Blocking Energy-Loss Pathways for Ideal Fluorescent Organic Light-Emitting Diodes with Thermally Activated Delayed Fluorescent Sensitizers

Article abstract of DOI:10.1002/adma.201705250

Organic light-emitting diodes (OLEDs) based on thermally activated delayed fluorescence-sensitized fluorescence (TSF) offer the possibility of attaining an ultimate high efficiency with low roll-off utilizing noble-metal free, easy-to-synthesize, pure org

Full text of DOI:10.1002/adma.201705250

CommuniCation
Organic Light-Emitting Diodes
www.advmat.de
Blocking Energy-Loss Pathways for Ideal Fluorescent
Organic Light-Emitting Diodes with Thermally Activated
Delayed Fluorescent Sensitizers
Dongdong Zhang,* Xiaozeng Song, Minghan Cai, and Lian Duan*
The dilemma for materials with TADF
Organic light-emitting diodes (OLEDs) based on thermally activated delayed
fluorescence-sensitized fluorescence (TSF) offer the possibility of attaining
an ultimate high efficiency with low roll-off utilizing noble-metal free, easy-
to-synthesize, pure organic fluorescent emitters. However, the performances
of TSF-OLEDs are still unsatisfactory. Here, TSF-OLEDs with breakthrough
efficiencies even at high brightnesses by suppressing the competitive deac-
tivation processes, including direct charge recombination on conventional
fluorescent dopants (CFDs) and Dexter energy transfer from the host to the
CFDs, are demonstrated. On the one hand, electronically inert terminal-
substituents are introduced to protect the electronically active core of the
CFDs; on the other hand, delicate device structures are designed to provide
multiple energy-funneling paths. As a result, unprecedentedly high maximum
as emitters is a small ΔEST intrinsically
accompanied by small radiative decay rates
5
7
−1
(
k , typically 10 –10 s ), invoking insta-
r
bility and efficiency roll-off at high current
[9,10]
density.
This tradeoff can be broken by
delivering the singlet energy of TADF to
conventional phosphorescent or fluorescent
dyes through the more efficient Förster res-
[11]
onance energy transfer (FET). With the
TADF-sensitized phosphorescence (TSP)
technology, ultimate efficiencies have been
achieved with a large critical current den-
sity (J1/2), which, however, cannot avoid
the utilization of noble heavy metal atoms.
Commonly, the commercial applications of
OLED technology require high brightness
−1
external quantum efficiency/power efficiency of 24%/71.4 lm W in a green
−1
TSF-OLED are demonstrated, which remain at 22.6%/52.3 lm W even at a
−2
−2
levels approaching 5000 cd m , given the
predicted per-m2 manufacturing cost of
large area illumination as well as the lim-
ited aperture ratio of the high-resolution
high luminance of 5000 cd m . The work unlocks the potential of
TSF-OLEDs, paving the way toward practical applications.
[12]
displays.
Also, as the next challenging
Outstanding properties of materials with thermally activated
delayed fluorescence (TADF) have triggered a new insight
into the next-generation organic semiconductors for organic
light-emitting diodes (OLEDs).[1–4] What distinguishes TADF
compounds from conventional organic materials is the small sin-
target in organic semiconductors, an electrically pumped organic
6
laser requires current densities ≈10 times larger than current
[12]
display applications. Theoretically, TADF-sensitized fluorescent
(TSF) OLEDs may offer prospects in realizing ultimate efficien-
cies with even lower efficiency roll-off, higher color-purity and
longer term-stabilities than TSP-OLEDs. Strategies to realize TSF
using materials with TADF as the sensitizing host or the assistant
dopant have been proposed by our group and Prof. Adachi and
co-workers separately, realizing dual enhancement of electrolumi-
glet–triplet energy gap (ΔE ), inducing efficient triplet-to-singlet
ST
spin-flip transition.[ Intriguingly, this so-called reverse inter-
system crossing process (RISC) has aroused many fascinating
properties, rendering materials with TADF suitable not only for
5]
[13–16]
next-generation emitters but also for ideal hosts in OLEDs with
the latter being believed to be more adequate for practical appli-
cations owing to their unique technological merits.[
nescence efficiency and operational stability.
However, the
performances of TSF-OLEDs are still far from satisfaction. How
to unlock the full potential of TSF-OLEDs is still unrevealed.
Here, we demonstrate ultimately efficient TSF-OLEDs even
at high brightnesses by blocking the competitive deactivations
processes. On one hand, electronically inert terminal-substit-
uents are introduced to substantially shield the electronically
active core of the should be changed to conventional fluores-
cent dopants (CFDs) from contact with adjacent molecules
while not affecting their evaporability. On the other hand, dedi-
cate device structures are exploited to provide multiple energy
funneling paths involving one from high-energy interfacial
exciplex to an intermediate TADF host and then to CFDs, pro-
moting the singlet excitons ratio while reducing triplet one.
The energy-loss paths, such as direct charge recombination
on the fluorescent dopants (CFDs) and Dexter energy transfer
from the host to the CFDs, are thereof blocked. The resulting
6–8]
Dr. D. D. Zhang, X. Z. Song, M. H. Cai, Prof. L. Duan
Key Lab of Organic Optoelectronics
and Molecular Engineering of Ministry of Education
Department of Chemistry
Tsinghua University
Beijing 100084, China
E-mail: 18810644695@163.com; duanl@mail.tsinghua.edu.cn
Prof. L. Duan
Center for Flexible Electronics Technology
Tsinghua University
Beijing 100084, P. R. China
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/adma.201705250.
DOI: 10.1002/adma.201705250
Adv. Mater. 2017, 1705250
1705250 (1 of 10)
© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.advmat.de
active cores of CFDs, not only increasing the intermolecular
distance but also reducing the orbital overlap of neighboring
molecules. Therefore, efficiently suppressed DET can be antici-
pated. Additionally, given that FET is long-range dipole–dipole
interaction in the order of several nms, not requiring orbital
[20]
overlap, it can be anticipated that steric bulk units on the
CFDs only shows limited influence on the Förster interaction,
which is the prerequisite for efficient TSF process. It should be
noted that too large terminal units may affect the evaporability of
the CFDs, rendering them only suitable for solution-processed
[21]
OLEDs, not beneficial for high efficiencies.
Therefore, the
size of the terminal substituents should be wisely tuned to be
suitable for highly efficient evaporated devices.
The CDFs testified in this work are presented in Figure 2a.
All those dyes involve N ,N ,N ,N -tetraphenylanthracene-
,10-diamine (PAD) as the luminophores, where anthracene
9
9
10 10
Figure 1. The energy loss paths are indicated by the red lines. FET is the
Förster energy transfer while DET is the Dexter energy transfer.
9
unit was chosen due to its large conjugation rigid planar, ben-
TSF-OLEDs exhibit maximum external quantum efficiency
efiting to increase the k s. Different size terminal-substituents,
r
−1
(
EQEmax)/power efficiency (PE ) of 24%/71.4 lm W , which
methyl, tert-butyl, and 2-phenylpropan-2-yl units, which are
intrinsically electronically inert moieties, were adopted to be
attached on the 9- and 10-position of the diamine units. The
larger the size of the terminal units, the more significant ster-
ical effect can be anticipated. Here the terminal substituents
are attached on the diamine units rather than directly on the
anthracene units to avoid the significant influence on geomet-
rical and electronic properties of the dyes. Also, the terminal
units chosen here do not influence the evaporability of the
CFDs as they can be purified by the sublimation.
All the materials were featured with the crystal structures as
shown in Figure 2b from both side- and plan-views. All dyes
show twisted structures with the dihedral angles between the
phenyl planes being 54.89°, 61.2°, 46.74°, and 69.65° for PAD,
MePAD, tBuPAD, and PhtBuPAD, respectively, indicating that
the terminal-substituents show slight influence on the mole-
cular structures. The distance from the center of the anthracene
planes to the farthest atoms of the terminal-substituents were
also measured to be 6.821, 8.180, 9.253, and 10.507 Å for PAD,
MePAD, tBuPAD, and PhtBuPAD, respectively, suggesting that
the molecular sizes are gradually increased. Therefore, it can be
anticipated that the distances between the anthracene cores and
the neighboring molecules are gradually increased from PAD
to PhtBuPAD.
The geometrical and electronic properties of the CFDs were
simulated using density functional theory (DFT) with the most
popular functional, B3LYP, to study the structure–property rela-
tionship of the fluorescent dyes. As illustrated in Figure 2c,
the lowest unoccupied molecule orbitals (LUMOs) of all dyes
are located on the anthracene units while the highest occupied
molecule orbitals (HOMOs) distributions are extended to the
diamine units owing to their strong electron-donating proper-
ties. Notably, no energy level distributes on the terminal-sub-
stituents, proving that those units only act as electronically
inert moieties surrounding the electronically active PAD cores
for all CFDs with different sterical effect. The DFT calculations
for the lowest excited triplet states of these dyes revealed that
the spin-density distributions also resided on the electronically
active PAD units, surrounded by the terminal units.
max
−1
remain at 22.6%/52.3 lm W even at a high luminance of
5
−
2
000 cd m . Those results testify the validity of such strategy
and may shed more light on developing more efficient
TSF-OLEDs suitable for practical applications.
The TSF process is illustrated in Figure 1. Ideally, the triplet-
to-singlet spin-flip transition happens on the TADF material
through efficient RISC process, and then the formed singlet
energy will be transferred to the CFDs through the long-range
FET.[ As indicated by the red arrows, direct charge recombina-
tion on the CFDs as well as Dexter energy transfer (DET) from
the triplet of the host to that of the CFDs are the two paths to
loss. Extremely low CFD concentrations (<1 wt%) are usually
adopted to reduce the loss paths, which, however, increase the
device fabrication difficulty in controlling accurate and constant
doping concentration and the effect is limitedly referred from
the low device efficiencies.[
13]
17]
In terms of direct recombination on CFDs, energy level
tuning of the host and the dyes to achieve Langevin recom-
bination can efficiently reduce this problems. For the Dexter
interaction in the emitting layers (EMLs), unfortunately, the
close proximity of the host materials and dyes mixed together
increases the likelihood of such deactivation process. Techni-
cally, DET happens when the excited electron is transferred
from the excited state of D (D*) to the ground state of the
acceptor (A) via a nonradiative path. It has been established that
the DET rate (kDET) decreases as exp(−2rDA/L), where rDA is the
donor–acceptor separation relative to their van der Waals radii,
L. This means that kDET is significantly influenced by the inter-
molecular distances, which drops to negligibly small values as
the intermolecular distance increases more than on the order
of one or two molecular diameters.[
18,19]
Besides, DET also
requires the overlap of the molecular orbitals of neighboring
[
20]
molecules since electron exchange process is involved.
Therefore, only donors in close proximity to acceptors may
transfer their energy by the Dexter mechanism.
Reasonably, increasing the intermolecular distance may effi-
ciently reduce the DET process. One way to increase intermo-
lecular distances through molecule design is the adoption of the
electronically inert terminal substituents. Owing to their large
steric effect, inert units can sterically shield the electronically
Based on the above calculations, it is reasonable to draw the
conclusion that the electronically inert terminal-substituents
Adv. Mater. 2017, 1705250
1705250 (2 of 10)
© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.advmat.de
Figure 2. a) The molecule structures of the dyes. b) The single-crystal structures of the fluorescent dyes observed from the side-view and the plan-view.
c) The calculated geometrical structures of the dyes as well as the HOMO and LUMO as well as the triplet spin density distributions.
appended to diamines units only exert the sterical effect to
substantially shield the electronically active core of the dyes
from contact with adjacent molecules. Thus, the terminal units
not only increase intermolecular distance between the lumi-
nophores of the dyes and the neighboring molecules but also
reduce their orbital overlaps, which, are the prerequisite for
efficient intermolecular electron exchange interactions. There-
fore, the Dexter interactions can be greatly suppressed by the
terminal-substituents. Since the shielding effect is related with
the size of the terminal units, it can be anticipated that from
Adv. Mater. 2017, 1705250
1705250 (3 of 10)
© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.advmat.de
Figure 3. a) The absorption and emission spectra of the dyes in toluene solutions with concentration of 10–5 mol L−1. b) The CV curves of the dyes
in dichloromethane. c) The emission spectra of the DPEPO: 20 wt% TADF: with or without 3 mol% dye films. d) The PL transient decay curves of the
4
CzTPN-Ph in the DPEPO: 20 wt% TADF: with or without 3 mol% dye films observed at 590 nm. The insert scheme is the energy transfer progress in
the doped films and the molecule structure of 4CzTPN-Ph.
PAD to PhtBuPAD, as the terminal size gradually increased,
the Dexter interactions can also be gradually suppressed.
The calculated HOMO and LUMO energy levels of the mole-
cules were also shown in Figure 2c, exhibiting that the energy
levels of other emitters are slightly shallower than those of PAD
due to the enhanced electron-donating ability of the side groups
by the introduction of the terminal-substituents. Additionally,
the singlet and triplet energies were also calculated using time-
dependent DFT (TD-DFT) with B3LYP basis. The singlet energy
of PAD is relatively higher than other dyes resulted from its rel-
atively larger energy gap, indicating that the enhanced electron-
donating ability of the side groups reduce the molecular energy
gaps. The triplet energies were calculated to be 1.60, 1.57,
with the calculated results. The φ s were also measured in tol-
PL
uene to be 0.89, 0.80, 0.80, and 0.83 for PAD, MePAD, tBuPAD,
and PhtBuPAD, respectively. The slight reduction in φ s of
PL
other emitters than that of PAD may be attributed to the rota-
tion of the terminal units which induces more relaxation pro-
cess in the solution. The PL transient decay curves of the emit-
ters were also measured and shown in Table 1 with lifetimes
(τs) of 10.96, 12.64, 13.29, and 13.16 ns for PAD, MePAD,
tBuPAD, and PhtBuPAD, respectively. The k s of the dyes can
r
also be calculated with the equation of k = φ /τ, which are all
r
PL
8
−1
in the range of 10 s , much higher than most of the TADF
emitters with k s of usually in the range of 10 –10 s .
5
7 −1
r
The cyclic voltammetry (CV) oxidation curves of the dyes
were illustrated in Figure 3b, showing that the electrochemical
properties of the dyes were significantly influenced by the ter-
minal units. Compared to that of PAD, the oxidation potentials
of the MePAD and tBuPAD are relatively lower, derived from
the strong electron-donating properties of the methyl and tert-
butyl units. Notably, the oxidation potential of PhtBuPAD is
higher than those of MePAD and tBuPAD, indicating that the
electron donating properties of the alkyl group are diluted by the
phenyl units which can disperse the rich electrons. The HOMO
energy levels of the emitters can be calculated to be 5.24, 5.15,
5.0, and 5.25 eV respectively for PAD, MePAD, tBuPAD, and
PhtBuPAD, respectively, determined from their anodic oxida-
tion potentials. The LUMO energy levels were calculated by
1
.58, and 1.58 eV for PAD, MePAD, tBuPAD, and PhtBuPAD,
respectively. The large ΔE s of those dyes facilitate to achieve
ST
the high photoluminance quantum yields (φ s) of the dyes by
PL
suppressing the intersystem crossing (ISC) process.
The photophysical properties of the emitters were investi-
gated in toluene with concentration of 10− mol L as shown
in Figure 3a. Intriguingly, all emitters show almost the same
absorption spectrum, derived from the same electronically
active cores which were not disturbed by the terminal-substit-
uents appended to diamines units. Furthermore, compared to
PAD with an emission peak at 510 nm, other emitters show
slightly red-shifted emission with emission peaks of 531 nm for
MePAD and 526 nm for tBuPAD and PhtBuPAD, inconsistent
5
−1
Adv. Mater. 2017, 1705250
1705250 (4 of 10)
© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.advmat.de
Table 1. Properties of the fluorescent emitters.
φPLa)
S1a)
[eV]
T1b)
[eV]
Ega)
[eV]
HOMOc)
[eV]
LUMOc)
[eV]
τ
[ns]
Emitters
Emissiona)
nm]
kr
[× 10 s−1]
8
[
PAD
510
531
526
526
0.89
0.80
0.80
0.83
2.43
2.34
2.36
2.36
1.60
1.57
1.58
1.58
2.42
2.40
2.42
2.42
5.24
5.15
5.0
2.82
2.75
2.58
2.83
10.96
12.64
13.29
13.16
8.1
6.3
5.9
6.3
MePAD
tBuPAD
PhtBuPAD
5.25
a)Measured in toluene solution with concentration of 10−5 mol L−1; b)Calculated using DFT with B3LYP basis; c)Measured in dichloromethane.
the HOMO values and the energy gaps (E s) obtained from the
onsets of the absorption spectra of these emitters, which are
listed in Table 1.
The suppressed DET by the terminal-substituents can
be directly testified by the PL transient decay curves of the
mixed films of bis[2-(diphenylphosphino)phenyl] ether oxide
The electroluminance (EL) performances of the CFDs
were examined with the device structures of indium tin oxide
(ITO)/4,4′-(cyclohexane-1,1-diyl)bis(N,N-di-p-tolylaniline) (TAPC,
30 nm)/tris(4-(9H-carbazol-9-yl)phenyl)amine (TCTA, 10 nm)/
10,10′-(sulfonylbis(4,1-phenylene)) bis(10H-phenoxazine) (PXZ-
DPS): 0.5 mol% CFDs (30 nm)/9,10-bis(4-(2-phenyl-1H-benzo[d]
imidazol-1-yl)phenyl) anthracene (BPBiPA, 40 nm)/LiF (0.5 nm)/
Al (150 nm) as is shown in Figure 4a (Device Type I). Here, PXZ-
DPS was chosen as the TADF sensitizing host, possessing a small
g
(
1
DPEPO): 20 wt% 2,3,5,6-tetrakis(3,6-diphenylcarbazol-9-yl)-
,4-dicyanobenzene (4CzTPN-Ph) with and without 3 mol%
CFDs, observed at the emission peak of the orange TADF emitter,
CzTPN-Ph.[ The singlet energy of 4CzTPN-Ph is reported to
1]
ΔEST of 0.08 eV with an emission peak at 508 nm. Though the
[23]
4
[1]
be about 2.2 V with an emission spectra peaked at 577 nm.
small ΔEST renders PXZ-DPS an efficient RISC efficiency, its k is
r
7
−1
Given that the lower singlet energy of 4CzTPN-Ph than CFDs,
the energy transfer between the singlet excitons is avoided.
Therefore, the only energy transfer process in the system is the
DET from the triplet of 4CzTPN-Ph to that of the dyes. Figure 3c
shows the emission spectra of the doped films, exhibiting only
the emission from 4CzTPN-Ph. The PL decay curves of 4CzTPN-
Ph in the doped films with or without CFDs were measured and
illustrated in Figure 3d, the lifetimes of the prompt parts (τps)
remain the same, indicating that no singlet excitons of 4CzTPN-
Ph involved in the energy transfer process in the system. Contra-
rily, compared with that of the film without CFDs, the lifetimes
of the delayed parts (τds) of the films with dyes were reduced,
assigned to the DET from the TADF material to the CFDs, which
induces an additional triplet exciton decay process.
only about 1.7 × 10 s , nearly two orders lower than that of the
CFDs. As can be seen from Figure S1 of the Supporting Informa-
tion, significant overlaps between the emission spectrum of PXZ-
DPS and the absorption spectra of the dyes are observed, indi-
cating efficient energy transfer. As elucidated by the energy level of
the devices (Figure 4a), due to the relatively deeper LUMO energy
level of PXZ-DPS than those of dyes, electrons from the adjacent
electron-transporting layer will be preferentially injected into the
PXZ-DPS host, and then combined with the holes injected from
TCTA layer owing to the small energy barriers. Therefore, direct
charges recombine on the dyes can be avoided, also evidenced by
the measured electroluminescence decay curves of the devices, as
illustrated in Figure S2 of the Supporting Information, where no
spike was observed after the voltage being turned off, indicating
[13]
The intrinsically decay curves of triplet after removing the
no charge trapping on the dyes.
[22]
excitation source can be written as
As illustrated in Figure 4b and Figure S3 of the Supporting
Information, when observed at the emission peaks of the dyes,
the PL transient decay curves of the PXZ-DPS: x mol% dye
films show clearly delayed components, evidencing the TSF
process in the films. Besides, the lifetimes of the delayed parts
are gradually decreased with the dye concentration increasing,
owing to the enhanced FET from the TADF to the fluorescent
dyes. The same trend was also observed in the EL decay curves
of the devices with increasing dye concentrations (Figure S2,
Supporting Information). The emission spectra of the TSF-
OLEDs are shown in Figure 4c, where the emissions of the
dyes with terminal units show slightly redshift compared with
that of PAD, corresponding to the spectra obtained with PL
excitation. The half-peak widths are 87, 76, 73, and 80 nm for
devices with PAD, MePAD, tBuPAD, and PhtBuPAD, respec-
tively. Contrarily, the device with PXZ-DPS as the emitter was
also fabricated, showing a half-peak width of 105 nm, wider
than that of the TSF-OLEDs. The relatively narrow spectrum
of TSF-OLEDs render them suitable for display applications,
where high color-purity is required.
T
d
[
T
1
]
/d
[
t
]
= −
(
k
nr +kRISC
)
[
T
1
]
+kISC
[
S
1
]
(1)
T
nr
where k stands for the nonradiative decay rates of the triplet.
kRISC is the RISC rates while kISC represents the ISC rates.
When a DET process is introduced, the decay curves of T1
can be expressed as
T
d
[
T
1
]
/d
[
t
]
= −
(
k
nr +kDET +kRISC
)
[
T
1
]
+kISC
[
S
1
]
(2)
where the kDET stands for the DET rates from the TADF
triple to that of the fluorescent dyes. According to those two
equations, the DET process will quicken the decay process
of triplet excitons of the TADF material, leading to relatively
shorter lifetimes. Additionally, as can be seen from Figure 3d,
the smaller the size of the terminal units of the dyes, the shorter
the τ s of the films were observed. According to Equation (2),
the shorter τ indicates the larger kDET. The films doped dyes
with relatively larger terminal-substituents show the longer τds
and thus the smaller kDETs, evidencing that the large terminal-
substituents can suppress Dexter interactions.
d
d
The device efficiencies with dopant concentration of
0.5 mol% are shown in Figure 4d, proving that the larger the
Adv. Mater. 2017, 1705250
1705250 (5 of 10)
© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.advmat.de
Figure 4. a) The TSF-OLED structures of Device Type I and some of the molecule structures. b) The PL decay curves of the host: PhtBuPAD doped
films. c) The EL-spectrum of the TSF-OLEDs. d) The EQE-brightness curves of the TSF-OLEDs based on different dyes. e) The normalized EQEmax and
the φPLs versus the dopant concentrations.
terminal-substituents of CFDs, the higher the device EQEmax
However, this cannot explain the inferior efficiencies based
can be obtained. The φ s of the doped films were measured
on MePAD and tBuPAD since their φ s are similar to that of
PL
PL
to be 0.77, 0.85, 0.85, and 0.86 for PAD, MePAD, tBuPAD, and
PhtBuPAD, respectively. The much lower device efficiencies of
PAD based device than that of PhtBuPAD can be partly attrib-
uted to the relatively lower φPL of PAD than that of PhtBuPAD.
PhtBuPAD. Considering no direct charge recombination on
the dyes, we believe the difference of the device efficiencies is
owing to the influence of the Dexter interactions in the EMLs.
As discussed above, the terminal-substituents of the dyes can
Adv. Mater. 2017, 1705250
1705250 (6 of 10)
© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.advmat.de
suppress the DET process in the devices and the suppression
effect greatly depends on size of the units. The sizes of the ter-
minal units from PAD to PbtBuPAD are gradually increased,
leading to enhanced suppression effect, and thus their device
efficiencies show the same trend.
sources, from where the energy will be transferred either to the
intermediate TADF material and then to the CFDs, or directly
to the CFDs, as illustrated in Figure 5d, constituting multiple
energy funneling paths.
The concentration of PXZ-DPS and PhtBuPAD in the
devices were optimized 30 wt% and 2 mol%, respectively. Both
device showed relatively higher efficiencies than the one with
only PXZ-DPS as the host, due to the suppressed annihilations
To further reveal the influence of the terminal-substituents
on device performances, the EQEmax of the devices and the
φPLs of the doped films with the gradually increased dopant
concentrations were measured and the normalized values
were summarized in Figure 4e. For all dyes, the concentration
influence on the φ s is quite limited. Contrarily, the EQE
induced by PXZ-DPS itself. Remarkably, a EQE
of 24% with
max
−
1
a PEmax of 71.4 lm W was observed for Device Type III. To
the best of our knowledge, this is the highest values for the
TSF-OLEDs reported so far. Also, the efficiency roll-off of the
devices was extremely small, with EQE/PE remaining at 22.6%
PL
max
of the devices are gradually reduced with the dye concentration
increasing. Ruling out the influence of φ s, it is reasonable to
PL
−
1
−2
make the conclusion that the reduction in EQEmax is due to the
enhanced DET derived from the increased dopant concentra-
tion, which shortens the intermolecular distances. Addition-
ally, it seems that the larger the terminal units, the slower the
reduction rate in EQEmax with dopant concentration increasing,
further proving that large terminal substituents can suppress
the DET and thus the exciton loss. However, though dyes with
large terminal units can suppress DET, EQEmax of only 13.7%
was achieved, still much lag behind the phosphorescent or the
TADF OLEDs.
and 52.3 lm W even at the brightness level of 5000 cd m ,
suitable for practical applications. Contrarily, Device Type II
showed inferior performances with the EQEmax/PEmax of only
−
1
19.0%/48.1 lm W and relatively more significant efficiency
roll-off. Since the current density of both devices is almost
the same, the influence of the charge balance on device per-
formances is negligible (Figure S6, Supporting Information).
Also, no spike was observed for the EL decay curves of both
devices after the voltage was turned-off, proving no direct
charge recombination on the dyes. Therefore, the main differ-
ence derived from interfacial exciplex.
Further efficiency promotion may be realized by intro-
ducing a wide-energy-gap host to dilute the TADF materials to
suppress the aggregation-caused quenching effect.[ A new
bipolar wide-energy-host, 3-(3-(4,6-diphenyl-1,3,5-triazin-2-yl)-
phenyl)-9-phenyl-9H-carbazole (PhCzTrz) was developed to
dilute PXZ-DPS to suppress exciton annihilation. Owing to the
deep LUMO and HOMO energy levels of PhCzTrz, which were
In Device Type II, only sensitizing path from PXZ-DPS to
dyes exists. Also, the formation of the PhCzTrz excitons adds
an additional energy loss through DET from triplet states of
PhCzTrz to dyes as shown in Figure 5c, especially under high
current density, which may account for the relatively high effi-
ciency roll-off. Contrarily, when interfacial exciplex formed, the
formation of the wide-energy-gap host excitons can be sup-
pressed. As illustrated in Figure 5d, on one hand, as competing
processes, the high RISC rate of the exciplex can suppress the
DET from the exciplex triplet to the fluorescent dyes. On the
other hand, the enhanced Förster energy transfer from the exci-
plex to the TADF material can greatly promote the ratio of the
singlet excitons (>25%) of the TADF material while suppressing
the triplet ones (<75%), which also suppresses DET from TADF
triplet to that of the fluorescent dyes. Therefore, the exciton loss
through DET can be greatly reduced with the second-order fun-
neling path, benefiting the device efficiencies.
14]
2
.93 and 6.15 eV, respectively, exciplex emission was observed
from PhCzTrz and TCTA mixed films as is shown in Figure S4
of the Supporting Information, exhibiting efficient TADF prop-
erties as indicated by the delayed part of the decay curve. But
there is no exciplex formed between PhCzTrz and 3,3-di(9H-
carbazol-9-yl)biphenyl (mCBP). Based on this, two kinds of
TSF-OLEDs (Device Type II and III) were fabricated for com-
parison as shown in Figure 5a,b. Also, it was demonstrated that
interfacial exciplex at the interface of PhCzTrz/TCTA layers was
formed in Device Type III as shown in Figure S5 (Supporting
Information).
The existing of the interfacial exciplex greatly modulates the
charge recombination ways in the devices. As elucidated by the
device energy levels, electrons from the adjacent electron-trans-
porting layer will be injected into the deep LUMO energy level
of PhCzTrz in both devices. However, the situation for holes
is different. In Device Type II, the holes from the mCBP layer
will be preferentially injected into the HOMOs of PXZ-DPS but
some still will be injected into the PhCzTrz, directly forming
PhCzTrz excitons, especially under high voltage. The accu-
mulated holes on the PXZ-DPS changes the internal electric
field distributions, shifting its LUMO to a deeper level. Elec-
trons thereafter become accessible from PhCzTrz to PXZ-DPS,
forming PXZ-DPS excitons. Only TSF process from PXZ-DPS
to the dyes can be anticipated, as shown in Figure 5c. In terms
of Device Type III, the electrons on PhCzTrz will be quickly
transferred to the TCTA/emitting layers interfaces to combine
with the holes in the TCTA layer to form low energy interfa-
cial exciplex excitons. The interfacial exciplex acts like exciton
Direct sensitizing process from the interfacial exciplex to the
fluorescent dyes can be demonstrated by the EL decay curves of
the fluorescent devices without PXZ-DPS as shown Figure S7
(Supporting Information). For the fluorescent device with inter-
facial exciplex, the decay lifetime is shorter than that of the
exciplex devices but higher than the devices without interfacial
exciplex, demonstrating the direct energy-funneling from the
exciplex to the dopant, which also enhances TSF process and
suppresses exciton loss through Dexter interactions. Besides,
clear overshoot was observed after the voltage was turned off in
the device without interfacial exciplex, attributing to the charge
trapping on the dye. Contrarily, there is no direct charge trap-
ping on the dyes when interfacial exciplex exists, indicating
that interfacial exciplex facilitates to reduce charge recombi-
nation on the fluorescent dyes. Those results greatly show the
vital importance of the formation of the interfacial exciplex,
not only achieving multiple energy funneling but also modu-
lation charge recombination in the emitting layers. Here, it is
[
24]
Adv. Mater. 2017, 1705250
1705250 (7 of 10)
© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.advmat.de
Figure 5. a,b) The energy level diagrams of Device Type II and III. c,d) The energy funneling mechanisms in Device Type II and III. e) The EQE–
brightness curves of Device Type II and III. f) The EQEmax–dopant concentration relationships of Device Type III with different CFDs.
worth noting that triplet fusion for anthracene derivatives has
size of the terminal substituents can suppress the excitons loss
through DET. Moreover, comparing with the trend of the device
with only PXZ-DPS as the hosts, the EQEmax reduction rate is
much slower, also evidencing that the exciton loss was further
reduced in the devices with multiple energy funneling paths.
In conclusion, we have shown that to boost the TSF-OLED
performances, the exciton loss paths must be blocked. The
CFDs with their electronically active cores shielded by the
electronically inert terminal-substituents as well as the device
structures with multiple energy funneling paths were exploited,
reducing charge recombination on the CFDs and suppressing
the Dexter interactions in the emitting layers. All those efforts
maximally suppress the exciton loss, leading to TSF-OLEDs
with breakthrough EQE of 24.0% with extremely low efficiency
roll-off. Additionally, unlike previous work adopting extremely
low dopant concentration (<1 wt%) to reduce DET, high dopant
[
25]
also been employed to promote the device efficiencies. How-
ever, the contribution of triplet fusion can be negligible for the
devices here due to the low doping concentrations as well as
the suppressed formation of the triplet states on the anthracene
derivatives.
Additionally, other dyes were also demonstrated in the
device structure of Device Type III, showing EQEmax/PEmax
of 18.6%/54.0, 20.8%/61.9, and 22.7%/66.8 lm W− for PAD,
MePAD, and tBuPAD, respectively, as can be seen from Table 2
and Figure S8 (Supporting Information). Those performances are
much higher than those of the TSF-OLEDs with only PXZ-DPS
as the host, proving the superiority of the devices with multiple
energy funneling paths. The EQEmax–dye concentration relation-
ship was also examined in such device structures as revealed in
Figure 5f. The same conclusion can also be drawn that the larger
1
Adv. Mater. 2017, 1705250
1705250 (8 of 10)
© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.advmat.de
Table 2. Device performances of the TSF-OLEDs.
Voltage
Emitter
EQE
[%]
PE
[lm W−1]
CIEd)
[
V]
0 cd m−
2.56
2
1000 cd m−2 5000 cd m−2
1000 cd m−2 5000 cd m−2
1000 cd m−2 5000 cd m−2
1
Max
18.6
20.8
22.7
24.0
19.3
13.7
max
54.0
61.9
66.8
71.4
48.5
41.6
PADa)
3.22
3.16
3.12
3.11
3.55
3.07
4.13
4.0
18.5
20.7
22.5
23.8
18.7
13.3
17.0
19.5
21.5
22.6
17.4
12.9
51.2
58.9
65.8
68.5
47.2
41.0
36.9
44.0
49.8
52.3
36.5
33.7
(0.36,0.57)
(0.37,0.58)
(0.36,0.58)
(0.36,0.58)
(0.36,0.58)
(0.36,0.58)
MePADa)
2.54
tBuPADa)
PhtBuPADa)
PhtBuPADb)
PhtBuPADc)
2.57
3.91
3.90
4.25
3.70
2.55
2.94
2.51
a)Device structure III; b)Device structure I; c)Device structure II; d)The Commission Internationale de L’Eclairage (CIEs) were obtained under the voltage of 5 V.
N ,N ,N10,N10-tetrakis(4-(2-phenylpropan-2-yl)phenyl)anthracene-
9
9
concentration (3–5 wt%) can be used by using CFDs with steric
terminal units, reducing the device fabrication difficulties and
paving their way toward practical applications. We can presume
that such concept can be used to fabricate highly efficient full-
color fluorescent OLEDs. Furthermore, with a fluorescent laser
dye, light amplification in an organic solid-state film with the
aid of TADF-sensitized approach has also been observed. Our
results here may shed more potential about the TSF-OLEDs
technique in current-injection organic semiconductor lasers.
1
9
(
(
,10-diamine (PhtBuPAD): H-NMR (600 MHz, CDCl ): δ (ppm) = 8.50
m, 4H), 7.45 (m, 4H), 7.33 (m, 16H), 7.22 (m, 4H), 7.03 (m, 16H), 0.32
t, 24H). High-resolution mass spectrometry (m/z): [M+H] calculated
for C74
found for C74H N ): C (90.20%, 90.17%), H (6.96%, 6.97%), N (2.84%,
2.86%).
Theoretical Calculations: The geometrical and electronic properties
of dyes were performed with the Gaussian 09 program package. The
calculation was optimized by means of B3LYP (Becke three parameters
hybrid functional with Lee–Yang–Perdew correlation functionals) with the
3
+
H N , 984.54; found, 984.53. Elemental analysis (calculated,
68 2
68
2
[
26]
6
–31G(d) atomic basis set. The singlet and triplet states were calculated
using TD-DFT calculations with B3LYP/6–31g (d). The molecular orbitals
were visualized using Gaussview.
Experimental Section
Optical Characterization of Organic Thin Films: Ultraviolet–
visible absorption spectra were recorded using an Agilent 8453
spectrophotometer. PL quantum efficiency was measured by an absolute
PL quantum yield measurement system (C9920-02, Hamamatsu
Photonics) in air atmosphere with an excitation wavelength of 360nm.
The PL spectrum and PL transient decay curves of the films were
measured using a transient spectrometer (Edinburg FL920P).
Electrochemical Measurement: The CV measurement were performed
Materials: DPEPO, 4CzTPN-Ph, TAPC, and TCTA were purchased from
Xi′an Polymer Light Technology Co. Ltd. (Xi′an, Shanxi province, China).
PXZ-DPS, BPBiPA were synthesized according to reported procedures.
Synthesis of Fluorescent Dyes: The following general procedure for
the synthesis of CDFs was used. Diphenylamine (1.02 g, 6.0 mmol),
di-p-tolylamine (1.19 g, 6.0 mmol), bis(4-(tert-butyl)phenyl)amine
(
6
(
1.69 g, 6.0 mmol), or bis(4-(2-phenylpropan-2-yl)phenyl)amine (2.44 g,
.0 mmol) was added to a stirred solution of 9,10-dibromoanthracene
1.01 g, 3.0 mmol) and potassium carbonate (2.48 g, 18.0 mmol)
in toluene (30 mL). With stirring, a solution of palladium(II) acetate
40.4 mg, 0.18 mmol) and tri-tert-butylphosphine (133.6 mg, 0.66 mmol)
with a Potentiostat/Galvanostat Model 283 (Princeton Applied
Research) electrochemical workstation, using Pt as working electrode,
platinum wire as auxiliary electrode, and a Ag/AgCl electrode as
reference electrode standardized against ferrocene/ferrocenium. The
oxidation potentials were measured in dichloromethane (CH Cl )
(
in 30 mL of toluene was then added. The reaction mixture was stirred
and refluxed for 10h. The cooled mixture was partitioned between
chloroform and water. The organic layer was separated, and the aqueous
layer was extracted with chloroform. The combined organic layers were
washed with brine, dried over Mg SO , and concentrated in vacuo.
2
2
solution containing 0.1 mn-Bu NPF as supporting electrolyte at a scan
4
6
−
1
rate of 100 mV s
.
Device Fabrication and Measurement: Before device fabrication,
the ITO glass substrates were precleaned carefully. Then the sample
was transferred to the deposition system. The devices were prepared
2
4
Column chromatography of the residue solid (eluent: chloroform/
−5
hexane = 1/4) afforded the target compounds with the yield of over 80%.
in vacuum at a pressure of 5 × 10 Torr. The hole-injection and hole-
transporting material TAPC, exciton blocking material TCTA, and electron-
transporting material BPBiPA were thermally evaporated at a rate of
9
9
10 10
1
N ,N ,N ,N -tetraphenylanthracene-9,10-diamine
(PAD):
H-
NMR (600 MHz, CDCl ): δ (ppm) = 8.47 (dd, 4H), 7.42 (dd, 4H), 7.23
3
−1
(
(
t, 8H), 7.12 (d, 8H), 6.86 (t, 4H). High-resolution mass spectrometry
m/z): [M+H] calculated for C H N , 513.23; found, 513.24. Elemental
1.0 Å s . After the organic film deposition, 0.5 nm of LiF and 150 nm of
aluminum were thermally evaporated onto the organic surface. All of the
organic materials used were purified by a vacuum sublimation approach.
The forward-viewing electrical characteristics of the devices were
measured using a Keithley 2400 source meter. The electroluminescence
spectra and luminance of the devices were obtained on a PR650
spectrometer. All the device fabrication and characterization steps were
performed at room temperature under ambient laboratory conditions.
For measurement of the transient electroluminescence characteristics,
short-pulse excitation with a pulse width of 15 µs was generated using
an Agilent 8114A pulse generator. The amplitude of the pulse was 9 V,
and the baseline was −3 V. The period was 50 µs, the delayed time was
25 µs, and the duty cycle was 30%. The decay curves of the devices were
detected using an Edinburg FL920P transient spectrometer.
+
3
8
28
2
analysis (calculated, found for C38H N ): C (89.03%, 89.01%), H
28
2
(
5.51%, 5.50%), N (5.46%, 5.49%).
N ,N ,N ,N -tetra-p-tolylanthracene-9,10-diamine (MePAD): 1H-
9
9
10 10
NMR (600 MHz, CDCl ): δ (ppm) = 8.16 (dd, 4H), 7.32 (dd, 4H), 6.98
3
(
s, 16H), 2.24 (s, 12H). High-resolution mass spectrometry (m/z):
+
[
M+H] calculated for C42H36N , 569.29; found, 569.27. Elemental
2
analysis (calculated, found for C H N ): C (88.69%, 88.68%), H
4
2
36
2
(
6.38%, 6.37%), N (4.93%, 4.95%).
N ,N ,N ,N10-tetrakis(4-(tert-butyl)phenyl)anthracene-9,10-diamine
9
9
10
1
(
7
tBuPAD): H-NMR (600 MHz, CDCl ): δ (ppm) = 8.18 (s, 4H),
3
.33 (s, 4H), 7.17 (s, 8H), 7.01 (s, 8H), 1.26 (s, 36H). High-resolution
+
mass spectrometry (m/z): [M+H] calculated for C H N , 737.48;
found, 737.45. Elemental analysis (calculated, found for C54H60N ): C
5
4
60
2
2
[CCDC 1564531; 1564532; 1564534; 1564535 contain the
supplementary crystallographic data for this paper. These data can
(87.99%, 87.97%), H (8.21%, 8.22%), N (3.80%, 3.81%).
Adv. Mater. 2017, 1705250
1705250 (9 of 10)
© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.advmat.de
be obtained free of charge from The Cambridge Crystallographic Data
Centre via www.ccdc.cam.ac.uk/data_request/cif.]
d) Y. Yuan, Y. Hu, Y. X. Zhang, J. D. Lin, Y. K. Wang, Z. Q. Jiang,
L. S. Liao, S. T. Lee, Adv. Funct. Mater. 2017, 27, 1700986.
[
[
4] 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) G. Z. Xie, X. L. Li, D. J. Chen,
Z. H. Wang, X. Y. Cai, D. C. Chen, Y. C. Li, K. K. Liu, Y. Cao, S. J. Su,
Adv. Mater. 2016, 28, 181; c) G. H. Xie, J. J. Luo, M. L. Huang,
T. H. Chen, H. L. Wu, S. L. Gong, C. L. Yang, Adv. Mater. 2017, 29,
Supporting Information
Supporting Information is available from the Wiley Online Library or
from the author.
1
604223.
5] Z. Y. Yang, Z. Mao, Z. L. Xie, Y. Zhang, S. W. Liu, J. Zhao, J. R. Xu,
Z. G. Chi, M. P. Aldred, Chem. Soc. Rev. 2017, 46, 915.
Acknowledgements
[6] a) D. D. Zhang, L. Duan, D. Q. Zhang, Y. Qiu, J. Mater. Chem. C
014, 42, 8983; b) D. D. Zhang, P. C. Wei, D. Q. Zhang, L. Duan,
2
D.D.Z. and X.Z.S. contributed equally to this work. The authors would
like to thank the National Basic Research Program of China (Grant
No. 2015CB655002), the National Science Fund of China (Grant
Nos. 51525304 and U1601651), the National Key Basic Research
and Development Program of China (Grant Nos. 2016YFB0400702,
ACS Appl. Mater. Interfaces 2017, 9, 19040; c) D. D. Zhang,
C. G. Zhao, Y. G. Zhang, X. Z. Song, P. C. Wei, M. H. Cai, L. Duan,
ACS Appl. Mater. Interfaces 2017, 9, 4769.
[
7] R. Nagata, H. Nakanotani, C. Adachi, Adv. Mater. 2017, 29,
1604265.
[8] K. H. Kim, S. H. Lee, C. K. Moon, S. Y. Kim, Y. S. Park, J. H. Lee,
2
016YFB0401003) and the Open Fund of the State Key Laboratory on
Integrated Optoelectronics (Grant No. IOSKL2015KF13) for financial
support.
J. W. Lee, J. Huh, Y. M. You, J. J. Kim, Nat. Commun. 2014, 5, 4769.
[
9] a) H. Kim, K. Kim, C. Moon, J. J. Kim, Adv. Optical. Mater. 2017, 5,
1
1
600749; b) H. G. Kim, K. H. Kim, J. J. Kim, Adv. Mater. 2017, 29,
702159.
Conflict of Interest
[
[
[
10] H. Nakanotani, T. Furukawa, T. Hosokai, T. Hatakeyama, C. Adachi,
Adv. Optical. Mater. 2017, 5, 1700051.
11] C. Li, L. Duan, D. D. Zhang, Y. Qiu, ACS Appl. Mater. Interfaces
The authors declare no conflict of interest.
2
015, 7, 15154.
12] C. Murawshi, K. Leo, M. C. Gather, Adv. Mater. 2013, 25, 6801.
Keywords
[13] D. D. Zhang, L. Duan, C. Li, Y. L. Li, H. Y. Li, D. Q. Zhang, Y. Qiu,
Adv. Mater. 2014, 26, 5050.
fluorescent organic light-emitting diodes, high efficiency, multiple
energy-funneling paths, terminal substituents, thermally activated
delayed fluorescent sensitizers
[
14] H. Nakanotani, T. Higuchi, T. Furukawa, K. Masui, K. Morimoto,
M. Numata, H. Tanaka, Y. Sagara, T. Yasuda, C. Adachi, Nat.
Common. 2014, 5, 4016.
[
15] T. Furukawa, H. Nakanotani, M. Inoue, C. Adachi, Sci. Rep. 2015, 5,
8429.
Received: September 12, 2017
Revised: October 8, 2017
Published online:
[16] X. K. Liu, Z. Chen, C. J. Zheng, M. Chen, W. Liu, X. H. Zhang,
C. S. Lee, Adv. Mater. 2015, 27, 2025.
[
[
[
17] W. Song, I. Lee, J. Y. Lee, Adv. Mater. 2015, 27, 4358.
18] D. L. Dexter, J. Chem. Phys. 1953, 21, 836.
19] Y. F. Zhang, S. R. Forrest, Chem. Phys. Lett. 2013, 590, 106.
[
[
1] H. Uoyama, K. Goushi, K. Shizu, H. Nomura, C. Adachi, Nature
012, 492, 7428.
2] a) D. Di, A. S. Romanov, L. Yang, J. M. Richter, J. H. Rivett, S. Jones,
T. Thomas, M. A. Jalebi, R. Friend, M. Linnolahti, M. Bochmann,
D. Credgington, Science 2017, 356, 159; b) D. D. Zhang,
M. H. Cai, Y. G. Zhang, D. Q. Zhang, L. Duan, Mater. Horiz. 2016,
[20] M. A. Baldo, M. E. Thompson, S. R. Forrest, Nature 2000, 403, 750.
[21] N. Aizawa, S. Shikita, T. Yasuda, Chem. Mater. 2017, 29, 7014.
[22] Y. Tao, K. Yuan, T. Chen, P. Xu, H. H. Li, R. F. Chen, C. Zheng,
L. Zhang, W. Huang, Adv. Mater. 2014, 26, 7931.
[23] Q. S. Zhang, B. Li, S. P. Huang, H. Nomura, H. Tanaka, C. Adachi,
Nat. Photonics 2014, 8, 326.
2
3
, 145; c) D. R. Lee, M. Kim, S. K. Jeon, S. Hwang, C. W. Lee,
[24] a) H. Sasabe, H. Nakanishi, Y. Waranabe, S. Yano, M. Hirasawa,
Y. J. Pu, J. Kido, Adv. Mater. 2013, 23, 5550; b) D. D. Zhang, J. Qiao,
D. Q. Zhang, L. Duan, Adv. Mater. 2017, 29, 1702847.
J. Y. Lee, Adv. Mater. 2015, 27, 5861; d) D. D. Zhang, M. H. Cai,
Z. Y. Bin, Y. G. Zhang, D. Q. Zhang, L. Duan, Chem. Sci. 2016, 7,
3
355.
[25] a) C.-J. Chiang, A. Kimyonok, M. Etherington, G. Griffiths, V. Jankus,
F. Turksoy, A. P. Monkman, Adv. Funct. Mater. 2013, 23, 739;
b) W. C. Chen, C. S. Lee, Q. X. Tong, J. Mater. Chem. C 2015, 3, 10957.
[26] H. Nakanotani, T. Furukawa, C. Adachi, Adv. Opt. Mater. 2015, 3,
13812.
[
3] a) M. K. Etherington, J. Gibson, H. F. Higginbotham, T. J. Penfold,
A. P. Monkman, Nat. Commun. 2016, 7, 13680; b) H. A. Al Attar,
A. P. Monkman, Adv. Mater. 2016, 28, 8014; c) J. Y. Lee, N. Y. Aizawa,
M. Numata, C. Adachi, T. Yasuda, Adv. Mater. 2017, 29, 1604856;
Adv. Mater. 2017, 1705250
1705250 (10 of 10)
© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim