Discovery of tunable and soluble organic emitters for solid-state lasers with a self-driving laboratory

Abstract

We have recently demonstrated the ability of using self-driving laboratories for AI-driven searches of organic emitters for solid-state lasing devices. Our past workflow featured solubility challenges for such large molecular moieties. In this next-generation study, we return to the drawing board to explore a family of compounds that are much solution processable and composed of a set of electronic cores that provide a broader color response. Out of 252 potential candidates, and with guidance from DFT calculations, we selectively perform a comprehensive study exploring 51 fluorene-based A-B-A-type organic laser oligomers, armed with our self-driving lab. The candidates range from simple hydrocarbon molecules to complex heteroatom-mixed molecules. As a result of this study, we highlight diketopyrrolopyrrole and benzodiazole derivatives for their largely red-shifted emissions. Furthermore, we investigate the effect of color change arising from heteroatom permutation, fluorine addition, thiophene coupling, and a combination of fluorine addition and thiophene coupling. Amplified spontaneous emission (ASE) measurements in the solid state further corroborate the lasing potential of selected candidates, reinforcing their suitability for future device applications. The computational study with density functional theory confirms the experimental results.

Introduction

Organic solid-state lasers (OSL) have emerged as a promising frontier in light-emitting materials, offering ultra-high emission color purity and narrow spectral linewidths¹. Their chemical tunability and large emission gain cross-sections make them particularly attractive for bio-integrated systems, cost-effective sensing, and lab-on-chip technologies². A major goal in this field is the development of tunable and solution-processable organic laser emitters, which remains a challenge despite recent progress. This study introduces fluorene-based organic laser emitters spanning a broad emission color range, from violet to the near-infrared (NIR) region, achieved through a systematic transition from hydrocarbon molecules to heteroatom-blended derivatives. The discovery campaign was powered by a Level 3 self-driving lab (SDL) system, where a single iteration of automated and interconnected experimentation, guided by quantum chemistry calculations, played a central role in the exploration process^3,4. Solution-phase screening provides a practical and scalable foundation for understanding fundamental organic photophysics, offering rapid insights before transitioning to solid-state investigations. Given that processing constraints often hinder solid-state material exploration, this approach serves as a critical step toward identifying candidates with high lasing potential for future device applications.

Despite recent advancements, exploration of OSL materials remains highly localized within the 4,4′,-bis[(N-carbazole)styryl]biphenyl (BSBCz)-like pentameric framework^5,6,7. A limitation of this scaffold lies in its poor solubility, restricting its processability in conventional solvents⁸. This presents a challenge in solid-state physics study, as physical vapor deposition (PVD)—a widely employed technique for fabricating thin films from low-solubility materials—is inherently time-intensive and less adaptable for large-scale device fabrication⁹. In contrast, the discovery of solution-processable laser-active molecules, particularly through the high-throughput capabilities of SDLs, remains uncharted. Efficient solid-state device fabrication methods are crucial for accelerating various optoelectronic applications, where molecular orientation and packing^10,11,12, charge transport^13,14, light amplification in microcavities^15,16,17, and host–guest interactions^18,19,20 all play a pivotal role. One promising strategy to improve solubility while preserving optical properties involves alkyl group functionalization^8,21,22. However, such structural modifications introduce an extra synthesis step. Moreover, while high-performance liquid chromatography (HPLC) is a commonly used purification technique, the introduction of alkyl groups often leads to longer retention times, due to their enhanced interactions with the stationary phase²³. These trade-offs highlight the need for a balanced molecular design approach that simultaneously enhances solubility, maintains laser performance, and ensures efficient synthesis.

Expanding the emission color range beyond the violet–blue–cyan region while exploring new chemical space presents another challenge for exploration. This limitation originates from the energy-gap law, which dictates that organic semiconductors with longer emission wavelengths generally display higher non-radiative decay rates, leading to lower fluorescent quantum yields (QYs)²⁴. While most organic laser-active molecules are confined to shorter emission wavelengths, strategic molecular engineering has demonstrated that incorporating benzothiadiazole moieties can successfully extend emissions into the yellow, red, and NIR regions^22,25. These heteroatom-containing structures act as strong electron acceptors, facilitating a reduction in the energy gap between the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO), which leads to longer emission wavelengths. Beyond electronic effects, structural rigidity also plays a key role in emission efficiency. Fluorenes, known for their planar and rigid conformations, exhibit high fluorescence efficiency due to their extended π-conjugation and minimal structural relaxation during charge transfer^21,26,27. Leveraging these properties, a trimeric fluorene demonstrated a remarkably low amplified spontaneous emission (ASE) threshold of 1.3 μJ/cm² in its neat film, highlighting the strong lasing potential of fluorene-based oligomers²⁶. Despite these isolated advancements, a comprehensive study systematically evaluating fluorene-based oligomers with diverse acceptor fragments and their impact on optical properties remains lacking.

A systematic experimental approach can provide valuable insights into structure-property relationships, which may inform the development of molecular design strategies for organic laser emitters. To enhance this effort, density functional theory (DFT) calculations were incorporated as a guiding tool to refine candidate selection and provide additional insights into electronic structure and optical transitions. This combined computational-experimental approach identified promising red-shifted materials, spanning from green to NIR, from a library of 252 candidates, with potential applications in electronic and photonic devices. The resulting dataset provides a detailed analysis of emission color shifts and reassesses traditional chemical design strategies, leading to a more extensive organic photophysics study. Two chemical families emerged as particularly promising for achieving large emission wavelength shifts, including a newly identified NIR-emitting candidate with a peak emission wavelength exceeding 700 nm, a notable extension beyond our previous SDL studies^6,7. The following sections first outline the SDL workflow and screening procedure, followed by a systematic exploration of the chemical space, where variable moieties were gradually modified to incorporate diverse fragments, ranging from purely carbon-based structures to heteroatom-functionalized derivatives. The culmination of this effort led to the discovery of fluorene-based oligomers with tunable emission colors, supported by computational insights that elucidate the underlying photophysical mechanisms.

Results

SDL workflow and candidate exploration

Figure 1 provides an overview of the experimental workflow employed in this study. Figure 1a illustrates the definition of A-B-A molecular framework, where fragment A is fixed as 9,9-dimethylfluorene, enabling efficient solution-phase screening, while fragment B is systematically varied with electron-accepting moieties to tune the emission color. This modular design facilitates a plug-and-play approach, allowing future substitution of fluorene fragments with longer alkyl chains (e.g., 9,9-dioctylfluorene) for upscaled synthesis and solid-state device physics investigations. Such modifications could be particularly advantageous given the commercial availability and affordability of these derivatives. To accelerate candidate exploration, a one-step synthetic approach based on Suzuki-Miyaura cross-coupling is employed, simplifying the construction of the A-B-A framework and enabling high-throughput screening of structurally diverse derivatives^6,7.

Fig. 1: Closed-loop high-throughput experimental screening of organic laser molecules.

a A framework of A-B-A-type oligomers to be screened (left). Fragment A is denoted with gray color as it is fixed 9,9-dimethylfluorene, and fragment B is denoted with white color as it is variable. Emission color tuning from the visible to the near-infrared (NIR), as predicted by computational pre-screening (right). b Parts of the self-driving experimental system. Quantos, MEDUSA²⁹, Chemspeed, high-performance liquid chromatography-mass spectroscopy (HPLC-MS), and in-house optical characterization setup, respectively (from left to right). c Experimental role of the parts introduced in (b).

A computational pre-screening approach was introduced to prioritize candidates before proceeding to automated experiments. A virtual library of 252 commercially available dihalides was assembled as potential fragment B units, and their corresponding A-B-A molecular structures were constructed. To refine this selection, DFT calculations were performed at the Franck–Condon region, evaluating key electronic and optical properties. Candidate selection was primarily guided by the absorption wavelength (corresponding to the S₁ energy level) and the oscillator strength of the low-lying S₁ excited state, with a focus on identifying molecules capable of long-wavelength emission (Fig. S1). As a lower-bound criterion, molecules with longer excitation wavelengths were preferentially selected, assuming that an absence of anti-Kasha’s rule would result in further red-shifted emission wavelengths²⁸. Assuming a limited distortion in the S₁ minima with respect to the Franck–Condon geometry, we performed a computational pre-screen. A DFT-guided analysis allowed us to select only candidates whose first excited states have sufficiently low excitation energies (i.e. long absorption wavelengths) and oscillator strength greater than 1.0 (Fig. S1b). To ensure robustness of the pre-screening and minimize the risk of false negatives, pre-screening calculations were performed using multiple implicit solvation models, specifically CPCM and SMD (Figs. S16, S17), to examine whether the choice of solvation model could inadvertently lead to the exclusion of viable candidates. In particular, candidates that have 450–600 nm of large theoretical absorption wavelengths to promote S₁ state were selected for further experimental exploration. Notably, many potentially high-performing candidates contained diketopyrrolopyrrole and benzodiazole moieties, reinforcing the rationale for focusing on these families to achieve substantial red shifts. Additionally, a diverse range of molecules incorporating fundamental structural motifs—such as pure carbon scaffolds and heteroatom-containing fragments (N, O, F, S, Se)—were selected to establish the baseline study. A few candidates declined in the pre-screening stage (e.g., O01 and BD06, Fig. S1(b)) were arbitrarily selected and experimentally explored to avoid the biased discovery.

To maintain a systematic labeling scheme, experimentally validated molecules were categorized based on their structural composition. Molecules composed exclusively of carbon scaffolds were designated as C[n], while those incorporating heteroatom substitutions in fragment B were labeled accordingly: N[n] for nitrogen-containing derivatives, S[n] for sulfur-blended analogs, O[n] for oxygen-containing variants, and X[n] for structures featuring a combination of heteroatoms. Here, [n] represents a unique identifier corresponding to a specific molecular variation. For particular families of interest, a distinct naming convention was introduced. Benzodiazole-based derivatives were assigned the notation BD[n], while amide derivatives were designated as AM[n]. Additionally, molecules that exhibited comparable oscillator strengths to the experimentally tested candidates but lacked experimental data were categorized as unsuccessful synthesis or characterization outcomes. Following this computational pre-screening, 51 molecules were shortlisted (Table S2), advancing to the high-throughput experimental phase, as outlined in the subsequent description.

The high-throughput experimental workflow implemented in this study is designed to synthesize and characterize the optical properties of the selected candidates through an SDL. This workflow integrates multiple automated platforms, including (i) a solid material dosing robot (Quantos) for precise reagent handling, (ii) a multi-channel organic synthesis machine (MEDUSA²⁹) for parallelized reactions, (iii) a solid-phase extraction (SPE) and solution transfer platform (ChemSpeed) for efficient purification, (iv) an high-performance liquid chromatography-mass spectroscopy (HPLC-MS) system for target molecule isolation, and (v) an in-house automated optical characterization system capable of measuring absorption and emission spectrum, fluorescence lifetime, and relative QY. Figure 1b provides photographic documentation of these automated synthesis and screening platforms, while Fig. 1c outlines the stepwise workflow, encompassing solid dosing, organic synthesis, target isolation, and optical characterization. A detailed description of each experimental configuration can be found in the Supplementary Information (SI). At the final stage of this workflow, candidates are evaluated based on the emission gain cross-section (σ), which serves as a quantitative measure of emission efficiency. The precise definition and calculation method of σ are provided in the Methods section.

Hydrocarbon backbones and heteroatom (N, S, O) blended scaffolds

As a first step in the high-throughput experimental screening, a diverse set of molecules was evaluated to determine their emission gain cross-section, with all measurements conducted in acetonitrile solvent. Figure 2 presents the molecular structures of the C, N, S, O, and X series, where only substituents R are explicitly shown for clarity, with 9,9-dimethylfluorene (fragment A) omitted. The systematic study begins with hydrocarbon-based molecules, establishing a baseline for emission properties before introducing heteroatom substitutions. A control set of molecules was selected to mitigate selection bias where the prescreening might merely filter a subset of generally good candidates without truly validating its effectiveness. Specifically, C06 with intrinsically low oscillator strength (0.68) was included for experimental exploration. This provides a benchmark for assessing the prescreening approach and offers an opportunity to explore and better understand the photophysical properties of these molecules. As an initial assessment, a subset of hydrocarbon molecules (C01–C06) was examined, revealing considerable variation in their measured emission gain cross-sections, as shown in Table S6. For instance, C02 in addition to C06 exhibited a poor QY (0.05) as well as a low emission gain cross-section (1.50 × 10⁻¹⁷ cm²). Regarding emission color, hydrocarbon molecules predominantly exhibited violet-to-blue emission, with peak wavelengths ranging from approximately 380–430 nm. Therefore, further structural modifications or the exploration of alternative molecular scaffolds will be necessary to achieve more efficient red-shifted fluorescence.

Fig. 2: Molecular structures of C, N, S, O, and X series and a summary of relative quantum yield (QY), fluorescence lifetime, and emission wavelength represented by color.

Source data are provided as a SourceData file.

Next, the photophysical properties of nitrogen (N01–N14) and sulfur (S01–S05) blended molecules were investigated by varying the number of heteroatoms (1–3) in fragment B. Notably, fragment permutations in nitrogen blended molecules provide the backbone of benzodiazole moiety, allowing the rationale baseline study. In this subset, distinct trends were observed in fluorescence lifetime and QY. Nitrogen blended molecules exhibited moderate fluorescence lifetimes (1.76 ± 1.25 ns) and relatively high QYs (0.51 ± 0.17), suggesting enhanced radiative transitions. In contrast, sulfur-blended molecules displayed shorter lifetimes (0.75 ± 0.09 ns) and lower QYs (0.35 ± 0.10), indicating more efficient non-radiative decay pathways. This trade-off between lifetime and emission efficiency underscores the importance of careful molecular engineering in tailoring emission properties. Despite these blends, the maximum emission wavelengths remained within the cyan region (~500 nm), indicating that sulfur and nitrogen incorporation alone is insufficient for inducing substantial red shifts. Further structural refinements, such as extended π-conjugation or additional donor-acceptor tuning, may be necessary to achieve longer-wavelength emissions with optimized fluorescence efficiencies.

In contrast, O01 and O02 were barely emissive. This lack of fluorescence is likely attributed to the presence of two carbonyl groups attached to the scaffold. These molecules exhibit relatively weak absorption in their first excited state, with a computed oscillator strength of approximately 0.2. Calculations at the Franck–Condon region for these molecules revealed intense absorptions in the S₂ and S₃ states (Fig. S13). Combined with the low fluorescence observed for these molecules, this suggests that non-radiative decay from S₂ and S₃ to the ground state may occur in these carbonyl compounds.

Building on the earlier findings, X01–X07 which features mixed heteroatom (N, S, and O) incorporated in fragment B, were investigated to assess their emission properties. Among the emissive derivatives, X07 exhibited efficient blue emission, with a high relative QY of 0.68, and a short fluorescence lifetime of 1.69 ns. Additionally, its high emission gain cross-section of 1.13  ×  10⁻¹⁶ cm² suggests its potential as a strong blue-emitting candidate. Besides X07, the X series molecules exhibited short lifetimes of  ~ 1ns, indicating efficient radiative decay. While these results highlight the strong emissive performance of the X series, they remain confined to the blue-cyan spectral region. To address this challenge, further molecular design strategies are required to extend emission into longer wavelengths.

Complicated heterocyclic scaffolds: amide and benzodiazole

As the previous experimental exploration demonstrates, molecules with carbonyl groups are often overlooked in organic laser molecule design, likely due to their tendency to undergo π−σ * favored excitation and non-radiative relaxation pathways³⁰. However, in the earlier pre-screening stage, cyclic amide derivatives exhibited high oscillator strengths, justifying further experimental validations (Fig. S1). Figure 3a presents 2D molecular structures of AM01-AM03 (substituents R with 9,9-dimethylfluorene omitted), while their absorption and emission spectra are displayed in Fig. 3b. AM01, an amide derivative lacking thiophene, exhibited UV-region emission at 381 nm. In contrast, the incorporation of a fused thiophene ring in AM02 resulted in a 64 nm red-shift in its emission peak wavelength. Another candidate, AM03, is a pentameric amide derivative featuring a 2,5-dihydropyrrolo[3,4-c]pyrrole-1,4-dione (diketopyrrolopyrrole, or DPP) core, coupled with two thiophene units. This molecular design resulted in a substantial red-shift of 258 nm compared to the emission peak of AM01 (Fig. 3b), demonstrating the effectiveness of DPP-based structural engineering in extending emission wavelengths. Photophysical characterization of AM03 revealed a relative QY of 0.33, a lifetime of 2.79 ns (Fig. 3c), and an emission gain cross-section of 4.73  ×  10⁻¹⁷ cm², indicating its strong lasing potential. This substantially red-shifted emission is particularly notable, as a previous study introducing dithiophenyl-DPP (DT-DPP) as a red emitter resulted in dual emission peaks at 560 nm and 620 nm³¹. Furthermore, another study on oligofluorene-DPP systems, including a star-shaped dendrimer, demonstrated that in the absence of thiophene coupling, emission peak wavelengths remained below 600 nm³². This comparison underscores the critical role of thiophene and fluorene units in achieving substantial red shifts while preserving efficient fluorescence properties.

Fig. 3: Steady-state optical properties of AM series.

a Emission color highlighted chemical structures with quantum yield (QY), ϕ), fluorescence lifetime (τ), b Absorption (dashed) and emission (solid) spectra, and c Transient photoluminescence decay. Source data are provided as a SourceData file.

Pre-screening result demonstrated that benzodiazole fragments have strong potential for red-shifted color tuning, therefore, the next exploration focuses on benzodiazole family. A previous study explored fluorene-based oligomeric benzothiadiazole derivatives, demonstrating that the trimeric form exhibited green color emission with an ASE threshold as low as 8.2 μJ/cm², highlighting its strong potential as a lasing emitter²⁵. Nevertheless, the broader chemical space of benzodiazole-based fragments within fluorene-capped A-B-A oligomeric framework has yet to be systematically investigated. To address this gap, the diverse benzodiazole derivatives were examined, inspired by the green-yellow emitting properties of poly(9,9-dioctylfluorene-alt-benzothiadiazole) (F8BT). A systematic synthesis and characterization campaign was performed, incorporating 14 distinct benzodiazole derivatives, including BD02 as a previously reported reference molecule. The structural modifications explored in this study followed four key design strategies: 1. heteroatom permutation, 2. fluorine addition, 3. thiophene coupling, and 4. a combination of fluorine addition and thiophene coupling. The influence of these modifications on emission wavelength is illustrated in Fig. 4a (substituents R with 9,9-dimethylfluorene omitted). The resulting optical properties of these molecules are summarized in Fig. 4b, where distinct structural modifications define three molecular classes: (i) fluorine addition (triangle), (ii) thiophene coupling (square), and (iii) a combined fluorine-thiophene system (diamond). Notably, molecules belonging to the third class (diamond symbols) can be synthesized via a stepwise approach, incorporating either fluorination followed by thiophene coupling, or vice versa, or through a direct one-step synthetic route. These molecular relationships and transformations are schematically represented with arrows indicating modification pathways and symbols distinguishing different molecular classes.

Fig. 4: Optical properties of benzodiazole family.

a Emission-color-highlighted chemical structures sorted by modifications, b Fluorescence lifetime, relative quantum yield (QY), and emission color shift, with structural modifications illustrated by arrows, and c Summary of emission gain cross-sections, where the larger shaded region highlights molecules emitting above 500 nm and the smaller shaded region marks molecules with emission gain cross-sections comparable to BD02²⁵. Source data are provided as a SourceData file.

The impact of heteroatom permutations on emission properties is illustrated in Fig. 4a, where BD02, the reference molecule, is positioned in the green-emitting region with a peak emission wavelength of 549.5 nm. A structurally similar derivative incorporating oxygen atom permutation, BD04, retained the same peak emission wavelength but exhibited a 20% higher emission gain cross-section, with a relative fluorescence QY of 0.67 and a fluorescence lifetime of 5.65 ns. These enhancements suggest that BD04 may offer improved lasing performance compared to BD02. Introducing a nitrogen atom into the framework, as in BD03, induced a 39 nm red-shift, leading to yellow emission at 588.5 nm. A different permutation approach involved replacing sulfur with selenium, as observed in BD05, which resulted in a moderate red-shift of 33.5 nm, yielding a peak emission at 583 nm. Conversely, replacing sulfur with nitrogen, as in BD01, resulted in a notable blue-shift of 98 nm, positioning its emission in the cyan region. This molecule also exhibited a shorter fluorescence lifetime (1.49 ns), a high relative QY (0.74), and a large emission gain cross-section of 1.64  ×  10⁻¹⁶ cm², indicative of strong radiative transitions. The introduction of two additional nitrogen atoms in another fused six-membered ring, as in BD06, induced a substantial red-shift of 83 nm. However, despite its extended emission, its relative QY (0.11) and fluorescence lifetime (11.77 ns) remained relatively limited, as shown in Figure 4b. Overall, heteroatom permutations enabled emission tuning across the blue-to-orange spectral regions, demonstrating their effectiveness in modifying photophysical properties. While these modifications extend the accessible color range, further structural changes are required to push emission into the red-to-NIR region.

Following the heteroatom permutation strategy, fluorine substitution was investigated as a second modification strategy . Owing to its high electronegativity, fluorine exerts a strong inductive effect that lowers the HOMO energy while introducing minimal steric perturbation due to its small atomic radius³³. Two fluorine atoms were introduced into three previously studied permutation fragments, yielding the molecule pairs BD02 → BD07 (28 nm blue-shift), BD04 → BD08 (21 nm blue-shift), and BD05 → BD09 (35 nm blue-shift). In all cases, fluorine substitution resulted in blue-shifted emission.

The third structural modification strategy involves thiophene coupling, in which two thiophene units are introduced at each bonding site of the benzodiazole fragments. This extension of π-conjugation effectively transformed the molecules into pentameric structures, and pronounced red shifts in peak emission wavelengths were observed: BD01 → BD10 (52 nm), BD02 → BD11 (116 nm), as shown in Fig. 4c. Among these, the largest red-shift occurred in BD05 → BD12 (137 nm). The red-shift observed here agrees with previous findings in amide derivatives, where thiophene-coupled systems exhibited bathochromic shifts relative to their non-thiophene counterparts. In terms of photophysical behavior, thiophene-coupled molecules exhibited longer fluorescence lifetimes, while their relative QYs were moderately reduced. This trend aligns with earlier observations in sulfur-blended molecules, as shown in Fig. 2. Despite this trade-off between lifetime and QY, their emission gain cross-sections remained high, comparable to that of the reference molecule BD02. These findings highlight thiophene coupling as an effective design strategy for achieving red-to-NIR emission, while preserving key fluorescence properties essential for OSL applications.

The final structural modification approach combined fluorine addition and thiophene coupling to investigate their combined effects on emission properties. This operation was applied to two molecular pairs: BD01 → BD13 (19.3 nm blue-shift) and BD02 → BD14 (73.6 nm red-shift). The varying alkyl chain lengths attached to the nitrogen atoms of BD01 and BD13 were assumed to have a negligible effect on peak emission wavelength, relative fluorescence QY, and lifetime³⁴. The combined modification yielded mixed spectral outcomes. In BD13, where benzotriazole was incorporated as fragment B, the expected red-shift from thiophene coupling was not observed. Instead, the fluorine-induced blue-shift dominated, resulting in an unexpected peak emission shift. In contrast, BD14 exhibited a moderate red-shift, though it was less pronounced than that observed in standalone thiophene-coupled derivatives. Despite the limited color shift, BD13 displayed a poor emission gain cross-section, which can be attributed to a decline in relative QY, with minimal compensation from its lifetime.

To analyze the impact of structural modifications, HOMO–LUMO energy gaps were calculated at the optimized ground-state geometries for the BD series (Fig. S21). Fluorine-added molecules (BD07, BD08, and BD09) exhibited lower HOMO levels, resulting in larger HOMO–LUMO gaps compared to their non-fluorinated counterparts (BD02, BD04, and BD05). In contrast, thiophene-coupled derivatives (BD10, BD11, and BD12) showed reduced HOMO–LUMO gaps relative to non-thiophene analogs (BD01, BD02, and BD05), consistent with the red-shifted emission observed experimentally. However, this correlation does not hold for BD13, where the calculated ground-state gap fails to reflect the blue-shifted emission. This discrepancy is likely attributable to geometric relaxation at the S₁ minimum, suggesting that the ground-state electronic structure alone may not fully capture the photophysical behavior in such cases. Overall, these findings underscore the molecular dependency of combined fluorine and thiophene modifications, emphasizing the need for careful structural design to balance color tuning, fluorescence efficiency, and emission gain properties in OSL materials.

Optical properties in thin-film device

To assess the solid-state optical properties relevant to organic laser applications, AM03 and BD12 were selected as representative emitters. These compounds were chosen due to their strongly red-shifted emission observed in the prior solution-phase screening. Thin films were prepared by doping each emitter into 4,4’-bis(carbazol-9-yl)biphenyl (CBP) host. CBP facilitates efficient energy transfer and suppresses aggregation-induced quenching, making it a suitable matrix for emissive film studies^6,7,19,35,36. In this study, CBP exhibited good spectral overlap with the absorption profiles of AM03 and BD12, suggesting efficient energy transfer from host to guest (Fig. S27a). Both AM03 and BD12 contain dioctyl-substituted fluorene as Fragment A for better solubility and therefore spin-coating compatibility^8,21,22. This structure enabled the reliable fabrication of emissive thin films suitable for comprehensive optical characterization.

Figure 5a, d presents the normalized absorption and photoluminescence (PL) spectra of AM03:CBP and BD12:CBP films across a range of doping concentrations. Both systems display red-shifted emission in the solid state, with AM03 retaining a narrow, vibronically structured PL profile even at higher dopant loading. In contrast, BD12 exhibits additional red-shifting with increasing dopant concentration, likely due to changes in the local dielectric environment or the formation of dipole-stabilized dimers, consistent with prior observations in polar organic matrices¹⁹. PLQY trends, shown in Fig. 5e, indicate that BD12:CBP films undergo a moderate efficiency drop with increased concentration, suggesting partial quenching. Interestingly, the BD12:CBP (1 wt%) film exhibited a high PLQY of 50%, despite incorporating a heavy atom (Se), which is typically associated with enhanced intersystem crossing (ISC) and increased non-radiative triplet formation³⁷. To investigate this apparent anomaly, spin-orbit coupling (SOC) values were calculated as an indicator of ISC rate (Table S8). Selenium-containing molecules without thiophene units, such as BD05 and BD09, exhibited SOC values on the order of tens of cm⁻¹. In contrast, BD12—a benzoselenadiazole derivative with thiophene conjugation—showed reduced SOC values in the single-digit range. This suggests that the incorporation of thiophene rings may effectively suppress ISC, thereby preserving singlet-state radiative decay and contributing to the high PLQY observed in BD12. In comparison, AM03:CBP films display a steeper decline in PLQY (Fig. 5b), pointing toward more non-radiative losses, possibly due to the self-absorption effect³⁸. Transient PL measurements (Fig. 5c and f) further support these distinctions. BD12-based films exhibit relatively long-lived emission lifetimes ranging from 4.76 ns to 5.94 ns, whereas AM03-based films show shorter decay times between 2.46 ns and 3.11 ns. These differences suggest that, despite stronger quenching at higher concentrations, AM03 exhibits more rapid exciton dynamics, which may be favorable for achieving efficient optical gain in laser applications.

Fig. 5: Steady-state and time-resolved optical properties of AM03 and BD12 in 4,4′-bis(carbazol-9-yl)biphenyl (CBP) thin films at various doping concentrations.

a, d Normalized absorption (solid lines with symbols) and photoluminescence (solid lines) spectra of AM03:CBP and BD12:CBP films, respectively. b, e Photoluminescence quantum yields (PLQYs) plotted as a function of dopant concentration for AM03 and BD12, respectively. c, f Time-resolved photoluminescence (PL) decay profiles showing concentration-dependent fluorescence lifetimes. Source data are provided as a SourceData file.

To evaluate the light-amplification potential of the selected emitters in the solid state, ASE measurements were carried out under pulsed optical excitation. Figure 6 presents the ASE characteristics of AM03:CBP and BD12:CBP films across varying doping concentrations. The most efficient ASE performance was observed at 1 wt% for AM03 and 5 wt% for BD12, which were selected as the optimal doping conditions for comparison (See Supporting Information for a summary of results). The AM03:CBP blend exhibited a low ASE threshold of 3.03 μJ/cm², accompanied by pronounced spectral narrowing centered at 720 nm, with the full-width at half-maximum (FWHM) reduced to 8.9 nm. These results indicate highly efficient emission light amplification and, therefore, a low ASE threshold. In comparison to previously reported DT-DPP systems, the AM03-based film demonstrates two key advantages: (1) a substantial red-shift of 100 nm in ASE peak emission, and (2) an even lower ASE threshold, achieved without relying on a TADF assistant dopant to facilitate triplet-to-singlet upconversion³¹. These features highlight the effectiveness of the molecular design of AM03 in enabling NIR lasing under solution-processed conditions. In contrast, BD12:CBP films exhibited a relatively high ASE threshold of 21.1 μJ/cm² and limited spectral narrowing, with FWHM confined to 25.8 nm at 720 nm. Although BD12 shows high PLQY in its doped film, its weaker ASE performance suggests the presence of non-radiative loss pathways, potentially associated with aggregation or triplet-state quenching. Altogether, the results demonstrate that AM03 achieves efficient NIR light amplification with a low ASE threshold, establishing it as a strong candidate for solid-state organic laser applications.

Fig. 6: Amplified spontaneous emission (ASE) characteristics of AM03 and BD12 doped in 4,4’-bis(carbazol-9-yl)biphenyl (CBP) thin films.

a, c Photoluminescence (PL) intensity and full-width at half-maximum (FWHM) as a function of excitation intensity for AM03:CBP (1 wt%) and BD12:CBP (5 wt%), respectively. The dashed lines indicate linear fits applied separately to the sub-threshold and above-threshold regimes, with the former obtained from linear scaling of emission intensity and the latter from a linear fit to the log–log representation of the data. The ASE threshold is determined as the intersection point of these two fitted lines. b, d Evolution of ASE spectra with increasing pump intensity, showing spectral narrowing and threshold behavior for each system. Source data are provided as a SourceData file.

Discussion

A computationally-guided virtual screening approach, applied to a library of 252 candidates, led to the discovery of 14 molecules exhibiting emission wavelengths beyond 500 nm (Fig. 4c). Tracing back to the pre-screening calculations, the estimation of absorption wavelength for S₁ state was aligned well with experimental measurements, supporting their utility in guiding candidate selection (Figs. S18, S19). Benchmarking against the brightest blue-emitting molecule reported in a previous study⁷ successfully supports that the newly identified molecular set exhibits comparable brightness, further validating the efficacy of the discovery campaign (Fig. S28). Figure 4c presents a hypothetical Pareto front, estimating the upper limit of emission gain cross-section across different color regions. From the virtual screening and experimental exploration results, a total of 51 organic laser emitter candidates were successfully identified, demonstrating a robust and adaptable molecular design strategy that extends beyond BSBCz-like pentamers to structurally diverse families. As a post-screening analysis, an inverse relationship was observed between the computed HOMO–LUMO energy gap and the experimentally measured emission wavelengths, particularly in A-B-A systems incorporating strongly electron-deficient fragment B units (Fig. S22). Additionally, natural transition orbitals (NTOs) associated with the S₁ state transition of the experimentally explored molecules were calculated, providing insight into the degree of charge-transfer characteristics and the spatial localization of the involved orbitals in representative cases. The corresponding electronic densities involved in absorption are visualized in Figs. S10–S15^39,40. (See “Methods” section for details). This comprehensive analysis of electronic density distributions provides an overview of orbital transitions depending on permutation fragments for future molecular design campaigns.

To achieve large red-shifted emissions, this study systematically explored molecular structures ranging from simple hydrocarbon scaffolds to complex heteroatom-enriched structures. Among them, two key structural classes emerged as particularly promising: DPP-based and benzodiazole derivatives. DPP has been widely utilized in organic field-effect transistors (OFETs) and organic photovoltaic devices, owing to its high charge carrier mobility and strong intermolecular π-π interactions^41,42. Notably, DPP-based polymers have achieved exceptionally high charge carrier mobilities, exceeding 1 cm² V⁻¹ s⁻¹ in OFET applications⁴³. While BSBCz derivatives remain widely studied in organic laser research, their relatively low charge carrier mobility (on the order of 10⁻³ cm² V⁻¹s^{−1 44}) may pose limitations for light-emitting performance in electrically driven devices. In this context, DPP-based systems represent a compelling alternative. Here we  demonstrate that DPP-based molecules can be engineered to exhibit deeper red-shifted emissions through incorporation of thiophene rings and fluorene units. An additional red-side emission peak at 696 nm as shown in Fig. 3b, along with spectral overlap between the absorption and blue-side emission peak, suggests the possibility of an ASE oscillation occurring near the red-side emission peak. This behaviour is attributed to self-absorption arising from the spectral overlap between absorption and the blue-side peak, which supresses the  emission light amplification. Steady-state photoluminescence and ASE measurements in the solid state confirmed this hypothesis, showing an additional bathochromic shift in the absence of solvent effects and consequently achieving a red-shifted ASE at 720 nm. This observation aligns with a previous study introducing dithiophenyl-DPP (DT-DPP) as a red emitter³¹, where a similar dual-emission behavior was reported. Overall, these findings highlight DPP-based emitters as a compelling class of organic laser materials, bridging the gap between optoelectronic and laser device applications.

Fluorene-based organic laser candidates exhibit remarkable color tunability, driven by structural versatility and π-conjugation engineering. Beyond small-molecule designs, an octafluorene derivative exhibited a low ASE threshold of 90 nJ/cm² in its neat film, demonstrating minimal concentration quenching²¹. This suggests that expanding these molecular systems into the larger systems of oligomeric or polymeric architectures could further enhance their solid-state performance. Starting from these findings, further investigation in solid-state is expected towards electrically driven lasing and other potential applications in high-performance fluorescence-based technologies, including bioimaging, display technologies, and optoelectronic devices. The suggested plug-and-play modular transition to the molecules with longer alkyl chains will provide a handy alternative to the solubility limitations, facilitating their integration into solution-processed organic light-emitting devices.

Methods

DFT configuration

A detailed computational protocol was employed for each of the molecules. First, the lowest energy conformers were found using CREST (version 2.12)^45,46 at the GFN2-xTB^47,48 level of theory. The resulting selected structures were further optimized with the B97-3c composite method⁴⁹, which can produce reliable ground state geometries for screening organic light-emitting diode (OLED) molecules with a comparably low computational cost^50,51,52. All positive harmonic frequencies with the same level of theory confirmed minimum-energy structures. Single-point vertical excitation energy calculations were performed on the optimized ground-state geometries using the ωB97X-D3⁵³/def2-TZVP⁵⁴ level of theory, incorporating an implicit solvation model (CPCM and SMD in acetonitrile). The computer oracle DELFI⁵⁵ suggested the functional and confirmed by a small benchmark on a subset of molecules. To evaluate the reliability of the workflow, its computed absorption wavelengths were compared with an experimental dataset published by Wu et al.⁶. The absorption wavelength corresponding to the lowest-energy peak in the experimental spectra was used for comparison with the computed values (Fig. S17). Finally, NTO calculation and analysis were performed using TheoDORE software (version 3.1.1)⁵⁶. The analysis of NTO was to see how much the orbital is localized in the permutation fragment (fragment B) and determine how electron-accepting the fragment is. Therefore, the orbital density of each fragment had a value within the 0-1 range. When analyzing the orbital density of each fragment, the density in hydrogen atoms was not considered, as it occupied  ~ 0.02 of negligible densities. All DFT calculations were performed with ORCA (version 5.0.3)⁵⁷.

High-throughput organic synthesis for small-scale screening

For the high-throughput screening, a one-step Suzuki-Miyaura cross-coupling reaction was conducted in which the only variable substrate was the dibromo fragment B. The equivalence ratio (eq) and the amount of fragment B were fixed at 1.0 eq and 0.05mmol, respectively (explicit masses provided in Table S2). In an 8 mL reaction vial, 33.6 mg (2.1 eq) of 2-(9,9-dimethyl-9H-fluoren-2-yl)-4,4,5,5-tetramethyl-1,3,2-dioxaborolane (fragment A), 2 mg (0.05 eq) of Pd XPhos G2, and 63.6 mg (12 eq) of Na₂CO₃ were added together with a magnetic stir bar. The vial was sealed and purged with nitrogen to remove residual oxygen. Separately, a 5:1 (v/v) mixture of 1,4-dioxane and deionized water was prepared and degassed by bubbling nitrogen for 15 min. The purged solvent mixture (2 mL) was then transferred into the reaction vial, and the reaction was performed at 90 ^∘C for 6 h under constant stirring at 400 rpm.

Emission gain cross-section and emission color

To quantify the laser performance of organic molecules, the emission gain cross-section (σ_em⁵⁸) is estimated as

$${\sigma }_{{{\rm{em}}}}({\lambda }_{\max })=\frac{{f}_{{{\rm{em}}}}({\lambda }_{\max })\,\cdot \phi }{{n}^{2}\cdot \tau }$$

(1)

where ${\lambda }_{\max }$ is the wavelength at which the emission spectrum reaches its maximum intensity, n is the refractive index of the organic solvent (e.g., n ≈ 1.5 for acetonitrile), ϕ is the fluorescent QY, and τ is the fluorescence lifetime. The parameter f_em is the spectral gain factor, which is calculated from the emission spectrum as follows:

$${f}_{{{\rm{em}}}}(\lambda )=\frac{{\lambda }^{4}\cdot {I}_{{{\rm{em}}}}(\lambda )}{8\pi c\cdot \int \,{I}_{{{\rm{em}}}}(\lambda )d\lambda }$$

(2)

where λ is the emission wavelength, I_em is the emission intensity at that wavelength, and c is the speed of light. According to Eq. (1) and Eq. (2), a higher QY and a shorter fluorescence lifetime lead to an increased emission gain cross-section, thereby improving the overall lasing performance of organic laser materials.

Scaled-up organic synthesis of soluble AM03

3,6-bis(5-bromothiophen-2-yl)-2,5-bis(2-ethylhexyl)-2,5-dihydropyrrolo[3,4-c]pyrrole-1,4-dione (339 mg, 0.5 mmol), 9,9-dioctyl-9H-fluorene-2-Bpin (570 mg, 1.1 mmol), K₃PO₃ (1.12 g, 5 mmol) and Pd XPhos G2 (8 mg, 0.01 mmol) were added into a 50 mL Schlenk tube with a magnetic stirbar. The Schlenk tube was capped, then evacuate-refilled three cycles to change to Ar atmosphere on a Schlenk line. Under a positive flow of Ar, the Schlenk tube cap was replaced with a septum, and 25 mL degassed 19:1 THF:H₂O was added using a syringe. The septum was removed and the Schlenk tube was recapped, then the reaction was heated and stirred at 50 ^∘C for 16 h. The reaction mixture was cooled to room temperature and dried over Na₂SO₄, then filtered through a Celite^® pad. The filter cake was washed with 2 × 20mL THF, and the combined liquid was concentrated on a rotavap. The mixture was then purified on a Buchi auto flash column with 120 g EcoMax column. The first peak was unconsumed 2-(9,9-dioctyl-9H-fluoren-2-yl)-4,4,5,5-tetramethyl-1,3,2-dioxaborolane, and the fourth peak was the desired product after removing the solvent on a rotavap to a dark blue solid (Fig. S6). The final yield of AM03 was 386 mg (59.6%).

Scaled-up organic synthesis of soluble BD12

4,7-bis(5-bromothiophen-2-yl)benzo[c][1,2,5]selenadiazole (256 mg, 0.5 mmol), 9,9-dioctyl-9H-fluorene-2-Bpin (570 mg, 1.1 mmol), K₃PO₃ (1.12 g, 5 mmol) and Pd XPhos G2 (8 mg, 0.01 mmol) were added into a 50 mL Schlenk tube with a magnetic stirbar. The Schlenk tube was capped, then evacuate-refilled three cycles to change to Ar atmosphere on a Schlenk line. Under a positive flow of Ar, the Schlenk tube cap was replaced with a septum, and 25 mL degassed 19:1 THF:H₂O were added using a syringe. The septum was removed and the Schlenk tube was recapped, then the reaction was heated and stirred at 50 ^∘C for 16 h. The reaction mixture was cooled to room temperature and dried over Na₂SO₄, then filtered through a Celite^® pad. The filter cake was washed with 2 × 20 mL THF, and the combined liquid was concentrated on a rotavap. The mixture was then purified on a Buchi auto flash column with 120 g EcoMax column. The first peak was unconsumed 2-(9,9-dioctyl-9H-fluoren-2-yl)-4,4,5,5-tetramethyl-1,3,2-dioxaborolane, and the second peak was the desired product after removing the solvent on a rotavap to a dark purple solid (Fig. S9). The final yield of BD12 was 451 mg (80.2%).

General NMR characterization procedure

¹H and ¹³C NMR spectra were recorded on a 500 MHz Agilent DD2 spectrometer in CDCl₃. Proton assignments were determined based on integration, multiplicity, and coupling patterns. Carbon assignments were supported by characteristic chemical shift ranges. Only unambiguously determined assignments are reported in the Supplementary Information.

Thin-film optical device fabrication and characterization

Thin films were fabricated on clean quartz substrates for optical measurements and on nonfluorescent glass (MATSUNAMI micro-slide glass S0313, 1 mm thickness) for ASE measurements. The substrates were rinsed with acetone and isopropyl alcohol and then treated with UV/ozone (Nippon Laser & Electronics Lab, NL-UV253) to clean adsorbed organics before film deposition. Each thin film was then prepared by spin-coating from a chloroform solution (24mg/mL including host and guest) at a rotational speed of 750 rpm for 60 s at room temperature. The measured thicknesses using a surface profiler (Bruker Dektak XT) were 170 ± 5nm for each doped film. Each thin film deposited on quartz substrates was measured using UV-Vis (Shimadzu Scientific UV-3600i Plus), PL (JASCO FP-8600 fluorometer), PL quantum yield (Hamamatsu Photonics Quantaurus-QY, C11347-01), and transient PL decay (Hamamatsu Photonics Quantaurus-Tau, C11367-03).

ASE measurements were conducted by monitoring the emission intensity and spectral line width of thin films as a function of excitation intensity with a nitrogen laser excitation source (USHO KEN-C6020; pulse width: 900 ps; repetition rate: 10 Hz). A cylindrical lens was used to shape the incident laser beam into a stripe, producing an excitation area of ~0.43 cm × 0.09 cm. All measurements were performed at room temperature and under a nitrogen atmosphere to minimize sample degradation. Emission from the film edge was collected using an optical fiber and recorded with a spectrometer (Hamamatsu Photonics PMA C10026 combined with Princeton Instruments SpectraPro HRS-300; groove density: 150 g/mm; blaze wavelength: 500 nm; spectral resolution: 1.7 nm). From these findings, ASE thresholds are determined by separately fitting the characteristic excitation intensity–emission intensity curves in the sub-threshold and above-threshold regime. For the sub-threshold regime, we assume a linear increase of the emission intensity I_em as a function of excitation intensity I_exc, following the equation below:

$${I}_{{{\rm{em}}}}={k}_{0}\cdot {I}_{{{\rm{exc}}}}$$

(3)

where k₀ is determined from a linear fit in the sub-threshold regime of the I_em vs. I_exc plot. For the above-threshold regime, a simplified power-law dependency of the form

$${I}_{{{\rm{em}}}}={k}_{1}\cdot {({I}_{{{\rm{exc}}}})}^{a}$$

(4)

is expected. The parameters k₁ and a were extracted by applying a linear fit to the above-threshold region in the log–log plot of emission intensity (I_em) versus excitation intensity (I_exc). The sub-threshold and above-threshold regimes were defined by identifying the breakpoint that minimizes the combined sum of squared residuals from both linear fits, thereby maximizing the overall fitting quality. The ASE threshold was then determined as the intersection point of the two fitted lines.

For comparison, ASE thresholds were also estimated using the FWHM halving method, in which the threshold is defined as the excitation intensity at which the FWHM of the emission spectrum decreases to half of its initial value at low excitation intensity.

Reporting summary

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