P ER O VS K ITE S

REVIEW morphs of the perovskite structure, either cubic

or partially distorted variants, offer the largest
possible extent of electronic delocalization within
Properties and potential optoelectronic the lead-halide framework and the correspond-
ing desirable semiconducting properties. Ortho-

applications of lead halide rhombic CsPbBr3 (Fig. 1B) is 1 of 15 possible

octahedral tilings of the cubic perovskite struc-
ture that maintains 3D connectivity (8). On the
perovskite nanocrystals contrary, 1D polymorphs of APbX3 compounds
(Fig. 1, C and D) are not appealing semicon-
Maksym V. Kovalenko,1,2* Loredana Protesescu,1,2 Maryna I. Bodnarchuk2* ductors because of their much larger band gaps
and poor electronic transport.
Semiconducting lead halide perovskites (LHPs) have not only become prominent thin-film In bulk LHPs, three 3D polymorphs are typically
absorber materials in photovoltaics but have also proven to be disruptive in the field observed: cubic, tetragonal, and orthorhombic,
of colloidal semiconductor nanocrystals (NCs). The most important feature of LHP NCs is in order of decreasing symmetry. The cubic phase
their so-called defect-tolerance—the apparently benign nature of structural defects, is always the highest temperature phase, and the
highly abundant in these compounds, with respect to optical and electronic properties. phase transitions have well-defined temperatures.
Here, we review the important differences that exist in the chemistry and physics of In the case of NCs, surface effects may adjust the
LHP NCs as compared with more conventional, tetrahedrally bonded, elemental, and relative stabilities of the various polymorphs,
binary semiconductor NCs (such as silicon, germanium, cadmium selenide, gallium and this matter is currently poorly understood.
arsenide, and indium phosphide). We survey the prospects of LHP NCs for optoelectronic At room temperature (RT), all as-synthesized

Downloaded from http://science.sciencemag.org/ on March 19, 2019

applications such as in television displays, light-emitting devices, and solar cells, LHP NCs crystallize into 3D phases as follows:
emphasizing the practical hurdles that remain to be overcome. MAPbI3 NCs are tetragonal; FAPbBr3, MAPbBr3,
and FAPbI3 NCs are pseudocubic; and CsPbBr3

F
ully inorganic lead halide perovskites (LHPs)
with CsPbX3 (X = Cl–, Br–, and I–) stoichi-
ometry have been known since the end of
the 19th century (1), although their perov-
skite crystal structure and semiconductive
nature were not reported until the 1950s (2).
Their hybrid organic-inorganic cousins, MAPbX3
(MA = CH3NH3+; methylammonium) and FAPbX3
[FA = CH(NH2)2+; formamidinium] have been
known since the late 1970s (3). In an optoelec-
tronic context, LHPs and similar tin (Sn)–halide–
based compounds received considerable attention
in the 1990s as channel layers for field-effect tran-
sistors or as active layers in light-emitting diodes
(LEDs) (4). Only a decade later, LHPs came into
the major research spotlight after the demonstra-
tion of highly efficient photovoltaic (PV) devices,
with LHPs as thin-film absorber layers; the cer-
tified power conversion efficiencies of these de-
vices have since improved from <10% to >22%
within the past several years. (5)
Colloidally synthesized nanometer-scale crys-
tals [nanocrystals (NCs)] represent the most recent
type of LHP materials with promising applica-
tions (6, 7). The major attribute of LHP NCs—
the bright and narrow-band photoluminescence
(PL) that is easily tunable from ultraviolet to
near-infrared wavelengths by either halide com-
position or NC size—is concomitant with facile
and low-cost synthesis, which motivates appli-
cations in television displays and lighting. Col-
loidal NCs can be easily processed in solution,
allowing the use of methods such as spin-coating
and additive manufacturing to achieve fast, flex-
1
Institute of Inorganic Chemistry, Department of Chemistry
and Applied Biosciences, ETH Zürich, Vladimir Prelog Weg 1,
CH-8093, Switzerland. 2Empa, Swiss Federal Laboratories for
Materials Science and Technology, Dübendorf, Überlandstrasse
129, CH-8600, Switzerland.
*Corresponding author. Email: mvkovalenko@ethz.ch (M.V.K.);
maryna.bodnarchuk@empa.ch (M.I.B.)
ible, large-area, and cost-effective production of
complex materials and devices by using, for
instance, self-assembly and three-dimensional
(3D) printing. Furthermore, colloidal NCs can
be easily combined with other classes of solution-
processable materials such as polymers, small
molecules, and carbon nanotubes and fullerenes
to produce composites with enhanced optical,
electronic, magnetic, or catalytic functionalities.
“…defect tolerance...is a major
enabling factor…for the
bright photoluminescence
of lead halide perovskite
nanocrystals.”
The saying goes, “You can’t teach an old dog
new tricks,” but for the case of LHP NCs, the
opposite applies: “You can’t teach a new dog old
tricks.” LHP NCs are vastly different from con-
ventional semiconductor NCs such as CdSe and
InP, also known as colloidal quantum dots (QDs).
Both a new theoretical mindset as well as a dis-
tinct experimental framework need to be devel-
oped in order to realize their full potential as
versatile photonic materials.
Basic structural chemistry, synthesis,
and self-assembly
The crystal structure of LHPs is analogous to
oxide perovskites. The parent motif is a cubic
lattice consisting of corner-sharing [PbX6] octahe-
dra connected in three dimensions (Fig. 1A). The
large cavity between octahedra (the A-site) is
occupied by one or a mixture of three large cat-
ions [Cs+, CH3NH3+, or CH(NH2)2+], yielding an
overall composition of APbX3. Only 3D poly-
and CsPbI3 NCs are orthorhombic. However, the
3D polymorphs of FAPbI3 and CsPbI3 NCs, two
compounds of primary interest for near-infrared
emission (with band gaps at 840 and 710 nm,
respectively), are metastable at RT. Because sizes
of A-site ions are suboptimal for 3D polymorphs
of FAPbI3 and CsPbI3 (FA+ being too large and Cs+
too small), bulk forms of these compounds crys-
tallize into 1D hexagonal and 1D orthorhombic
structures, respectively (Fig. 1, C and D). This prob-
lem, termed the “perovskite red wall,” has been
addressed by forming mixed-cation compounds
such as Cs1–xFAxPbX3 NCs (9).
LHPs are multinary halide salts with substan-
tially ionic bonding character that enables their
facile formation at low temperatures. The syn-
thesis of colloidal LHP NCs (defined here as
freely suspended crystallites <20 nm in at least
one dimension) is a surfactant-controlled co-
precipitation of ions that proceeds with fast
kinetics even at RT. The widespread study of
LHP NCs stems from the simplicity of their syn-
thesis, which is typically possible just by mixing
the reagents in an open beaker under ambient
atmosphere, and from their excellent emissive
properties. At RT, the PL of LHP NCs spans the
entire visible spectral range and exhibits narrow
linewidths not exceeding 100 meV [measured as
the full width at half maximum (FWHM)]—for
example, 12-nm width in the blue limit of the
visible range (CsPbCl3), 20-nm width in green at
~520 nm (CsPbBr3 NCs), and 40- to 45-nm width
in red at ~690 nm (CsPbI3). The PL peak position
(color) is tunable by adjusting the NC compo-
sition (Fig. 1, E and F)—the Cl:Br or Br:I ratio
and A-cation—as well as by altering the size and
shape. The corresponding absorption spectra
are well-structured and offset by a slight Stokes
shift (Fig. 1G), as in conventional CdSe QDs. The
PL lifetimes of LHP NCs fall in the nanosecond
range and generally increase with decreasing
band-gap energy. A multitude of synthesis routes
have been developed to control the size, shape,

Kovalenko et al., Science 358, 745–750 (2017) 10 November 2017 1 of 6

 NCs, synthesized either directly (16) or doped by
means of postsynthetic cation exchange (17). The
characteristic property of such NCs is the efficient
transfer of energy from the exciton to the im-
purity atom followed by the emergence of PL
at lower energies, creating a large Stokes shift.

Structural lability, intrinsic defects, and

defect tolerance
Perovskite oxides have had enormous techno-
logical impact in the past 100 years, serving in
catalysis, high-critical temperature superconduc-
tivity, dynamic random-access memory, colossal
magnetoresistance, ferromagnets, piezoelectrics,
ferroelectrics, and multiferroics. In part, their
commercial success arises from the rigid, ther-
mally stable nature of their crystal structure.
LHPs are rather different; the lower charges of
the constituent ions in halide perovskites (half
that of oxides) can alone reduce the crystal lattice
energy by a factor of approximately four based

Downloaded from http://science.sciencemag.org/ on March 19, 2019

on the Glasser generalization of the Kapustinskii
equation for mixed ion systems (18). This difference
leads to much lower melting points, Tm, in halide
perovskites [for example, Tm ≈ 570°C for CsPbBr3
(19) and Tm ≈ 2000°C for CaTiO3]. A further reduc-
tion in lattice energy is observed in LHPs with larger
ions (for example, Tm ≈ 460°C for CsPbI3) (20).
Similarly, the formation energy of vacancies,
the major type of point defect in halide perov-
skites, is also drastically reduced, to <0.5 eV for
Schottky defects versus at least 2 to 3 eV in
Fig. 1. Basics of colloidal LHP NCs. (A) Cubic (MAPbBr3, FAPbBr3, Pm3m  space group) and CaTiO3, leading to an enormous concentration
(B) orthorhombically distorted (CsPbBr3, Pnma space group) 3D perovskite lattices and, for of Schottky-type A-site and halide vacancies in
comparison, nonperovskite 1D polymorphs, formed by the (C) face- or (D) edge-sharing of LHPs (up to several atomic percent) (21). This
octahedra. (E) Survey PL spectra and (F) the corresponding photographs (under mixed daylight large concentration of halide vacancies enables
and UV excitation) of colloids of composition-tuned APbX3 NCs. (G) Absorption and PL spectra the high mobility of halide anions, leading to
of 8-nm colloidal CsPbBr3 NCs, exhibiting quantum-size effects and three well-resolved optical consequences that can be either interpreted as
transitions. (H) High-resolution image of a single CsPbBr3 NC by means of high-angle annular positive (such as facile anion exchange) or neg-
dark-field scanning transmission electron microscopy (HAADF-STEM). (I) Photograph (excitation ative (such as photoinduced ion separation, ionic
wavelength lexc = 365 nm) of highly luminescent CsPbX3 NC-polymethylmethacrylate monoliths conduction and related electronic noise, and hys-
obtained by use of Irgacure 819 as the photoinitiator for polymerization (7). [Reproduced with teresis in PV cells). Other defects such as grain
permission from (7)] (J) Photograph of mesoporous silica impregnated with CsPbBr3 under UV boundaries, dislocations, stacking faults, and twin
illumination (10). [Reproduced with permission from (10)] planes are also promoted by the low lattice energy.
There is ample experimental evidence for
highly dynamic disorder in LHP NCs (Fig. 3) (22),
and composition of LHP NCs, of which the most Beyond the mixed-cation or mixed-anion which is seen as the formation of multiple crys-
common morphology is that of cubes (Fig. 1H). compositions, LHP NCs can also be doped with talline domains separated by twin planes. Over-
Noncolloidal synthesis pathways are possi- external impurities. Doping is the controlled, all, this dynamic disorder often makes even the
ble as well, such as templated growth within intentional incorporation of impurity atoms so basic assignment of the crystal structure to a
the nanosized pores of mesoporous silica (Fig. as to alter favorably the electronic and optical particular space group a nontrivial task with
1J) (10). properties of the material. For example, bulk, high- conventional laboratory powder x-ray diffraction
LHP NCs, primarily CsPbX3, have recently purity semiconductors are insulating but, upon (XRD) and accounts for the initially incorrect
been commercialized by a number of companies, the addition of a small quantity of dopant, be- assignment of CsPbX3 NCs to a cubic crystal
such as Quantum Solutions (Saudi Arabia) (11), come highly conductive with a corresponding structure, by our group and by others. Techniques
Avantama AG (Switzerland) (12), and PlasmaChem p- or n-type conductivity. The optical properties such as pair-distribution function or Debye total
GmbH (Germany) (13). of such a material, including its PL, can also be scattering analysis are expected to assist in un-
Whether embedded in polymer films or in po- governed through recombination at the impurity veiling the atomistic structure of LHP NCs.
rous silica (Fig. 1, I and J), LHP NCs hold great atoms. In NCs, adding one impurity atom per Intrinsic point defects, just like external im-
promise for various applications exploiting their sub-10-nm NC corresponds to doping levels as purities, have always been very important in
bright PL. Monodisperse ensembles of cubic high as 1018 to 1019 cm−3. With magnetic dopants, traditional semiconductors (such as Si and GaAs)
LHP NCs (exhibiting a standard deviation in size enhanced interactions with other carriers, or because they act as electronic traps or electronic
of <10%) can readily form NC superlattices by quantum mechanical spins, spin-orientation– dopants (23) and can be highly detrimental even
means of drying-mediated self-assembly (Fig. 2) dependent transport thought to be needed for in low concentrations [parts per million (ppm) or
(14). The packing of individual cubic CsPbBr3 NCs future spintronic devices can be envisaged (15). parts per billion levels for point defects]. CsPbBr3
into a simple cubic superlattice leads to the for- LHP NCs appear to be relatively easily doped, as has also been thoroughly examined with ab
mation of supercubes as large as 10 mm. demonstrated recently for Mn-doped CsPbCl3 initio methods in order to elucidate the atomistic

Kovalenko et al., Science 358, 745–750 (2017) 10 November 2017 2 of 6

 P ER O VS K ITE S
details of its inherently defect-prone structure,
especially to determine the defect formation
energies and their resulting electronic effects.
In particular, point defects in the bulk material
(24), at grain boundaries (25), and on the NC
surfaces (26) were investigated. All three studies
showed that despite being abundant because of
their low formation energy (for example, as
vacancies on A- and X-sites, grain boundaries of
various crystallographic and mutual orientation)
or inherently high specific surface (for example,
as surface sites on NCs), defects in CsPbBr3 NCs
are benign with respect to the electronic and op-
tical properties; they do not form mid-gap trap
states. This defect tolerance, also commonly ob-
served in other LHPs, is a major enabling factor
for highly efficient perovskite PV and for the
bright PL of LHP NCs. LHP NCs are highly lumi-
nescent without any electronic surface passiva-
tion, which is otherwise an absolute necessity for
conventional QDs (such as CdSe and InP).
The difference between the effects of defects
on the electronic properties of conventional
(defect-intolerant) semiconductors and LHPs can
be rationalized as shown in Fig. 4A (defect-related
electronic states indicated in red) (27). In CdSe,
the removal or displacement of a Cd ion leads to
localized nonbonding or weakly bonding orbitals
of Se (28); these orbitals reside deep within the
band gap and act as trap states. Trap state for-
mation is the usual scenario because the band
gap is normally formed between the bonding
[valance band (VB)] and antibonding states [con-
duction band (CB)]. However, in LHPs the band
gap is formed between two sets of antibonding
orbitals, so the vacancies form states residing
within the VB and CB, or at worst are shallow de-
fects. Dangling bonds at the surface of LHP NCs
have similar effects, leading to localized, non-
bonding states. The formation of benign vacancies
suggests the existence of rather benign surfaces,
which was subsequently confirmed for CsPbBr3
with computational studies (Fig. 4B) (26). Last,
an important difference between LHPs and metal
chalcogenides is that the perovskite structure is
highly immune to the formation of antisite and
interstitial point defects, both of which are very
likely to form trap states.
Processability and long-term stability
Four major forms of instability are characteristic
of LHP NCs. First, all LHPs are partially or highly
soluble in polar solvents, which is favorable for
the fabrication of thin films for perovskite PV
and for the convenient low-cost growth of single
crystals of high optoelectronic quality (29). How-
ever, it is troublesome for the long-term struc-
tural integrity of NCs. The solubility is lowest for
CsPbX3. However, even a low but finite solubility
threatens the structural integrity of NCs. Second,
not only is the internal bonding in LHP NCs high-
ly ionic, but the NC-ligand binding is as well, in
addition to being highly dynamic in solutions (30).
The resulting relatively fast ligand desorption,
contrary to the covalent and more static ligand
binding at the surfaces of conventional NCs (31),
renders severe difficulties in the retention of the
Kovalenko et al., Science 358, 745–750 (2017) 10 November 2017
Fig. 2. Self-assembly of CsPbBr3 NCs. (A) Drying-mediated self-assembly of colloidal CsPbBr3 NCs
into a primitive cubic superlattice (supercubes) under ambient conditions. (B) Optical microscopy
image of large supercubes formed upon drying of colloids on a silicon substrate. (C to E) Transmission
electron micrograph of (C) small supercubes formed on thin carbon films, and (D) mono- and (E)
bilayer 2D assemblies (14).
colloidal state and, eventually, also to the reten-
tion of structural integrity during the intense
postsynthetic purification and processing steps.
Third, although LHPs are oxidatively stable com-
pounds, the long-term stability of CsPbX3 may
still be limited in the presence of a combination
of light, moisture, and oxygen. For instance, the
decomposition of MAPbI3 involves not only dis-
sociation into volatile CH3NH2 and HI, but also
oxidation, presumably facilitated by the diffusion
of the photogenerated superoxide anion O2– on
halide vacancy sites (32). Last, the low melting
points of LHPs render densely packed arrays of
LHP NCs prone to sintering.
For the same reason, certain shapes of LHP
NCs, such as platelets and wires, readily lose
their structural integrity. We expect that the sta-
bilization of LHP NCs against these four pro-
cesses will become the major research focus in
this field. This task is somewhat simplified by
their aforementioned defect tolerance; epitaxial
encapsulation by a protective shell is not man-
datory, as is required for conventional QDs. Con-
finement of individual LHP NCs within the thin
dielectric layers of silica or alumina might be
sufficient for preventing sintering and to impart
environmental stability. Conventional synthetic
routes to achieve this end, such as sol-gel methods,
have thus far been limited by the sensitivity of
LHP NCs to alcohols and water. However, this
problem can be mitigated, as recently shown for
integrating CsPbBr3 NCs into silica-alumina
monoliths, by the controlled diffusion of water
traces into a toluene solution containing LHP NCs
and a single-source silica-alumina precursor, di-sec-
butoxyaluminoxytriethoxysilane (33). Atomic-layer
deposition, a solventless method typically used to
deposit thin films of Al2O3 and similar oxides, was
Downloaded from http://science.sciencemag.org/ on March 19, 2019
also successfully used to encapsulate LHP NCs
within an alumina matrix, imparting higher sta-
bility toward air, heat, light, and moisture (34).
Specially designed gas- and moisture-impermeable
polymers have also been shown to be highly effec-
tive matrices for LHP NC stabilization—for exam-
ple, by enabling the full retention of their PL after
several weeks of direct immersion in water (35).
Applications in displays, lighting,
and light-emitting diodes and
environmental impacts
LHP NCs offer high PL quantum yields (QYs)
and highly saturated colors because of their nar-
row emission bandwidths, the attributes required
for wide-color-gamut liquid-crystal displays (LCDs)
and for lighting with a high color-rendering in-
dex. The symmetric and narrow emission bands
of semiconductor NCs offer remarkable color pu-
rity in comparison with that of organic dye mol-
ecules, which are characterized by asymmetrically
broadened PL (with a red tail) caused by the
strong coupling of electronic and vibrational
states. In contrast, LHP NCs offer blue, green,
and red primary colors with an impressive gamut,
achieving up to ~140% of the North American
National Television Standard Committee (NTSC)
specification (Fig. 5A) and even up to 100% of the
newer International Telecommunication Union
Rec. 2020 standard. In the context of LCDs, LHP
NCs can be used as color downconverters in back-
lighting (Fig. 5B); the blue light generated by
standard InGaN LEDs (~460 to 470 nm) can be
transformed into lower-energy emission in green
(530 nm) and in red (630 to 640 nm). For this
purpose, polymer films containing mixtures of
polymer beads with embedded red and green
LHP NCs can be used. To ensure the long-term
3 of 6
retention of high PL QY, barrier films with low
oxygen and water transmission rates can be ap-
plied in such devices.
At present, several major display manufac-
turers (such as Sony, Samsung, and LG) have
already adopted QD technologies based on CdSe-
or InP-based NCs. Other prominent semiconduc-
tors with the potential to emit in the green and
red spectral regions include Si and GaAs; how-
ever, their PL properties are challenging to control
because of difficulties in synthesizing high-quality
colloidal NCs. Overall, there still exists a large
performance gap between green-emissive InP NCs
and CdSe-based or LHP NCs, with respect to lu-
minance, color gamut, operational stability, and
power consumption (36, 37). InP-based NCs have
arguably reached their maximum performance
potential (emission FWHM of 38 to 40 nm), given
the inherent limitations of morphological con-
trol and compositional homogeneity in that sys-
tem. CdSe-based NCs and LHP NCs thus remain
These restrictions apply to every inseparable com-
ponent of the device, such as a polymer film with
embedded NCs. Products that contain higher
quantities of these heavy metals may be subject
to exemption (Exemption 39), each of which is
reviewed regularly to determine whether an al-
ternative heavy metal–free technology has been
developed in the meantime. Cd-based backlight-
ing films have difficulties complying with the
RoHS; thus far, the only RoHS-compliant Cd-
containing QD product is produced by Nanosys
(Hyperion QDs; ≤95 ppm Cd), which combines
green CdSe-based and red InP-based QDs to
achieve emission FWHMs of 25 and 42 nm,
respectively.
LHP NCs appear to be much more likely to be
compliant with the RoHS because of the much
higher limit of 1000 ppm for Pb. This amount is
sufficient to achieve excellent optical perform-
ance, and Pb is already widespread in lead-acid
batteries, solders, and piezoelectric materials.
which are relative newcomers in this applica-
tion, degradation models still need to be devel-
oped, first by individual effects and then combined
effects of the major relevant parameters: temper-
ature, radiation flux, and for applications that
cannot afford the additional expense of protec-
tive barrier films, humidity and oxygen levels.
In a conceptually identical application, solid-
state lighting, the stability requirements for
LHP NCs are substantially more stringent than
in displays and may never be accessible. There
are two major strategies for integrating a color-
converting emitter into lighting devices, referred
to as “on-chip” and “remote phosphor.” In on-
chip configurations, the NC emitter would be
deposited directly onto a powerful LED, leading
to a substantial transfer of heat to the NCs by a
very high local flux of ultraviolet (UV) or blue
light. In remote phosphor configurations, as im-
plied by the name, the NC emitter would be lo-
cated remotely with respect to the LED light

Downloaded from http://science.sciencemag.org/ on March 19, 2019

strong contenders to outperform InP. LHP NCs
already offer emission FWHM values of 20 and
35 nm in green and in red, respectively. In ad-
dition, the so-called Helmholtz-Kohlrausch effect
states that high saturation of colors, caused by
small FWHM, are perceived by human eyes as
brighter than less-saturated ones of equivalent
luminance.
Another major obstacle to the commerciali-
zation of semiconductor NCs, besides competi-
tive technologies, arises from legislation in the
European Union (EU) such as RoHS (“Restriction
of Hazardous Substances,” Directive 2011/65/EU
of the European Parliament). The RoHS directive
regulates the use of heavy metals in electrical
and electronic devices. Other countries, especially
the United States and China, usually eventually
enact legislation that follows EU regulations,
with a delay of several years. As per the RoHS,
any Cd- or Pb-containing technology must con-
tain <100 ppm Cd and <1000 ppm Pb (by weight).
Less than 5 mg of Pb is required to manufacture
a typical LCD TV display (40 to 50 inch) based
on LHP NCs, which corresponds to only several
hundred parts per million in the necessary NC
films.
Although LHP NCs are often reported to have
superior optoelectronic materials for LCD dis-
plays, these claims are rarely reinforced by ap-
propriate tests of their operational stability under
relevant thermal and radiative conditions imposed
by the blue LEDs. Such displays are typically re-
quired to maintain front-of-screen luminance and
color specifications for 20,000 to 30,000 hours
(2.25 to 3.5 years). Various methods exist to sim-
ulate accelerated aging (combining experiment
and theory), such as those proposed by 3M for
testing InP- and CdSe-based NCs (38). For ex-
ample, it is suggested to test CdSe-based NCs in
polymer films under a blue radiative flux of
400 mW/cm2 at 50°C for 150 hours to simulate
30,000 hours of normal operation. For LHP NCs,
source; for example, an NC film could be used
to cover the surface area of the light fixture. Al-
though this design might require a much larger
quantity of NCs, the remote phosphor design
enables a lower operation temperature. In either
case, temperatures of above 100°C are anticipated.
Far greater challenges for LHP NCs lie in the
direct electrical excitation of their emission, as
required for LEDs. Electroluminescence (EL),
the radiative recombination of electrons and
holes injected from electrodes into a thin layer
of LHP NCs (Fig. 5C), can perhaps eventually
replace down-conversion in both displays and
in lighting. In this regard, semiconductor NCs
offer several promising possibilities: solution-
processability and the engineerability of their
electronic properties. For example, although low
exciton binding energies are required for solar
cells, as commonly observed in iodide-based LHPs,
the opposite is true for LEDs. The exciton binding
energies of bulk CsPbBr3 and CsPbI3 are estimated

A B C
twinned regular

A B B A A B A B

Fig. 3. Dynamic structural disorder in LHP NCs. (A) Atomistic (undistorted) orthorhombic structure of a LHP NC. The density and
representation of a single CsPbBr3 NC with polydomain structure. (B) A crystallographic and mutual orientation of these planar defects determine
single twin boundary connecting domains, highlighting the discontinuity of the the observed diffraction pattern; this can cause an inherently orthorhombic
halide sublattice and the coherence of the Pb sublattice. (C) The regular lattice to appear cubic in a powder XRD experiment (22). [Adapted from (22)]

Kovalenko et al., Science 358, 745–750 (2017) 10 November 2017 4 of 6

 P ER O VS K ITE S

Fig. 4. Defect tolerance in LHP NCs. (A) Schematics comparing corresponds to a molecular orbital. Each color indicates the contribution
electronic structures that are defect-intolerant, such as for conventional of a type of atom (or moiety) for a given molecular orbital. (Left) A
semiconductors (for example, CdSe, GaAs, and InP), and defect- charge-neutral NC, generally terminated by Cs+, Br–, and MA ions. The

Downloaded from http://science.sciencemag.org/ on March 19, 2019

tolerant, such as for LHPs (27). Defects do not act as trap states
in LHPs and are therefore benign toward their electronic and optical
properties. [Adapted from (27)] (B) Electronic structure diagrams
for CsPbBr3 NCs at the DFT/PBE level of theory (26), where PBE is
Perdew-Burke-Ernzerhof exchange-correlation functional. Each line
MA ions emulate the oleylammonium ligand, a major capping ligand
in such NCs (30). Upon removal of the MA ions from the NC surface
in the form of MABr (middle) and the additional removal of CsBr
(right) (simulating the effect of washing), a trap-free bandgap
is maintained. [Adapted from (26)]

to be 33 and 15 meV, respectively, by the magneto-
optical measurements (39). Similarly, binding en-
ergies of ~30 meV were measured for large cubic
CsPbBr3 NCs (~11 nm) by using ultrafast tran-
sient absorption measurements (40). The binding
energy increased to 50 meV for smaller (5.5 nm)
NCs (41). Strong quantum confinement in one
dimension, as observed in 3.4-nm-thick (~3 unit
cells) and ~20-nm-wide CsPbBr3 nanoplatelets,
leads to much higher exciton binding energies
of 120 meV, which is associated with the giant
oscillator strengths (42). The higher structural
lability of 2D LHP NCs as compared with cube-
shaped LHP NCs, however, limits their efficient
purification and processing into devices. LEDs
fabricated from colloidal LHP nanoplatelets
(2D NCs) often exhibit EL characteristics, espe-
cially peak positions, similar to those observed in
devices made from molecular precursors (thin
films) or from standard cube-shaped NCs. This
highlights the common difficulties encountered
in retaining preengineered quantum-confined
morphologies in thin-film devices.
Presently, most reported studies of LHP NC
devices have been devoted to green LEDs that
contain cube-shaped CsPbBr3 NCs, showing ex-
ternal quantum efficiencies (EQEs) of up to 6 to
9% and peak luminance of up to >15,000 cd/m2
(43, 44). Red LEDs containing CsPbI3 NCs have
exhibited an EQE of 5.7% at 698 nm (45) and an
EQE of 7.25%, with a peak luminance of 435 cd/m2
at 688 nm (46). Making efficient blue-emissive
LEDs has proven more challenging; in the blue-
green (cyan) region, a maximum EQE of 1.9%
was observed for CsPbBrxCl3–x NCs showing a
rather low peak luminance of 35 cd/m2 at 490 nm
(47). CsPbCl3 NCs, which exhibit the widest band-
gap in the LHP family, have achieved a maxi-
mum EQE of only 0.61%, with a corresponding
luminance of 11 cd/m2 at a deep-blue wavelength
of 404 nm (46).
Looking forward
“I don’t mind your thinking slowly; I mind
your publishing faster than you think.”
—Wolfgang Pauli (48)
The scientific research of LHPs is currently
happening at a fast rate; several thousand publica-
tions were accepted in the peer-reviewed literature
between January 2016 and July 2017 (source:
Web of Science, https://apps.webofknowledge.
com), of which up to 1000 concern nanoscale
forms of LHPs. Typical consequences of this “pub-
lish or perish” environment include, for example,
initially inaccurate crystal structure assignment
of CsPbBr3 and CsPbI3 NCs, as previously dis-
cussed. Another issue concerns inaccurate nomen-
clature; hybrid organic-inorganic LHPs are often
referred to as “organometal,” “organolead,” or
“organometallic,” which are not strictly correct
given the lack of metal-carbon bonding or similar
direct coordination motifs between Pb atoms and
organic moieties (several authors have already
emphasized this point of confusion) (49, 50).
A great challenge with respect to the applica-
tions of LHPs and LHP NCs is for the research
community to present balanced assessments of
the truly relevant performance parameters, rather
than make strong claims as to the future com-
mercial prospects of materials or the superiority
of one class of materials over another. LHPs and
LHP NCs are soft and chemically unstable sub-
stances, and therefore the most obvious research
necessity is to establish methods of their stabili-
zation with respect to light, temperature, and the
environment. These methods should be tested with
appropriate accelerated aging tests, as already exist
for display and photovoltaic technologies.
Several near-future research frontiers for LHP
NCs can be clearly identified. With respect to their
synthesis, progress can be greatly accelerated
by using high-throughput continuous-flow or
segmented-flow microfluidic methods, equipped
with in situ optical characterization, especially
with effective algorithms for the targeted syn-
thesis of NCs with desired optical properties.
Furthermore, the synthesis of NC heterostruc-
tures, in which LHP and other inorganic ma-
terials are combined into a single NC, remains
largely undeveloped. Efficient ligand-exchange,
encapsulation, and matrix-integration appear
harder to accomplish than with conventional
QDs, which are more structurally rigid and chem-
ically robust. To this end, classification of LHP NC
surface coordinations—for example, by using
nomenclature that invokes X-, Y- and Z-ligand
types (28) or its LHP-specific alternative—and
their rational engineering at the molecular level
are paramount.
The lability of LHP NCs can be exploited as an
advantageous property with respect to synthesis.
For example, in the low-cost deposition of thin
films for PV applications (51), the crystallinity
and compositional modulations can be ratio-
nally preprogrammed via the NC surface chem-
istry and deposition conditions. The potential
for LHP NCs in photocatalysis applications in
nonaqueous media is an unexplored area. Be-
cause LHP NCs do not require encapsulation
by wide-band-gap materials to prevent carrier
trapping, the photogenerated carriers are highly
accessible to drive photocatalytic redox reactions

Kovalenko et al., Science 358, 745–750 (2017) 10 November 2017 5 of 6

 10. D. N. Dirin et al., Nano Lett. 16, 5866–5874 (2016).
11. www.qdream.co
12. http://avantama.com
13. www.plasmachem.com
14. M. V. Kovalenko, M. I. Bodnarchuk, Chimia 71, 461–470 (2017).
15. D. J. Norris, A. L. Efros, S. C. Erwin, Science 319, 1776–1779
(2008).
16. A. K. Guria, S. K. Dutta, S. D. Adhikari, N. Pradhan, ACS Energy
Lett. 2, 1014–1021 (2017).
17. G. Huang et al., Adv. Mater. 2017, 29 (2017).
18. L. Glasser, Inorg. Chem. 34, 4935–4936 (1995).
19. C. C. Stoumpos et al., Cryst. Growth Des. 13, 2722–2727
(2013).
20. R. J. Sutton et al., Adv. Energy Mater. 6, 1502458 (2016).
21. J. M. Ball, A. Petrozza, Nat. Energy 1, 16149 (2016).
22. F. Bertolotti et al., ACS Nano 11, 3819–3831 (2017).
23. H. J. Queisser, E. E. Haller, Science 281, 945–950 (1998).
24. J. Kang, L.-W. Wang, J. Phys. Chem. Lett. 8, 489–493 (2017).
25. Y. Guo, Q. Wang, W. A. Saidi, J. Phys. Chem. C 121, 1715–1722
(2017).
26. S. ten Brinck, I. Infante, ACS Energy Lett. 1, 1266–1272 (2016).
27. R. E. Brandt et al., Chem. Mat. 29, 4667–4674 (2017).
28. A. J. Houtepen, Z. Hens, J. S. Owen, I. Infante, Chem. Mater.
29, 752–761 (2017).
29. D. Shi et al., Science 347, 519–522 (2015).
30. J. De Roo et al., ACS Nano 10, 2071–2081 (2016).
Fig. 5. Toward applications of LHP NCs in television displays and LEDs. (A) PL spectra of

Downloaded from http://science.sciencemag.org/ on March 19, 2019

31. J. Owen, Science 347, 615–616 (2015).
CsPbX3 NCs plotted on CIE chromaticity coordinates (black points) compared with common color 32. N. Aristidou et al., Nat. Commun. 8, 15218 (2017).
standards (LCD television, dashed white line, and NTSC television, solid white line), reaching 140% 33. Z. Li, L. Kong, S. Huang, L. Li, Angew. Chem. Int. Ed. 56,
8134–8138 (2017).
of the NTSC color standard (solid black line) (7). [Reproduced with permission from (7)] (B) Operation 34. A. Loiudice, S. Saris, E. Oveisi, D. T. L. Alexander, R. Buonsanti,
principle of a QD LCD display, showing blue emission from standard InGaN LEDs transmitted by the Angew. Chem. Int. Ed. 56, 10696–10701 (2017).
diffuser into a polymer film containing LHP NCs, undergoing partial conversion into green and red PL. The 35. S. N. Raja et al., ACS Appl. Mater. Interfaces 8, 35523–35533
mixture of colors is then incident upon a standard LCD matrix, containing liquid crystals and color filters (2016).
36. http://informationdisplay.org/IDArchive/2016/
to define the mixing ratios of the three primary colors so as to achieve any color within the color gamut. NovemberDecember.aspx
Green and red LHP NCs are proposed to be separated into different polymer layers or beads in order 37. S. Sadasivan, K. Bausemer, S. Corliss, R. Pratt, Proc. SID
to avoid inter-NC anion exchange. (C) Schematic of a three-color LED pixel with LHP NCs as the emissive Symp. Dig. Tech. Papers 47, 333–335 (2016).
layer. The hole and electron injecting materials can be inorganic (such as conductive oxides or metals) 38. J. Thielen et al., Proc. SID Symp. Dig. Tech. Papers 47,
336–339 (2016).
or organic (such as small molecules or conductive polymers). LEDs have fewer layers in their device 39. Z. Yang et al., ACS Energy Lett. 2, 1621–1627 (2017).
architecture than LCDs and can therefore afford thinner devices and make more efficient use of the light. 40. J. Aneesh et al., J. Phys. Chem. C 121, 4734–4739 (2017).
41. A. Shinde, R. Gahlaut, S. Mahamuni, J. Phys. Chem. C 121,
14872–14878 (2017).
42. J. Li et al., J. Phys. Chem. Lett. 8, 1161–1168 (2017).
of various substrates. An additional feature of photophysics. However, faithful optical and elec- 43. J. Li et al., Adv. Mater. 29, 1603885 (2017).
LHPs is their much slower cooling of photo- tronic analogs of LHPs remain elusive. Some of 44. T. Chiba et al., ACS Appl. Mater. Interfaces 9, 18054–18060
(2017).
generated hot carriers (at ~1 to 10 meV/ps) than the major difficulties encountered thus far have 45. G. Li et al., Adv. Mater. 28, 3528–3534 (2016).
that in conventional semiconductors (for example, been in the oxidative instabilities of Sn and Ge 46. X. Zhang et al., J. Phys. Chem. Lett. 7, 4602–4610 (2016).
up to 1 eV/ps in GaAs) (52). This slower cooling analogs; the inability of Sb and Bi halides to 47. J. Pan et al., Adv. Mater. 28, 8718–8725 (2016).
may open new possibilities to harness the energy form 3D extended frameworks; and in so-called 48. S. Ratcliffe, Little Oxford Dictionary of Quotations (Oxford Univ.
Press, 2012).
of hot carriers for efficient PV and other applica- double perovskites of composition A2M+M3+X6 49. B. Saparov, D. B. Mitzi, Chem. Rev. 116, 4558–4596 (2016).
tions. The peculiarities of the exciton fine structure (M+ = Ag+ or Cu+ and M3+ = In3+, Sb3+, or Bi3+, 50. P. R. Varadwaj, Helv. Chim. Acta 100, e1700090 (2017).
of LHP NCs, such as bright triplet excitons—leading the structural analogs of 3D-APbX3), the prohib- 51. A. Swarnkar et al., Science 354, 92–95 (2016).
to ~20 and ~1000 times faster emission than any itively large or indirect band gaps, oxidative in- 52. K. Miyata, T. L. Atallah, X.-Y. Zhu, Sci. Adv. 3, e1701469 (2017).
53. M. A. Becker et al., arXiv:1707.03071 [cond-mat.mes-hall]
other semiconductor NCs at room and cryogenic stability (for M+ = In+), or difficulty in synthesis (10 July 2017).
temperatures, respectively—become the focus of because of competition with more thermody- 54. J. Calabrese et al., J. Am. Chem. Soc. 113, 2328–2330
theoretical studies (53). Last, a topic that remains namically stable ternary phases (such as Cs3Bi2I9). (1991).
completely unexplored is the rational control of Another obstacle is that the predictive power 55. C. C. Stoumpos et al., Chem. Mater. 28, 2852–2867 (2016).
56. Z. Yuan, Y. Shu, Y. Xin, B. Ma, Chem. Commun. 52, 3887–3890
charge transport in densely packed assemblies of of high-throughput computational screening (2016).
LHP NCs. Beyond APbX3-type (3D) perovskites, is generally limited by the inability of density 57. S. Gonzalez-Carrero, G. M. Espallargas, R. E. Galian,
an extremely active area of research is in 2D functional theory–based methods to discover J. Perez-Prieto, J. Mater. Chem. A Mater. Energy Sustain. 3,
perovskites, such as Ruddlesden-Popper phases, metastable phases. However, most inorganic 14039–14045 (2015).
58. M. C. Weidman, A. J. Goodman, W. A. Tisdale, Chem. Mater. 29,
(RNH3)2(MA)n–1PbnX3n+1 (R = C4H9, C9H19–, compounds are actually metastable, which leaves 5019–5030 (2017).
or Ph–CH2CH2– and X– = Br– or I–) (54, 55), in ample opportunity for future experimental seren-
which the potential library of compositions and dipity in the discovery of new LHP-like materials. AC KNOWLED GME NTS
structures is believed to be much greater. The M.V.K. is very grateful to his former and present co-workers and
synthesis of 2D perovskites in the form of col- RE FERENCES AND NOTES collaborators, whose names can be found on joint publications.
1. H. L. Wells, Z. Anorg. Allg. Chem. 3, 195–210 (1893). This work was financially supported by the European Research
loidal NCs becomes an additional exciting op- Council (ERC) under the European Union’s Seventh Framework
2. C. K. Møller, Nature 182, 1436 (1958).
portunity (56–58). 3. D. Weber, Z. Naturforsch. C 33, 1443–1445 (1978). Program (grant agreement 306733, ERC Starting Grant
There is an urgent need to explore alternative 4. D. B. Mitzi, in Progress in Inorganic Chemistry (John Wiley & “NANOSOLID”). M.I.B. acknowledges the Swiss National Science
metal halide compounds that comprise environ- Sons, 2007), pp. 1–121. Foundation (SNF Ambizione Energy grant PZENP2_154287). We thank
5. www.nrel.gov/pv/assets/images/efficiency-chart.png N. Stadie for reading the manuscript; N. Schwitz for providing
mentally friendly elements instead of Pb. The photographs of colloidal LHP NCs; and F. Bertolotti and I. Infante for
6. L. C. Schmidt et al., J. Am. Chem. Soc. 136, 850–853 (2014).
success of LHPs in PV has naturally led to an ex- 7. L. Protesescu et al., Nano Lett. 15, 3692–3696 (2015). the help in preparing Figs. 3 and 4B, respectively.
tensive experimental and computational search 8. C. J. Howard, H. T. Stokes, Acta Crystallogr. B 54, 782–789 (1998).
for new compounds with similar defect-tolerant 9. L. Protesescu et al., ACS Nano 11, 3119–3134 (2017). 10.1126/science.aam7093

Kovalenko et al., Science 358, 745–750 (2017) 10 November 2017 6 of 6

Properties and potential optoelectronic applications of lead halide perovskite nanocrystals
Maksym V. Kovalenko, Loredana Protesescu and Maryna I. Bodnarchuk

Science 358 (6364), 745-750.

DOI: 10.1126/science.aam7093

Downloaded from http://science.sciencemag.org/ on March 19, 2019

ARTICLE TOOLS http://science.sciencemag.org/content/358/6364/745

RELATED http://science.sciencemag.org/content/sci/358/6364/732.full
CONTENT
http://science.sciencemag.org/content/sci/358/6364/734.full
http://science.sciencemag.org/content/sci/358/6364/739.full
http://science.sciencemag.org/content/sci/358/6364/751.full
http://science.sciencemag.org/content/sci/358/6364/768.full
http://science.sciencemag.org/content/sci/358/6367/1192.full
REFERENCES This article cites 47 articles, 6 of which you can access for free
http://science.sciencemag.org/content/358/6364/745#BIBL

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Use of this article is subject to the Terms of Service

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement of
Science, 1200 New York Avenue NW, Washington, DC 20005. 2017 © The Authors, some rights reserved; exclusive
licensee American Association for the Advancement of Science. No claim to original U.S. Government Works. The title
Science is a registered trademark of AAAS.