Review Article

https://doi.org/10.1038/s41563-020-0784-7

Metal halide perovskites for light-emitting diodes

Xiao-Ke Liu   1,5, Weidong Xu   1,5, Sai Bai   1,5, Yizheng Jin   2, Jianpu Wang   3, Richard H. Friend   4
and Feng Gao   1 ✉

Metal halide perovskites have shown promising optoelectronic properties suitable for light-emitting applications. The develop-
ment of perovskite light-emitting diodes (PeLEDs) has progressed rapidly over the past several years, reaching high external
quantum efficiencies of over 20%. In this Review, we focus on the key requirements for high-performance PeLEDs, highlight
recent advances on materials and devices, and emphasize the importance of reliable characterization of PeLEDs. We discuss
possible approaches to improve the performance of blue and red PeLEDs, increase the long-term operational stability and reduce
toxicity hazards. We also provide an overview of the application space made possible by recent developments in high-efficiency
PeLEDs.

M
etal halide perovskites, which have led to great advances this Review by discussing strategies for achieving highly emissive
in photovoltaic devices, have also proved to be promis- perovskite emitters through effective control of both radiative and
ing candidates for light-emitting diodes (LEDs)1. They non-radiative recombination processes. We then focus on critical
have shown excellent optoelectronic properties suitable for LEDs, device prerequisites to achieve high-efficiency PeLEDs, highlight-
such as high photoluminescence quantum yields (PLQYs), widely ing similarities and differences between the device engineering of
tunable bandgap, narrow emission width and high charge-carrier PeLEDs and other LEDs, and emphasizing the importance of reli-
mobility2. Although early reports on perovskite LEDs (PeLEDs) able device characterization for this emerging technology. We also
date back to the 1990s3,4, room-temperature PeLEDs were not overview specific challenges and possible solutions in realizing
demonstrated until 20145. Since then, benefiting from established high-performance red and blue PeLEDs, improving operational sta-
experience in both perovskite materials and solution-processed bility and reducing the toxicity of these devices. Finally, we discuss
optoelectronic devices, the community has quickly boosted the future applications of perovskite emitters.
external quantum efficiencies (EQEs) of PeLEDs to reach more
than 20% (Box 1)6–10. Towards highly emissive metal halide perovskites
The rapid development of PeLEDs could lead to a new gen- From a material point of view, the prerequisite for achieving highly
eration of low-cost and high-performance LEDs for applications efficient LEDs is to use light emitters with high PLQYs. For any
including displays, lighting and optical communications2,11,12. luminescent material, its PLQY is defined as the number of photons
Compared with other emitters used in commercial devices, such as emitted divided by the number of photons absorbed, and can also
III–V inorganic semiconductors, organic emitters and conventional be expressed by the ratio between its radiative recombination rates
colloidal quantum dots (QDs), perovskites have several promising and the overall recombination rates:
characteristics. Specifically, perovskite emitters with high PLQYs P
can be straightforwardly fabricated from low-cost precursor solu- Rr
PLQY ¼ P P ð1Þ
tions, potentially reducing manufacturing costs. Synthesis of colloi- Rr þ Rnr
dal perovskite nanocrystals (PNCs) is also simplified, as PNCs can
reach near-unity PLQYs without delicate shell passivation, owing where Rr and Rnr are the radiative and non-radiative recombination
to their unique defect-tolerant nature13,14. Furthermore, the opto- rates, respectively. Equation (1) indicates that the general principle
electronic properties of perovskite emitters can be readily tailored for achieving a highly efficient light emitter is to enhance its radiative
by engineering composition and dimensionality, enabling continu- recombination rate(s) and to suppress the non-radiative recombina-
ously tunable light emission from violet to near-infrared (NIR) tion rate(s). In metal halide perovskites, the radiative recombination
regions2,13. In addition, light emission from perovskites shows can be bimolecular or excitonic, whereas trap-assisted monomo-
narrow linewidths (<100 meV), resulting in high colour purity: lecular and Auger recombination are non-radiative processes17,18.
for example, the photoluminescence full-width at half-maximum In this section, we discuss strategies for achieving highly efficient
(FWHM) is around 12, 20 and 40 nm for CsPbCl3, CsPbBr3 and perovskite emitters through effective control over both the radiative
CsPbI3 PNCs, respectively15. The colour gamut of displays made and non-radiative recombination processes.
by PNCs can cover up to 140% of the National Television System
Committee standard and is close to 100% of the new International Enhancing the radiative bimolecular recombination rate.
Telecommunication Union Rec. 2020 standard, making them prom- Three-dimensional (3D) perovskite emitters generally possess small
ising for ultra-high-definition displays15,16. exciton binding energies close to the room-temperature thermal
Yet the deployment of perovskite emitters in commercial devices energy, indicating that the excitons in perovskites readily dissoci-
presents barriers that need to be carefully addressed. We open ate into free charge carriers18. Because bimolecular recombination

Department of Physics, Chemistry and Biology (IFM), Linköping University, Linköping, Sweden. 2Center for Chemistry of High-Performance and Novel
1

Materials, State Key Laboratory of Silicon Materials, and Department of Chemistry, Zhejiang University, Hangzhou, China. 3Key Laboratory of Flexible
Electronics (KLOFE), Institute of Advanced Materials (IAM), Nanjing Tech University (NanjingTech), Nanjing, China. 4Cavendish Laboratory, University of
Cambridge, Cambridge, UK. 5These authors contributed equally: Xiao-Ke Liu, Weidong Xu, Sai Bai. ✉e-mail: [email protected]

Nature Materials | www.nature.com/naturematerials

Review Article NaTuRE MaTERIaLs

Box 1 | The development of perovskite emitters and PeLEDs

Metal halide perovskites refer to a large family of crystalline ma- emitted broad electroluminescence from the organic moieties
terials with structures similar to that of the natural mineral cal- in the material structure, attracting limited attention. PeLEDs
cium titanate, which was discovered by Gustav Rose in 1839 and demonstrating room-temperature electroluminescence with
named after Russian mineralogist Lev Perovski. 3D metal halide narrow emission were developed in 20145, setting a new start for the
perovskites can be described with a general chemical formula of evolution of this research field (panel b of the figure). Early-stage
ABX3, where A is a monovalent cation, B is a divalent metal cation development of PeLEDs mostly involved 3D perovskites, with a
and X is a halogen anion that bonds with metal cations, forming focus on uniform film morphology and reduced grain sizes20,21.
corner-sharing [BX6]4− octahedra cavities embedding the A cati- More recently, it was found that discontinuous perovskite films
ons. Limited by the Goldschmidt tolerance factor of the perovs- could increase light outcoupling and improve efficiency, as long
kite phase, A-site cations in such 3D structures are mostly caesium as leakage currents could be effectively suppressed8. Along with
(Cs+), methylammonium (MA, CH3NH3+) or formamidinium the development of 3D perovskite emitters, perovskite films made
(FA, CH(NH2)2+) cations with suitable sizes. of mixtures of 2D/quasi-2D/3D phases (hereafter referred to as
From the 3D structures, nanostructures or even mixed-dimensional perovskites) have also attracted attention for
quantum-confined structures can be derived by decreasing the high-efficiency PeLEDs26,27. Such perovskites demonstrate much
crystal size or reducing the structural dimensionality (panel a of enhanced confinement of charge carriers as well as effective
the figure). Through colloidal approaches, perovskite nanocrystals radiative recombination within the lowest-energy perovskite phase.
have been successfully synthesized with different sizes and shapes, In addition to the polycrystalline thin films deposited directly
including nanocubes, nanoplatelets, nanosheets, nanowires and from precursor solutions, assemblies of colloidal PNCs have been
quantum dots13. Alternatively, reduced structural dimensionality used in PeLEDs. Since the PNC-based PeLEDs reported in 201548,
— namely 2D or quasi-2D layered perovskites such as Ruddlesden– intensive efforts have been dedicated to optimizing material
Popper and Dion–Jacobson phases — has been obtained by compositions (for instance through A-site substitution and metal ion
incorporating large-sized organic cations into the perovskite doping), purification protocols and surface ligands13. These efforts
frameworks. These materials share the formula LmAn−1BnX3n+1, have led to controllable luminescent properties of PNCs and rapid
where L is typically a large-sized monovalent (m = 2) or divalent performance improvement of the ensuing PNC-based PeLEDs9,13,36.
(m = 1) cation, and n is the number of inorganic octahedra sheets In all these different perovskite emitters, it is of critical
between the large-sized organic spacers, with the case of n = ∞ importance to effectively eliminate non-radiative recombination
representing the 3D structures. The large compositional and pathways for achieving high-performance PeLEDs. A wide range of
dimensional diversity obtained by combining different cations effective strategies have been developed, including quasi-core–shell
and anions in the crystal structures offers wide tunability of the structures7, perovskite–polymer heterostructures10 and rational
optoelectronic properties of perovskite emitters. hydrogen bonding manipulation6. These efforts have pushed the
Early reports of PeLEDs date back to the 1990s3,4. The EQEs of PeLEDs over 20%, approaching those of organic LEDs
devices at that time either operated at cryogenic temperatures or (OLEDs) and conventional QD-based LEDs (QLEDs).

a Decreasing crystal size b

25
3D perovskites
Mixed-dimensional perovskites
Perovskite nanocrystals
20

15
EQE (%)

Increasing binding energy

10

A+ X– [BX6]4–

3D structure (n = ∞) 0
2014 2015 2016 2017 2018 2019 2020
Year
Reducing structural dimensionality

The development of perovskite emitters and PeLEDs. a, Material diversity of metal halide perovskite emitters. Beyond 3D structures, nanostructures
(including quantum-confined structures) can be obtained by decreasing the crystal size or reducing the structural dimensionality, resulting in
increased electron–hole (exciton) binding energy. b, Development of PeLEDs with different emitters and different light emissions; the red, green and
blue symbols denote red/near-infrared, green and blue PeLEDs, respectively. The data used for this figure are listed in Supplementary Table 1.

rates are usually slow in typical 3D perovskites, free carriers can in 3D perovskites is strongly dependent on charge-carrier density,
be easily captured by non-radiative recombination centres at low and can be described by equation (2)18,19,
charge-carrier densities (as illustrated in Fig. 1a). At relatively high
charge-carrier densities, traps are saturated, and radiative recombi- dN
� ¼ AN þ BN 2 þ CN 3 ð2Þ
nation can be dominant. The charge-carrier recombination kinetics dt

Nature Materials | www.nature.com/naturematerials

NaTuRE MaTERIaLs Review Article
a 3D perovskite b c
1.2 1.2
A = 106 s–1; C = 10–28 cm6 s–1 B = 10–10 cm3 s–1; C = 10–28 cm6 s–1
+ 1.0 1.0
B (cm3 s–1) A (s–1)
5 × 10–11 1 × 104
0.8 0.8
1 × 10–10 1 × 105
hv
–

PLQY

PLQY
–+ 0.6 5 × 10–10 0.6 1 × 106
–9
5 × 10 1 × 107
0.4 0.4
–
0.2 0.2
+
0.0 0.0
– Electron + Hole Trap 1011 1012 1013 1014 1015 1016 1017 1018 1019 1011 1012 1013 1014 1015 1016 1017 1018 1019
Radiative recombination Excitation density (cm–3) Excitation density (cm–3)

d Quantum-confined perovskite e
1.2
k = 108 s–1; k2 = 10–8 cm3 s–1
ktrap (s–1)
1.0
1 × 106
0.8 1 × 107
5 × 107
–

PLQY
hv
+ 0.6 1 × 108

0.4

0.2

0.0
– + Exciton Trap 1011 1012 1013 1014 1015 1016 1017 1018 1019

Radiative recombination Excitation density (cm–3)

Fig. 1 | Charge-carrier recombination kinetics. a, A schematic illustration of charge-carrier recombination in 3D perovskites. b,c, Related PLQY-excitation
density characteristics with (b) varying bimolecular recombination coefficients and (c) varying non-radiative monomolecular recombination coefficients.
d, A schematic illustration of charge-carrier recombination in quantum-confined perovskites. e, Related PLQY-excitation density plots with varying
trap-assisted non-radiative recombination rate constants. Figure adapted with permission from: a,d, ref. 17, under a Creative Commons licence
(http://creativecommons.org/licenses/by/4.0/).

where N is the charge-carrier density, t is time, A is the trap-assisted Using radiative excitonic recombination. In addition to enhanc-
non-radiative monomolecular recombination coefficient, B is the radi- ing radiative bimolecular recombination, another strategy to
ative bimolecular recombination coefficient and C is the three-body increase PLQYs is to make use of radiative monomolecular recom-
Auger (non-radiative) recombination coefficient. The PLQYs of 3D bination pathways. This could be readily achieved through devel-
perovskite emitters can hence be quantified by equation (3): oping quantum-confined perovskites such as 2D perovskites,
quasi-2D perovskites or PNCs (Box 1). In contrast to free
BN charge carriers generated in 3D perovskites, stable excitons can
PLQY ¼ ð3Þ form in these quantum-confined perovskites at room temperature
A þ BN þ CN 2
(as illustrated in Fig. 1d) owing to their high exciton binding ener-
Figure 1b and c illustrate the excitation-density-dependent gies, up to hundreds of millielectronvolts17. This feature results
PLQYs of 3D perovskites with different recombination coefficients. in fast radiative monomolecular recombination rates, and hence
These calculations indicate the importance of increasing B (Fig. 1b) recombination dynamics in this case is less dependent on charge-
and decreasing A values (Fig. 1c), so that the PLQYs can approach carrier densities.
unity in a wide range of excitation density relevant for practical LED In quantum-confined perovskites, first-order radiative excitonic
operations. recombination competes with first-order non-radiative recombina-
Although bimolecular recombination rates in typical 3D tion through charge- or exciton-trapping, and with second-order
perovskites are usually slow under the LED operating conditions19, non-radiative Auger recombination of exciton-electron/hole or
they can be efficiently enhanced by spatially confining the charge exciton–exciton annihilation24. Therefore, charge-carrier recombi-
carriers in perovskite emitters. Early work5 on room-temperature nation kinetics and PLQYs can be described by equations (4) and
PeLEDs demonstrated a useful strategy for confining the charge car- (5), respectively:
riers by using thin perovskite layers (~30 nm). Similar confinement
effects can also be obtained by decreasing the perovskite grain sizes; dN  
� ¼ ktrap þ k N þ k2 N 2 ð4Þ
this has been obtained by constraining crystal growth through, for dt
instance, engineering the precursor stoichiometry20, incorporating
ammonium halides with large organic cations21 and embedding the and
perovskite grains in polymer matrixes22. In addition, confining one
of the carriers through local variations in band-edge energies can k
switch the decay to first-order kinetics and give enhanced PLQYs, as PLQY ¼   ð5Þ
reported for ‘photodoping’ in a mixed cation/anion system23. ktrap þ k þ k2 N

Nature Materials | www.nature.com/naturematerials

Review Article NaTuRE MaTERIaLs

a b c
P HO- P -OH
Vacancies

=
=
R-NH2 R-COOH O O
Diamine-based molecules

ZnX2/MnX2

Stronger passivation ability with B2+

Weaker hydrogen bonding with A+
NH 2
H 2N
Interstitials

O NH 2
H 2N O

NaSCN
O NH 2
H 2N O

NH4SCN O O
H 2N O NH 2

HOOC COOH + -
Metal Pb cluster Anti-sites H
H
N
N Zwitterionic salts
A+ B2+ X-

Fig. 2 | Defects in perovskites and defect passivation. a, Point defects of perovskites including vacancies, interstitials, anti-sites and metal Pb clusters.
b, Categories of passivating ligands for colloidal PNCs; the right side demonstrates the hard–soft acids–bases theory that was used in ref. 32 to predict the
passivation effects of Lewis-base ligands for CsPbX3 PNCs. c, An illustration demonstrating the modulation of passivation effects by rational molecular
design reported in ref. 6.

where ktrap, k and k2 are rate constants of non-radiative monomo- The basic concept of defect passivation in perovskites is to heal
lecular recombination, radiative monomolecular recombination the dangling bonds by using agents that provide extra coordina-
and Auger recombination, respectively. Accordingly, under low tion or ionic bonding to neutralize the charged defects, leading to
excitation densities where the second-order Auger process is neg- annihilation of electronic trap states. Generally, useful passivation
ligible, the PLQY is equal to k/(ktrap + k), which is independent of agents can be introduced by loading them in precursors for film
the excitation density (Fig. 1e). With finely controlled trap densities, deposition or colloidal synthesis of perovskites, or by performing
first-order excitonic radiative recombination could outperform the post-treatment on the crystallized perovskites6,32. A wide range
non-radiative recombination, leading to high PLQYs even at low of agents or ligands, including molecules containing Lewis acid/
excitation densities (Fig. 1e). Indeed, near-unity PLQYs have been base moieties, positively/negatively charged organic components,
demonstrated in a range of PNCs, 2D and quasi-2D perovskites10,25. and alkali metal ions, have been demonstrated to be efficient in
We note that, in practical cases, the radiative processes of reducing the defect density of perovskites, leading to increased
perovskite emitters might be more complicated. For example, in PLQYs of perovskite emitters and improved performance of
quasi-2D perovskites with large n values and PNCs with crystal PeLEDs33.
sizes close to exciton Bohr diameters, both excitonic and free-carrier Such studies have also been instrumental in understanding the
recombination may contribute to the light emission and collec- role of chemical structures of passivation agents on their effective-
tively determine the PLQYs. In addition, in mixed-dimensional ness. For instance, the successful passivation of halide vacancies
perovskites (mixtures of 2D/quasi-2D/3D phases), radiative in CsPbX3 PNCs using Lewis base ligands can be predicted by the
recombination can happen at different phases, and there is usually hard–soft acids–bases theory32: harder species such as alkylcarbox-
energy transfer from the large-bandgap phases to the low-bandgap ylates and carbonates are ineffective owing to a hard–soft mismatch
phase26,27. Furthermore, the presence of materials with excited-state between hard groups and soft Pb binding sites, whereas softer spe-
energies lower than those of the perovskite emitters (such as cies such as alkylphosphonates and sulfonates are effective owing to
additives in 3D perovskites, organic spacers in 2D or quasi-2D their higher binding affinity (Fig. 2b). Hydrogen bonding between
perovskites) can result in energy transfer from the perovskite emit- the passivation agents and the organic cations of the perovskite
ters to these materials4,28,29, leading to lower PLQYs or emission structure has a critical role in determining the passivation effec-
from these lower-energy materials. tiveness6: diamine-based agents, whose amino groups form weaker
hydrogen bonding with organic cations, are more effective in coordi-
Suppressing trap-assisted non-radiative recombination. In nating with unsaturated Pb sites (Fig. 2c). Moreover, molecules with
addition to the above strategies for enhancing the radiative P=O and As=O end groups have shown strong binding energies
recombination rates, minimizing trap-assisted non-radiative with unsaturated Pb sites in reduced-dimensional perovskites, lead-
recombination pathways is equally important for maximizing the ing to high PLQYs34. An in-depth understanding of such structure–
PLQYs of perovskite emitters. Trap states are usually associated with passivation relationships would be useful for rational design of opti-
intrinsic point defects, such as vacancies, interstitials and anti-site mal passivation agents.
occupations (Fig. 2a). So far, several types of point defects, includ- We note that PNCs often require additional considerations for
ing halide and A-site vacancies, Pb–halide anti-sites and metallic the passivation agents although they have similar defect passivation
Pb0, have been experimentally demonstrated to be associated with mechanisms to bulk perovskites. For example, the commonly used
non-radiative trap states30,31. surface ligands (such as oleic acid and oleylamine) are highly labile
Eliminating the negative effects of trap states has been inten- and easily desorb from the surface during the purification, lead-
sively investigated in perovskite photovoltaic materials, and vari- ing to a high density of dangling bonds at the surface of PNCs35. In
ous useful strategies for defect passivation have been developed30. addition, the residual long-chain organic ligands usually deteriorate

Nature Materials | www.nature.com/naturematerials

NaTuRE MaTERIaLs Review Article
the charge-transport properties of the resulting emissive layers. them43. For example, studies on solar cells have revealed that the
Therefore, alternative passivation ligands with stronger binding surface wetting properties of bottom interlayers affect the crystal-
affinity to PNCs and less negative effects on the electrical proper- lization process of the subsequently deposited perovskite films44. In
ties of the emissive layers (for instance, phosphonates, zwitter- addition, different surface properties of the perovskite layers can
ionic salts and metal halides; see Fig. 2b) are highly desired and affect the stacking of organic molecules on top of them, resulting
have been explored13. These useful strategies have improved the in morphological and electrical changes of the charge-transport
optoelectronic properties and stability of PNCs, benefiting the contacts45.
resulting PeLEDs13,36. Chemical interactions at the perovskite–interlayer interfaces and
their impacts on the performance of PeLEDs are more complicated
Controlling non-radiative Auger recombination. Auger recombi- and deserve further investigation. Functional passivating moieties
nation is a non-radiative process in which an electron−hole pair (such as Lewis base/acid groups), if grafted on interfacial materials,
passes recombination energy to another charge carrier. In the case are likely to induce additional healing of surface defects46. Beyond
of a high excitation density where Auger recombination is domi- passivation, other chemical reactions can occur between perovskite
nant with CN2 >> (BN + A) in equation (3) or k2N >> (k + ktrap) precursors and interlayers; for instance, the alkaline nature of the
in equation (5), the PLQY is inversely related to the excitation den- ZnO interlayer induces efficient deprotonation of the organic cat-
sity (Fig. 1b,c and e). It has been reported that Auger recombination ions and promotes perovskite crystallization47. As such, rational
can result in noticeable efficiency roll-off at high current densities control of the precursor stoichiometry has enabled high-quality
in LEDs37. FAxCs1−xPbI3 perovskite films on a ZnO interlayer, whereas the
In 3D bulk perovskites, for which the C value is around 10−28 same precursor failed to produce perovskite films on TiOx and SnO2
to 10−29 cm6 s−1 (refs. 18,38), Auger recombination is usually domi- interlayers (Fig. 3a)47.
nant at excitation densities higher than 1017 cm−3 (Fig. 1b,c), indi- In addition to the impacts on device engineering, these strong
cating that 3D bulk perovskites are promising for high-brightness physical/chemical interactions between perovskites and interlayers
LEDs and laser diodes. In quantum-confined perovskites, Auger suggest that special attention should be paid to the characteriza-
recombination becomes dominant at a lower charge-carrier density tion of perovskite films, especially polycrystalline films crystallized
(Fig. 1e) and could be one or two orders of magnitude faster17. A prom- in situ on different substrates. For instance, single-carrier devices,
ising mitigation strategy in PeLEDs based on mixed-dimensional which are commonly used to evaluate charge injection/transport
perovskites is to introduce a higher number of phases with large n balance in OLEDs and QLEDs, might not be reliable in PeLEDs
values to reduce local carrier density, ensuring high EQEs at high owing to the possible changes in the perovskite active layers depos-
current densities39. ited on different charge-transport interlayers.
Although Auger recombination in nanocrystals usually increases For emissive layers assembled from PNCs, which are more
remarkably with reducing crystal sizes, the situation is different in sensitive than conventional colloidal QDs to most of the com-
PNCs, where Auger recombination shows a much weaker volume monly used organic solvents, vacuum deposition of the subsequent
dependence compared with conventional QDs, deviating from charge-transport layers is often used to avoid damage to the emis-
the common volume scaling (1/volume)40,41. This phenomenon is sive layers48. For successful solution processing of charge-transport
more pronounced in the weak confinement region, resulting in layers on top, rationally designed surface ligands or crosslinking of
Auger recombination rates of PNCs with little difference from bulk the PNCs are necessary to improve the resistance of these emissive
perovskites42. Auger recombination was demonstrated to be slower layers to organic solvents49,50.
in organic-cation-containing PNCs than in all-inorganic ones42.
These observations suggest that PNCs may outperform conven- Light outcoupling in PeLEDs. Perovskite emissive layers generally
tional QDs under high excitation densities, and they call for fur- possess a high refractive index, which potentially limits the outcou-
ther understanding of the mechanisms determining Auger losses in pling efficiency of PeLEDs as predicted by the ray-optics theory.
perovskites. However, detailed optical simulations indicate higher outcoupling
efficiency in PeLEDs, leading to maximum EQEs of 25% and 20%
Device engineering and characterizations of PeLEDs for PeLEDs based on 3D perovskites and on mixed-dimensional
Highly emissive perovskite layers are only one ingredient needed perovskites, respectively, under the assumption of planar struc-
for high-efficiency PeLEDs; other key elements include balanced tured films without any light-extraction technologies51. To fur-
and efficient charge injection as well as high light outcoupling ther improve the external efficiencies, the same light-extraction
efficiency. Typically, PeLEDs have a device structure in which the technologies developed in OLEDs and QLEDs (for instance, the use
perovskite emissive layer is sandwiched between electron- and of a nanophotonic substrate52 or nanoimprinted moth-eye nano-
hole-transport interlayers. The device structure is similar to those structures combined with a half-ball lens53) have proven useful in
of solution-processed organic LEDs (OLEDs) and conventional enhancing light outcoupling and peak EQEs of PeLEDs as well. In
QD-based LEDs (QLEDs), from which we can readily borrow rich addition, for PeLEDs with strong self-absorbing perovskite emissive
experience, including the selection of interfacial materials, engi- layers, various filtering structures and reduced electrode area are
neering of energy level alignments at the interfaces, manipulation of expected to show a greatly enhanced light-outcoupling efficiency,
charge injection, and design of light-outcoupling techniques. owing to the benefits of high photon recycling combined with a
Despite these similarities, two features are unique to PeLEDs: reduction in the amount of photons trapped in the substrate and
strong physical/chemical interactions between perovskites and waveguide modes54.
interlayers; and ion migration. These two features have impacts on A further strategy to improve light outcoupling relies on modu-
both the device optimization and reliable characterization of the lating the morphology of perovskite films. Perovskite emissive lay-
device performance. In addition, perovskites provide opportunities ers consisting of isolated submicrometre-sized platelets (Fig. 3b)
for facile control of light outcoupling, which will also be discussed exhibit a much-enhanced light-outcoupling efficiency approach-
in this section. ing 30%, leading to NIR PeLEDs with a high peak EQE of 20.7%
(ref. 8). The simple light-extraction strategy through morphology
Interface engineering of PeLEDs. The physical properties of inter- control of the perovskite emissive layers retains the wavelength- and
layers can affect the crystallinities, morphologies, defect properties viewing-angle-independent electroluminescence and avoids the use
and energy level structures of perovskite layers deposited on top of of complicated manufacturing processes8. We note that, in contrast

Nature Materials | www.nature.com/naturematerials

Review Article NaTuRE MaTERIaLs

a b c
1.2 24 350
Normalized PL intensity (a.u.)

20 300
1.0
MoOx /Au

Radiance (W sr–1 m–2)

TiOx /PEIE 250
0.8 SnO2/PEIE TFB 16
200

EQE (%)
ZnO/PEIE Perovskite
0.6 12
Organic layer 150
0.4 ZnO–PEIE 8 100
EQE
ITO 4 50
0.2 Radiance
Glass substrate
0 0
0.0
500 600 700 800 900 1,000 0 100 200 300 400
Wavelength (nm) Current density (mA cm–2)

d e f
24 100 140 1.1
25 °C
22 1.0 35 °C

Normalized EL intensity (a.u.)

80 120 45 °C
20 0.9
Radiance (W sr–1m–2)

18 100 Tramp 0.8

L/L0 (%)

60
EQE (%)

16 0.7
80
14 40 0.6
12 EQE
60 0.5
Radiance 20
10 Tdecay
@ 50 mA cm–2 0.4
8 40
0 0.3
0 1 2 3 4 5 6 Time 0 5 10 15 20 25
Time (s) Time (min)

Fig. 3 | Device aspects of PeLEDs. a, Normalized photoluminescence (PL) spectra and photographs (inset) of perovskite films on TiOx/PEIE, SnO2/PEIE
and ZnO/PEIE substrates prepared from the same precursor solution, indicating that the interlayers can strongly affect the properties of the perovskite
layer; PEIE is polyethylenimine ethoxylated. b, An illustration (left panel) of a device with submicrometre-sized perovskite platelets, where rays A, B
and C represent light, otherwise trapped in devices with a continuous emitting layer, that can be extracted by the submicrometre structure. TFB is
poly(9,9-dioctyl-fluorene-co-N-(4-butylphenyl)diphenylamine). Top-view scanning electronic microscopic (SEM) image (right panel) of the perovskite
film, with a scale bar of 1 μm. c, d, Examples of the commonly used EQE– and radiance–current density curves (c), and stabilized EQE and radiance at a
constant current density for the same device (d). e, A schematic operational stability curve of PeLEDs with an obvious increase in the luminance during the
measurement. Tramp, the needed time for the initial luminance to increase to the peak value; Tdecay, the needed time for the peak luminance to drop to 50%
of the value. f, Operational stability of an FAPbI3 PeLED measured at various temperatures at 100 mA cm−2. Figure adapted with permission from:
a, ref. 47, under a Creative Commons licence (http://creativecommons.org/licenses/by/4.0/); b, ref. 8, Springer Nature Ltd; c, ref. 6, Springer Nature Ltd;
f, ref. 63, under a Creative Commons licence (http://creativecommons.org/licenses/by/4.0/).

to early-stage film optimization strategies that emphasize the film In addition, EQE enhancement by repeated voltage scans has been
uniformness, discontinuous film morphology does not necessarily observed, ascribed to defect filling caused by ion migration under
result in electrical issues, as long as leakage current and electrical electrical stress59. It has also been demonstrated that the peak values
shunts through the gaps of the isolated perovskite grains can be and shapes of the EQE curves are dependent on the scan speed dur-
properly suppressed6,8,55. ing the measurements21.
Another approach for achieving high light-outcoupling effi- As an emerging research field that is attracting increasing atten-
ciency is to use light emitters with horizontally oriented (in-plane) tion, PeLEDs development critically requires reliable measurements
transition dipole moments, a strategy that is widely used in OLEDs of the device parameters. We suggest providing details of the mea-
and yet to be explored in PeLEDs. Theoretical simulations predict surements, such as scan step, delay time, scan speed and the use of
a maximum light-outcoupling efficiency of around 46% for OLEDs pre-bias stress protocols. In addition to the commonly used EQE–
based on light emitters with both PLQYs and orientation factor and brightness–current density curves (Fig. 3c), we propose to pro-
approaching unity56. Encouragingly, anisotropic photon emission vide stabilized EQEs and brightness at a constant current density
has been observed in PNCs57, suggesting that high light-outcoupling to minimize the effect of scanning methods on the device perfor-
efficiencies in PeLEDs are possible by using perovskites with mance (Fig. 3d). Although the values measured under a constant
in-plane transition dipole moments. current density could be slightly different from those measured with
scanning current densities, the former represents the values under
Device characterizations of PeLEDs. The characterization meth- practical operating conditions of LEDs. Note that measuring stabi-
ods for PeLEDs follow the standards established in other LED lized power output has been widely adopted in the perovskite pho-
technologies58. However, the well-acknowledged ion migration in tovoltaic community to accurately determine the performance60. In
perovskites brings challenges in reliably determining the param- addition, for spectrally unstable devices, it is necessary to provide
eters, and this calls for further attention. In particular, the perfor- both the electroluminescence spectra and EQEs at different bright-
mance parameters of PeLEDs, including current densities, EQEs, ness, or at different driving voltages or current densities.
luminance/radiance and efficiency roll-off, often depend on how Device operational stability is another critical parameter of
the measurements are performed. For example, a hysteresis in EQE– PeLEDs. In the LED luminaire industry, operational stability (life-
voltage curves under different scanning directions was reported5. time) is typically recognized as the operating time until lumen

Nature Materials | www.nature.com/naturematerials

NaTuRE MaTERIaLs Review Article
maintenance has diminished to a certain level. Similar to this segregate into Br- and I-rich domains under the electric field, resulting
industrial standard, in literature, the operational stability of a vis- in shifts of the electroluminescence spectra and changes of emission
ible LED is often evaluated based on the time required for the lumi- colours during device operation (Fig. 4b)66; a similar phenomenon is
nance to drop to a certain level (for instance, 50%; that is, T50) of observed in blue PeLEDs with mixed chloride–bromide perovskite
its initial value under continuous operating conditions7,9,28,61. As emissive layers49.
shown in Fig. 3e, slow increase of the luminance ranging from min- Although they are challenging, we believe that it is too early to
utes to hours (Tramp) is often observed in stability measurements of rule out the possibility of using mixed halide perovskites for effi-
PeLEDs7,28,43,46,62. In this case, the time needed to drop to 50% of the cient colour-stable red/blue emission. Experimental and theoreti-
peak luminance (Tdecay) would be a more reasonable value to evalu- cal calculations suggest that phase segregation is mediated by ionic
ate the device’s operational stability (Fig. 3e), as there is no evidence defects, especially by vacancies67. Therefore, attention should be
showing that the ramp process is part of the degradation. In addi- paid to defect passivation to eliminate ion migration pathways. In
tion, the operational stability of PeLEDs strongly depends on the addition, previous investigations on perovskite photovoltaics sug-
environmental temperature (Fig. 3f)63, which should be clearly gested that photo-induced phase segregation is thermodynamic in
stated along with the device stability results. nature; hence it can be driven by free-energy differences linked to
For devices with relatively poor stability, which is the case for composition or electronic structure68. Local strains induced by lat-
most PeLEDs demonstrated at this stage, we would suggest directly tice mismatch may also facilitate ion migration69. Therefore, a high
measuring the T50. For highly stable PeLEDs, accelerated tests of the compositional homogeneity in mixed halide perovskite films with
operational stability at a high luminance can be used to predict the low free energy and low local strain could be favourable for sup-
T50 at a lower luminance (for instance, 100 cd m−2). The prediction pressing phase segregation, although it is a challenging task owing
is based on an empirical scaling law described in equation (6): to widely observed non-equilibrium crystal growth in these materi-
als70. It has also been proposed that light-induced phase separation
T50 ´ La0 ¼ constant ð6Þ in mixed halide perovskites is negligible when the charge-carrier
density is below a certain threshold71. This might explain previous
where a is the acceleration factor, fitted based on measured lifetimes results showing improved spectral stability for PeLEDs based on
at different initial luminance (L0) values64. We emphasize that the mixed halide perovskites operated at low current densities9,62. These
value of a depends on the materials and the device architectures. findings indicate that achieving colour-stable PeLEDs with mixed
As a result, for highly stable PeLEDs where we need to use equation halide compositions might be feasible, and call for further under-
(6), it is necessary to determine the a value based on sufficient data- standing of the ion segregation mechanisms, especially under the
sets collected from stability tests under different initial luminance. electric field.
We also note that the degradation mechanisms of PeLEDs might be An alternative strategy to approach the Rec. 2020 primary blue
more complicated than other LEDs, and hence systematical stud- (red) colours is to introduce quantum confinement effects in pure
ies and in-depth understanding of the degradation mechanisms are bromide (iodide)-based perovskites. The enhanced quantum con-
required. For example, ion migration could screen external electric finement can be achieved through either using mixed-dimensional
fields, causing changes in diode characteristics. In addition, the ions perovskites or synthesizing perovskite QDs. However, considerable
tend to localize at interfaces or grain boundaries, leading to changes challenges also accompany these quantum-confined structures.
in interfacial properties and formation of defect sites. In mixed-dimensional perovskites, emission mostly occurs from
For non-visible PeLEDs (for instance, NIR ones), where lumi- the phase with the smallest bandgap, owing to efficient energy
nance is not relevant, we propose to measure the radiance decay transfer from the large-bandgap phases26,27. Because there is usu-
instead when reporting their operational stability. This would ally a wide dimensional distribution, a high amount of large-sized
allow the community to reasonably compare the device lifetimes of organic cations is required to push the emission towards the
non-visible PeLEDs at a certain level of light output, as already done desired blue and red colours61,72; however, poor charge-transport
for visible PeLEDs. properties, multiphase-induced peak broadening and even elec-
troluminescence peak shifts are widely observed in these emis-
Key challenges ahead in PeLEDs sive layers72,73. In addition, this can bring about more n = 1 phases
Rapid progress in PeLEDs has been made in the past several years, (Box 1), which are usually less efficient emitters owing to strong
yet there are critical challenges to be tackled to foster further electron–phonon coupling. Therefore, precise dimensional con-
advances of this promising technology. In this section, we iden- trol towards narrow dimensional distribution, which would reduce
tify key challenges and discuss possible strategies to address them: the need for large organic cations and improve charge transport,
improving the performance of red and blue PeLEDs, enhancing the is highly desired to realize Rec. 2020 primary blue and red emis-
stability of PeLEDs and solving the toxicity issue of lead. sion in mixed-dimensional perovskites (Fig. 4c)74. Crystallization
control approaches, such as using anti-solvent techniques or care-
High-performance red and blue PeLEDs. Despite the great success fully screening large cations and additives, would be promising to
in high-performance green and NIR PeLEDs, it is still challenging decrease the dimensional distribution74,75. In addition to modulat-
to achieve high-efficiency red and blue devices with high luminance ing dimensional distribution, controlling the vertical orientation of
(Fig. 4a), especially those with the CIE coordinates approaching the the perovskite phases, favourable for charge transport in PeLEDs,
Rec. 2020 specified red (0.708, 0.292) and blue (0.131, 0.046) pri- might also help to improve device performance76.
mary colours that are desired for ultra-high-definition displays. For QDs, the quantum confinement is typically obtained when
A straightforward strategy to approach these desired red and the crystal size of a material is comparable to or smaller than its
blue colours is to use perovskite emissive layers with well-adjusted exciton Bohr diameter. The exciton Bohr diameters for CsPbBr3
mixed halide compositions: mixed chloride–bromide for blue and and CsPbI3 are 7 nm and 12 nm, respectively15; the values are even
mixed bromide–iodide for red emission. Although continuous smaller for their organic–inorganic hybrid counterparts77. This
bandgap tuning by varying halide ions is one of the most attrac- imposes strict limits on the crystal size, making it challenging to
tive properties of perovskite emitters, practical applications of this synthesize such small QDs with desired blue/red emission and good
strategy are limited by the colour instability of the resulting stability. Poor size uniformity might be another challenge for these
PeLEDs9,65. For instance, in the case of red PeLEDs based on mixed small QDs, inducing spectral broadening and leading to poor colour
bromide–iodide perovskite emissive layers, Br and I ions tend to purity13. To mitigate these limits of QDs, shape engineering, such

Nature Materials | www.nature.com/naturematerials

Review Article NaTuRE MaTERIaLs

a b c
25 Green 1.2
Increasing n
Red 0 min (5 V)

Normalized EL intensity (a.u.)

1.0
20 Blue 0 min (7 V)
2 min (7 V)
0.8

Phase population
4 min (7 V)
15
6 min (7 V)
EQE (%)

0.6
10
0.4

5 n=1
0.2

0 0.0
101 102 103 104 105 106 500 600 700 800 900 400 450 500
Luminance (cd m–2) Wavelength (nm) Wavelength (nm)

d e f
x = 0.8 CsPbBr3
PEA2(RbxCs1–x)2Pb3Br10
x = 0.6 2.8 CsPb1–xCdxBr3
x = 0.4
Normalized intensity (a.u.)

x = 0.2 CsPb1–xZnxBr3
x=0 2.7

PL energy (eV)
2.6

2.5

2.4
50 nm
400 450 500 550 5.70 5.75 5.80 5.85 5.90
Wavelength (nm) Lattice vector (Å)

Fig. 4 | Challenges in high-performance blue and red PeLEDs. a, Summary of EQE and luminance of state-of-the-art blue, green and red PeLEDs; the data
used for this figure are listed in Supplementary Table 1. b, Normalized electroluminescence (EL) spectra of a CsPb(Br/I)3-based PeLED during operation.
c, A schematic diagram of the dimensional control in mixed-dimensional perovskites towards narrow dimensional distribution with the aim of realizing
Rec. 2020 primary blue emission. Reducing the domains composed of n = 1 phase and large n phases (red arrows) helps achieving narrow dimensional
distribution. d, Low-resolution transmission electron microscopic (TEM) image (left panel) of stacked nanoplatelets (thicknesses are around 3.5 nm
and 6.5 nm without and with the ligand shell, respectively) and TEM image (right panel) of CsPbBr3 cuboid nanocrystals assembled from nanoplatelets.
e, Normalized photoluminescence spectra of PEA2(RbxCs1−x)2Pb3Br10 perovskites, demonstrating the effect of Rb+ incorporation on the bandgap.
f, Photoluminescence energy as a function of lattice vector in doped CsPb1–xMxBr3 (M = Cd and Zn) nanocrystals and their parent CsPbBr3 nanocrystals,
indicating that impurity metal ions in the perovskite lattice can tune the bandgap of perovskite nanocrystals. Figure adapted with permission from: b, ref. 66,
American Chemical Society; c, ref. 74, American Chemical Society; d (left panel), ref. 79, American Chemical Society; d (right panel), ref. 81, under a Creative
Commons licence (http://creativecommons.org/licenses/by/4.0/); e, ref. 83, under a Creative Commons licence; f, ref. 84, American Chemical Society.

as the use of colloidal nanoplatelets, also called colloidal quantum impurity metal ions into the perovskite lattice provides an additional
wells (CQWs), could be a useful approach78. As the quantum con- synthetic tool to tune the bandgaps of the PNCs. For instance, zinc-
finement in these systems is hardly affected by the lateral size distri- or cadmium-doped CsPbBr3 PNCs present considerably blueshifted
bution, the size requirement is only on the thickness of the CQWs emission owing to lattice contraction (Fig. 4f)84. This approach would
and hence less stringent, beneficial for achieving narrow emission be difficult to achieve in polycrystalline films formed in situ, because
linewidth. For example, a sharp photoluminescence spectrum with of the high formation energies of the doped perovskites, in which the
the FWHM of 18 nm has been realized in blue emissive CsPbBr3 impurity metal ions tend to locate on grain surfaces.
CQWs with well controlled thickness (Fig. 4d, left panel)79. These
CQWs can also form large nanocrystals through self-assembly Stability of PeLEDs. As a young technology, the operational sta-
(Fig. 4d, right panel) while preserving narrow emission from the bility of PeLEDs remains a big challenge limiting their practical
CQWs, which is promising for PeLEDs80,81. applications. Currently, the most stable PeLEDs are recorded with
Successful fabrication of high-performance PeLEDs with blue/red NIR devices8. Nevertheless, visible PeLEDs show T50 of hundreds of
emission that satisfies display purposes might require a combination hours at most at a low luminance of 100 cd m−2 (ref. 85), far behind
of strategies leveraging on mixed halides, the quantum confinement those of OLEDs and QLEDs. Most of the blue PeLEDs reported so
effect, and other materials composition engineering (such as A- or far can only last for tens of seconds to minutes61,83. Inspired by the
B-site alloying or doping). For instance, the introduction of Rb+ into development of perovskite photovoltaics86, we consider that the
CsPbBr3 increases the bandgap owing to the tilting of the [PbBr6]4− factors responsible for the poor stability of PeLEDs could mainly
octahedra and the reduced overall orbital overlap82. It is therefore include ion migration, thermal instability and interfacial instability.
attractive to shift the emission towards the blue by incorporating Besides the impact on phase stability of mixed halide perovskites
Rb+, which is also compatible with other strategies for bandgap tun- discussed above, ion migration has also been regarded as a critical
ing. In fact, blue emission within the range of about 450–490 nm has issue affecting the long-term stability of perovskite optoelectronic
been demonstrated in PEA2(RbxCs1−x)2Pb3Br10 mixed-dimensional devices67. Ion migration is generally associated with defect for-
perovskites, where PEA is phenethylammonium (Fig. 4e)83. Introducing mation, interfacial charge and ion accumulation, degradation of

Nature Materials | www.nature.com/naturematerials

NaTuRE MaTERIaLs Review Article
interlayers and corrosion of electrodes67. It is generally believed that attempted in LEDs, these materials (such as tin-based perovskites94,
point defects, especially halide vacancies, serve as ion migration Cs2AgInCl6 derivatives12, A3B2X9-type materials95 and copper
channels87. Therefore, reducing defect density with rational passiv- halides96) show relatively low efficiencies for now. Although the
ation and crystallization control would be effective ways to mitigate design rules to target promising materials from such a big library
ion migration6,55. Fortunately, PeLEDs are more tolerant to additives are currently lacking, these lead-free compounds might have bet-
than perovskite photovoltaics, owing to less stringent requirements ter chances for LEDs than photovoltaics, considering the larger tol-
on charge-carrier mobility. Hence, various additives are allowed erance on exciton binding energies, bandgaps and charge-carrier
in perovskite films at a relatively high content, making it possible mobilities of the materials in LEDs. A combination of experimen-
to heal the defects or immobilize the mobile ions by a variety of tal and theoretical investigations, maybe involving machine learn-
chemical bonding or physical interactions. In addition to long-term ing, may well help to search for the right lead-free materials for
stability of PeLEDs, ion migration might affect the response time high-performance LEDs97.
of the devices during practical display operations. Encouragingly,
similar dynamic behaviours to those observed in OLEDs were Conclusions and outlook
demonstrated by transient electroluminescence investigations of Recent advances in PeLEDs indicate their bright future in the
PeLEDs88, and the operation frequency of PeLEDs can reach tens areas of displays, lighting and optical communications. Their
of megahertz11. properties potentially go beyond these areas; for example, the
Thermal instability of perovskite emitters is not only related to excellent performance of PeLEDs and of hybrid LEDs based on
materials decomposition, but also to the change of dimensional inorganic QDs and perovskites98,99 in the NIR region may bring
distribution on heating. The latter is particularly serious in some about new applications in biomedical diagnosis and a variety of
mixed-dimensional perovskite films, which usually undergo a wearable electronic devices. Here we discuss the outlook for fur-
dimensional transition toward large-n phases upon heating89. ther development of white PeLEDs and the potential applications
In addition, we notice that most of the highly emissive perovskite of perovskite emitters in printable and flexible electronics as well
films are prepared with a short-term annealing procedure at as electrically pumped lasers, aiming to shed light on these promis-
a low temperature (<100 °C for several minutes)61,90. Further ing future directions.
increasing the annealing temperature or duration usually leads
to PLQY dropping90. Therefore, Joule heating during device White PeLEDs. In addition to the three primary colours (blue,
operation might cause permanent degradation of the perovskite green and red), perovskites also offer great opportunities for direct
emissive layers. white light emission. White light can be generated by using a single
In addition to the issues caused by mobile ions and thermal insta- white-emitting material or a set of luminophores of distinct colours.
bility of perovskites, the selection of electrodes and charge-transport Luminophores are assembled in several ways, such as single- or
layers can also considerably affect the device operational multi-emissive-layer white LEDs, or hybrid LEDs combining colour
stability55,63,91. Previous investigations have revealed that drift and down-conversion emitters with blue/ultraviolet chips. Although
diffusion of metal ions from the electrodes into perovskites can result efficient hybrid white LEDs combining perovskite emitters with
in device degradation due to redox reactions (for instance, between commercial inorganic LED chips have been demonstrated, there are
Al0 and Pb2+)91. This might be more serious in PeLEDs because of few examples of white PeLEDs that show white electroluminescence
the electric-field-induced diffusion of metal ions. In addition, it has directly from the perovskite emitters12,100. One promising strategy
been demonstrated that the electroluminescence decay could be is to take advantage of the emission from self-trapped excitons
much quicker than the photoluminescence decay in PeLEDs, indi- (STEs). In the simplified mechanism for STE emission (Fig. 5a),
cating degradation of the interfaces between the perovskite emissive free excitons undergo fast relaxation to various self-trapped states
layer and charge-transport layers55,63. Ideal charge-transport layers with different energies lying within the bandgap, yielding broad-
should have excellent interfacial and morphological stability upon band emission that is desired for indoor lighting. STEs have been
heating and could also protect the perovskite layers from metal ions widely observed in 2D perovskites, lead-free perovskites and other
in the electrodes. Currently, the device structures and interlayer metal halides101. For instance, white light emission was achieved
materials for PeLEDs follow those established in OLEDs. However, from STEs in doped double perovskites (Fig. 5b)12. However, the
perovskites are much more sensitive to interfacial chemical and performance of these white PeLEDs needs further improvement,
physical environment47, and hence more targeted designs are neces- and more efforts are required to exploit the potential of perovskite
sary to improve the device lifetime. materials for white light generation.

Lead-free perovskites. The presence of lead in perovskite opto- Printable and flexible electronics. Perovskite emitters are compati-
electronic devices raises concerns about their toxicity, even though ble with low-temperature solution-based manufacturing techniques,
the lead content in perovskite optoelectronic devices meets current such as inkjet printing, roll-to-roll and 3D printing, providing
criteria for commercial products as previously evaluated in great potential in scale-up production and flexible electronics. For
perovskite photovoltaics92. One critical reason is that lead halide example, multicolour pixel arrays based on 3D-printed perovskite
perovskites have high solubility in water, making them readily leach nanowires have been demonstrated, where the emission colour can
into the environment in the case of encapsulant failure92. Before a be controlled by leveraging anisotropy property of the emitters with
clear consensus on the toxicity issue is reached, increasing efforts polarizing filters102. High-performance flexible PeLEDs have also
should be made in parallel to search for alternative compounds that been fabricated on various lightweight substrates, such as carbon
are environmentally friendly and perform equally well. nanotubes and silver nanowires103,104. These devices showed excel-
During the past several years, a range of promising lead-free mate- lent mechanical robustness with negligible performance loss after
rials have been developed, including 3D ABX3 perovskites (mainly a bending test of up to 10,000 cycles104. Pixelated touch-responsive
based on Sn2+), vacancy-ordered perovskites (such as Cs2SnI6), PeLEDs for instantaneous visualization of pressure mapping fur-
halide double perovskites (such as Cs2AgInCl6, Cs2AgBiBr6), and ther expand the potential of this technology to applications includ-
non-perovskite metal halides with various metal ions and dimen- ing robotics, motion detection and fingerprint identification105.
sionalities93. These lead-free perovskites and perovskite deriva- For high-resolution devices with enhanced performance, further
tives, some of which have shown comparable PLQYs to lead halide improvements in ink design, device engineering and associated
perovskites, represent a rich library for LED applications. When printing and encapsulation technologies are still necessary.

Nature Materials | www.nature.com/naturematerials

Review Article NaTuRE MaTERIaLs

a b
FC 1.2

FE 1.0

Normalized intensity (a.u.)

0.8
STE
Energy
0.6

0.4
GS
0.2

0.0
400 500 600 700 800
Nuclear coordinate
Wavelength (nm)
c d
100 PW = 800 / 250 ns
/ Glass
100 80 / Sapphire
Threshold fluence (μJ cm–2)

Radiance (kW sr–1 m–2)

/ Sapphire + graphite

60

10 40

20

1 0
150 200 250 300 0.0 0.5 1.0 1.5 2.0 2.5 3.0
Temperature (K) Current density (kA cm–2)

Fig. 5 | Prospects for perovskites in white PeLEDs and electrically pumped lasers. a, A schematic illustration of STE emission; FC, free carrier state; FE,
free exciton state; STE, self-trapped exciton state; GS, ground state. b, Normalized photoluminescence spectrum of Cs2Ag0.60Na0.40InCl6 measured at
293 K; the inset shows a 300-nm-thick Cs2Ag0.60Na0.40InCl6 film under 254-nm ultraviolet illumination. c, Pump threshold fluence as a function of
temperature for MAPbI3. d, Radiance–current density curves of pulse-driven PeLEDs on glass or sapphire substrates, with or without a graphite heat spreader
and copper heat sink. PW, pulse width. Adapted with permission from: b, ref. 12, Springer Nature Ltd; c, ref. 109, American Chemical Society; d, ref. 110, Wiley.

Electrically pumped perovskite lasers. Solution-processed elec- high-performance blue and red PeLEDs, long-term device stability
trically pumped laser diodes have been intensively studied since and lead-free PeLEDs. These challenges require an interdisciplin-
the 1990s but are hard to achieve with organic emitters and conven- ary approach for further breakthroughs to create industrial paths.
tional QDs106. Perovskites, which show low trap density (<1016 cm−3), Combining the considerable advances already made by the commu-
high charge-carrier mobility (>10 cm2 V−1 s−1) and low Auger recom- nity and the outstanding properties of perovskites, we believe that
bination coefficient (about 10−28 to 10−29 cm6 s−1)2, may offer greater PeLEDs offer unique advantages for future applications.
opportunities towards this ambitious goal. So far, a range of optically
pumped perovskite lasers with tunable emission (from 390 to 800
nm), ultra-low threshold (around 0.2–40 µJ cm−2) and high optical Received: 12 December 2019; Accepted: 24 July 2020;
Published: xx xx xxxx
gain (as high as 3,200 cm–1 for CsPbBr3 PNCs) have been demon-
strated24. The achievement of lasing usually requires a high thresh-
old carrier density of around 1018 cm−3, corresponding to a roughly References
estimated threshold current density of at least hundreds of A cm−2 1. Stranks, S. D. & Snaith, H. J. Metal-halide perovskites for photovoltaic and
under electrical excitation107. Encouragingly, short-pulse driven light-emitting devices. Nat. Nanotechnol. 10, 391–402 (2015).
2. Quan, L. N. et al. Perovskites for next-generation optical sources. Chem.
PeLEDs have shown a negligible Auger loss although they were Rev. 119, 7444–7477 (2019).
operated at high current densities exceeding 150 A cm−2; in 3. Era, M., Morimoto, S., Tsutsui, T. & Saito, S. Organic–inorganic
this case, their efficiency roll-off is dominated by Joule heat- heterostructure electroluminescent device using a layered perovskite
ing and charge imbalance108. In addition, the excitation thresh- semiconductor (C6H5C2H4NH3)2PbI4. Appl. Phys. Lett. 65, 676–678 (1998).
old increases considerably with rising temperature (Fig. 5c)109. 4. Chondroudis, K. & Mitzi, D. B. Electroluminescence from an organic−
inorganic perovskite incorporating a quaterthiophene dye within lead halide
Rational designs are thus desired to develop perovskite materials perovskite layers. Chem. Mater. 11, 3028–3030 (1999).
with temperature-insensitive thresholds and devices with thermal 5. Tan, Z.-K. et al. Bright light-emitting diodes based on organometal halide
dissipation, respectively. Along this line, thermal management perovskite. Nat. Nanotechnol. 9, 687–692 (2014).
in devices has successfully resulted in PeLEDs with ultra-high 6. Xu, W. et al. Rational molecular passivation for high-performance
radiance at around 2 kA cm−2 (Fig. 5d)110. perovskite light-emitting diodes. Nat. Photon. 13, 418–424 (2019).
7. Lin, K. et al. Perovskite light-emitting diodes with external quantum
To conclude, we have discussed the fundamental require- efficiency exceeding 20 per cent. Nature 562, 245–248 (2018).
ments for achieving high-performance PeLEDs from both mate- 8. Cao, Y. et al. Perovskite light-emitting diodes based on spontaneously
rial and device perspectives, and highlighted the key challenges for formed submicrometre-scale structures. Nature 562, 249–253 (2018).

Nature Materials | www.nature.com/naturematerials

NaTuRE MaTERIaLs Review Article
9. Chiba, T. et al. Anion-exchange red perovskite quantum dots with 41. Castañeda, J. A. et al. Efficient biexciton interaction in perovskite quantum
ammonium iodine salts for highly efficient light-emitting devices. Nat. dots under weak and strong confinement. ACS Nano 10, 8603–8609 (2016).
Photon. 12, 681–687 (2018). 42. Eperon, G. E., Jedlicka, E. & Ginger, D. S. Biexciton Auger recombination
10. Zhao, B. et al. High-efficiency perovskite–polymer bulk heterostructure differs in hybrid and inorganic halide perovskite quantum dots. J. Phys.
light-emitting diodes. Nat. Photon. 12, 783–789 (2018). Chem. Lett. 9, 104–109 (2018).
11. Bao, C. et al. Bidirectional optical signal transmission between two identical 43. Zhang, L. et al. Ultra-bright and highly efficient inorganic based perovskite
devices using perovskite diodes. Nat. Electron. 3, 156–164 (2020). light-emitting diodes. Nat. Commun. 8, 15640 (2017).
12. Luo, J. et al. Efficient and stable emission of warm-white light from 44. Bi, C. et al. Non-wetting surface-driven high-aspect-ratio crystalline grain
lead-free halide double perovskites. Nature 563, 541–545 (2018). growth for efficient hybrid perovskite solar cells. Nat. Commun. 6,
13. Shamsi, J., Urban, A. S., Imran, M., De Trizio, L. & Manna, L. Metal halide 7747 (2015).
perovskite nanocrystals: synthesis, post-synthesis modifications, and their 45. Jung, E. H. et al. Efficient, stable and scalable perovskite solar cells using
optical properties. Chem. Rev. 119, 3296–3348 (2019). poly(3-hexylthiophene). Nature 567, 511–515 (2019).
14. Huang, H., Bodnarchuk, M. I., Kershaw, S. V., Kovalenko, M. V. & Rogach, 46. Yang, X. et al. Efficient green light-emitting diodes based on
A. L. Lead halide perovskite nanocrystals in the research spotlight: stability quasi-two-dimensional composition and phase engineered perovskite with
and defect tolerance. ACS Energy Lett. 2, 2071–2083 (2017). surface passivation. Nat. Commun. 9, 570 (2018).
15. Protesescu, L. et al. Nanocrystals of cesium lead halide perovskites (CsPbX3, 47. Yuan, Z. et al. Unveiling the synergistic effect of precursor stoichiometry
X = Cl, Br, and I): novel optoelectronic materials showing bright emission and interfacial reactions for perovskite light-emitting diodes. Nat. Commun.
with wide color gamut. Nano Lett. 15, 3692–3696 (2015). 10, 2818 (2019).
16. Yu, D., Cao, F., Gao, Y., Xiong, Y. & Zeng, H. Room-temperature 48. Song, J. et al. Quantum dot light-emitting diodes based on inorganic
ion-exchange-mediated self-assembly toward formamidinium perovskite perovskite cesium lead halides (CsPbX3). Adv. Mater. 27, 7162–7167 (2015).
nanoplates with finely tunable, ultrapure green emissions for achieving Rec. 49. Li, G. et al. Highly efficient perovskite nanocrystal light-emitting
2020 displays. Adv. Funct. Mater. 28, 1800248 (2018). diodes enabled by a universal crosslinking method. Adv. Mater. 28,
17. Xing, G. et al. Transcending the slow bimolecular recombination in lead- 3528–3534 (2016).
halide perovskites for electroluminescence. Nat. Commun. 8, 14558 (2017). 50. Shi, Z. et al. Strategy of solution-processed all-inorganic heterostructure for
18. Herz, L. M. Charge-carrier dynamics in organic–inorganic metal halide humidity/temperature-stable perovskite quantum dot light-emitting diodes.
perovskites. Annu. Rev. Phys. Chem. 67, 65–89 (2016). ACS Nano 12, 1462–1472 (2018).
19. Manser, J. S. & Kamat, P. V. Band filling with free charge carriers in 51. Shi, X.-B. et al. Optical energy losses in organic–inorganic hybrid
organometal halide perovskites. Nat. Photon. 8, 737–743 (2014). perovskite light-emitting diodes. Adv. Opt. Mater. 6, 1800667 (2018).
20. Cho, H. et al. Overcoming the electroluminescence efficiency limitations of 52. Zhang, Q. et al. Efficient metal halide perovskite light-emitting diodes with
perovskite light-emitting diodes. Science 350, 1222–1225 (2015). significantly improved light extraction on nanophotonic substrates. Nat.
21. Xiao, Z. et al. Efficient perovskite light-emitting diodes featuring Commun. 10, 727 (2019).
nanometre-sized crystallites. Nat. Photon. 11, 108–115 (2017). 53. Shen, Y. et al. High-efficiency perovskite light-emitting diodes with
22. Li, G. et al. Efficient light-emitting diodes based on nanocrystalline synergetic outcoupling enhancement. Adv. Mater. 31, 1901517 (2019).
perovskite in a dielectric polymer matrix. Nano Lett. 15, 2640–2644 (2015). 54. Cho, C. et al. The role of photon recycling in perovskite light-emitting
23. Feldmann, S. et al. Photodoping through local charge carrier accumulation diodes. Nat. Commun. 11, 611 (2020).
in alloyed hybrid perovskites for highly efficient luminescence. Nat. Photon. 55. Wang, H. et al. Perovskite-molecule composite thin films for efficient and
14, 123–128 (2020). stable light-emitting diodes. Nat. Commun. 11, 891 (2020).
24. Wei, Q. et al. Recent progress in metal halide perovskite micro- and 56. Kim, K.-H. et al. Phosphorescent dye-based supramolecules for
nanolasers. Adv. Opt. Mater. 7, 1900080 (2019). high-efficiency organic light-emitting diodes. Nat. Commun. 5,
25. Dutta, A., Behera, R. K., Pal, P., Baitalik, S. & Pradhan, N. Near-unity 4769 (2014).
photoluminescence quantum efficiency for all CsPbX3 (X = Cl, Br, and I) 57. Jurow, M. J. et al. Tunable anisotropic photon emission from self-organized
perovskite nanocrystals: a generic synthesis approach. Angew. Chem. 131, CsPbBr3 perovskite nanocrystals. Nano Lett. 17, 4534–4540 (2017).
5608–5612 (2019). 58. Anaya, M. et al. Best practices for measuring emerging light-emitting diode
26. Wang, N. et al. Perovskite light-emitting diodes based on solution-processed technologies. Nat. Photon. 13, 818–821 (2019).
self-organized multiple quantum wells. Nat. Photon. 10, 699–704 (2016). 59. Zhao, L. et al. Electrical stress influences the efficiency of CH3NH3PbI3
27. Yuan, M. et al. Perovskite energy funnels for efficient light-emitting diodes. perovskite light emitting devices. Adv. Mater. 29, 1605317 (2017).
Nat. Nanotechnol. 11, 872–877 (2016). 60. Snaith, H. J. et al. Anomalous hysteresis in perovskite solar cells. J. Phys.
28. Qin, C. et al. Triplet management for efficient perovskite light-emitting Chem. Lett. 5, 1511–1515 (2014).
diodes. Nat. Photon. 14, 70–75 (2020). 61. Liu, Y. et al. Efficient blue light-emitting diodes based on quantum-confined
29. Gao, Y. et al. Molecular engineering of organic–inorganic hybrid perovskites bromide perovskite nanostructures. Nat. Photon. 13, 760–764 (2019).
quantum wells. Nat. Chem. 11, 1151–1157 (2019). 62. Wang, Q. et al. Efficient sky-blue perovskite light-emitting diodes via
30. Chen, B., Rudd, P. N., Yang, S., Yuan, Y. & Huang, J. Imperfections and photoluminescence enhancement. Nat. Commun. 10, 5633 (2019).
their passivation in halide perovskite solar cells. Chem. Soc. Rev. 48, 63. Miao, Y. et al. Stable and bright formamidinium-based perovskite
3842–3867 (2019). light-emitting diodes with high energy conversion efficiency. Nat. Commun.
31. Ball, J. M. & Petrozza, A. Defects in perovskite-halides and their effects in 10, 3624 (2019).
solar cells. Nat. Energy 1, 16149 (2016). 64. Féry, C., Racine, B., Vaufrey, D., Doyeux, H. & Cinà, S. Physical mechanism
32. Nenon, D. P. et al. Design principles for trap-free CsPbX3 nanocrystals: responsible for the stretched exponential decay behavior of aging organic
enumerating and eliminating surface halide vacancies with softer Lewis light-emitting diodes. Appl. Phys. Lett. 87, 213502 (2005).
bases. J. Am. Chem. Soc. 140, 17760–17772 (2018). 65. Li, C.-H. A., Zhou, Z., Vashishtha, P. & Halpert, J. E. The future is blue
33. Lee, S. et al. Versatile defect passivation methods for metal halide (LEDs): why chemistry is the key to perovskite displays. Chem. Mater. 31,
perovskite materials and their application to light-emitting devices. Adv. 6003–6032 (2019).
Mater. 31, 1805244 (2019). 66. Vashishtha, P. & Halpert, J. E. Field-driven ion migration and color
34. Quan, L. N. et al. Edge stabilization in reduced-dimensional perovskites. instability in red-emitting mixed halide perovskite nanocrystal
Nat. Commun. 11, 170 (2020). light-emitting diodes. Chem. Mater. 29, 5965–5973 (2017).
35. De Roo, J. et al. Highly dynamic ligand binding and light absorption 67. Yuan, Y. & Huang, J. Ion migration in organometal trihalide perovskite and
coefficient of cesium lead bromide perovskite nanocrystals. ACS Nano 10, its impact on photovoltaic efficiency and stability. Acc. Chem. Res. 49,
2071–2081 (2016). 286–293 (2016).
36. Chiba, T. & Kido, J. Lead halide perovskite quantum dots for light-emitting 68. Brivio, F., Caetano, C. & Walsh, A. Thermodynamic origin of
devices. J. Mater. Chem. C. 6, 11868–11877 (2018). photoinstability in the CH3NH3Pb(I1–xBrx)3 hybrid halide perovskite alloy.
37. Bae, W. K. et al. Controlling the influence of Auger recombination on the J. Phys. Chem. Lett. 7, 1083–1087 (2016).
performance of quantum-dot light-emitting diodes. Nat. Commun. 4, 69. Zhao, J. et al. Strained hybrid perovskite thin films and their impact on the
2661 (2013). intrinsic stability of perovskite solar cells. Sci. Adv. 3, eaao5616 (2017).
38. Wehrenfennig, C., Eperon, G. E., Johnston, M. B., Snaith, H. J. & Herz, L. 70. Tennyson, E. M., Doherty, T. A. S. & Stranks, S. D. Heterogeneity at
M. High charge carrier mobilities and lifetimes in organolead trihalide multiple length scales in halide perovskite semiconductors. Nat. Rev. Mater.
perovskites. Adv. Mater. 26, 1584–1589 (2014). 4, 573–587 (2019).
39. Zou, W. et al. Minimising efficiency roll-off in high-brightness perovskite 71. Draguta, S. et al. Rationalizing the light-induced phase separation of mixed
light-emitting diodes. Nat. Commun. 9, 608 (2018). halide organic–inorganic perovskites. Nat. Commun. 8, 200 (2017).
40. Makarov, N. S. et al. Spectral and dynamical properties of single excitons, 72. Li, Z. et al. Modulation of recombination zone position for
biexcitons, and trions in cesium–lead-halide perovskite quantum dots. Nano quasi-two-dimensional blue perovskite light-emitting diodes with efficiency
Lett. 16, 2349–2362 (2016). exceeding 5%. Nat. Commun. 10, 1027 (2019).

Nature Materials | www.nature.com/naturematerials

Review Article NaTuRE MaTERIaLs
73. Vashishtha, P., Ng, M., Shivarudraiah, S. B. & Halpert, J. E. High efficiency 96. Jun, T. et al. Lead-free highly efficient blue-emitting Cs3Cu2I5 with 0D
blue and green light-emitting diodes using Ruddlesden–Popper inorganic electronic structure. Adv. Mater. 30, 1804547 (2018).
mixed halide perovskites with butylammonium interlayers. Chem. Mater. 97. Lu, S. et al. Accelerated discovery of stable lead-free hybrid organic–
31, 83–89 (2019). inorganic perovskites via machine learning. Nat. Commun. 9, 3405 (2018).
74. Yantara, N. et al. Designing the perovskite structural landscape for efficient 98. Gong, X. et al. Highly efficient quantum dot near-infrared light-emitting
blue emission. ACS Energy Lett. 5, 1593–1600 (2020). diodes. Nat. Photon. 10, 253–257 (2016).
75. Si, J. et al. Efficient and high-color-purity light-emitting diodes based on 99. Vasilopoulou, M. et al. Efficient colloidal quantum dot light-emitting diodes
in situ grown films of CsPbX3 (X = Br, I) nanoplates with controlled operating in the second near-infrared biological window. Nat. Photon. 14,
thicknesses. ACS Nano 11, 11100–11107 (2017). 50–56 (2020).
76. Tsai, H. et al. Stable light‐emitting diodes using phase‐pure Ruddlesden– 100. Mao, J. et al. All-perovskite emission architecture for white light-emitting
Popper layered perovskites. Adv. Mater. 30, 1704217 (2018). diodes. ACS Nano 12, 10486–10492 (2018).
77. Tanaka, K. et al. Comparative study on the excitons in lead-halide-based 101. Li, S., Luo, J., Liu, J. & Tang, J. Self-trapped excitons in all-inorganic halide
perovskite-type crystals CH3NH3PbBr3 CH3NH3PbI3. Solid State Commun. perovskites: fundamentals, status, and potential applications. J. Phys. Chem.
127, 619–623 (2003). Lett. 10, 1999–2007 (2019).
78. Bekenstein, Y., Koscher, B. A., Eaton, S. W., Yang, P. & 102. Zhou, N. et al. Perovskite nanowire–block copolymer composites
Alivisatos, A. P. Highly luminescent colloidal nanoplates of perovskite with digitally programmable polarization anisotropy. Sci. Adv. 5,
cesium lead halide and their oriented assemblies. J. Am. Chem. Soc. 137, eaav8141 (2019).
16008–16011 (2015). 103. Bade, S. G. R. et al. Fully printed halide perovskite light-emitting diodes
79. Bertolotti, F. et al. Crystal structure, morphology, and surface termination with silver nanowire electrodes. ACS Nano 10, 1795–1801 (2016).
of cyan-emissive, six-monolayers-thick CsPbBr3 nanoplatelets from X-ray 104. Zhao, L. et al. Influence of bulky organo-ammonium halide additive choice
total scattering. ACS Nano 13, 14294–14307 (2019). on the flexibility and efficiency of perovskite light-emitting device. Adv.
80. Jagielski, J. et al. Scalable photonic sources using two-dimensional lead Funct. Mater. 28, 1802060 (2018).
halide perovskite superlattices. Nat. Commun. 11, 387 (2020). 105. Chou, S.-Y. et al. Transparent perovskite light-emitting touch-responsive
81. Bi, C. et al. Spontaneous self-assembly of cesium lead halide perovskite device. ACS Nano 11, 11368–11375 (2017).
nanoplatelets into cuboid crystals with high intensity blue emission. Adv. 106. Xing, G. et al. Low-temperature solution-processed wavelength-tunable
Sci. 6, 1900462 (2019). perovskites for lasing. Nat. Mater. 13, 476–480 (2014).
82. Linaburg, M. R., McClure, E. T., Majher, J. D. & Woodward, P. M. 107. Gunnarsson, W. B. & Rand, B. P. Electrically driven lasing in metal halide
Cs1–xRbxPbCl3 and Cs1–xRbxPbBr3 solid solutions: understanding octahedral perovskites: challenges and outlook. APL Mater. 8, 030902 (2020).
tilting in lead halide perovskites. Chem. Mater. 29, 3507–3514 (2017). 108. Kim, H. et al. Hybrid perovskite light emitting diodes under intense
83. Jiang, Y. et al. Spectra stable blue perovskite light-emitting diodes. Nat. electrical excitation. Nat. Commun. 9, 4893 (2018).
Commun. 10, 1868 (2019). 109. Jia, Y. et al. Diode-pumped organo-lead halide perovskite lasing in a
84. van der Stam, W. et al. Highly emissive divalent-ion-doped colloidal metal-clad distributed feedback resonator. Nano Lett. 16, 4624–4629 (2016).
CsPb1–xMxBr3 perovskite nanocrystals through cation exchange. J. Am. 110. Zhao, L. et al. Thermal management enables bright and stable perovskite
Chem. Soc. 139, 4087–4097 (2017). light-emitting diodes. Adv. Mater. 32, 2000752 (2020).
85. Wang, H. et al. Trifluoroacetate induced small-grained CsPbBr3 perovskite
films result in efficient and stable light-emitting devices. Nat. Commun. 10, Acknowledgements
665 (2019). We thank O. Inganäs, J. Qin and N. K. Kumawat for discussions. We acknowledge
86. Snaith, H. J. & Hacke, P. Enabling reliability assessments of pre-commercial financial support from a European Research Council Starting Grant (no. 717026), the
perovskite photovoltaics with lessons learned from industrial standards. Swedish Energy Agency Energimyndigheten (no. 48758-1), the Swedish Foundation for
Nat. Energy 3, 459–465 (2018). International Cooperation in Research and Higher Education (no. CH2018-7736), and
87. Senocrate, A. et al. The nature of ion conduction in methylammonium the Swedish Government Strategic Research Area in Materials Science on Functional
lead iodide: a multimethod approach. Angew. Chem. Int. Ed. 56, Materials at Linköping University (Faculty Grant SFO-Mat-LiU No. 2009-00971).
7755–7759 (2017). R.H.F. acknowledges support from the UK Engineering and Physical Sciences Research
88. Xu, M. et al. A transient-electroluminescence study on perovskite Council. J.W. acknowledges financial support from the Joint Research Program between
light-emitting diodes. Appl. Phys. Lett. 115, 041102 (2019). China and the European Union (2016YFE0112000). Y.J. acknowledges support from
89. Sutanto, A. A. et al. In situ analysis reveals the role of 2D perovskite in the National Key Research and Development Program of China (2016YFB0401600)
preventing thermal-induced degradation in 2D/3D perovskite interfaces. and the National Natural Science Foundation of China (21975220, 91833303, 91733302
Nano Lett. 20, 3992–3998 (2020). and 51911530155). X.-K.L. is a Marie Skłodowska-Curie Fellow (no. 798861). F.G. is a
90. Yu, J. C. et al. Improving the stability and performance of perovskite Wallenberg Academy Fellow.
light-emitting diodes by thermal annealing treatment. Adv. Mater. 28,
6906–6913 (2016).
91. Zhao, L. et al. Redox chemistry dominates the degradation and Competing interests
decomposition of metal halide perovskite optoelectronic devices. ACS R.H.F. is a director of Helio Display Materials Ltd. The other authors declare no
Energy Lett. 1, 595–602 (2016). competing interests.
92. Babayigit, A., Ethirajan, A., Muller, M. & Conings, B. Toxicity of
organometal halide perovskite solar cells. Nat. Mater. 15, 247–251 (2016).
93. Ning, W. & Gao, F. Structural and functional diversity in lead-free halide Additional information
perovskite materials. Adv. Mater. 31, 1900326 (2019). Supplementary information is available for this paper at https://doi.org/10.1038/
94. Lanzetta, L., Marin-Beloqui, J. M., Sanchez-Molina, I., Ding, D. & Haque, S. s41563-020-0784-7.
A. Two-dimensional organic tin halide perovskites with tunable visible Correspondence should be addressed to F.G.
emission and their use in light-emitting devices. ACS Energy Lett. 2, Reprints and permissions information is available at www.nature.com/reprints.
1662–1668 (2017).
95. Ma, Z. et al. Electrically-driven violet light-emitting devices based on highly Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
stable lead-free perovskite Cs3Sb2Br9 quantum dots. ACS Energy Lett. 5, published maps and institutional affiliations.
385–394 (2020). © Springer Nature Limited 2020

Nature Materials | www.nature.com/naturematerials