SOLAR CELLS
Stabilizing heterostructures
of soft perovskite semiconductors
Yanbo Wang^1 , Tianhao Wu^1 , Julien Barbaud^1 , Weiyu Kong^1 , Danyu Cui^1 , Han Chen^1 ,
Xudong Yang1,2, Liyuan Han1,2,3
Here we report a solution-processing strategy to stabilize the perovskite-based heterostructure.
Strong Pb–Cl and Pb–O bonds formed between a [CH(NH 2 ) 2 ]x[CH 3 NH 3 ] 1 −xPb1+yI 3 film
with a Pb-rich surface and a chlorinated graphene oxide layer. The constructed
heterostructure can selectively extract photogenerated charge carriers and impede the
loss of decomposed components from soft perovskites, thereby reducing damage to
the organic charge-transporting semiconductors. Perovskite solar cells with an aperture
area of 1.02 square centimeters maintained 90% of their initial efficiency of 21% after
operation at the maximum power point under AM1.5G solar light (100 milliwatts per
square centimeter) at 60°C for 1000 hours. The stabilized output efficiency of the aged
device was further certified by an accredited test center.
T
he performance of perovskite solar cells
(PSCs) relies on generation and extraction
of charge carriers in working devices ( 1 – 6 ).
Generally, it is the semiconductor hetero-
structure formed by perovskites and organic
or inorganic electron-transporting layers (ETLs)
and hole-transporting layers (HTLs) that enables
photogenerated charge-carrier extraction. Con-
struction of defect-free heterostructures is criti-
cal for large-scale application of perovskite-based
optoelectronic devices ( 7 – 9 ). On the perovskite
side of the heterostructure, component elements
are assembled by relatively weak chemical bonds,
including ionic bonds, hydrogen bonds, and van
der Waals interactions ( 10 – 12 ). Because of their
weak bonding nature, the soft crystal lattice of
perovskitesiseasilydecomposed,normallystart-
ing at the surface, and leads to great difficulties
in forming stable heterostructures on perovskite
surfaces. Several reports have demonstrated that
perovskite components can permeate through
theETLsandHTLs,disorderingthefavored
heterostructure and decreasing the charge extrac-
tion ( 13 , 14 ). Degradation of the heterostructure
is accelerated by multiple factors such as illumi-
nation, heat, and electric fields ( 15 – 17 ).
Here we report a solution-processing strategy
to stabilize the perovskite heterostructure by
forming strong chemical bonds at the surface
of soft perovskite films that can largely impede
the loss of perovskite components, resulting in
less damage to the organic HTL. In addition, the
band offset of the heterostructure can benefit
theholeextractionbetweentheperovskiteandthe
HTLs. We fabricated PSCs with efficiencies ap-
proaching 21% on an aperture area of 1.02 cm^2.
The PSC with stabilized heterostructure exhibited
excellent operational stability, maintaining 90%
of its initial value after aging under operation
conditions of AM1.5G solar light, 100 mW cm−^2
at the maximum power point, under 60°C for
1000 hours. We also sent the aged device to a
public test center [Calibration, Standards and
Measurement Team at the Research Center for
Photovoltaics, National Institute of Advanced In-
dustrial Science and Technology (AIST), Japan];
a certified stabilized efficiency of 18.6% was ob-
tained, still maintaining ~90% of its initial value.
The heterostructure consists of (i) a perovskite
film with a surface rich in Pb and (ii) a chlorinated
graphene oxide (Cl-GO) layer, where strong
Pb–Cl and Pb–O bonds are formed to join the two
layers. Figure 1A illustrates the formation of the
Pb surface–rich perovskite film of [CH(NH 2 ) 2 ]x
[CH 3 NH 3 ] 1 −xPb1+yI 3 (also noted as FAxMA 1 −xPb1+yI 3 ).
We used a dilute Pb(SCN) 2 solution as the Pb
source that was spin-coated onto the perovskite
film surface. The perovskite film with a surface
rich in Pb was formed by heat treatment to
remove volatile organic components like FASCN
(formamidinium thiocyanate) or MASCN (meth-
ylamine thiocyanate).
We probed the surface of the perovskite film by
x-ray photoelectron spectroscopy (XPS) (Fig. 1B).
The peak intensity of Pb 4f core levels increased
when more Pb(SCN) 2 was spin-coated onto the
perovskite surface, whereas the peak intensity of
I 3d core levels decreased (Fig. 1C), indicating the
fabrication of a perovskite layer with a surface
rich in Pb. The top morphology of the perovskite
layer was not changed, as determined by scanning
electron microscopy (Fig. 1D and fig. S1, A to E).
TheroughnessofthePbsurface–rich perovskite
layer was 25.1 ± 2.1 nm (Fig. 1E). The correspond-
ing surface potential mapping and phase image
are shown in fig. S2, A and B. Notably, Pb(COOH) 2
had an effect similar to Pb(SCN) 2 ,forming
Pb surface–rich perovskite layers.To form a stable heterostructure on the perov-
skitelayer,weconstructedstrongchemicalbonds
of Pb–OandPb–Cl by deposition of Cl-GO. A C-Cl
peak at the binding energy of 289 eV appeared
in the x-ray photoelectron spectrum of the C 1s
corelevelinCl-GO,incontrasttothex-rayphoto-
electron spectrum of GO (fig. S3, A and B) ( 18 ). We
used scanning electronmicroscopy (SEM) to ob-
serve the perovskite/Cl-GO layer in comparison
with the perovskite/GO layer. The surface of the
perovskite/GO film was rough with poor uniform-
ity (Fig. 2A), but Cl-GO spread well on the sur-
face of perovskite film (Fig. 2B). Atomic force
microscopy (AFM) over a large area (25mmby
25 mm) on the film surface (Fig. 2, C and D) revealed
surface roughnesses of 62.6 ± 24.31 nm and 24.6 ±
2.37 nm for perovskite/GO and perovskite/Cl-GO,
respectively.
We also conducted Kelvin probe force micros-
copy (KPFM) to compare the work functions
(WFs) of the corresponding three samples (figs.
S2A and S4, A and B) ( 19 ). The surface potential
distribution of perovskite/GO was relatively non-
uniform compared with the potential distribu-
tion of the other two samples. After the calibration
of the surface potential of the tip with Au reference
(yAu=5.10 eV), the WFs of each sample were 5.36,
5.33, and 5.43 eV, for perovskite, perovskite/GO,
and perovskite/Cl-GO, respectively. The small
difference in WFs between the perovskite and
perovskite/GO samples was likely the result of the
poor coverage or contact of GO that would not
obviously change the WFs of perovskite. How-
ever, the WF of perovskite/Cl-GO was much dif-
ferent from that of perovskite and was near the
Fermi level (EF) of Cl-GO measured by ultraviolet
photoelectron spectroscopy (UPS) (fig. S5, A and B)
( 20 ), indicating that Cl-GO made uniform contact
with the surface of perovskite with high coverage.
We measured the x-ray photoelectron spectra
of perovskite, perovskite/GO, and perovskite/
Cl-GO to evaluate whether the good contact and
coverage of perovskite/Cl-GO was the result of
strong chemical bonding.Noobviouschangein
the Pb 4f core level was observed between pe-
rovskite and perovskite/GO (Fig. 2E) ( 21 ), and we
attributed this result to the poor coverage or con-
tact of GO. However, the Pb 4f core level shifted
up by 0.3 eV for perovskite/Cl-GO. We analyzed
the O 1s core levels in x-ray photoelectron spec-
tra for the pure GO, Cl-GO, perovskite/GO, and
perovskite/Cl-GO (Fig. 2F). The O 1s core level of
Cl-GO shifted from 532.21 to 532.52 eV, which
indicates that O in Cl-GO would have stronger
electron-withdrawing properties. In addition, the
O 1s binding energy of perovskite/Cl-GO is ~0.10 eV
lower than that of perovskite/GO, indicating that
the oxidation state of O in Cl-GO is even lower
than that in GO. These XPS results indicate that
the Pb–O bond within perovskite/Cl-GO is stron-
ger than that of perovskite/GO. In addition, we
also observed the Cl 2p core levels of Cl-GO at
198.85 and 200.30 eV shift to 197.85 and 199.40 eV
in perovskite/Cl-GO, respectively (Fig. 2G), indi-
cating the formation of Pb–Cl bonds ( 22 ).
We used density functional theory (DFT) to
calculate the electron density profiles of O in GORESEARCH
Wanget al.,Science 365 , 687–691 (2019) 16 August 2019 1of5
(^1) State Key Laboratory of Metal Matrix Composites, Shanghai
Jiao Tong University, Shanghai 200240, China.^2 Center of
Hydrogen Science, School of Materials Science and
Engineering, Shanghai Jiao Tong University, Shanghai
200240, China.^3 Photovoltaic Materials Group, Center for
Green Research on Energy and Environmental Materials,
National Institute for Materials Science, Tsukuba, Ibaraki
305-0047, Japan.
*Corresponding author. Email: [email protected] (X.Y.);
[email protected] (L.H.)