Exploring the possibility of using fluorine-involved non-conjugated electron-withdrawing groups for thermally activated delayed fluorescence emitters by TD-DFT calculation

  1. and
  2. § ORCID Logo
Organic Semiconductor Centre, EaStCHEM School of Chemistry, University of St Andrews, St Andrews, Fife, KY16 9ST, UK
  1. Corresponding author email
§ Fax: +44-1334 463808; Tel: +44-1334 463826
Guest Editor: D. O'Hagan
Beilstein J. Org. Chem. 2021, 17, 210–223. https://doi.org/10.3762/bjoc.17.21
Received 11 Nov 2020, Accepted 05 Jan 2021, Published 21 Jan 2021
Full Research Paper
cc by logo


The trifluoromethyl group has been previously explored as a non-conjugated electron-withdrawing group in donor–acceptor thermally activated delayed fluorescence (TADF) emitters. In the present study, we investigate computationally the potential of other fluorine-containing acceptors, trifluoromethoxy (OCF3), trifluoromethylthio (SCF3), and pentafluorosulfanyl (SF5), within two families of donor–acceptor TADF emitters. Time-dependent density functional theory calculations indicate that when only two ortho-disposed carbazole donors are used (Type I molecules), the lowest-lying triplet state possesses locally excited (LE) character while the lowest-lying singlet state possesses charge-transfer character. When five carbazole donors are present in the emitter design (Type II molecules), now both S1 and T1 states possess CT character. For molecules 2CzOCF3 and 5CzOCF3, the singlet energies are predicted to be 3.92 eV and 3.45 eV; however, the singlet-triplet energy gaps, ΔESTs, are predicted to be large at 0.46 eV and 0.37 eV, respectively. The compounds 2CzCF3, 2CzSCF3, and 2CzSF5, from Type I molecules, show significant promise as deep blue TADF emitters, possessing high calculated singlet energies in the gas phase (3.62 eV, 3.66 eV, and 3.51 eV, respectively) and small, ΔESTs, of 0.17 eV, 0.22 eV, and 0.07 eV, respectively. For compounds 5CzSCF3 and 5CzSF5, from Type II molecules, the singlet energies are stabilized to 3.24 eV and 3.00 eV, respectively, while ΔESTs are 0.27 eV and 0.12 eV, respectively, thus both show promise as blue or sky-blue TADF emitters. All these six molecules possess a dense number of intermediate excited states between S1 and T1, thus likely leading to a very efficient reverse intersystem crossing in these compounds.


Organic thermally activated delayed fluorescence (TADF) materials have generated significant attention recently, particularly for their use as emitters in organic light-emitting diodes (OLEDs). This is due to their ability to utilize both singlet excitons and triplet excitons, thereby increasing the theoretical internal quantum efficiency (IQE) to 100% from 25% for fluorescent compounds [1-4]. For TADF materials, a small energy gap between the lowest singlet and triplet excited states (ΔEST) is essential to permit the efficient up-conversion of triplet excitons to singlet excitons via reverse intersystem crossing (rISC) [5-7]. The rISC process can happen by hyperfine coupling when the ΔEST is sufficiently small (<10 meV) or spin orbit coupling (SOC), which requires different symmetry between the two states coupled with a relatively small singlet–triplet energy gap, ΔEST, (<300 meV) [8,9]. The ΔEST is directly dependent on the magnitude of the electron exchange energy J (Equation 1), which itself is dependent on the electron density overlap between the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) (Equation 2) [10,11]. Compounds possessing a donor–acceptor (D–A) structure could satisfy the requirements for efficient TADF if the donor and acceptor moieties are poorly conjugated with each other in order to minimize J. The HOMO/LUMO separation that controls J can be modulated by introducing strong and bulky electron donors and electron acceptors to produce large torsions between the donor and acceptor groups so as to localized the HOMO on the electron-donating moiety and to confine the LUMO on the electron-withdrawing moiety [12,13].


According to the Fermi’s golden rule, the reversed intersystem crossing rate (krISC) can be expressed as [14,15]:


Where |VSOC|2 is the spin-orbit coupling matrix element between S1 and T1 and ρFCWD is the Franck–Condon-weighted density of states, which can be expressed as [16]:


where λ is the Marcus reorganization energy associated with the intermolecular and intramolecular low-frequency vibrations; kB is Boltzmann’s constant; and T is temperature. Combing Equation 3 and Equation 4, it is evident that krISC is proportional to |VSOC|2 × exp[−(ΔEST2)]. Further, judicious molecular design in terms of the identity, position, and number of donor to acceptor moieties can also contribute to the modulation of ΔEST, leading to faster rISC. Typical donors include a small group of structurally related N-heterocycles such as carbazole [5], dimethylacridine [13], phenoxazine [17], and phenothiazine [18].

Prior studies have shown that placing the donor groups ortho to the acceptor can lead to more limited conjugation between the two, resulting in emitters with relatively smaller ΔEST compared to analogous compounds where the donor is positioned para to the acceptor [19,20]. Duan et al. have investigated the properties of D–A TADF benzonitrile-based emitters containing two carbazole donors disposed at different positions about the phenylene bridge [19]. The results showed that when the carbazoles were both located ortho to the cyano acceptor the molecule (2,6-2CzBN) possessed a highly twisted structure and a corresponding small ΔEST (0.27 eV in toluene). The ΔESTs increased to 0.41 (2,4-2CzBN) and 0.40 eV (3,5-2CzBN) in toluene when at least one of the carbazoles was disposed meta or para to the cyano acceptor [19]. OLEDs fabricated using 2,6-2CzBN as the emitter exhibited deep blue emission with λEL = 418 nm and CIE coordinate of (0.15, 0.05); however, due to the low photoluminescence quantum yields (ΦPLs) (28% in 10 wt % DPEPO films) and relatively slow krISC (0.86 × 105 s−1) in the DPEPO host, the EQEmax was only 2.5%, and showed significant efficiency roll-off, reducing to 0.1% at 50 cd·m−2 [21]. A similar study by Monkman, Lee and co-workers investigated the compound 2,6-2CzTRZ, which possessed the smallest ΔEST (0.02 eV) amongst the family of emitters possessing a diphenyltriazine as the acceptor and different regiochemistry of the carbazole donors; the ΔESTs increased to 0.10 eV for 2,4-2CzTRZ and 0.29 eV for 3,4-2CzTRZ. The single crystal structure of 2,6-2CzTRZ revealed a highly twisted structure with large torsions (81.0o and 76.3o) between the carbazole moieties and the central benzene ring; the same torsions are appreciably smaller at 45.6o and 69.6o for the molecule 2,4-2CzTRZ where one of the carbazole donors is situated at the para position and another one situated at the ortho position [20]. Compound 2,6-2CzTRZ possessed a very small ΔEST (0.02 eV) and short delayed fluorescence lifetime (τd = 16.4 μs) in zeonex [20]. These two studies illustrate that ortho-substituted D–A molecules possess highly twisted geometries, leading to spatially separated HOMO/LUMO distributions and, thus, small ΔESTs, while maintaining high energy excited states.

The presence of intermediate triplet states lying above T1 and below S1 have been shown to facilitate rISC and render TADF more efficient by opening up a reverse internal conversion (RIC) pathway that is mediated by spin-vibronic coupling between T1 and one or more of the intermediate states, followed by rISC [22]. This situation typically occurs when there are multiple donors about a single acceptor as exists in the molecules 5CzBN and 5CzTRZ. For 5CzBN, time-dependent density functional theory (TD-DFT) calculation revealed the existence of three intermediate triplet states [22]. The presence of these states helped to explain the short τd of 3.7 μs and the high EQEmax of 17% and good device stability with a T50 of 176 hours for the OLED [CIE coordinate (0.22, 0.40)] [23]. In an analogous manner, TD-DFT calculations predicted 5CzTRZ to possess a small ΔEST (0.02 eV) as well as a small energy gap (≈0.24 eV) between T2 and T1 [24]. In an analogous manner, 5CzTRZ showed very fast krISC of ≈1.5 × 107 s−1 in toluene, and the device based on 5CzTRZ exhibited superior EQEmax = 29% with λEL = 486 nm and very low efficiency roll-off with the EQE at 5,000 cd·m−2 remaining high at 27% [24]. Huang et al. also adopted a multiple donor strategy in concert with the weak trifluoromethyl (CF3) acceptor group in their TADF emitter design. The blue-emitting TADF emitter 5CzCF3 possessed a miniscule measured ΔEST of 0.02 eV and ΦPL of 43% in oxygen-free toluene [25]. The solution-processed device based on 5CzCF3 exhibited sky-blue emission with CIE coordinates of (0.21, 0.33) and an EQEmax of 5.2% at 1 cd·m−2 [25].

The promising performance of emitters possessing a CF3 acceptor group prompted us to investigate other fluorinated weakly-conjugated acceptor units in order to assess their potential within TADF emitter design (Figure 1) [25-27]. In the present study, we report on the impact of incorporating other fluorine-containing electron-withdrawing groups beyond trifluoromethyl (CF3), including trifluoromethoxy (OCF3), trifluoromethylthio (SCF3), and pentafluorosulfanyl (SF5) groups, and explore their potential computationally within TADF emitter design. We cross-compare their optoelectronic properties with analog materials using well-studied conjugated electron-withdrawing groups (cyano, benzophenone, and triazine). We investigated two families of structures. The first family consists of D–A–D (Type I) molecules containing two carbazole donors disposed each ortho to the acceptor group, while the second family consists of five carbazole donors substituted about a central benzene ring and the sixth position occupied by the acceptor moiety (Type II). Adachi et al. have shown that compounds that fall within the Type I family can simultaneously show high singlet and triplet energies and small ΔEST while compounds that are a part of Type II family possess a more dense number of low-lying excited states [22], the presence of which has been shown to assist in the rISC process through spin-vibronic coupling [23,24,27]. The energy levels and electronic configurations of S1 and T1 in these molecules were analysed and we found that compounds possessing either SCF3 and SF5 groups as acceptors (2CzSCF3/2CzSF5 in Type I, 5CzSCF3/5CzSF5 in Type II), possessed LUMOs that are mainly located on the central benzene ring and the acceptor group while the HOMOs are mainly localized on the carbazoles, thereby leading to small ΔESTs. The calculated ΔESTs for 2CzSCF3/2CzSF5 are 0.22 eV and 0.07 eV, respectively, which are comparable to the calculated results for 2CzBN (0.18 eV) and 2CzTRZ (0.08 eV); likewise, the calculated ΔESTs for 5CzSCF3/5CzSF5 are 0.27 eV and 0.12 eV, respectively, which are close to the calculated results of 5CzBN (0.20 eV) and 5CzTRZ (0.17 eV). The molecules incorporating an OCF3 acceptor (2CzOCF3 in Type I, 5CzOCF3 in Type II), however, exhibited relatively larger ΔESTs (0.46 eV for 2CzOCF3, 0.37 eV for 5CzOCF3). The calculated S1 energies of 2CzOCF3 (3.92 eV), 2CzSCF3 (3.62 eV), 2CzSF5 (3.51 eV), and 5CzOCF3 (3.45 eV) demonstrate that these molecules show potential as deep blue emitters as their S1 states are higher in energy than that of 2CzBN (3.34 eV calculated in gas phase in this work), which was reported as deep blue emitter with λEL = 418 nm and CIE coordinate of (0.15, 0.05) when doped in DPEPO [21]. DFT calculations for 5CzOCF3, 5CzSCF3, and 5CzSF5 predicted dense populations of excited states between T1 and S1, which should assist in rISC process [28,29].


Figure 1: Molecular structures of emitters discussed in this work.

Results and Discussion

We employed density functional theory (DFT) and TD-DFT calculations to predict the photophysical properties of these emitters in order to assess their potential as TADF emitters for OLEDs. All ground-state calculations were performed using PBE0/6-31G(d,p) in the gas phase [30,31]. The lowest energy structures from these DFT calculations were used as input geometries for excited-state calculations using the Tamm–Dancoff approximation (TDA) to TD-DFT, which provide computed energies of the excited singlet and triplet states [32,33]. The nature of the lowest singlet and triplet states were ascertained by an analysis of the natural transition orbitals (NTO) obtained from the TDA-DFT calculations [34].

We first investigated the strength of the acceptor groups by modelling phenyl-substituted acceptors and compared their LUMO energies as well as the energies of the S1 and T1 states (Figure 2). Among the fluorinated electron-withdrawing groups in the study, PhOCF3 possesses the shallowest LUMO at −0.22 eV while PhSF5 possess the deepest LUMO at −0.90 eV, with PhSCF3 (−0.78 eV) and PhCF3 (−0.57 eV) possessing intermediate values. The LUMO energies of these four acceptors correlate linearly to the Hammett substituent constant, σp, (Figure 2c) [35]. All of these fluorinated acceptors are much weaker than the more commonly investigated benzonitrile (BN, −1.30 eV), triphenyltriazine (TRZ, −1.72 eV) and benzophenone (BP, −1.58 eV) acceptors. These results indicate that the use of the fluorinated acceptor groups in donor–acceptor TADF emitters should lead to a pronounced blue-shift in the emission, as reflected in the higher-energy singlet states of the model systems in Figure 2.


Figure 2: a) Calculated HOMO, LUMO, S1 and T1 energies, as well as HOMO and LUMO topologies of PhCF3, PhOCF3, PhOSCF3, and PhSF5, b) Calculated HOMO, LUMO, S1 and T1 energies, as well as HOMO and LUMO topologies of BN, TRZ and BP (isovalue = 0.02). c) Hammett para substituent values (σp) relationship with the calculated LUMO energies for fluorine-containing acceptors PhCF3, PhOCF3, PhOSCF3, and PhSF5.

We next modelled the Type I emitters (Figure 3 and Figure 4). The DFT-calculated geometries indicate that the carbazoles adopt a significantly twisted conformation (dihedral angles > 50o) in order to minimize their interaction with the acceptor group. Specifically, for 2CzCF3 the carbazoles are twisted to 60.2o and 70.5o with respect to the bridging phenyl ring while for 2CzSF5, due to the increased bulkiness of the SF5 group, the corresponding twist angle increased to 78.5o and 78.7o. These highly twisted conformations contribute to the spatial separation of the HOMO and LUMO.


Figure 3: Calculated HOMO, LUMO, S1 and T1 energies, as well as HOMO and LUMO topologies of 2CzCF3, 2CzOCF3, 2CzSCF3, and 2CzSF5 (isovalue = 0.02).


Figure 4: Calculated HOMO, LUMO, S1 and T1 energies, as well as HOMO and LUMO topologies of 2CzBN, 2CzTRZ, and 2CzBP (isovalue = 0.02).

Figure 3 shows the energies of the HOMOs and LUMOs and the S1 and T1 states for the fluorinated acceptor-containing emitters 2CzCF3, 2CzOCF3, 2CzSCF3, and 2CzSF5. The HOMOs in these compounds are mainly located on the two carbazole moieties and a small part on the bridging central benzene ring. The LUMOs of 2CzCF3, 2CzSCF3, and 2CzSF5 are mainly located on the benzene ring and a small distribution onto the electron-withdrawing group, whereas the LUMO of 2CzOCF3 is localized essentially only on the central benzene. Emitters 2CzCF3, 2CzOCF3, and 2CzSCF3 show similarly deep HOMO values at around −5.80 eV, while the HOMO level of 2CzSF5 is more stabilized at −5.89 eV. The trend in LUMO energies matches that observed for the model acceptors (Figure 2) where 2CzOCF3 possesses the shallowest LUMO of −0.95 eV while 2CzSF5 possesses the deepest LUMO level of −1.46 eV. 2CzOCF3 possesses the largest energy gap (ΔEg) at 4.83 eV while the ΔEg for 2CzSF5 is the smallest at 4.43 eV amongst these four compounds. Figure 4 shows the corresponding data for the Type I reference compounds 2CzBN, 2CzTRZ, and 2CzBP. In these three compounds the HOMOs are located mostly on the two carbazole moieties, with only a small contribution from the bridging benzene ring; this latter contribution is most pronounced for 2CzBN, which leads to the greatest stabilization of the HOMO level at −5.89 eV. 2CzTRZ, and 2CzBP possess destabilized HOMO levels of −5.69 and −5.60 eV, respectively. The LUMOs of 2CzBN, 2CzTRZ and 2CzBP are each located on the bridging benzene ring and the electron-acceptor groups. The LUMO levels for 2CzBN, 2CzTRZ, and 2CzBP of −1.70 eV, −1.63 eV, and −1.67 eV, respectively, are much deeper those of the fluorine-containing emitters in Figure 3, which is a reflection of the greater conjugation length present in compounds with an extended π-accepting framework. The corresponding ΔEg of 2CzBN (4.19 eV), 2CzTRZ (4.06 eV), and 2CzBP (3.93 eV) are all significantly smaller compared to those of 2CzCF3, 2CzOCF3, 2CzSCF3, and 2CzSF5.

The emissive S1 state for the seven Type I molecules is characterized mainly by a HOMO to LUMO transition, while the distribution of highest occupied natural transition orbitals (HONTOs) and the lowest unoccupied natural transition orbitals (LUNTOs) show good agreement with the HOMOs and LUMOs (Figure 5 and Figure 6). As the HOMOs and LUMOs of the seven molecules are sufficiently separated, the nature of the S1 is charge-transfer (CT) in character. The S1 energies of 2CzCF3, 2CzOCF3, 2CzSCF3, and 2CzSF5 are much higher than those of 2CzBN, 2CzTRZ, and 2CzBP. 2CzOCF3 possesses the highest S1 at 3.92 eV followed by 2CzSCF3 (3.66 eV) and 2CzCF3 (3.62 eV). The S1 of 2CzSF5 at 3.51 eV is relatively more stabilized due to the stronger electron-withdrawing ability of the SF5 group. The S1 states of 2CzBN, 2CzTRZ, and 2CzBP are 3.34 eV, 3.22 eV, and 3.09 eV, respectively. The calculated S1 values are slightly destabilized relative to the literature reported values for 2CzBN (3.27 eV in toluene [19]) and 2CzTRZ (3.12 eV in zeonex [20]).


Figure 5: HOMO and LUMO distribution, HONTO and LUNTO of lowest singlet (S1) and triplet excited (T1) states for compounds 2CzCF3, 2CzOCF3, 2CzSCF3, and 2CzSF5 (isovalue = 0.02).


Figure 6: HOMO and LUMO distribution, HONTO and LUNTO of lowest singlet (S1) and triplet excited (T1) states for compounds 2CzBN, 2CzTRZ, and 2CzBP (isovalue = 0.02).

The nature of the T1 state of 2CzCF3, 2CzOCF3, and 2CzSF5 is of locally excited (LE) character on the carbazole, while for 2CzSCF3 the T1 state is also LE, but also involving the bridging benzene ring. These assignments are reflected in very similar T1 energies of around 3.45 eV. The corresponding ΔEST values are 0.17 eV for 2CzCF3, 0.46 eV for 2CzOCF3, 0.22 eV for 2CzSCF3 and 0.07 eV for 2CzSF5; thus, with the exception of 2CzOCF3, the small singlet-triplet energy gaps coupled with the large difference in symmetry between S1 and T1 augers well for efficient deep blue TADF emitters. By contrast, the triplet states of 2CzBN, 2CzTRZ, and 2CzBP are best characterized by HOMO to LUMO CT-type transition. The calculated T1 values for 2CzBN, 2CzTRZ, and 2CzBP are 3.16 eV, 3.14 eV, and 3.00 eV, respectively. These values are slightly destabilized compared to the literature reported values for 2CzBN (3.03 eV in toluene [19]) and 2CzTRZ (3.05 eV in zeonex [20]). The corresponding ΔEST values are generally smaller than those of the Type I fluorinated compounds with values of 0.08 eV for 2CzTRZ, 0.09 eV for 2CzBP and 0.18 eV for 2CzBN; however, the similar orbital symmetries between S1 and T1 would render rISC between these two states less efficient. The calculated ΔEST values are close to the literature reported values for 2CzBN (0.27 eV in toluene [19]) and 2CzTRZ (0.07 eV in zeonex [20]).

Inspired by these results, we next extended our theoretical study to Type II compounds where we increased the number of carbazole donor groups from two to five. We expect this design to lead to improved spatial separation of the electron density distributions between the HOMO and LUMO, thereby strengthening the CT character of the S1 state and leading to smaller ΔEST values, and thus more efficient TADF. The HOMO and LUMO distributions and energies for the Type II emitters are shown in Figure 7 and Figure 8. The HOMOs of 5CzCF3, 5CzOCF3, and 5CzSCF3 are mainly located on the carbazole moieties located ortho and meta to the acceptor group, with only a small distribution on the para-carbazole. For 5CzSF5, the HOMO is evenly distributed over the five carbazole moieties. The LUMOs of 5CzCF3, 5CzSCF3, and 5CzSF5 are mainly located on the bridging benzene ring and the electron-withdrawing groups along with a small contribution from the para-disposed carbazole, whereas the LUMO of 5CzOCF3 is located only on the central benzene ring, a similar behavior to 2CzOCF3. Compounds 5CzCF3, 5CzSCF3, and 5CzSF5 showed similarly deep HOMO values of around −5.65 eV, while the HOMO value of 5CzOSF3 is more stabilized at −5.73 eV. The 5CzOCF3 possesses the most destabilized LUMO level at −1.41 eV, while 5CzSF5 possesses the deepest LUMO level at −1.80 eV. The LUMO values for 5CzCF3 and 5CzSCF3 are −1.61 eV and −1.63 eV, respectively. 5CzOCF3 has, therefore, the largest energy gap (ΔEg) at 4.32 eV while 5CzSF5 has the smallest at 3.85 eV; both 5CzCF3 and 5CzOCF3 possess ΔEg of 4.03 eV. The trends for the HOMO and LUMO energies for these five Type II emitters mirror those observed for their Type I analogues; however, the HOMO and LUMO values in the Type II emitters are more stabilized and the energy gaps are reduced.


Figure 7: Calculated HOMO, LUMO, S1 and T1 energies, as well as HOMO and LUMO topologies of 5CzCF3, 5CzOCF3, 5CzSCF3, and 5CzSF5 (isovalue = 0.02).


Figure 8: Calculated HOMO, LUMO, S1 and T1 energies, as well as HOMO and LUMO topologies of 5CzBN, 5CzTRZ, and 5CzBP (isovalue = 0.02).

The HOMO of 5CzBN is symmetrically distributed across the ortho- and meta-disposed carbazoles while the HOMO of 5CzTRZ is located mostly on the meta- and para-carbazoles. For 5CzBP, due to the asymmetric structure, the HOMO is located on one side of ortho- and meta-disposed carbazoles while the pseudo-degenerate HOMO−1 is located on the other ortho- and meta-disposed carbazoles. The LUMOs of 5CzBN, 5CzBP, and 5CzTRZ are each located on the central benzene ring and extending onto the electron-withdrawing group. The HOMO of 5CzBN is deepest at −5.74 eV, similar to that calculated for 5CzOCF3, while the HOMOs of 5CzBP and 5CzTRZ are −5.59 and −5.60 eV, respectively. The LUMO values of 5CzBN, and 5CzBP are −1.98 eV, and −1.89 eV, respectively, which are significantly more stabilized than the fluorinated Type II emitters while the LUMO of 5CzTRZ at −1.67 eV is similar to those predicted for 5CzCF3 (−1.61 eV) and 5CzSCF3 (−1.63 eV). The ΔEg values of 5CzBN (3.76 eV), 5CzTRZ (3.93 eV), and 5CzBP (3.70 eV) are all slightly smaller than those of the fluorinated Type II emitters.

The HONTOs and LUNTOs for the Type II emitters are shown in Figure 9 and Figure 10. These generally reflect the HOMO and LUMO distributions, save for 5CzTRZ where the HONTO of S1 is located on the ortho-carbazoles. Due to the sufficiently large separation of the electron densities between the HOMO and LUMO of each of the seven Type II emitters, the S1 state for each of these possesses CT character, analogously to those calculated for the Type I compounds. 5CzOCF3 possesses the highest S1 energy (3.45 eV) among Type II molecules, followed by 5CzSCF3 (3.24 eV) and 5CzCF3 (3.20 eV). The S1 of 5CzSF5 is 3.00 eV, which is close to the values of 5CzBN (2.98 eV), 5CzTRZ (3.08 eV) and 5CzBP (2.91 eV). The calculated S1 values are more destabilized than the literature reported values of 5CzBN (2.90 eV in toluene [23]), 5CzTRZ (2.85 eV in toluene [24]) and 5CzCF3 (2.82 eV in toluene [25]). The nature of the T1 state for each of these compounds is CT where the HONTOs of T1 are mainly located on the carbazole moieties (and sometimes the central benzene) while the LUNTOs of T1 are mainly located on the benzene ring and electron-withdrawing groups, except for 5CzOCF3 where the LUNTO is located only on the benzene. 5CzOCF3 possesses the highest T1 energy (3.08 eV), while the T1 energies of 5CzCF3, 5CzSCF3, and 5CzSF5 are stabilized at 2.99 eV, 2.96 eV, and 2.88 eV, respectively. The T1 energy of 5CzTRZ is 2.91 eV while those of 5CzBN and 5CzBP are more stabilized at 2.78 eV and 2.80 eV, respectively. The calculated T1 energies match the literature reported value of 5CzBN (2.78 eV in toluene [23]) and are slightly destabilized relative to the literature reported value of 5CzTRZ (2.79 eV in toluene [24]) and 5CzCF3 (2.82 eV in toluene [25]). The corresponding ΔEST value of 5CzOCF3 is 0.37 eV, which is reduced by 0.11 eV compared to 2CzOCF3 (0.46 eV). This reduction results from the greater CT character in both S1 and T1. However, as the HOMO/LUMO overlap includes a small distribution on para-disposed carbazole in the Type II emitters with the exception of 5CzOCF3, the ΔEST values of Type II emitters are generally slightly larger compared to their Type I congeners. The ΔESTs of 5CzCF3, 5CzSCF3 and 5CzSF5 are 0.21 eV, 0.27 eV, and 0.12 eV, respectively, which are 0.04 eV, 0.05 eV, and 0.05 eV, respectively larger compared to 2CzCF3 (0.17 eV), 2CzSCF3 (0.22 eV), and 2CzSF5 (0.07 eV). The ΔESTs of 5CzBN and 5CzBP are 0.20 eV and 0.11 eV, which are only 0.02 eV larger compared to 2CzBN (0.18 eV) and 2CzBP (0.09 eV), while the ΔEST for 5CzTRZ is 0.17 eV, which is 0.09 eV larger than that of 2CzTRZ (0.08 eV). The calculated ΔEST values are slightly larger than the literature reported values for 5CzBN (0.12 eV in toluene [23]) and 5CzTRZ (0.06 eV in toluene [24]).


Figure 9: HOMO and LUMO distribution, HONTO and LUNTO of lowest singlet (S1) and triplet excited (T1) states for compounds 5CzCF3, 5CzOCF3, 5CzSCF3, and 5CzSF5 (isovalue = 0.02).


Figure 10: HOMO and LUMO distribution, HONTO and LUNTO of lowest singlet (S1) and triplet excited (T1) states for compounds 5CzBN, 5CzTRZ, and 5CzBP (isovalue = 0.02).

The spin-orbit coupling (SOC) values between excited singlet and triplets were calculated by considering the three T1 substates (m = 0, ±1) are degenerate and the |VSOC|2 as the average of the three spin-orbit coupling matrix elements (SOCME) between singlet and the triplet states [36]. The results are summarized in Table 1. Among the Type I molecules, 2CzSCF3 possesses the highest |VSOC|2 value as 0.148 cm−2, followed by 2CzBP (0.070 cm−2) and 2CzSF5 (0.053 cm−2). The |VSOC|2 values for 2CzCF3 and 2CzOCF3 are 0.011 cm−2 and 0.019 cm−2, respectively, which are still much higher than 2CzBN (0.002 cm−2) and 2CzTRZ (3 × 10−4 cm−2). The Type II molecules show an increase in |VSOC|2 values compared to their Type I counterparts. 5CzSCF3 possesses the highest |VSOC|2 value at 0.750 cm−2 which is five times higher than 2CzSF5, and 5CzSF5 possesses the second highest |VSOC|2 value as 0.718 cm−2, which is more than thirteen times higher than 2CzSF5. The higher |VSOC|2 values of 2CzSCF3/5CzSCF3 and 2CzSF5/5CzSF5 can be ascribed to the presence of the relatively heavier chalcogen, which has also been attributed by Duan et al. to much higher SOCME values in a sulfur-containing emitter than in analogs without the sulfur atom present [37]. The |VSOC|2 values of 5CzBN and 5CzBP increased to 0.298 cm−2 and 0.267 cm−2, respectively, which are more than one hundred times higher than 2CzBN and four times higher than 2CzBP. The |VSOC|2 values of 5CzCF3 and 5CzOCF3 are also higher at 0.090 cm−2 and 0.060 cm−2, respectively. The |VSOC|2 value of 5CzTRZ also increased to 0.001 cm−2 from 3 × 10−4 cm−2 for 2CzTRZ; however, the predicted |VSOC|2 value between S1 and T2 (0.107 cm−2) is much higher (Table S14, Supporting Information File 1). A measure of the magnitude of krISC can be ascertained from |VSOC|2 × exp[−(ΔEST2)]. The trends align here are consistent with the SOCME calculations. By comparison, the experimentally inferred krISC for 2CzBN, 5CzBN and 5CzTRZ are 0.86 × 105 s−1 in DPEPO film [21], 2.2 × 105 s−1 in toluene [22], and 1.5 × 107 s−1 in toluene [24], respectively. The trend in experimental krISC for 2CzBN and 5CzBN match our SOCME calculations as 5CzBN possesses the third highest |VSOC|2 × exp[−(ΔEST2)] while 2CzBN has the third lowest value. Clearly, for 5CzTRZ there is a lack of correlation between the computed |VSOC|2 × exp[−(ΔEST2)] and the experimentally determined krISC values. The significantly higher experimental krISC can be explained by the presence of intermediate triplet states leading to second order spin-vibronic coupling to mediate rISC in 5CzTRZ [24]; indeed, the |VSOC|2 value was predicted to be much higher by the SOCME calculations between S1 and T2 at 0.107 cm−2.

Table 1: S1 and T1 energies, ΔEST, and average |VSOC|2 values of Type I and Type II molecules.

Compound S1 [eV] T1 [eV] ΔEST [eV] |VSOC|2 [cm−2] |VSOC|2 × exp[−(ΔEST2)]
2CzCF3 3.62 3.45 0.17 0.011 1.48 × 10−10
2CzOCF3 3.92 3.46 0.46 0.019 2.20 × 10−10
2CzSCF3 3.66 3.44 0.22 0.148 2.03 × 10−9
2CzSF5 3.51 3.44 0.07 0.053 7.54 × 10−10
2CzBN 3.34 3.16 0.18 0.002 3.07 × 10−11
2CzBP 3.22 3.14 0.08 0.070 1.00 × 10−9
2CzTRZ 3.09 3.00 0.09 3 × 10−4 4.29 × 10−12
5CzCF3 3.20 2.99 0.21 0.090 1.24 × 10−9
5CzOCF3 3.45 3.08 0.37 0.060 7.51 × 10−10
5CzSCF3 3.24 2.96 0.27 0.750 1.00 × 10−8
5CzSF5 3.00 2.88 0.12 0.718 1.02 × 10−8
5CzBN 2.98 2.78 0.20 0.298 4.12 × 10−9
5CzBP 3.08 2.91 0.17 0.267 3.74 × 10−9
5CzTRZ 2.91 2.80 0.11 0.001 1.57 × 10−11

Prior studies on 5CzBN and 5CzTRZ showed that intermediate excited states between S1 and T1 can facilitate the rISC process by providing extra rISC transition channels from the higher intermediate excited triplet states to S1 thereby improving the rISC rate [22,24]. The presence of multiple donors, each possessing slightly different conformations, and thereby presenting slightly different electronic coupling with the central acceptor guarantees a dense population of excited states [22,24]. We analysed the higher excited states of the fluorinated acceptor-containing emitters in both Type I and Type II structures. For 2CzCF3, the T1 is locally excited; further, T2 (3.46 eV) to T6 (3.58 eV) all exhibited significant LE character. The lowest triplet state that exhibits charge transfer characteristics is T7 at 3.72 eV (Figure 11). By contrast, the T1 of 5CzCF3 exhibited CT character and the higher triplet states from T2 to T6 also exhibited CT character, which is a similar picture to the literature reported calculated electronic structure of 5CzBN using TD-DFT/ωB97XD [22] (Figure 12). This change from mostly low-lying LE triplet states in Type I emitters to mostly low-lying CT states in Type II emitters is prevalent in 2CzOCF3/5CzOCF3, 2CzSCF3/5CzSCF3, and 2CzSF5/5CzSF5 (Figures S1–S6, Supporting Information File 1). Both Type I and Type II molecules are predicted to possess multiple intermediate excited states between S1 and T1. For example, for 2CzCF3 T2 to T6 lie between S1 and T1 and the energy gap between T6 and S1ES1T6) is 0.04 eV while for 5CzCF3 the T2 to T4 are intermediate states with energies below S1 and the energy gap between T4 and S1ES1T4) is 0.02 eV. This phenomenon is also observed in 2CzOCF3ES1T6 = 0.08 eV)/5CzOCF3ES1T8 = 0.02 eV), and 2CzSCF3ES1T6 = 0.09 eV)/5CzSCF3ES1T4 = 0.00 eV), 2CzSF5ES1T3 = 0.01 eV)/5CzCF3ES1T4 = 0.00 eV). We thus contend that the intermediate excited states present in the fluorinated acceptor-containing emitters will assist in the rISC process, and improve the TADF characteristics, mitigating the somewhat larger ΔEST values in these compounds.


Figure 11: HONTOs and LUNTOs of 2CzCF3 in higher excited states (isovalue = 0.02).


Figure 12: HONTOs and LUNTOs of 5CzCF3 in higher excited states (isovalue = 0.02).


This computational study demonstrates the high potential of fluorinated acceptors in TADF emitter design. In particular, we showed that OCF3, SCF3 and SF5 groups should all be considered when designing deep blue TADF emitters. Type II emitters, with five carbazole donors, showed the most promise in terms of suitable small ΔEST values, high spin-orbit coupling values coupled with a relatively large density of intermediate excited triplet states that can be recruited to render TADF more efficient. Present efforts are ongoing to synthesize promising candidates from this theoretical study.

Supporting Information

The research data underpinning this publication can be accessed at https://doi.org/10.17630/b8f9f445-60a0-4c0a-808e-ce27cfcbf48a

Supporting Information File 1: Calculation details, Cartesian coordinates of all the molecules, SOCME calculation result, and HONTOs and LUNTOs of 2CzCF3/5CzCF3, 2CzOCF3/5CzOCF3, 2CzSCF3/5CzSCF3, and 2CzSF5/5CzSF5 in higher-lying excited states are available in supporting information.
Format: PDF Size: 3.6 MB Download


We thank Oliver Lee for help with the SOCME calculations.


Dongyang Chen thanks the China Scholarship Council (201603780001). We acknowledge support from the Engineering and Physical Sciences Research Council of the United Kingdom (grant EP/P010482/1).


  1. Wong, M. Y.; Zysman-Colman, E. Adv. Mater. 2017, 29, 1605444. doi:10.1002/adma.201605444
    Return to citation in text: [1]
  2. Liu, Y.; Li, C.; Ren, Z.; Yan, S.; Bryce, M. R. Nat. Rev. Mater. 2018, 3, 18020. doi:10.1038/natrevmats.2018.20
    Return to citation in text: [1]
  3. Cai, X.; Su, S.-J. Adv. Funct. Mater. 2018, 28, 1802558. doi:10.1002/adfm.201802558
    Return to citation in text: [1]
  4. Teng, J.-M.; Wang, Y.-F.; Chen, C.-F. J. Mater. Chem. C 2020, 8, 11340–11353. doi:10.1039/d0tc02682d
    Return to citation in text: [1]
  5. dos Santos, P. L.; Chen, D.; Rajamalli, P.; Matulaitis, T.; Cordes, D. B.; Slawin, A. M. Z.; Jacquemin, D.; Zysman-Colman, E.; Samuel, I. D. W. ACS Appl. Mater. Interfaces 2019, 11, 45171–45179. doi:10.1021/acsami.9b16952
    Return to citation in text: [1] [2]
  6. Li, W.; Li, B.; Cai, X.; Gan, L.; Xu, Z.; Li, W.; Liu, K.; Chen, D.; Su, S.-J. Angew. Chem., Int. Ed. 2019, 58, 11301–11305. doi:10.1002/anie.201904272
    Return to citation in text: [1]
  7. Wang, Y.-K.; Huang, C.-C.; Ye, H.; Zhong, C.; Khan, A.; Yang, S.-Y.; Fung, M.-K.; Jiang, Z.-Q.; Adachi, C.; Liao, L.-S. Adv. Opt. Mater. 2020, 8, 1901150. doi:10.1002/adom.201901150
    Return to citation in text: [1]
  8. Gibson, J.; Monkman, A. P.; Penfold, T. J. ChemPhysChem 2016, 17, 2956–2961. doi:10.1002/cphc.201600662
    Return to citation in text: [1]
  9. Etherington, M. K.; Gibson, J.; Higginbotham, H. F.; Penfold, T. J.; Monkman, A. P. Nat. Commun. 2016, 7, 13680. doi:10.1038/ncomms13680
    Return to citation in text: [1]
  10. Agou, T.; Matsuo, K.; Kawano, R.; Park, I. S.; Hosoya, T.; Fukumoto, H.; Kubota, T.; Mizuhata, Y.; Tokitoh, N.; Yasuda, T. ACS Mater. Lett. 2020, 2, 28–34. doi:10.1021/acsmaterialslett.9b00433
    Return to citation in text: [1]
  11. Penfold, T. J.; Gindensperger, E.; Daniel, C.; Marian, C. M. Chem. Rev. 2018, 118, 6975–7025. doi:10.1021/acs.chemrev.7b00617
    Return to citation in text: [1]
  12. Sharma, N.; Spuling, E.; Mattern, C. M.; Li, W.; Fuhr, O.; Tsuchiya, Y.; Adachi, C.; Bräse, S.; Samuel, I. D. W.; Zysman-Colman, E. Chem. Sci. 2019, 10, 6689–6696. doi:10.1039/c9sc01821b
    Return to citation in text: [1]
  13. Meng, G.; Chen, X.; Wang, X.; Wang, N.; Peng, T.; Wang, S. Adv. Opt. Mater. 2019, 7, 1900130. doi:10.1002/adom.201900130
    Return to citation in text: [1] [2]
  14. Robinson, G. W.; Frosch, R. P. J. Chem. Phys. 1963, 38, 1187–1203. doi:10.1063/1.1733823
    Return to citation in text: [1]
  15. Lawetz, V.; Orlandi, G.; Siebrand, W. J. Chem. Phys. 1972, 56, 4058–4072. doi:10.1063/1.1677816
    Return to citation in text: [1]
  16. Schmidt, K.; Brovelli, S.; Coropceanu, V.; Beljonne, D.; Cornil, J.; Bazzini, C.; Caronna, T.; Tubino, R.; Meinardi, F.; Shuai, Z.; Brédas, J.-L. J. Phys. Chem. A 2007, 111, 10490–10499. doi:10.1021/jp075248q
    Return to citation in text: [1]
  17. Chen, Z.; Wu, Z.; Ni, F.; Zhong, C.; Zeng, W.; Wei, D.; An, K.; Ma, D.; Yang, C. J. Mater. Chem. C 2018, 6, 6543–6548. doi:10.1039/c8tc01698d
    Return to citation in text: [1]
  18. Wang, K.; Shi, Y.-Z.; Zheng, C.-J.; Liu, W.; Liang, K.; Li, X.; Zhang, M.; Lin, H.; Tao, S.-L.; Lee, C.-S.; Ou, X.-M.; Zhang, X.-H. ACS Appl. Mater. Interfaces 2018, 10, 31515–31525. doi:10.1021/acsami.8b08083
    Return to citation in text: [1]
  19. Zhang, D.; Cai, M.; Bin, Z.; Zhang, Y.; Zhang, D.; Duan, L. Chem. Sci. 2016, 7, 3355–3363. doi:10.1039/c5sc04755b
    Return to citation in text: [1] [2] [3] [4] [5] [6]
  20. Oh, C. S.; de Sa Pereira, D.; Han, S. H.; Park, H.-J.; Higginbotham, H. F.; Monkman, A. P.; Lee, J. Y. ACS Appl. Mater. Interfaces 2018, 10, 35420–35429. doi:10.1021/acsami.8b10595
    Return to citation in text: [1] [2] [3] [4] [5] [6]
  21. Chan, C.-Y.; Cui, L.-S.; Kim, J. U.; Nakanotani, H.; Adachi, C. Adv. Funct. Mater. 2018, 28, 1706023. doi:10.1002/adfm.201706023
    Return to citation in text: [1] [2] [3]
  22. Noda, H.; Chen, X.-K.; Nakanotani, H.; Hosokai, T.; Miyajima, M.; Notsuka, N.; Kashima, Y.; Brédas, J.-L.; Adachi, C. Nat. Mater. 2019, 18, 1084–1090. doi:10.1038/s41563-019-0465-6
    Return to citation in text: [1] [2] [3] [4] [5] [6] [7]
  23. Zhang, D.; Cai, M.; Zhang, Y.; Zhang, D.; Duan, L. Mater. Horiz. 2016, 3, 145–151. doi:10.1039/c5mh00258c
    Return to citation in text: [1] [2] [3] [4] [5]
  24. Cui, L.-S.; Gillett, A. J.; Zhang, S.-F.; Ye, H.; Liu, Y.; Chen, X.-K.; Lin, Z.-S.; Evans, E. W.; Myers, W. K.; Ronson, T. K.; Nakanotani, H.; Reineke, S.; Bredas, J.-L.; Adachi, C.; Friend, R. H. Nat. Photonics 2020, 14, 636–642. doi:10.1038/s41566-020-0668-z
    Return to citation in text: [1] [2] [3] [4] [5] [6] [7] [8] [9] [10]
  25. Mei, L.; Hu, J.; Cao, X.; Wang, F.; Zheng, C.; Tao, Y.; Zhang, X.; Huang, W. Chem. Commun. 2015, 51, 13024–13027. doi:10.1039/c5cc04126k
    Return to citation in text: [1] [2] [3] [4] [5]
  26. Liang, X.; Han, H.-B.; Yan, Z.-P.; Liu, L.; Zheng, Y.-X.; Meng, H.; Huang, W. New J. Chem. 2018, 42, 4317–4323. doi:10.1039/c7nj04482h
    Return to citation in text: [1]
  27. Yuan, W.; Yang, H.; Duan, C.; Cao, X.; Zhang, J.; Xu, H.; Sun, N.; Tao, Y.; Huang, W. Chem 2020, 6, 1998–2008. doi:10.1016/j.chempr.2020.04.021
    Return to citation in text: [1] [2]
  28. Ward, J. S.; Kukhta, N. A.; dos Santos, P. L.; Congrave, D. G.; Batsanov, A. S.; Monkman, A. P.; Bryce, M. R. Chem. Mater. 2019, 31, 6684–6695. doi:10.1021/acs.chemmater.9b01184
    Return to citation in text: [1]
  29. dos Santos, P. L.; Ward, J. S.; Congrave, D. G.; Batsanov, A. S.; Eng, J.; Stacey, J. E.; Penfold, T. J.; Monkman, A. P.; Bryce, M. R. Adv. Sci. 2018, 5, 1700989. doi:10.1002/advs.201700989
    Return to citation in text: [1]
  30. Adamo, C.; Barone, V. J. Chem. Phys. 1999, 110, 6158–6170. doi:10.1063/1.478522
    Return to citation in text: [1]
  31. Pople, J. A.; Binkley, J. S.; Seeger, R. Int. J. Quantum Chem. 1976, 10 (Suppl. 10), 1–19. doi:10.1002/qua.560100802
    Return to citation in text: [1]
  32. Grimme, S. Chem. Phys. Lett. 1996, 259, 128–137. doi:10.1016/0009-2614(96)00722-1
    Return to citation in text: [1]
  33. Hirata, S.; Head-Gordon, M. Chem. Phys. Lett. 1999, 314, 291–299. doi:10.1016/s0009-2614(99)01149-5
    Return to citation in text: [1]
  34. Jesser, A.; Rohrmüller, M.; Schmidt, W. G.; Herres-Pawlis, S. J. Comput. Chem. 2014, 35, 1–17. doi:10.1002/jcc.23449
    Return to citation in text: [1]
  35. Hansch, C.; Leo, A.; Taft, R. W. Chem. Rev. 1991, 91, 165–195. doi:10.1021/cr00002a004
    Return to citation in text: [1]
  36. Gao, X.; Bai, S.; Fazzi, D.; Niehaus, T.; Barbatti, M.; Thiel, W. J. Chem. Theory Comput. 2017, 13, 515–524. doi:10.1021/acs.jctc.6b00915
    Return to citation in text: [1]
  37. Cai, M.; Auffray, M.; Zhang, D.; Zhang, Y.; Nagata, R.; Lin, Z.; Tang, X.; Chan, C.-Y.; Lee, Y.-T.; Huang, T.; Song, X.; Tsuchiya, Y.; Adachi, C.; Duan, L. Chem. Eng. J. 2021, 127591. doi:10.1016/j.cej.2020.127591
    Return to citation in text: [1]
Other Beilstein-Institut Open Science Activities