Introduction

The study of room temperature phosphorescence (RTP) in purely organic compounds has a long history1,2,3, yet it remains a topic of interest due to its critical applications in OLEDs, photocatalysis, optical sensors, and security encryption, just name a few. Early efforts primarily explored the underlying mechanisms of RTP, with a particular focus on benzophenone4, which exhibits phosphorescence in degassed or frozen solution, as well as in the crystal state5,6. Since the early 2010s, RTPs by Kim, Tang, Fraser, and others have been observed in a diverse range of organic compounds beyond benzophenone7,8,9,10,11,12,13,14,15. As its underlying principles have been elucidated, the development of purely organic phosphors with varied structural designs has accelerated, along with extensive research into their broad applications, particularly in data encryption and anti-counterfeiting15,16,17,18,19,20,21,22,23,24,25. For these security-related applications, the defining feature of RTP materials is their intrinsically long-lived afterglow, driven by weak spin–orbit coupling (SOC), which slows the excited state deactivation associated with RTP15,17,26,27,28,29. This distinctive “time-gated” emission offers a key advantage over conventional fluorescent systems, whose luminescence fades immediately upon removal of the excitation source30,31,32,33,34,35.

However, its promise, relying solely on long phosphorescence lifetimes, is often inadequate for robust data protection, as a single security layer remains susceptible to counterfeiting or unauthorized duplication31,32,33,34,35. To address this, additional protective features—especially stimuli-responsive capabilities—are essential. These properties enable reversible modulation of emission in response to external stimuli such as heat, solvent exposure, or mechanical stress3,36,37,38,39,40,41,42,43,44,45. When integrated with the materials’ long-lasting emission, such reversible emission shifts introduce a second layer of complexity, heightening the challenge of counterfeiting.

Despite the clear value of reversible RTP for advanced encryption, achieving phosphorescence-color switching in purely organic materials under mild and practical conditions remains challenging. Most reported systems rely on multi-dopant matrices or require changes in excitation wavelength, complicating both the fabrication process and practical implementation (Supplementary Fig. 1a)31,32,33. Moreover, in single-component systems, external stimuli (e.g., temperature, solvent, pH, mechanical force) often lead to irreversible structural disruption or quenched emission (Supplementary Fig. 1b)46,47,48,49,50,51,52,53,54,55,56,57. For instance, Li et al. developed a stimulus-responsive RTP system by embedding phosphorescent and fluorescent dyes in a PVA matrix, enabling afterglow color modulation from blue to green to orange through triplet-to-singlet Förster resonance energy transfer31. However, exposure to water vapor compromised the rigid environment essential for phosphorescence, resulting in severe emission quenching. Similarly, Song et al. demonstrated a single-component phosphorescent material capable of shifting from blue to green under mechanical stress48. However, this transformation came at the cost of a significant reduction in phosphorescence intensity—from 68.4% (blue) to 7.6% (green)—as the looser packing after grinding failed to adequately suppress molecular motion. Recently, Ogawa and co-workers have reported a series of single-component organic crystals that exhibit RTP-to-RTP color changes in response to external stimuli via molecular conformational rearrangement58,59,60,61,62. However, these systems often suffer from limited reversibility, low emission efficiency in one of the states, or require complex external triggers, thereby limiting their practicality and broader applicability.

Herein, we address these challenges by introducing an organic phosphorescent crystal, 1,1′-(2,5-dibromoterephthaloyl)bis(glutarimide) (BrGlu), which exhibits fully reversible, pseudopolymorph-dependent phosphorescence-to-phosphorescence color switching under mild external conditions (Fig. 1a)63,64,65. Under standard crystallization, BrGlu forms green-emitting crystals (G-crystal), while crystallization in the presence of CHCl3, yields blue-emitting solvent-inclusion crystals (B-crystal). This color change stems from syn–anti rotamer conformational rearrangements triggered by solvation. Syn/anti rotamers have commonly only a small impact on the emission properties66, but were shown to be relevant in the fluorescence process of the parent bromine-free TGlu series67. Structural analysis reveals that solvent-mediated hydrogen bonding flips the bromine–carbonyl orientation into a syn-rotamer, thereby increasing the triplet-state energy and resulting in a blue-shifted emission. Our system achieves reversible switching between G-crystal and B-crystal through two distinct pathways: dissolution–recrystallization in the presence of specific halogenated solvents and mild heating–driven solid-state desolvation. This transformation between well-defined conformational rotamers affords high photostability, high phosphorescence quantum yields (43% and 93% for B- and G-crystals, respectively), and complete reversibility. Unlike previously reported systems requiring complex fabrication or suffering emission quenching, BrGlu’s single-molecule design ensures reversible switching with high quantum yields (Supplementary Table 1). Quantum-chemical calculations further indicate a low energy barrier for this rotameric interconversion, allowing rapid and complete reversion to the original anti-rotamer upon mild heating. As desired for advanced data encryption, this color change is entirely reversible upon mild heating, facilitating repeated rapid switching between the two phosphorescent states.

Fig. 1: Phosphorescence color switching under mild external conditions for enhanced data encryption.
Fig. 1: Phosphorescence color switching under mild external conditions for enhanced data encryption.
Full size image

a Reversible conformational transition of BrGlu between G-crystal and B-crystal states via solvation and heating, leading to phosphorescence color switching. b Schematic representation of the enhanced data encryption process, requiring keys 1 and 2 for sequential decryption and encryption through external stimuli. c Chemical structure of BrGlu alongside its DFT-optimized structure.

On the basis of these findings, we have developed a proof-of-concept data encryption platform that takes full advantage of both BrGlu’s phosphorescence and its reversible emission switching (Fig. 1b). By intermittently applying solvent and mild heating, we achieve “advanced data encryption,” wherein information is encoded in both spatial patterns and emission lifetime in a manner that is not readily discernible through simple observation. Furthermore, the sub-millisecond phosphorescent lifetimes restrict visibility of the encoded signal to detection systems equipped for microsecond-scale readout, thereby adding another layer of security. This fully reversible, single-component RTP highlights the potential of stimulus-responsive organic phosphors for next-generation secure data storage and anti-counterfeiting applications.

Results

Synthesis

We designed and synthesized a purely organic phosphor, 1,1′-(2,5-dibromoterephthaloyl)bis(glutarimide) (BrGlu), incorporating two main design elements (Fig. 1c). Bromine atoms were here introduced to enhance SOC and promote intersystem crossing (ISC) from the lowest singlet state to triplet state manifold68. A sterically hindered acyl-glutarimide (Ac–Glu) moiety was introduced to suppress intramolecular motion and reduce π–π stacking in the solid state. This design aims to limit close intermolecular interactions that could facilitate nonradiative deactivation pathways, in particular exciton diffusion, towards defect sites; this was recently reviewed in the context of phosphorescent crystals3. Density functional theory (DFT) calculations of the optimized molecular geometry reveal that the glutarimide groups adopt a nearly orthogonal orientation relative to the dibromoterephthaloyl core (Fig. 1c, right), which minimizes orbital overlap between neighboring chromophores and supports efficient RTP.

Preparation and analyses of phosphorescent crystals

Crystals of BrGlu were grown from eight different solvents, resulting in diverse morphological and emissive properties (Supplementary Note 5 and Supplementary Fig. 2). Most solvents yielded green-emissive crystalline solids with varying forms, whereas CHCl3 produced a blue-emissive counterpart (Fig. 2a). Powder X-ray diffraction (PXRD) analysis supports these findings (Supplementary Fig. 3): crystals grown in CH2Cl2 (dichloromethane), ethyl acetate, tetrahydrofuran, and methanol exhibited nearly identical diffraction profiles, suggesting a common crystal structure. In contrast, the CHCl3-derived solid displayed a different PXRD pattern, indicating a distinct packing arrangement. These findings underscore the critical role of solvent selection in directing crystal structure, which in turn governs the resulting emission color—revealing a direct correlation between molecular packing and phosphorescent behavior.

Fig. 2: Structural characterization of BrGlu in two crystalline states.
Fig. 2: Structural characterization of BrGlu in two crystalline states.
Full size image

a Schematic showing the bromine–carbonyl conformational change in BrGlu when crystallized from CHCl3 (syn-rotamer) versus CH2Cl2 (anti-rotamer), together with fluorescent micrographs of the resulting crystals (scale bar = 2.0 mm). b, d Packing diagrams of BrGlu in the B-crystal and G-crystal states perpendicular to the c-a plane, revealing their ordered arrangements with interlayer distances of 7.58 Å and 6.01 Å, respectively. Hydrogen atoms are omitted for clarity. Red dots denote the centroids of the aromatic rings. c Key intermolecular interactions in the B-crystal state, highlighting hydrogen bonding distances between BrGlu and CHCl3 solvent molecules at 2.28 Å, 3.41 Å, and 4.09 Å. e, f Thermal analysis (TGA and DSC) indicates a 14.4% weight loss for B-crystal at 65 °C, absent in the G-crystal. g Raman spectra exhibit distinct peaks, emphasizing structural differences between the two states.

To elucidate the relationship between emissive behavior and structural variations, we performed single-crystal (SC) XRD analysis (Fig. 2b–d, Supplementary Figs. 4 and 5, and Supplementary Table 2). The blue-emissive crystalline solid (B-crystal, solvate form) incorporates CHCl3 molecules and crystallizes in the I 2/a space group (a = 16.9 Å, b = 10.3 Å, c = 15.1 Å, β = 120°, Fig. 2b, c). In contrast, the green-emissive crystal (G-crystal), grown from CH2Cl2, is solvent-free and adopts the P 21/c space group (a = 12.1 Å, b = 6.0 Å, c = 13.3 Å, β = 106°, Fig. 2d). Notably, both crystals exhibit relatively large interlayer distances—7.58 Å for B-crystal and 6.01 Å for G-crystal (measured perpendicular to the c–a plane)—supporting the design principle that sterically bulky Glu units hinder close molecular packing. This structural feature is expected to reduce intermolecular excitonic coupling3, thereby limiting triplet exciton diffusion and effectively enhancing solid-state phosphorescence; this will be discussed in more detail further down.

Thermogravimetric (TGA) and differential scanning calorimetric (DSC) analyses reveal a clear distinction between the B- and G-crystals with respect to solvent incorporation and thermal stability (Fig. 2e, f). The B-crystal exhibits a mass loss of approximately 14.4 wt% upon heating to ~65 °C, corresponding to the release of CHCl3 (Fig. 2e). This is consistent with a 1:1 BrGlu:CHCl3 solvate, supported by the calculated theoretical CHCl3 content of 18.8 wt% based on the molar masses of BrGlu (514.13 g/mol) and CHCl3 (119.38 g/mol). SCXRD confirms this stoichiometry, showing CHCl3 molecules occupying disordered but well-defined lattice sites, stabilized by C–H···O interactions.

In contrast, the G-crystal displays no observable mass loss and remains structurally stable until decomposition begins above 260 °C (Fig. 2f and Supplementary Fig. 6). This exceptional thermal stability was further validated by repeated thermal cycling experiments. When subjected to five cycles of heating and cooling between 0 and 200 °C, the G-crystal retained its structural integrity, as confirmed by PXRD analysis (Supplementary Fig. 7). These results demonstrate the BrGlu’s robustness under operational conditions relevant to reversible switching applications.

Raman spectroscopy (Fig. 2g) further underscores these structural differences. The C = O stretching mode shifts from 1749 cm−1 in the G-crystal to 1763 cm−1 in the B-crystal, correlating with a slight shortening of the average C = O bond (1.20 Å vs. 1.19 Å, Supplementary Fig. 8). This shift reflects bond reinforcement induced by CHCl3–lattice interactions, rather than classical hydrogen bonding. Additionally, a new peak at 1586 cm−1 appears exclusively in the B-crystal, which can be attributed to a change in the lattice leading to a local vibration of CHCl3–BrGlu interaction69. This peak reflects solvent-induced modulation of the lattice dynamics, not Br···O halogen bonding, which typically appears below 300 cm−1.

The B-crystal contains CHCl3 molecules that form hydrogen bonds with BrGlu (HCHCl3 to OBrGlu distances: 2.280 Å, 3.405 Å, 4.085 Å; Fig. 2c). These hydrogen bonds rearrange the local packing, affecting the orientation of the C = O and Br substituents (Fig. 3a)70. In the B-crystal, these groups largely point in the same direction (syn-rotamer; adjacent Br1 to O1 distance (da): 3.1 Å; other Br2 to O1 distance (do): 6.3 Å; carbonyl dihedral angle (θc): 36°). By contrast, in the G-crystal, an anti-rotamer of the C = O and Br substituents is found; da is 4.5 Å, do is 5.2 Å, and θc is 26°. While conformation-dependent RTP behavior is not widely reported, recent studies have shown that subtle conformational variations can affect triplet emission colors and rates60,61,62,63,64,65,70,71,72. In the BrGlu system, a single molecule exhibits a distinct conformational change between two emissive pseudopolymorphs, triggered by solvent inclusion. To explore this mechanism in detail, we performed further quantum calculations based on the crystal structures. Single point (SP) DFT calculations of the syn- and anti-rotamers in their x-ray crystal conformations reveal that the anti-form (i.e., the solvent-free G-crystal) is slightly favoured by about ΔG = 5.0 kJ/mol (52 meV; see Fig. 3b and Supplementary Table 3); thus, the occurrence of the syn-form in the B-crystal is evidently stabilized by chromophore-solvent interactions. We note that both syn- and anti-conformations found in the crystals are not very different from the fully DFT-optimized ones in vacuum, see Fig. 3a and Table 1. For the latter, syn is slightly more stable than anti (5.6 kJ/mol (58 meV) in Fig. 3b).

Fig. 3: DFT analyses of structural parameters and energetic relationships in B- and G-crystals.
Fig. 3: DFT analyses of structural parameters and energetic relationships in B- and G-crystals.
Full size image

a The syn-rotamer (B-crystal) and anti-rotamer (G-crystal) X-ray structures define the inner (θi) and outer (θo) torsional angles. Hydrogen atoms are omitted for clarity. See Table 1 for a comparison of the experimentally observed (X-ray) and DFT-optimized values in both the ground (S₀) and first triplet (T₁) states. b The relative Gibbs free energy ΔG (kJ/mol) of the B– vs. G-crystal is assessed via three approaches: (i) the SC-XRD geometry, (ii) a restricted DFT optimization (fixing the torsion angles to those found in SC-XRD), and (iii) a fully DFT optimization (free molecule).

Table 1 DFT analyses of structural parameters
Full size table

In summary, SC XRD, Raman, and thermal analyses confirm that CHCl3-solvated BrGlu contains solvent molecules involved in hydrogen bonding. Comparing the B-crystal and G-crystal (see Fig. 3 and Table 1) reveals substantial differences in substituent orientation: in the B-crystal, the C = O groups are co-aligned with the bromine atoms (syn-rotamer), whereas in the G-crystal they are oriented in the opposite direction (anti-rotamer).

Phosphorescence color switching under mild external conditions

We first examined the optical properties of BrGlu in Ar-purged CHCl3 solution at room temperature (RT, 298 K) and low temperature (LT, 78 K), as shown in Fig. 4 and Supplementary Fig. 9. The absorption spectrum exhibits a main peak at 327 nm; according to our time-dependent (TD) DFT results (see Supplementary Fig. 10), this is ascribed to the electronic transition to the lowest excited singlet state (S0→S1). The latter is essentially described by a one-electron excitation between the highest occupied and lowest unoccupied molecular orbitals (HOMO, LUMO). According to the calculations, the lowest excited triplet state (T1), which is not visible in absorption, is located 0.74 eV below S1; the T1 state exhibits a complex electronic description, which thus requires natural transition orbitals, NTOs (Supplementary Fig. 10). At RT, the solution exhibits neither fluorescence nor phosphorescence under steady-state conditions; in contrast, at LT conditions pronounced phosphorescence occurs at 492 nm, whereas no fluorescence could be detected (Supplementary Fig. 9). The results evidence that nonradiative deactivation, being highly effective at RT, is efficiently blocked in the frozen solution through the rigid environment10. Moreover, the results show that ISC from S1 to the triplet manifold (Tn) is highly efficient in BrGlu, largely exceeding the fluorescence channel. This is in line with the TD DFT calculations, which indicate energetically accessible triplet states with considerable SOC elements (see Supplementary Fig. 10); this is a result of the participation of the free electron pairs of bromide in the respective NTOs, such providing a significant heavy atom effect15,65. Consequently, consistent with its behaviour in frozen solution, BrGlu is expected to exhibit phosphorescence also in its distinct crystalline forms, as indeed observed, see Fig. 4 and Table 2.

Fig. 4: Optical and photophysical properties and TD-DFT analyses of B-crystal and G-crystal.
Fig. 4: Optical and photophysical properties and TD-DFT analyses of B-crystal and G-crystal.
Full size image

a Absorption (Abs; black line) and steady-state emission (ssEm; blue for B-crystal, green for G-crystal, red for solution), and delayed emission spectra of B-crystal (blue dots) and G-crystal (green dots) at room temperature under ambient condition. The B-crystal exhibits ΦP = 43% at λP = 480 nm, while the G-crystal shows ΦP = 93% at λP = 510 nm (λex = 330 nm). b Phosphorescence lifetimes for B-crystal (blue, 150 µs) and G-crystal (green, 220 µs) states (λex = 330 nm). c The transition energetic difference of G-crystal relative to B-crystal calculated adiabatic and vertical transition energies E00, Evert, for the fully optimized anti and syn conformations in the ground and first excited triplet states (S0, T1). d Representative NTOs of B-crystal (top) and G-crystal (bottom) at S1, T4, and T1. e, f Reversible phosphorescence color switching upon solvation by CHCl3 and heating: Photographic images under 365 nm irradiation and corresponding spectra show the transition from G-crystal to B-crystal states in CHCl3 while stirring. G-crystals (100 mg) were placed in a vial with a stir bar and 3 mL of CHCl3 was added. Upon heating at 60 °C, solvation is disrupted, causing the emission to revert to the G-crystal state. g CIE 1931 chromaticity diagram depicting emission color coordinates, demonstrating reversible color switching.

Table 2 Photophysical properties of BrGlu in the different phases
Full size table

Building on these observations, we next explored how solvation-driven structural reorganization leads to distinctly different photophysical properties (Fig. 4). Steady-state PL measurements reveal emission maxima at 480 nm for the B-crystal and 510 nm for the G-crystal; low-temperature measurements are consistent with the experiments at ambient conditions, maintaining the bathochromic shift from B to G (Supplementary Fig. 12). As we will see below, this spectral shift reflects the different rotameric conformations, see Fig. 1a. Time-resolved PL spectra (Fig. 4a, delay: 0.2–1.0 ms) closely mirror the steady-state profiles, confirming triplet-state emission, with lifetimes in the microsecond regime (Fig. 4b). The phosphorescence quantum yields (Φp) are 93% for the G-crystal and 43% for the B-crystal, with corresponding triplet lifetimes (τp) of 220 µs and 150 µs, respectively—demonstrating that both phases remain highly emissive despite the reduced efficiency of the B-crystal.

In order to rationalize the experimentally observed bathochromic shift of phosphorescence when going from B to G, we first conducted single-point TD-DFT calculations (i.e., giving the vertical transition energies Evert(S0→T1)) of the molecules in those conformations, which were found in the crystal structures (see Fig. 4c and computational part for details). According to the calculations, Evert(S0→T1) are somewhat higher in G-crystal compared to B-crystal (Supplementary Table 4), which seems to, at first glance, contradict the experiment. Similar results are found for Evert(S0→T1) on the fully relaxed syn and anti conformations, see Supplementary Table 5. Nevertheless, we note that the figure of merit for the phosphorescence process is not the vertical energy for absorption, but that for emission, i.e., Evert(T1→S0). Therefore, we optimized the T1 states, to calculate the latter; in fact, here the anti-conformation (G-phase) is found at 1.85 eV, i.e., 0.09 eV below the syn conformation (B-phase; at 1.94 eV), see Fig. 4c and Supplementary Table 6. This inversion is driven by the much larger structural relaxation in the T1 state of anti in comparison syn. The TD-DFT results on Evert(T1→S0) are in reasonable agreement with the experimentally observed shift from 2.43 eV (510 nm) for G to 2.58 eV (480 nm) for B, that is 0.15 eV. The agreement is in particular good, as an additional small hypsochromic shift for the B-phase is expected to arise from the lower polarizability due to dilution by the solvent molecules in the B-crystal73, which is not accounted for in the TD-DFT calculations. Finally, we rule out excitonic contributions to the spectral shift, as this becomes only significant for high oscillator strength (f) of the lowest excited state and small intermolecular separation3,74; in the BrGlu crystal, however, f of the emitting state is self-evidently very small due to its triplet nature, and, at the same time, intermolecular separations are large, as evidenced from the crystal structure (vide supra).

To clarify the excited state deactivation processes, we derived the radiative and nonradiative rates from T1 via kP = ΦP/ τP and knr = (1-ΦP)/ τP (Table 2) by assuming that the ISC efficiency ηISC from S1 to the triplet manifold is about unity. The latter assumption is, on one hand, consistent with the absence of fluorescence at any temperature, while the nonradiative deactivation from S1 is expected to be very small due to the restricted access to the conical intersection for such emitters64. On the other hand, the spin-orbit coupling matrix elements are high for ISC from S1 to the triplet manifold Tn, due to the very significant participation of the free electron pairs of Br in S1 (see Fig. 4d for representative NTOs and Supplementary Figs. 10 and 13 for details); therefore, this heavy-atom contribution enables efficient ISC via effective SOC. Deriving kp, notably, the B-crystal and G-crystal show both high kP values (2.9 × 103 s−1 and 4.2 × 103 s−1, respectively), which arise from effective SOC through participation of the free electron pairs of Br also in the description of the T1 state, see Supplementary Fig. 13. This can be directly evidenced from quantum chemistry; for this, we calculated the oscillator strength f(T1) by TD-DFT under inclusion of SOC, and subsequently estimated kP via the Strickler-Berg equation3,75 (see Supplementary Note 10). This yields ca. 1.0 × 103 s−1 for B and 3.4 × 103 s−1 for G, being in reasonable qualitative agreement with the experimental values; in particular, it confirms the larger value for kP in the G-crystal in comparison with B. The main factor for the enhancement of kp in G vs. B is essentially due to the higher f(T1) in G, indicating stronger SOC of T1 with S0 through the heavy atom effect of bromine.

The non-radiative rates knr from T1 are notably low ( = 3.8 × 103 s−1 for B-crystal, 0.32 × 103 s−1 for G-crystal). On one hand, phosphorescence quenching may be due to nonradiative recovery of the ground state S0 through inefficient reverse ISC03; the latter is commonly proceeding through a T1/S0 crossing point (CP). On the other hand, quenching may be due to trapping (in particular at interfaces/surfaces)63. These traps are reached by triplet exciton diffusion through long exciton lifetimes and effective exciton coupling via an exchange mechanism, which thus requires orbital overlap3. The latter is effectively suppressed in the BrGlu crystals due to the larger interlayer distances dIL of 7.58 Å in B vs. 6.01 Å in G as induced by the sterically bulky acyl-glutarimide groups. The larger knr for the B-crystal compared to G thus clearly evidences that trapping is not the predominant quenching mechanism, as the larger dIL in B vs. G should slow down exciton coupling (and thus trapping) even more and not vice versa. Instead, the enhanced knr for B vs. G must be ascribed to more effective nonradiative ground state recovery via ISC0. In fact, although the densities of the B-and G-crystals hardly differ (see Supplementary Table 2), the presence of the solvent molecule in the B crystal may give BrGlu more structural flexibility in B vs. G, and thus facilitate opening the path to the T1/S0 CP. Additionally, the higher-lying T1 state in B compared to G should further promote the access to the CP, in analogy to the internal conversion (via a conical intersection) in the TGlu series64.

Importantly, the transition between the two phosphorescent phases is fully reversible. When G-crystal (100 mg) samples are suspended in 3 mL of CHCl3, they dissolve and recrystallize into the B-crystal over approximately 60 min (Fig. 4e, f and Supplementary Fig. 14). Photographic images in Fig. 4e under 365 nm irradiation and corresponding spectra show the transition from G-crystal to B-crystal states in CHCl3 while stirring. Subsequent heating to ~60 °C removes the solvent, as confirmed by TGA and DSC, thereby restoring the original green emission. By contrast, exposure of the G-crystal to CHCl3 vapor for up to 2 days did not induce any phase transformation (Supplementary Fig. 15), indicating that solvent-assisted recrystallization from solution is necessary for pseudopolymorph interconversion. PXRD analyses of the post-heating sample further validate the transition between the B- and G-crystal states (Supplementary Fig. 7), with the final green-emissive material reverting to the same packing structure as the initial G-crystal. This complete reversibility is also illustrated by the emission switching in the CIE 1931 color space—from (0.216, 0.513) to (0.148, 0.257) (Fig. 4g)—which remains stable through at least ten solvation–desolvation cycles (Supplementary Fig. 16), with minimal variation in emission intensity ( < 5%) and λmax (<1 nm). In addition, both crystal forms exhibit excellent environmental stability under ambient conditions (25 °C, 30% relative humidity). The G-crystal maintains a consistent emission profile over 7 days, indicating high structural robustness (Supplementary Fig. 17a). For the B-crystal, the emission remains largely unchanged during the initial 24 h, after which a gradual red-shift in λmax is observed, corresponding to a slow conversion to the G-phase as CHCl3 slowly evaporates (Supplementary Fig. 17b). These results confirm the high durability, reversibility, and practical reliability of the BrGlu system, supporting its potential for real-world applications in optical data storage and anti-counterfeiting technologies.

The facile conversion between the B- and G-phase is rationalized by DFT relaxed scans around the inner torsional angle θi (C1 − C2 − C4 − N1), which converts the syn-rotamer (B-phase) into the anti-rotamer (G-phase), see Fig. 5; here, we additionally allowed for the torsion around the outer torsional angle θo (C1 − C2 − C4 − N1) to release the steric constraints. A low-energetic path on the potential energy surface is evident from Fig. 5, which governs the transformation via a saddle point (X) at θi = 98° and θo = 109° with a barrier height of about 34 kJ/mol (0.35 eV), which explains why reversible interconversion of the two phases is rather easily accomplished in this material.

Fig. 5: DFT-calculated energy landscape for the interconversion of syn-rotamer (B-phase) and anti-rotamer (G-phase).
Fig. 5: DFT-calculated energy landscape for the interconversion of syn-rotamer (B-phase) and anti-rotamer (G-phase).
Full size image

a 2D contour map indicates the Gibbs free energy (kJ/mol) over a range of inner (θᵢ) and outer (θₒ) torsional angles, whereas (b and c) portray 3D perspectives of this same surface, highlighting the key saddle point and the distinct minima corresponding to each rotamer. Blue marker denotes the syn-rotamer, while green marker denotes the anti-rotamer.

To further explore the solvent dependence of solvation-induced phosphorescence color switching, we investigated several halogenated solvents—including CDCl3, CHBr3, C2H2Cl4, C2H4Cl2, and CCl4; among them, besides CHCl3 (vide supra), only CDCl3, CHBr3, and C2H2Cl4 triggered the solvation–desolvation cycle (Fig. 6a–d). By contrast, solvents with negligible or no hydrogen-bond donation (e.g., C2H4Cl2 or CCl4) failed to induce a conformational reorganization of the crystal lattice, underscoring the pivotal role of hydrogen bonding in crystal dissolution, re-nucleation, and thus the phosphorescence color change. Interestingly, the solvation rates decrease in the following order, CDCl3 (10 min) > C2H2Cl4 (20 min) > CHCl3 (60 min) > CHBr3 (120 min), reflecting the interplay of isotope effects, solvent polarity, and molecular size. For instance, deuterated chloroform (CDCl3) accelerates solute–solvent interactions more effectively, whereas the increased number of chlorine atoms (electron-withdrawing groups) in C2H2Cl4 enhance its polarity, which partially compensates for its higher mass. Conversely, the larger van der Waals volume of CHBr3 slows down its diffusion into the G-crystal, thereby extending the time required for hydrogen-bonding interactions to form and resulting in the longest overall solvation time.

Fig. 6: Photophysical properties of BrGlu in halogenated solvents before and after solvation.
Fig. 6: Photophysical properties of BrGlu in halogenated solvents before and after solvation.
Full size image

Solvation-driven phosphorescence transitions of BrGlu in halogenated solvents: CDCl3 (a), C2H2Cl4 (b), CHBr3 (c), CCl4, and C2H4Cl2 (d). Photographic images under 365 nm irradiation and corresponding spectra illustrate the transition from the G-crystal to the B-crystal state in each solvent while stirring under ambient condition (λex = 330 nm). G-crystals (100 mg) were placed in a vial with a stir bar, and 3 mL of the corresponding solvent was added.

Furthermore, electrostatic potential (ESP) calculations (Supplementary Fig. 18) revealed that CDCl3, CHBr3, and C2H2Cl4 contain sufficiently “activated” hydrogen atoms capable of hydrogen bonding with BrGlu. In contrast, solvents such as C2H4Cl2 or CCl4 either possess minimal hydrogen-bonding capability or lack hydrogen entirely. Additional tests with polar protic and aprotic solvents—including methanol, ethanol, and acetonitrile—also failed to induce any phase transition, showing only minimal changes in PL spectra (Supplementary Fig. 19). These results suggest that the phase transition is governed not by polarity or hydrogen-bonding capability alone, but by a combination of factors, including ESP distribution, molecular size, and diffusivity, as summarized in Supplementary Table 7. PXRD and Raman spectra (Supplementary Figs. 20 and 21, respectively) confirm that each of these halogenated solvents with activated hydrogen induces a distinct solvated crystal form relative to the unsolvated G-crystal. Supporting this, TGA (Supplementary Fig. 22) and DSC (Supplementary Fig. 23) measurements on the solvated crystals with CDCl3-, C2H2Cl4-, and CHBr3 show mass losses in the range of 67–71 °C, indicating that solvent acidity, polarity, and molecular size collaboratively influence the solvation/de-solvation behavior and associated emission characteristics.

Advanced data encryption

Based on BrGlu’s distinctive capability to toggle between two phosphorescent states, we engineered a multilayer encryption scheme built on two decryption keys (key 1 and key 2) for secure data access (Fig. 7a). In this configuration, B-crystal ink functions as the authentic cipher, whereas the FL dye ink adds a deceptive layer of visible camouflage. Both inks were formulated by dispersing the respective luminescent materials in mineral oil to form stable, solid-state dispersions suitable for precise patterning. The molds used for patterning were fabricated using a PDMS casting method, and hand-cut to produce features down to ~2 × 8 mm (see Supplementary Note 9 and Supplementary Fig. 24).

Fig. 7: Advanced data encryption platforms.
Fig. 7: Advanced data encryption platforms.
Full size image

a Schematic overview of a multilayer encryption system employing heat and long-lasting phosphorescence imaging for controlled decryption. b Under 365 nm UV light, a misleading blue display (“8888”) appears, concealing the true message. Heating (~ 60 °C, key 1) and cooling induce a phase transformation to the G-crystal. Turning off UV (key 2) sustains the green phosphorescence and revealing “UM.” Scale bar: 1 cm. c, d A complementary “3D” encryption method capitalizes on the distinct solvation behaviors of CCl4, CDCl3, and CHCl3 (key 1), while controlled time intervals (key 2) enable stepwise decryption. The binary color code is sequentially converted into ASCII and then into text. A brief heating step removes solvent residues, restoring the initial green state and allowing iterative encryption-decryption cycles. Scale bar: 2 cm.

In Fig. 7b, we observe that under 365 nm UV light, the visible pattern appears as the spurious “8888” in blue (Supplementary Movie 1). Heating (~60 °C, key 1) briefly suppresses BrGlu’s phosphorescence through increased molecular vibrations, distorting the display. Upon cooling, BrGlu transitions to its G-crystal phase, glowing blue and green mixed message. Turning off the UV lamp (key 2) maintains this green long-lasting phosphorescence emission, confirming the correct information, being “UM” (Supplementary Movie 2). Notably, this thermal cycle is fully reversible, enabling repeated encrypt-decrypt operations.

Additionally, a “3D” encryption strategy (see Fig. 7c, d, Supplementary Note 9 and Supplementary Movies 3–5) augments the security by harnessing the pseudopolymorpic kinetics of CCl4, CDCl3, and CHCl3 (key 1 for spatial color code), alongside controlled time intervals (key 2). A 1 × 1 cm grid was prepared using a silicone well plate to define pixel positions for selective solvent exposure. By segmenting G-crystal regions and exposing them to different solvents, TD phosphorescence color switching from green (0) to blue (1) was achieved, effectively generating a binary optical code.

At time zero, the entire grid remains green. After 10 min, only the CDCl3-exposed region converts to blue, corresponding to the ASCII-encoded message “GOBLUE.” Further exposure (e.g., 60 min) results in undesired color changes and scrambled output. A brief heating step effectively removes residual solvent, restoring all pixels to the original green-emitting G-crystal state, and enabling system reset for repeated use. This layered encryption framework underscores the synergy of thermal modulation, solvent-mediated color switching, and phosphorescent emission for secure data storage.

Collectively, these demonstrations establish BrGlu’s significant adaptability for complex data encryption protocols. By selecting specific solvents, timing, and thermal treatments, our platform delivers multiple security layers—from immediate visual decoys to time-gated or thermally controlled readouts—thereby showcasing the potential of reversible phosphorescence for sophisticated, stimuli-responsive encryption technologies.

Discussion

This study presents a single-component organic phosphor, BrGlu, that exhibits fully reversible RTP color switching between bright blue (B-crystal) and green (G-crystal) states upon mild external conditions, such as solvent exposure and gentle heating. Structural and photophysical analyses show that the forward transformation from G-crystal to B-crystal occurs through a solvent-triggered dissolution–recrystallization, while the reverse transformation from B-crystal to G-crystal proceeds by solid-state desolvation upon mild heating. These processes involve syn–anti conformational reorientation of the bromine–carbonyl substituents stabilized by hydrogen bonding. Computational analysis reveals a low energy barrier (34 kJ/mol) for this interconversion, enabling rapid and complete reversibility under mild conditions. Notably, both crystalline states exhibit exceptionally high phosphorescence quantum yields—93% for the G-crystal and 43% for the B-crystal—without significant quenching or lifetime degradation. These results highlight the effectiveness of our molecular design, which integrates heavy-atom-assisted ISC with a sterically rigid Ac–Glu framework to suppress nonradiative decay pathways. Utilizing its characteristic combination of microsecond-level long-lasting and robust, repeatable color switching, we developed sophisticated encryption strategies that greatly enhance data security in practical encryption and anti-counterfeiting technology. These findings underscore the transformative potential of stimulus-responsive RTP materials in real-world encryption scenarios, offering both practical and secure solutions for emerging information-protection needs.

Methods

Optical spectroscopy

To study the optical properties of BrGlu in solution, samples were prepared in spectroscopic-grade CHCl3 (Aldrich Chemical Co.) at a concentration of 1 × 10–5 M. To prevent phosphorescence quenching by O2, each solution was deaerated via three freeze–thaw cycles: the solution was frozen in liquid nitrogen, placed under vacuum for 10 min, then allowed to thaw. UV–visible absorption spectra were recorded with a Varian Cary 50 spectrophotometer. Photoluminescence (PL) emission spectra of both solutions and crystals were collected on a Photon Technologies International QuantaMaster spectrofluorometer (QM-400), equipped with a continuous 75 W Xe lamp. Emission was detected at a 90° angle using a Horiba PMT-928 photomultiplier tube; the spectra were corrected for the λ-dependence of the detector. For solid-state measurements, crystals were placed in Wilmad quartz NMR tubes (Aldrich Chemical Co.) on a quartz mount. Low-temperature measurements were carried out by immersing the quartz mount in liquid nitrogen. Except for low-temperature experiments, all other measurements of crystalline samples were conducted under ambient atmospheric conditions. For color-switching experiments (e.g., Fig. 4e, f), approximately 100 mg of G-crystals were placed in a Hellma fluorescence cuvette (QS grade) with 3 mL of anhydrous (non-deaerated) CHCl3 and stirred at room temperature. The reversible transition to B-crystal was monitored over time by PL spectroscopy under ambient air. Time-resolved PL measurements were performed on the same spectrofluorometer using a xenon flash lamp as the excitation source. Lifetimes were extracted by mono- or bi-exponential fitting (as appropriate) through deconvolution using the instrumental response function with the Felix GX software. PL quantum yields were determined using a calibrated integrating sphere attached to the spectrometer, with data measured against a blank sample. Each measurement was performed four times on freshly prepared samples to ensure consistency. All measurements showed high repeatability, and errors are reported as ±1 standard deviation. To verify measurement reliability, rhodamine 6G and coumarin 153 were studied as reference dyes following standard literature protocols76.

Crystal preparation

Crystals of BrGlu were grown by slow solvent evaporation (3 mg/mL) at ambient conditions using various organic solvents. CHCl3 and CH2Cl2 yielded single crystals within 48 h, referred to as B- and G-crystals, respectively. The B-crystals from CHCl3 exhibited blue phosphorescence, while G-crystals from CH2Cl2 showed green emission. Both formed well-defined crystals with uniform sizes ranging from 0.1 to 0.5 mm, as confirmed by optical microscopy (Supplementary Fig. 2).

In contrast, ethyl acetate, THF, and acetone generated larger crystals—typically a few millimeters across—also emitting green phosphorescence. On the other hand, toluene, ethanol, and methanol produced thin, needle-shaped crystals with smaller dimensions, yet still maintained visible green phosphorescence under UV light (Supplementary Figs. 2 and 3).

Structural analysis

PXRD patterns were measured using a Rigaku MiniFlex 600 with Cu Kα radiation, operating at 40 kV and 15 mA. X-ray detection was performed using a D/TeX Ultra2 one-dimensional silicon strip detector. PXRD data were collected over a 2θ range of 2–50° with a scan rate of 0.0100° per step. SC XRD data were collected using a Rigaku XtaLAB Synergy-S X-ray diffractometer with a kappa goniometer geometry configuration. The X-ray source is a PhotonJet-S microfocus Cu source (λ = 1.54184 Å) operated at 50 kV and 1 mA. X-ray intensities were measured with a HyPix-6000HE detector held 34 mm from the sample. The data were processed using CrysAlisPro v40.82 (Rigaku Oxford Diffraction) and were corrected for absorption. The structures were determined and refined using OLEX277 v1.5-ac5-024 with SHELXT78 and SHELXL79. All non-hydrogen atoms were refined anisotropically with hydrogen atoms located in idealized positions. Crystallographic data for BrGlu have been deposited in the Cambridge Crystallographic Data Centre (CCDC) under deposition numbers 2431976 (B-crystal) and 2431977 (G-crystal). ORTEP-style illustrations with probability ellipsoids and detailed notes on A- and B-level alerts are provided in Supplementary Fig. 26 (B-crystal) and 27 (G-crystal). Raman spectra were measured using a Renishaw Qontor Raman spectrometer (1800 lines/mm grating) with a 100 mW 532 nm laser through a 50× Leica long working distance microscope objective. Static scans (2010–270 cm−1) were collected by focusing a single crystal of each sample with 10 s exposure time for 20 acquisitions using cosmic ray removal at 1% laser power. The collected data was smoothed, truncated (2000-900 cm-1), baseline-corrected, and normalized via WiRE 5.6. The single crystal of BrGlu solvated in CDCl3 was measured with static scans (2732 ~ 1164 cm−1) following the same procedure. The obtained spectra were truncated (2400 ~ 1700 cm−1), baseline-corrected, and normalized via WiRE 5.6.

Thermal Analysis

Thermogravimetric analysis (TGA) was performed on a TGA 5500 Discovery thermal analyzer (TRIOS 5.6). Approximately 1 mg of sample was placed in a platinum TGA pan and heated from 25 °C to 600 °C at a rate of 5 °C/min. Differential scanning calorimetry (DSC) measurements were carried out on a Q20 DSC instrument. About 2 mg of sample was sealed in a Thermal Support aluminum hermetic pan with a lid, which was then pierced using a BD PrecisionGlide™ Needle (26G × 5/8). The DSC instrument was calibrated with an indium standard using the same pan and pierced lid. For each run, the sample was first cooled to 0 °C, then heated to 200 °C at 5 °C/min, and subsequently cooled back to 0 °C at the same rate. This heating–cooling cycle was repeated five times.

Quantum-chemical calculations

Ground and excited state (S0, S1) optimization, MO and NTO analysis, and SP TD calculations were done at the DFT level in vacuum, using the B3LYP functional and 6–311 G* basis set, as defined in the Gaussian 16 program package80. The SP TD-DFT calculations were either performed (i) directly on the molecules taking from the X-ray structures, (ii) under restricted optimization, where the torsional angles θi, θ0 were fixed to those found in the X-ray structures, and (iii) on the fully optimized structures in syn- and anti-conformations, respectively. MO topologies were plotted with ChemCraft, using a contour value of 0.02. Additionally, SOC analyses, as well as the calculation of the oscillator strengths f(T1), were conducted in Orca 6.0.0, utilizing the B3LYP functional and 6–311 G* basis set, in vacuum, based on the ground-state geometry previously optimized in Gaussian 1681. Specifically for the f(T1) calculation, the structure used was the crystal geometry (ground-state), which was subsequently optimized to the T1 state. Rate constants for phosphorescence kP were estimated from the Strickler-Berg equation (see Supplementary Note 10 for details)75, using an approximated form suitable for quantum-chemical output (utilizing f(T1) and Evert for absorption and emission)3, under consideration of the state degeneracy, and inserting a refractive index of n ≈ 2 to account for the crystal environment in the spectral region of emission73,82.

Fabrication of data encryption platform

To demonstrate the practical applicability of our system for data encryption, we fabricated luminescent patterns using two complementary approaches with distinct spatial and temporal encoding strategies.

For Fig. 7a, b, a PMMA mold was prepared using Silgard 184 silicone elastomer. The base and curing agent were mixed in a 10:1 weight ratio, degassed in a desiccator for 10 min to remove air bubbles, and then cured in an oven at 110 °C. The mold was manually cut into the desired pattern (see Supplementary Fig. 24), with typical feature dimensions of approximately 2 × 8 mm. The encryption ink was formulated by dispersing the phosphorescent BrGlu material in mineral oil to increase viscosity and improve durability against solvent-induced degradation. The ink was applied using a micropipette to fill the mold cavities precisely.

For Fig. 7c, d, a 1 × 1 cm grid array was constructed using a silicone well plate sealed with a silicone mat. G-crystal suspensions were selectively deposited into individual wells, and time-resolved phosphorescence switching was triggered by exposure to different solvents: CCl4 (no switching), CDCl3 (~10 min), and CHCl3 (~60 min) (see Supplementary Fig. 25). After solvent application, the plate was sealed and incubated to allow controlled solvent diffusion and rotameric conformational switching.

Afterglow images and Movies (Fig. 7b and Supplementary Movies 1–5) and dynamic pattern evolution were recorded using a high-frame-rate camera (1000 fps, ISO 12800, shutter speed 1/12800), enabling real-time visualization of emission changes.

These demonstrations confirm that the system can encode multistage, time-gated information with moderate spatial resolution ( ~ mm scale) and maintain high durability through repeated solvent exposure. Further refinement using microfluidics, photopatterning, or inkjet printing techniques could enhance pixel resolution.