A quantitative, predictive account of thermally activated delayed fluorescence in carbazolyl-dicyanobenzene emitters: singlet–triplet spin mixing, low-temperature decay heterogeneity, kinetic-model validity, and structure–efficiency relationships

The 2012 discovery that metal-free carbazolyl-dicyanobenzene (CDCB) donor–acceptor emitters such as 4CzIPN convert dark triplet excitons into light with internal efficiencies near 100% founded the modern field of thermally activated delayed fluorescence (TADF), but the original work left the underlying photophysics largely phenomenological: the spin–orbit mixing rationale was qualitative and uncomputed, low-temperature decay curves could not be fit by the standard two-level kinetic model, the extracted rate constants rested on untested simplifying assumptions, and the origins of low photoluminescence yield in substituted derivatives were unexplained. This problem asks for a quantitative, predictive account linking molecular structure and film environment to TADF observables — spin-mixing (ISC/RISC) rates, decay-curve shapes, and photoluminescence quantum yields — for the CDCB emitter family. Later literature substantially transformed the mechanistic picture (second-order spin–vibronic coupling via locally excited triplet states, multi-triplet kinetics, heterogeneous conformational ensembles) yet, as leading groups still state, quantitative prediction from first principles remains incomplete, so the problem is open.

Background

In an organic light-emitting diode (OLED), recombining charge carriers form spin-singlet and spin-triplet excitons in a 1:3 ratio, so three quarters of the excitons are born in the triplet manifold, which in purely organic molecules normally decays non-radiatively. Metal-free emitters showing thermally activated delayed fluorescence (TADF) recover these losses: the molecule is designed so that the energy gap $\Delta E_{ST}$ between the lowest singlet excited state $S_1$ and the lowest triplet state $T_1$ is very small (roughly tens of meV), so that the thermally activated reverse intersystem crossing (RISC) process, the spin flip $T_1 \rightarrow S_1$, competes with non-radiative triplet decay, after which the molecule fluoresces. Intersystem crossing (ISC) is the forward $S_1 \rightarrow T_1$ spin flip; the photoluminescence quantum yield (PLQY, denoted $\Phi$) is the fraction of absorbed photons re-emitted, split experimentally into a nanosecond 'prompt' component and a microsecond-to-millisecond 'delayed' component.

Uoyama, Goushi, Shizu, Nomura and Adachi (Nature 492, 234–238 (2012), DOI 10.1038/nature11687) introduced the CDCB (carbazolyl-substituted dicyanobenzene) family, most famously 4CzIPN (1,2,3,5-tetrakis(carbazol-9-yl)-4,6-dicyanobenzene): four electron-donating carbazolyl groups twisted about 60° out of the plane of an electron-accepting dicyanobenzene core, so that the highest occupied and lowest unoccupied molecular orbitals (HOMO and LUMO) are spatially separated and $\Delta E_{ST}$ is small. Doped in a wide-triplet host such as CBP (4,4′-bis(carbazol-9-yl)-1,1′-biphenyl), 4CzIPN showed PLQY of 83% and OLEDs reached an external quantum efficiency (EQE, the fraction of generated photons emitted from the device) of 19.3%, i.e. roughly 100% internal quantum efficiency under an assumed 20% light out-coupling factor (the fraction of internally generated photons that escape the device). The paper's quantitative footing, however, was limited: geometries came from density functional theory and its time-dependent extension for excited states (DFT/TD-DFT, at the PBE0/6-31G(d) functional and basis-set level), and the central spin-physics argument was the first-order perturbation statement that the singlet–triplet mixing coefficient $\lambda = H_{SO}/\Delta E_{ST}$ (where $H_{SO}$ is the spin–orbit interaction matrix element) grows as $\Delta E_{ST}$ shrinks — no spin–orbit matrix elements were computed, and no design rule beyond 'make $\Delta E_{ST}$ small' was offered.

Four limitations of the 2012 analysis motivate this problem. First, the spin-mixing rationale was qualitative: a predictive, quantitative strategy for maximizing spin conversion while preserving a large radiative decay rate (which requires HOMO–LUMO overlap, the very thing small $\Delta E_{ST}$ destroys) was not established. Later work sharpened this tension decisively: for charge-transfer (CT) states, first-order spin–orbit coupling between ${}^1$CT and ${}^3$CT of the same orbital character essentially vanishes (an instance of El-Sayed's rule, the photochemical selection rule that intersystem crossing is efficient only between states of different orbital character), so RISC in donor–acceptor TADF proceeds mainly through second-order, vibronically coupled pathways mediated by locally excited triplet (${}^3$LE) states (Etherington et al., Nat. Commun. 7, 13680 (2016), DOI 10.1038/ncomms13680; Gibson, Monkman and Penfold, ChemPhysChem 17, 2956–2961 (2016), DOI 10.1002/cphc.201600662), a mechanism the 2012 first-order $\lambda$ picture does not capture. Second, below about 200 K the photoluminescence decay of 4CzIPN:CBP films departs from the double-exponential (prompt + delayed) model and becomes multi-exponential, which the authors attributed to a widened distribution of $\Delta E_{ST}$ caused by inhomogeneous molecular environments in the amorphous film — an explanation offered in one sentence, with no microscopic model. Third, the kinetic analysis carried untested simplifying assumptions: the Arrhenius extraction of the RISC rate constant $k_{RISC} = k_p k_d \Phi_d/(k_{ISC}\Phi_p)$ from data between 200 and 300 K (yielding an 83 meV activation energy) assumed that the ISC rate constant $k_{ISC}$ is temperature-independent, and the internal efficiency was inferred from EQE through an assumed out-coupling factor. Fourth, within the same CDCB series the PLQY collapses upon substitution (26% for the orange emitter 4CzTPN-Ph versus 94% for 4CzIPN in solution), and the interplay of substituent effects, molecular conformation, $\Delta E_{ST}$ and non-radiative channels was not systematically analyzed.

Problem Statement

Establish a quantitative, predictive account of thermally activated delayed fluorescence in the carbazolyl-dicyanobenzene (CDCB) emitter family, as left qualitative by Uoyama et al. (2012), by resolving four linked open issues:

1. Spin–triplet mixing and design rules. Replace the qualitative first-order argument that the singlet–triplet mixing coefficient $\lambda = H_{SO}/\Delta E_{ST}$ grows as $\Delta E_{ST}$ shrinks with a quantitative, predictive theory of the spin conversion rates $k_{ISC}$ and $k_{RISC}$ in these metal-free donor–acceptor molecules — including the now-known second-order spin–vibronic pathway through locally excited triplet states and the possibility of direct CT-to-CT coupling enhanced by conformational flexibility — and derive from it design rules that maximize spin conversion while preserving a high radiative decay rate, validated against measured rate constants for CDCB-type emitters.

2. Low-temperature decay heterogeneity. Determine the microscopic origin of the multi-exponential photoluminescence decay observed below about 200 K in 4CzIPN films (attributed in the source only to a widened distribution of $\Delta E_{ST}$ from inhomogeneous molecular environments, and in later work to intermediate triplet states and conformational ensembles), and provide a quantitative kinetic model that describes the full temperature-dependent decay.

3. Validity of the kinetic analysis. Test and remove the simplifying assumptions used to extract TADF rate constants: the temperature independence of $k_{ISC}$ assumed in the Arrhenius analysis of $k_{RISC}$, and the reliance on assumed light out-coupling factors when converting external to internal quantum efficiency; establish which extracted quantities (e.g. the 83 meV activation energy of 4CzIPN) are robust.

4. Structure–efficiency relationships. Explain quantitatively why similarly constructed CDCB derivatives differ so strongly in photoluminescence quantum yield (e.g. 26% for 4CzTPN-Ph versus 94% for 4CzIPN in solution), i.e. how substituent effects, molecular conformation, $\Delta E_{ST}$, and non-radiative decay channels interact.

The four issues are independent and each admits a resolution on its own.

References

1. Uoyama, H.; Goushi, K.; Shizu, K.; Nomura, H.; Adachi, C. Highly efficient organic light-emitting diodes from delayed fluorescence. Nature 492, 234–238 (2012). DOI: 10.1038/nature11687, https://doi.org/10.1038/nature11687
2. Etherington, M. K.; Gibson, J.; Higginbotham, H. F.; Penfold, T. J.; Monkman, A. P. Revealing the spin–vibronic coupling mechanism of thermally activated delayed fluorescence. Nature Communications 7, 13680 (2016). DOI: 10.1038/ncomms13680, https://doi.org/10.1038/ncomms13680
3. Gibson, J.; Monkman, A. P.; Penfold, T. J. The Importance of Vibronic Coupling for Efficient Reverse Intersystem Crossing in Thermally Activated Delayed Fluorescence Molecules. ChemPhysChem 17, 2956–2961 (2016). DOI: 10.1002/cphc.201600662, https://doi.org/10.1002/cphc.201600662
4. Kim, I.; Jeon, S. O.; Jeong, D.; Choi, H.; Son, W.-J.; Kim, D.; Rhee, Y. M.; Lee, H. S. Spin–Vibronic Model for Quantitative Prediction of Reverse Intersystem Crossing Rate in Thermally Activated Delayed Fluorescence Systems. Journal of Chemical Theory and Computation 16, 621–632 (2020). DOI: 10.1021/acs.jctc.9b01014, https://doi.org/10.1021/acs.jctc.9b01014
5. Wada, Y.; Nakagawa, H.; Matsumoto, S.; Wakisaka, Y.; Kaji, H. Organic light emitters exhibiting very fast reverse intersystem crossing. Nature Photonics 14, 643–649 (2020). DOI: 10.1038/s41566-020-0667-0, https://doi.org/10.1038/s41566-020-0667-0
6. Wada, Y.; Wakisaka, Y.; Kaji, H. Efficient Direct Reverse Intersystem Crossing between Charge Transfer-Type Singlet and Triplet States in a Purely Organic Molecule. ChemPhysChem 22, 625–632 (2021). DOI: 10.1002/cphc.202001013, https://doi.org/10.1002/cphc.202001013
7. Kobayashi, T.; Niwa, A.; Takaki, K.; Haseyama, S.; Nagase, T.; Goushi, K.; Adachi, C.; Naito, H. Contributions of a Higher Triplet Excited State to the Emission Properties of a Thermally Activated Delayed-Fluorescence Emitter. Physical Review Applied 7, 034002 (2017). DOI: 10.1103/PhysRevApplied.7.034002, https://doi.org/10.1103/PhysRevApplied.7.034002
8. Kobayashi, T.; Kawate, D.; Niwa, A.; Nagase, T.; Goushi, K.; Adachi, C.; Naito, H. Intersystem Crossing Rate in Thermally Activated Delayed Fluorescence Emitters. physica status solidi (a) 217, 1900616 (2020). DOI: 10.1002/pssa.201900616, https://doi.org/10.1002/pssa.201900616
9. Haase, N.; Danos, A.; Pflumm, C.; Morherr, A.; Stachelek, P.; Mekic, A.; Brütting, W.; Monkman, A. P. Kinetic Modeling of Transient Photoluminescence from Thermally Activated Delayed Fluorescence. Journal of Physical Chemistry C 122, 29173–29179 (2018). DOI: 10.1021/acs.jpcc.8b11020, https://doi.org/10.1021/acs.jpcc.8b11020
10. Vázquez, R. J.; Yun, J. H.; Muthike, A. K.; Howell, M.; Kim, H.; Madu, I. K.; Kim, T.; Zimmerman, P.; Lee, J. Y.; Goodson, T. New Direct Approach for Determining the Reverse Intersystem Crossing Rate in Organic Thermally Activated Delayed Fluorescent (TADF) Emitters. Journal of the American Chemical Society 142, 8074–8079 (2020). DOI: 10.1021/jacs.0c01225, https://doi.org/10.1021/jacs.0c01225
11. Chen, G.; Swartzfager, J. R.; Asbury, J. B. Matrix Dynamics and Their Crucial Role in Non-radiative Decay during Thermally Activated Delayed Fluorescence. Journal of the American Chemical Society 145, 25495–25504 (2023). DOI: 10.1021/jacs.3c11719, https://doi.org/10.1021/jacs.3c11719
12. Beer, D.; Weiser, J.; Gabler, T.; Zeitler, K.; Deibel, C.; Wiebeler, C. Beyond the Static Approximation: Assessing the Impact of Conformational and Kinetic Broadening on the Description of TADF Emitters. arXiv:2604.08483 (2026). arXiv ID: 2604.08483, https://arxiv.org/abs/2604.08483
13. Dos Santos, J. M.; Hall, D.; Basumatary, B.; Bryden, M.; Chen, D.; Choudhary, P.; Comerford, T.; Crovini, E.; Danos, A.; De, J.; Diesing, S.; Fatahi, M.; Griffin, M.; Gupta, A. K.; Hafeez, H.; Hämmerling, L.; Hanover, E.; Haug, J.; Heil, T.; Karthik, D.; Kumar, S.; Lee, O.; Li, H.; Lucas, F.; Mackenzie, C. F. R.; Mariko, A.; Matulaitis, T.; Millward, F.; Olivier, Y.; Qi, Q.; Samuel, I. D. W.; Sharma, N.; Si, C.; Spierling, L.; Sudhakar, P.; Sun, D.; Tankelevičiūtė, E.; Tonet, M. D.; Wang, J.; Wang, T.; Wu, S.; Xu, Y.; Zhang, L.; Zysman-Colman, E. The Golden Age of Thermally Activated Delayed Fluorescence Materials: Design and Exploitation. Chemical Reviews 124, 13736–14110 (2024). DOI: 10.1021/acs.chemrev.3c00755, https://doi.org/10.1021/acs.chemrev.3c00755
14. Kaji, H. (Invited) Quantitative Prediction of Exciton Dynamics. ECS Meeting Abstracts MA2024-01, 1044 (2024). DOI: 10.1149/ma2024-01131044mtgabs, https://doi.org/10.1149/ma2024-01131044mtgabs (meeting abstract)