Submitted Manuscript: Confidential 1 February 2014
Overcoming the Electroluminescence Efficiency Limitations of Perovskite
Light-Emitting Diodes
Authors: Himchan Cho
1†
, Su-Hun Jeong
1†
, Min-Ho Park
1†
, Young-Hoon Kim
1
, Christoph
Wolf
1
, Chang-Lyoul Lee
2
, Jin Hyuck Heo
3
, Aditya Sadhanala
4
, NoSoung Myoung
2
, Seunghyup
Yoo
5
, Sang Hyuk Im
3
, Richard H. Friend
4
, Tae-Woo Lee
1,6
*
Affiliation:
1
Department of Materials Science and Engineering, Pohang University of Science and
Technology (POSTECH), 77 Cheongam-Ro, Pohang, Gyungbuk 790-784, Republic of Korea.
2
Advanced Photonics Research Institute (APRI), Gwangju Institute of Science & Technology
(GIST), 1 Oryong-dong, Buk-gu, Gwangju, 500-712, Republic of Korea.
3
Department of Chemical Engineering, College of Engineering, Kyung Hee University, 1
Seochon-dong, Giheung-gu, Youngin-si, Gyeonggi-do 446-701, Republic of Korea.
4
Cavendish Laboratory, University of Cambridge, JJ Thomson Avenue, Cambridge CB3 0HE,
UK.
5
Department of Electrical Engineering, Korea Advanced Institute of Science and Technology
(KAIST), 373-1 Guseong-dong, Daejeon 305-701.
6
Department of Chemical Engineering, Division of Advanced Materials Science, School of
Environmental Science and Engineering, Pohang University of Science and Technology
(POSTECH), 77 Cheongam-Ro, Nam-Gu, Pohang, Gyungbuk 790-784, Republic of Korea.
*Corresponding author. E-mail: twlee@postech.ac.kr, taewlees@gmail.com
†: These authors contributed equally to this work.
Abstract: Organic-inorganic hybrid perovskites are emerging low-cost emitters with very high
color purity, but their low luminescent efficiency is a critical drawback. We boosted the current
efficiency (CE) of perovskite LEDs with a simple bilayer structure to 42.9 candela per ampere,
similar to CE of phosphorescent OLEDs, with two modifications. We prevented the formation of
metallic Pb atoms that cause significant exciton quenching through small increase in
methylammonium bromide (MABr) molar proportion, and we spatially confined the exciton in
uniform MAPbBr
3
nanograins (average diameter = 99.7 nanometers) formed by a nanocrystal
pinning process and concomitant reduction of exciton diffusion length to 67 nanometers. These
changes caused substantial increases in steady-state photoluminescence intensity and lifetime of
MAPbBr
3
nanograin layers.
One Sentence Summary: Ultrahigh-efficiency organic/inorganic hybrid perovskite light-
emitting diodes of 42.9 cd A
-1
was achieved using stoichiometric tuning and nanograin
engineering.
Main Text: Organic-inorganic hybrid perovskites (OIPs) have recently established as important
class of materials in photovoltaic devices with rapid progress in increasing their power
conversion efficiency (1-5). OIPs are being emerged also as promising light emitters because
they can provide very high color purity (full width at half maximum ~ 20 nm) irrespective of the
crystal size, unlike conventional inorganic quantum dots, because their intrinsic crystal structure
is similar to a multiple quantum well (6, 7). Also, OIPs have low material cost and simply-
tunable band gap with a reasonable ionization energy (IE) comparable to common hole injection
materials (7-11). Thus, OIPs are attractive materials as alternative emitters that can overcome the
disadvantages of organic light-emitting diodes (OLEDs) (e.g., complex synthesis, high cost, and
poor color purity) and inorganic quantum dot LEDs (e.g., complex synthesis, high cost and high
IE).
Bright electroluminescence (EL) (> 100 cd m
-2
) at room temperature from perovskite
light-emitting diodes (PeLEDs) with methylammonium lead halides (MAPbX
3
, where X is I, Br
or Cl) emission layer was demonstrated recently (6,7,12-18). As an emission layer, MAPbBr
3
has higher air stability (7, 19) and exciton binding energy (76 or 150 meV) than does MAPbI
3
(30 or 50 meV) (20, 21). However, PeLEDs have much lower current efficiency (CE) at room
temperature than do OLEDs or quantum dot LEDs. Existing methods have not overcome the
substantial luminescence quenching in MAPbX
3
caused by facile thermal ionization of excitons
generated in the OIP layer, which has a low exciton binding energy. Spin-coating of MAPbBr
3
solution creates a rough, non-uniform surface with many cuboids with large grain size (22),
which leads to a substantial leakage current and large exciton diffusion length L
D
that reduces CE
in PeLEDs. To improve the CE of PeLEDs, the OIP grain size must be decreased, and OIP films
should be flat and uniform. Smaller grains can spatially limit L
D
of excitons or charge carriers
and reduce the possibility of exciton dissociation into carriers. This fabrication goal differs from
that of the OIP layers in solar cells, which should be dense films with large grain size to achieve
good exciton diffusion and dissociation. Thus, processes designed to achieve uniform OIP film
morphology with large grain size in solar cells such as solvent dripping (23, 24) are not
applicable to PeLEDs, which require small L
D
.
Here, we report a systematic approach for achieving highly bright and efficient green
PeLEDs with CE = 42.9 cd A
-1
and external quantum efficiency (EQE) = 8.53 % even in a
simplified bilayer structure. These high efficiencies represent >20,000-fold increase compared
with that of the control devices and are higher than the best EQE of a previous report regarding
visible PeLEDs using OIP films by factors of > 10.6 (Table S1, Fig. S1) (15). The high
efficiency PeLEDs were constructed based on effective management of exciton quenching by a
modified MAPbBr
3
emission layer that was achieved using (i) fine and controllable
stoichiometry modification and (ii) optimized nanograin engineering by nanocrystal pinning
(NCP) (Fig. S2). Furthermore, we demonstrated a flexible PeLED using a self-organized
conducting polymer (SOCP) anode and the first large-area PeLED (2 cm by 2 cm pixel).
A fundamental problem that must be solved to achieve high CE in PeLEDs is minimizing
the presence of metallic Pb atoms in MAPbBr
3
that limits the efficiency of PeLEDs. Metallic Pb
atoms can emerge in MAPbBr
3
even if MABr and PbBr
2
are mixed with 1:1 (mol:mol) ratios
because of the decomposition of MABr or incomplete reaction between MABr and PbBr
2
(7, 19).
Excess Pb atoms degrade luminescence by increasing the nonradiative decay rate and decreasing
the radiative decay rate (25). Prevention of the formation of metallic Pb atoms was achieved by
finely increasing the molar proportion of MABr by 2 to 7 % in MAPbBr
3
solution (Fig. S2A).
Use of excess MABr suppressed exciton quenching and reduced hole-injection barrier from
SOCP layers (Table S2) to MAPbBr
3
layers with decreased IE and greatly increased steady-state
photoluminescence (PL) intensity and PL lifetime of MAPbBr
3
films. We propose that radiative
decay of the PL process in MAPbBr
3
nanograins originates from shallow-trap-assisted radiative
recombination at grain boundaries and radiative recombination inside the grains. Second, the CE
in PeLEDs can be increased by decreasing MAPbBr
3
grain sizes, which improves uniformity and
coverage of MAPbBr
3
nanograin layers and radiative recombination by confining the excitons in
the nanograins (leading to small L
D
). An optimized NCP process (Fig. S3) helped change in the
morphology of MAPbBr
3
layers from scattered micrometer-sized cuboids to well-packed
nanograins with uniform coverage, which greatly reduced leakage current and increased CE.
We fabricated MAPbBr
3
films by spin-coating using stoichiometrically-modified
perovskite solutions on prepared glass/SOCPs or silicon wafer/SOCPs substrates later used in
devices (Fig. 1, A and B), and then characterized the films’ morphologies and optoelectronic
properties. The solutions had different molar ratios of MABr to PbBr
2
(MABr:PbBr
2
= 1.05:1,
1:1, or 1:1.05). To achieve uniform surface coverage and reduced grain size, NCP was used
instead of normal spin-coating (Fig. S3). This process washed out the “good” solvents
[dimethylformamide or dimethyl sulfoxide (DMSO)], and causes pinning of NCs by inducing
fast crystallization. Chloroform was chosen as the solvent for NCP because a highly volatile
nonpolar solvent is suitable to reduce the size and increase the uniformity of MAPbBr
3
grains by
reducing solvent evaporation time. In addition, to further reduce grain size, we devised additive-
based NCP (A-NCP) which uses an organic small molecule, 2,2',2"-(1,3,5-benzinetriyl)-tris(1-
phenyl-1-H-benzimidazole) (TPBI), as an additive to chloroform, whereas pure chloroform is
used in solvent-based NCP (S-NCP).
Use of NCP affected film morphology (Fig. 2). Without NCP, micrometer-sized
MAPbBr
3
cuboids were scattered on the SOCP layer (Fig. 2A). They were only interconnected
with a few other cuboids, and so a large amount of space remained uncovered. This high surface
roughness and the formation of pinholes in OIP films result in formation of bad interface with
the electron transport layer and electrical shunt paths, and thus severely limit CE in PeLEDs. In
contrast, when NCP was used, perfect surface coverage was obtained, and the MAPbBr
3
crystal
morphology changed to a well-packed assembly of tiny grains ranging from 100 to 250 nm (Fig.
2, B to E, Fig. S4). MAPbBr
3
grain size was little affected by the stoichiometric modification of
MAPbBr
3
solutions (Fig. 2, B to D, Fig. S4, A to C). Furthermore, MAPbBr
3
grain size was
further reduced to 50-150 nm (average = 99.7 nm) by A-NCP (Fig. 2E, Fig. S4D). This reduction
can be attributed to hindrance of crystal growth by TPBI molecules during crystal pinning. The
thickness of MAPbBr
3
layer was ~400 nm (Fig. 1B).
The crystal structures of MAPbBr
3
films were analyzed by measuring x-ray diffraction
(XRD) patterns (Fig. 2F, Fig. S5, Table S3). The XRD patterns of MAPbBr
3
films (1:1) exhibit
peaks at 15.02°, 21.3°, 30.28°, 33.92°, 37.24°, 43.28° and 46.00° that can be assigned to (100),
(110), (200), (210), (211), (220) and (300) planes respectively, by using Bragg’s law to convert
the peak positions to interplanar spacings (Fig. 2F). The lattice parameter is in accordance with a
previous report (19), and demonstrates that MAPbBr
3
films had a stable cubic Pm3
m phase.
Using the Scherrer equation, the crystallite size was calculated to be 23.5 ± 2.0 nm, and the
variation with stoichiometric change was not large (Table S3). Because the crystallite sizes were
much smaller than the apparent grain sizes (Fig. 2, A to E), we conclude that all grains consisted
of many crystallites. The stoichiometric changes had very little effect on the peak positions (Fig.
S5A). Furthermore, A-NCP did not change the peak positions when compared to S-NCP (Fig.
S5); this stability in positions indicates that the stoichiometric changes of MAPbBr
3
solution and
the use of TPBI additive did not affect the crystal structure of MAPbBr
3
films.
To study chemical changes in the MAPbBr
3
layers fabricated using perovskite solutions
with different stoichiometries, x-ray photoelectron spectroscopy (XPS) was conducted. The
survey spectra showed strong peaks of Br (~ 68 eV), Pb (~ 138 and 143 eV), C (~ 285 eV) and N
(~ 413 eV); these results agree with values in previous reports (Fig. S6A) (7, 26-28). Systematic
deconvolution of Pb4f, Br3d and N1s spectra into summations of Gaussian-Lorentzian curves
revealed the nature of chemical bonds in MAPbBr
3
(Fig. S6, B to D, Fig. S7). The gradual
increase of MABr molar proportion in the films was confirmed by observing the gradual increase
in N1s peak intensities as MABr:PbBr
2
increased from 1:1.05 to 1.05:1 (Fig. S7, C and D) and
the gradual decrease in Br:Pb atomic ratio (Supplementary Text F). In the Pb4f spectra (Fig. S6,
B to F), large peaks were observed at ~138.8 and ~143.6 eV (caused by the spin orbit split) that
correspond to Pb4f
7/2
and Pb4f
5/2
levels, respectively (26-28). Each of these peaks was associated
with a smaller peak that was shifted to 1.8-eV lower binding energy; these small peaks can be
assigned to metallic Pb (26-28). The height of peaks that represent metallic Pb decreased as
MABr:PbBr
2
increased from 1:1.05 to 1:1 (Fig. S6, E and F); this peak was absent in the film
with MABr:PbBr
2
= 1.05:1 (Fig. S6F). This trend indicates that the presence of metallic Pb
atoms on the films was successfully prevented by fine stoichiometry control. In contrast, the high
peak intensity of the metallic Pb peak in the films with MABr:PbBr
2
= 1:1 and 1:1.05 suggests
that numerous metallic Pb atoms were formed on the film surfaces.
We measured the work functions (WFs) and IEs of the MAPbBr
3
films using ultraviolet
photoelectron spectroscopy (UPS) (Fig. S8). The WFs were obtained by subtracting the energies
at secondary cut-offs of the UPS spectra from the UV radiation energy of 21.2 eV when a Fermi
level of 0 eV was the common reference for all energies. The IEs were determined by adding the
WF (Fig. S8A) to the energy offset between WFs and IEs of MAPbBr
3
(Fig. S8B) (29). The IE
gradually decreased with increasing MABr molar proportion from 6.01 eV in the film with
MABr:PbBr
2
= 1:1.05 to 5.86 eV in the film with MABr:PbBr
2
= 1.1:1 (Fig. 1C, Table S4). The
gradual decrease in IEs with decreasing PbBr
2
molar proportion can be understood based on the
IE being greater in PbBr
2
than in MAPbBr
3
(30). In PeLEDs, this decrease can help alleviate
hole-injection barriers from SOCP layers to MAPbBr
3
layers (Fig. 1C).
The luminescent properties of the MAPbBr
3
films were investigated by steady-state PL
measurement (Fig. 3A). The measurement was conducted using a spectrofluorometer with
excitation from monochromatic light with wavelength of 405 nm (xenon lamp). The MAPbBr
3
films fabricated from MABr:PbBr
2
= 1.05:1 had a ~ 5.8 times increase in PL intensity (Fig. 3A)
compared with 1:1 films, and had much higher PL quantum efficiency (PLQE; 36 % vs 3 %). In
addition, the reduction of grain size with A-NCP vs S-NCP increased the PL intensity by ~2.8
times. The PL intensity of the films with MABr:PbBr
2
= 1:1.05 was greater than in those with
MABr:PbBr
2
= 1:1, although the PbBr
2
molar proportion had increased in the former. We
suspect that this departure from the expected trend is due to PbBr
2
-induced surface passivation of
the film; this process reduces nonradiative recombination at the trap sites (31).
To understand the kinetics of excitons and free carriers in MAPbBr
3
films and how the
presence of metallic Pb atoms affects the PL lifetime, time-correlated single-photon counting
(TCSPC) measurements were conducted (Fig. 3B). The PL decay curves were fitted using a bi-
exponential decay model, in which the PL lifetime is considered as the summation of fast and
slow decay components that give short lifetime
1
and long lifetime
2
, respectively. To
investigate the quality of quenching sites, we prepared the layers (MABr:PbBr
2
= 1.05:1) with
and without sealing with a 50-nm-thick poly(methyl methacrylate) (PMMA) layer. The fraction
f
2
of
2
decreased from 91 % to 77 % in the film without sealing (Table S5). Oxygen and
moisture can diffuse quickly into grain boundaries when the top PMMA layer is not used;
oxygen or moisture at grain boundaries provides quenching sites. The fast decay is related to
trap-assisted recombination at grain boundaries, whereas the slow decay is related to radiative
recombination inside the grains (Fig. S9) (32).
This proposition was supported by analyzing the change in
and f of MAPbBr
3
films
with varying stoichiometric ratio. As MABr:PbBr
2
increased from 1:1 to 1.05:1, the average
lifetime
avg
gradually increased from 12.1 to 51.0 ns (Table S5). The short
avg
(12.1 ns) in the
film with MABr:PbBr
2
= 1:1 originated from the significant reduction of
2
. This implies that
uncoordinated metallic Pb atoms at grain boundaries inhibit radiative recombination and cause
strong nonradiative recombination (Fig. S9). The MAPbBr
3
films fabricated using PbBr
2
-rich
perovskite solution (MABr:PbBr
2
= 1:1.05) had a longer lifetime than films with MABr:PbBr
2
=
1:1, possibly through PbBr
2
-induced surface passivation (31). We calculated the average L
D
using a model similar to that in previous reports
(Fig. S10) (33). The films (MABr:PbBr
2
=
1.05:1) underneath a PMMA layer exhibited a much smaller L
D
(67 nm) than those previously
reported (>1 m)
(33, 34). We attribute this reduction in L
D
to the reduced grain sizes in which
excitons are spatially stronger confined, thereby reducing dissociation and enhancing radiative
recombination; this compensates the plausible adverse effect of larger grain boundary area (6).
The PeLED fabricated without using NCP showed poor luminous characteristics
(maximum CE = 2.03×10
-3
cd A
-1
) mainly due to high leakage current (Fig. S11). In contrast,
maximum CE was significantly increased (0.183 cd A
-1
) when a full-coverage uniform MAPbBr
3
nanograin layer with decreased grain size was achieved using S-NCP (Fig. 4, A and B, Table 1).
The maximum CE was boosted up to 21.4 cd A
-1
in the PeLEDs fabricated with perovskite
solutions with excess MABr (1.07:1, 1.05:1, 1.03:1 and 1.02:1) (Fig. 4A, Table 1). As
MABr:PbBr
2
increased from 1:1 to 1.05:1, the maximum CE varied from 0.183 to 21.4 cd A
-1
.
Without stoichiometric modifications of MAPbBr
3
to avoid metallic Pb atoms (molar ratios 1:1
and 1:1.05), the achieved maximum CEs was only 0.183 and 4.87×10
-2
cd A
-1
, respectively.
We further increased CE of PeLEDs by using A-NCP. The PeLEDs based on A-NCP had
a maximum CE of 42.9 cd A
-1
(Fig. 4, C and D, Table 1), which represents an EQE of 8.53 %
when the angular emission profile is considered (Fig. S12). The EL spectra of PeLEDs were very
narrow; full width at half maximum was ~20 nm for all spectra; this high color purity of OIP
emitters show great potential when used in displays (Fig. 4E). A pixel of the PeLED of based on
MABr:PbBr
2
= 1.05:1 exhibited strong green light emission (Fig. S13A). Furthermore, the
proposed processes and materials used therein are compatible with flexible and large-area
devices; a high-brightness flexible PeLED (Fig. 4, F and G) and a large-area (2 cm by 2 cm
pixel) PeLED (Fig. S13B) were fabricated. Our study reduces the technical gap between PeLEDs
and OLEDs or quantum dot LEDs, and is a big leap towards the development of efficient next-
generation emitters with high color purity and low fabrication cost based on perovskites.
References and Notes:
1. W. S. Yang, J. H. Noh, N. J. Jeon, Y. C. Kim, S. Ryu, J. Seo, S. I. Seok, High-performance
photovoltaic perovskite layers fabricated through intramolecular exchange. Science 348,
1234-1237 (2015).
2. M. Liu, M. B. Johnston, H. J. Snaith, Efficient planar heterojunction perovskite solar cells by
vapour deposition. Nature 501, 395-398 (2013).
3. N. J. Jeon, J. H. Noh, W. S. Yang, Y. C. Kim, S. Ryu, J. Seo, S. I. Seok, Compositional
