplementary Table 4.

In contrast, BPDSB MPs were obtained by uniformly mixing 1 mL of a monomer solution of BPDSB in THF with 1 mL EtOH and the mixture was allowed to stand for 30 min at 25 °C. While 4PPy MPs were synthesized on a hot plate by a microspacing physical vapor transport (PVT) technique. Prior to sublimation, a drop of a monomer solution of 4PPy in THF (C_4PPy = 2 mM) was spin-coated on a glass coverslip and then slowly evaporated. The resulting microparticles were used as the precursor source for vapor-phase growth. The sublimation temperature and time were set at 230 °C and 40 s, respectively.

Preparation of host/guest microcrystals

Host/guest microcrystals derived from hosts 1−4 were synthesized following two different synthetic procedures (Methods I and II), and the detailed experimental parameters are shown in Supplementary Table 7.

Method Ⅰ

1·SB, 1·ST, 1·DSB^a, 1·DSB^b, 1·4S, 2·4P, 3·DSB, and 4·DPA microcrystals were prepared by a facile solution co-assembly route. For 1·SB, solid powders of host 1 and guest SB with 1:3 molar ratio were first dispersed in a THF/Dio mixture (v/v = 3:2, C₁ = 5 mM, C_SB = 15 mM). After the addition of an equal volume of IPA, such a ternary mixture (THF/Dio/IPA, v/v/v = 3:2:5) was allowed to evaporate slowly, in which IPA acts as an anti-solvent. Finally, 1·SB MRs were deposited on a silicon or a quartz substrate at room temperature. To enable the complete binding of 1 to SB, an excess of SB molecules was used to enhance the intermolecular interactions between them. Similarly, 1·ST MRs were obtained using THF, ST, and IPA as co-solvents. Specifically, solid powders of host 1 were initially dissolved in a THF/ST mixed solution (v/v = 3:2, C₁ = 5 mM), and an equal volume of IPA was then added as an anti‑solvent to induce crystallization. The ternary mixture (THF/ST/IPA, v/v/v = 3:2:5) was allowed to evaporate slowly at ambient temperature, yielding 1·ST MRs deposited on silicon or quartz substrates. Note that ST is a liquid at room temperature and was used directly as a solvent component in the initial solution. 1·DSB^a MRs and 1·DSB^b MPs were obtained at different host/guest molar ratios. Specifically, 1·DSB^a MRs were obtained by slow evaporation of a Dio/EtOH mixture (v/v = 1:1, C₁ = C_DSB = 0.5 mM), while 1·DSB^b MPs were prepared at a host/guest molar ratio of 1:2 using a Dio/PA mixture (v/v = 1:1, C₁ = 0.5 mM, C_DSB = 1 mM). Following the aforementioned approach, 1·4S MRs were prepared from a Dio/EtOH mixture (v/v = 1:1, C₁ = C_4S = 2.5 mM), while 2·4P MPs were synthesized by employing THF as the good solvent and EtOH as the anti-solvent under slower evaporation conditions (v/v = 1:1, C₂ = C_4P = 0.5 mM). For 3·DSB MPs, a ternary solvent system (THF/Dio/IPA, v/v/v = 3:2:15, C₃ = C_DSB = 0.4 mM) was utilized, where IPA served as the anti-solvent during slow evaporation. In the case of 4·DPA MPs, a binary mixture (Dio/MeOH, v/v = 1:1, C₄ = C_DPA = 1 mM) was gradually evaporated, with MeOH serving as the anti‑solvent during crystallization.

Method ⅠⅠ

1·BPDSB, 1·4P, and 1·4PPy microcrystals were prepared by a solution-phase coprecipitation method. For 1·BPDSB MPs, solid powders of host 1 and guest BPDSB with a 2:1 molar ratio were firstly dispersed in a THF/Tol mixture, and then EtOH was introduced as an anti-solvent (v/v/v = 3:2:5, C_BPDSB = 0.5 mM). The mixture solution was transferred into a petri dish, which was placed on a hot plate to allow for slow evaporation at 50 °C. Shortly afterwards, the clear solution became turbid and crystalline precipitates of 1·BPDSB were collected by centrifugation. For 1·4P MRs, solid powders of host 1 and guest 4P in a 1:1 molar ratio were initially dissolved in Dio to prepare a precursor solution (C₁ = 1 mM). An equal volume of EtOH was then added as an anti-solvent. After approximately 1 min, a large quantity of 1·4P MRs precipitated. In contrast, 1·4PPy MRs were obtained without the use of an anti-solvent to induce crystallization. Specifically, equimolar amounts of host 1 and guest 4PPy were co-dissolved in THF, and the solution was allowed to stand at room temperature for about 3 h, after which a large quantity of 1·4PPy MRs precipitated spontaneously.

Preparation of large-sized host/guest single crystals

Large-sized host/guest single crystals suitable for single-crystal XRD analysis (SCXRD) were obtained following three different preparation procedures (Methods i−iii), and the detailed experimental parameters are shown in Supplementary Table 7.

Method I

Centimeter- or millimeter-sized single crystals of 1·ST, 1·SB, 1·DSB^a, 1·4PPy, 1·4S, and 4·DPA were prepared via solution-phase evaporation. Take 1·SB as an example, its single crystals were obtained by slow evaporation of a monomer solution of 1 in THF/Dio/PA containing a certain amount of SB (v/v/v = 3:2:1, C₁ = 2.5 mM, C_SB = 7.5 mM) in a sealed glass bottle. Single crystals of 1·ST were grown by slow evaporation from a ST/IPA mixture (v/v = 1:1, C₁ = 1 mM) in a sealed plastic centrifuge tube. 1·DSB^a and 1·4S single crystals were both obtained via slow evaporation from a Dio/EtOH mixture (C₁ = C_DSB = 0.5 mM; v/v = 1:1, C₁ = C_4S = 1 mM). For 1·4PPy, its single crystals were obtained by slow evaporation from a THF solution (C₁ = C_4PPy = 1 mM). Single crystals of 4·DPA were grown by slow evaporation from a Dio/MeOH mixture (C₄ = C_DPA = 1 mM).

Method II

For poorly soluble guests, CAC single crystals were obtained via thermal evaporation-induced crystallization. As a representative example, for 1·DSB^b, a mixture of host 1 and guest DSB in a 1:2 molar ratio was dissolved in a THF/Tol/EtOH mixture (v/v = 1:1:2), with an initial concentration of C₁ = 2C_DSB = 0.25 mM. The resulting dilute solution was transferred to a petri dish and slowly evaporated at 60 °C on a hot plate. After complete solvent removal, large single crystals of 1·DSB^b were deposited on the bottom of the dish. Similarly, single crystals of 1·BPDSB suitable for X‑ray diffraction analysis were prepared using a 1:1 molar ratio of host 1 and guest BPDSB in a THF/Tol/EtOH mixture (v/v/v = 3:2:5) under identical concentration and temperature conditions.

Method III

Centimeter- or millimeter-sized single crystals of 1·4P were obtained via an anti-solvent diffusion method combined with solvent-evaporation induced crystallization. Specifically, solid powders of host 1 and guest 4P with a 1:1 molar ratio were first dispersed in Dio (C₁ = C_4P = 0.5 mM), and then 2 mL of a mixed solution was sealed in a 50 mL weighing bottle that contained an unsealed glass bottle loaded with 6 mL of EtOH. After one week, large-sized single crystals of 1·4P were formed at the bottom of the weighing bottle. Similarly, high-quality single crystals of 2·4P and 3·DSB were also realized by the above cocrystallization route. In detail, host 2 and guest 4P were dissolved in THF at equal concentrations (C₂ = C_4P = 0.5 mM). Then, 2 mL of this mixture was placed into a 50 mL weighing bottle that contained an unsealed glass bottle with 6 mL of EtOH. After about one week, large-sized single crystals of 2·4P had formed at the bottom of the weighing bottle. In the case of 3·DSB, an equimolar mixture of host 3 and guest DSB was initially dissolved in a THF/Dio mixture (v/v = 3:2, C₃ = C_DSB = 0.5 mM). And then, 2 mL of this solution was placed into a 50 mL weighing bottle containing an unsealed glass bottle filled with 6 mL of IPA, and high-quality crystals of 3·DSB were formed at the bottom of the weighing bottle over a period of one week.

Sample characterization

¹H and ¹³C NMR spectra were recorded on a JEOL JNM-A500/ECZ400S instrument with chemical shifts being referenced against tetramethylsilane as the internal standard. The morphologies and sizes of a variety of host/guest microcrystals and their constituent materials were examined using field-emission scanning electron microscopy (FESEM, Apreo S LoVac) at an acceleration voltage of 5 kV. Prior to observation, the samples were deposited with a thin gold layer using an Edwards Sputter Coater. TEM images were obtained using a JEOL-2100 electron microscope operated at 80 kV. Each of the as-prepared samples was transferred onto a carbon-coated copper grid and dried at room temperature. The X-ray diffraction (XRD) patterns were measured on a D/max 2400 X-ray diffractometer with Cu Kα radiation (λ = 1.54050 Å) operated in the 2θ range from 3 to 40°, by using the samples spin-coated on the surface of a quartz substrate. The photoluminescence (PL) spectra of the samples were measured on a HORIBA JOBTN YVON FLUOROMAX-4 spectrofluorimeter with a slit width of 1 nm. The PL decay measurements were performed by a steady-state and transient-state fluorescence spectrometer (FLS-1000). The samples were both deposited on the surface of a quartz substrate. The fluorescence microscopy (FM) images were obtained using an Olympus bx53 fluorescence microscope with a spot-enhanced charge-coupled device (CCD). Each of the host/guest microcrystals dispersed on a glass substrate was excited locally by variable pumping pulses via a home-built micro-PL system. A fs laser system (Pharos, Light Conversion) delivered laser pulses at 800 nm (50 fs, 1 KHz), which were then sent to an optical parametric amplifier to generate the pump pulses (400 nm). The fs laser spot, with a diameter of ~50 µm, is focused onto the samples through a 40x objective lens. The PL spectra were collected using an optical spectrometer (Princeton Instruments) with a spectral resolution of 0.2 nm.

Theoretical calculations

The geometries of the crystals were optimized using density functional theory (DFT) under periodic boundary conditions in the CP2K software package^48,49. The PBE exchange-correlation functional and GTH pseudopotentials with the DZVP-MOLOPT-GTH basis set were employed. The convergence criteria of minimization for the maximum geometry change and force difference were set at 3 × 10^-3 Bohr and 4.5 × 10^-4 Hartree/Bohr, respectively. The vertical electron excitations were calculated using time-dependent density functional theory (TD-DFT) at the PBE/TZVP-GTH theoretical level. The energy decomposition analysis was performed at the B3LYP-D3(BJ)/6-311 + G(2 d,p) theoretical level in Gaussian 09 software, in combination with the sobEDA approach^50,51. In addition, the electron and hole density distributions were visualized using the Multiwfn software package^52,53.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Y.L., Y.Z. (Yanli Zhao), and Z.M. supervised and conceived the project. Z.F., Y.Z. (Yihui Zhu), Z.M., C.H., and L.Z. performed the experimental studies. Y.L., Z.F., Y.Z. (Yihui Zhu), and Z.M. contributed to the data analyses and discussion. X.C. and H.N. performed the theoretical calculation. C.W. and Z.F. performed tensile testing and data analyses. Y.L., Z.F., and Y.Z. (Yihui Zhu) wrote the manuscript, and all authors have agreed to the content of the manuscript.

Peer review

Peer review information

Nature Communications thanks Changsik Song, Zengqi Xie and the other anonymous reviewer for their contribution to the peer review of this work. A peer review file is available.

Data availability

The experiment data are available from the corresponding author upon request. Crystallographic data for the structures reported in this Article have been deposited at the Cambridge Crystallographic Data Centre, under deposition numbers CCDC 2379309 (1), CCDC 2379310 (2), CCDC 2379311 (4), CCDC 2512247 (1·Dio), CCDC 2379313 (1·Tol), CCDC 2442680 (1·ST), CCDC 2379314 (1·SB), CCDC 2379315 (1·DSB^a), CCDC 2512248 (1·DSB^b), CCDC 2450068 (1·BPDSB), CCDC 2379319 (1·4P), CCDC 2379326 (1·4PPy), CCDC 2379338 (1·4S), CCDC 2379342 (1·^2-ClAn), CCDC 2379328 (1·Tc), CCDC 2379359 (2·4P), CCDC 2261020 (3·DSB), and CCDC 2379360 (4·DPA). Copies of the data can be obtained free of charge via https://www.ccdc.cam.ac.uk/structures. All data are available in this Article and its Supplementary Information. 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 10.1038/s41467-026-69483-0.