Universal Approach toward Hysteresis-Free Perovskite Solar Cell via Defect Engineering

Article abstract of DOI:10.1021/jacs.7b10430

Organic-inorganic halide perovskite is believed to be a potential candidate for high efficiency solar cells because power conversion efficiency (PCE) was certified to be more than 22%. Nevertheless, mismatch of PCE due to current density (J)-voltage (V) hysteresis in perovskite solar cells is an obstacle to overcome. There has been much lively debate on the origin of J-V hysteresis; however, effective methodology to solve the hysteric problem has not been developed. Here we report a universal approach for hysteresis-free perovskite solar cells via defect engineering. A severe hysteresis observed from the normal mesoscopic structure employing TiO₂ and spiro-MeOTAD is almost removed or does not exist upon doping the pure perovskites, CH₃NH₃PbI₃ and HC(NH₂)₂PbI₃, and the mixed cation/anion perovskites, FA_0.85MA_0.15PbI_2.55Br_0.45 and FA_0.85MA_0.1Cs_0.05PbI_2.7Br_0.3, with potassium iodide. Substantial reductions in low-frequency capacitance and bulk trap density are measured from the KI-doped perovskite, which is indicative of trap-hysteresis correlation. A series of experiments with alkali metal iodides of LiI, NaI, KI, RbI and CsI reveals that potassium ion is the right element for hysteresis-free perovskite. Theoretical studies suggest that the atomistic origin of the hysteresis of perovskite solar cells is not the migration of iodide vacancy but results from the formation of iodide Frenkel defect. Potassium ion is able to prevent the formation of Frenkel defect since K⁺ energetically prefers the interstitial site. A complete removal of hysteresis is more pronounced at mixed perovskite system as compared to pure perovskites, which is explained by lower formation energy of K interstitial (-0.65 V for CH₃NH₃PbI₃ vs -1.17 V for mixed perovskite). The developed KI doping methodology is universally adapted for hysteresis-free perovskite regardless of perovskite composition and device structure.

Full text of DOI:10.1021/jacs.7b10430

Subscriber access provided by UNIVERSITY OF ADELAIDE LIBRARIES
Article
Universal Approach toward Hysteresis-Free
Perovskite Solar Cell via Defect Engineering
Dae-Yong Son, Seul-Gi Kim, Ja-Young Seo, Seon-Hee
Lee, Hyunjung Shin, Donghwa Lee, and Nam-Gyu Park
J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.7b10430 • Publication Date (Web): 04 Jan 2018
Downloaded from http://pubs.acs.org on January 4, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of the American Chemical Society is published by the American Chemical
Society. 1155 Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 24
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Universal Approach toward Hysteresis-Free Perovskite
Solar Cell via Defect Engineering
2
†
1†
1
2
2
DaeꢀYong Son , SeulꢀGi Kim , JaꢀYoung Seo , SeonꢀHee Lee , Hyunjung Shin ,
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
1
Donghwa Lee and NamꢀGyu Park *
1
School of Chemical Engineering, Sungkyunkwan University, Suwon 16419, Korea
2
Department of Energy Science, Sungkyunkwan University, Suwon 16419, Korea
3
Department of Materials Science and Engineering Pohang University of Science and
Technology (POSTECH), Pohang 37673, Korea
†
These authors equally contributed to this work
*
Corresponding author
Eꢀmail: npark@skku.edu, Tel: +82ꢀ31ꢀ290ꢀ7241
ABSTRACT
Organicꢀinorganic halide perovskite is believed to be potential candidate for high efficiency
solar cell because power conversion efficiency (PCE) was certified to be more than 22%.
Nevertheless, mismatch of PCE due to current density (J)ꢀvoltage (V) hysteresis in perovskite
solar cell is obstacle to overcome. There has been much lively debate on the origin of JꢀV
hysteresis, however, effective methodology to solve the hysteric problem has not been
developed. Here we report universal approach for hysteresisꢀfree perovskite solar cell via
defect engineering. A severe hysteresis observed from the normal mesoscopic structure
employing TiO
pure perovskites, CH
FA0.85MA0.15PbI2.55Br0.45 and FA0.85MA Cs PbI Br , with potassium iodide. Substantial
2
and spiroꢀMeOTAD is almost removed or does not exist upon doping the
3
NH PbI and HC(NH PbI , and the mixed cation/anion perovskites,
3
3
2
)
2
3
0.1
0.05
2.7
0.3
reductions in lowꢀfrequency capacitance and bulk trap density are measured from the KIꢀ
1
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 2 of 24
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
doped perovskite, which is indicative of trapꢀhysteresis correlation. A series of experiments
with alkali metal iodides of LiI, NaI, KI, RbI and CsI reveals that potassium ion is the right
element for hysteresisꢀfree perovskite. Theoretical studies suggest that the atomistic origin of
the hysteresis of perovskite solar cells is not the migration of iodide vacancy but results from
the formation of iodide Frenkel defect. Potassium ion is able to prevent the formation of
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
+
Frenkel defect since K energetically prefers the interstitial site. A complete removal of
hysteresis is more pronounced at mixed perovskite system as compared to pure perovskites,
3 3 3
which is explained by lower formation energy of K interstitial (ꢀ0.65 V for CH NH PbI vs ꢀ
1.17 V for mixed perovskite). The developed KI doping methodology is universally adapted
for hysteresisꢀfree perovskite regardless of perovskite composition and device structure.
Keywords: doping, hysteresis, perovskite solar cell, potassium iodide, Frenkel defect
2
ACS Paragon Plus Environment

Page 3 of 24
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
INTRODUCTION
Perovskite solar cells (PSCs) based on organicꢀinorganic halide materials attract a
great deal of attention due to superb photovoltaic properties. The first report on solidꢀstate
perovskite solar cell with a power conversion efficiency (PCE) of 9.7% and 500 hꢀlongꢀtermꢀ
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
stability in 2012, following the two reports at the early stage on perovskiteꢀsensitized liquid
2
3
junction solar cells in 2009 and 2011, has an enormous impact. It certainly opens up new
vistas of the nextꢀgeneration photovoltaics with unprecedented performances. PCE was
swiftly improved up to 22.1% through wellꢀengineered devices followed by fundamental
4
understanding on perovskite layers. Although the certified PCE of PSC is comparable to or
even higher than conventional thin film solar cells based on semiconductors of
5
multicrystalline Si (~22%), CIGS (~21.7%) or CdTe (~21%), PSC is suffering from
anomalous currentꢀvoltage hysteresis observed at different voltage scan direction. The
6
hysteresis was first issued by Snaith et al. in 2014, where hysteresis is associated with
perovskite material itself, selective contact materials, and external scanning parameters such
as scanning rate and external electric field. Recently, it was suggested that hysteresis could
7,8
have an adverse effect on stability. Thus hysteresisꢀfree technology is of importance in
PCSs. It has been proposed that several factors might be related to the origin of hysteresis
including (1) ferroelectricity due to the switchable polarization induced by breakdown of
9,10
centrosymmetry,
(2) modulation of charge transport by filling (trapping) and releasing
11
(detrapping) at deep trap sites created by defects, and/or (3) ion migration associated with
1
2ꢀ17
change in interfacial field and barriers resulted from accumulation of ions at interfaces,
Technical approaches were proposed to suppress the hysteresis: promoting the interfacial
18
charge transfer process, hindering the ion migration, and reducing the amount of defects.
19
20,21
Growth of grain size and grain boundary engineering
3
were found to reduce hysteresis.
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 24
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Impact of selective contacts on hysteresis was reported, where the normal structure with the
electron transporting TiO layer and the hole transporting spiroꢀMeOTAD layer exhibited
2
relatively noticeable hysteresis as compared to the NiO and PCBM based inverted structure
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
22,23
with almost no hysteresis.
Although ion migration has been regarded as a dominant factor
18
affecting the hysteresis, charge diffusion length and surface recombination were suggested
to play more critical role in the hysteresis from the viewpoint that most of highly efficient
24
devices showed small hysteresis. Despite much effort to better understand the origin of the
hysteresis and minimize the hysteresis, the hysteresis in PSCs is still under lively debate and
no effective methodology is discovered.
Here, we report a universal method to minimize currentꢀvoltage hysteresis observed
in the mesoscopic and planar perovskite solar cells having TiO
2
electron transporting layers.
We have considered that trap states generated by defects are one of potential factors causing
2
5
hysteresis because the defect deep in the band gap can capture electrons and/or holes. In our
previous work, we found a strong correlation between trap density and hysteresis in which
hysteresis was gradually reduced as trap density decreased as evidenced by reduction in nonꢀ
20
radiative recombination. However, the surface and/or grain boundary engineering does not
seem to be sufficient to remove completely the hysteresis, from which we learn that
engineering trap states of bulk perovskite will be more effective way. Since perovskite
structure can be described in terms of cubic close packing model forming the basic lattice
comprising 75% of X ions and 25% of A ions on the same lattice plane and placing B ions in
26
h
octahedral sites, O , between the lattice planes in ABX
3
perovskite, energetically favorable
25
point defects viz vacancy, interstitial, and antisite are available. Frenkel and Schottky
defects are also expected in case that an ion moves into an interstitial site and creates a
vacancy and stoichiometric cationꢀanion vacancy, respectively. Among the point defects,
4
ACS Paragon Plus Environment

Page 5 of 24
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
3 3 3
interstitial of Pb and antisites of MA, Pb and I in MAPbI (MA = CH NH ) are related to
deep level being responsible for nonꢀradiative recombination and associated with the
hysteresis, which motivated us to develop a universal method to minimize hysteresis by
defect engineering. Since doping has been known to be one of promising and rather facile
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
27
methods for defect engineering, for instance thermoelectric figureꢀofꢀmerit of tin selenide or
28
photovoltaic performance of Cu
2
ZnSn(S,Se)
4
was enhanced by alkali ion doping, we have
designed doping of perovskite with alkali iodide (LiI, NaI, KI, RbI and CsI) for engineering
defects causing the hysteresis of PSCs and discovered that potassium ion plays important role
in controlling the defects.
RESULTS AND DISCUSSION
1
0 ꢁmol of potassium iodide (KI) was mixed with perovskite precursor solutions,
where MAPbI , FAPbI (FA = HC(NH ) and mixed cation/anion perovskites of
3
3
2 2
)
FA0.85MA0.15PbI2.55Br0.45 and FA0.85MA0.1Cs0.05PbI2.7Br0.3 were studied in order to investigate
effect of KI on the hysteresis of different compositional perovskites. All the devices contain a
2 2
blocking TiO layer on FTO substrate and a mesoporous TiO (diamtere of about 50 nm)
layer on the blocking film. Figure 1 shows current density (J)ꢀvoltage (V) curves of
perovskite solar cells with and without KI doping. Little difference in JꢀV curves between the
reverse scan and the forward scan are obeserved after 10 ꢁmol KI doping, while significant
mismatch in JꢀV curvers are obtained from the pristin perovskites. This emphasizes that KI
doping eliminates the hysteresis of perovskite solar cells regardless of compisition of
perovskite materials. For the mixed FA/MA and I/Br perovskite case
(FA0.85MA0.15PbI2.55Br0.45) in Figure 1A and B, average shortꢀcircuit current density (Jsc),
5
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 6 of 24
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
opneꢀcircuit voltage (Voc), fill fcator (FF) and PCE of the pristine perovskite are 22.04
2
2
mA/cm , 1.114 V, 0.698 and 17.14% at the reverse scan, respectively, and 22.10 mA/cm ,
.093 V, 0.621 and 15.01% at the forward scan, respectively (see Table S1 in supporting
1
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
information). Difffernce in FF between the reverse and forward scans (ꢀFF = [FF(reverse) –
FF(forward)]/FF(reverse)) is as large as 11.0%, which is substaintially reduced to 1.9% after
2
KI doping. As a result, the PCE of 17.55% (Jsc = 21.47 mA/cm , Voc = 1.128 V, FF = 0.725)
2
observed at the reverse scan is almost identical with that of 17.14% (J = 21.49 mA/cm , V
sc
oc
=
1.122 V, FF = 0.711) at the forward scan. For the triple cation of
FA0.85MA0.1Cs0.05PbI2.7Br0.3 in Figure 1C and D, KI doping shows the same tendency. As
listed in Table S1, a significant difference in PCE between the reverse scan (17.99%) and the
forward scan (15.08%) is observed for the pristine perovskite due to a large difference in FF
(0.730 for reverse vs 0.637 for forward, ꢀFF = 12.7%). Hysteresis nearly disappears by KI
dopping, where a PCE of 18.20% for the reverse scan is again similar to that of 17.97% for
the forward scan because of negliable ꢀFF = 0.9% (0.7302 for reverse vs 0.732 for forward).
Large hysteresis in MAPbI
doping as can be seen in Figure 1F and H. Difference in PCE (ꢀPCE = [PCE(reverse)ꢀ
PCE(forward)]/PCE(reverse)) is reduced from 26.2% to 14.5% for MAPbI and from 43.9%
3 3
and FAPbI (Figure 1E and G) is considerably reduced by KI
3
to 14.9% for FAPbI after KI doping (Table S1). Besides great reduction in ꢀFF after KI
3
doping for all the perovskites, it is interesting to note that the Voc values for the forward scan
are increased and merged with the ones for the reverse scan after KI doping. A large
19
difference in Jsc at a given bias volatge is related to timeꢀdepedent capacitive current,
leading to lower Voc at forward scan, in which unextracted current is likely to have infleunce
on Voc. Such a capcitive current might be originated from junction interface and/or bulk
6
ACS Paragon Plus Environment

Page 7 of 24
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
perovskite. Improvement of forward scanned Voc after KI doping indicates that all the current
at the given bias voltage is extracted at the forward scan. Since neither the device
configurations nor selective contact materials were not chnaged, it is valid that capacitive
current is dominately atrributed to perovskite itself. Nevetheless, modification of interface
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
between the perovskite and the electrode materials, such as TiO , cannot be ruled out in case
that precipitation of KI happens prior to foming perovskite phase. In this case, interface
modification can also take part in reducing hysteresis. When considering the the hysteresis
29
occurs not only in perovskite solar cell but also dyeꢀensitized solar cell and silicon solar cell,
3
0
latter two cases are related to trappingꢀdetrapping process and internal capacitance at short
scan time, respectively, removal (or significantly reduced) of hysteresis in the postssiumꢀ
doped perovskite materials could be related to removal of trapped charges in perovskite. Scan
rateꢀindependent hysteresisꢀfree behavior is also observed for the KIꢀdoped mixed cation
perovskites (Figure S1 and Table S2 in the supporting infromation) as compared to strong
dependence of photovoltaic performance on scan rate for the pristine perovskites without KI.
7
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 8 of 24
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1. Current density (J)ꢀvoltage (V) curves of perovskite solar cells employing different
perovskite materials FA0.85MA0.15PbI2.55Br0.45, FA0.85MA0.1Cs0.05PbI2.7Br0.3, MAPbI
3
and
doped with and without 10 ꢁmol KI, measured at reverse (filled circles) and forward
empty circles) scans at the scan rate of 130 mV/s (= voltage settling time of 200 ms) under
FAPbI
3
(
2
2
AM 1.5G one sun illumination (100 mW/cm ). Aperture mask area was 0.125 cm .
8
ACS Paragon Plus Environment

Page 9 of 24
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
We further invetigate effect of alkali metal ions on hysteresis of perovskite solar cells.
Perovskite precursor solution with 10 ꢁmol of alkali metal iodide (LiI, NaI, KI, RbI or CsI)
was prepared to study effect of alkali metal ions on hysteresis, where
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3 3
(FAPbI )0.875(CsPbBr )0.125 was chosen as pristine perovskite composition. We have selected
FAꢀbased perovskite with a elemetal substitution with Cs and Br as a control perovskite
because incorporation of Cs in FA site improves not only PCE but also moistureꢀ and photoꢀ
3
1
32
stability and is suitable for top cell in tandem structure. Optimal composition in (FAPbI )
3
1ꢀ
x
3 x
(CsPbBr ) is determined by the Urbach energy estimated from absorbance and Tauc plot in
Figure S2, where x = 0.125 shows minimum Urbach energy among the studied ratio (x
33
values are varied by considering tolerance factor ). In addiiton, this compisition of x = 0.125
leads to largest grains as confimed by scanning electron microscpic (SEM) images in Figure
S3 (grain size can be more exactly determined by electron back scattering technique).
Hysteresis is strongley dependent on alkali metal ions as shown in Figure 2A-E.
+
+
When increasing ionic radius of alkali metal ions from Li to K (Figure 2A-C), hysteresis
+
tends to decrease and disappears at K . Hysteresis appears again upon futher increase in ionic
+
+
+
radius such as Rb and Cs (Figure 2 D and E) and is apprent in Cs . Tiny amount of alkali
metal ions is expected to be placed in cation vacancy and/or interstitial. In this case, chemical
stability of alkali metal ions can be simply expected from cation/anion radius ratio or
estimated by theoretical calculation, which will be discussed in detail. Since it is obvious that
possasium ion removes the hysteresis, effect of KI concentration on hysteresis is further
examined. In Figure 2 F-I, no significant difference in Jsc and Voc between the reverse and
forward scans is observed as KI concentration increases from 2 ꢁmol to 10 ꢁmol, while
difference in FF is distinct at low concentration but becomes smaller and eventually negliable
9
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 10 of 24
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
1
9
at 8ꢀ10 ꢁmol. Since hysteresis was reported to be influenced by morphology, change in
morphology by KI concentration may affect hysteresis, which however can be ruled out
because of little change in morphology (Figure S4). We have examined carefully XRD
patterns (Figure S5a) and found gradual change in lattice parameter with KI addition (Figure
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
34
S5b). XRD pattern is analyzed using a freeware of EXPO2014 program, where
FAPbI 0.125 perovskite is able to be indexed as cubic phase with space group
(
3 3
)0.875(CsPbBr )
of Pmꢀ3m. Peaks sift to lower angle by 0.04 ~ 0.11 degree after doping as compared to w/o
KI perovskite. The lattice parameters are calculated to be 6.27700 ±0.002167 Å (w/o KI),
6
.29346±0.001334 Å (2 ꢁmol KI), 6.29162±0.000820 Å (4 ꢁmol KI), 6.29856±0.000877 Å
(
6 ꢁmol KI), 6.30098±0.001249 Å (8 ꢁmol KI), and 6.29306±0.000924 Å (10 ꢁmol KI). It
3
5,36,37
was reported that the lattice parameters were changed by impurity doping.
For instance,
the lattice parameter of cꢀSi was changed by 0.0007 Å by boron doping (dopant concentration
19
37
level was 2.3×10 per cubic centimeters). Variation of lattice parameter in GaAs was about
ꢀ4
18
~10 Å by Si doping (dopant concentration level was 6×10 per cubic centimeters). Thus, a
slight but gradual increase in lattice parameter (0.016 ~ 0.024 Å) indicates that potassium ion
is crystallized in the perovskite lattice. The increase in lattice parameter also underlines that
potassium ion might be placed in interstitial site rather than substitution, which is reasonable
+
+
when considering ionic size of potassium ion much smaller than FA and Cs . So, it is likely
that KI is doped in perovskite lattice and plays crucial role in eliminating the hysteresis. It is
noted that KI concetration of 15 ꢁmol degrades photovoltaic performance and higher
concentration of 20 ꢁmol generates again the hysteresis, which indicates that KI doping
amount for controlling hysteresis seems to be saturated at 10 ꢁmol. JꢀV curves and the
relavent photovoltaic parameters depending on KI concetration are displayed in Figure S6
and listed in Table S3, respectively.
10
ACS Paragon Plus Environment

Page 11 of 24
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3 3 0.125
Figure 2. (A)ꢀ(E) JꢀV curves of perovskite solar cells based on (FAPbI )0.875(CsPbBr )
doped with 10 ꢁmol of LiI, NaI, KI, RbI and CsI, measured at reverse and forward scans at
2
the scan rate of 130 mV/s under AM 1.5G one sun illumination (100 mW/cm ). The aperture
2
mask area was 0.125 cm . (F)ꢀ(I) Effect of KI concentration on statistical photovoltaic
parameters of shorꢀcircuit current density (J ), openꢀcircuit voltage (V ), fill factor (FF) and
sc
oc
power conversion efficiency (PCE) obtained at reverse (green) and forward (pink) scans,
respectively.
J
sc as a function of time at maximum power point (steadyꢀstate photocurrent) can
inform charge extarction kinetics. It takes much longer time to stabilize photocurrent for the
pristine perovskite without KI doping, whereas Jsc is immediately stabilized upon KI doping,
especially for 10 ꢁmol (Figure S7). This underlines that fast and effective extraction of
charges are realized by KI doping. In Figure 3A and B, frequencyꢀdependent capacitance is
11
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 12 of 24
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
comapred with and without KI doping. The device with KI regardless of selective contacts
shows the saturated plateau (not new peak) of capacitance in the region of 10ꢀ100 Hz, where
little variation of capacitance with frequency is consistent with no IꢀV hysteresis after KI
doping. The capacitanceꢀfrequency profile with plateau for the KI doped perovskite is similar
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
2
to that of hysteresisꢀfree inverted device. In Figure 3B, we prepared the device without any
selective contacts (FTO/PSK/Au) to confirm dependence of IꢀV hysteresis on bulk perovskite
rather than interface. Capacitanceꢀfrequency behavior of this specific structure shows similar
behavior, which supports the IꢀV hysteresis is closely related to bulk defect.
t
Trap density (n ) is evaluated based on spaceꢀchargeꢀlimited current (SCLC) and
impedance measurements using a device structure of FTO/perovskite/Au (Figure S8) in the
ꢈ
38
dark. The
represents the onset voltage of trap filled limit,
thickness of the active layer, is the vacuum permittivity (8.8542 × 10 F/cm) and ε
ꢆ
t TFL
n values are calculated from the relation of ꢀꢁꢂꢃ = ꢄꢅ ꢇ /2ꢉꢉꢊ, where V
ꢀ19
e
is electric charge (1.602 × 10 C),
d
is
is
ꢀ14
ε
0
dielectric constant. The dielectric constants are calculated from the capacitance measured in
4
39,40
high frequency region (~10 Hz).
S4, TFL decreases from 0.788 V for the pristine perovskite to 0.617 V for the 10 ꢁmol KI
doping. In addition, is similarly decreased from 42.45 to 33.53 upon 10 ꢁmol KI doping
(Table S4). Since is proportional to at the given perovskite film thickness of 520
nm (Figure S9), decreases from 1.367 × 10 cm to 0.846 × 10 cm after 10 ꢁmol KI
doping. The estimated
noted that increases again to 1.399 × 10 cm and to 1.560 × 10 cm upon higher doping
As can be seen in Figure 3C and D along with Table
V
ε
n
t
TFL
V ×ε
16
ꢀ3
16
ꢀ3
n
t
16
ꢀ3
41
t
n of ~ 10 cm is in good agreement with the reported value. It is
1
6
ꢀ3
16
ꢀ3
n
t
level of 15 ꢁmol and 20 ꢁmol, respectively. Change in trap density upon KI doping and
doping concentration is well consistent with the tendency of hysteresis, which is indicative of
12
ACS Paragon Plus Environment

Page 13 of 24
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
strong correlation between trap density and hysteresis. Interfacial trap can be shown using a
42
high frequency capacitanceꢀvoltage (CꢀV) curve (Figure S10). The increase in charge
accumulation (capacitance) at negative bias voltage of mixed (FAPbI
3 3 0.125
)0.875(CsPbBr )
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
perovskite is indicative of pꢀtype semiconducting material. As the bias voltage is applied
from +1 V to – 1 V, the CꢀV curve of pristine perovskite (w/o KI) shifts to negative voltage
due to the positive charges trapped on the interface of perovskite layer. The CꢀV curve of 10
ꢁmol KI doped perovskite shows a reduced interfacial trap state, which indicates that the
potassium iodide reduces not only the bulk trap but also interfacial trap of the perovskite.
Under the bias voltages (at either ꢀ1 or +1 voltage) it is clearly showed that higher current can
flow through the perovskite layer with KI doping as shown in Figure S11. No carriers can be
further trapped by defects in bulk as well as at the interfaces as shown by experimental results
above on trap density (nt) and CꢀV for interfacial traps. As a result, higher conductance
through the layers has been observed by CꢀAFM. Interestingly, it is noted that no
microstructural dependence on the conduction in between without and with KI doping is
observed. Both cases under the either – or + biases always show grain boundaries are better
conduction pathways.
13
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 14 of 24
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3. Capacitanceꢀfrequency (CꢀV) plots of (A) FTO/TiO
2 3 3 0.125
/(FAPbI )0.875(CsPbBr ) /
spiroꢀMeOTAD/Au and (B) FTO/(FAPbI 0.125/Au under one sun illumination.
3 3
)0.875(CsPbBr )
FTO, PSK and spiro represent fluorineꢀdoped tin oxide, perovskite and spiroꢀMeOTAD,
respectively. Dark currentꢀvoltage curve of (C) pristine (w/o KI) and (D) 10 ꢁmol KI doped
3 3
(FAPbI )0.875(CsPbBr )0.125 with the structure of FTO/Perovskite/Au.
In order to understand physical origin of the hysteresisꢀfree KIꢀdoped PSCs, the firstꢀ
principles density function theory (DFT) calculations are employed for MAPbI as a model
3
compound. We first perform various geometrical relaxations to identify any structural
3
changes of MAPbI during photovoltaic operation. Our DFT study finds that I_Frenkel defect
can be formed under more than two excess electrons; although cation Frenkel defect is
generally more common, I anion leads to Frenekel defect because of smaller size than MA
molecule. Figure 4A shows the schematic view of 2×2 perovskite MAPbI structure, where
3
14
ACS Paragon Plus Environment

Page 15 of 24
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
MA cation and I anion occupy one fourth and three fourth of FCC lattice sites while Pb
cation sits on the center interstital site of FCC lattice. Thus, among four O interstitial sites (3
h
edges and 1 center) in FCC lattice, the center of I
octahedron (black octahedron). Figure 4B shows the schematic view after the formation of
I_Frenkel defect in MAPbI . As can be clearly seen, the migration of I ion into the O
interstitial site (I ) leaves I vacancy on the lattice site (V ) and leads to the formation of
6 6
cage is only occupied to form PbI
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
h
i
I
I_Frenkel defect. Figure 4C shows the variation in energetics during the formation of
I_Frenkel defect. Initial state (IS) represents the perfect perovskite structure with I ion sitting
on lattice site, while final state (FS) indicates the I_Frenkel defect structure in which I ion
h
moves 2.05 Å toward the O interstitial site (see Figure 4B). As can be seen from the
energetics, the formation of I_Frenkel defect occurs along two steps. The first step is
associated with the rotation of MA molecule, while the second step is the result of actual
migration of I ion from the lattice to the interstitial site. The molecular rotation is necessary
since it can open up space for I migration and stabilize I ion sitting on the interstitial site by
forming hydrogen bonds. The energy barrier associated with the molecular rotation is 0.16
eV and the intermediate state is only 0.01 eV energetically higher than the initial ground state.
Thus, the reorientation of molecules can easily occur at room temperature and so the
molecular rotational step does not restrain the formation of I_Frenkel defect. Instead, the I
migration step can limit the formation of I_Frenkel defect. Our DFT calculation predicts that
the energy barrier associated with I migration is 0.47 eV and the FS is 0.06 eV energetically
higher than the IS with the presence of two excess electrons. Thus, the formation of I_Frenkel
defect is limited by the second step associated with the migration of I ion from lattice to
interstitial site.
15
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 16 of 24
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
We now discuss about the reason why the I_Frenkel defect can be stabilized under
two excess electrons, while I ion goes back to its original lattice site spontaneously without
two excess electrons. In order to understand the physical origin of the formation of I_Frenkel
defect under two excess electrons, we further look at the variation of electronic charge
density after the formation of I_Frenkel defect. Figure 4D and E shows the spatial location
of two excess electrons before and after I_Frenkel defect is formed. Before I_Frenkel defect
has formed, excess electrons are distributed on conduction band (p orbital of Pb) of MAPbI3.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
I
With I_Frenkel defect, however, the excess electrons are localized near the V and Pb, see
cyon region in Figure 4E. The localized excess electrons can explain the stabilization of
I_Frenkel defect under two excess electrons. Without two excess electrons, the existence of
I
positively charged V between two Pb cations is energetically highly unfavorable and so the
negatively charged I will spontaneously go back to the original lattice site. With two excess
i
electrons, however, the charge state of V
excess electrons. Accordingly, the negatively charged V
forms ionic bond with them. As a consequence, PbꢀV
I
is changed from positive to negative by trapping
attracts two nearby Pb cations and
ꢀPb arrangement is energetically
I
I
favored under two excess electrons. The estimated distance between two Pb cations is 3.38 Å
which is similar to the bond length of metallic Pb (3.5 Å). Thus, the defective I_Frenkel
structure can be stabilized by forming PbꢀV
I
ꢀPb configuration, soꢀcalled Pb dimer. In
summary, two excess electrons can stabilize I_Frenkel defect by forming Pb dimer on V
I
site.
Our DFT calculation predicts that the energetic preference of I_Frenkel defect can be further
enhanced by increasing the number of excess electrons. With three excess electrons, the
I_Frenkel defect structure is even energetically 0.29 eV lower than the perfect perovskite
structure. It seems that the I_Frenkel defect structure become energetically more preferred by
strengthening the formation of the Pb dimer by trapping more excess electrons. Thus, we
16
ACS Paragon Plus Environment

Page 17 of 24
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
believe that the formation of I_Frenkel defect is inevitable near the cathode at which
electrons are accumulated, which is associated with origin of hysteresis in bulk perovskite.
The question is why the hysteresis disappears by KI doping. Once light is illuminated on
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
MAPbI , photoꢀexcited charge carriers will be generated and transported into the opposite
side of electrodes. During this process, excess electrons can be accumulated on cathode and
lead to the formation of I_Frenkel defect. Since I_Frenkel defects are generated by trapping
two excess electrons, it is detrimental for accurate determination of actual PCE due to
hysteresis. In addition, the opposite charged Frenkel defect can be separated and affect long
term stability of MAPbI
stability and hysteresis of MAPbI
possible that the migration of V luckily explains the hysteresis behavior since the energetics
3
. We note here that previous attempts tried to interconnect the
3
with the migration of V may not be correct. It can be
I
I
for the formation of I_Frenkel defect is similar to the migration of V . Thus, we suggest that
I
the atomistic origin of the hysteresis of PSCs correlates with the formation of I_Frenkel
I
defect rather than the migration of V . Hence, the hysteresis can be removed by preventing
the formation of I_Frenkel defect. Since I_Frenkel defect can be formed by electron
accumulation, we can minimize the formation of I_Frenkel defect by reducing electron
accumulation at the cathode or by preventing I ion migration into the O interstitial site. In
h
that sense, the KIꢀdoping is an ideal method, since K ion fits well into the O
h
ꢀ
interstitial site
+
and prevents I ion migration. The cation/anion radius ratio between K and I (0.616) shows
+
+
+
h
how well K ion fits into the interstial O site (Figure 4F). Although Na and Rb are suitable
for doping elements in order to reduce the hysteresis, as observed in Figure 2B and D, they
+
are not as effective as K ion because of overalp with 4 coordination and 8 coordination,
+
respectively. Li ions are too small so that they can be mobile and cannot completely hinder I
17
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 18 of 24
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
+
migration into interstitial sites; cation/anion ratio is 0.315 for Li ion. On the other hand, Cs
+
ions are too large to fit into the interstitial site since the cation/anion ratio is 0.782. Thus, K
is expected to best element to cure the I_Frenkel defect, which is experimentally proved.
More strong effect of KI doping is observed for the mixed cation/anion perovskites than pure
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3 3
perovskite (MAPbI and FAPbI ), as observed in Figure 1. In order to understand the effect
of the mixed cation anion, formation energy of K interstitial is compared between pure and
mixed cation perovskite. For the pure system, the formation energy of K interstitial is ꢀ0.65
3 3
eV for MAPbI and ꢀ0.98 eV for FAPbI . Negative sign represents that K ion energetically
prefers the interstitial site. For the mixed cation system (FA0.875 MA0.125PbI2.55Br0.45), the
formation energy of K interstitial is lowered to ꢀ1.17 eV. The lowered formation energy can
be understood from enhanced structural distortion induced by the shape difference between
MA and FA. The enhanced structral distortion accommodates K ion into the interstitial site
more easily and so the defect formation energy of K interstitial is lowered for the mixed
cation anion perovskites. In conclusion, the mixed cation anion system shows better
performance for removing hysterisis by accommodating K ion into the interstitial site more
readily.
18
ACS Paragon Plus Environment

Page 19 of 24
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 4. Schematic view of MAPbI
3
hybrid perovskite structure (A) for perfect and (B) with
I_ Frenkel defect. Different species are shown as Green(C), Yellow(N), Whte(H), Purple(I),
and Black(Pb). (C) Relative energy profile of the formation of I_Frenkel defect under two
excess electrons; initial state (IS) and final state (FS) represents the structure before and after
the formation of I Frenkel defect. Two step processes are shown with molecular rotational
barrier (TS1) and I migration barrier (TS2). Spatial location of two excess electrons (D) for
perfect and (E) with I_Frenkel defect is shown; Blue and red represent the region where
electron density increases and decreases; isosurfaces are plotted at the level of +/ꢀ0.03e. (F)
ꢀ
Cation/anion radius ratio between alkali metal cation and I anion.
CONCLUSIONS
Addition of tiny amount of potassium iodide in perovskite materials (pure perovskites and
mixed cation/anion perovskites) was found to reduce currentꢀvoltage hysteresis to great
extent or even removed completely the hysteresis in case of the mixed cation/anion
perovskites. A series of experiments with alkali metal iodides (LiI, NaI, KI, RbI and CsI)
confirmed that K ion is right element for this purpose. Bulk and interfacial trap densities were
reduced and lowꢀfrequency capacitance was lowered by KI doping. Theoretical calculation
revealed that atomistic origin of hysteresis was not related to iodide migration but associated
with Frenkel defect. Potassium ion was best for preventing Frenkel defect as compared to
19
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 20 of 24
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
other alkali metal cations and KI doping effect was more much more pronounced at mixed
cation perovskites because of energetically more stabilized K interstitial. According to the
computational simulation combined with experimental observations, we propose that defect
engineering via KI doping is universal method to solve the hysteric problem in perovskite
solar cells, regardless of perovskite composition and device structure.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACKNOWLEDGEMENTS
This work was supported by the National Research Foundation of Korea (NRF) grants funded
by the Ministry of Science, ICT Future Planning (MSIP) of Korea under contracts NRFꢀ
2
012M3A6A7054861 and NRFꢀ2014M3A6A7060583 (Global Frontier R&D Program on
Center for Multiscale Energy System) and NRFꢀ2016M3D1A1027663 and NRFꢀ
016M3D1A1027664 (Future Materials Discovery Program). This work was also supported
2
by Basic Science Research Program through the NRF under contact 2016R1A2B3008845
and NRFꢀ2017H1A2A1046990 (NRFꢀ2017ꢀFostering Core Leaders of the Future Basic
Science Program/Global Ph.D. Fellowship Program).
AUTHOR CONTRIBUTIONS
N.G.P. conceived of concept and experiments, performed data analysis and prepared the
manuscript. D.Y.S. and S.G.K. prepared materials, fabricated devices, and measured optoꢀ
electronic properties. J.Y.S performed and analyzed Impedance spectroscopy. H.J.S. and
S.H.L. measured CꢀAFM and wrote the relevant part. D.L. performed DFT calculation and
wrote the relevant part. All authors discussed the results and commented on the manuscript.
References
1
. Kim, H.ꢀS.; Lee, C.ꢀR.; Im, J.ꢀH.; Lee, K.ꢀB.; Moehl, T.; Marchioro, A.; Moon, S.ꢀJ.;
HumphryꢀBaker, R.; Yum, J.ꢀH.; Moser, J. E.; Grätzel, M.; Park, N.ꢀG. Sci. Rep. 2012, , 591.
. Kojima, A.; Teshima, K.; Shirai, Y.; Miyasaka, T. J. Am. Chem. Soc. 2009, 131, 6050–
2
2
6
3
4
051.
. Im, J.ꢀH.; Lee, C.ꢀR.; Lee, J.ꢀW.; Park, S.ꢀW.; Park, N.ꢀG. Nanoscale 2011, 3, 4088.
. Yang, W. S.; Park, B.ꢀW.; Jung, E. H.; Jeon, N. J.; Kim, Y. C.; Lee, D. U.; Shin, S.
S.; Seo, J.; Kim, E. K.; Noh, J. H. Science 2017, 356, 1376–1379.
5. Green, M. A.; Hishikawa, Y.; Warta, W.; Dunlop, E. D.; Levi, D. H.; HohlꢀEbinger, J.;
20
ACS Paragon Plus Environment

Page 21 of 24
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
HoꢀBaillie, A. W. H. Prog. Photovolt. Res. Appl. 2017, 25, 668–676.
6
. Snaith, H. J.; Abate, A.; Ball, J. M.; Eperon, G. E.; Leijtens, T.; Noel, N. K.; Stranks, S. D.;
Wang, J. T.; Wojciechowski, K.; Zhang, W. J Phys. Chem. Lett. 2014, , 1511–1515.
. Park, N.ꢀG.; Grätzel, M.; Miyasaka, T.; Zhu, K.; Emery, K. Nat. Energy 2016, , 16152.
. Yuan, Y.; Huang, J. Acc. Chem. Res. 2016, 49, 286–293.
5
7
8
1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
9. Wei, J.; Zhao, Y.; Li, H.; Li, G.; Pan, J.; Xu, D.; Zhao, Q.; Yu, D. J. Phys. Chem. Lett
.
2014,
5, 3937–3945.
10. Seol, D.; Jeong, A.; Han, M. H.; Seo, S.; Yoo, T. S.; Choi, W. S.; Jung, H. S.; Shin, H.;
Kim, Y. Adv. Funct. Mater. 2017, 27, 1701924.
1. Miller, D. W.; Eperon, G. E.; Roe, E. T.; Warren, C. W.; Snaith, H. J.; Lonergan, M. C.
Appl. Phys. Lett. 2016, 109, 153902.
2. Xiao, Z.; Yuan, Y.; Shao, Y.; Wang, Q.; Dong, Q.; Bi, C.; Sharma, P.; Gruverman, A.;
1
1
Huang, J. Nat. Mater. 2015, 14, 193–198.
13. Li, C.; Tscheuschner, S.; Paulus, F.; Hopkinson, P. E.; Kiessling, J.; Kohler, A.; Vaynzof,
Y.; Huettner, S. Adv. Mater. 2016, 28, 2446–2454.
1
4. Zhang, T.; Chen, H.; Bai, Y.; Xiao, S.; Zhu, L.; Hu, C.; Xue, Q.; Yang, S. Nano Energy
2016, 26, 620–630.
5. Richardson, G.; O'Kane, S. E. J.; Niemann, R. G.; Peltola, T. A.; Foster, J. M.; Cameron,
P. J.; Walker, A. B. Energy Environ. Sci. 2016, (4), 1476–1485.
6. Meloni, S.; Moehl, T.; Tress, W.; Franckevicius, M.; Saliba, M.; Lee, Y. H.; Gao, P.;
Nazeeruddin, M. K.; Zakeeruddin, S. M.; Rothlisberger, U.; Graetzel, M. Nat. Commun. 2016,
1
9
1
7
, 10334.
17. Calado, P.; Telford, A. M.; Bryant, D.; Li, X.; Nelson, J.; O'Regan, B. C.; Barnes, P. R.
Nat. Commun. 2016,
7
, 13831.
8. Li, C.; Guerrero, A.; Zhong, Y.; Huettner, S. J. Phys.: Condens. Matter. 2017, 29, 193001.
9. Kim, H. S.; Park, N. G. J. Phys. Chem. Lett. 2014, , 2927–2934.
1
1
5
20. Lee, J. W.; Kim, S. G.; Bae, S. H.; Lee, D. K.; Lin, O.; Yang, Y.; Park, N. G. Nano Lett.
2017, 17, 4270–4276.
21. Son, D.ꢀY.; Lee, J.ꢀW.; Choi, Y. J.; Jang, I.ꢀH.; Lee, S.; Yoo, P. J.; Shin, H.; Ahn, N.;
Choi, M.; Kim, D.; Park, N.ꢀG. Nat. Energy 2016,
1, 16081.
22. Kim, H. S.; Jang, I. H.; Ahn, N.; Choi, M.; Guerrero, A.; Bisquert, J.; Park, N. G. J. Phys.
21
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 22 of 24
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Chem. Lett. 2015,
3. Seo, S.; Park, I. J.; Kim, M.; Lee, S.; Bae, C.; Jung, H. S.; Park, N. G.; Kim, J. Y.; Shin,
H. Nanoscale 2016, , 11403–11412.
4. Neukom, M. T.; Züfle, S.; Knapp, E.; Makha, M.; Hany, R.; Ruhstaller, B. Sol. Energ.
6, 4633–4639.
2
8
2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Mat. Sol. 2017, 169, 159–166.
25. Yin, W.ꢀJ.; Shi, T.; Yan, Y. Appl. Phys. Lett. 2014, 104, 063903.
26. Johnsson, M.; Lemmens, P. J. Phys.: Condens. Matter. 2008, 20, 264001.
27. Wei, T. R.; Tan, G.; Zhang, X.; Wu, C. F.; Li, J. F.; Dravid, V. P.; Snyder, G. J.;
Kanatzidis, M. G. J. Am. Chem. Soc. 2016, 138, 8875–8882.
8. Hsieh, Y.ꢀT.; Han, Q.; Jiang, C.; Song, T.ꢀB.; Chen, H.; Meng, L.; Zhou, H.; Yang, Y.
Adv. Energy Mater. 2016, , 1502386.
9. Wu, F.; Li, X.; Tong, Y.; Zhang, T. J. Power Sources 2017, 342, 704–708.
0. Herman, M.; Jankovec, M.; Topič, M. Int. J. Photoenergy 2012, 151452, 1–11.
2
6
2
3
31. Lee, J.ꢀW.; Kim, D.ꢀH.; Kim, H.ꢀS.; Seo, S.ꢀW.; Cho, S. M.; Park, N.ꢀG. Adv. Energy
Mater. 2015, , 1501310.
2. McMeekin, D. P.; Sadoughi, G.; Rehman, W.; Eperon, G. E.; Saliba, M.; Horantner, M.
T.; Haghighirad, A.; Sakai, N.; Korte, L.; Rech, B.; Johnston, M. B.; Herz, L. M.; Snaith, H. J.
Science 2016, 351, 151–155.
5
3
A
3
2
3
3. Li, Z.; Yang, M.; Park, J.ꢀS.; Wei, S.ꢀH.; Berry, J. J.; Zhu, K. Chem. Mater. 2016, 28,
84–292.
4. Altomare, A.; Cuocci, C.; Giacovazzo, C.; Moliterni, A.; Rizzi, R.; Corriero, N.;
Falcicchio, A. J. Appl. Cryst. 2013, 46, 1231ꢀ1235.
3
3
3
3
5. Fewster, P. F.; Willoughby, A. F. W. J. Cryst. Growth 1980, 50, 648ꢀ653.
6. Celotti, G; Nobili, D.; Ostoja, P. J. Mater. Sci. 1974, , 821ꢀ828
9
7. Kucytowski, J.; Wokulska, K. Cryst. Res. Technol. 2005, 40, 424 ꢀ 428.
8. Rose, A. Phys. Rev. 1955, 97, 1538−1544.
39. Brus, V. V.; Kyaw, A. K. K.; Maryanchuk, P. D.; Zhang, J. Prog. Photovolt. Res. Appl.
2015, 23, 1526–1535.
40. Zhumekenov, A. A.; Saidaminov, M. I.; Haque, M. A.; Alarousu, E.; Sarmah, S. P.;
Murali, B.; Dursun, I.; Miao, X.ꢀH.; Abdelhady, A. L.; Wu, T.; Mohammed, O. F.; Bakr, O.
M. ACS Energy Lett. 2016, , 32–37.
1
22
ACS Paragon Plus Environment

Page 23 of 24
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
41. Stranks, S. D.; Burlakov, V. M.; Leijtens, T.; Ball, J. M.; Goriely, A.; Snaith, H. J. Phys.
Rev. Appl. 2014,
2, 034007.
42. S. M. Sze, Semiconductor devices, physics and technology, Wiley: New York, 1985.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
23
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 24 of 24
TOC
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
2
1
1
0
5
0
5
0
(
b)
0.76 Å
KI
KI
+
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
I
Li
K+
0
.0 0.2 0.4 0.6 0.8 1.0
2
1
1
0
5
0
5
0
I
1.38 Å
K+
25
20
25
20
15
10
5
0
.0 0.2 0.4 0.6 0.8 1.0
1
1
5
0
5
0
20
1.67 Å
1
5
10
5
Cs+
0
0
0
.0 0.2 0.4 0.6 0.8 1.0 1.2
0.0 0.2 0.4 0.6 0.8 1.0 1.2
0.0 0.2 0.4 0.6 0.8 1.0
V (V)
V (V)
V (V)
ACS Paragon Plus Environment