LETTER https://doi.org/10.

1038/s41586-018-0451-1

All-inorganic perovskite nanocrystal scintillators

Qiushui Chen1, Jing Wu2, Xiangyu Ou3,4, Bolong Huang5, Jawaher Almutlaq6, Ayan A. Zhumekenov6, Xinwei Guan6,
Sanyang Han1, Liangliang Liang1, Zhigao Yi1, Juan Li3,4, Xiaoji Xie7, Yu Wang8, Ying Li8, Dianyuan Fan8, Daniel B. L. Teh9,
Angelo H. All9,10, Omar F. Mohammed6, Osman M. Bakr6, Tom Wu11, Marco Bettinelli12, Huanghao Yang3,4*, Wei Huang7,13,14,15*
& Xiaogang Liu1,8,16*

The rising demand for radiation detection materials in many highly sensitive to X-rays and integrable to flexible substrates remains
applications has led to extensive research on scintillators1–3. The a daunting challenge.
ability of a scintillator to absorb high-energy (kiloelectronvolt- Recently, bulk crystals of organic–inorganic hybrid perovskites have
scale) X-ray photons and convert the absorbed energy into low- been found to exhibit large X-ray stopping power11–14 and the ability
energy visible photons is critical for applications in radiation of efficiently converting X-ray photons into charge carriers15–18. The
exposure monitoring, security inspection, X-ray astronomy and direct photon-to-current conversion can be attributed to the heavy Pb
medical radiography4,5. However, conventional scintillators are atom19 and large electron–hole diffusion lengths available in organic–
generally synthesized by crystallization at a high temperature inorganic hybrid perovskites20–25. We reason that caesium lead halide
and their radioluminescence is difficult to tune across the visible perovskite nanocrystals26, which feature heavy constituent elements
spectrum. Here we describe experimental investigations of a series and tunable electronic bandgaps in the visible range, could be a prom-
of all-inorganic perovskite nanocrystals comprising caesium ising candidate for high-efficiency X-ray scintillation. An appealing
and lead atoms and their response to X-ray irradiation. These aspect of these perovskite nanocrystals is that their unique electronic
nanocrystal scintillators exhibit strong X-ray absorption and structures render highly emissive triplet excited states27 and anomalous
intense radioluminescence at visible wavelengths. Unlike bulk fast emission rates28. By virtue of the effect of quantum confinement
inorganic scintillators, these perovskite nanomaterials are solution- and increased overlap of electron and hole wavefunctions, the spatial
processable at a relatively low temperature and can generate X-ray- distribution of luminescence centres and X-ray-generated excitons
induced emissions that are easily tunable across the visible spectrum can be confined within the Bohr radius of the nanocrystals. Here we
by tailoring the anionic component of colloidal precursors during report experimental investigations of multicolour X-ray scintillation
their synthesis. These features allow the fabrication of flexible from a series of all-inorganic perovskite nanocrystals and demonstrate
and highly sensitive X-ray detectors with a detection limit of their use for ultrasensitive X-ray sensing and low-dose digital X-ray
13 nanograys per second, which is about 400 times lower than technology.
typical medical imaging doses. We show that these colour-tunable In a typical bulk scintillator material, incident X-ray photons can
perovskite nanocrystal scintillators can provide a convenient interact with heavy atoms (for example, Pb, Tl or Ce) to produce a large
visualization tool for X-ray radiography, as the associated image number of hot electrons through the photoelectric effect8. These charge
can be directly recorded by standard digital cameras. We also carriers are quickly thermalized to form low-energy excitons, which can
demonstrate their direct integration with commercial flat-panel subsequently be transported to defect centres or activators for radia-
imagers and their utility in examining electronic circuit boards tive luminescence (Extended Data Fig. 1a). We thus predict that high-
under low-dose X-ray illumination. energy (kiloelectronvolt-scale) X-ray photons can be converted to
The nature of the atomic constituents of a scintillator plays an numerous low-energy visible photons via direct bandgap emissions in
important role in the radioluminescence process of the material lead halide perovskite nanocrystals (Fig. 1a). To validate this hypothesis,
because X-ray absorption increases exponentially with atomic num- we prepared a series of perovskite nanocrystals (CsPbX3, with X = Cl,
ber6. Although a wide range of scintillation materials containing heavy Br or I) by controlling the reaction of Cs-oleate with different PbX2 pre-
atoms have been characterized in detail for efficient X-ray scintil- cursors via a hot-injection solution method29 (Extended Data Fig. 2).
lation, almost all of these materials are bulk crystals and grown by Transmission electron micrograph imaging reveals a cubic shape of the
the Czochralski method7 at temperatures above 1,700 °C. For bulk- as-synthesized nanocrystals, with an average size of 9.6 nm (Fig. 1b).
form scintillators, such as PbWO4 and Bi4Ge3O12, a certain distance Remarkably, under X-ray beam excitation the perovskite quantum dots
of exciton migration is typically needed to transport charge carriers (QDs) yield narrow and colour-tunable emissions (Fig. 1c, Extended
for subsequent trapping by luminescence centres8. However, exces- Data Fig. 3). This unique property allows multicolour, high-efficiency
sive exciton migration is detrimental because it can cause either X-ray scintillation to be realized (Fig. 1d, e, Extended Data Table 1). By
radioluminescence afterglow or low-efficiency X-ray scintillation. In contrast, the radioluminescence spectrum of conventional bulk scintil-
addition, conventional activator-doped scintillators, such as thalli- lators (CsI:Tl, PbWO4, YAlO3:Ce and Bi4Ge3O12) is almost invariable
um-activated CsI (CsI:Tl) and cerium-activated YAlO3 (YAlO3:Ce), and exhibits a wide emission peak with a large full-width at half-
cannot produce tunable scintillation because of their fixed transition maximum (Extended Data Fig. 1b). This inherent limitation of conven-
energies9,10. Despite enormous efforts, the development of scintil- tional scintillators makes it difficult to achieve multicolour visualization
lating materials that are low-temperature- and solution-processable, of X-ray irradiation.
1
Department of Chemistry, National University of Singapore, Singapore, Singapore. 2School of Science, China University of Geosciences, Beijing, China. 3MOE Key Laboratory for Analytical Science
of Food Safety and Biology, Fuzhou University, Fuzhou, China. 4State Key Laboratory of Photocatalysis on Energy and Environment, Fuzhou University, Fuzhou, China. 5Department of Applied
Biology and Chemical Technology, The Hong Kong Polytechnic University, Hong Kong, China. 6Division of Physical Sciences and Engineering, King Abdullah University of Science and Technology,
Thuwal, Saudi Arabia. 7Institute of Advanced Materials, Nanjing Tech University, Nanjing, China. 8SZU-NUS Collaborative Innovation Center for Optoelectronic Science and Technology, Shenzhen
University, Shenzhen, China. 9Singapore Institute for Neurotechnology, National University of Singapore, Singapore, Singapore. 10Department of Biomedical Engineering, Johns Hopkins School
of Medicine, Baltimore, MD, USA. 11School of Materials Science and Engineering, University of New South Wales, Sydney, New South Wales, Australia. 12Luminescent Materials Laboratory, DB,
University of Verona, Verona, Italy. 13Key Laboratory for Organic Electronics and Information Displays, Nanjing University of Posts and Telecommunications, Nanjing, China. 14Institute of Advanced
Materials, Nanjing University of Posts and Telecommunications, Nanjing, China. 15Shaanxi Institute of Flexible Electronics, Northwestern Polytechnical University, Xi’an, China. 16Center for
Functional Materials, NUS Suzhou Research Institute, Suzhou, Jiangsu, China. *e-mail: hhyang@fzu.edu.cn; iamwhuang@nwpu.edu.cn; chmlx@nus.edu.sg

N A T U R E | www.nature.com/nature
© 2018 Springer Nature Limited. All rights reserved.
 RESEARCH LETTER

a b c Emission wavelength (nm)

010 400 500 600 700 800
110
100 Excitation: 278 μGy s–1 X-rays
0.58 nm Sample 1 Sample 12

Radioluminescence intensity
X-ray 2 nm

hQ = 1.7–3.0 eV

Cs Pb Cl, Br or I
10 nm

d e
CsI:Tl 185 520 CsPbBr3
0.8
CsPbBr3 QDs 100 540 3.2 3.0 2.8 2.6 2.4 2.2 2.0 1.8 1.6
5 6 Emitted photon energy (eV)
YAlO3:Ce crystal 19.4 560 f
0.6 Bright field X-ray illumination
Bi4Ge3O12 crystal 7.0 4 580
7
y

PbWO4 crystal 0.35

0.4 8 600
CH3NH3PbBr3 crystal 0.042 9
10–12
CdTe QDs 0.016 700
0.2
3 CsPbI3
CsPbBr3 crystal 0.013
2
470 1
Carbon dots 0 0 CsPbCl3
380
10 mm 10 mm
X-ray-induced emission 0 0.2 0.4 0.6 0.8
x
Fig. 1 | Full-colour radioluminescence from perovskite nanocrystal CsPbBr3 (6), CsPbBr2I (7), CsPbBr1.8I1.2 (8), CsPbBr1.5I1.5 (9), CsPbBr1.2I1.8
scintillators. a, Schematic representation of X-ray-induced luminescence (10), CsPbBrI2 (11) and CsPbI3 (12). The insets show photographs of
of energy hν (where h is the Planck constant and ν is the frequency), the thin-film samples 3, 6 and 9, which emit blue, green and red colours,
generated in an all-inorganic perovskite lattice with a cubic crystal respectively, upon X-ray irradiation. d, Comparison of the optical
structure (see main text for details). b, Low-resolution transmission sensitivity of various scintillator materials in response to exposure to
electron microscopy (TEM) image of the as-synthesized CsPbBr3 X-rays produced at a voltage of 10 kV. e, CIE (Commission Internationale
nanocrystals. The inset shows a high-resolution TEM image of a single de l’Eclairage) chromaticity coordinates of the X-ray-induced visible
CsPbBr3 nanocrystal and the corresponding electron diffraction pattern emissions measured for samples 1–12. f, Multicolour X-ray scintillation
along the [100] zoom axis. c, Tunable luminescence spectra of the (left, bright-field imaging; right, X-ray illumination at a voltage of 50 kV)
perovskite QDs under X-ray illumination with a dose rate of 278 µGy s−1 at from three types of perovskite nanocrystal scintillator (orange, CsPbBr2I;
a voltage of 50 kV. The material compositions of samples 1–12 are CsPbCl3 green, CsPbBr3; blue, CsPbClBr2).
(1), CsPbCl2Br (2), CsPbCl1.5Br1.5 (3), CsPbClBr2 (4), CsPbCl2.5Br0.5 (5),

Inspired by the bandgap-tunable perovskite nanocrystal scintillators, that typical bulk single crystals of CsPbBr3 and CH3NH3PbBr3 do not
we successfully developed a flexible prototype device for multicolour exhibit noticeable visible emission under the same experimental con-
X-ray scintillation through a combination of solution processing and ditions (Fig. 1d, Extended Data Fig. 5). The noteworthy scintillation
soft lithography (Fig. 1f, Extended Data Fig. 3d, e). The fabrication performance of CsPbBr3 nanocrystals with respect to their bulk coun-
of the X-ray-sensing device was made possible by casting the oleate- terparts presents a compelling case for investigating the origins of the
capped perovskite nanocrystals onto the flexible substrate of interest. scintillation process in our system. This process can be explained in
This flexible substrate allowed rapid X-ray multicolour visualiza- part by the lack of exciton confinement in the bulk form, in which
tion (Supplementary Video 1), which is inaccessible by current bulk discrete or quantized energy levels that give access to visible emission
scintillators. cannot be generated31.
We then compared the sensitivity of the perovskite nanocrystals to We further investigated experimentally and theoretically the physical
X-ray illumination with that of several of the most widely used com- processes that govern the interaction between X-rays and perovskite
mercial bulk scintillators (CsI:Tl, PbWO4, YAlO3:Ce and Bi4Ge3O12). nanocrystals. As shown in Fig. 2a, we compared the absorption coeffi-
We used low-dose irradiation of 5.0 µGy s−1 (all doses refer to doses in cient of the CsPbBr3 nanocrystals (highest atomic number Zmax = 82;
air) at 10 kV and 5 µA and found that the ability of CsPbBr3 nanocrystal Kα = 88.0 keV) as a function of X-ray photon energy with two types of
thin films (thickness of about 0.1 mm) to convert X-ray photons into conventional QD (CdTe, Zmax = 52, Kα = 31.8 keV; carbon, Zmax = 6,
visible luminescence is comparable to that of high-efficiency CsI:Tl Kα = 0.285 keV). The nature of heavy atomic constituents is critically
bulk scintillators (thickness of 5.0 mm), whereas it compares much important for efficient X-ray scintillation, because X-ray absorption
4
more favourably (more intense by a factor of 5 or more) than other bulk scales with the effective atomic number, Zeff, as Zeff /AE 3 , where A is
scintillators, including PbWO4, YAlO3:Ce and Bi4Ge3O12 (Fig. 1d). This the atomic mass and E is the X-ray photon energy6. We thus speculate
superior performance is attributed to the large X-ray stopping power that the Pb-based perovskite nanocrystals are much more suitable for
and high emission quantum yields of the lead halide QDs. Notably, efficient X-ray absorption than QDs without the Pb component. We
conventional QDs (for example, CdTe QDs and carbon dots) exhibit carried out an X-ray photoelectron spectroscopic investigation to
low-efficiency X-ray-induced luminescence possibly due to weak X-ray record the kinetic process of electrons escaping from the CsPbBr3 nano-
absorption30, and thus are not suitable for practical use as scintillators crystal upon irradiation with soft X-rays (Fig. 2b). To reveal the photo-
(Fig. 1d, Extended Data Fig. 4). As a point of comparison, we also found ionization nature of the X-ray scintillation process under study, we

N A T U R E | www.nature.com/nature
© 2018 Springer Nature Limited. All rights reserved.
 LETTER RESEARCH

a 104 b e
Heavy atom
3 e–
Cs 3d
Absorbance (cm2 g–1)

Secondary
102 electron

Intensity (a.u.)
X-rays Ejection
2 Pb 4f e–
Carbon CdTe CsPbBr3 K
L
M Thermalization
100 1 Br 3d

Photoelectric effect
10–2 0 e–

Energy
100 101 102 103 800 600 400 200 e– e– e– S e–
Electron binding energy (eV) T
X-ray energy (keV)

c d 6
Pb L1
Cs K
4 3p hQ
M
Scintillation intensity

M 3p
Energy (eV)

2 Fast radiative
R recombination
Bandgap
X 2p
Pb L2 0 Γ L
Br K 2s

–2 X-rays
Pb L3
Br K h+ Hole
K 1s
10 15 20 25 30 35 Γ R M
X-ray energy (keV) Position in the Brillouin zone Ionizing radiation Radioluminescence

f K = 0.9752
g h
X-rays X-rays, 1.4 × 107 photons mm−2 s−1
Energy density (meV Å–1)

15
–200 Without X-rays
d Au
e– Current (nA)
Au 10
Pe
rov h+
–204 ski
te
QD
s 5
10.32 Å

–208 0
5 10 15 20 25 0 20 40 60
Particle distance, d (Å) Voltage (V)

Fig. 2 | Mechanistic investigation of X-ray energy conversion by radiation creates an energetic electron and a hole in an inner electronic
perovskite nanocrystals. a, Measured absorption spectra of CsPbBr3, CdTe shell). Subsequently, the ejected high-energy electron produces secondary
and carbon as a function of X-ray energy. The attenuation coefficients high-energy electrons. The generated hot charge carriers then undergo
were obtained from ref. 33. b, X-ray photoelectron spectroscopic data thermalization and produce low-energy excitons. Next, fast radiative
of the CsPbBr3 nanocrystals plotted against the binding energy of the recombination takes place, producing radioluminescence of energy hν in
electron. The photoemission peaks Cs 3d, Pb 4f and Br 3d are indicated. either a singlet (S) or triplet (T) state at the electronic band edge. f, Energy
a.u., arbitrary units. c, Measurement of X-ray-induced luminescence from density on the surface of a CsPbBr3 cluster as a function of particle distance,
the perovskite nanocrystals using synchrotron radiation. The electronic d, in the lattice. The red line is a fit with a Gaussian distribution function with
edge energies of Pb L, Cs K and Br K (shown as red squares) fall in the a fitting coefficient of η = 0.9752. The particle distance corresponding to the
X-ray energy range 10–38 keV. The line is a guide for the eye. d, Calculated maximum energy density is 10.32 Å. g, Schematic showing the basic design
electronic band structures of the CsPbBr3 nanocrystal. The inset shows the of a perovskite-nanocrystal-based photoconductor used for X-ray sensing. A
Brillouin zone of the cubic-phased crystal lattice (see Methods for details). 10-µm-thick layer of CsPbBr3 QDs is spin-coated onto the substrate for X-ray
e, Proposed mechanism of X-ray scintillation in a lead halide perovskite photon–carrier conversion. Gold (Au) electrodes are placed onto the QDs for
nanocrystal. Upon X-ray irradiation, a high-energy electron (red circles, hole–electron extraction. h, Current–voltage characteristics of the
e−) is ejected from a lattice atom through photoelectric ionization (ionizing as-fabricated photoconductor, recorded with and without X-ray illumination.

measured the radioluminescence of the perovskite nanocrystals in absorbed dose of ionizing radiation into visible light (Fig. 2d). In addi-
response to synchrotron radiation (Fig. 2c, Extended Data Fig. 6, tion, the orbital contour plots of the CsPbBr3 nanocrystal indicate that
Supplementary Video 2). We observed an abrupt enhancement in the the presence of hole-like surface-vacancy-induced Coulomb-trapping
scintillation intensity upon excitation at 14 keV, 16 keV and 36 keV, states near the Fermi level beyond the valence band maximum is
indicating an X-ray absorption resonance at the electronic edge of the responsible for the electronically energetic confinement of excitons in
Pb L, Cs K and Br K shells in the CsPbBr3 structure. Density functional the perovskite nanocrystal (Extended Data Fig. 8b).
theory calculations confirmed that the electronic band structure of Figure 2e presents a plausible mechanism for the high-intensity
these perovskite nanocrystal scintillators is tunable, which is associated radioluminescence from the perovskite nanocrystals. At the initial
with the tailorability of their valence band through control of the halide conversion stage, an incident X-ray photon with energy lower than
composition (Extended Data Fig. 7). The bandgap energy of the a few hundred kiloelectronvolts interacts with the lattice atoms of a
perovskite nanocrystal under study is located in the range 1.7–3 eV, perovskite nanocrystal, predominantly through the photoelectric effect.
suggesting the feasibility of using such a nanomaterial to convert an During this process a large number of high-energy electrons and holes

N A T U R E | www.nature.com/nature
© 2018 Springer Nature Limited. All rights reserved.
 RESEARCH LETTER

a b c
Radioluminescence intensity (a.u.)

y = 4.79x + 18.7 Detection limit: 137Cs O = 530 nm

source, 661 keV

Radioluminescence intensity
Radioluminescence intensity
6 Intensity
R2 = 0.996
13 nGy s–1

Day 1 Day 2 Day 3 Day 4

4 X-ray photons 7.4 h 9.1 h 9.5 h 9.3 h
0 40 80 120 160
Dose rate (nGy s–1)
Perovskite QDs W = 44.6 ns
120 on/off cycles
2 MS
PD
T
PM

0
0 20 40 60 0 50 100 150 0 2,000 4,000 6,000 8,000
Dose rate (μGy s–1) Time (ns) Irradiation time (s)

d e Bright field f X-ray imaging

Front
O = 530 nm

80 μm

Camera

Sample
X-ray source Perovskite QDs
y

Holder
z
Scintillation platform 9 mm 5 mm
x

Fig. 3 | Ultrasensitive X-ray sensing and radiography using CsPbBr3 of the CsPbBr3-based scintillator under excitation with a 137Cs source
nanocrystals. a, Radioluminescence measurements for a CsPbBr3- (photon energy, 661 keV). The scintillation decay time is τ = 44.6 ns.
based scintillator as a function of dose rate. The left inset shows c, Photostability of the CsPbBr3-based scintillator against continuous
radioluminescence profiles measured at low dose rates. The detection X-ray irradiation (wavelength λ = 530 nm, 50 kV; top) and repeated cycles
limit of 13 nGy s−1 is derived from the slope of the fitting line, with of X-ray excitation at 30 kV with a time interval of 30 s (λ = 530 nm;
a signal-to-noise ratio of 3. The right inset shows a schematic of the bottom). d, Schematic of the experimental setup used for real-time X-ray
X-ray photodetector, which consists of a CsPbBr3 nanocrystal thin film diagnostic imaging of biological samples. A beetle is placed between the
(about 120 µm thickness), a polydimethylsiloxane (PDMS) layer and a X-ray source and a scintillation platform covered with perovskite QDs.
photomultiplier tube (PMT). All measurements were performed three e, f, Bright-field (e) and the X-ray (f) images of the sample, recorded with a
times. Error bars are mean ± s.d. b, Measured radioluminescence decay digital camera. The X-ray images were recorded at a voltage of 50 kV.

can be created, and electronic transport occurs between the perovskite is critical to scintillation performance in medical radiography. The
nanocrystals (Fig. 2f). The hot electrons and holes are then quickly photostability of the perovskite nanocrystals was further examined
thermalized in the conduction and valence band edges. The X-ray- under continuous or repeated cycles of X-ray illumination, as shown
induced charge carriers in the perovskite nanocrystals were experimen- in Fig. 3c.
tally confirmed by measuring the current through a photoconductor To assess the suitability of the perovskite nanocrystals as scintilla-
upon X-ray illumination (Fig. 2g, h). The trapping and radiative recom- tors for X-ray phase-contrast imaging, we implanted a metallic needle
bination of electron–hole pairs can be controlled to produce a desired into a green scarab beetle and imaged the biological specimen with
luminescence colour by adjusting the bandgap energy. The mechanism X-rays against a background substrate comprising a thin film of
of intense X-ray scintillation could be attributed in part to the strong solution-processed CsPbBr3 nanocrystals (Fig. 3d). We note that the
X-ray stopping power and quantum confinement effects of perovskite CsPbBr3 nanocrystals were chosen for this demonstration because their
nanocrystals. Additionally, the scintillation process is dominated by green emission at 530 nm matches well with the maximum wavelength
the presence of highly emissive triplet excited states (Fig. 2e), large response of a complementary metal-oxide-semiconductor sensor. As
absorption cross-section within the bandgap (Extended Data Fig. 9a, b) shown in Fig. 3e, f, owing to the large difference between the X-ray
and fast emission output (Extended Data Fig. 9c–e), which are charac- stopping powers of the needle and the beetle, the needle inside the
teristics of perovskite nanocrystals27,28,32. beetle is clearly revealed by phase-contrast imaging recorded using a
The solution-processability of the perovskite nanocrystals makes it common digital camera. The concept of direct X-ray contrast imaging
possible to fabricate a thin-film scintillator device for ultrasensitive through the use of high-efficiency perovskite nanocrystals is readily
X-ray detection. In this device (Fig. 3a), spin-coated CsPbBr3 nano- applicable to high-throughput electronics inspection and tissue
crystals are used for X-ray sensing by converting high-energy X-ray imaging, where common digital cameras can be conveniently used
photons into visible emission, which is readily detectable by a photo- (Extended Data Fig. 11; Extended Data Table 2).
multiplier tube. A favourable characteristic of the prototype X-ray We took a step further and tested the compliance of the perovskite
detector is its linear response to the X-ray dose rate, covering a range nanocrystals to commercial flat-panel X-ray detectors equipped
as broad as four orders of magnitude (Extended Data Fig. 10). The with α-Si photodiode arrays (Fig. 4a, b). As shown in Fig. 4c, the
lowest detectable dose rate for X-ray detection is demonstrated to be perovskite-nanocrystal-based X-ray detector shows a modulation
13 nGy s−1. This value is about 420 times lower than the dose typically transfer function of 0.72 at a spatial resolution of 2.0 line pairs per
used for X-ray diagnostics (5.5 µGy s−1)14. This scintillation photo- millimetre, which is much higher than the spatial resolution of com-
detector also exhibits a very fast response (scintillation decay time, mercially used CsI:Tl-based flat-panel X-ray detectors (0.36 at 2.0 line
τ = 44.6 ns) upon excitation with pulsed photons (661 keV) from a pairs per millimetre). This high spatial resolution could be ascribed to
portable 137Cs source (Fig. 3b). The fast response to X-ray photons the lower degree of light scattering in the nanoparticle-based thin film

N A T U R E | www.nature.com/nature
© 2018 Springer Nature Limited. All rights reserved.
 LETTER RESEARCH

a X-ray photons b c
Protective Al foil
Perovskite Measured
1.2

Modulation transfer function

scintillator Fitted

0.8

0.4
Pixel α-silicon
photodiode array
0
TFT sensor panel 0 1 2 3 4 5
Spatial frequency (mm–1)
d e f
Perovskite + Only
α-Si panel α-Si panel

X-rays 10 mm X-rays 70 kV

Fig. 4 | Prototype perovskite-nanocrystal-based flat-panel detector function under 15 µGy of X-ray exposure. The blue circles and purple line
for digital radiography. a, Multilayered design of the flat-panel X-ray show measured values and the black line is a fit to the data. d, e, Digital
imaging system consisting of a thin-film-transistor (TFT) sensor panel, photograph of a network interface card (d) and corresponding X-ray
a pixelated α-silicon photodiode array, a CsPbBr3 perovskite nanocrystal image obtained using the flat-panel detector (70 kV and 2.5 mGy s−1
thin film (about 75 µm thick) and a protective Al foil cover (40 µm). exposure for 6 ms) (e). f, Comparison of X-ray images of an Apple iPhone
b, Photograph of the packaged flat-panel detector. c, Spatial resolution acquired with the perovskite scintillator deposited on an α-Si photodiode
of the X-ray imaging system, characterized by the modulation transfer panel (left) and only with an α-Si photodiode (right).

compared with that occurring in commercial bulk-scintillator-based Received: 24 October 2017; Accepted: 21 June 2018;
films made of thick polycrystalline ceramics or long micropillars. We Published online xx xx xxxx.
further used the prototype device to image the internal structures of
electronic circuits and an Apple iPhone with a low X-ray dose of 15 µGy 1. Röntgen, W. C. On a new kind of rays. Science 3, 227–231 (1896).
(Fig. 4d–f). Unlike CsI:Tl scintillators, which have the issue of after- 2. Moretti, F. et al. Radioluminescence sensitization in scintillators and phosphors:
trap engineering and modeling. J. Phys. Chem. C 118, 9670–9676 (2014).
glow luminescence (scintillation decay time of 1,000 ns), our perovskite 3. Büchele, P. et al. X-ray imaging with scintillator-sensitized hybrid organic
nanocrystals have a very fast response (44.6 ns) to X-rays, making them photodetectors. Nat. Photon. 9, 843–848 (2015).
ideal for dynamic real-time X-ray imaging. 4. Yaffe, M. J. & Rowlands, J. A. X-ray detectors for digital radiography. Phys. Med.
Biol. 42, 1–39 (1997).
In conclusion, we have demonstrated inorganic perovskite nano- 5. Durie, B. G. & Salmon, S. E. Scintillator distribution in high-speed
crystals as a new class of scintillators that are capable of converting autoradiography. Science 190, 1093–1095 (1975).
small doses of X-ray photons into multicolour visible light. When 6. Nikl, M. & Yoshikawa, A. Recent R&D trends in inorganic single-crystal
scintillator materials for radiation detection. Adv. Opt. Mater. 3, 463–481 (2015).
considering the material’s solution-processability and practical scal- 7. Weber, M. J. Inorganic scintillators: today and tomorrow. J. Lumin. 100, 35–45
ability, it is envisioned that these scintillators are suitable for the mass (2002).
production of ultrasensitive X-ray detectors and large-area, flexible 8. Rodnyi, P. A. Physical Processes in Inorganic Scintillators (CRC Press, Boca Raton,
X-ray imagers. Compared to conventional CsI:Tl scintillators—whose 1997).
9. Nagarkar, V. V. et al. Structured CsI(Tl) scintillators for X-ray imaging
use is constrained by the risk of thallium poisoning, the presence of applications. IEEE Trans. Nucl. Sci. 45, 492–496 (1998).
afterglow and high-temperature synthesis—perovskite nanocrystals 10. Baccaro, S. et al. Scintillation properties of YAP:Ce. Nucl. Instrum. Methods
offer several outstanding attributes, including relatively low toxicity, A 361, 209–215 (1995).
11. Rowlands, J. A. Material change for X-ray detectors. Nature 550, 47–48 (2017).
low-temperature solution synthesis, fast scintillation response and high 12. Kim, Y. C. et al. Printable organometallic perovskite enables large-area, low-dose
emission quantum yield. Although there is still much to be learned X-ray imaging. Nature 550, 87–91 (2017).
regarding the origin of nanocrystal scintillation, these perovskite nano- 13. Wei, W. et al. Monolithic integration of hybrid perovskite single crystals with
heterogenous substrate for highly sensitive X-ray imaging. Nat. Photon. 11,
crystals may hold substantial promise for advancing X-ray sensing and 315–321 (2017).
imaging industry. The thermal and environmental instability issues 14. Wei, H. et al. Sensitive X-ray detectors made of methylammonium lead
that are often associated with perovskite materials in photovoltaic and tribromide perovskite single crystals. Nat. Photon. 10, 333–339 (2016).
15. Yakunin, S. et al. Detection of X-ray photons by solution-processed organic-
light-emitting-diode applications could be largely avoided through the inorganic perovskites. Nat. Photon. 9, 444–449 (2015).
X-ray scintillation settings. 16. Shrestha, S. et al. High-performance direct conversion X-ray detectors based on
sintered hybrid lead triiodide perovskite wafers. Nat. Photon. 11, 436–440
(2017).
Online content 17. Wei, H. et al. Dopant compensation in alloyed CH3NH3PbBr3−xClx perovskite
Any methods, additional references, Nature Research reporting summaries, source single crystals for gamma-ray spectroscopy. Nat. Mater. 16, 826–833 (2017).
data, statements of data availability and associated accession codes are available at 18. Pan, W. et al. Cs2AgBiBr6 single-crystal X-ray detectors with a low detection
https://doi.org/10.1038/s41586-018-0451-1. limit. Nat. Photon. 11, 726–732 (2017).

N A T U R E | www.nature.com/nature
© 2018 Springer Nature Limited. All rights reserved.
 RESEARCH LETTER

19. Birowosuto, M. D. et al. X-ray scintillation in lead halide perovskite crystals. Sci. (Singapore); the National Research Foundation, Prime Minister’s Office,
Rep. 6, 37254 (2016). Singapore under its Competitive Research Program (CRP award number
20. Dong, Q. et al. Electron–hole diffusion lengths >175 µm in solution-grown NRF-CRP15-2015-03); the National Basic Research Program of China (973
CH3NH3PbI3 single crystals. Science 347, 967–970 (2015). Program, grant number 2015CB932200); the National Natural Science
21. Shi, D. et al. Low trap-state density and long carrier diffusion in organolead Foundation of China (21635002, 21471109, 21210001 and 21405143); and
trihalide perovskite single crystals. Science 347, 519–522 (2015). the Natural Science Foundation of Jiangsu Province (BE2015699). We thank
22. Tan, H. et al. Efficient and stable solution-processed planar perovskite solar cells H. Jiang, B. Deng, Z. Fang, Z. Zhou, Y. Zhang, X. Ling, M. Sun and A. Malko for
via contact passivation. Science 355, 722–726 (2017). technical assistance.
23. Li, X. et al. A vacuum flash-assisted solution process for high-efficiency
large-area perovskite solar cells. Science 353, 58–62 (2016). Reviewer information Nature thanks R. Comin, W. Heiss and the other
24. Im, J.-H., Jang, I.-H., Pellet, N., Grätzel, M. & Park, N.-G. Growth of CH3NH3PbI3 anonymous reviewer(s) for their contribution to the peer review of this work.
cuboids with controlled size for high-efficiency perovskite solar cells. Nat.
Nanotechnol. 9, 927–932 (2014). Author contributions Q.C. and X.L. conceived and initiated the project. X.L.,
25. Son, D.-Y. et al. Self-formed grain boundary healing layer for highly efficient H.Y. and W.H. supervised the project and led the collaboration efforts. Q.C., X.L.,
CH3NH3PbI3 perovskite solar cells. Nat. Energy 1, 16081 (2016). H.Y. and W.H. designed the experiments. Q.C., J.W., L.L. and S.H. performed the
26. Protesescu, L. et al. Nanocrystals of cesium lead halide perovskites (CsPbX3, nanocrystal synthesis. Q.C., X.O. and J.L. carried out the spectral measurements.
X=Cl, Br, and I): novel optoelectronic materials showing bright emission with Q.C., X.O., Y.W., Y.L., D.F., Z.Y., D.B.L.T. and A.H.L. contributed to the design
wide color gamut. Nano Lett. 15, 3692–3696 (2015). and implementation of the X-ray sensing experiments. B.H., M.B. and O.F.M.
27. Becker, M. A. et al. Bright triplet excitons in caesium lead halide perovskites. carried out the theoretical calculations. J.A., A.A.Z. and O.M.B. prepared the
Nature 553, 189–193 (2018). perovskite single crystals. X.G. and T.W. fabricated the photoconductor devices
28. Hu, F. et al. Superior optical properties of perovskite nanocrystals as single and performed the photocurrent measurements. X.X. fabricated the PDMS
photon emitters. ACS Nano 9, 12410–12416 (2015). moulds and measured the low-temperature scintillation spectra. Q.C. and X.L.
29. Swarnkar, A. et al. Quantum dot-induced phase stabilization of α-CsPbI3 wrote the manuscript. All authors discussed the results and commented on the
perovskite for high-efficiency photovoltaics. Science 354, 92–95 (2016). manuscript.
30. Hossu, M., Liu, Z., Yao, M., Ma, L. & Chen, W. X-ray luminescence of CdTe quantum
dots in LaF3:Ce/CdTe nanocomposites. Appl. Phys. Lett. 100, 013109 (2012). Competing interests The authors declare no competing interests.
31. Saidaminov, M. I. et al. Pure Cs4PbBr6: highly luminescent zero-dimensional
perovskite solids. ACS Energy Lett. 1, 840–845 (2016). Additional information
32. Kovalenko, M. V., Protesescu, L. & Bodnarchuk, M. I. Properties and potential Extended data is available for this paper at https://doi.org/10.1038/s41586-
optoelectronic applications of lead halide perovskite nanocrystals. Science 358, 018-0451-1.
745–750 (2017). Supplementary information is available for this paper at https://doi.org/
33. Berger, M. J. et al. XCOM: Photon Cross Sections Database; https://www.nist.gov/ 10.1038/s41586-018-0451-1.
pml/xcom-photon-cross-sections-database (2013). Reprints and permissions information is available at http://www.nature.com/
reprints.
Acknowledgements This work is supported by the King Abdullah University Correspondence and requests for materials should be addressed to H.Y. or
of Science and Technology; the Singapore Ministry of Education (grants W.H. or X.L.
R143000627112 and R143000642112); the Agency for Science, Technology Publisher’s note: Springer Nature remains neutral with regard to jurisdictional
and Research (A*STAR) under contracts 122-PSE-0014 and 1231AFG028 claims in published maps and institutional affiliations.

N A T U R E | www.nature.com/nature
© 2018 Springer Nature Limited. All rights reserved.
 LETTER RESEARCH

METHODS 250Xi instrument equipped with Al Kα monochromatized X-rays at 1,486.6 eV.

Chemicals. Caesium carbonate (Cs2CO3, 99.9%), lead(ii) chloride (PbCl2, 99.99%), Absorption spectra were measured by an ultraviolet–visible spectrophotometer
lead(ii) bromide (PbBr2, 99.99%), lead(ii) iodide (PbI2, 99.99%), oleylamine (UV-2450, Shimadzu, Japan). Photoluminescence and radioluminescence spectra
(technical grade 70%), oleic acid (technical grade 90%), 1-octadecene (technical were obtained by an Edinburgh FS5 fluorescence spectrophotometer (Edinburgh
grade 90%) and cyclohexane (chromatography grade 99.9%) were purchased from Instruments Ltd, UK) equipped with a miniature X-ray source (AMPEK, Inc.).
Sigma-Aldrich. Silicon wafers were obtained from Xilika Crystal Polishing Material Photographs of the X-ray-induced luminescence were acquired with a digital
Co., Ltd (Tianjin, China). SU-8 photoresist (2050) and developer solution were camera (Nikon D7100 with AF Micro-Nikkor 60mm f/2.8D). For the time-
purchased from Microchem Corp. (Newton, MA). A Sylgard 184 silicone elasto- resolved photoluminescence measurements, a pulsed excitation source was used.
mer kit was purchased from Dow Corning for the preparation of polydimethyl- The scintillation decay measurement was carried out at the Institute of High Energy
siloxane (PDMS) substrates. Crystals of CsI:Tl, Bi4Ge3O12, YAlO3:Ce and PbWO4 Physics of the Chinese Academy of Sciences with a 137Cs source used for the pulsed
scintillators were purchased from Zhonghelixin Co., Ltd (Chengdu, China). CdTe excitation. The effective scintillation decay time (τeff ) can be calculated using the
QDs were obtained from Xingzi New Material Technology Development Co., Ltd following formula:
(Shanghai, China). Unless otherwise noted, all the reagents were used without
1 ∞
additional treatment.
Synthesis of Cs-oleate as a caesium precursor. In a typical synthesis procedure,
τeff =
I0 ∫0 I(t)dt

Cs2CO3 (0.4 g, 1.23 mmol), oleic acid (1.25 ml) and octadecene (15 ml) were added to
where I(t) and I0 denote the radioluminescence (or photoluminescence) intensity
a two-neck round-bottom flask (50 ml). The resulting mixture was heated to 100 °C
as a function of time, t, and the maximum intensity, respectively.
under vigorous stirring and vacuum conditions for 0.5 h. After that, a nitrogen
Measurement of photoluminescence quantum yield. The quantum yield was
purge and vacuum were alternately applied to the flask three times to remove
determined with an optical spectrometer equipped with an integrating sphere.
moisture and O2. Subsequently, the mixture was heated to 150 °C and the solution
Perovskite QDs were dispersed in cyclohexane. The excitation and luminescence
became clear, indicating the completion of the reaction between Cs2CO3 and oleic
emission were detected by a photomultiplier tube (PMT) through total internal
acid. The Cs-precursor solution was kept at 150 °C in a nitrogen atmosphere before
reflection in the integration sphere. The photoluminescence quantum yield (PLQY)
the synthesis of perovskite nanocrystal.
was calculated according to PLQY = Psample/(Sref − Ssample), where Sref and Ssample
Synthesis of CsPbX3 (X = Cl, Br or I) nanocrystals. The CsPbX3 perovskite QDs
are the excitation light intensities not absorbed by the solvent and the sample,
were synthesized by a modified hot-injection procedure29. In a typical experi-
respectively, and Psample is the integrated emission intensity of the sample (Extended
ment, PbX2 (0.36 mmol each for X = Cl, Br or I), oleic acid (1.0 ml), oleylamine
Data Fig. 5e).
(1.0 ml) and octadecene (10 ml) were added to a two-neck round-bottom flask
Measurement of exciton binding energy in perovskite nanocrystal scintillator.
(50 ml). The resulting mixture was heated to 100 °C under vigorous stirring and
The exciton binding energy (Ea) was estimated by measuring the temperature-
vacuum conditions for 0.5 h, at which time the moisture residue was removed
dependent radioluminescence intensity. By fitting data derived from the integrated
by purging with nitrogen and vacuum suction. Then the mixture was heated to
luminescence intensity of the CsPbBr3 QD scintillator with the Arrhenius formula:
160 °C until the PbX2 precursors dissolved completely. A hot Cs-oleate precursor
solution (1 ml) was injected quickly into the above reaction mixture. After 5 s of I(T0)
reaction, the flask was transferred into an ice bath. The CsPbX3 QDs were obtained I (T ) =
1 + CT exp[ − Ea/(k BT )]
by centrifugation at 13,000 r.p.m. for 10 min and stored in 4 ml of cyclohexane
before further use. Mixed-halide perovskite QDs were synthesized to tune the where I(T0) is the radioluminescence intensity at the low-temperature (T0) limit,
luminescence colour. Samples 1–12 correspond to the as-synthesized CsPbCl3 (1), kB is the Boltzmann constant, C is a constant and T is the temperature, we obtain
CsPbCl2Br (2), CsPbCl1.5Br1.5 (3), CsPbClBr2 (4), CsPbCl2.5Br0.5 (5), CsPbBr3 (6), an exciton binding energy of 49 meV.
CsPbBr2I (7), CsPbBr1.8I1.2 (8), CsPbBr1.5I1.5 (9), CsPbBr1.2I1.8 (10), CsPbBrI2 (11) Fabrication of perovskite nanocrystal scintillator films on PDMS substrates.
and CsPbI3 (12) QDs. The PDMS substrates were fabricated by a standard soft lithography microfabri-
Growth of lead halide perovskite single crystals. The growth of CsPbBr3 single cation technique. Briefly, a photomask was first designed using Adobe Illustrator
crystals was carried out according to a method described in the literature34. In a CS6. A 60-µm-thick layer of negative photoresists (SU-8 2015; 2,500 r.p.m., 60 s)
typical procedure, CsBr (0.64 g, 3 mmol) and PbBr2 (2.2 g, 6 mmol) were dissolved was spin-coated onto a silicon wafer (3 inch; 1 inch = 1.54 cm). The wafer was
in 3 ml of dimethyl sulfoxide and stirred for 1 h. Subsequently, 1.5 ml of the mix- prebaked at 60 °C for 10 min and then at 85 °C for 5 min. The resulting photoresist
ture was transferred into a vial heated at 60 °C, and the temperature was raised to on the wafer was irradiated by an ultraviolet lamp for 20 s, followed by a post-
100 °C with a heating rate of 10 °C h−1. At 100 °C, the solution was filtered and baking treatment in an oven at 75 °C for 5 min. Next, the desired microstructure on
then gradually heated to 120 °C. We observed the growth of small-sized crystals the silicon wafer was produced using a developer solution. The PDMS substrates
with increasing temperature. The resulting crystals were washed with hot dimethyl were fabricated with a premixed PDMS prepolymer and curing agent (10:1 by
sulfoxide and dried in vacuo at 100 °C for 1 h. For the synthesis of CH3NH3PbBr3 mass) under vacuum conditions, followed by heat treatment at 80 °C for 2 h. The
(MAPbBr3) single crystals, a mixture of PbBr2 and CH3NH3Br (1.5 mol each) was PDMS replicas were carefully peeled off from the master. Finally, perovskite QDs
dissolved in a solution of N,N,-dimethyl formamide (1 ml) at room temperature. dispersed in cyclohexane were coated onto the PDMS substrate.
The solution was purified by passing through a polytetrafluoroethylene filter with Radioluminescence measurement for perovskite nanocrystal scintillators. The
a pore size of 0.22 µm. The growth of the MAPbBr3 single crystals was carried out measurement of X-ray-induced luminescence was performed using a solid film
in an oil bath heated at 60 °C and under ambient pressure. comprising perovskite QDs. We note that perovskite QDs dispersed in solution
Synthesis of fluorescent carbon dots. Fluorescent carbon dots were synthesized are not suitable for scintillation characterization under X-ray excitation, because a
by a hydrothermal method. In a typical experiment, ammonium citrate (0.972 g, low population of QDs in solution is inefficient for X-ray absorption. Unlike under
4.0 mmol) was first dissolved in a 20-ml water solution. The solution was then visible-light excitation, a quartz cuvette is not used for measuring scintillation
transferred into a 30-ml Teflon-lined vessel at room temperature while stirring. luminescence under X-ray excitation, because the excitation can be strongly
Subsequently, the solution was heated to 190 °C and kept at that temperature for absorbed by the cuvette. The scintillation decay times36 of CsI:Tl, Bi4Ge3O12,
10 h. After cooling to room temperature, the product was purified using dialysis, YAlO3:Ce and PbWO4 crystal scintillators are listed in Extended Data Table 1.
and the cut-off molecular weight of the dialysed membrane was equivalent to X-ray photoconductor devices. To fabricate the X-ray photoconducting device,
about 2,000. The carbon dot solution was concentrated using a rotary evaporator. silica wafers with a 300-nm-thick SiO2 layer were first cleaned by sonication in
Preparation of silica-coated perovskite nanocrystals. The stability of perovskite acetone, ethanol and deionized water separately. After drying with flowing nitro-
nanocrystals was improved by coating with a silicon dioxide (silica) layer according gen, the substrates were treated with oxygen plasma for 6 min. The solution of
to the literature35. In a typical procedure, the CsPbBr3 QDs, dispersed in cyclohexane CsPbBr3 QDs was spin-coated onto the Si/SiO2 substrates at 500 r.p.m. for 30 s and
(2 ml), were introduced into a 50-ml flask containing 10 ml of toluene solution subsequently annealed at 100 °C for 5 min. This procedure was repeated three times
(≥99.5%; AnalaR NORMAPUR). 100 µl of tetramethoxysilane was injected quickly to produce a film with a thickness of about 10 µm. After that, 100-nm-thick gold
into the mixture at room temperature. After stirring for 2 h, the products were electrodes were deposited onto the CsPbBr3 QD film by thermal evaporation, using
isolated through centrifugation at 13,000 r.p.m. for 8 min. The silica-coated CsPbX3 a shadow mask to control the size of the deposition. For the X-ray photon-to-cur-
(X = Br or I) QDs were either dispersed in cyclohexane or dried in air. rent measurement, we used a commercially available, miniaturized X-ray tube
Physical characterization. Powder X-ray diffraction characterization was carried (Amptek). The target in the X-ray tube was made of gold and the maximum output
out by an ADDS wide-angle X-ray powder diffractometer with Cu Kα radiation was 4 W. In our measurement, the X-ray tube voltage was kept at 50 kV while the
(wavelength, λ = 1.54184 Å). TEM imaging was performed using a FEI Tecnai G20 peak X-ray energy was set at 10 keV with an Al/W filter and a 2-mm-diameter brass
transmission electron microscope with an accelerating voltage of 200 kV. X-ray collimator. The distance between the X-ray source and the X-ray photoconducting
photoelectron spectroscopy analysis was carried out using a Thermo escalab device was about 30 cm. The current–voltage measurement of the devices was

© 2018 Springer Nature Limited. All rights reserved.

 RESEARCH LETTER

performed using a Signotone Micromanipulator S-1160 probe station equipped reciprocal Brillouin zone, which leads to a difference between the electron effective
with a Keithley 4200 Semiconductor Parametric Analyzer. All the experiments mass along different paths and the excitonic binding energy. This difference illus-
were carried out at ambient conditions. trates the probability that electronic transport within reciprocal band structures is
X-ray scintillation detector and imaging. The X-ray scintillation detector was directionally selected for luminescence. Another plausible route is the annihilation
constructed by coating of a PDMS substrate with perovskite QDs (layer thickness of shallow acceptor levels (Pb vacancies), which induces an absence of recombina-
of 120 µm), followed by the attachment of a PMT. In a typical procedure, a solution tion centres. Such intrinsic lattice defects usually produce low-excitation energy
of CsPbBr3 QDs was spin-coated onto the PDMS substrate. The PDMS substrate levels compared with the ideal lattice and consequently hinder energy transfer
was then coupled to the PMT for maximized collection of visible photons. For during the light-absorption process.
X-ray detection, a range of X-ray dose rates (0.013–278 µGy s−1) was applied by Anisotropic transport-induced luminescence contrast. Using DFT calculations, the
adjusting the current and voltage of the X-ray source. For X-ray imaging, a plastic bandgap of a bulk CsPbBr3 crystal was calculated to be about 2.02 eV, whereas in
disk coated with CsPbBr3 nanocrystals was used. A green scarab beetle implanted the CsPbBr3 QDs the bandgap increases slightly to 2.22 eV (Fig. 2d). This is because
with a metallic needle was employed as a specimen for X-ray imaging. the large surface-to-volume ratio in the CsPbBr3 nanocrystal induces an evident
In vivo multicolour optical bioimaging. All the animal experiments were per- quantum confinement, thus leading to an enlarged vacuum Coulomb barrier for
formed in compliance with institutional guidelines. Silica-coated perovskite QDs electronic transitions.
(CsPbBr3, CsPbBr1.5I1.5, CsPbBr1.2I1.8; 100 µg, 50 µl) dispersed in a phosphate- We chose reciprocal, highly symmetrical points and lined up two different paths
buffered saline buffer solution were subcutaneously injected into the Balb/c (X→R→M→Γ→R and Γ→R→M) within the Brillouin zone. As shown in the
nude mice (age, 4–6 weeks; weight, 18 g). An animal imaging system (Advanced electronic band structure plot in Fig. 2d, the valence band edge and the conduc-
Molecular Imager, Cold Spring Biotech Corp., Shanghai) equipped with an X-ray tion band edge are located at the same point R(1/2, 1/2, 1/2). Using effective mass
source was used to carry out in vivo radioluminescence imaging of the mice. The theory, the effective mass for electrons and holes was found to be anisotropic along
exposure time for in vivo imaging was set at 1 s. For in vivo multicolour optical these two directions. In the directions X→R→M→Γ→R and Γ→R→M, the effec-
imaging, optical filters (530 nm, 630 nm and 670 nm) were used to selectively tive masses for electrons were calculated to be 0.03m0 and 0.11m0 and the effective
record the X-ray-induced luminescence at different emission wavelengths masses for holes were calculated to be 0.12m0 and 0.24m0, respectively, where m0
(Extended Data Fig. 11). is the rest mass of the electron. Thus, the Wannier–Mott exciton binding energy
Construction of perovskite-based flat-panel X-ray imaging system. The α-Si and radius are different in these two directions.
photodiode assay backplane was customized for commercial α-Si/CsI:Tl detectors By converting the reciprocal Brillouin zone area into a real-space diagram,
supplied by iRAY Technology Shanghai, Inc. The active area of a photodiode array we found that the Cs site at Γ(0, 0, 0) in the body-centred area is different in
is 43.0 cm × 43.0 cm, consisting of 3,072 × 3,072 square pixels with a pixel pitch of these two paths. Point R(1/2, 1/2, 1/2) denotes the position of the Pb site, whereas
139 µm. CsPbBr3 nanocrystals were first dispersed in cyclohexane. We coated the M(1/2, 1/2, 0) represents the location of the Br site. Owing to the different effective
photodiode arrays (8.0 cm × 8.0 cm) with a thin film (thickness, 75 µm) of nano- electron masses of the two paths, the path Γ→R→M is energetically favourable to
crystals using a solution-processing method. After evaporation of cyclohexane, the transport of electrons and holes. By contrast, the X→R→M direction is ruled
an aluminium film (40 µm thick) was added under vacuum, in a packaging process out because the binding energies are too large to release electrons and holes for
similar to that used in commercial CsI:Tl-based X-ray imaging systems. The alu- recombination. This implies a charge transfer process from the Cs site to the Pb site
minium film was used to protect the scintillators from moisture and light soaking. at the cubic apex point, finally reaching the Br site at the middle point of the cubic
We note that a reflecting layer was coated on the surface of the aluminium film to edge, namely, through the path Γ→R→M (Extended Data Fig. 7a).
enhance the light collection into the photodiode elements. The power consump- Furthermore, we used an orbital calculation to retrieve the electronic and hole
tion was 25 W for full-image acquisition and the X-ray source was operated at a orbitals from the electronic band structure. Our results show that bound electrons
voltage of 70 kV. X-ray imaging of electronic circuit boards was acquired with an stay at the Br sites at a non-bonding state in the p–π orbital level (Extended Data
X-ray exposure of 2.5 mGy s−1 for 6 ms, resulting in a dose of 15 µGy. The spatial Fig. 7b). Meanwhile, bound holes were found to stay at the Pb site with an s-orbital
resolution was determined by measurement of the modulation transfer function. spherical distribution (Extended Data Fig. 7c). The orbital contour plots reveal the
Radioluminescence analysis using synchrotron radiation. The characterization localization of electrons and holes at perfect lattices. The stabilized charge state of
of the yield of X-ray-induced luminescence near electronic-shell edges was con- the body-centred Cs site is Cs+ because electrons are transferred from the Pb site
ducted using the synchrotron beamline in the Shanghai Synchrotron Radiation to the Br site through the ionization of one s-state electron.
Facility. A thin film of CsPbBr3 nanocrystals was cast onto a PDMS substrate. Intrinsic lattice defects in perovskite nanocrystal scintillators. Intrinsic lattice defects
The X-ray excitation energies were 10–38 keV, and a portable spectrophotometer in perovskite nanocrystal scintillators are responsible for both luminescence and
(Ocean Optics) was used to measure the radioluminescence. the quenching effect. Here we consider the low-energy native defects of a Br
Density functional theory calculation. For the calculation of the projected partial vacancy (VBr) and a Pb vacancy (VPb). For VBr, the absence of one Br atom leaves
density of states (PDOS), density functional theory (DFT) calculations were carried one electron occupying the empty p orbitals of the nearest neighbouring Pb site.
out. We used the Cambridge Serial Total Energy Package (CASTEP) source code Accordingly, localized electronic orbitals were modelled for VBr in the neutral (V 0Br)
to perform the calculations with the rotation-invariant DFT+U method. In a typ- and singly positive (V + Br ) states (Extended Data Fig. 7d, e). Because the charge
ical procedure, a simple cubic phase with Pm3m symmetrical lattice arrangement bound at the nearby Pb site is positively ionized, the p electronic orbitals of the V + Br
was modelled for bulk-phase CsPbBr3. Norm-conserving pseudopotentials of the site show a transition from the correlated state to a repulsive behaviour between
Cs, Pb and Br atoms were generated by the OPIUM code in the Kleinman– the two neighbouring Pb sites. The PDOS analysis also shows that the electronic
Bylander projector form. A nonlinear partial core correction and a scalar relativ- level of VBr, which is localized at the bottom of the conduction band edge, serves
istic averaging scheme were used to treat the spin–orbit coupling effect. In as a shallow donor. The VPb lattice defects produce an acceptor trap centre and a
particular, we treated the 4s, 4p and 4d states of the Br atoms as valence states, the spin-polarized state (Extended Data Fig. 7f, g). The singly negative state of a Pb
5s, 5p and 5d states for Cs atoms, and the 5d, 6s and 6p states for Pb atoms. The site (V−Pb) with one electron already captured could partially passivate the acceptor
Rappe–Rabe–Kaxiras–Joannopoulos method was chosen to optimize the pseudo- trap site with weakened charge localization. The process of local geometrical relax-
potentials during electronic minimization, particularly using a blocked-Davidson- ation on different charge states of VPb indicates that the Cs+ sites centrosymmet-
scheme matrix diagonalization. rically move towards the VPb centre (Extended Data Fig. 7h). Upon the occurrence
For the calculations of the electronic states in the CsPbBr3 material, we used of a strong clustering effect to form V2Pb− near the VPb site, the electronically active
self-consistent determination for on-site U correction on the localized p orbitals acceptor trap centre can be completely terminated. In this case, formation of local
of Br sites to correct the on-site Coulomb energy of the electron spurious self- Cs–Br motifs is possible. We also considered VCs and V− Cs sites in the lattice and
energy. The on-site electronic self-energy and related wavefunction relaxation found no effects on the electronic properties of the host lattice. Accordingly, VBr
in the semicore p, d or f orbitals in mixed-valence elements were used to obtain and VPb produce the electronic and hole levels for luminescence recombination in
accurate orbital eigenvalues for the electronic structures and transition levels. An the form of photon emission (Extended Data Fig. 7i).
ab initio two-way crossover searching calculation was performed by two func- Further, we calculated the excited energy and thermodynamic transition levels
tionally compiled CASTEP-17 source codes. Using the self-consistent determina- in the CsPbBr3 QDs. From the internal bulk lattice to the surface region, the dimen-
tion, on-site Hubbard U parameters for different orbitals of Br and Pb sites were sions of the material decrease but its surface-to-volume radio is increased. The
obtained. Further, a time-dependent DFT calculation was performed with a two- electronic donor trap levels experience a transition from a localized state below
electron-based Tamm–Dancoff approximation imported from the self-consistently the conduction band edge to a delocalized state in the conduction band (Extended
corrected ground-state wavefunctions. Data Fig. 7j). During the process of radiation ionization and release, the number of
Luminescence mechanisms in perovskite nanocrystal scintillator. Two energy- bound electrons is increased accordingly. By contrast, with a decreased dimension
transfer mechanisms for recombination luminescence exist in the perovskite of the host lattice, the trapping ability of the acceptor is decreased where the hole
nanocrystal scintillator. One is anisotropic electron and hole transport within the level shifts from a delocalized state in the valence band to a localized state above

© 2018 Springer Nature Limited. All rights reserved.

 LETTER RESEARCH

the valence band. Therefore, the quenching effect in the bulk CsPbBr3 materials on the surface of the QDs as a function of particle distance (Extended Data Fig. 8d).
for luminescence recombination is caused by annihilation of a hole level that is To investigate surface confinement, we further calculated the relative energy level
deeply buried in the valence band. The structural transition from the QD to the of the SVIC state as a function of particle distance (Extended Data Fig. 8d). It is
bulk form occurs from the surface to the bulk, thus the hole level is annihilated. obvious that at a distance of 12.06 Å from the particle surface, the SVIC energy
Intrinsic quantum confinement in CsPbBr3 nanocrystals. The intrinsic effect of implies a strong hole-like confinement, merely 0.056 eV above the valence band
quantum confinement in CsPbBr3 nanocrystals was examined by additional the- maximum. Our results suggest that the mean confinement path of electronic trans-
oretical study of their surface electronic properties. In a typical procedure, we port within the lattice is approximately 10.32 Å (Fig. 2f). Indeed, the intrinsic
first built a simplified model of the CsPbBr3 structure composed of 293 atoms energetics on the surface of the QDs is reasonable for energy confinement of the
with a particle size of 12.06 Å, namely, a lattice group (6 × 6 × 6) truncated from thermalized low-energy excitons inside the nanocrystal, resulting in a high yield
bulk CsPbBr3 crystal, using the radial coordinated structural formation program of X-ray scintillation light.
(RCSFP) (Extended Data Fig. 8a). DFT calculations yield that the orbital contour Data availability. The data that support the findings of this study are available
plots of the CsPbBr3 QD show the highest occupied molecular orbital (HOMO), from the corresponding authors upon reasonable request.
the lowest unoccupied molecular orbital (LUMO) and surface-vacancy-induced
Coulomb trapping (SVIC) states (Extended Data Fig. 8b). The electronic struc- 34. Saidaminov, M. I. et al. Inorganic lead halide perovskite single crystals:
tures show that the SVIC state is formed owing to unsaturated p orbitals of sur- phase-selective low-temperature growth, carrier transport properties, and
face Br sites and is electronically localized at the apex corner regions of the QD. self-powered photodetection. Adv. Opt. Mater. 5, 1600704 (2017).
Additionally, the PDOS of the CsPbBr3 QD indicates that such SVIC sites are 35. Huang, S. et al. Enhancing the stability of CH3NH3PbBr3 quantum dots by
mainly distributed near the Fermi level, beyond the valence band maximum, thus embedding in silica spheres derived from tetramethyl orthosilicate in
“waterless” toluene. J. Am. Chem. Soc. 138, 5749–5752 (2016).
exhibiting a hole-like feature and being strongly confined by the LUMO orbitals
36. Nikl, M. Scintillation detectors for X-rays. Meas. Sci. Technol. 17, 37–54 (2006).
(Extended Data Fig. 8c). This leads to the suppression of the long-distance 37. Miyata, A. et al. Direct measurement of the exciton binding energy and effective
transport of electron–hole pairs across the particle surface or between particles. masses for charge carriers in organic–inorganic tri-halide perovskites. Nat.
Furthermore, a model was built to perform a simulation of the energetic evolution Photon. 11, 582–587 (2015).

© 2018 Springer Nature Limited. All rights reserved.

 RESEARCH LETTER

Extended Data Fig. 1 | Schematic representation of scintillation frequency during excitonic luminescence; νA, photon frequency during
mechanism and X-ray-induced luminescence in bulk inorganic activator (A) luminescence. b, Scintillation properties of CsPbBr3 QDs
scintillators. a, In bulk inorganic materials, X-ray photons absorbed by and commercial bulk inorganic materials under X-ray excitation. The
the lattice atoms can generate hot charge carriers, followed by exciton full-width at half-maximum (FWHM) represents the spectral width of
thermalization and subsequent transport to defect sites or activator the scintillation spectra of the crystal. The insets show photographs of the
centres (‘traps’), where radiative emission occurs. VB, valence band; corresponding samples acquired under X-ray excitation. BGO, Bi4Ge3O12.
CB, conduction band; Vk, self-trapped hole; F, Farbe centre; νexc, photon

© 2018 Springer Nature Limited. All rights reserved.

 LETTER RESEARCH

Extended Data Fig. 2 | Physical characterization of as-synthesized (PDF) number 54-0752). c, d, Dark-field scanning transmission
perovskite QDs. a, TEM images of the as-prepared cubic-phased electron micrographs (STEM; JEM-F200HR) of CsPbBr3 nanocrystals.
nanocrystals (left) and the corresponding size distribution of the e, Elemental mapping of Br, Pb and Cs for the nanocrystals, obtained
nanocrystals (right). The samples are CsPbCl3, CsPb(Cl/Br)3, CsPbBr3, from the area marked by the rectangular box in d using energy-dispersive
CsPb(Br/I)3 and CsPbI3 nanocrystals (from top to bottom). Insets are X-ray spectroscopy. f, Atomically resolved dark-field STEM image
images of perovskite nanocrystals dispersed in cyclohexane, recorded of a single CsPbBr3 nanocrystal, showing Cs and Pb lattice atoms.
under 365-nm ultraviolet light excitation. b, Powder X-ray diffraction g, Energy-dispersive X-ray spectrum of the as-prepared CsPbBr3
patterns for typical ternary and mixed-halide CsPbX3 (X = Cl, Br or I) perovskite nanocrystals, confirming the stoichiometric composition of the
nanocrystals. All peaks are indexed in accordance with the cubic-phased CsPbBr3 nanocrystals. We note that strong Cu signals come from the TEM
CsPbBr3 structure (Joint Committee on Powder Diffraction Standards file copper grid.

© 2018 Springer Nature Limited. All rights reserved.

 RESEARCH LETTER

Extended Data Fig. 3 | Multicolour-emitting perovskite QD scintillators 278 µGy s−1. c, Typical photographs of radioluminescence from
upon X-ray irradiation. a, Demonstration of X-ray-induced luminescence CsPb(Cl/Br)3, CsPbBr3 and CsPb(Br/I)3 QDs under X-ray excitation.
modulation using CsPbX3 QDs of different compositions (X = Cl, Br or I). d, e, Multicolour visualization of as-developed perovskite QD scintillators
b, Multicolour X-ray scintillation from CsPb(Cl/Br)3, CsPbBr3 and using X-rays (d) and the corresponding bright-field image (e).
CsPb(Br/I)3 nanocrystals cast on a PDMS substrate. The X-ray dose rate is

© 2018 Springer Nature Limited. All rights reserved.

 LETTER RESEARCH

Extended Data Fig. 4 | Comparison of X-ray-induced luminescence intensity for the as-prepared carbon dots, CdTe QDs and CsPbBr3
intensity for QDs. a–c, Radioluminescence spectra of carbon dots (a), nanocrystals under 278 µGy s−1 X-ray excitation. e, Scintillation
CdTe QDs (b) and CsPbBr3 nanocrystals (NCs; c) under X-ray excitation performance and maximal K-edge energy of lead halide perovskite
at 5.0 µGy s−1 and 278 µGy s−1. d, Comparison of radioluminescence nanocrystals, carbon dots and CdTe QDs.

© 2018 Springer Nature Limited. All rights reserved.

 RESEARCH LETTER

Extended Data Fig. 5 | Comparison of X-ray-induced luminescence 278 µGy s−1. d, Comparison of radioluminescence intensity for three types
for lead halide perovskite materials. a–c, Radioluminescence spectra of of perovskite material under 278 µGy s−1 X-ray excitation. e, PLQY and
bulk single-crystal CH3NH3PbBr3 (a) and CsPbBr3 (b) and of CsPbBr3 exciton binding energy of perovskite materials at 300 K31,37. We note that
nanocrystals (c) under X-ray excitation at 5.0 µGy s−1 and 278 µGy s−1. the thermal energy at 300 K is kBT ≈ 25 meV. Nanocrystalline perovskites
The insets are photographs of CsPbBr3 bulk single crystal (b) and are highly luminescent materials, whereas bulk perovskite crystals are
nanocrystal powders (c) taken under ambient light (top) and X-ray most suitable for the generation of free charge carriers owing to their low
illumination (bottom). The X-ray dose used for the experiments was exciton binding energy.

© 2018 Springer Nature Limited. All rights reserved.

 LETTER RESEARCH

Extended Data Fig. 6 | Measurement of exciton binding energy and the CsPbBr3 nanocrystals at 532 nm. c, Experimental setup for the
synchrotron-radiation-induced radioluminescence of CsPbBr3 synchrotron-radiation-induced radioluminescence measurements at the
nanocrystals. a, Temperature-dependent scintillation spectra of the X-ray beamline of the Shanghai Synchrotron Radiation Facility (SSRF).
CsPbBr3 nanocrystals at 77–300 K under X-ray illumination at 50 µGy s−1. The X-ray energy is 10–38 keV. d, The electronic edge energies for Pb L, Cs
b, Arrhenius plot of the X-ray-induced luminescence intensities of K and Br K.

© 2018 Springer Nature Limited. All rights reserved.

 RESEARCH LETTER

Extended Data Fig. 7 | Electronic structure and scintillation mechanism nanocrystal through the influence of the valence orbitals. Cs, purple; Pb,
of the CsPbBr3 nanocrystal along selected reciprocal high-symmetry grey; Br, orange; electron orbital, blue; hole orbital, green. d–g, Localized
points within the Brillouin zone. a, The Brillouin zone of the cubic- electronic and hole levels for VBr and VPb at different charge states.
phased crystal lattice in CsPbBr3, calculated using relativistic corrections. h, i, Schematic diagram of energy transfer for radioluminescence induced
Γ, M, R and X denote high-symmetry points within the reciprocal space by intrinsic lattice defects and the quenching effect caused by evident ion
(blue). b, c, Calculated electron density associated with the valence band movement (Cs+). V δ− Pb denotes a Pb vacancy with small (|δ| < 1) charge
(b) and the conduction band (c) of the cubic CsPbBr3. We note that the transfer. j, Thermodynamic transition levels of the perovskite nanocrystal.
halide ions contribute to the changes of the bandgap in the perovskite Ec, maximum valence band energy; Ev, minimum conduction band energy.

© 2018 Springer Nature Limited. All rights reserved.

 LETTER RESEARCH

Extended Data Fig. 8 | Theoretical studies of the surface electronic level (EF). d, The relative energy of the SVIC state to the valence band
properties of CsPbBr3 nanocrystals. a, Simplified model of CsPbBr3 (VB) maximum as a function of inter-particle distance, d. The red line is
structure composed of 293 atoms with a particle size of 12.06 Å. calculated by fitting with the Gaussian distribution function. The top inset
b, Calculated orbital contour plots of a CsPbBr3 nanocrystal, showing shows the simulation model used to calculate the energy evolution of the
the HOMO (blue), LUMO (green) and SVIC trapping (red) states. The CsPbBr3 nanocrystal as a function of particle distance. CB, conduction
SVIC states are formed owing to unsaturated p orbitals of surface Br sites. band.
c, PDOS of the CsPbBr3 QD. The SVIC state is located near the Fermi

© 2018 Springer Nature Limited. All rights reserved.

 RESEARCH LETTER

Extended Data Fig. 9 | Characterization of absorption cross-section and σ, was determined by ε = NAσ/(1,000 × ln10), where NA is Avogadro’s
transient luminescence spectra of the CsPbBr3 nanocrystals. number. c, Photoluminescence (PL) lifetime of a single CsPbBr3
a, Absorption spectra of CsPbBr3 nanocrystals dispersed in cyclohexane perovskite nanocrystal. d, Second-order correlation function, g2(τ), of
at different concentrations. b, Absorption as a function of concentration the nanocrystal. The value g2(0) = 0.16 confirms the single-quantum-
for CsPbBr3 nanocrystals. The molar extinction coefficient, ε, was emitter nature of the photon emission. e, Fluorescence intermittency
determined by applying the Beer–Lambert law A = εcL, where A is the trace recorded for a single CsPbBr3 perovskite nanocrystal. The recorded
absorbance, c is the molar concentration (mol l−1) and L is the optical photoluminescence intensity reaches more than 2,000 counts per
path length (1 cm) through the sample. The absorption cross-section, 20-ms bin.

© 2018 Springer Nature Limited. All rights reserved.

 LETTER RESEARCH

Extended Data Fig. 10 | Performance of X-ray detection system using radioluminescence intensity at 532 nm in response to an X-ray dose
CsPbBr3 nanocrystal scintillators. a, Normalized radioluminescence rate of 0.013–278 µGy s−1. e, Emission spectra of CsPbBr3 nanocrystal
intensity of CsPbBr3 nanocrystals of different thickness under X-ray scintillator in response to an X-ray dose rate of 0.013–278 µGy s−1.
excitation at a voltage of 50 kV. b, Normalized radioluminescence intensity f, Radioluminescence (RL) intensity as a function of the X-ray dose rate
as a function of perovskite nanocrystal film thickness. c, X-ray output shown in e.
spectra recorded at 10, 20, 30, 40 and 50 kV. d, Kinetic measurement of

© 2018 Springer Nature Limited. All rights reserved.

 RESEARCH LETTER

Extended Data Fig. 11 | Direct X-ray imaging and multiplexed nanoparticles. g, Multicolour luminescence spectra of the perovskite
labelling for in vivo optical imaging using perovskite nanocrystal nanocrystals under X-ray irradiation at a voltage of 50 kV. The materials’
scintillators. a–d, A flexible flat cable (a) and needle-implanted pork compositions are CsPbBr3, CsPbBr1.5I1.5 and CsPbBr1.2I1.8 for green (G),
tissue (b) were imaged with bright-field and X-ray imaging (c, d). orange (O) and red (R) emissions, respectively. We note that the risk
We note that the CsPbBr3 nanocrystal scintillator platform shown in of lead toxicity must be considered during experimentation. h, Bright-
Fig. 3d was used for the X-ray phase contrast imaging. In both cases, field and multicolour luminescent in vivo imaging in mice under X-ray
the X-ray images clearly reveal the presence of metallic wires embedded excitation at a voltage of 50 kV. The X-ray-induced luminescence was
in the cable and pork tissue. e, Synthesis of CsPbBr3/SiO2 core-shell recorded by a charge-coupled-device camera equipped with three optical
nanoparticles with a hydrophobic surface for protection against moisture. filters at 530 nm, 630 nm and 670 nm.
RT, room temperature. f, TEM image of the as-prepared CsPbBr3/SiO2

© 2018 Springer Nature Limited. All rights reserved.

 LETTER RESEARCH

Extended Data Table 1 | Scintillation characteristics for different materials

© 2018 Springer Nature Limited. All rights reserved.

 RESEARCH LETTER

Extended Data Table 2 | Properties of perovskite nanocrystals and bulk crystals used for X-ray detection

Data are from this work and from refs 12–16,18. MA, CH3NH3.
a
We note that scintillators are a special class of luminescence materials that have been most widely used for radiation detection by converting high-energy X-ray photons into visible light—a sensing
process that is different in semiconductor materials, where the dominant process is photon-to-current conversion. The as-developed perovskite nanocrystal scintillators are solution-processable and
can respond to X-rays with multicolour output, which can be readily recorded by a common digital camera.
b
We note that the organic–inorganic hybrid QDs are less photostable than their purely inorganic counterparts and their synthesis is relatively difficult to scale up owing to the need for stringent control
over the reaction conditions.

© 2018 Springer Nature Limited. All rights reserved.