Kinetics of Ion Transport in Perovskite Active Layers and Its Implications for Active Layer Stability

Article abstract of DOI:10.1021/jacs.5b08535

Solar cells fabricated using alkyl ammonium metal halides as light absorbers have the right combination of high power conversion efficiency and ease of fabrication to realize inexpensive but efficient thin film solar cells. However, they degrade under prolonged exposure to sunlight. Herein, we show that this degradation is quasi-reversible, and that it can be greatly lessened by simple modifications of the solar cell operating conditions. We studied perovskite devices using electrochemical impedance spectroscopy (EIS) with methylammonium (MA)-, formamidinium (FA)-, and MA_xFA_1-x lead triiodide as active layers. From variable temperature EIS studies, we found that the diffusion coefficient using MA ions was greater than when using FA ions. Structural studies using powder X-ray diffraction (PXRD) show that for MAPbI₃ a structural change and lattice expansion occurs at device operating temperatures. On the basis of EIS and PXRD studies, we postulate that in MAPbI₃ the predominant mechanism of accelerated device degradation under sunlight involves thermally activated fast ion transport coupled with a lattice-expanding phase transition, both of which are facilitated by absorption of the infrared component of the solar spectrum. Using these findings, we show that the devices show greatly improved operation lifetimes and stability under white-light emitting diodes, or under a solar simulator with an infrared cutoff filter or with cooling.

Full text of DOI:10.1021/jacs.5b08535

Subscriber access provided by NEW YORK MED COLL
Article
Kinetics of Ion Transport in Perovskite Active
Layers and its Implications for Active Layer Stability
Monojit Bag, Lawrence A. Renna, Ramesh Adhikari, Supravat Karak, Feng Liu, Paul
M. Lahti, Thomas P. Russell, Mark T. Tuominen, and Dhandapani Venkataraman
J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.5b08535 • Publication Date (Web): 28 Sep 2015
Downloaded from http://pubs.acs.org on October 3, 2015
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 10
Journal of the American Chemical Society
1
2
3
4
5
6
Kinetics of Ion Transport in Perovskite Active Layers and its
Implications for Active Layer Stability
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
†
†
‡
†
⊥
⊥
§
Monojit Bag , Lawrence A. Renna , Ramesh Adhikari , Supravat Karak , Feng Liu , Paul M.
†
‡
†*
§∥
Lahti , Thomas P. Russell , Mark T. Tuominen , and D. Venkataraman
†
Department of Chemistry, University of Massachusetts Amherst, Amherst, Massachusetts 01003-9303, United
States.
‡
Department of Physics, University of Massachusetts Amherst, Amherst, Massachusetts 01003-9303, United States.
§
Materials Science Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720-1730, United
States.
∥Department of Polymer Science & Engineering, University of Massachusetts, Amherst, Massachusetts 01003-9303,
United States.
ABSTRACT: Solar cells fabricated using alkyl ammonium metal halides as light absorbers have the right combination of
high power conversion efficiency and ease of fabrication to realize inexpensive but efficient thin film solar cells. However,
they degrade under prolonged exposure to sunlight. Herein we show that this degradation is quasi-reversible, and that it
can be greatly lessened by simple modifications of the solar cell operating conditions. We studied perovskite devices us-
ing electrochemical impedance spectroscopy (EIS) with methyl ammonium (MA)–, formamidinium (FA)–, and MA FA
x
1-x
lead triiodide as active layers. From variable temperature EIS studies, we found that the diffusion coefficient using MA
ions was greater than when using FA ions. Structural studies using powder x-ray diffraction (PXRD) show that for MAPbI3
a structural change and lattice expansion occurs at device operating temperatures. Based on EIS and PXRD studies, we
postulate that in MAPbI the predominant mechanism of accelerated device degradation under sunlight involves thermal-
3
ly activated fast ion transport coupled with a lattice-expanding phase transition; both of which are facilitated by absorp-
tion of the infrared component of the solar spectrum. Using these findings, we show that the devices show greatly im-
proved operation lifetimes and stability under white-light emitting diodes, or under a solar simulator with an infrared
cut-off filter or with cooling.
INTRODUCTION
that facilitate undesired ion transport in the materials.
Toward clarifying the possible role of ion transport in
the degradation of perovskite solar cells, we studied
1
-10
Lead halide-based perovskite solar cells
drawn considerable interest because of their high effi-
ciency and low temperature solution processability.
But, a major practical limitation of these systems is their
poor stability under ambient conditions
have
1
1
them in operando, by electrochemical impedance spec-
25-27
troscopy (EIS)
combined with structural analysis of
1
2,13
active layers by powder x-ray diffraction (PXRD), using
varied illumination and temperature conditions.
and their
degradation upon prolonged exposure to heat and sun-
1
4
light. Previous reports have focused on degradation
pathways attributed to moisture and temperature ef-
EXPERIMENTAL SECTION
Methyl ammonium iodide (MAI) was synthesized
by the following procediure. Methylamine (24 mL, 33%)
was dissolved in EtOH (100mL) under Ar . HI (5 mL,
1
4-16
fects.
But thus far, the origin and nature of photo-
induced degradation is largely unexplored.
(g)
In inorganic-organic hybrid lead iodide perovskites
an organic counterion — typically alkyl ammonium —
occupies voids within a 3D framework created by vertex-
55%) was added drop-wise to stirred methylamine solu-
tion. After ca. 15 minutes, the solution was concentrated
in vacuo. The yellow solid was washed several times
with diethyl ether, recrystallized from diethyl
ether/EtOH to form a white solid and then dried in vac-
uo at 70 °C.
shared PbI octahedra. Migration of ions within the 3D
6
framework has been speculated to be an important fac-
tor leading to device instability and hysteresis in device
1
7
current-voltage (J-V) profiles. Some reports have ex-
plored ionic drift/transport in the perovskite active lay-
Formamidinium iodide (FAI) was synthesized by
the following procedure. Formamidine acetate (2.5g)
1
8-24
er.
But, there is no clear understanding of the factors
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 2 of 10
was dissolved in EtOH (100 mL) under Ar . HI (10.77 fractometer having a Ni filter, ½” diverging slit, vertical
(g)
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
mL, 57%) was added drop-wise to stirred formamidine
acetate solution. After ca. 2 h at ~50 °C, the solution was
concentrated in vacuo. The yellow solid was washed
several times with diethyl ether, recrystallized from di-
ethyl ether/EtOH to form a white solid and then dried
in vacuo at 70 °C.
goniometer, and X’Celerator detector. Measurements
were made from 2θ = 5°-60° under Cu K-Alpha (1.542 Å).
RESULTS AND DISCUSSION
J-V curves of planar heterostructure (Figure 1a inset)
perovskite solar cells shown in Figure 1a were measured
under inert atmosphere inside a glove box by scanning
Devices were fabricated as follows. ITO substrates
the voltage from the short circuit current (J ) to the
SC
-1
-
1
(
~20 Ω·sq ) were cleaned by ultrasonication of the sub-
open circuit voltage (VOC) at a scan rate of 250 mV·s .
strates submerged in soap solution, then acetone, and
then isopropyl alcohol followed by drying in a hot air
oven at 140 °C for 2 h. poly(3,4-ethylenedioxythiophene)
polystyrene sulfonate (PEDOT:PSS) (Clevios PVP AI
We obtained an average power conversion efficiency
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
(
PCE) of 10.3% over 32 MAPbI devices with a maximum
3
PCE of 13.6% (without using a photo-mask) under an
-
2
AM 1.5G solar simulator at 100 mW·cm light intensity.
The average over 32 devices and maximum efficiency
obtained using perovskite having both counterions
4
083) was filtered through a 0.45 micron polyvinylidene
fluoride (PVDF) filter (Wilkem Scientific), then spin
coated onto the substrates at 3000 rpm for 30 s The
PEDOT:PSS coated ITO substrates were then annealed
(
MA FA PbI ) were 11.6% and 13.3%, respectively. By
x 1-x 3
comparison, the maximum efficiency obtained from
FAPbI devices was 8.9%. Low PCE in FAPbI devices has
at 140 °C for 20 min in a hot air oven. PbI (40 wt. %)
2
3
3
28
was dissolved in dry N,N-dimethylformamide (DMF)
and filtered through a 0.45 micron polytetrafluoroeth-
ylene (PTFE) (Wilkem Scientific) filter. The hot solution
been attributed to their weak near-IR absorption. The
J-V characteristics, reproducibility, and device crystallin-
ity of these devices are given in Figure S1. For the main
(
~80 °C) was then spin coated at 6000 rpm onto a hot
discussion below, we focus on results from MAPbI and
3
substrate (~80 °C) for 60 s The PbI coated film was then
2
MA FA PbI devices because of their high PCEs and
x
1-x
3
kept on a hot plate (90 °C) to dry for 1 h. MAI, FAI and
mixtures of MAI and FAI (1:1 by wt.) were dissolved in
isopropyl alcohol (40 mg·mL ) and spin coated on the
reproducibility compared to FAPbI devices.
3
-1
Under prolonged illumination, the overall PCE of
the devices decreased over time (Figure 1b), with the
MAPbI devices degrading faster than MA FA PbI de-
PbI coated substrates at 6000 rpm for 60 s at room
2
3
x
1-x
3
temperature. MAPbI and MA FA PbI samples were
3
x
1-x
3
vices. In 20 h, MAPbI device PCEs dropped by 40%,
3
annealed on a hot plate (85 °C) for 1 h. FAPbI samples
3
whereas the MA FA PbI device PCEs dropped only by
x
1-x
3
were annealed at 160 °C for 10 min. The devices were
then kept in the dark for over 12 h. [6,6]-phenyl-C -
1
5% (see Figure S2). Surprisingly, when a previously il-
61
luminated MAPbI device was kept in the dark for ~15
3
butyric acid methyl ester (PCBM) in chlorobenzene (20
-1
min and then re-illuminated, much of the lost PCE was
recovered, indicating that the device degradation was
quasi-reversible (Figure 1b). After being kept in the dark
under inert atmosphere inside a glove box for 70 h, the
mg mL ) as an electron-transporting layer was then spin
coated atop the devices at 1000 rpm for 60 s inside a
glove box. Next, a 15 nm thick Ca electrode was thermal-
-
1
ly deposited at a rate of 0.5 Å·s followed by 100 nm of
-
1
PCE of MAPbI
more impressively, from a similar dark-storage test for
0 days, MA FA PbI device shows no appreciable loss
3
devices dropped only marginally. Much
Al electrode deposited at a rate of 1 – 3 Å·s , all at a
-
6
chamber pressure of 1×10 mbar.
3
x
1-x
3
Current-voltage (J-V) curves of the perovskite solar
in PCE (Figure S4). MAPbI devices did show degrada-
3
cells were tested under an AM 1.5G solar simulator at
tion (see Figure S4), which can be attributed to the ex-
posure to light and applied bias when making J-V meas-
-2
1
00 mV·cm light intensity inside a N filled glove box,
2
at time intervals over 70 h. Impedance was measured
using an Agilent 4294A Precision Impedance Analyzer
at different bias voltages after each efficiency measure-
ment, using a frequency analyzer under AM 1.5G simu-
lated illumination, as well as under dark condition.
Temperature dependent study was conducted using a
thermoelectric heater/cooler with a thermocouple at-
tached to measure in-situ temperature. J-V measure-
ments were made without a photomask to compare with
EIS experiments were a photomask cannot be used. De-
vice stability experiments were performed using a cool
urements. Compared to MAPbI devices, we also ob-
3
served a reduced J-V hysteresis in MA FA PbI devices
x
1-x
3
(
Figure S5). Based on these data, we conclude that (a)
the device degradation is accelerated by the AM1.5G
illumination of the device and not by trace oxygen or
moisture inside the glove box (Table 1), and (b) the
mixed counterion MA FA PbI devices exhibit greatly
x
1-x
3
improved stability under dark and under illumination
conditions, compared to the corresponding single-
counterion devices.
To understand the origin of photo-induced, yet
quasi-reversible degradation in the perovskite films, we
carried out EIS measurements on devices under dark
conditions as well as under AM1.5G illumination in a
nitrogen-filled glove box. In these EIS measurements,
the device impedance was measured by applying 20 mV
white-LED array (12V, 10 W) light source and an AM
-2
1
.5G solar simulator at 100 mWcm light intensity and
the same solar simulator with an infrared cut off, New-
port-Oriel KG5 filter.
PXRD measurements were made using samples on
3
cleaned glass slides with a PANalytic X’Pert X-Ray dif-
ACS Paragon Plus Environment

Page 3 of 10
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1. (a) MAPbI
3
perovskite solar cell light J-V characteristics for best cell without photomask. Inset: Device architecture. (b) Device
efficiency as a function of light exposure time. EIS was carried out at each points represented as red circle. Black circle represents the
efficiency just after the device was kept in dark for 15 min during dark impedance measurement.
Table 1. MAPbI perovskite solar cell metrics
3
Device
JSC
VOC
FF
η
Rs
Rsh
-2
2
2
Condition
(mA·cm )
(V)
(%)
(%)
(Ω·cm )
(kΩ·cm )
Sample 1, as prepared
After 70 h in light
After 15 min in dark
Sample 2, as prepared
After 5 days
18.85
8.66
12.67
17.57
17.55
0.865
0.929
0.851
0.983
0.970
70.2
58.1
62.8
66.4
61.5
11.45
4.68
6.78
5.8
11.8
9.2
0.857
0.328
0.330
0.896
0.686
11.47
10.48
7.5
10.6
AC voltage whose frequency (ω) is swept from 100 Hz to
MHz in 100 equal steps on a logarithmic scale. EIS re-
attributed predominantly to impedance arising from the
electronic transport. The second semicircle is associated
with the low frequency spectrum and is attributed to the
impedance arising either from charge or mass transfer at
1
sults are conventionally presented as a Nyquist plot,
with the real component of the impedance (Z’) as the
abscissa and the imaginary part (Z’’) as the ordinate.
29
30
the interface or Warburg ion diffusion. Additional da-
Nyquist plots of a MAPbI sample under dark conditions
ta, including light intensity dependence and dark meas-
urements, are shown in Figure S6. As a control experi-
ment, we measured the EIS of a bulk heterojunction
organic solar cell having no ions in the active layer, a
cell fabricated from poly(3-hexylthiophene) and PCBM.
The EIS of this system — as expected — shows only one
semicircle in the high frequency regime, under both
dark and illuminated conditions (Figure S7c & d). Based
on these results, and literature precedent describing the
impedance spectroscopy of systems with two types of
3
(
Figure 2a) versus AM 1.5G solar simulator conditions at
-2
1
00 mWcm light intensity (Figure 2b) show different
characteristics. The plots under dark conditions showed
a single charge transport regime at low forward bias
voltages (Vapp) with an estimated transit time (τ ) of ~90
29
d
μs (Figure 2a). When Vapp was increased, a single recom-
bination semicircle was observed, indicating only one
type of charge transport (electronic) is predominant un-
der dark conditions. The Nyquist plot from a device un-
der illumination shows two distinct charge transport
regimes, with two semicircles associated with different
time constants (τ). The semicircle closest to the origin is
associated with the higher frequency spectrum and is
3
1-33
charge carriers,
we assert that the low frequency EIS
component arising only in the illuminated perovskite is
predominately associated with the diffusion of ions.
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 10
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 2. (a) Nyquist plot of MAPbI
3
samples under dark. Data was taken from 100 Hz to 1 MHz oscillation frequency at 20 mV AC
3
amplitude and 0 V DC bias. Inset: High frequency component showing transport regime. (b) EIS plot of MAPbI sample at 45 °C at 100
-
2
mW·cm light intensity and 0 V applied bias. (c) Warburg impedance plot of same data. (d) Equivalent circuit diagram of perovskite
solar cells showing combined charge and ion transport impedance. (e) Typical Nyquist plot for mixed conductor system with double-
layer capacitance showing semi-circular feature in low frequency regime. (f) Typical Nyquist plot for mixed conductor system with
Warburg diffusion as evidence by the linear portion of the low frequency regime.
Prior EIS studies of perovskite-based solar cells at-
tribute the low frequency component to double layer
capacitance due to charge accumulation at the interface,
amination of the published EIS spectra from which
charge accumulation effects were concluded, shows the
presence of a linear component in low frequency
1
8,34-40
41
not to ion migration.
To account for charge accu-
Nyquist plot region. This feature is just what we ob-
mulation at the interface in these studies, ion migration
to the interface can be invoked. But, this low frequency
component can also arise from ion diffusion or Warburg
diffusion. The tell-tale signature of ion/Warburg diffu-
sion is the presence of a linear region as in Figure 2f in
the Nyquist plot instead of just a semicircle as in Figure
tained in our in operando EIS measurements.
To support our ion diffusion degradation mecha-
nism hypothesis further, we modelled the ionic diffusion
in the low frequency regime as a Warburg element
4
2
(W ), whose impedance is represented in equation (1):
S
2e that indicates double-layer capacitance. A close ex-
ACS Paragon Plus Environment

Page 5 of 10
Journal of the American Chemical Society
anion may be the mobile ion other studies indicate that
P
tanh
(
j
ω
T
)
P
W
22-24
1
alkyl ammonium counterion
is mobile. Recent first
W = A
(1)
S
W
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
(
jω
TW
)
principles, state-of-the-art computational analysis indi-
cated that the time scale associated with iodide ion mo-
bility is < 1μs, but for methyl ammonium ions to be tens
where AW is the Warburg impedance, T is the Warburg
time constant and exponent P is ~0.5.
planation of the model derivation is given in the Sup-
porting Information. Plots of |Z’| and |Z”| as functions
of ω show linear behavior in the low frequency re-
W
2,43
23
3
of ms. Given the fast time scale predicted for iodide
A detailed ex-
transport, evidence of iodide transport in EIS would be
hidden under the first semicircle, which is dominated by
electronic processes. From our EIS measurements, the
time constant associated with the Nyquist semicircle
that only appears during illumination is on the order of
tens of ms. Moreover, we find that diffusion coefficient
-
1/2
4
1
gime, with the slope of both lines being equal to AW
Figure 2c). The imaginary component line intersects
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
(
the ordinate at zero. This observation is unequivocal
proof that this sort of low frequency component arises
from Warburg ion diffusion.
obtained in MAPbI devices ions is greater than that in
3
4
2
FAPbI devices. Both these observations are consistent
3
with the computational studies reported by De Angelis
To evaluate the kinetics of the photo-induced ion
transport, the Nyquist plots were modelled as mixed
conductor systems based on the equivalent circuit mod-
el in Figure 2d, which contains both high frequency
electronic elements and low frequency ionic
23
and co-workers, with all of these results supporting our
hypothesis that the mobile ion in illuminated perovskite
solar cells of this type is the alkyl ammonium counteri-
on.
3
0,31
The activation energy (E ) for ionic diffusion can be
a
elements.
In the model circuit, the high frequency
calculated by its relationship to T shown in equation
w
component consists of charge and ion transport re-
sistance (R ) coupled with interfacial charge transfer
resistance (R ). Any ion accumulation at the interface is
modelled as an interfacial Debye-layer capacitance (C ),
(
3):
tr
2
CT
1
kT a
α

−Ea


=
exp
(3)
dl
2
TW
h L
D
 kT 
while chemical capacitance (C ) represents the stored
μ
charge in the bulk perovskite layer. Free carrier recom-
bination/transport (electrons and holes both) is mod-
elled using another resistance term (Relectr) in parallel
where a is the lattice parameter for the perovskite, α
alpha) is the coordination factor, h is Planck’s constant
(
3
2
and E is the activation energy . By measuring imped-
with the ion diffusion term. Series resistance (R ) repre-
a
s
ance spectra at various temperatures (Figure S8) and
sents contact resistance arising from the electron and
hole transport layers, as well as contributions from the
measurement apparatus itself. The details of the model-
ling are provided in the supporting information.
-1
plotting ln[1/(T·T )] vs. T as shown in Figure 3, we cal-
W
-1
culated for MAPbI that E = ~56 kJ·mol (~0.58 eV) in
3
a
the low temperature regime (<320 K). Interestingly, De
23
Angelis and co-workers calculated E = ~0.5 eV and
a
Using ZView version 3.4c (Scribner Associates Inc.),
we fit the impedance spectra in Figure 2b and all other
Nyquist plots to the circuit model shown in Figure 2d,
and obtained R , R , C , C , R
From the value of T , we calculated the effective chemi-
24
Tateyama and co-workers calculated E = 0.57 eV for
a
+
MA ion migration in a tetragonal MAPbI supercell. In
3
the high temperature regime (>320 K), we calculate that
E drops significantly to ~22 kJ·mol . We speculate this
a
, R , A , T , and P.
s w w
tr
CT
dl
μ
electr
-1
W
discontinuity in Arrhenius-like behavior of ionic
cal diffusion coefficient (D) using the equation (2):
transport in MAPbI arises from a discontinuous volume
3
2
44
L
change induced by a phase transition, as is discussed
D
D =
(2)
-1
below. For MA FA PbI the E = ~61 kJ·mol was ob-
x
1-x
3
a
TW
tained at all temperatures. MA FA PbI exhibits contin-
x
1-x
3
uous Arrhenius-like ionic transport because a phase
where LD is the effective ion diffusion length. In our
study, a maximum diffusion length was assumed to be
equal to the perovskite film thickness (~300 nm) meas-
ured by profilometry. At the 45 °C (318 K) operating
temperature of the device, under AM1.5G illumination,
D was calculated to be ~3.6×10 cm ·s for MAPbI ,
~2×10 cm ·s for MA FA PbI and ~3×10 cm ·s for
FAPbI The ionic diffusion coefficient follows the order
MA>MA FA >>FA, decreasing with increasing size of
transition does not occur as in MAPbI , again as shown
3
-1
below. The E = ~21 kJ·mol for FAPbI samples at oper-
a
3
-1
ating temperatures, rising to ~76 kJ·mol at higher tem-
perature. These combined results show that under de-
vice operating conditions, the activation energy for ion
diffusion is higher for MA FA , consistent with our ob-
-
12
2
-1
3
x
1-x
-
12
2
-1
-13
2
-1
x
1-x
3,
servation that devices with mixed ions have a higher
3
.
stability than those with MA ions alone. FAPbI devices
3
x
1-x
also shows improved stability (see Figure S4) compared
the ions and being about an order of magnitude higher
for the increased stability MA and mixed counterion
materials. While some recently reported studies on
to compared to MAPbI devices despite a low E ; howev-
3
a
er the D is also very low, which can explain the en-
hanced stability.
20,21
MAPbI perovskite solar cells conclude that iodide
3
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 6 of 10
diffraction (PXRD) under different light and tempera-
ture conditions, to gauge the effects of visible versus IR
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
illumination components. The PXRD of MAPbI device
3
after solar simulator illumination for 15 min showed an
increase in lattice dimensions (Figure S13). No such
structural lattice change was observed when the sample
was exposed to white-LED light alone at a measured
operating temperature of 28 °C (301 K). A similar lattice
expansion to that induced under AM1.5G solar simulator
illumination, was also observed for illumination with an
IR-only source (150 W commercial IR lamp), as shown in
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 4a. The tetragonal (I4cm) MAPbI perovskite (211)
3
4
4
PXRD peak at 2θ = 23.48° actually disappears as the
temperature rises above 45 °C (see Figure S14), while
peaks showing a transition to cubic perovskite (Pm3m)
are observed. However, this transition reverses when the
IR lamp is switched off. PXRD of both MA FA PbI and
x
1-x
3
FAPbI show the PXRD peaks of a trigonal (P3m1) struc-
3
4
5
-1
ture at 45 °C, consistent with the literature. These
show no phase transitions, but they do show an increase
in lattice parameters with IR illumination, similar to
MAPbI3.
Figure 3. Arrhenius-like plot of 1/(T·T
FAPbI and MA
W 3
) vs T for MAPbI ,
3
x 3
FA1-xPbI .
The temperature dependent ionic conductivity
σion) can also be independently obtained from the War-
(
To extract lattice parameters and unit cell volumes
burg impedance (A ). A plot of ln(σ ) vs. 1/T and its
W
ion
before and after heating, the PXRD data was refined
derivation are shown in Figure S9. Despite varying tem-
peratures from 301K to 353K, Nyquist plots of devices
under dark conditions showed only high frequency
charge transport regime behavior (Figure S8d), con-
sistent with electronic-only transport. No measurable
dark performance ion transport regime was observed,
unlike for devices under illumination (Figure S8a-c),
indicating that the ion transport is indeed induced by
light.
46,47
using the Rietveld method using known structures
as
starting points. Upon heating with the IR lamp (~45 °C,
18 K), the volume change observed in MAPbI was 9.1
3
3
3
Å per unit cell, a 3.66% relative expansion. These obser-
vations indicate reversible perovskite lattice expansion
under IR heating that correlates with the quasi-
reversible thermal/IR PCE cycling that we observed in
these solar cells (Figure 1b). The volume increase that
accompanies the tetragonal to cubic structural phase
In order to identify the regions in the AM 1.5G solar
simulator spectrum that induce the destabilizing ion
transport, we first measured EIS for devices operating
under a commercial white-LED that does not emit IR or
UV components (Figure S10). The EIS Nyquist plots for
change in MAPbI , has been reported to occur discon-
3
44
tinuously, which also correlates with the ion transport
-1
activation energy barrier discontinuity from 56 kJ·mol
-1
to 22 kJ·mol that we find at ~320 K. The PXRD analyses
3
3
for MA FA PbI and FAPbI indicate 4.6 Å and 2.2 Å
x
1-x
3
3
MAPbI devices comparing white-LED versus AM1.5G
3
per unit cell volume increases upon heating, which are
relative expansions of 0.63% and 0.23%, respectively
-
2
illuminated performance at 79 mW·cm light intensity,
show significant decrease in the low frequency ion-
transport regime under white-LED illumination, indicat-
ing slower ion transport compared to the device illumi-
nated by the solar simulator (Figure S11). The ion diffu-
(
Figure S15). MA FA PbI does not undergo a structural
x 1-x 3
phase transition, and its thermal expansion is quite low
compared to that of MAPbI : thus, ionic diffusion in this
3
mixed system follows continuous Arrhenius behavior.
-
12
2
-1
sion coefficient under white-LED was 1.4×10 cm ·s ,
There was no observed phase change for FAPbI , nor is
3
-
12
2
-1
compared to 3×10 cm ·s under AM1.5G solar simula-
tion. This difference in the diffusion coefficient is con-
sistent with differences in the operating temperatures of
the devices, 301 K under LED and 320 K under AM1.5G
solar simulator. It further indicates that the ion
transport is induced and accelerated by visible light: the
IR illumination component seems to contribute only to
thermal activation of the visible-light induced ion
transport.
there any literature evidence of structural changes at
high temperatures, so the increase in E observed for
a
FAPbI samples requires further investigation
3
The EIS and PXRD results correlated with device
performance in this study provide a comprehensive and
revised picture of the origin of the photo-induced deg-
radation of perovskites. The EIS studies show that that
ion diffusion is associated with the degradation, and
that the diffusion is light induced and thermally activat-
ed, and that the alkyl ammonium-type counterions are
the diffusion ions. PXRD proves that the perovskite lat-
tices expand on heating, but that this process is reversi-
ble. There is a strong correlation between the ionic dif-
fusion coefficient and the activation energy barrier for
Grazing incidence wide-angle X-ray scattering
(
GIWAXS) indicates that all the perovskite films in this
study are polycrystalline (Figure S12). We therefore ex-
amined perovskite samples
glass/PEDOT:PSS/perovskite/PCBM) by powder X-ray
(
ACS Paragon Plus Environment

Page 7 of 10
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4. (a) PXRD of MAPbI
3
sample at room temperature (28 °C) and after heating with infrared lamp (~45 °C) in dark. The process is
reversible once the device gets cooled. (b) PXRD pattern of perovskite sample after prolonged heating in presence of visible light,
showing degradation of perovskite to PbI structure.
2
Figure 5. (a) Mixed counterions perovskite solar cell stability after 17 h continually under dark at 45 °C, white-LED light at 45 °C, and
-
2
AM1.5G solar simulator at 100 mW·cm light intensity. J-V curves averaged over 4 devices before and after (b) heating in dark at 45 °C,
c) heating at 45 °C under white-LED, and (d) AM1.5G solar simulator.
(
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 8 of 10
ion transport with device stability. In MAPbI lower ac- lated with alkyl ammonium transport. This may be due
3
,
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
tivation energy barrier coupled with a higher diffusion
coefficient of ions leads to poor device stability under
to accumulation of vacancies, which may destabilize the
lattice, and causes collapse to PbI2.
light and dark. Moreover, the MAPbI activation energy
3
decreases at high temperature, further promoting ion
transport. By contrast, in mixed MA FA PbI devices, a
CONCLUSION
x
1-x
3
If perovskite-based photovoltaics are truly the fu-
ture of cheap and efficient solar energy, they cannot
degrade under sunlight. It is therefore imperative to
develop a fundamental understanding of the light in-
duced degradation pathways in perovskite active layers
so that this problem can be solved. In this communica-
tion we have clarified the degradation mechanism of
alkyl ammonium lead triiodide-based perovskite solar
cells under continuous photo-illumination. We find that
the degradation of perovskite solar cells under sunlight
requires both an IR and a visible component of the solar
spectrum. Further, the quasi-reversibility of the photo-
induced degradation phenomenon is observed for the
first time in these systems. EIS measurements provide
evidence of fast ion transport of the ammonium coun-
terions in the perovskite thin film, which for MAPbI3
can be facilitated by lattice expansion accompanied with
a phase transition under simulated solar spectrum con-
taining infrared component. Thus, the photo-stability
can be improved by reducing the infrared component
incident on the perovskite solar cells. Mixing a larger
counterion (FA) with MA provides the right balance
between device performance and stability by circum-
venting the phase change and increasing the activation
energy barrier for ion transport.
higher activation energy barrier coupled with the lower
diffusion coefficient of ions leads to a better stability
under light and dark. This last finding has important
practical consequences in the search for more stable
perovskite solar cell formulations.
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
Figure 6. MAPbI perovskite device stability (normalized PCE)
-2
under AM1.5G solar simulator at 79 mW·cm and under
-2
AM1.5G solar simulator at 100 mW·cm light intensity with a
ASSOCIATED CONTENT
Supporting Information.
Additional details on device performance and stability,
AFM, EIS measurements and modeling, XRD, and GIWAXS
can be found in the Supporting Information. This material
is available free of charge via the Internet at
http://pubs.acs.org.
KG5 IR cut off filter, AM1.5G solar simulator with a fan cooler,
-
2
and white-LED at 20 mW·cm . 3% error is accounted for J-V
measurement hysteresis and lamp power fluctuation.
Since the ion diffusion is light induced but thermal-
ly activated, we hypothesized that the device stability
would be further increased if the devices were operated
under conditions that do not increase the device tem-
perature. Indeed, with MA FA PbI , the PCE under
AUTHOR INFORMATION
x
1-x
3
white-LED light (sans IR component) illumination was
stable over the full 20 h duration of an experiment, as
shown in Figure S2. The device stability was also vastly
improved when a KG5 filter was used to block the near-
IR component of the AM1.5G illuminator. Finally, the
device was also stable under standard AM 1.5G illumina-
tion when the device was air-cooled with a fan (Figure
Corresponding Author
* dv@umass.edu
Author Contributions
⊥These authors contributed equally.
ACKNOWLEDGMENT
6
). Devices kept at 45 °C for 17 h under dark showed a
% decrease in PCE when re-illuminated with an AM
.5G illuminator. After 17 h at 45 °C, PCE of devices
showed a 19% decrease under white-LED light and a
3% decrease with AM1.5G solar simulator (Figure 5a).
This work was supported as part of Polymer-Based Materi-
als for Harvesting Solar Energy (PHaSE), an Energy Frontier
Research Center funded by the U.S. Department of Energy,
Office of Science, Basic Energy Sciences under Award # DE-
SC0001087. We would like to thank Dr. Akshay Kokil
5
1
2
All the device J-V characteristics were averaged over 4
samples and are shown in Figure 5b-d. These results
conclusively show that a combination of heat and light
facilitates rapid ion diffusion and device degradation. The
enhanced stability under white LED along with the de-
crease in ion diffusion coefficient (vide supra) indicates
that perovskite active layer degradation is indeed corre-
(
UMass Lowell) for assistance with device preparation.
REFERENCES
(1) Burschka, J.; Pellet, N.; Moon, S.-J.; Humphry-Baker, R.;
Gao, P.; Nazeeruddin, M. K.; Grätzel, M. Nature 2013, 499, 316.
ACS Paragon Plus Environment

Page 9 of 10
Journal of the American Chemical Society
(
2) Etgar, L.; Gao, P.; Xue, Z.; Peng, Q.; Chandiran, A. K.; Liu,
(27) Li, M.; Pietrowski, M. J.; De Souza, R. A.; Zhang, H.;
Reaney, I. M.; Cook, S. N.; Kilner, J. A.; Sinclair, D. C. Nat.
Mater. 2014, 13, 31.
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
B.; Nazeeruddin, M. K.; Grätzel, M. J. Am. Chem. Soc. 2012, 134,
17396.
(
8
(
3) Green, M. A.; Ho-Baillie, A.; Snaith, H. J. Nat. Photon. 2014,
, 506.
4) Heo, J. H.; Im, S. H.; Noh, J. H.; Mandal, T. N.; Lim, C.-S.;
(28) Hu, M.; Liu, L.; Mei, A.; Yang, Y.; Liu, T.; Han, H. J. Mater.
Chem. A 2014, 2, 17115.
(29) Bisquert, J.; Mora-Sero, I.; Fabregat-Santiago, F.
ChemElectroChem 2014, 1, 289.
(30) Jamnik, J.; Maier, J. J. Electrochem. Soc. 1999, 146, 4183.
(31) Lai, W.; Haile, S. M. J. Am. Ceram. Soc. 2005, 88, 2979.
(32) Jasinski, P.; Petrovsky, V.; Suzuki, T.; Anderson, H. U. J.
Electrochem. Soc. 2005, 152, J27.
(33) Javier, A. E.; Patel, S. N.; Hallinan, D. T.; Srinivasan, V.;
Balsara, N. P. Angew. Chem. Int. Edit. 2011, 50, 9848.
(34) Pascoe, A. R.; Duffy, N. W.; Scully, A. D.; Huang, F.;
Cheng, Y.-B. J. Phys. Chem. C 2015, 119, 4444.
(35) Juarez-Perez, E. J.; Wuβler, M.; Fabregat-Santiago, F.;
Lakus-Wollny, K.; Mankel, E.; Mayer, T.; Jaegermann, W.;
Mora-Sero, I. J. Phys. Chem. Lett. 2014, 5, 680.
(36) Pockett, A.; Eperon, G. E.; Peltola, T.; Snaith, H. J.; Walker,
A.; Peter, L. M.; Cameron, P. J. J. Phys. Chem. C 2015, 119, 3456.
(37) Shao, Y.; Xiao, Z.; Bi, C.; Yuan, Y.; Huang, J. Nat. Commun.
2014, 5, 5784.
(38) Almora, O.; Zarazua, I.; Mas-Marza, E.; Mora-Sero, I.;
Bisquert, J.; Garcia-Belmonte, G. J. Phys. Chem. Lett. 2015, 6,
1645.
(39) Kim, H. S.; Mora-Sero, I.; Gonzalez-Pedro, V.; Fabregat-
Santiago, F.; Juarez-Perez, E. J.; Park, N. G.; Bisquert, J. Nat.
Commun. 2013, 4, 2242.
(40) Christians, J. A.; Fung, R. C.; Kamat, P. V. J. Am. Chem.
Soc. 2014, 136, 758.
(41) For example in Figures 3(b) and 3(c) of Ref. 34, we can see
Warburg-like linear component in the low frequency regime.
However the equivalent circuit shown in Figure 3(a) of this
reference does not include Warburg component.
(42) Ho, C.; Raistrick, I. D.; Huggins, R. A. J. Electrochem. Soc.
1980, 127, 343.
(43) Sugimoto, W.; Iwata, H.; Yokoshima, K.; Murakami, Y.;
Takasu, Y. J. Phys. Chem. B 2005, 109, 7330.
(44) Baikie, T.; Fang, Y.; Kadro, J. M.; Schreyer, M.; Wei, F.;
Mhaisalkar, S. G.; Grätzel, M.; White, T. J. J. Mater. Chem. A
2013, 1, 5628.
(45) Binek, A.; Hanusch, F. C.; Docampo, P.; Bein, T. J. Phys.
Chem. Lett. 2015, 6, 1249.
(46) Stoumpos, C. C.; Malliakas, C. D.; Kanatzidis, M. G. Inorg.
Chem. 2013, 52, 9019.
(47) Mashiyama, H.; Kurihara, Y.; Azetsu, T. J. Korean Phys.
Soc. 1998, 32, S156.
Chang, J. A.; Lee, Y. H.; Kim, H.-j.; Sarkar, A.; Nazeeruddin, M.
K.; Grätzel, M.; Seok, S. I. Nat. Photon. 2013, 7, 486.
(5) Jeon, N. J.; Noh, J. H.; Kim, Y. C.; Yang, W. S.; Ryu, S.; Seok,
S. I. Nat. Mater. 2014, 13, 897.
(6) Lee, M. M.; Teuscher, J. l.; Miyasaka, T.; Murakami, T. N.;
Snaith, H. J. Science 2012, 338, 643.
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
(
(
(
(
7) Liu, M.; Johnston, M. B.; Snaith, H. J. Nature 2013, 501, 395.
8) McGehee, M. D. Nat. Mater. 2014, 13, 845.
9) Snaith, H. J. J. Phys. Chem. Lett. 2013, 4, 3623.
10) Zhou, H.; Chen, Q.; Li, G.; Luo, S.; Song, T.-b.; Duan, H.-S.;
Hong, Z.; You, J.; Liu, Y.; Yang, Y. Science 2014, 345, 542.
11) You, J.; Hong, Z.; Yang, Y.; Chen, Q.; Cai, M.; Song, T.-B.;
(
Chen, C.-C.; Lu, S.; Liu, Y.; Zhou, H. ACS Nano 2014, 8, 1674.
(12) Grätzel, M. Nat. Mater. 2014, 13, 838.
(13) Niu, G.; Guo, X.; Wang, L. J. Mater. Chem. A 2015, 3, 8970.
(
14) Misra, R. K.; Aharon, S.; Li, B.; Mogilyansky, D.; Visoly-
Fisher, I.; Etgar, L.; Katz, E. A. J. Phys. Chem. Lett. 2015, 6, 326.
15) Christians, J. A.; Miranda Herrera, P. A.; Kamat, P. V. J.
Am. Chem. Soc. 2015, 137, 1530.
16) Yang, J.; Siempelkamp, B. D.; Liu, D.; Kelly, T. L. ACS Nano
015, 9, 1955.
17) Unger, E. L.; Hoke, E. T.; Bailie, C. D.; Nguyen, W. H.;
(
(
2
(
Bowring, A. R.; Heumuller, T.; Christoforo, M. G.; McGehee, M.
D. Energy Environ. Sci. 2014, 7, 3690.
(
18) Dualeh, A.; Moehl, T.; Tétreault, N.; Teuscher, J.; Gao, P.;
Nazeeruddin, M. K.; Grätzel, M. ACS Nano 2014, 8, 362.
19) Tress, W.; Marinova, N.; Moehl, T.; Zakeeruddin, S. M.;
Nazeeruddin, M. K.; Grätzel, M. Energy Environ. Sci. 2015, 8,
95.
20) Eames, C.; Frost, J. M.; Barnes, P. R. F.; O'Regan, B. C.;
(
9
(
Walsh, A.; Islam, M. S. Nat. Commun. 2015, 6, 7497.
(21) Yang, T.-Y. Y.; Gregori, G.; Pellet, N.; Grätzel, M.; Maier, J.
Angew. Chem. Int. Edit. 2015, 54, 7905.
(
22) Xiao, Z.; Yuan, Y.; Shao, Y.; Wang, Q.; Dong, Q.; Bi, C.;
Sharma, P.; Gruverman, A.; Huang, J. Nat. Mater. 2015, 14, 193.
23) Azpiroz, J., M.; Mosconi, E.; Bisquert, J.; De Angelis, F.
Energy Environ. Sci. 2015, 8, 2118.
24) Haruyama, J.; Sodeyama, K.; Han, L.; Tateyama, Y. J. Am.
Chem. Soc. 2015, 137, 10048.
25) Freger, V.; Bason, S. J. Membrane Sci. 2007, 302, 1.
(
(
(
(26) Zhuang, Q.-C.; Wei, T.; Du, L.-L.; Cui, Y.-L.; Fang, L.; Sun,
S.-G. J. Phys. Chem. C 2010, 114, 8614.
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 10 of 10
Table of Contents artwork
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
ACS Paragon Plus Environment
10