High-Performance Dibenzoheteraborin-Based Thermally Activated Delayed Fluorescence Emitters: Molecular Architectonics for Concurrently Achieving Narrowband Emission and Efficient Triplet–Singlet Spin Conversion

Article abstract of DOI:10.1002/adfm.201802031

Thermally activated delayed fluorescence (TADF) materials, which enable the full harvesting of singlet and triplet excited states for light emission, are expected as the third-generation emitters for organic light-emitting diodes (OLEDs), superseding the conventional fluorescence and phosphorescence materials. High photoluminescence quantum yield (Φ_PL), narrow-band emission (or high color purity), and short delayed fluorescence lifetime are all strongly desired for practical applications. However, to date, no rational design strategy of TADF emitters is established to fulfill these requirements. Here, an epoch-making design strategy is proposed for producing high-performance TADF emitters that concurrently exhibiting high Φ_PL values close to 100%, narrow emission bandwidths, and short emission lifetimes of ≈1 μs, with a fast reverse intersystem crossing rate of over 10⁶ s^?1. A new family of TADF emitters based on dibenzoheteraborins is introduced, which enable both doped and non-doped TADF-OLEDs to achieve markedly high external electroluminescence quantum efficiencies, exceeding 20%, and negligible efficiency roll-offs at a practical high luminance. Systematic photophysical and theoretical investigations and device evaluations for these dibenzoheteraborin-based TADF emitters are reported here.

Full text of DOI:10.1002/adfm.201802031

FULL PAPER
Organic Light-Emitting Diodes
www.afm-journal.de
High-Performance Dibenzoheteraborin-Based Thermally
Activated Delayed Fluorescence Emitters: Molecular
Architectonics for Concurrently Achieving Narrowband
Emission and Efficient Triplet–Singlet Spin Conversion
In Seob Park, Kyohei Matsuo, Naoya Aizawa, and Takuma Yasuda*
field because of its great promise for
Thermally activated delayed fluorescence (TADF) materials, which enable
high internal electroluminescence (EL)
the full harvesting of singlet and triplet excited states for light emission, are
quantum efficiencies (ηint) in organic light-
expected as the third-generation emitters for organic light-emitting diodes
(OLEDs), superseding the conventional fluorescence and phosphorescence
materials. High photoluminescence quantum yield (ΦPL), narrow-band
emission (or high color purity), and short delayed fluorescence lifetime are
all strongly desired for practical applications. However, to date, no rational
design strategy of TADF emitters is established to fulfill these requirements.
Here, an epoch-making design strategy is proposed for producing high-per-
formance TADF emitters that concurrently exhibiting high ΦPL values close to
100%, narrow emission bandwidths, and short emission lifetimes of ≈1 µs,
with a fast reverse intersystem crossing rate of over 106 s−1. A new family of
TADF emitters based on dibenzoheteraborins is introduced, which enable
both doped and non-doped TADF-OLEDs to achieve markedly high external
electroluminescence quantum efficiencies, exceeding 20%, and negligible
efficiency roll-offs at a practical high luminance. Systematic photophysical
and theoretical investigations and device evaluations for these dibenzoheter-
aborin-based TADF emitters are reported here.
emitting diodes (OLEDs) without the use
of precious metals.[1–3] OLEDs based on
metal-free purely organic TADF emitters
can produce emissive bright singlet (S1)
excitons from non-emissive dark triplet
(T1) excitons via spin-converting reverse
intersystem crossing (RISC), leading
to fully fluorescent ηint of up to 100%
(Figure 1). To produce cost-effective
OLEDs, high-efficiency TADF systems are
thus expected to be a favorable alternative
to the current phosphorescent systems,[4]
which employ iridium or platinum com-
plexes. Since the seminal report on OLEDs
using 4CzIPN (1,2,3,5-tetrakis(carbazol-
9-yl)-4,6-dicyanobenzene) as
a
green
TADF emitter by Adachi and co-workers
in 2012,[1] considerable research effort has
been devoted to the development of high-
efficiency red,[5] green,[6] and blue[7] (RGB)-
emitting TADF-OLEDs. Over the last few
1. Introduction
years, high external EL quantum efficiencies (ηext) exceeding
20%, corresponding to ηint of ≈100%, have been achieved in
these devices.
Thermally activated delayed fluorescence (TADF) is a phenom-
enon of growing research interest in the organic electronics
Nevertheless, two major challenges remain to be solved for their
practical applications. One is that narrow-emission TADF mate-
rials are difficult to design. Because commercial OLED displays
utilize color filters to cut off the margin region of EL emission
and to compensate the high color purity defined by broadcasting
standards, the broadening of the EL spectra largely depresses
the actual ηext values. For most TADF emitters reported to date,
the full width at half-maxima (FWHM) of their emission spectra
are as wide as 0.45–0.55 eV (typically, 80–100 nm or even wider).
This is because typical TADF molecules with a small S1–T1 energy
splitting (ΔEST) rely on donor–acceptor (D–A) electronic systems
to generate TADF via intramolecular charge-transfer (ICT) excited
states, which typically causes a broadening and bathochromic shift
of the emission spectra. Recently, using the multiple resonance
effect, Hatakeyama et al. successfully developed pure-blue TADF
emitters that exhibit extremely narrow EL emissions with an
FWHM of 28 nm.[7f] However, this strategy is limited to a specific
molecular framework and inapplicable to common D–A-structured
TADF emitters. Hence, a new universal design principle for
Dr. I. S. Park, Dr. K. Matsuo, Dr. N. Aizawa, Prof. T. Yasuda
INAMORI Frontier Research Center (IFRC)
Kyushu University
744 Motooka, Nishi-ku, Fukuoka 819-0395, Japan
E-mail: yasuda@ifrc.kyushu-u.ac.jp
Dr. I. S. Park, Prof. T. Yasuda
Department of Applied Chemistry
Graduate School of Engineering
Kyushu University
744 Motooka, Nishi-ku, Fukuoka 819-0395, Japan
Dr. N. Aizawa
PRESTO
Japan Science and Technology Agency (JST)
Honcho, 4-1-8 Kawaguchi, Saitama 332-0012, Japan
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/adfm.201802031.
DOI: 10.1002/adfm.201802031
©
Adv. Funct. Mater. 2018, 1802031
1802031 (1 of 12)
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.afm-journal.de
the exciton deactivation processes, not only ΔEST must be mini-
mized, but also the SOC between the S1 and T1 (or upper Tn, n ≥ 2)
states must be significantly enhanced in these TADF systems.
Despite its importance in the RISC process, little consideration
has been given to the magnitude of SOC in the design of TADF
emitters to date. This is because the calculated SOC matrix ele-
ment values for representative TADF molecules consisting of
light main-group elements are all near or less than 1 cm−1,[11–14]
which are much smaller than those for phosphorescent emit-
ters (>100 cm−1) including heavy transition metals.[15] Recently,
internal[16] and external[17] heavy-atom effects derived from heavy
halogen atoms or iridium cores have been explored to shorten
the TADF lifetime. However, it remains unclear whether the
perturbation of non-metallic elements can indeed enhance SOC
and accelerate the RISC process in purely organic TADF mole-
cules, without depressing their intrinsic high EL properties.
Here, we propose an epoch-making approach for the design
of TADF emitters that exhibit narrower ICT emission and
faster RISC. To validate our strategy in this study, we focused
on dibenzo[b,e][1,4]heteraborins as A units, bearing an electron-
deficient boron center and an adjustable bridging heteroatom
(i.e., sulfur,[18a,b] oxygen,[7c,19] or nitrogen[18,20]), and developed
a new family of TADF emitters 1–4 (Figure 2), by combining
these A units with 2,7-dimethyl-9,9-diphenylacridan (MPAc) as
a common D unit. The structural versatility of the heteroatomic
moiety and the D–A linking mode among 1–4 allowed us to sys-
tematically investigate their unique photophysical properties,
geometrical and electronic characteristics, and performance in
OLED devices. While the sulfur atom (ZN = 16), like other light
elements such as oxygen (ZN = 8) and nitrogen (ZN = 7), is not
usually considered as a heavy atom, we revealed that incorpo-
rating a phenothiaborin unit (BS in 1), instead of a phenoxa-
borin (BO in 2) or phenazaborin unit (BN in 3), could largely
enhance SOC, thereby accelerating RISC and shortening the
TADF emission lifetime to ≈1 µs, which is among the shortest
value ever reported for efficient TADF emitters and also compa-
rable to organometallic phosphorescent emitters. Furthermore,
1 and 2 exhibited very favorable TADF properties with high
photoluminescence (PL) quantum yields (ΦPL) of 96–100% and
narrow-band ICT emissions with an FWHM of 0.31–0.36 eV
(or 56–74 nm) in solutions and in solid thin films (even in non-
doped neat films). As a result, both doped and non-doped TADF-
OLEDs based on the BS-containing 1 achieved a high maximum
ηext of 25.3% with negligible EL efficiency roll-off, even at the
practical high luminance of 100 cd m−2 (display relevant) and
1000 cd m−2 (illumination relevant).
Figure 1. Schematic representation of the TADF process in OLEDs. Upon
hole and electron (h–e) recombination, 25% of the generated excited
states are S1 and 75% are T1. With D–A-structured TADF emitters pos-
sessing a very small ΔEST, the T1 excitons are harvested through RISC to
emit TADF.
narrow-band EL emission in simple D–A-structured TADF emit-
ters remains in high demand.
The other key challenge is the serious EL efficiency roll-off
at high current density and rapid degradation of TADF-OLEDs.
This efficiency roll-off stems from exciton deactivation pro-
cesses, such as triplet–triplet annihilation (TTA),[8] that are
primarily caused by the relatively long excited-state lifetime of
over 101–104 µs, generating very active high-energy excited spe-
cies. In general, RISC involves a spin-forbidden, upconversion
process from lower-energy T1 to higher-energy S1 states and is
a rate-determining step in current TADF systems; hence, the
rate of RISC (kRISC) is the most prominent factor governing the
degree of efficiency roll-off in TADF-OLEDs.[9] According to first-
order perturbation theory, namely Fermi’s golden rule, kRISC is
described as[10,11]
2
2π
ꢀ
ˆ
kRISC
=
ρFC S1|HSOC|T1
(1)
ˆ
where 〈S1|HSOC|T1 〉 is the spin–orbit coupling (SOC) matrix
element between the S1 and T1 states with different spin mul-
tiplicities, and ρFC denotes the Frank–Condon-weighted density
of states. In the high-temperature regime (e.g., at room temper-
ature), ρFC can be given by the following semiclassical Marcus
theory expression[10,11]
2. Results and Discussion
2.1. Design Strategy and Computational Simulations

2 

∆E + λ
4λkBT
(
)
1
ST
ρFC
=
exp −
(2)

We first started by investigating the geometric structures and
electronic transition characteristics of the newly designed D–A
molecules 1–4, as illustrated in Figure 2. Using density func-
tional theory (DFT), the ground-state (S0) geometries of 1–4
were initially optimized at the PBE0/6-31G(d) level in the gas
phase, and then the configurations and energies in the excited
4πλkBT




where λ is the Marcus reorganization energy associated with
the RISC process, kB is the Boltzmann constant, and T is tem-
perature. Accordingly, to attain a large kRISC value and suppress
©
Adv. Funct. Mater. 2018, 1802031
1802031 (2 of 12)
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.afm-journal.de
Figure 2. Molecular structures, frontier orbital distributions, and optimized geometries at So, S1, and T1 states, with their respective energy levels, for
TADF emitters 1–4, calculated at the PBE0/6-31G(d) level. H → L represents the HOMO to LUMO transition.
S1 and T1 states and associated oscillator strengths ( f ) were
computed using time-dependent DFT (TDDFT) at the same
level. As designed, 1–3 adopt nearly orthogonal D–A confor-
mations between the MPAc and their respective dibenzoheter-
aborin units, with large torsion angles (θ) of 76°–89° in their
optimized S0 structures. Consequently, the highest occupied
molecular orbitals (HOMOs) are primarily localized on the
electron-donating MPAc moiety (except for the non-conjugated
peripheral diphenyl substituents), whereas the lowest unoccu-
pied molecular orbitals (LUMOs) are entirely localized in the
vicinity of the electron-accepting dibenzoheteraborin moiety.
Because the lowest-excited S1 and T1 states for these mole-
cules are dominated by the HOMO → LUMO ICT transitions
abatic ΔEST values of 12–35 meV,[21] enabling efficient RISC.
Note that among 1–3, the calculated S1 energies decreased in the
order 3 (2.737 eV) > 2 (2.544 eV) > 1 (2.478 eV), which can be
attributed to the lowering of the LUMO energy level (or incre-
mental changes in electron-accepting ability). The same trend
was also found in the T1 energy arrangement for 1–3 along with
the variation in the bridging heteroatoms. Unlike 1–3, if a locally
excited triplet state (3LE) dominated by π–π* transition(s) on
each D or A constituent lies in a lower energy state than the 3CT
state, the T1 state is represented by LE instead of by CT, even
in the D–A molecules, resulting in enlarged ΔEST. For 4, as
an analog of 1, a similar orthogonal connectivity appears
between the MPAc unit (or BS unit at the other side in the
framework) and the adjacent central 2,6-diisopropylphenylene
(Dip) π-linker (θ1 = 83° and θ2 = 90°, respectively), leading to
3
3
1
3
(denoted as CT and CT, respectively), this clear spatial sepa-
ration of the frontier molecular orbitals leads to very small adi-
©
Adv. Funct. Mater. 2018, 1802031
1802031 (3 of 12)
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.afm-journal.de
λS value for a representative high-efficiency
TADF emitter (5)[7b] consisting of 9,3′:6,9″-
tercarbazole and 2,4,6-triphenyl-1,3,5-triazine
units (Figure 3c). The images comparing its
S0 and S1 structures manifested a large D–A
internal twisting, with increasing θ from 52°
to 88° and elongating l by 0.04 Å, thereby
resulting in a larger calculated λS of 0.32 eV.
2.2. Synthesis and Characterization
Syntheses of 1–4 were achieved through
palladium-catalyzed
Buchwald–Hartwig
reactions between the corresponding mono-
brominated dibenzoheteraborin derivatives
and MPAc. Detailed synthesis procedures
and characterization data are described in
the Experimental Section and Supporting
Information. To sterically protect the tri-
coordinate boron center from attack by
nucleophiles,
a
bulky 2,4,6-triisopropyl-
phenyl (Tip) group (for 1–3) or Dip linker
(for 4) was introduced in each dibenzoheter-
aborin unit. This design feature rendered
1–4 highly tolerant to air and moisture. All
these materials were purified using column
chromatography, and their robust structural
Figure 3. a) Potential energy surfaces of the S0, S1, and T1 states of TADF molecules and character allowed for further purification by
b) schematic representation of the excitation and emission spectra in the 1CT process.
means of temperature-gradient vacuum sub-
limation to obtain high-purity materials for
the device evaluations.
c) Molecular structure and optimized S0 and S1 geometries of a representative triazine-based
TADF emitter (5)[7b] with a large reorganization energy (λS), calculated by TDDFT.
X-ray crystallographic analyses for
1
more restricted HOMO–LUMO overlap and a much smaller cal-
culated ΔEST of 2 meV. Conversely, this electronic feature of 4
causes a smaller oscillator strength ( f ) of 0.0002 for the S1 exci-
tation in comparison with 1–3 ( f = 0.0023–0.0120).
and 4 further elucidated the highly twisted molecular struc-
tures in their solid states. As can be seen from Figure 4,
the BS frameworks in 1 and 4 are highly planar structures, and
the Tip group and Dip linker are connected orthogonally at the
B1 atoms. The geometries around the B1 atoms in 1 and 4 are
almost planar (∑(C–B1–C)3 = 360°). On the other hand, the
MPAc moiety in both molecules tends to have a slightly bent
shape along the central N1–C42 or N1–C39 axis, and each N1
atom adopts a slightly pyramidal geometry (∑(C–N1–C)3 = 353.5°
for 1 and 356.7° for 4), presumably because of the sp3 character
of the C42 or C39 atom and the mutual steric hindrance arising
As mentioned above, with a few exceptions,[7c,f,k,22] most
reported TADF emitters in which the S1 state is represented by
1CT exhibit broad Gaussian-shaped PL and EL spectra, resulting
from the large structural (vibrational) relaxation in the ICT tran-
sition. As schematically represented in Figure 3a, reorganization
energies, λS and λS*, associated with the emission and excita-
tion processes should have a crucial impact on the resulting
spectral shape and FWHM. In principle, the sum of λS and λS*
corresponds to the Stokes shift (Figure 3b), and hence, TADF
emitters with smaller λS will exhibit narrower ICT emission.
Based on the TDDFT calculations, rather small λS values of
0.20–0.27 eV were estimated for 1–4 (see Figure 2), which can
be rationalized by the small degree of structural deformation
between the optimized S0 and S1 structures, with a minimal
variation in θ and the relevant bond length (l). We expect that
large steric repulsion between the peri-hydrogen atoms in the D
and A units effectively restricts such structural deformation. In
addition, the significant contribution from the vacant p orbital
of the boron center to the LUMO also plays an important role
in minimizing the change in the relevant bond lengths and low-
ering λS upon the S1 excitation. To prove our design concept,
we also calculated the S0 and S1 geometries with the associated
Figure 4. Crystal structures of a) 1 and b) 4 (CCDC 1825382 and 1825383,
respectively) with thermal ellipsoids at 50% probability. Hydrogen atoms
are omitted for clarity. Selected bond length (Å): for 1, B1–C6, 1.545(9);
B1–C13, 1.58(1); N1–C9, 1.436(8); for 4, B1–C1, 1.552(5); B1–C13,
1.585(5); N1–C16, 1.440(4).
©
Adv. Funct. Mater. 2018, 1802031
1802031 (4 of 12)
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.afm-journal.de
from the peripheral diphenyl substituents in the MPAc unit.
These multiple bulky substituents in both the D and A units may
contribute to suppressed concentration quenching[6g] and exciton
annihilation[9] in the condensed states. Overall, the highly twisted
D–A structures of the single crystals in 1 and 4 are basically con-
sistent with the optimized molecular structures obtained by the
DFT calculations. For 1, the torsion angles between the BS and
MPAc units (θ) are 87.1° (for C10–C9–N1–C35) and 64.1° (for
C10–C9–N1–C28); whereas for 4, those between the Dip and
MPAc units (θ1) are 62.3° (for C17–C16–N1–C32) and 96.7° (for
C17–C16–N1–C25).
2.3. Steady-State Photophysical Properties
Figure 5. a) UV–vis and b) PL spectra of 1–4 in toluene (10−5 m). The
inset of (a) shows a magnified view of the lower-energy ICT absorption
bands. The inset of (b) shows the curves (dashed lines) fitted according
to the Marcus theory.
The UV–vis absorption and PL spectra of 1–4 in dilute toluene
solutions are depicted in Figure 5, and their detailed photo-
physical data are compiled in Table 1. The intense higher-energy
absorption bands below 400 nm, shown in Figure 5a are mainly
attributed to the locally excited (1LE) absorptions associated with
the π–π* transitions of the respective dibenzoheteraborin moie-
ties. In addition, 1–3 showed weak and broad shoulder bands
in a lower-energy region (400–460 nm), with molar absorption
ICT emissions upon photoexcitation (Figure 5b), with the PL
maxima (λPL) ranging from 449 nm (deep blue) to 493 nm
(green). In comparison with the BN-based 3, the PL emissions
of the BO-based 2 and BS-based 1 in toluene were bathochro-
mically shifted by 28 and 44 nm, respectively, agreeing with the
TDDFT results.
coefficients (ε) on the order of 103 −1 cm−1, which can be attrib-
m
uted to the ICT (1CT) absorption originating from their char-
acteristic D–A electronic systems. For 4, the fully electronically
separated connection between the D and A units through the
Dip π-linker inevitably produces a forbidden HOMO → LUMO
transition with f ≈ 0. Nevertheless, a corresponding ICT absorp-
The FWHM for the solution PL spectra of 1–4 were indeed
significantly smaller (EFWHM = 0.32–0.37 eV and λFWHM
=
56–74 nm; Table 1) than those of typical blue TADF emitters
such as 5[7b] (EFWHM = 0.45 eV). In the framework of the Marcus
theory, spectral line shapes for the ICT emission can be fitted
using the following formula[23]
tion with a very small ε of ≈102 −1 cm−1 was detected at around
m
440 nm. Depending on the different dibenzoheteraborin units
in 1–4, their deoxygenated toluene solutions exhibited tunable
Table 1. Photophysical data of dibenzoheteraborin-based TADF emitters 1–4.
f)
b,j)
b,k)
b,l)
d)
e)
f)
g)
h)
h)
b,m)
Emitter
EFWHM
[eV]
krb,i)
[s−1
kISC
[s−1
kRISC
[s−1
knr,T
λPL
ΦPL
λFWHM
[nm]
λS
τp
τd
ΔEST
]
]
]
[s−1
]
[nm]
493
497
481
477
483
466
449
465
448
487
485
471
[%]
[eV]
0.40
0.38
[ns]
5.9
7.9
7.5
23
[µs]
1.0
1.3
1.7
2.0
1.8
2.4
35
[meV]
1
Sola)
Dopeb)
Neatc)
Sola)
96
0.35
0.36
0.34
0.32
0.35
0.31
0.34
0.35
0.37
0.37
0.39
0.37
70
74
65
60
68
56
56
62
63
74
76
68
2.8 × 107
1.9 × 107
2.4 × 107
9.5 × 106
9.9 × 107
1.7 × 107
9.1 × 107
3.8 × 107
3.5 × 106
1.0 × 106
1.8 × 105
2.3 × 105
≈0
100
99
23
n)
–
2
3
4
100
99
0.34
0.33
Dopeb)
Neatc)
Sola)
28
24
50
30
1.2 × 104
1.8 × 104
3.6 × 104
n)
98
–
27
50
0.40
0.36
5.6
8.7
8.6
47
Dopeb)
Neatc)
Sola)
75
18
n)
32
–
16
91
0.46
0.49
19
Dopeb)
Neatc)
65
21
12
n)
21
–
21
2.0
a)Measured in oxygen-free toluene solution (10−5 m) at 300 K; b)Measured as a 50 wt% doped thin film in a PPF host matrix at 300 K; c)Measured as a pristine neat film at
300 K; d)PL emission maximum; e)Absolute PL quantum yield evaluated using an integrating sphere under N2; f)Full width at half-maximum of the PL spectrum given in
energy or wavelength; g)Reorganization energy along the S1 → S0 ICT emission estimated from the PL spectrum; h)PL lifetimes of prompt fluorescence (τp) and delayed flu-
orescence (τd); i)Rate constant of fluorescence radiative decay (S1 → S0): kr = Φp/τp; j)Rate constant of ISC (S1 → T1): kISC = (1 − Φp)/τp; k)Rate constant of RISC (T1 → S1):
kRISC = Φd/(kISC ∙τp ∙τd ∙Φp); l)Rate constant of non-radiative decay in the triplet states; m)Singlet–triplet energy splitting determined from the Arrhenius plots for kRISC
(cf. Figure 6d); n)Not determined.
©
Adv. Funct. Mater. 2018, 1802031
1802031 (5 of 12)
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.afm-journal.de
(PPF)[6g,7i,j,9,20] was employed as a suitable
high-T1 host to prevent the reverse excited-
energy transfer from the T1 states of the
dopant emitters and confine the excitons
within the emitters. As depicted in Figure 6a,b,
the heavily doped thin films of 1–4 at
50 wt% dopant concentrations exhibited blue
to green PL emissions (λPL = 465–497 nm),
the absolute PL quantum yields (ΦPL) of
which reached 100%, 99%, 75%, and 65%,
respectively, under N2 atmosphere. Intrigu-
ingly, with the non-doped neat films, the λPL
positions were blue-shifted by 14–17 nm (see
Table 1 and the Supporting Information),
with respect to those obtained with the PPF-
hosted doped films. This phenomenon can
be explained by the different polarities of the
solid host media. Compared with the highly
polar PPF, which possesses a large dipole
moment (µ = 5.8 D, estimated by DFT calcu-
lations), the less polar nature of 1–4 (µ = 1.3,
1.4, 3.4, and 2.4 D, respectively) as host media
1
could offer a lower degree of CT state stabi-
lization via solid-state solvation,[24] resulting
in such unusual blue-shifted PL emissions in
the neat films. Furthermore, narrow PL emis-
sion bandwidths (EFWHM = 0.31–0.37 eV and
λFWHM = 56–68 nm) were maintained in the
neat films.
It is noteworthy that 1 and 2 underwent
no obvious concentration quenching; ΦPL did
not decrease at all, retaining values as high
as 99% for 1 and 98% for 2 even in the host-
free neat films. This negligible concentration
quenching behavior is primarily attributable
to their substantially shorter excited-state
lifetimes within ≈10 µs (Figure 6c), which
suppresses the diffusion and quenching of
the triplet excitons through intermolecular
electron-exchange interactions (i.e., Dexter
energy transfer).[6g,25] A few TADF emitters
that are free of concentration quenching have
recently been reported by us[6g,7c,9,26] and other
Figure 6. a) Steady-state PL spectra, b) fractional quantum yields for prompt and delayed
fluorescence (Φp and Φd), and c) transient PL decay profiles of 1–4 in 50 wt% emitter:PPF-
doped films measured at 300 K under N2. The inset of (a) shows the curves (dashed lines)
fitted according to the Marcus theory. d) Arrhenius plots of the rate constants of RISC (kRISC
)
obtained with the doped films of 1–4, where the solid lines denote the least-squares fittings.

2 

E
− λ −E
4λSkBT
groups.[6d,7h,27] In contrast, the ΦPL values of 3 and 4 with much
longer excited-state lifetimes sharply decreased to 32% and 21%
in the neat films, demonstrating general aggregation-caused
quenching (ACQ) behavior, similar to most TADF emitters.
(
exp −

)
E
CT
S
IPL
∝
(3)
4πλSkBT




where IPL is the PL intensity as a function of photon energy E,
and ECT is the energy difference between the S0 and S1 states
(cf. Figure 3a,b). The fitted curves matched well with the experi-
mentally obtained PL spectra (Figure 5b, inset). The λS values
as small as 0.34–0.40 eV were thus deduced for 1–3 in solution,
whereas the experimental λS values were ≈0.1–0.2 eV larger
than those estimated from the theoretical calculations because
of solvation effects.
To factually assess the performance of 1–4 as TADF emit-
ters, their solid-state photophysical properties were investi-
gated in doped thin films as well as non-doped neat films. In
the doped films, 2,8-bis(diphenylphosphoryl)dibenzo[b,d]furan
2.4. TADF Characteristics and Exciton Dynamics
To further elucidate the TADF behavior of 1–4, the transient PL
characteristics of their doped films were examined. As shown
in Figure 6c, each transient PL curve demonstrated two distinct
components: a nanosecond-order prompt decay and a micro-
second-order delayed decay at 300 K, which could be fitted using
the biexponential model IPL = A1 exp(−t/τp) + A2 exp(−t/τd),
where A1 and A2 are frequency factors, and τp and τd are the life-
times of the prompt and delayed decay components, respectively.
©
Adv. Funct. Mater. 2018, 1802031
1802031 (6 of 12)
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.afm-journal.de
The extracted τp and τd values are included in Table 1. Their frac-
tional quantum yields for the prompt and delayed fluorescence
(Φp and Φd, respectively) were then determined from the overall
ΦPL value and the proportions of the integrated PL intensities
of these two components (Figure 6b). With increasing tempera-
ture from 5 to 300 K, the delayed fluorescence components for
1–4 were greatly intensified (see the Supporting Information),
revealing their typical TADF features. Remarkably, the τd of
1 and 2 in the doped films (1.3 and 1.8 µs) were 1 order of mag-
nitude shorter than those of 3 and 4 (18 and 12 µs, respectively),
which was expected to originate from the faster RISC in 1 and 2.
We further analyzed the exciton dynamics of the S1 and T1
states in 1–4 on the basis of their photophysical transition rates.
As listed in Table 1, despite only subtle variations in the bridging
heteroatom, the RISC of the thia-bridged 1 was found to be dras-
tically enhanced, with an increased kRISC of 3.5 × 106 s−1, which
is 3.5- and 19-fold those of the oxa-bridged 2 and aza-bridged
3, respectively. To the best of our knowledge, the kRISC value
obtained with 1, far surpassing 106 s−1, is among the highest to
have been reported for TADF emitters. According to the semi-
classical TADF model, kRISC is given as kRISC = A exp(−ΔEST/kBT),
where A is the frequency factor involving the SOC constant.
Therefore, so far, a high kRISC has been achieved by decreasing
ΔEST in common TADF systems. However, from the Arrhenius
plots of kRISC of 1–4 (Figure 6d), almost the same small activation
energies (corresponding to the actual ΔEST) were experimentally
obtained for 1 and 2 (ΔEST = 23 and 24 meV, respectively). These
results unambiguously indicate that kRISC is not dominated only
by ΔEST and can be largely enhanced by replacing the bridging
heteroatom from second-period oxygen to third-period sulfur,
without lowering ΔEST. Furthermore, the experimentally deter-
mined ΔEST values for both 1 and 2 were indeed smaller than
the thermal energy at 300 K (kBT ≈ 25.9 meV ≈ 209 cm−1), mani-
festing efficient RISC assisted by ambient thermal energy.
Figure 7. Schematic diagram for plausible RISC and TADF mechanism
involving the SOC-induced spin conversion in BS-based 1. The energy levels
for CT and LE were estimated from the onsets of the low-temperature
3
3
phosphorescence spectra of 1 and its precursor Br-BS, respectively.
computed SOC matrix elements between the T1 and S1 states of
ˆ
〈S |HSOC|T1 〉
1
1–4 were all very tiny (
≤ 0.1 cm−1), supporting the
foregoing prediction. The T2 states of 1–4, which had LE charac-
ters, were found to afford much larger SOCs with the S1 states
than their T1 counterparts. Moreover, the T2 state of 1 exhib-
ˆ
ited a significantly large SOC with its S1 state (〈S1|HSOC|T2 〉=
4.67 cm−1), which was more than 10 times larger than those of
ˆ
2 and 3 (
= 0.41 and 0.38 cm−1, respectively). There-
〈S1|HSOC|T2 〉
fore, the thia-bridged 1 is expected to undergo RISC more
effectively than oxa-bridged 2 and aza-bridged 3 through the
interstate coupling with 3LE because the SOC in the former
is greatly enhanced by the internal heavy-atom effect from the
bridging sulfur atom. Unlike 1, BS-based 4 was found to afford
a much smaller SOC, with 〈S1|HSOC|T2 〉 of 0.03 cm−1, in spite of
ˆ
similarity in the electronic configuration of the S1 and T2 states
(being dominated by 1CT and 3LE, respectively). This result may
arise from the coplanar geometries between the MPAc and BS
units connected through the twisted Dip π-linker, where the
3LE–1CT transition may not be accompanied by the clear orbital
angular momentum change required for spin flip. For the pre-
sent dibenzoheteraborin-based TADF molecules, including 1,
the calculated SOC constants were still ≈2 orders of magnitude
smaller than the values estimated for phosphorescent emitters
containing much heavier transition metals.[15] However, we
found that even a slight increase in the SOC strength induced
by a non-metallic “moderate” heavy atom can greatly enhance
the overall rate of RISC, and hence shorten the resulting delayed
fluorescence lifetime to ≈1 µs.
Here, our primary interest lies in what happens in the RISC
process of the thia-bridged 1 with a high kRISC, that is, why its
T1 excitons convert so efficiently into S1 excitons. The com-
monly accepted scheme for the TADF process is a cyclic direct
spin conversion between the energetically close 1CT and 3CT
states via ISC and RISC (see Figure 1).[1–3] However, in view of
the large kRISC value of 1 (>106 s−1), the S1 state is unlikely to
harvest the population of the T1 state directly via RISC because
3
1
the SOC between their CT and CT states is formally vanished
(i.e., 〈S1|HSOC|T1 〉 ≈ 0).[28] Therefore, we envision that other elec-
ˆ
tronic transition state(s) are involved in the triplet–singlet spin
conversion process to accelerate RISC. Recent theoretical and
spectroscopic studies by several groups[12–14] have proposed an
alternative mechanism of RISC mediated by mixing the lowest
1CT state with an energetically close-lying 3LE state (residing on
the dibenzoheteraborin fragment in our TADF molecules 1–4).
In this mechanism, as illustrated in Figure 7, a weakly spin-
2.5. OLED Device Performance
The performances of 1–4 as TADF emitters were tested in
OLEDs. As depicted in Figure 8a, we first fabricated multilayer
TADF-OLEDs (devices A–D) employing 50 wt% 1–4:PPF-doped
films as the emission layer (EML) with the following struc-
ture: indium tin oxide (ITO, 50 nm)/HAT-CN (10 nm)/TAPC
(50 nm)/CCP (10 nm)/EML (30 nm)/PPF (10 nm)/B3PyPB
(40 nm)/Liq (1 nm)/Al (100 nm). In this device architecture,
3
1
allowed transition from LE to CT involving an orbital angular
momentum change is facilitated via non-adiabatic vibronic
coupling between the 3CT and 3LE states, 〈T2|HVIB|T1 〉,
ˆ
which is considered to be much larger than SOC in TADF
molecules.[12] To gain more insight into the triplet–singlet spin
conversion processes involved in 1–4, the effective SOC matrix
elements between the T1 (or T2) and S1 states were calculated
using TDDFT (see the Supporting Information for details). The
HAT-CN
(2,3,6,7,10,11-hexacyano-1,4,5,8,9,12-hexaazatriphe-
nylene) and TAPC (1,1-bis(4-di-p-tolylaminophenyl)cyclohexane)
served as the hole-injection layer and the hole-transporting layer
©
Adv. Funct. Mater. 2018, 1802031
1802031 (7 of 12)
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.afm-journal.de
Figure 8. a) Energy level diagram and chemical structures of the materials for the doped TADF-OLEDs (devices A–D) based on 1–4 as emitters. b) EL
spectra measured at 10 mA cm−2 and photos of EL emissions from the devices. c) Current density–voltage–luminance (J–V–L) characteristics and
d) external EL quantum efficiency (ηext) versus L plots of devices A–D. The inset of c) shows the angular dependence of the EL intensities with the
Lambertian distributions.
(HTL), whereas B3PyPB[29] (1,3-bis[3,5-di(pyridine-3-yl)phenyl]
benzene) and Liq (8-hydroxyquinoline lithium) were used as the
electron-transporting layer (ETL) and electron-injection mate-
rial, respectively. In addition, thin layers (10 nm thicknesses) of
CCP[7c] (9-phenyl-3,9′-bicarbazole) and PPF with high T1 ener-
gies (3.0 and 3.1 eV, respectively) were inserted at the HTL/EML
and EML/ETL interfaces as exciton-blocking layers to prevent
unfavorable triplet exciton quenching and thereby confine all
electrogenerated excitons within the EML.[7i,j,9]
C and D compared with devices A and B is attributed to the substan-
tially lower ΦPL values of 3 and 4. Most remarkably, ultra-low effi-
ciency roll-off was indeed accomplished for device A (Figure 8d).
When driven at practical high luminance of 100 and 1000 cd m−2,
most reported TADF-OLEDs have exhibited serious efficiency
roll-offs, similar to those exhibited by devices B–D. However,
device A was capable of retaining very high EL efficiencies
with ηext of 25.0%, 23.7%, and 20.5% at 100, 1000, and even at
5000 cd m−2, respectively. These ηext values achieved with device A
at this practical luminance level surpass those of reported TADF-
OLEDs, confirming that BS-based 1 outperforms as a state-of-the-
art TADF emitter. Device B showed a similarly high maximum
ηext reaching 25%, but a larger efficiency roll-off at high lumi-
nance. Given the comparable ΦPL and ΔEST values of 1 and 2, the
accelerated RISC process for 1, with the larger kRISC (3.5 × 106 s−1),
is thus considered to play the dominant role in alleviating the
efficiency roll-off of the device. In contrast, with 3 and 4, the one
order smaller kRISC (≈105 s−1) led to the more severe efficiency
roll-offs of these devices. In devices C and D, the few upconverted
T1 excitons of 3 and 4 with the longer excited-state lifetimes con-
tributed much less to TADF emission and rather caused exciton
deactivation via bimolecular processes such as TTA, lowering the
overall EL efficiencies in high luminance regimes.
The EL spectra, current density–voltage–luminance (J–V–L)
characteristics, and ηext versus L plots of the fabricated doped
devices A–D are depicted in Figure 8b–d, respectively, and their
key EL parameters are listed in Table 2. Devices A–D turned on
at low voltages of 2.8–3.2 V and emitted stable blue to green EL
with emission peaks (λEL) in the range of 475–503 nm. Their
EL spectra coincided with the corresponding PL spectra of the
doped films, supporting the supposition that the EL emissions
1
were generated solely from the lowest CT states of 1–4. Among
the four devices with the same device architecture, the 1-based
device A exhibited the highest EL performance with a maximum
η
ext of 25.3%, a current efficiency (ηc) of 72.1 cd A−1, and a power
efficiency (ηp) of 77.4 lm W−1, without any light out-coupling
enhancement. The angle-resolved EL intensities measured for
devices A–D were basically Lambertian (Figure 8c). Comparing
the EL performance of these doped devices, the maximum ηext
values were found in the order of device A (25.3%) ≈ B (24.9%) >
D (16.4%) ≈ C (16.0%). The reason for the lower ηext of devices
As mentioned earlier, 1 and 2 were free from the concentra-
tion quenching (or ACQ) and required no host for device fab-
rication. Thus, we fabricated non-doped TADF-OLEDs (devices
E and F) by employing the neat films of 1 and 2 as the EML,
©
Adv. Funct. Mater. 2018, 1802031
1802031 (8 of 12)
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.afm-journal.de
Table 2. EL performance of the TADF-OLEDs based on 1–4.
Device
EMLa)
1:PPF
2:PPF
3:PPF
4:PPF
1
Vonb) [V]
2.8
EFWHMd) [eV]
0.34
CIEi) [x,y]
λELc) [nm]
503
λFWHMd) [nm]
ηext,maxe) [%]
25.3
ηext,100/1000f) [%]
25.0/23.7
22.8/14.7
7.6/–j)
ηcg) [cd A−1
72.1
]
ηph) [lm W−1
77.4
]
A
B
C
D
E
71
67
68
78
63
59
(0.20, 0.51)
(0.16, 0.38)
(0.14, 0.23)
(0.19, 0.45)
(0.15, 0.36)
(0.14, 0.23)
3.0
489
0.34
24.9
57.0
55.6
3.2
475
0.36
16.0
23.7
23.3
3.0
500
0.38
16.4
12.7/2.3
42.0
40.7
3.4
487
0.32
22.8
22.6/19.0
18.3/8.0
49.9
41.4
F
2
3.5
474
0.32
21.3
33.6
29.3
a)Emission layer consisting of a codeposited 50 wt% emitter:PPF-doped film (for devices A–D) or a non-doped neat film (for devices E and F); b)Turn-on voltage at a lumi-
nance over 1 cd m−2 c)EL emission maximum at 10 mA cm−2 d)Full width at half-maximum of the EL spectrum given in energy or wavelength; e)Maximum external EL
;
;
quantum efficiency; f)External EL quantum efficiency at the luminance of 100 and 1000 cd m−2 g)Maximum current efficiency; h)Maximum power efficiency; i)Commission
;
Internationale de l’Éclairage (CIE) chromaticity coordinates recorded at 10 mA cm−2
;
j)Not determined.
with the same architecture of ITO (50 nm)/HAT-CN (10 nm)/
TAPC (50 nm)/CCP (10 nm)/1 or 2 (30 nm)/PPF (10 nm)/
B3PyPB (40 nm)/Liq (1 nm)/Al (100 nm). The detailed EL
characteristics are presented in Figure 9 (see also Table 2).
The non-doped devices E and F exhibited bluer EL emissions
(λEL = 487 and 474 nm, respectively) and narrower bandwidths
(EFWHM = 0.32 eV and λFWHM = 59–63 nm) compared to the cor-
responding doped devices. Because of the suitable energy levels
of 1 and 2 (with HOMOs and LUMOs of −5.6 to −5.7 eV and
−2.8 to –2.9 eV), the hole- and electron-injection barriers at the
both the CCP/EML and PPF/EML junctions appeared to be
minimized. These well-matched energy levels are preferable for
balanced charge transport and efficient recombination within
host-free neat EMLs, leading to even low turn-on voltages
and high EL efficiencies for these devices. Considerably high
ηext values of up to 22.8% and 21.3% were attained with the
non-doped devices E and F, respectively; these values were
only slightly lower than those obtained for the optimal doped
devices. More importantly, when the luminance increased to
100 and 1000 cd m−2, the 1-based device E likewise retained ηext
as high as 22.6% and 19.0%, respectively (Figure 9c), demon-
strating greatly advanced efficiency stability for non-doped
blue TADF-OLEDs. In contrast, the 2-based device F showed
a relatively large efficiency roll-off, and its ηext values declined
more seriously as luminance increased. The most prominent
difference between 1 and other TADF emitters, including 2–4,
is its much faster RISC due to enhanced SOC. This feature is
vital for suppressing exciton deactivation processes, particularly
at high luminance, and for achieving extremely low efficiency
roll-off in actual OLED devices featuring TADF.
3. Conclusions
In this study, we successfully developed a new family of
high-performance TADF emitters based on dibenzo[b,e][1,4]heter-
aborins, typified by phenothiaborin, phenoxaborin, and phenaza-
borin, with an electronically adjustable bridging heteroatom. We
systematically investigated their intrinsic ICT characteristics and
TADF properties by steady-state and time-resolved photophysical
measurements together with quantum chemical simulations.
Lowering the reorganization energy in the emission process by
restricting the structural deformation between the ground and
excited states was demonstrated to produce TADF molecules
that exhibited narrow-band EL emissions, with FWHM as small
as 0.32 eV (≈60 nm), in actual devices. We also demonstrated
that TADF lifetimes can be shortened to ≈1 µs via the moderate
heavy-atom effect of the incorporated sulfur atom in the pheno-
thiaborin-based emitter 1, without reducing the intrinsically high
ΦPL values. Thus, the key RISC process in TADF molecules can
be drastically accelerated via enhanced SOCs by simple chemical
modulations in the integrated heteroatomic moiety. The
Figure 9. EL characteristics of the non-doped TADF-OLEDs (devices E and F) based on 1 and 2: a) EL spectra measured at 10 mA cm−2 and photos of
blue EL emissions, b) J–V–L characteristics, and c) ηext versus L plots of devices E and F.
©
Adv. Funct. Mater. 2018, 1802031
1802031 (9 of 12)
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.afm-journal.de
126.30, 124.96, 124.84, 124.66, 124.49, 120.07, 115.54, 57.12, 35.44,
34.27, 24.30, 24.25, 24.19, 20.88. MS (MALDI-TOF): m/z calcd 757.39
[M]+; found 757.58. Anal. calcd (%) for C54H52BNS: C 85.58, H 6.92, N
1.85; found: C 85.73, H 6.84, N 1.87.
TADF-OLEDs based on 1 achieved excellent EL efficiencies,
with a maximum ηext of 25.3%, ηc of 72.1 cd A−1, and ηp of
77.4 lm W−1. While most reported TADF-OLEDs continue to
suffer from severe efficiency roll-offs, both the doped and non-
doped TADF-OLEDs based on 1 retained high ηext over 20%
even at practical high luminance, demonstrating extremely low
efficiency roll-off behavior. This work thus makes critical contri-
butions to understanding and controlling the underlying mecha-
nism of RISC and TADF. Our present results provide a new
strategy to alleviate efficiency roll-off and degradation in high-
efficiency TADF-OLEDs by further exploring such strategically
designed TADF emitters involving the heavy-atom effect.
MPAc-BO (2): Compound 2 was synthesized according to the same
procedure described above for the synthesis of 1, except that 7b (1.00 g,
2.17 mmol) and 6 (0.82 g, 2.28 mmol) were used as the reactants,
yielding 2 as a light-yellow solid (yield = 1.47 g, 95%). 1H NMR (400 MHz,
CDCl3): δ 7.78 (td, J = 7.2, 2.4 Hz, 2H), 7.71 (td, J = 7.6, 1.8 Hz, 1H), 7.54
(d, J = 8.0 Hz, 1H), 7.34 (d, J = 2.0 Hz, 1H), 7.26-7.18 (m, 7H), 7.06 (s,
2H), 6.98 (dd, J = 7.8, 1.4 Hz 4H), 6.90 (dd, J = 8.0, 1.6 Hz, 2H), 6.76 (dd,
J = 8.0, 2.0 Hz, 1H), 6.69 (d, J = 1.6 Hz, 2H), 6.57 (d, J = 8.4 Hz, 2H), 2.98
(sep., 1H), 2.34 (sep., 2H), 2.21 (s, 6H), 1.34 (d, J = 6.8 Hz, 6H), 1.03
(dd, J = 6.6, 1.4 Hz, 12H). 13C NMR (100 MHz, CDCl3): δ 159.97, 158.87,
150.24, 148.49, 147.35, 146.10, 139.90, 138.69, 136.83, 134.49, 131.80,
130.43, 130.31, 129.98, 127.58, 127.41, 126.27, 123.01, 122.51, 119.92,
117.46, 117.36, 115.50, 57.13, 35.32, 34.30, 24.39, 24.34, 24.16, 20.89. MS
(MALDI-TOF): m/z calcd 741.41 [M]+; found 740.80. Anal. calcd (%) for
C54H52BNO: C 87.43, H 7.07, N 1.89; found: C 87.44, H 7.06, N 1.93.
4. Experimental Section
Synthesis: The synthetic routes used to prepare TADF emitters 1–4
are outlined in Scheme 1. All reactions were performed under N2
atmosphere. The detailed synthesis procedures and characterization data
for the intermediates are given in the Supporting Information. The final
products were purified using temperature-gradient vacuum sublimation
with a P-100 system (ALS Technology), before the measurements and
device fabrication.
MPAc-BS (1): A mixture of 7a (1.00 g, 2.10 mmol), 6 (0.80 g,
2.21 mmol), Pd(OAc)2 (0.011 g, 0.05 mmol), P(t-Bu)3H∙BF4 (0.015 g,
0.01 mmol), and sodium tert-butoxide (0.40 g, 4.20 mmol) in dry
toluene (30 mL) was refluxed for 4 h under N2. After cooling to room
temperature, the reaction mixture was filtered through a Celite pad,
and then filtrate was concentrated under reduced pressure. The crude
product was purified by column chromatography on silica gel (eluent:
hexane/chloroform = 4:1, v/v) to afford 1 as a light yellow solid (yield =
1.51 g, 95%). 1H NMR (400 MHz, CDCl3): δ 7.93 (d, J = 8.4 Hz, 2H), 7.72
(d, J = 8.0 Hz, 1H), 7.60 (td, J = 7.6, 1.5 Hz, 1H), 7.39 (d, J = 1.6 Hz, 1H),
7.31 (td, J = 7.4, 1.2 Hz 1H), 7.24-7.18 (m, 6H), 7.07 (s, 2H), 6.97 (dd,
J = 8.0, 1.6 Hz 4H), 6.92 (dd, J = 8.2, 1.8 Hz 1H) 6.89 (dd, J = 8.4, 1.6 Hz
2H), 6.68 (d, J = 1.6 Hz, 2H), 6.51 (d, J = 8.4 Hz 2H) 2.98 (sep., 1H), 2.25
(sep., 2H), 2.20 (s, 6H), 1.35 (d, J = 7.2 Hz, 6H), 1.00 (d, J = 6.8 Hz 12H).
13C NMR (100 MHz, CDCl3): δ 149.99, 148.42, 146.06, 145.21, 145.04,
143.26, 142.28, 140.31, 139.86, 132.02, 131.88, 130.32, 130.07, 127.42,
MPAc-BN (3): Compound
3 was synthesized according to the
same procedure described above for the synthesis of 1, except that
7c (1.00 g, 1.86 mmol) and 6 (0.70 g, 1.95 mmol) were used as the
reactants, yielding 3 as a light-yellow solid (yield = 1.42 g, 93%). 1H NMR
(400 MHz, CDCl3): δ 7.84 (d, J = 8.4 Hz, 2H), 7.61 (tt, J = 7.2, 1.5 Hz,
2H), 7.55 (tt, J = 7.2, 1.5 Hz, 1H), 7.47 (td, J = 7.8, 1.8 Hz, 1H), 7.39 (dd,
J = 6.8, 162 Hz 2H), 7.18-7.15 (m, 6H), 7.12-7.09 (m, 3H), 6.93-6.88 (m,
4H), 6.79 (td, J = 8.8, 1.8 Hz, 3H), 6.39 (td, J = 8.4, 1.7 Hz, 4H), 6.36
(d, J = 8.0 Hz, 2H), 3.01 (sep., 1H), 2.50 (sep., 2H), 2.16 (s, 6H), 1.37
(d, J = 6.8 Hz, 6H), 1.05 (dd, J = 6.8, 3.6 Hz, 12H). 13C NMR (100 MHz,
CDCl3): δ 150.50, 147.87, 147.55, 146.32, 146.07, 144.91, 141.35, 139.98,
139.83, 137.62, 132.61, 130.87, 130.65, 130.34, 130.06, 130.04, 129.24,
128.95, 127.47, 127.28, 126.14, 120.90, 119.82, 119.69, 117.88, 116.76,
114.49, 56.93, 35.13, 34.27, 24.50, 24.46, 24.22, 20.80. MS (MALDI-TOF):
m/z calcd 816.46 [M]+; found 816.69. Anal. calcd (%) for C60H57BN2: C
88.21, H 7.03, N 3.43; found: C 88.20, H 7.02, N 3.48.
MPAc-Dip-BS (4): A mixture of 12 (0.65 g, 2.10 mmol), 6 (0.65 g,
1.80 mmol), Pd2(dba)3 (0.028 g, 0.03 mmol), P(t-Bu)3H∙BF4 (0.035 g,
0.12 mmol), and sodium tert-butoxide (0.22 g, 2.25 mmol) in dry toluene
(15 mL) was stirred at 90 °C for 16 h under N2. After cooling to room
temperature, the reaction mixture was added to an aqueous solution of
NH4Cl and then extracted with toluene. The combined organic layers were
washed with brine and dried over anhydrous Na2SO4. After filtration and
Scheme 1. Synthetic routes for dibenzoheteraborin-based TADF emitters 1–4.
©
Adv. Funct. Mater. 2018, 1802031
1802031 (10 of 12)
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.afm-journal.de
evaporation, the crude product was purified by column chromatography
on silica gel (eluent: hexane/toluene = 5:1, v/v). After removal of the
solvents by evaporation, the resulting solid was washed with AcOEt
and hexane to afford 4 as a light yellow solid (yield = 0.78 g, 73%).
1H NMR (400 MHz, CDCl3): δ 7.92 (dd, J = 7.6 Hz, 1.2 Hz, 2H), 7.79
(d, J = 7.6 Hz, 2H), 7.65 (ddd, J = 7.6 Hz, 7.6 Hz, 1.6 Hz, 2H), 7.37 (ddd,
J = 7.6 Hz, 7.6 Hz, 1.2 Hz, 2H), 7.31-7.22 (m, 6H), 7.07 (m, 4H), 6.97
(dd, J = 8.0 Hz, 1.2 Hz, 2H), 6.84 (s, 2H), 6.70 (d, J = 2.0 Hz, 2H), 6.56
(d, J = 8.0 Hz, 2H), 2.26 (sep., J = 6.8 Hz, 2H), 2.23 (s, 6H), 0.89 (d,
J = 6.8 Hz, 12H). 13C NMR (100 MHz, CDCl3): δ 152.60, 146.75,
143.87, 141.37, 140.82, 139.96, 132.24, 130.16, 129.72, 128.65, 127.47,
126.12, 125.24, 124.51, 124.33, 114.03, 56.94, 35.23, 24.05, 20.86. MS
(MALDI-TOF): m/z calcd 715.34 [M]+; found 715.42. Anal. calcd (%) for
C51H46BNS: C 85.58, H 6.48, N 1.96; found: C 85.66, H 6.51, N 2.01.
Photophysical Measurements: Organic thin films for photophysical
measurements were deposited under high vacuum (≈7 × 10−5 Pa)
onto quartz or Si(100) substrates. UV−vis absorption and PL spectra
were measured with a V-670 spectrometer (Jasco) and a FP-8600
spectrophotometer (Jasco), respectively. The absolute PL quantum yields
were determined using an ILF-835 integrating sphere system (Jasco).
The transient PL decay measurements for the thin films were performed
using a C11367 Quantaurus-tau fluorescence lifetime spectrometer
(Hamamatsu Photonics; λ = 340 nm, pulse width = 100 ps, and repetition
rate = 20 Hz) under N2, and a C9300 streak camera (Hamamatsu
Photonics) with a N2 gas laser (λ = 337 nm, pulse width = 500 ps, and
repetition rate = 20 Hz) under vacuum (<4 × 10−1 Pa).
Fabrication and Evaluation of OLED Devices: ITO-coated glass
substrates were cleaned with detergent, deionized water, acetone, and
isopropanol. The substrates were then subjected to UV–ozone treatment
for 30 min before being loaded into an E-200 vacuum evaporation
system (ALS Technology). The organic layers and a cathode Al layer were
thermally evaporated on the substrates under vacuum (<6 × 10−5 Pa) with
a deposition rate of <0.3 nm s−1 through a shadow mask, defining a pixel
size of 0.04 cm2. The thickness and deposition rate were monitored in situ
during deposition by an oscillating quartz thickness monitor. The J–V–L
characteristics of the fabricated OLEDs were measured using a Keithley
2400 source meter and a CS-2000 spectroradiometer (Konica Minolta).
[1] H. Uoyama, K. Goushi, K. Shizu, H. Nomura, C. Adachi, Nature
2012, 492, 234.
[2] a) A. Endo, K. Sato, K. Yoshimura, T. Kai, A. Kawada, H. Miyazaki,
C. Adachi, Appl. Phys. Lett. 2011, 98, 083302; b) S. Y. Lee, T. Yasuda,
H. Nomura, C. Adachi, Appl. Phys. Lett. 2012, 101, 093306.
[3] For reviews, see: a) Y. Tao, K. Yuan, T. Chen, P. Xu, H. Li, R. Chen,
C. Zheng, L. Zhang, W. Huang, Adv. Mater. 2014, 26, 7931;
b) M. Y. Wong, E. Zysman-Colman, Adv. Mater. 2017, 29, 1605444;
c) Y. Im, M. Kim, Y. J. Cho, J.-A. Seo, K. S. Yook, J. Y. Lee, Chem.
Mater. 2017, 29, 1946; d) N. Aizawa, I. S. Park, T. Yasuda, AAPPS
Bull. 2016, 26, 9.
[4] a) M. A. Baldo, D. F. O’Brien, Y. You, A. Shoustikov, S. Sibley,
M. E. Thompson, S. R. Forrest, Nature 1998, 395, 151; b) S. Reineke,
F. Lindner, G. Schwartz, N. Seidler, K. Walzer, B. Lüssem, K. Leo,
Nature 2009, 459, 234; c) T. Fleetham, G. Li, J. Li, Adv. Mater. 2017,
29, 1601861.
[5] a) J. Li, T. Nakagawa, J. MacDonald, Q. Zhang, H. Nomura,
H. Miyazaki, C. Adachi, Adv. Mater. 2013, 25, 3319; b) Q. Zhang,
H. Kuwabara, W. J. PotscavageJr., S. Huang, Y. Hatae, T. Shibata,
C. Adachi, J. Am. Chem. Soc. 2014, 136, 18070; c) S. Wang, X. Yan,
Z. Cheng, H. Zhang, Y. Liu, Y. Wang, Angew. Chem., Int. Ed. 2015,
54, 13068; d) C. Xie, X. Li, D. Chen, Z. Wang, X. Cai, D. Chen,
Y. Li, K. Liu, Y. Cao, S.-J. Su, Adv. Mater. 2016, 28, 181; e) W. Zeng,
H.-Y. Lai, W.-K. Lee, M. Jiao, Y.-J. Shiu, C. Zhong, S. Gong, T. Zhou,
G. Xie, M. Sarma, K.-T. Wong, C.-C. Wu, C. Yang, Adv. Mater.
2018, 30, 1704961; f) R. Furue, K. Matsuo, Y. Ashikari, H. Ooka,
N. Amanokura, T. Yasuda, Adv. Opt. Mater. 2018, 6, 1701147.
[6] a) H. Kaji, H. Suzuki, T. Fukushima, K. Shizu, K. Suzuki, S. Kubo,
T. Komino, H. Oiwa, F. Suzuki, A. Wakamiya, Y. Murata, C. Adachi,
Nat. Commun. 2015, 6, 8476; b) K. Suzuki, S. Kubo, K. Shizu,
T. Fukushima, A. Wakamiya, Y. Murata, C. Adachi, H. Kaji,
Angew. Chem., Int. Ed. 2015, 54, 15231; c) D. R. Lee, M. Kim,
S. K. Jeon, S.-H. Hwang, C. W. Lee, J. Y. Lee, Adv. Mater. 2015, 27,
5861; d) W.-L. Tsai, M.-H. Huang, W.-K. Lee, Y.-J. Hsu, K.-C. Pan,
T.-H. Huang, H.-C. Ting, M. Sarma, Y.-Y. Ho, H.-C. Hu, C.-C. Chen,
M.-T. Lee, K.-T. Wong, C.-C. Wu, Chem. Commun. 2015, 51, 13662;
e) P. Rajamalli, N. Senthilkumar, P. Gandeepan, P.-Y. Huang,
M.-J. Huang, C.-Z. Ren-Wu, C.-Y. Yang, M.-J. Chiu, L.-K. Chu,
H.-W. Lin, C.-H. Cheng, J. Am. Chem. Soc. 2016, 138, 628; f) K. Wu,
T. Zhang, L. Zhan, C. Zhong, S. Gong, N. Jiang, Z.-H. Lu, C. Yang,
Chem. – Eur. J. 2016, 22, 10860; g) J. Lee, N. Aizawa, M. Numata,
C. Adachi, T. Yasuda, Adv. Mater. 2017, 29, 1604854; h) M. Liu,
R. Komatsu, X. Cai, K. Hotta, S. Sato, K. Liu, D. Chen, Y. Kato,
H. Sasabe, S. Ohisa, Y. Suzuri, D. Yokoyama, S.-J. Su, J. Kido,
Chem. Mater. 2017, 29, 8630; i) T.-L. Wu, M.-J. Huang, C.-C. Lin,
P.-Y. Huang, T.-Y. Chou, R.-W. Chen-Cheng, H.-W. Lin, R.-S. Liu,
C.-H. Cheng, Nat. Photonics 2018, 12, 235.
[7] a) Q. Zhang, B. Li, S. Huang, H. Nomura, H. Tanaka, C. Adachi, Nat.
Photonics 2014, 8, 326; b) S. Hirata, Y. Sakai, K. Masui, H. Tanaka,
S. Y. Lee, H. Nomura, N. Nakamura, M. Yasumatsu, H. Nakanotani,
Q. Zhang, K. Shizu, H. Miyazaki, C. Adachi, Nat. Mater. 2015, 14,
330; c) M. Numata, T. Yasuda, C. Adachi, Chem. Commun. 2015,
51, 9443; d) J. W. Sun, J. Y. Baek, K.-H. Kim, C.-K. Moon, J.-H. Lee,
S.-K. Kwon, Y.-H. Kim, J.-J. Kim, Chem. Mater. 2015, 27, 6675;
e) D. Zhang, M. Cai, Y. Zhang, D. Zhang, L. Duan, Mater. Horiz.
2016, 3, 145; f) T. Hatakeyama, K. Shiren, K. Nakajima, S. Nomura,
S. Nakatsuka, K. Kinoshita, J. Ni, Y. Ono, T. Ikuta, Adv. Mater. 2016,
28, 2777; g) S. Y. Lee, C. Adachi, T. Yasuda, Adv. Mater. 2016, 28,
4626; h) T.-A. Lin, T. Chatterjee, W.-L. Tsai, W.-K. Lee, M.-J. Wu,
M. Jiao, K.-C. Pan, C.-L. Yi, C.-L. Chung, K.-T. Wong, C.-C. Wu, Adv.
Mater. 2016, 28, 6976; i) I. S. Park, J. Lee, T. Yasuda, J. Mater. Chem. C
2016, 4, 7911; j) I. S. Park, H. Komiyama, T. Yasuda, Chem. Sci. 2017,
8, 953; k) P. Rajamalli, N. Senthilkumar, P.-Y. Huang, C.-C. Ren-Wu,
H.-W. Lin, C.-H. Cheng, J. Am. Chem. Soc. 2017, 139, 10948.
[8] C. Murawski, K. Leo, M. C. Gather, Adv. Mater. 2013, 25, 6801.
Supporting Information
Supporting Information is available from the Wiley Online Library or from
the author.
Acknowledgements
I.S.P. and K.M. contributed equally to this work. This work was supported
in part by Grant-in-Aid for JSPS KAKENHI Grant Nos. JP17K17937 (K.M.)
and JP18H02048 (T.Y.), the Research Foundation for Opto-Science
and Technology (T.Y.), and the Hoso Bunka Foundation (T.Y.). I.S.P.
acknowledges the support from the Rotary Yoneyama Scholarships.
The authors are grateful for the support of the Cooperative Research
Program “Network Joint Research Center for Materials and Devices.” The
computations were mainly performed using the computer facilities at
Research Institute for Information Technology, Kyushu University.
Conflict of Interest
The authors declare no conflict of interest.
Keywords
dibenzoheteraborins, electroluminescence, organic light-emitting diodes,
spin conversion, thermally activated delayed fluorescence
Received: March 22, 2018
Revised: May 28, 2018
Published online:
©
Adv. Funct. Mater. 2018, 1802031
1802031 (11 of 12)
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.afm-journal.de
[9] J. Lee, N. Aizawa, T. Yasuda, Chem. Mater. 2017, 29, 8012.
[10] a) K. Schmidt, S. Brovelli, V. Coropceanu, D. Beljonne, J. Cornil,
C. Bazzini, T. Caronna, R. Tubino, F. Meinardi, Z. Shuai, J.-L. Brédas,
J. Phys. Chem. A 2007, 111, 10490; b) J.-L. Brédas, D. Beljonne,
V. Coropceanu, J. Cornil, Chem. Rev. 2004, 104, 4971.
[19] a) Y. Kitamoto, T. Namikawa, D. Ikemizu, Y. Miyata, T. Suzuki,
H. Kita, T. Sato, S. Oi, J. Mater. Chem. 2015, 3, 9122; b) Y. Kitamoto,
T. Namikawa, T. Suzuki, Y. Miyata, H. Kita, T. Sato, S. Oi, Tetrahedron
Lett. 2016, 57, 4914; c) H. Hirai, K. Nakajima, S. Nakatsuka,
K. Shiren, J. Ni, S. Nomura, T. Ikuta, T. Hatakeyama, Angew. Chem.,
Int. Ed. 2015, 54, 13581.
[20] I. S. Park, M. Numata, C. Adachi, T. Yasuda, Bull. Chem. Soc. Jpn.
2016, 89, 375.
[21] The comparatively smaller ΔEST values for 1 and 2, compared to that
of 3, can be associated with their enhanced CT character in the T1
state (97% vs 87%).
[11] P. K. Samanta, D. Kim, V. Coropceanu, J.-L. Brédas, J. Am. Chem.
Soc. 2017, 139, 4042.
[12] a) J. Gibson, A. P. Monkman, T. J. Penfold, ChemPhysChem 2016, 17,
2956; b) F. B. Dias, J. Santos, D. R. Graves, P. Data, R. S. Nobuyasu,
M. A. Fox, A. S. Batsanov, T. Palmeira, M. N. Berberan-Santos,
M. R. Bryce, A. P. Monkman, Adv. Sci. 2016, 3, 1600080;
c) M. K. Etherington, J. Gibson, H. F. Higginbotham, T. J. Penfold,
A. P. Monkman, Nat. Commun. 2016, 7, 13680.
[22] Y. J. Cho, S. K. Jeon, S.-S. Lee, E. Yu, J. Y. Lee, Chem. Mater. 2016, 28,
5400.
[13] X.-K. Chen, S.-F. Zhang, J.-X. Fan, A.-M. Ren, J. Phys. Chem. C 2015,
119, 9728.
[14] C. M. Marian, J. Phys. Chem. C 2016, 120, 3715.
[15] a) M. Kleinschmidt, C. van Wüllen, C. M. Marian, J. Chem. Phys.
2015, 142, 094301; b) E. Y.-T. Li, T.-Y. Jiang, Y. Chi, P.-T. Chou, Phys.
Chem. Chem. Phys. 2014, 16, 26184.
[16] a) A. Kretzschmar, C. Patze, S. T. Schwaebel, U. H. F. Bunz, J. Org.
Chem. 2015, 80, 9126; b) Y. Xiang, Y. Zhao, N. Xu, S. Gong, F. Ni,
K. Wu, J. Luo, G. Xie, Z.-H. Lu, C. Yang, J. Mater. Chem. C 2017, 5,
12204.
[17] a) W. Zhang, J. Jin, Z. Huang, S. Zhuang, L. Wang, Sci. Rep. 2016, 6,
30178; b) M. Einzinger, T. Zhu, P. de Silva, C. Belger, T. M. Swager,
T. Van Voorhis, M. A. Baldo, Adv. Mater. 2017, 29, 1701987.
[18] a) T. Agou, J. Kobayashi, T. Kawashima, Chem. – Eur. J. 2007, 13,
8051; b) J. Kobayashi, K. Kato, T. Agou, T. Kawashima, Chem. –
Asian J. 2009, 4, 42; c) T. Agou, J. Kobayashi, T. Kawashima, Chem.
Commun. 2007, 3204; d) T. Agou, J. Kobayashi, T. Kawashima,
Org. Lett. 2006, 8, 2241; e) T. Agou, M. Sekine, J. Kobayashi,
T. Kawashima, Chem. Commun. 2009, 1894.
[23] a) R. A. Marcus, J. Phys. Chem. 1989, 93, 3078; b) I. R. Gould,
D. Noukakis, L. Gomez-Jahn, R. H. Young, J. L. Goodman, S. Faid,
Chem. Phys. 1993, 176, 439; c) K. Vandewal, K. Tvingstedt, A. Gadisa,
O. Inganäs, J. V. Manca, Phys. Rev. B 2010, 81, 125204.
[24] C. F. Madigan, V. Bulovi , Phys. Rev. Lett. 2003, 91, 247403.
[25] S. M. Menke, R. J. Holmes, J. Phys. Chem. C 2016, 120, 8502.
[26] a) R. Furue, T. Nishimoto, I. S. Park, J. Lee, T. Yasuda, Angew. Chem.,
Int. Ed. 2016, 55, 7171; b) N. Aizawa, C.-J. Tsou, I. S. Park, T. Yasuda,
Polym. J. 2017, 49, 197.
[27] a) J. Guo, X.-L. Li, H. Nie, W. Luo, S. Gan, S. Hu, R. Hu, A. Qin,
Z. Zhao, S.-J. Su, B. Z. Tang, Adv. Funct. Mater. 2017, 27, 1606458;
b) J. Huang, H. Nie, J. Zeng, Z. Zhuang, S. Gan, Y. Cai, J. Guo,
S.-J. Su, Z. Zhao, B. Z. Tang, Angew. Chem., Int. Ed. 2017, 56,
12971.
[28] B. T. Lim, S. Okajima, A. K. Chandra, E. C. Lim, Chem. Phys. Lett.
1981, 79, 22.
[29] H. Sasabe, E. Gonmori, T. Chiba, Y.-J. Li, D. Tanaka, S.-J. Su,
T. Takeda, Y.-J. Pu, K. Nakayama, J. Kido, Chem. Mater. 2008, 20,
5951.
©
Adv. Funct. Mater. 2018, 1802031
1802031 (12 of 12)
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim