Pressure-driven steric hindrance engineering for maximizing photoluminescence in covalent organic frameworks

Abstract

Covalent organic frameworks (COFs) are promising platforms for smart photoluminescent (PL) materials, but their emission is often quenched by π-π stacking–induced nonradiative transitions. Here, we use a pressure-treatment strategy on a series of sterically engineered pyrene-based imine COFs—Py-Da-COF, Py-Da-2CH₃-COF, and Py-Da-4CH₃-COF—to achieve steric-hindrance–responsive PL enhancement. Notably, the pressure-treated Py-Da-4CH₃-COF exhibits an increase in PL quantum yield, reaching a record-high value of 91.5% from the initial 14.7%. Experimental and theoretical analyses reveal that the bulky methyl substituents elevate the phase transition barrier, locking the COF into an irreversible a quasi–AB stacking configuration. This structural rearrangement suppresses π-π interactions and restricts carbon-hydrogen vibrations, minimizing nonradiative decay. Our work establishes a generalizable approach to designing high-performance PL COFs for practical optoelectronic applications.

Pressure-treated Py-Da-4CH₃-COF exhibits a record-high PLQY of 91.5% from the initial 14.7% using steric hindrance engineering.

INTRODUCTION

The surging demand for the light-emitting applications of optoelectronic devices, biological imaging, and chemical sensing underlines the importance of developing high-performance smart photoluminescence (PL) materials (1–4). Covalent organic frameworks (COFs), constructed from organic building blocks via dynamic covalent bonds, represent a class of crystalline porous materials featuring long-range order, well-defined topology, and structural tunability (5–10). These attributes enable COFs to be promising candidates to address the growing requirement, owing to the integration of various molecular luminophore building blocks into two-dimensional (2D) or 3D frameworks (11–13). However, realizing superior emission performance in COFs encounters a considerable challenge. There are usually various nonirradiative transition processes in COFs arising from π-π stacking and intramolecular rotation, giving rise to intensified PL quenching and inevitably compromising the practical application (14, 15). As a consequence, the quest to develop design strategies that can realize efficient emission in COFs represents a highly anticipated area in the COF field.

To this end, endeavors including reducing the π-π interactions by exfoliation and staggered stacking, as well as introducing intra/interlayer hydrogen bonding or vinyl linkage to suppress bond rotation, hold great importance for improving radiative transitions (16–19). Furthermore, a recent study has demonstrated that isotope substitution can inhibit bond vibrations and thus result in a substantial PL quantum yield (PLQY) enhancement (20). Nevertheless, the predesign of monomer to directly manipulate topological structures has rendered the synthesis process quite challenging. Pressure emerges as a clean and powerful tool that can manipulate the crystal structure, intermolecular interactions, and optical properties without altering the chemical composition (21–24). Recent studies have detailed that pressure-induced emission enhancement and piezochromic behavior in COFs have led to preliminary progress (25–27). However, harvesting enhanced emission of COF materials through pressure treatment has remained elusive. Typically, interlayer interactions response more sensitively to pressure than intralayer interactions governed by strong covalent bond (28). Hence, extreme pressure can induce interlayer sliding and structural evolutions of COFs, which tend to trigger tremendous changes in optical property. Notably, introducing steric hindrance may result in large steric repulsion, which affects the stacking configuration (29, 30). Thereby, steric hindrance has been used to optimize π-π stacking, and its impact on emission enhancement has been limited. We hypothesize that combining steric hindrance design with pressure treatment could synergistically maximize PL efficiency (Fig. 1, A and B).

Fig. 1. Steric hindrance design strategy for high-performance COFs.

(A) Increased methylation introduces steric hindrance and weakens interlayer interactions, promoting pressure-induced interlayer sliding in COFs. (B) Schematic illustration of the PL enhancement mechanism. Abs., absorption; Fluo., fluorescence; N.R., nonradiation.

In light of this, we engineer a series of pyrene-based imine COFs, including Py-Da-COF, Py-Da-2CH₃-COF, and Py-Da-4CH₃-COF, with different numbers of methyl groups based on the enhanced steric hindrance strategy. Pressure treatment engineering approach is leveraged to enhance PLQY from 14.7 to 91.5% in Py-Da-4CH₃-COF, surpassing previously reported COF materials. This high-emission state is attributed to the increased potential barrier of phase transition caused by the large steric hindrance, enabling the irreversible structure. The reduced π-π interactions and restricted C─H vibrations derived from a pressure-treated a quasi–AB stacking model facilitate the intensity of the radiative transition oscillator. In contrast, the pressure-treated Py-Da-2CH₃-COF with minor steric hindrance maintains a similar PL intensity comparable to its initial state under the synergistic effect of a smaller slipped AA stacking model and a closer interlayer distance, whereas the reduced interlayer distance enhances the π-π interaction in Py-Da-COF, thereby leading to a decrease in emission. A phosphor-converted light-emitting diode (pc-LED) based on pressure-treated Py-Da-4CH₃-COF is fabricated with magnificent chromaticity and fatigue stability, which provides a platform for the potential in solid-state lighting and displays.

RESULTS

Constructions and characterizations

Pyrene, a robust PL unit with extended π-conjugation, serves as an ideal building block for constructing emissive COFs (31, 32). To explore the regulation of steric hindrance effects on framework stacking and pressure-responsive behavior, we synthesized a series of pyrene-based COFs via Schiff-base polycondensation between 1,3,6,8-tetra(4-formylphenyl)pyrene (Py-CHO) and diamine linkers bearing increasing methyl substitutions. The parent Py-Da-COF was constructed from p-phenylenediamine (Da-NH₂), while Py-Da-2CH₃-COF and Py-Da-4CH₃-COF were obtained using 2,5-dimethyl-p-phenylenediamine (Da-2CH₃-NH₂) and 2,3,5,6-tetramethyl-p-phenylenediamine (Da-4CH₃-NH₂), respectively (Fig. 2A). All three COFs adopt a 2D sql net topology, constructed from four-connected pyrene nodes and linear diamine linkers. The structures of as-synthesized COFs were determined by powder x-ray diffraction (PXRD) together with structural modeling in Materials Studio 7.0 (33). The first and most intense diffraction peaks at 3.80°, 3.78°, and 3.88° could be observed in Py-Da-COF, Py-Da-2CH₃-COF, and Py-Da-4CH₃-COF, consistent with the strong reflection from (110) planes (Fig. 2, B to D). For Py-Da-COF and Py-Da-2CH₃-COF, additional reflections were indexed to the (020), (220), (310), (330), (240), and (440) planes, respectively. In contrast, Py-Da-4CH₃-COF displayed additional reflections such as (120), (130), (040), and (211), as well as high-angle peaks including (060), (361), and (461), indicating a more complex stacking pattern and reduced symmetry relative to the other two COFs. Py-Da-COF and Py-Da-2CH₃-COF adopt 2D eclipsed (AA) stacking mode belonging to the C2/m space group, whereas Py-Da-4CH₃-COF crystallizes in the P1 space group (figs. S1 to S4). The crystalline layers of Py-Da-4CH₃-COF are slightly shifted along both a and b axes, forming a slipped AA stacking configuration, more specifically, a serrated AA stacking arrangement. This pronounced interlayer offset and a reduction in symmetry are primarily due to the steric repulsion from the four methyl groups. Pawley refinement results matched well with the experimental PXRD patterns of Py-Da-COF [weighted profile factor (R_wp) = 1.89%, profile factor (R_P) = 3.45%], Py-Da-2CH₃-COF (R_wp = 5.63%, R_P = 4.65%), and Py-Da-4CH₃-COF (R_wp = 3.43%, R_P = 1.99%). The simulation results yielded optimal unit cell parameters of a = 31.9786 Å, b = 34.6198 Å, c = 3.8314 Å, α = γ = 90°, and β = 95.9077° for Py-Da-COF; a = 33.1750 Å, b = 33.7007 Å, c = 7.6794 Å, α = γ = 90°, and β = 95.1011° for Py-Da-2CH₃-COF; and a = 35.3981 Å, b = 29.8019 Å, c = 6.6509 Å, α = 89.5651, γ = 90.5489°, and β = 89.8775° for Py-Da-4CH₃-COF. We also performed Rietveld refinement on these three COFs, and the results agreed well with the experimental PXRD patterns (fig. S5 and tables S1 to S3). Together, these PXRD analyses and refinement results conclusively demonstrate the high crystallinity and distinguishable stacking architectures of the three COFs, highlighting the structural tunability enabled by methyl substitution.

Fig. 2. Chemical structures and characterizations of COFs.

(A) Synthetic pathway of Py-Da-COF, Py-Da-2CH₃-COF, and Py-Da-4CH₃-COF. Experimental and simulated PXRD patterns of (B) Py-Da-COF, (C) Py-Da-2CH₃-COF, and (D) Py-Da-4CH₃-COF. Insets exhibit the structural models viewed normal and parallel to the ab plane (purple, blue, and yellow spheres represent C, H, and N atoms, respectively). a.u., arbitrary units. High-resolution transmission electron microscopy (HRTEM) images of (E) Py-Da-COF, (F) Py-Da-2CH₃-COF, and (G) Py-Da-4CH₃-COF showing periodic (110) lattice fringes with spacings of ~2.47, ~2.47, and ~2.16 nm. Insets exhibit the corresponding fast Fourier transform patterns.

We used scanning electron microscopy (SEM) and high-resolution transmission electron microscopy (HRTEM) to elucidate the morphology and the crystal structures of Py-Da-COF, Py-Da-2CH₃-COF, and Py-Da-4CH₃-COF. SEM characterization revealed that Py-Da-COF and Py-Da-2CH₃-COF formed elongated rodlike structures, while Py-Da-4CH₃-COF appeared as irregular sheetlike aggregates (fig. S6). The (110) lattice fringes of the first two materials exhibited an interplanar spacing of 2.47 nm, which was consistent with the typical AA stacking configuration (Fig. 2, E and F). In contrast, the tetramethyl-substituted Py-Da-4CH₃-COF showed a reduced lattice spacing of 2.16 nm due to interlayer slipping induced by the increased steric hindrance (Fig. 2G). This structural deviation aligns well with the simulated models and supports the presence of perturbed stacking in response to methyl substitution. The aforementioned observations strongly corroborate the long-range ordering and structural tunability of the COF frameworks. The porosity properties of as-prepared COFs were conducted by the nitrogen adsorption-desorption isotherm experiments at 77 K (fig. S7). All three COFs were classified as reversible type-I isotherms with steep uptake at low relative pressure (P/P₀ < 0.1), characteristic of microporous materials. Py-Da-COF, Py-Da-2CH₃-COF, and Py-Da-4CH₃-COF showed the calculated Brunauer-Emmett-Teller (BET) surface areas of 1596, 814, and 684 m² g⁻¹, respectively. Pore size distribution analyses were based on nonlocal density functional theory. The Py-Da-COF and Py-Da-2CH₃-COF had dominant pore sizes around 16 Å, whereas Py-Da-4CH₃-COF manifested a slightly narrower distribution centered at ~13 Å, which is attributed to interlayer sliding induced by the steric bulk of the four methyl substituents (fig. S8). The chemical structures of the COFs were also investigated by Fourier transform infrared (FTIR) spectroscopy and solid-state ¹³C nuclear magnetic resonance (¹³C NMR) spectroscopy. FTIR spectra displayed the appearance of C═N stretching vibration peaks at 1619, 1623, and 1633 cm⁻¹ in Py-Da-COF, Py-Da-2CH₃-COF, and Py-Da-4CH₃-COF, accompanied by the sharp diminishment of C═O (~1690 cm⁻¹) and N─H stretching vibrations (3211 to 3422 cm⁻¹) (fig. S9). This demonstrated the completion of Schiff-base condensation in all three COFs. Solid-state ¹³C NMR spectra further supported this transformation, showing diagnostic imine carbon resonances at 156.5, 155.3, and 160.5 parts per million for the respective COFs (figs. S10 to S12). Thermogravimetric analysis (TGA) showed that all three COFs were thermally stable up to 350°C under a N₂ atmosphere (fig. S13).

PL properties during a full compression and decompression cycle

Under the 355-nm laser excitation, we found that the pristine Py-Da-COF, Py-Da-2CH₃-COF, and Py-Da-4CH₃-COF presented chartreuse and yellow light emission with PLQYs of 7.7, 10.6, and 14.7% (Fig. 3, A to C, and fig. S14). This weak PL performance is attributed to the nonradiative energy loss originating from the strong π-π interactions, which stem from the 2D eclipsed (AA) stacking mode and the slipped AA stacking configuration with a slight shift along both a and b axes of COFs. Therefore, we used pressure as an external trigger to manipulate the stacking configuration of molecules for boosting the PL properties in COFs. We conducted in situ high-pressure PL experiments to investigate the emission properties of all three COFs. It was found that the target Py-Da-COF exhibited weaker emission than its atmospheric pressure state after an entire cycle of 1 atm to 7.9 GPa, while the PL intensity returned to the initial state in the pressure-treated Py-Da-2CH₃-COF (Fig. 3, A and B). The corresponding PLQYs have changed from the original 7.7 and 10.6 to 5.0 and 10.6%, respectively (fig. S14). After the same compression cycle, we obtained the dazzling yellow emission in Py-Da-4CH₃-COF, and the PL intensity experienced a substantial 5.7-fold enhancement compared to the initial value (Fig. 3C). Under this condition, the PLQY of pressure-treated Py-Da-4CH₃-COF increased from the initial 14.7 to 91.5%, which was the highest value among all state-of-the-art PL COFs reported so far (fig. S14 and table S4) (14, 15, 18–20, 34–36). In addition, we performed the excitation spectra of these three COFs before and after applying pressure (fig. S15). It was found that the excitation spectra of each sample were not entirely consistent before and after pressure treatment, which could be mainly attributed to the structural changes. The PLQYs at both the optimal excitation wavelength and 355 nm were summarized (figs. S14 and S16 and table S5). We further probe the optical property evolution of all three COFs during the compression and decompression process. With increasing pressure, the emission intensities of these three COFs decreased progressively until they disappeared at 7.9 GPa (Fig. 3, D to F). During the decompression process, the PL intensity of Py-Da-COF gradually recovered but eventually remained weaker than the initial state (fig. S17A). In Py-Da-2CH₃-COF, the PL intensity upon decompression changed in a similar way as during the pressurization process, ultimately returning to its original intensity (fig. S17B). In contrast, an increase in emission emerged, as the pressure gradually released and the PL intensity of pressure-treated Py-Da-4CH₃-COF was 5.7 times that of the initial value (Fig. 3G). In this regard, we speculate that pressure effectively overcomes PL quenching induced by π-π stacking in COF materials with greater steric hindrance from methyl substituents, achieving a pronounced emission. Meanwhile, the wavelengths of PL exhibited continuous redshifts up to 154, 142, and 132 nm for Py-Da-COF, Py-Da-2CH₃-COF, and Py-Da-4CH₃-COF and from 1 atm to 6.0 GPa. In situ PL images visually displayed that the naked-eye visible color of Py-Da-COF, Py-Da-2CH₃-COF, and Py-Da-4CH₃-COF manifested stepwise changes from chartreuse and yellow to red (fig. S18). Complementing these observations, pressure-dependent Commission International de l’Eclairage (CIE) coordinates were also measured (Fig. 3H and fig. S19). The evolution of the CIE coordinates from (0.40, 0.57), (0.41, 0.54), and (0.45, 0.53) at 1 atm to (0.45, 0.35), (0.53, 0.35), and (0.59, 0.32) at 6.0 GPa in Py-Da-COF, Py-Da-2CH₃-COF, and Py-Da-4CH₃-COF confirms that these three COFs are appealing candidates as pressure-responsive materials. A prominent emission enhancement harvested from the initial weak yellow light in Py-Da-4CH₃-COF through pressure treatment has never been reported previously. In particular, this exceeds the PLQYs of all other reported COF materials (Fig. 3I).

Fig. 3. Pressure-dependent PL properties.

PL spectra and photographs of (A) Py-Da-COF, (B) Py-Da-2CH₃-COF, and (C) Py-Da-4CH₃-COF at 1 atm and upon complete release of the pressure. PL spectra of (D) Py-Da-COF, (E) Py-Da-2CH₃-COF, and (F) Py-Da-4CH₃-COF during compression. h, hours. (G) Pressure-dependent PL intensity of Py-Da-4CH₃-COF. (H) The CIE chromaticity diagram of Py-Da-4CH₃-COF in the pressure range of 1 atm to 7.9 GPa. (I) Benchmark of PLQYs for reported COFs versus this work.

DISCUSSION

Evolution of crystal structure and emission mechanism after pressure treatment

To gain a deeper understanding of the relationship between the optical properties and structure, we performed the PXRD experiments of all three COFs after a compression cycle from 1 atm to 7.9 GPa. In the pressure-treated Py-Da-4CH₃-COF, the Bragg diffraction peak of the (110) plane nearly disappeared, while the (020) and (220) peaks showed marked intensity enhancement and peak shift compared to the pristine sample, substantiating the distortion of the structure (Fig. 4A and fig. S20). It displayed the identified peaks at ~4.7°, 6.0°, 6.7°, 11.9°, 13.0°, and 14.4° for the (110), (020), (220), (320), (301), and (420) facets, respectively, which matched well with a quasi–AB stacking model (37, 38). Pawley refinement afforded a P1 space group with the cell unit parameters of a = 37.8103 Å, b = 30.4432 Å, c = 6.8838 Å, α = 86.9984°, β = 116.0043°, and γ = 109.0034° and the minor error values of R_p = 0.46% and R_wp = 0.99%. In contrast, the diffraction peak of the (110) crystal face demonstrated a slight attenuation in the pressure-treated Py-Da-2CH₃-COF, whereas the (020), (310), and (220) reflections intensified and shifted (Fig. 4B and fig. S21). The diffraction peaks at 2θ = 3.8°, 6.0°, 6.5°, 7.6°, 11.3°, 12.2°, 13.6°, 14.7°, and 15.0° affiliated to the (110), (020), (310), (220), (330), (520), (240), (211), and (440) crystal planes. We denoted that the pressure-treated sample exhibited a slipped AA stacking configuration with moderate offset along both a and b axes. Pawley refined cell parameters were established as a = 42.9985 Å, b = 28.3883 Å, c = 6.6474 Å, α = 95.9110°, β = 87.0024°, and γ = 89.9904° with negligible residuals R_p = 0.41% and R_wp = 0.78%. These observations signified a tremendous transformation in Py-Da-2CH₃-COF from the C2/m space group at ambient conditions to the P1 space group after pressure treatment. Nonetheless, all diffraction peaks of the pressure-treated Py-Da-COF were essentially in accordance with the original state (Fig. 4C and fig. S22). The identical stacking model and similar unit cell parameters before and after pressure treatment suggested that the structure had not experienced marked deformation, with only the interlayer spacing becoming closer. In addition, Rietveld refinements also yielded excellent agreement between the simulated and experimental patterns (fig. S23 and tables S6 to S8). With the above consequences, we found that the stacking model of targeted Py-Da-4CH₃-COF varied from the initial slipped AA stacking configuration to a quasi–AB stacking model, leading to a complete interlayer displacement of pyrene groups with larger π electrons (Fig. 4D). This avoided PL quenching derived from π-π stacking arrangement, thereby eventually benefiting the yellow light emission enhancement. There was a phase transition in Py-Da-2CH₃-COF from C2/m to P1, during which the initial eclipsed stacking mode changed to a slipped AA stacking mode (Fig. 4E). Although the pyrene groups in pressure-treated Py-Da-2CH₃-COF exhibited partial displacement, the interlayer spacing also became correspondingly closer. We proposed that the aforementioned two collaboration effects gave rise to the restoration of the PL intensity to its initial state. In pressure-treated Py-Da-COF, the stacking mode remained unchanged, while the interlayer distance was reduced, resulting in the reduction of PL intensity. Hence, it is clear that the substantial steric effect of methyl groups with a greater number was anticipated to increase the barrier of phase transition, thus preserving the pressure-derived metastable states. This reduced the PL quenching arising from π-π stacking of pyrene groups with larger π electrons, ultimately achieving highly efficient PL. The above PXRD findings are fully corroborated by HRTEM analyses. The (020) fringe spacing reduced from the initial 1.42 to 1.39 nm in Py-Da-4CH₃-COF, while the (110) lattice fringe spacing in Py-Da-2CH₃-COF and Py-Da-COF contracted from the incipient 2.47 to 2.19 and 2.24 nm, respectively (figs. S24 and S25). Collectively, these results provide direct evidence of pressure-induced lattice compression and stacking reorganization, revealing the structural tunability and pressure-responsive nature of the crystalline porous frameworks.

Fig. 4. Structural evolution under pressure.

PXRD patterns of (A) Py-Da-4CH₃-COF, (B) Py-Da-2CH₃-COF, and (C) Py-Da-COF before and after pressure treatment. Schematic models of irreversible structural transitions for (D) Py-Da-4CH₃-COF and (E) Py-Da-2CH₃-COF. Amplified (F) IR and (G) Raman spectra of Py-Da-4CH₃-COF showing vibration mode changes at 1 atm and after pressure was completely released. (H) Hirshfeld surfaces for the Py-Da-4CH₃-COF at 1 atm and after pressure was released from 7.9 GPa mapped with a d_norm distance, highlighting intermolecular contact modifications.

We then analyzed in situ high-pressure IR and Raman spectra up to 7.9 GPa to further explore the detailed structural characteristics of these three COFs (figs. S26 to S29). In Py-Da-4CH₃-COF, the signals at the range of 800 to 900, 2871, and 2958 cm⁻¹ were assigned to δ(C─H)_ring (the C─H out-of-plane bending vibration of phenyl), ν_s(C─H) (the C─H symmetric stretching vibrations of methyl group), and ν_as(C─H) (the C─H asymmetric stretching vibrations of methyl group) (fig. S26) (25, 26). Moreover, the spectral region between 1430 and 1550 cm⁻¹ corresponded to the C═C stretching vibration of the benzene ring. These bands located in the range of 800 to 900 and 1430 to 1550 cm⁻¹ gradually broadened and split upon compression, typically suggesting the enhanced π-π interactions (39, 40). Most vibrational peaks in IR spectroscopy exhibited a blue shift with increasing pressure, which originated from the enhancement of interatomic interactions. This promoted electron-phonon coupling. The enhanced π-π interaction and electron phonon coupling led to PL quenching and redshift under compression. There was a peak emerging at 822 cm⁻¹ in the high-frequency side of δ(C─H)_ring when the pressure was increased to 4.2 GPa, accompanied by peak splitting of 1460 and 2966 cm⁻¹ modes. These observations imply that the enormous distortion and deformation of the structure are responsible for altering these chemical environments of these bonds and disrupting the symmetry of these vibrational modes. We found that the ν_s(C─H) and ν_as(C─H) shifted toward lower wave numbers from 2871 to 2868 and 2958 to 2950 cm⁻¹ after pressure treatment, proving that the distorted structure under high pressure was preserved to ambient conditions (Fig. 4F). This represented the restriction of corresponding vibrations. It facilitated the reduction of nonradiative transitions and the enhancement of PL intensity, which is related to the large steric hindrance in the target structure. The evolution of the IR absorption bands in Py-Da-2CH₃-COF under high pressure was analogous to that of Py-Da-4CH₃-COF, that is, its structure underwent distortion and deformation when pressurized to 4.2 GPa (fig. S27). After pressure was completely released, the shift of ν_s(C─H) by 2 cm⁻¹ to lower wave numbers demonstrated that the distorted structure was retained at 1 atm (fig. S28). The factors accountable for piezochromic behavior and PL quenching in the two COFs mentioned above can also be generalized to Py-Da-COF (fig. S29). These phenomena aligned with the PXRD results after pressure treatment. In addition, in situ high-pressure Raman spectra displayed that the β(C─H) (the C─H in-of-plane bending vibration of phenyl) and ν(C═N) (C═N stretching vibration) bands of the pressure-treated Py-Da-4CH₃-COF were redshifted from 1062 to 1060 and 1624 to 1619 cm⁻¹ (Fig. 4G) (41, 42). The restriction of Raman bands is beneficial for reducing nonradiative energy dissipation and enhancing emission. Validation of these behaviors was further performed using the Hirshfeld surface of the system, where strong interactions are highlighted in red areas and weaker interactions in blue. After pressure treatment, the reduced red regions of π-π interactions in Py-Da-4CH₃-COF signify weaker π-π interactions (Fig. 4H). This confirms that the engineered sample with a quasi–AB stacking model renders pyrene groups with larger π electrons to occur a complete interlayer displacement. On the contrary, the enlarged red regions of π-π interactions in the pressure-treated Py-Da-COF suggest boosted π-π interactions, which arise from a decrease in interlayer distance (fig. S30).

Calculation and application after pressure treatment

To further visually elucidate the mechanism of the emission enhancement and magnitude of steric hindrance in Py-Da-4CH₃-COF, an interaction region indicator (IRI) analysis was used in this study. The function of the IRI is defined as following equation (43)

$$\text{IRI}\left(\mathbf{r}\right)=\frac{\mid \nabla \mathrm{\rho }\left(\mathbf{r}\right)\mid }{{\left[\mathrm{\rho }\left(\mathbf{r}\right)\right]}^a}$$

where ρ represents the electron density, r is the coordinate vector, a is an adjustable parameter, and a = 1.1 is adopted for standard definition of IRI. The sign(λ₂) denotes the sign of Hessian second largest eigenvalue of ρ, which are used to distinguish attractive and repulsive interactions. The magnitude of sign(λ₂)ρ is mapped on the IRI isosurface by different colors to vividly present the nature of the interaction region. We displayed the isosurface maps and scatter maps of Py-Da-4CH₃-COF in Fig. 5A, where the light red areas correspond to the presence of steric hindrance effects. In particular, the red region of recovered Py-Da-4CH₃-COF is more intense than those ones before pressure treatment, implying an increased steric hindrance effect after through pressure engineering (44). In the scatter maps, the higher point density in the red area of the pressure-treated Py-Da-4CH₃-COF further confirm the above conclusion. The larger steric hindrance contributes to increasing the potential barrier, thereby preventing the high-pressure metastable state from returning to the original stable state after pressure treatment. This played a crucial role in achieving emission enhancement of Py-Da-4CH₃-COF (Fig. 5B). The oscillation intensity after pressure treatment [2.18 arbitrary unit (au)] surpassed the initial value (0.017 au) by two orders of magnitude, indicating an enhanced transition dipole moment (Fig. 5C) (45). It is greatly associated with the improved PL intensity.

Fig. 5. Emission mechanism and device applications of the pressure-treated Py-Da-4CH₃-COF.

(A) The distribution of IRI isosurface maps and scatter plots of IRI versus sign(λ₂)ρ for Py-Da-4CH₃-COF at 1 atm and upon complete release of the pressure. vdW, van der Waals. (B) Illustration of steric hindrance. (C) The oscillator strengths of Py-Da-4CH₃-COF before and after pressure treatment. (D) The emission spectrum of the fabricated yellow pc-LED. The insets show the photographs of operating yellow pc-LED device. (E) CIE chromaticity coordinates of the fabricated yellow pc-LED. (F) Emission spectra of the yellow pc-LED devices at various operating currents from 10 to 90 mA. (G) Integrated PL intensity of the yellow pc-LED as a function of aging time. The inset displays visual time-quenching behavior.

Furthermore, a yellow pc-LED device was fabricated using 365-nm ultraviolet (UV) LED chip to illustrate the potential application and stability of pressure-treated Py-Da-4CH₃-COF (the insert of Fig. 5D). The fabricated pc-LED device exhibited favorable yellow light characteristics under a current of 30 mA with CIE chromaticity coordinates of (0.47, 0.47) (Fig. 5, D and E). The corresponding color temperature was 2969 K. This soft warm yellow light within the range of 2700 to 3000 K complies with road lighting standards and offers good protection for eyes. We further examined the stability of the yellow pc-LED using a low voltage (3 V) and various operating currents. As the current was raised from 10 to 90 mA, the steady increase in the emission intensity of yellow pc-LED demonstrated outstanding color stability (Fig. 5F). The yellow pc-LED exhibited high fatigue stability with slight changes in emission intensity for 72 hours at ambient conditions (Fig. 5G).

In summary, we developed a steric hindrance strategy by incorporating methyl groups into diamine linkers to precisely control the stacking configurations of pyrene-based COFs under pressure. The Py-Da-4CH₃-COF, with the highest steric demand, undergoes an irreversible transition from an initial slipped AA stacking arrangement to a more strongly slipped configuration approaching an AB-type stacking, driven by an increased phase transition barrier. This structural shift suppresses π-π interactions and C─H vibrations, yielding a record PLQY of 91.5%. In contrast, Py-Da-2CH₃-COF, with moderate steric hindrance, reverts to its original PL intensity due to a balance between a slipped AA stacking configuration and reduced interlayer distance, while Py-Da-COF suffers from emission quenching due to enhanced π-π interactions. Furthermore, we demonstrate the practical utility of pressure-treated Py-Da-4CH₃-COF by fabricating a pc-LED with exceptional chromaticity and fatigue resistance. This work highlights how synergistic molecular design and pressure engineering can unlock unprecedented PL performance in COFs, paving the way for advanced optoelectronic materials.

MATERIALS AND METHODS

Sample preparation

All starting materials and solvents were obtained from Jilin Chinese Academy of Sciences–Yanshen Technology Co. Ltd. and used without further purifications.

Py-Da-COF

The synthesis of Py-Da-COF was adapted from a previously described method, with modifications (46). In a typical synthesis, a mixture of Py-CHO (0.01 mmol, 6.18 mg), Da-NH₂ (0.02 mmol, 2.2 mg), 1,2-dichlorobenzene (2.7 ml), n-butanol (0.3 ml), and 12 M acetic acid (0.5 ml) was sonicated for 3 min in a 10-ml Pyrex tube and degassed by three freeze-pump-thaw cycles before vacuum sealing. Subsequently, the tube was heated at 120°C for 3 days. Upon complete cooling naturally to room temperature, the resulting precipitate was filtered and washed with tetrahydrofuran. Then, the sample was further extracted in methanol using a Soxhlet extractor for 10 hours. Ultimately, the powder was dried at 100°C in a vacuum overnight.

Py-Da-2CH₃-COF

In a typical synthesis, a mixture of Py-CHO (0.01 mmol, 6.18 mg), Da-2CH₃-NH₂ (0.02 mmol, 2.72 mg), 1,2-dichlorobenzene (2.7 ml), n-butanol (0.3 ml), and 12 M acetic acid (0.5 ml) was sonicated for 3 min in a 10-ml Pyrex tube and degassed by three freeze-pump-thaw cycles before vacuum sealing. Subsequently, the tube was heated at 120°C for 3 days. Upon complete cooling naturally to room temperature, the resulting precipitate was filtered and washed with tetrahydrofuran. Then, the sample was further extracted in methanol using a Soxhlet extractor for 10 hours. Ultimately, the powder was dried at 100°C in a vacuum overnight.

Py-Da-4CH₃-COF

In a typical synthesis, a mixture of Py-CHO (0.01 mmol, 6.18 mg), Da-4CH₃-NH₂ (0.02 mmol, 3.28 mg), 1,2-dichlorobenzene (2.7 ml), n-butanol (0.3 ml), and 6 M acetic acid (0.3 ml) was sonicated for 3 min in a 10-ml Pyrex tube and degassed by three freeze-pump-thaw cycles before vacuum sealing. Subsequently, the tube was heated at 120°C for 3 days. Upon complete cooling naturally to room temperature, the resulting precipitate was filtered and washed with tetrahydrofuran. Then, the sample was further extracted in methanol using a Soxhlet extractor for 10 hours. Ultimately, the powder was dried at 100°C in a vacuum overnight. Notably, we used 6 M acetic acid in this process, which slowed down the condensation rate and extended the time window for reversible bond exchange. This allowed the monomers with large steric hindrance to overcome repulsive interactions and gradually assemble into thermodynamically ordered frameworks, ultimately yielding Py-Da-4CH₃-COF with higher crystallinity.

In situ high-pressure measurements

We used a symmetrical diamond anvil cell (DAC) apparatus with a pair of 400-μm culet diamonds to carry out all high-pressure experiments. The sample was loaded into a 150-μm-diameter DAC chamber that was composed of a T301 stainless steel compressible gasket preindented to a thickness of 45 μm. The pressure determination was performed using the ruby fluorescent technique (47).

In situ high-pressure PL spectra were collected by the Ocean Optics QE65000 spectrometer. A 355-nm line of a UV diode-pumped solid-state laser was used for PL measurements. PL photographs of the samples were obtained using a camera (Canon EOS 5D Mark II) mounted on a microscope (Eclipse Ti-U, Nikon). In situ high-pressure IR absorption experiments were performed on a Nicolet iN10 microscope spectrometer (Thermo Fisher Scientific, USA) equipped with a nitrogen-cooled mercury-cadmium-telluride detector. KBr was used as the pressure transmitting medium for IR absorption measurements. We gathered in situ high-pressure Raman spectra using a spectrometer (iHR550, Symphony II, HORIBA Jobin Yvon) with an excitation laser of 785 nm and a power output of 0.5 mW.

Characterization

TEM images were obtained using a Thermo Fisher Scientific Talos F200i S/TEM operated at 200 kV, while HRTEM images were acquired on a Thermo Fisher Scientific Spectra 300 (S)TEM equipped with a double spherical aberration corrector at 300 kV. We subjected the COF powders to ultrasonic treatment in ethanol and then dropcast the supernatant onto the copper grids. Before imaging, it is necessary to perform beam alignment, center the aperture, and automatically compensate for the spherical aberration of the HRTEM. Specimens were located at ×2550× magnification and recorded at 245,000× for TEM. For HRTEM, we aligned the samples to the zone axis at 13,000× and captured high-resolution images in energy-filtered transmission electron microscopy mode using a Gatan K3 camera. The SEM images were measured on a Hitachi Regulus 8100. Nitrogen sorption isotherms were measured at 77 K using a Micromeritics ASAP 2020 plus analyzer. Before measurements, we carried out solvent exchange on the ~50-mg sample with tetrahydrofuran (3 × 20 ml) and acetone (3 × 5 ml) for 12 hours each and then activated it under vacuum at 100°C for 12 hours. BET surface areas were determined on the adsorption branch within the International Union of Pure Applied Chemistry (IUPAC)–recommended linear region (P/P₀ ≈ 0.05 to 0.30), validating the range by positive C constant and minimum χ². Pore size distributions were obtained using nonlocal density functional theory (N₂ at 77 K, carbon slit pore, and nonnegative regularization). We took total pore volume at P/P₀ = 0.99 and estimated micropore volume by the t-plot method. Gas uptake, surface area, and pore sizes are reported in cubic centimeters per gram (standard temperature and pressure), square meters per gram, and angstrom, respectively. Isotherm types and hysteresis loops follow the IUPAC (2020) classification. TGA were obtained on a SHIMADZU DTG-60 thermal analyzer under nitrogen flow (30 ml min⁻¹) using ~5 to 10 mg of sample, from 30° to 800°C at a heating rate of 10°C min⁻¹ in Al pans. We recorded the solid-state ¹³C NMR spectra by a Bruker AVIII 500-MHz solid-state NMR spectrometer. The PXRD data were collected using Cu Kα radiation (λ = 1.5418 Å) on the Rigaku R-AXIS RAPID II. The experimental data were recorded over a 2θ range of 2.0° to 40.0° with a 0.02° step size and 2 s per step. We measured the excitation spectra using a SHIMADZU rf6000 spectrometer. The PLQY was carried out using an integrating sphere on the HORIBA Scientific Canada QuantaMaster 8000 System #3715. First, we measured the spectrum of the blank substrate under 355-nm laser excitation. Subsequently, the powder sample was uniformly spread in the sample cell, and the corresponding spectrum was obtained under the same experimental conditions. Ultimately, we selected the appropriate wavelength ranges as the calculation ranges for the excitation and the emission spectra. We obtained the PLQYs based on the ratio of emitted-to-absorbed photon numbers. The wavelengths of 363 and 380 nm were generated by a 75-W xenon lamp light source through a monochromator for spectral separation. These excitation sources are the LPS 100 model from HORIBA Scientific.

Theoretical calculations

We analyzed the interactions in the crystal structure using the Multiwfn 3.8 program for IRIs based on reduced density gradients (48, 43). All IRI maps were rendered by the VMD 1.9.3 program (49). Calculations on the oscillation intensity were carried out using the Gaussian 09 software package. In the DFT B3LYP method, the Becke’s three-parameter hybrid function with the Lee-Yang-Parr nonlocal correlation, 6-31G(d) basis set were chosen to calculated all molecular structures (50–52).

Supplementary Materials

This PDF file includes:

Figs. S1 to S30

Tables S1 to S8