pith. sign in

arxiv: 2511.11606 · v2 · submitted 2025-11-02 · ⚛️ physics.app-ph

Data-Driven Design Rules for TADF Emitters from a High-Throughput Screening of 747 Molecules

Jean-Pierre Tchapet Njafa , Elvira Vanelle Kameni Tcheuffa , Aissatou Maghame Foumkpou , Serge Guy Nana Engo This is my paper

Pith reviewed 2026-05-18 01:56 UTC · model grok-4.3

classification ⚛️ physics.app-ph
keywords TADF emittershigh-throughput screeningD-A-D architecturesinglet-triplet energy gaptorsional angleoscillator strengthmolecular designblue emission
0
0 comments X

The pith

D-A-D architectures with 50-90 degree twists give the smallest singlet-triplet energy gaps in TADF emitters

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper computationally screens 747 experimentally known TADF molecules to extract quantitative design guidelines for better emitters. It finds that donor-acceptor-donor frameworks produce the smallest energy difference between the first singlet and triplet excited states. A torsional angle range of 50 to 90 degrees balances that small gap with the spin-orbit coupling needed for reverse intersystem crossing. The results also flag 127 molecules predicted to meet thresholds of ΔE_ST below 0.1 eV and oscillator strength above 0.1, with clustering highlighting multi-resonance structures for blue emission. These guidelines can direct chemists toward more efficient designs for organic light-emitting devices.

Core claim

By applying a validated xTB-based high-throughput workflow to 747 known TADF molecules, the study shows that D-A-D architectures achieve the smallest ΔE_ST. Torsional angles of 50°-90° provide an effective balance between a reduced singlet-triplet gap and adequate spin-orbit coupling for reverse intersystem crossing. Clustering separates high-performance candidates and identifies multi-resonance emitters suited to blue emission, resulting in the selection of 127 molecules with predicted ΔE_ST < 0.1 eV and oscillator strength f > 0.1.

What carries the argument

D-A-D molecular architecture and 50°-90° torsional angles as the parameters that minimize ΔE_ST while preserving spin-orbit coupling for reverse intersystem crossing

If this is right

  • D-A-D frameworks should be prioritized when designing new TADF emitters to achieve smaller singlet-triplet energy gaps.
  • Torsional angles kept between 50° and 90° optimize the trade-off between small ΔE_ST and sufficient coupling for reverse intersystem crossing.
  • Multi-resonance structures stand out as promising for blue-emitting TADF materials.
  • The 127 identified candidates become direct targets for synthesis and testing in devices.
  • Clustering methods can be reused to separate high-performance molecules from the broader chemical space.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The design rules could be combined with charge-transport simulations to predict full device efficiency rather than isolated molecular properties.
  • Applying the same screening approach to solvent or solid-state environments would test how the guidelines hold under realistic operating conditions.
  • Extending the workflow to larger virtual libraries might reveal additional candidates outside the original 747.
  • Controlling molecular rigidity around the identified torsional range could improve long-term stability in operating OLEDs.

Load-bearing premise

The xTB workflow accurately ranks ΔE_ST and oscillator strength across the full chemical diversity of the 747 molecules without systematic errors that would alter the extracted design rules or candidate list.

What would settle it

Measuring ΔE_ST experimentally for several of the 127 flagged candidates and finding values above 0.1 eV would show the screening predictions do not hold.

Figures

Figures reproduced from arXiv: 2511.11606 by Aissatou Maghame Foumkpou, Elvira Vanelle Kameni Tcheuffa, Jean-Pierre Tchapet Njafa, Serge Guy Nana Engo.

Figure 1
Figure 1. Figure 1: FIG. 1. Principal component analysis score plots for the 747 TADF emitters in the gas phase. The tight clustering of data along [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. TADF performance stratified by molecular architecture. The box plots show the distribution of the singlet-triplet [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: • Cluster 0 (60%, n=449): ”Standard TADF”. These molecules exhibit moderate ∆EST(mean 0.29 eV) and visible emission (500 nm to 600 nm), representing the bulk of typical D-A and D-A-D systems. • Cluster 1 (17%, n=125): ”High-Efficiency Candidates”. This group is characterized by a small ∆EST(mean 0.18 eV), strong emission (f = 0.62), and blue-green wavelengths (450 nm to 520 nm). It is significantly enriche… view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. K-means clustering (k=4) of the 747 TADF emitters based on their standardized photophysical properties in both the [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. Validation of the optimal torsional angle design rule. The bar chart shows the percentage of molecules that are efficient [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. Correlation analysis between HOMO-LUMO overlap and key photophysical properties. The strong negative correlation [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
read the original abstract

TADF emitter performance depends on both thermodynamic and kinetic factors. We analyze 747 experimentally known TADF molecules computationally to extract quantitative design guidelines. Using a validated xTB-based workflow, we examine how architecture, geometry, and electronic structure affect photophysical properties. Among architectures, D-A-D frameworks achieve the smallest \deltaest. A favorable torsional angle of $50{\deg}-90{\deg}$ balances small $\Delta E_{\text{ST}}$ with the spin--orbit coupling needed for reverse intersystem crossing. Clustering separates high-performance candidates and highlights multi-resonance emitters for blue emission. From these results, we identify 127 candidates with predicted $\Delta E_{\text{ST}} < 0.1 eV$ and oscillator strength $f > 0.1$. These HTVS-derived design guidelines and candidates can guide future TADF emitter development.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript reports a high-throughput virtual screening of 747 experimentally known TADF molecules using a validated xTB-based workflow. It extracts quantitative design rules, finding that D-A-D architectures yield the smallest ΔE_ST, that a 50°–90° torsional angle optimally balances small ΔE_ST with sufficient spin-orbit coupling for reverse intersystem crossing, and that 127 molecules satisfy the thresholds ΔE_ST < 0.1 eV and oscillator strength f > 0.1. Clustering is used to separate high-performance candidates and highlight multi-resonance emitters for blue emission.

Significance. If the xTB rankings prove reliable across architectures, the work supplies concrete, data-driven guidelines that could accelerate rational design of efficient TADF emitters, particularly for blue OLEDs. The scale of the experimental-molecule dataset and the focus on both thermodynamic and kinetic factors constitute a useful contribution to applied organic electronics.

major comments (2)
  1. [Abstract and Methods] Abstract and Methods (validation subsection): The claim that the xTB workflow is 'validated' is not supported by reported error bars, cross-validation statistics, or an assessment of architecture-dependent systematic errors in ΔE_ST. Because the central design rules (D-A-D superiority, 50°–90° torsion optimum) and the selection of the 127 candidates rest directly on these rankings, the absence of such diagnostics leaves the quantitative thresholds and architecture comparisons vulnerable to bias that correlates with the very variables used to derive the rules.
  2. [Results] Results (architecture and torsion analysis): No quantitative comparison is shown between xTB-predicted ΔE_ST values and higher-level reference calculations (e.g., TD-DFT or experimental values) stratified by architecture class or by torsional angle. Without this, it is unclear whether the reported ordering of D-A-D versus other frameworks or the identification of the 50°–90° window survives possible method-specific errors in charge-transfer character or spin-orbit coupling.
minor comments (2)
  1. [Methods] The manuscript should specify the exact criteria and algorithm used for post-screening clustering and architecture classification, including any thresholds applied to separate multi-resonance emitters.
  2. [Figures] Figure captions and text should consistently report the number of molecules in each architectural class and the distribution of torsional angles to allow readers to judge the statistical robustness of the extracted trends.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed feedback. The points raised regarding validation metrics and stratified method comparisons are important for strengthening the reliability of the reported design rules. We respond to each major comment below and will revise the manuscript to incorporate additional supporting analyses.

read point-by-point responses

  1. Referee: [Abstract and Methods] Abstract and Methods (validation subsection): The claim that the xTB workflow is 'validated' is not supported by reported error bars, cross-validation statistics, or an assessment of architecture-dependent systematic errors in ΔE_ST. Because the central design rules (D-A-D superiority, 50°–90° torsion optimum) and the selection of the 127 candidates rest directly on these rankings, the absence of such diagnostics leaves the quantitative thresholds and architecture comparisons vulnerable to bias that correlates with the very variables used to derive the rules.

    Authors: We agree that the current manuscript would benefit from more explicit validation diagnostics. The description of the workflow as 'validated' draws on benchmarks established in our earlier publications for TADF-relevant systems, but these details are not reproduced here. In the revised manuscript we will expand the Methods section with a new validation subsection that reports MAE/RMSE values against TD-DFT and experimental ΔE_ST for a representative subset, includes error bars, and provides a breakdown of any systematic deviations by architecture class. This addition will directly support the robustness of the D-A-D and torsional-angle trends. revision: yes

  2. Referee: [Results] Results (architecture and torsion analysis): No quantitative comparison is shown between xTB-predicted ΔE_ST values and higher-level reference calculations (e.g., TD-DFT or experimental values) stratified by architecture class or by torsional angle. Without this, it is unclear whether the reported ordering of D-A-D versus other frameworks or the identification of the 50°–90° window survives possible method-specific errors in charge-transfer character or spin-orbit coupling.

    Authors: We acknowledge that a stratified comparison would increase confidence in the architecture and geometry trends. Performing higher-level calculations on the full set of 747 molecules is not feasible, but we will add to the revised Results section a quantitative comparison for a curated subset of approximately 40 molecules that evenly samples the different architecture classes and torsional-angle ranges. This will include side-by-side xTB versus TD-DFT (and available experimental) ΔE_ST values, with explicit discussion of any architecture- or angle-dependent discrepancies. We anticipate that the D-A-D preference and the 50°–90° optimum will be preserved, but any observed deviations will be reported transparently. revision: yes

Circularity Check

0 steps flagged

No significant circularity: data-driven statistical patterns from external experimental molecules

full rationale

The paper performs high-throughput computational screening of 747 experimentally known TADF molecules using an xTB workflow to extract empirical design rules regarding architectures (e.g., D-A-D), torsional angles, and property thresholds. All central claims derive from statistical analysis of computed properties on this independent external dataset rather than from internal parameter fitting, self-referential definitions, or load-bearing self-citations. No derivation step reduces by construction to the inputs (e.g., no fitted quantities renamed as predictions or ansatzes smuggled via prior self-work). The workflow is presented as validated, but the validation and screening results remain externally anchored to known molecules, keeping the chain self-contained and non-circular.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Abstract-only review limits visibility into exact computational parameters; the workflow implicitly relies on standard xTB approximations and clustering choices whose sensitivity is not quantified here.

axioms (1)
  • domain assumption xTB semi-empirical method provides sufficiently accurate relative rankings of ΔE_ST and oscillator strength for TADF molecules
    Invoked by the phrase 'validated xTB-based workflow' in the abstract; no explicit error metrics supplied.

pith-pipeline@v0.9.0 · 5706 in / 1331 out tokens · 28541 ms · 2026-05-18T01:56:59.967184+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

37 extracted references · 37 canonical work pages

  1. [1]

    Highly efficient organic light-emitting diodes from delayed fluorescence.Nature, 492(7428):234–238, December 2012

    Hiroki Uoyama, Kenichi Goushi, Katsuyuki Shizu, Hiroko Nomura, and Chihaya Adachi. Highly efficient organic light-emitting diodes from delayed fluorescence.Nature, 492(7428):234–238, December 2012. ISSN 1476-4687. doi: 10.1038/nature11687

    work page doi:10.1038/nature11687 2012

  2. [2]

    Third-generation organic electroluminescence materials.Japanese Journal of Applied Physics, 53(6): 060101, May 2014

    Chihaya Adachi. Third-generation organic electroluminescence materials.Japanese Journal of Applied Physics, 53(6): 060101, May 2014. ISSN 1347-4065. doi:10.7567/jjap.53.060101

    work page doi:10.7567/jjap.53.060101 2014

  3. [3]

    Yuchao Liu, Chensen Li, Zhongjie Ren, Shouke Yan, and Martin R. Bryce. All-organic thermally activated de- layed fluorescence materials for organic light-emitting diodes.Nature Reviews Materials, 3(4):18020, apr 2018. doi: 10.1038/natrevmats.2018.20

    work page doi:10.1038/natrevmats.2018.20 2018

  4. [4]

    Photophysics of thermally activated delayed fluorescence molecules.Methods and applications in fluorescence, 5(1):012001, mar 2017

    Fernando B Dias, Thomas J Penfold, and Andrew P Monkman. Photophysics of thermally activated delayed fluorescence molecules.Methods and applications in fluorescence, 5(1):012001, mar 2017. doi:10.1088/2050-6120/aa537e

    work page doi:10.1088/2050-6120/aa537e 2017

  5. [5]

    Xian-Kai Chen, Dongwook Kim, and Jean-Luc Br´ edas. Thermally activated delayed fluorescence (tadf) path toward efficient electroluminescence in purely organic materials: Molecular level insight.Accounts of Chemical Research, 51(9):2215, aug

    work page

  6. [6]

    doi:10.1021/acs.accounts.8b00174

    work page doi:10.1021/acs.accounts.8b00174

  7. [7]

    Moral, L

    M. Moral, L. Muccioli, W.-J. Son, Y. Olivier, and J. C. Sancho-Garc´ ıa. Theoretical rationalization of the singlet–triplet gap in oleds materials: Impact of charge-transfer character.Journal of Chemical Theory and Computation, 11(1):168–177, dec

    work page

  8. [8]

    doi:10.1021/ct500957s

    work page doi:10.1021/ct500957s

  9. [9]

    Thomas J. Penfold. On predicting the excited-state properties of thermally activated delayed fluorescence emitters.Journal of Physical Chemistry C, 119(24):13535, jun 2015. doi:10.1021/acs.jpcc.5b03530. 45

    work page doi:10.1021/acs.jpcc.5b03530 2015

  10. [10]

    Yoann Olivier, M´ onica Moral, Luca Muccioli, and Juan-Carlos Sancho-Garc´ ıa. Dynamic nature of excited states of donor-acceptor tadf materials for oleds: How theory can reveal structure-property relationships.Journal of Materials Chemistry C, 5(23):5718–5729, 2017. doi:10.1039/c6tc05075a

    work page doi:10.1039/c6tc05075a 2017

  11. [11]

    Monkman, and Thomas J

    Jamie Gibson, Andrew P. Monkman, and Thomas J. Penfold. The importance of vibronic coupling for efficient reverse intersystem crossing in thermally activated delayed fluorescence molecules.ChemPhysChem, 17(19):2956–2961, July 2016. ISSN 1439-7641. doi:10.1002/cphc.201600662

    work page doi:10.1002/cphc.201600662 2016

  12. [12]

    Haitao Sun, Cheng Zhong, and Jean-Luc Br´ edas. Reliable prediction with tuned range-separated functionals of the singlet-triplet gap in organic emitters for thermally activated delayed fluorescence.Journal of Chemical Theory and Computation, 11(8):3851–3858, jul 2015. doi:10.1021/acs.jctc.5b00431

    work page doi:10.1021/acs.jctc.5b00431 2015

  13. [13]

    Comparative study of tddft and tddft-based steom-dlpno-ccsd calculations for predicting the excited-state properties of mr-tadf.Heliyon, 10(10):e30926, May 2024

    Sunwoo Kang and Taekyung Kim. Comparative study of tddft and tddft-based steom-dlpno-ccsd calculations for predicting the excited-state properties of mr-tadf.Heliyon, 10(10):e30926, May 2024. ISSN 2405-8440. doi:10.1016/j.heliyon.2024.e30926

    work page doi:10.1016/j.heliyon.2024.e30926 2024

  14. [14]

    Olivier, B

    Y. Olivier, B. Yurash, L. Muccioli, G. D’Avino, O. Mikhnenko, J. C. Sancho-Garc´ ıa, C. Adachi, T.-Q. Nguyen, and D. Beljonne. Nature of the singlet and triplet excitations mediating thermally activated delayed fluorescence.Physical Review Materials, 1(7):075602, dec 2017. doi:10.1103/PhysRevMaterials.1.075602

    work page doi:10.1103/physrevmaterials.1.075602 2017

  15. [15]

    T. J. Penfold, E. Gindensperger, C. Daniel, and C. M. Marian. Structure-property relationships in donor-acceptor tadf emitters.Chemical Reviews, 118:6975–7025, 2018. doi:10.1021/acs.chemrev.7b00617

    work page doi:10.1021/acs.chemrev.7b00617 2018

  16. [16]

    Dingyun Huang and Jacqueline M. Cole. A database of thermally activated delayed fluorescent molecules auto-generated from scientific literature with chemdataextractor.Scientific Data, 11(1), January 2024. ISSN 2052-4463. doi:10.1038/s41597- 023-02897-3

    work page doi:10.1038/s41597- 2024

  17. [17]

    Pracht, F

    Philipp Pracht, Fabian Bohle, and Stefan Grimme. Automated exploration of the low-energy chemical space with fast quantum chemical methods.Physical Chemistry Chemical Physics, 22(14):7169–7192, 2020. ISSN 1463-9076. doi:10.1039/c9cp06869d

    work page doi:10.1039/c9cp06869d 2020

  18. [18]

    Bannwarth, S

    Christoph Bannwarth, Sebastian Ehlert, and Stefan Grimme. Gfn2-xtb—an accurate and broadly parametrized self-consistent tight-binding quantum chemical method with multipole electrostatics and density-dependent dispersion contributions. Journal of chemical theory and computation, 15(3):1652–1671, 2019. ISSN 1549-9618. doi:10.1021/acs.jctc.8b01176

    work page doi:10.1021/acs.jctc.8b01176 2019

  19. [19]

    Bannwarth, E

    Christoph Bannwarth, Eike Caldeweyher, Sebastian Ehlert, Andreas Hansen, Philipp Pracht, Jakob Seibert, Sebastian Spicher, and Stefan Grimme. Extended tight-binding quantum chemistry methods.WIREs Computational Molecular Science, 11(2):e1493, August 2021. ISSN 1759-0884. doi:10.1002/wcms.1493

    work page doi:10.1002/wcms.1493 2021

  20. [20]

    Robust and efficient implicit solvation model for fast semiempirical methods.Journal of Chemical Theory and Computation, 17(7):4250–4261, 2021

    Sebastian Ehlert, Marcel Stahn, Sebastian Spicher, and Stefan Grimme. Robust and efficient implicit solvation model for fast semiempirical methods.Journal of Chemical Theory and Computation, 17(7):4250–4261, 2021. ISSN 1549-9618. doi:10.1021/acs.jctc.1c00471

    work page doi:10.1021/acs.jctc.1c00471 2021

  21. [21]

    Stefan Grimme and Christoph Bannwarth. Ultra-fast computation of electronic spectra for large systems by tight-binding based simplified tamm-dancoff approximation (stda-xtb).The Journal of Chemical Physics, 145(5):054103, August 2016. ISSN 0021-9606. doi:10.1063/1.4959605

    work page doi:10.1063/1.4959605 2016

  22. [22]

    The exact integral simplified time-dependent density functional theory (xstd-dft)

    Marc de Wergifosse and Stefan Grimme. The exact integral simplified time-dependent density functional theory (xstd-dft). The Journal of Chemical Physics, 160(20), May 2024. ISSN 1089-7690. doi:10.1063/5.0206380

    work page doi:10.1063/5.0206380 2024

  23. [23]

    Jan-Michael Mewes. Modeling tadf in organic emitters requires a careful consideration of the environment and going beyond the franck–condon approximation.Physical Chemistry Chemical Physics, 20(18):12454–12469, 2018. doi:10.1039/c8cp01792a

    work page doi:10.1039/c8cp01792a 2018

  24. [24]

    A new design strategy for efficient thermally activated delayed fluorescence organic emitters: From twisted to planar structures.Advanced Materials, 29(46):1702767, oct 2017

    Xian-Kai Chen, Youichi Tsuchiya, Yuma Ishikawa, Cheng Zhong, Chihaya Adachi, and Jean-Luc Br´ edas. A new design strategy for efficient thermally activated delayed fluorescence organic emitters: From twisted to planar structures.Advanced Materials, 29(46):1702767, oct 2017. doi:10.1002/adma.201702767

    work page doi:10.1002/adma.201702767 2017

  25. [25]

    Isomer engineering of universal electron transport materials for suppressed exciton quenching in organic light-emitting diodes.Dyes and Pigments, 203:110319, July 2022

    Ji Ae Kang, Seung Chan Kim, and Jun Yeob Lee. Isomer engineering of universal electron transport materials for suppressed exciton quenching in organic light-emitting diodes.Dyes and Pigments, 203:110319, July 2022. ISSN 0143-7208. doi:10.1016/j.dyepig.2022.110319. URLhttp://dx.doi.org/10.1016/j.dyepig.2022.110319

    work page doi:10.1016/j.dyepig.2022.110319 2022

  26. [26]

    Lu, A comprehensive electron wavefunction analysis toolbox for chemists, Multiwfn, J

    Tian Lu. A comprehensive electron wavefunction analysis toolbox for chemists, Multiwfn.The Journal of Chemical Physics, 161(8):082503, 08 2024. ISSN 0021-9606. doi:10.1063/5.0216272. URLhttps://doi.org/10.1063/5.0216272

    work page doi:10.1063/5.0216272 2024

  27. [27]

    Toward a quantitative assessment of electronic transitions’ charge-transfer character.Journal of Chemical Theory and Computation, 10(9):3896–3905, aug 2014

    Thibaud Etienne, Xavier Assfeld, and Antonio Monari. Toward a quantitative assessment of electronic transitions’ charge-transfer character.Journal of Chemical Theory and Computation, 10(9):3896–3905, aug 2014. doi:10.1021/ct5003994

    work page doi:10.1021/ct5003994 2014

  28. [28]

    M. A. El-Sayed. Triplet state. its radiative and nonradiative properties.Accounts of Chemical Research, 1:8–16, 1968. doi:10.1021/ar50001a002

    work page doi:10.1021/ar50001a002 1968

  29. [29]

    Approaching 800 ˆA nm near-infrared thermally activated delay fluorescence materials via simple malononitrile condensation strategy.Organic Electronics, 110:106645, November

    Qian Li, Jingyi Xu, Shuer Tan, Yu Dai, Jie Xue, and Juan Qiao. Approaching 800 ˆA nm near-infrared thermally activated delay fluorescence materials via simple malononitrile condensation strategy.Organic Electronics, 110:106645, November

    work page

  30. [30]

    doi:10.1016/j.orgel.2022.106645

    ISSN 1566-1199. doi:10.1016/j.orgel.2022.106645. URLhttp://dx.doi.org/10.1016/j.orgel.2022.106645

    work page doi:10.1016/j.orgel.2022.106645 2022

  31. [31]

    Ultrapure blue thermally activated delayed fluorescence molecules: Efficient homo-lumo separation by the multiple resonance effect.Advanced Materials, 28(14):2777–2781, feb 2016

    Takuji Hatakeyama, Kazushi Shiren, Kiichi Nakajima, Shintaro Nomura, Soichiro Nakatsuka, Keisuke Kinoshita, Jingping Ni, Yohei Ono, and Toshiaki Ikuta. Ultrapure blue thermally activated delayed fluorescence molecules: Efficient homo-lumo separation by the multiple resonance effect.Advanced Materials, 28(14):2777–2781, feb 2016. doi:10.1002/adma.201505491

    work page doi:10.1002/adma.201505491 2016

  32. [32]

    Stefan Grimme. Exploration of chemical compound, conformer, and reaction space with meta-dynamics simulations based on tight-binding quantum chemical calculations.Journal of chemical theory and computation, 15(5):2847–2862, 2019. doi:10.1021/acs.jctc.9b00143

    work page doi:10.1021/acs.jctc.9b00143 2019

  33. [33]

    Marc de Wergifosse. Computing excited states of very large systems with range-separated hybrid functionals and the exact integral simplified time-dependent density functional theory (xstd-dft).The Journal of Physical Chemistry Letters, 15(51): 12628–12635, December 2024. ISSN 1948-7185. doi:10.1021/acs.jpclett.4c03193

    work page doi:10.1021/acs.jpclett.4c03193 2024

  34. [34]

    Thibaud Etienne, Xavier Assfeld, and Antonio Monari. New insight into the topology of excited states through de- tachment/attachment density matrices-based centroids of charge.Journal of Chemical Theory and Computation, 10(9): 46 3906–3914, aug 2014. doi:10.1021/ct500400s

    work page doi:10.1021/ct500400s 2014

  35. [35]

    R. A. Marcus and N. Sutin. Electron transfer reactions in chemistry. theory and experiment.Biochimica et Biophysica Acta, 811:265–322, 1985. doi:10.1016/0304-4173(85)90014-X

    work page doi:10.1016/0304-4173(85)90014-x 1985

  36. [36]

    Friederich, F

    P. Friederich, F. H¨ ase, J. Proppe, and A. Aspuru-Guzik. Machine-learned potentials for next-generation matter simulations. Nature Materials, 20:750–761, 2021. doi:10.1038/s41563-020-0777-6

    work page doi:10.1038/s41563-020-0777-6 2021

  37. [37]

    Y. H. Lee, S. Park, J. H. Oh, S. J. Shin, J. Dell, P. Cronin, T. Earmme, S. H. Woo, and S. J. Yoo. Machine learning-based design of thermally activated delayed fluorescence emitters.The Journal of Physical Chemistry Letters, 12:6970–6977, 2021. doi:10.1021/acs.jpclett.1c01553

    work page doi:10.1021/acs.jpclett.1c01553 2021