Continuous-wave organic laser enabled by monolayer molecular crystal

Abstract

Organic lasers, by virtue of their inherent flexibility, cost-effectiveness, and tunable wavelength, offer potential applications in wearable optoelectronics, biocompatible sensing, and full-colour flexible display. However, continuous-wave (CW) solid-state organic lasers have been obstructed owing to the nonradiative losses induced by triplet exciton accumulation as well as the photo-thermal degradation of organic materials under continuous optical excitations. Here, we have accomplished room-temperature CW organic lasing by employing a monolayer organic molecular crystal (MOMC) as the gain medium. The J-aggregate configuration in the MOMC induces super-radiant emission with an ultrashort radiative lifetime, which outcompetes the intersystem crossing (ISC) timescale, thus effectively suppressing the triplet exciton accumulation. Furthermore, the thermal stability of the MOMC is enhanced by encapsulating it within a hexagonal boron nitride microdisk, which serves as both the optical resonator and the thermal management material. Consequently, we obtain CW organic lasing at room temperature, featured by the reduced emission linewidth, threshold behaviour, as well as the build-up of linear polarization and coherence. Our findings break the long-term challenge in the organic semiconductor community and pave the way for the practical applications of organic lasers.

Continuous-wave (CW) organic lasers remain challenging due to triplet accumulation and thermal degradation. Here, authors employ monolayer organic molecular crystal gain medium within hexagonal boron nitride microdisk to overcome those difficulties, enabling CW organic lasing at room temperature.

Introduction

Since the first demonstration, semiconductor lasers have emerged as the focus of optoelectronics^1–7, enabling wide applications in high-speed optical communications, advanced laser display, ultraprecision manufacturing, and minimally invasive biomedical therapeutics. Compared to their inorganic counterparts, organic semiconductors exhibit unique advantages including cost-effectiveness, low-temperature solution processibility, inherent mechanical flexibility, and seamless integration with flexible substrates, making them promising candidates as the gain media for ultralow-threshold laser architectures^8–18. Moreover, the structural diversity in molecular design and controllable supramolecular packing configurations facilitate the energy-level engineering for achieving population inversion and lasing¹¹. However, despite the exciting progresses in full-colour organic laser displays¹³ and electrically pumped organic lasers^12,18, so far, continuous-wave (CW) solid-state organic lasers still suffer from the “lasing death”, which remains as a long-term challenge in this field¹⁹.

The realization of CW organic lasers has been fundamentally limited by two critical issues. First, the singlet excitons in organic materials can be converted into long-lived triplet excitons (~microsecond lifetime) via intersystem crossing (ISC) processes^20–22. With the CW optical excitations, the triplet excitons will be accumulated with time, leading to triplet absorption losses or singlet-triplet annihilations, which impede the occurrence of CW lasing^23–26. Secondly, organic materials also suffer from their low thermal conductivity²⁷ (typical <1.0 W m⁻¹ K⁻¹) and photostability²⁸. Under high-power CW optical pumping, the temperature rising in the organic gain medium could reach 100 K²⁹, resulting in irreversible material degradation that terminate lasing. By rational molecular engineering and optical resonator design, organic lasers now could operate in the quasi-CW (qCW) regime with long pulse durations ranging from milliseconds to seconds^30–32. However, further optimization of triplet related gain losses and thermal management is required towards the true CW organic lasers with extended operation lifetime.

Recently, the emergence of J-aggregated monolayer organic molecular crystal (MOMC) has attracted significant attention^33–40. The self-assembly of single-layer molecular monomers into J-aggregate configuration leads to super-radiant emission from the MOMC at room temperature, featuring high photoluminescence (PL) quantum yield (PLQY) close to 100% and shortened radiative lifetime down to 27 ps³³. In this case, the radiative decay of singlet excitons becomes much faster than ISC, which largely suppresses the generation and accumulation of triplet excitons. On the other hand, in terms of stability, it has been found that the photostability of the J-aggregated MOMC could be well improved using two-photon excitations with suppressed exciton-exciton annihilation³⁵. Alternatively, it is also possible to optimize the photostability of the MOMC by encapsulating it using transparent dielectric materials with high thermal conductivity, such as hexagonal boron nitride^41–44 (hBN, with thermal conductivity^45–47 up to 750 W m ⁻¹ K ⁻¹).

Here, we demonstrate room-temperature CW organic lasing by employing a J-aggregated MOMC (dimethyl-3,4,9,10-perylenetetracarboxilic diimide, in short Me-PTCDI) as the gain medium. The Me-PTCDI monolayer was encapsulated within an hBN microdisk, which serves as not only the optical resonator but also the thermal management material. Under a CW excitation, lasing from the Me-PTCDI monolayer is observed, with clear evidence including linewidth narrowing, threshold behaviour, and the build-up of linear polarization and coherence. Meanwhile, benefitting from the ultralow threshold (1.4–3.4 W/cm²) and excellent thermal management, the operation half lifetime of the organic laser under CW excitations extends from several minutes to 8 h. Our study thus overcomes the longstanding challenge of CW lasing based on organic semiconductors, which represents a breakthrough towards the practical applications of organic lasers.

Results

Selection of the gain medium

Because of the different in-plane and out-of-plane coupling, the monolayer Me-PTCDI and its bulk crystal feature distinct spectral characteristics. With only in-plane coupling, J-aggregated monolayer Me-PTCDI exhibits dominant 0–0 electronic transitions at ~550 nm, yielding bright green-color emission in wide-field fluorescence imaging (Supplementary Fig. 1a-b). Conversely, as a result of out-of-plane coupling, bulk Me-PTCDI crystals manifest interlayer charge-transfer excitons that hybridize with Frenkel excitons, inducing spectral broadening and red-shifted emission (Supplementary Fig. 1c).

Besides the spectral signatures, the monolayer and bulk Me-PTCDI crystals also exhibit distinct excited state kinetics (Supplementary Fig. 1d). Figure 1a shows the transient absorption (ΔT/T) spectra of monolayer Me-PTCDI, with a clear photoinduced bleaching (PIB) signal at 547 nm, corresponding to the 0–0 transition. With a mono-exponential fit to the decay curve (Fig. 1b), we obtained a lifetime of 1.1 ± 0.1 ps for monolayer Me-PTCDI. This ultrafast decay rate (> ISC rate) effectively suppresses triplet accumulation and associated gain losses, establishing monolayer Me-PTCDI as a promising gain medium for CW laser. In stark contrast, bulk Me-PTCDI crystals exhibit much slower kinetics extending until nanosecond timescale (Fig. 1c), which could be divided into three steps (Supplementary Fig. 2), including decay of singlet excited state (0–3 ps), intersystem crossing (3–17 ps) and decay of triplet excited state (>17 ps). Figure 1d shows the ΔT/T spectra in the selected time window of 3–17 ps, where we observed a rising photoinduced absorption (PIA) signal originating from triplet excited state generated via ISC. In this case, the triplet accumulation fundamentally impedes stable CW lasing based on bulk crystals. Consequently, we selected monolayer Me-PTCDI as the gain medium for subsequent studies.

Fig. 1. Excited state kinetics of the gain medium.

a Transient absorption (ΔT/T) spectra of monolayer Me-PTCDI. b Excited-state decay kinetics of monolayer (1 L) Me-PTCDI (extracted from 547 nm). Red curve: exponential fit (t = 1.1 ± 0.1 ps). Inset: proposed quasi-four-level lasing mechanism of monolayer Me-PTCDI with suppressed ISC. E1-E4: four energy levels. S₀: singlet ground state. S₁: singlet excited state. T₁: triplet excited state. ISC: intersystem crossing. c Transient absorption (ΔT/T) spectra of bulk Me-PTCDI. d ΔT/T spectra of bulk Me-PTCDI with selected delay time (3, 5, 10, 15 and 17 ps). The black arrow indicates the increasing PIA of triplet excited state generated via ISC. PIA: photoinduced absorption.

Design of the hBN-based microdisk resonator

Towards organic lasers under CW operations, we designed the hBN-based microdisk resonators as illustrated in Fig. 2a, which could support whispering gallery mode (WGM) lasing^48–51. The hBN serves not only as a low-loss photonic platform^51,52 but also provides an atomically smooth surface for the epitaxial growth of single-crystal Me-PTCDI monolayers. Moreover, the organic monolayer is sandwiched between two hBN microdisks to achieve a better optical mode overlap, such that the larger modal gain will facilitate the lasing of organic monolayers. By combining dry-transfer, physical vapor deposition (PVD), microsphere mask, and reactive ion etching (RIE), we have fabricated the hBN microdisk resonators on 300 nm SiO₂/Si substrate (see Methods and Supplementary Fig. 3 for more details). Figure 2b is the atomic force microscopy (AFM) image of the microdisk resonator. According to the height profile (Fig. 2c), the total thickness of the hBN microdisk is around 650 nm, with a diameter of ~4 μm. In addition, beneath the hBN, there is another SiO₂ pedestal with a slightly larger diameter and thickness of ~300 nm. The pedestal is formed as the etching gases we used for hBN also work for SiO₂, while the etching self-terminates at the surface of silicon. In Supplementary Fig. 4, we have verified the elemental compositions (B, N, Si, O) of the concentric hBN/SiO₂ microdisks using energy-dispersive X-ray spectroscopy (EDX) in scanning electron microscope (SEM).

Fig. 2. Device structure of the CW organic laser.

a Design of the CW organic laser with monolayer Me-PTCDI encapsulated within the hBN microdisk. AFM image (b) and height profile (c) of the microdisk resonator. Top-view (d) and cross-sectional view (e) of the electric-field distribution of the WGM mode (TE₂₆) inside the hBN microdisk with a diameter of ~4 μm and height of 650 nm. The yellow dotted line indicates the position of the Me-PTCDI monolayer with a thickness of 0.3 nm. The white dashed lines indicate the boundary of each material. f PL spectra of the Me-PTCDI monolayer before (blue curve) and after (red curve) embedded inside the hBN microdisk. The inset image shows the annular red-color emission at the microdisk periphery.

Further, Fig. 2d, e presents the top and cross-sectional views of the simulated electric-field distribution for WGM resonances (using TE₂₆ as an example) in the hBN microdisk with a diameter of ~4 μm (see Supplementary Figs. 5–6 and Supplementary Table 1), where the optical mode confined within the hBN microdisk overlaps well with the embedded Me-PTCDI monolayer. Crucially, the SiO₂ pedestal plays an important role by mitigating the substrate-induced absorption losses, leading to an enhanced cavity field intensity by ~10³ times compared to the hBN microdisk directly on the Si substrate (Supplementary Fig. 7). Experimental verification through micro-PL spectroscopy shows discrete WGM resonances emerging post-RIE processing (Fig. 2f), which are absent in the unpatterned regions. Cavity integration induces clearly enhanced emission from vibronic transitions, accompanied by the appearance of discrete WGM resonances. Consequently, in the fluorescence image, we visualized an annular red-color emission at the microdisk periphery, consistent with WGM field localization at microdisk boundary. Transient absorption spectroscopy further reveals a reduced excited-state lifetime at the microdisk periphery (0.6 ± 0.1 ps) compared to the center (1.4 ± 0.2 ps), implying a 2.33× enhancement in the radiative decay rate due to the Purcell effect (Supplementary Fig. 8). The observed radial modulation of both emission profiles and decay dynamics directly demonstrates efficient coupling between the organic monolayer and the WGM cavity.

Power-dependent emission intensity and linewidth under CW pumping

To verify the WGM lasing in the organic monolayer, we have performed power-dependent PL spectroscopy on the microdisk device under 532 nm CW excitation (Supplementary Fig. 9). Figure 3a displays the magnified emission spectra of a 4-μm microdisk near lasing threshold (627–641 nm spectral window). As the excitation power increases, a sharp WGM peak emerges at 634.7 nm, dominating the emission profile, while the background emission from uncoupled Me-PTCDI monolayer has a much lower intensity. The log-log plot of output intensity versus excitation power (L-L curve, Fig. 3b) reveals a S shape with clear threshold of ~1.5 W/cm². Beyond this threshold, the superlinear increase of the emission intensity accompanied with the spectral narrowing serves as the evidence for the transition from spontaneous emission into amplified spontaneous emission and finally lasing regime. Based on a standard rate equation fitting, we extracted a spontaneous emission factor β of 0.1 (Supplementary Note 10). Meanwhile, the normalized high-resolution emission spectra clearly illustrate the suppression of spontaneous emission background with increasing excitation powers (Supplementary Fig. 10), serving as another evidence for the lasing process. Furthermore, the real-space PL imaging exhibits different emission patterns, i.e., uniform emission from the whole microdisk area below the threshold versus annular emission from the boundary of the microdisk in the lasing regime (Supplementary Fig. 11).

Fig. 3. Power-dependent intensity and linewidth of the CW organic laser.

a PL emission spectra of the microdisk device (4 μm diameter) near the lasing threshold (532 nm CW excitation), highlighting the WGM peak at 634.7 nm. The purple line is a Lorentzian fit to the emission peak. b Plot of emission intensity and FWHM as a function of excitation power density based on the WGM peak at 634.7 nm. The emission intensity is integrated over 631.6–637.8 nm. The red line is a fitting with the rate equation. FWHM: full width at half maximum. c PL emission spectra for the microdisk devices with three different diameters of 2.5, 4 and 5 μm recorded at 137 W/cm². The insets are the corresponding optical micrographs of the microdisks. d Fourier transform of the PL emission spectra in (c).

We also varied the diameter of hBN microdisks (see Supplementary Fig. 12 for the power-dependent emission spectra and L-L curve), and the representative emission spectra were summarized in Fig. 3c. As the diameter decreases from 5 μm to 2.5 μm, the free spectral range (FSR) increases accordingly from 13.9 nm to 28.1 nm, which agrees with the theory ($FSR\approx \frac{{\lambda }^2}{2\pi Rn}$, where λ is the resonant wavelength, R is the microdisk radius, and n is the refractive index of hBN). Alternatively, with the Fourier transform of the spectra (Fig. 3d), an average frequency of laser-peak-spacing can be resolved, which is reciprocal to the FSR value. With this method, the diameters of these microdisk devices were determined to be 2.4 μm, 3.8 μm and 4.8 μm, respectively. Furthermore, the extracted full width at half maximum (FWHM) for the WGM peaks is 1.2 ± 0.4 nm, corresponding to an average cavity quality factor Q of 580. We note that the current low Q is mainly limited by the rough edge of our hBN microdisk, while higher Q factors (>2000) are achievable through high-resolution electron beam lithography combined with optimized reactive ion etching processes⁵¹.

Build-up of linear polarization and coherence in the lasing regime

Beyond threshold characteristics and spectral narrowing, to fulfil the criteria for lasing, we also investigated the linear polarization and coherence of light emission in the lasing regime. Figure 4a shows the PL spectra recorded at polarization angles of 0° and 90°, where the 90° polarization features clearly higher intensity than 0° polarization. The polar plot in Fig. 4b was extracted from the WGM peak at 643.6 nm, yielding a degree of polarization (DoP, $DoP=\frac{I_{90^{\circ }}-I_{0^{\circ }}}{I_{90^{\circ }}+I_{0^{\circ }}}$) of 0.48 in the lasing regime (see Supplementary Fig. 13 for more polar plots). In contrast, the polar plot for the background emission in the spontaneous emission regime (Fig. 4c) exhibits a circular shape with a negligible DoP (Supplementary Fig. 14). The clear enhancement in polarization anisotropy provides extra support for the occurrence of lasing.

Fig. 4. Linear polarization and coherence of the emission in the lasing regime.

a PL emission spectra of the CW organic laser with 0° and 90° polarization under 532 nm CW excitation (74 W/cm²). Polar plot for the WGM peak emission at 643.6 nm (b) and the background emission at 648 nm (c). d–f Experimental Fourier-space angle-resolved PL spectra of the microdisk devices with different diameters D recorded at 137 W/cm². g–i Theoretical modelling of the Fourier-space interference with different microdisk diameters. The inset in panel a defines the in-plane polarization angle α and the out-of-plane emission angle θ.

The coherence of the light emission was confirmed via the wavelength-dependent far-field interference in the Fourier-space. Figure 4d–f show the angle-resolved PL spectra recorded in the lasing regime, where pronounced interference fringes appear at the WGM resonant frequencies. These interference fringes are quite similar to those in Young’s interference experiments, where the diameter D functions as the double-slit distance and the light emission from opposite side of the disk interferes with each other. The periodicity of the interference fringes is found to inverse proportional to the diameter of the microdisk D, where the fringe periodicity decreases as D increases. To be more quantitative, we have modelled the Fourier-space fringes based on double-slit interference (see Supplementary Note 16 for more details). Figure 4g–i shows the interference fringes for D = 2.5 μm, 4 μm, and 5 μm, respectively. The modelled Fourier-space fringes fit well with the experimental cases, indicating the physical origin of the observed interference feature. Moreover, power dependence measurements reveal that the visibility of interference fringes significantly relies on the excitation power density (Supplementary Figs. 15–17), which becomes nonvisible below the threshold. Thus, these interference fringes clearly show the coherence of the light emission, which is the critical character of laser.

Operational reliability of the organic lasers under CW excitations

Finally, we systematically evaluated the operational reliability of the organic lasers under CW excitations. Figure 5a–c display representative WGM lasing spectra for 2.5 μm, 4 μm, and 5 μm microdisks recorded in the lasing regime. Statistical analysis across 15 devices (Fig. 5d) reveals threshold spanning 1.4-3.4 W/cm², showing no diameter-dependent trend. This threshold range represents at least one order of magnitude reduction compared to state-of-the-art organic lasers under CW or qCW excitations (Supplementary Table 2), and is close to the best 2D lasers using 2D perovskites or monolayer transition metal dichalcogenides as gain media (Supplementary Table 3).

Fig. 5. Operational reliability of the organic lasers under CW excitations.

Normalized emission spectra for 2.5 μm (a), 4 μm (b), and 5 μm (c) microdisks recorded at 137 W/cm². 15 devices were measured independently, with 5 devices for each microdisk diameters. d Summary of the thresholds for the 15 microdisk devices with different diameters D. e Emission spectra of the same microdisk measured before (yellow curve) and after (blue curve) 12-h CW laser illumination (198 W/cm²). f Normalized emission intensity time trace for the WGM peak at 672 nm. Data points represent averaged intensities of 600 data points recorded within 10 mins (1 s/spectrum), with error bars representing standard deviations of intensity fluctuations. The standard deviation for the WGM peak wavelength is 0.2 nm. g The intensity retention ratio I_End/I_Start as a function of wavelength for different WGM peaks. I_start: intensity recorded at ~0 h. I_End: intensity recorded after continuous operation over ~12 h.

Moreover, the emission stability of the organic laser under CW operations was followed in real time at different timescales (Supplementary Fig. 18). Figure 5e presents the PL spectra of the same microdisk recorded at the initial and final state, which confirms preserved WGM spectral features after 12-h CW operation. Quantitative analysis of the intensity time trace (Fig. 5f) reveals intensity retention ratio (I_End/I_Start) ranging from 41% to 80% across different WGM peaks (Fig. 5g). This corresponds to an operational half-life (τ_1/2) around 8 h at room temperature (Supplementary Fig. 19), i.e., 120–160× improvement over prior report⁵³ for organic lasers under CW excitations (τ_1/2 ≈ 3–4 mins). The enhanced stability originates from both suppressed triplet accumulation and optimized thermal dissipation. With full encapsulation of hBN, the photodegradation rate of the Me-PTCDI monolayer was largely reduced (Supplementary Fig. 20).

Discussion

In summary, we demonstrate a CW solid-state organic laser through integration of a J-aggregated MOMC with a hBN microdisk resonator. The MOMC exhibits near-unity photoluminescence quantum yield and superradiance-enabled ultrafast radiative lifetime, which effectively suppresses triplet-state accumulation. Concurrently, the low-loss hBN resonator with high thermal conductivity effectively prevents photodegradation of the MOMC. This synergistic combination enables robust CW lasing operation, as evidenced by: (i) S-shape power-dependent emission intensity; (ii) linewidth narrowing; (iii) build-up of polarization anisotropy; (iv) coherent emission confirmed by Fourier-space interference. Our work thus resolves the longstanding challenge in organic semiconductor community, establishing a new platform for realization of CW operations of organic lasers. Furthermore, monolithic integration of the WGM laser with organic light-emitting diodes (OLEDs) may enable electrically-pumped organic semiconductor lasers driven by direct current, which will be a step further than current pulsed electrical excitations.

Methods

Microdisk fabrication

The fabrication of Me-PTCDI monolayer encapsulated within hBN microdisk involves the following procedures. First, mechanically exfoliated hBN flakes (300-350 nm thick) were transferred onto a 300 nm SiO₂/Si substrate using a PDMS stamp under the optical microscope. Then, monolayer Me-PTCDI crystals were directly grown on the hBN surface using the PVD method³³. In the PVD process, Me-PTCDI powder (Rhawn, 99%) was positioned at the tube furnace’s central heating region, with hBN substrate placed ∼10 cm downstream. Before growth, we use Ar gas (100 sccm) to flush the chamber for ~10 minutes to establish an inert environment. The system was then evacuated to ∼2.5 × 10⁻³ Pa using a turbo molecular pump. Growth commenced at 230 °C, with monolayer crystals forming in 15 min. After that, another hBN flake (300–350 nm thick) was transferred onto the as-grown Me-PTCDI monolayer as the top capping layer, followed by spin-coating of polymethyl methacrylate (PMMA). The spin-coating process was performed at 2000 rpm for 60 seconds, after which the sample was heated on a hotplate at 180 °C for 90 s. Finally, polystyrene (PS) microspheres (dispersed in anhydrous ethanol) with different diameters were spin-coated (3000 rpm, 30 s) onto the hBN/Me-PTCDI/hBN region as etching masks. RIE was conducted using Ar/SF₆ gas mixtures for 20 min. Post-etching, the PMMA layer and the PS microspheres were removed via acetone immersion with ultrasonication for 5 s, yielding the final microdisk samples ready for optical characterizations.

Steady-state optical characterizations

The steady-state PL measurements were performed using a commercial confocal microscopy system (WITec Alpha300R) with a 100× objective (Zeiss EC Epiplan-Neofluar, NA = 0.9). Samples were excited by a 532 nm CW laser (WITEC-532, linear polarization extinction ratio 200:1, output power 75 mW) with a focused spot diameter of 1 μm. The spectral resolution is 0.32 nm with 150 line/mm diffraction grating and 0.08 nm with 600 line/mm diffraction grating. The polarization-resolved PL spectra were acquired by rotating a linear polarizer in 15° increments. The Fourier-space PL characterization was performed on a custom angle-resolved spectroscopy system incorporating a spectrometer (Andor Shamrock 500i) and a spectroscopic imaging CCD camera (Andor iDus 416). Samples were excited by a 532 nm CW laser (Changchun Laser MW-ZGL-532, linear polarization extinction ratio 100:1, output power 100 mW) with a focused spot diameter of 2 μm.

Kinetics measurement

Ultrafast transient absorption spectroscopy was performed using a commercial system (mFemto-TR100, Time-Tech Spectra LLC). The laser source consisted of a femtosecond laser (PHAROS PH2-20W-SP, Light Conversion Ltd.) operating at 1030 nm with 190 fs pulse duration and 100 kHz repetition rate. The fundamental output was frequency-converted via harmonic generation and optical parametric amplification modules to produce the 532 nm pump beam. The probe beam is a broadband white-light continuum covering 480–950 nm. Both beams were focused to a spot diameter of 2.5 μm at the sample plane. Measurements for monolayer and bulk Me-PTCDI (on quartz substrate) were performed in transmission mode (yielding ΔT/T signal), while the measurement for microdisk device (on SiO₂/Si substrate) was performed in reflection mode (yielding ΔR/R signal). All experiments were conducted at room temperature with a pump fluence of 3 mJ/cm².

Simulation of E-field distribution

The electrical field distributions of the microdisks were simulated using Lumerical FDTD (finite-difference time-domain) Solutions. In the simulations, the microdisks were modelled as hBN cylinders with the height of 650 nm, which were put on the SiO₂/Si substrate. Perfect matching layers were used as the boundary conditions and the dipolar source was used as the light source located at the edge of the middle plane of the hBN cylinders. The refractive index of the materials is detailed in Supplementary Note 5.

Theoretical modelling of the Fourier-space interference

In the Fourier space, the interference fringes originate from the interference of the light emitted at the same out-of-plane angle (the angle to the optical axis). The light emission from our microdisks was considered as the emission from the WGM mode position (as in Fig. 2d) with a diameter about 0.5 μm smaller than the microdisk diameter. The Fourier-space interference fringes were modelled via Young’s double-slit interference with the detailed description in Supplementary Note 16 and refs. ^54–56.

Reporting summary

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

Author contributions

W.D. conceived the idea. L.Z. and B.Z. Fabricated the microdisk device. The lasing characteristics measurement and data analysis were conducted by L.Z. and B.Z., with assistance from B.R., G.Z., W.L., W.X. and X.X. The kinetics measurements were performed by B.R and L.Z. The simulations and theoretical analysis were performed by T.W. and H.H. All authors discussed the results together. W.D. and T.W. wrote the manuscript with comments from W.Z. and Q.X. The project is supervised by W.D., T.W. and Q.X.

Peer review

Peer review information

Nature Communications thanks Isaac Suarez and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

W.D. discloses support for the research of this work from National Natural Science Foundation of China (grant number 62575196). T.W. discloses support for the research of this work from National Natural Science Foundation of China (grant number 62571358). W.Z. discloses support for the research of this work from National Natural Science Foundation of China (grant number 62574044). Q.X. discloses support for the research of this work from National Natural Science Foundation of China (grant number 12434011 and W2541001) and the New Cornerstone Science Foundation via New Cornerstone Investigator Programme (grant number NCI202508).

Data availability

All the data used for plots generated in this manuscript have been deposited in Figshare.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-73984-3.