www.advenergymat.

de
www.MaterialsViews.com

FULL PAPER
Oxygen Degradation in Mesoporous Al2O3/CH3NH3PbI3−xClx
Perovskite Solar Cells: Kinetics and Mechanisms
Andrew J. Pearson, Giles E. Eperon, Paul E. Hopkinson, Severin N. Habisreutinger,
Jacob Tse-Wei Wang, Henry J. Snaith,* and Neil C. Greenham*

unexpectedly clean electronic properties of

The rapid pace of development for hybrid perovskite photovoltaics has these semiconductors have led to the dem-
recently resulted in promising figures of merit being obtained with regard onstration of proof-of-concept PV devices
with an initial efficiency that outperforms
to device stability. Rather than relying upon expensive barrier materials,
competing ‘next-generation’ technolo-
realizing market-competitive lifetimes is likely to require the development gies such as dye-sensitized, organic, and
of intrinsically stable devices, and to this end accelerated aging tests can quantum dot solar cells, and even matches
help identify degradation mechanisms that arise over the long term. Here, that of commercially deployed PV.[6] Other
oxygen-induced degradation of archetypal perovskite solar cells under attractive characteristics of hybrid per-
ovskites include the low embedded costs
operation is observed, even in dry conditions. With prolonged aging, this
of materials processing and the ability to
process ultimately drives decomposition of the perovskite. It is deduced that tune the properties of the semiconductor
this is related to charge build-up in the perovskite layer, and it is shown that through changes in chemical composi-
by efficiently extracting charge this degradation can be mitigated. The results tion.[1,7–10] All of these factors have led to
confirm the importance of high charge-extraction efficiency in maximizing the hybrid perovskites being considered as
tolerance of perovskite solar cells to oxygen. disruptive materials in a wide range of
technology applications;[11–16] interest from
the materials science research community
is both substantial and rapidly maturing.
1. Introduction Realizing the technology potential of hybrid perovskites
necessitates a thorough understanding of the degradation
Solution-processed organic–inorganic (hybrid) halide per- mechanisms that limit the lifetime of devices, and identifying
ovskites have emerged as a promising family of semicon- solutions to mitigate these. This exercise is nontrivial because
ducting materials for photovoltaic (PV) applications.[1–5] The of the various factors that can affect device performance: light,
heat, moisture, oxygen, mechanical and electrical stresses, and
Dr. A. J. Pearson, Prof. N. C. Greenham combinations thereof.[17,18] Currently, perovskite solar cells
Cavendish Laboratory (PSCs) based on the “triple layer” architecture with a thick
J. J. Thomson Avenue, Cambridge CB3 0HE, UK carbon back electrode have demonstrated perhaps the most
E-mail: [email protected]
stable performance characteristics of all PSCs.[19] These devices,
Dr. G. E. Eperon,[+] S. N. Habisreutinger, J. T.-W. Wang,
based on the mixed cation perovskite (5-AVA)X(MA)1−XPbI3
Prof. H. J. Snaith
Clarendon Laboratory (where 5-AVA corresponds to 5-ammoniumvaleric acid), have
University of Oxford been shown to maintain their initial power conversion efficiency
Parks Road, Oxford OX1 3PU, UK (PCE) of approximately 10% over 1000 h continuous simulated
E-mail: [email protected] solar illumination,[20] dark storage for three months at elevated
Dr. P. E. Hopkinson temperatures and humidity (85 °C/85%) and a minimum of 7
Centre for Advanced Materials
Universität Heidelberg d operation in real-world conditions (Jeddah, Saudi Arabia).[21]
Heidelberg 69120, Germany Although clearly commendable, it is apparent that significant
Dr. P. E. Hopkinson work is required to realize PSCs with high stability and high
Kirchhoff-Institut für Physik performance, where promising stability metrics are combined
Universität Heidelberg with the certified PCE values for champion devices.[22]
Heidelberg 69120, Germany
To elucidate the underlying mechanisms for degradation in
[+]Present Address: Department of Chemistry, University of Washington
hybrid perovskites and guide the development of intrinsically
Seattle, Washington 98195, United States.
stable devices, considerable work has been undertaken on the
The copyright line was changed 8 June 2016 after initial publication. archetypal semiconductor of the research field: CH3NH3PbI3.
This is an open access article under the terms of the Creative Commons It is known that this material (and other single or mixed-halide
Attribution License, which permits use, distribution and reproduction in derivatives) are particularly sensitive to moisture,[1,7,17,23–26] where
any medium, provided the original work is properly cited.
the proposed chemical reactions involving water result in decom-
DOI: 10.1002/aenm.201600014 position of the perovskite into its precursor components via

Adv. Energy Mater. 2016, 6, 1600014 © 2016 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com (1 of 10) 1600014
 16146840, 2016, 13, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.201600014 by Universitatsbibliothek, Wiley Online Library on [30/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advenergymat.de
www.MaterialsViews.com

intermediate and semi-reversible monohydrate phases.[17,26] The

FULL PAPER

the factors affecting the stability of this promising technology.

rate of decomposition is dependent on the preparation method of Alongside standard IV measurements, transient photocurrent/
the perovskite sample and is accelerated by oxygen and heat.[27– photovoltage (TPC/TPV) measurements were conducted in situ,
29]
However, the presence of some moisture during device pro- i.e., in sequence with the IV scans, to gain partial access to the
cessing has been shown to be somewhat beneficial for material dynamics of photocurrent generation during accelerated aging.
quality and solar cell performance, owing to a reduced trap den- Combined with optical microscopy and XRD measurements,
sity in films. However, the stability of devices prepared in the pres- a relatively detailed insight into the evolution of the solar cells
ence of moisture is still an open question owing to the absence of during our controlled aging tests is obtained. These comple-
promising lifetime data.[30,31] Hybrid perovskite films containing mentary experiments provide sufficient information to iden-
aprotic organic cations have been suggested[32] and shown[33] to tify the underlying causes of degradation in our model system
exhibit improved intrinsic moisture tolerance, the latter inferred and provide a framework for engineering PSCs with enhanced
through powder X-ray diffraction (XRD) studies. Nevertheless, the oxygen tolerance, an important contribution to the realization
moisture stability of CH3NH3PbI3 can be improved at the device of PSCs with market-competitive lifetimes.
level by employing various engineering strategies during fabrica-
tion of the PSC. Substitution of the doped spiro-MeOTAD hole
transport layer for undoped tetrathiafulvalene derivatives[34] or 2. Results and Discussion
carbon nanotubes embedded in a polymer matrix[27] have been
demonstrated as effective modifications for imparting greater In Figure 1, we show the initial photovoltaic characteristics of
moisture tolerance, the latter also improving thermal stability. CH3NH3PbI3−XClX PSCs employing a mesoporous alumina scaf-
In dry conditions, understanding the effects of oxygen on fold (the reference solar cells)[2,36] and lifetime data during aging
PSCs is perhaps less well understood. Device stability work in the presence of O2. Compact TiO2 and doped Spiro-MeOTAD
in this specific area has so far been limited to dark storage were used as electron- and hole-transporting layers, respectively,
tests[2,35] which provide no insight into the prolonged inter- in the reference solar cells (see the Experimental Section for
play between light, oxygen, and biasing on a working PSC. To details). For each device, an initial stabilized PCE was determined
address this issue, we characterize the behavior of a model PSC by first measuring the cell photocurrent during a reverse bias scan
in the presence of oxygen, improving our understanding of and subsequently monitoring the photocurrent at the maximum

20
(a) (b) 15 12

15 10
Current density / mA cm-2

Current density / mA cm-2

12

10 JSC: 18.5 mA cm-2 8

VOC: 0.92 V 9

PCE / %
5 FF: 55% 6
PCE: 9.4% 6
0 4

-5 3
2

-10 0 0
0.0 0.2 0.4 0.6 0.8 1.0 0 5 10 15 20 25 30
Applied bias / V Time / s

(c) 1 (d) 1.6

VOC
JSC
1.2 FF
Metric / Normalized
Relative PCE / a.u.

PCE

0.1 0.8

1% O2 0.4
5% O2
10% O2
0.01 0.0
0 4 8 12 16 20 24 0 4 8 12 16 20 24
Ageing time / hours Ageing time / hours

Figure 1. Solar cell characteristics for the device architecture studied in this work. a) Current–voltage curve of a typical device, as measured via reverse
bias scan. b) Photocurrent and power conversion efficiency of the same device over 30 s continuous illumination at maximum power point, used to
determine an initial quasi-stabilized PCE. c) Relative changes in solar cell power conversion efficiency (PCE) for different vol% of O2 in the testing
chamber atmosphere. Here, devices were first stabilized for 24 h under light and N2 before stress testing. Each quantity is normalized to the efficiency of
the solar cell after the stabilizing step. d) Relative changes in the photovoltaic metrics of a reference solar cell operating in an atmosphere of 5 vol% O2.

1600014 (2 of 10) wileyonlinelibrary.com © 2016 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Energy Mater. 2016, 6, 1600014
 16146840, 2016, 13, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.201600014 by Universitatsbibliothek, Wiley Online Library on [30/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advenergymat.de
www.MaterialsViews.com

FULL PAPER
power point until stable (Figure 1b). According to this protocol, sequential fast and slow bias scans in both a forward and reverse
the tested solar cells exhibited an initial PCE of 9 ± 1% across two direction. As shown in the Supporting Information, Figure S5,
device batch fabrications (see Figure 1a for typical metrics), with a once the solar cell is stabilized under N2 the slow scans converge
champion efficiency of 11.1%. The distribution of solar cell PCEs to give a reproducible PCE value. Fast bias scans result in an
is shown in the Supporting Information, Figure S1. apparent enhancement (reverse direction) or reduction (forward
The stability of the reference solar cells under simulated AM direction) in PCE relative to the slow bias scans, in qualitative
1.5 G solar illumination was monitored in a custom-built atmos- agreement with previous reports.[38] During accelerated aging
pheric chamber. Because our study is focused on the effects of under N2/O2, solar cell PCE undergoes a reduction regardless
oxygen, devices were first measured for 24 h under N2 to con- of the scan conditions used, however the slow scans eventually
firm that the remaining environmental conditions do not trigger give a higher apparent PCE than the fast scans irrespective of the
kinetic processes (such as “burn-in”) that overlap on a com- scan direction used. Work is ongoing to rationalize this behavior,
parable timescale. Specifically, this “stabilizing” step provided however we hypothesize that it reflects a progressive increase of
time for the PSCs to undergo limited burn-in (either positive or the forward bias stabilization time required to reach peak voltage
negative) due to continuous illumination, voltage biasing, and and current, as compared to the time spent at forward bias in
equilibrating to the chamber temperature (40–45 °C). Further- the fast scan routine. It has been empirically observed that cells
more, a slow voltage scan (0–1 V, 0.05 V step size, 10 s wait) was which exhibit the most severe hysteresis generally require longer
employed throughout each experiment to balance the monitoring to reach stabilized photocurrent.[51] The increased timescale may
of individual solar cell metrics with conditions likely to be expe- be due to a combination of increased density of traps and mobile
rienced in real world applications (i.e., maximum-power-point ionic species, in addition to a change in the activation energies
tracking). Unless explicitly stated, solar cell efficiency metrics underlying migration of ionic species.
correspond to an average of values obtained from slow forward To gain additional information about the effects of O2 expo-
and reverse bias scans. Figure S2–S4 in the Supporting Informa- sure, we consider the evolution in solar cell characteristics as
tion provide additional information on our testing procedure. measured through TPC and TPV. Individual measurements
Figure 1c,d presents the evolution in PCE of perovskite solar were averaged over ≈10 s, a period that is commensurate with
cells exposed to oxygen, where data correspond to testing chamber the wait time used in our standard IV measurements. Results
atmospheres with a fixed vol% of O2 relative to N2. Both gases from TPC are shown in Figure 2.
were dry as confirmed by an external gas sensor. Throughout all In Figure 2a, the TPC current for the stabilized device is
aging tests, the atmospheric concentration of H2O was typically characterized by a prompt rise to a constant value when the
10–30 ppm, which should be insignificant for degradation over LED is switched on. After the LED is switched off, the cur-
the timescales of the experiment. From Figure 1c, it can be seen rent undergoes a rapid fall with a small tail extending for a
that the PCE of the reference cells undergoes a dramatic reduc- few tens of microseconds. This type of response is qualitatively
tion due to operating in an N2/O2 atmosphere. The severity of indicative of a solar cell with good charge transport and extrac-
this process is clearly linked to the absolute vol% of O2; after 24 h tion characteristics.[52–54] With increasing exposure time of the
stress testing, 70% of the initial efficiency is lost when O2 is cell to O2, we see the emergence of a decay in photocurrent
present at 1 vol%. When the O2 concentration is increased to at the end of the LED pulse. Following LED switch off, here
10 vol%, the same relative performance drop occurs after 2.7 h. highlighted in Figure 2b, the transient current becomes nega-
Figure 1d gives further insight into the origin of the performance tive with increased aging of the solar cell (i.e., charge is being
decrease, where data for the intermediate stress testing case of injected into the solar cell rather than being extracted). We
5 vol% O2 are presented. As shown, power conversion efficiency clarify the kinetics of these processes in Figure 2c,d. Figure 2c
is primarily lost via a reduction in photocurrent density (JSC). shows a comparison between current values extracted immedi-
Losses in open circuit voltage (VOC) are also observed, however the ately after the LED is switched on and just before it is switched
device fill factor undergoes a temporary increase during the early off. Figure 2d shows the total extracted charge during the decay
stages of stress testing before reducing, this behavior potentially after the LED is switched off. From these figures it can be seen
reflecting changes in ionic mobility in the sample on a timescale that the photocurrent immediately after LED switch on does
that is commensurate with our slow voltage scans.[37] not change until about 12 h into the stress test. Note that by
We recognize that the perceived kinetics of degradation in this time the quasi-stabilized photocurrent has already under-
PSCs may be influenced by measurement history alongside the gone a substantial reduction: 64% as measured through TPC
conditions of the aging environment. PSCs routinely exhibit (photocurrent reading at the end of the LED pulse) and 80%
unstable current–voltage characteristics, described as either a as measured through IV, with the mismatch potentially arising
“hysteresis effect”[38] or a “temporary current enhancement by from variations in photocurrent dependency on light intensity.
bias”.[39] The underlying origins for this behavior are still under Beyond 12 h, further reductions in photocurrent are correlated
debate,[38,40–42] however it is thought to involve the motion of with the extracted charge becoming less negative.
ions or defects within the perovskite and their ability to adjust We interpret these measurements as an indication that the
the built-in field of a solar cell during illumination or prebiasing, PSC has undergone two distinct phases of degradation over the
alongside the presence of electronic traps which facilitate trap- course of the experiment. For low cumulative O2 exposure (i.e.,
assisted recombination.[2,43–47] The timescales and temperatures before ≈12 h in this case), photocurrent loss is due to the forma-
over which hysteresis effects are observable depend strongly on tion of charge barriers at the PSC electrode interface layers that
the architecture of the PSC.[37,48–50] To appreciate how hysteresis act to screen the built-in field, thereby reducing the efficiency of
may evolve during accelerated aging, one device was subject to charge extraction.[41] The simultaneous emergence of a negative

Adv. Energy Mater. 2016, 6, 1600014 © 2016 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com (3 of 10) 1600014
 16146840, 2016, 13, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.201600014 by Universitatsbibliothek, Wiley Online Library on [30/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advenergymat.de
www.MaterialsViews.com
FULL PAPER

(a) 0.07 (b) 0.008

Stabilized
0.06 1 hour
2 hours 0.004
0.05
4 hours

Current / mA
Current / mA

0.04 6 hours
8 hours
0.03 12 hours 0.000
16 hours
0.02
20 hours
0.01 24 hours -0.004

0.00

-0.01 -0.008
-0.2 0.0 0.2 0.4 0.6 0.8 1.0 0.40 0.45 0.50 0.55 0.60
Time / ms Time / ms

0.06
(c) (d) 0.4
0.05

Extracted charge / nC
0.2
0.04
Current / mA

0.0
0.03
-0.2
0.02
-0.4
0.01 LED switch on
LED switch off -0.6
0.00
0 4 8 12 16 20 24 0 4 8 12 16 20 24
Ageing time / hours Ageing time / hours

Figure 2. Transient photocurrent data for the reference solar cell aged under 5% O2. a) TPC signal at distinct stages in the stress test, where 0 ms
corresponds to the start of the 0.4 ms LED illumination pulse. b) A magnified region corresponding to the current decay (t > 0.4 ms, LED off). c) The
evolution in TPC current throughout the aging test at two distinct points (LED switch on: averaged between 0.004 and 0.014 ms; LED switch off: quasi-
stabilized photocurrent measured just before switch-off at 0.4 ms). d) The integrated current during the transient decay is plotted against aging time.

compensation current in the transient decay suggests that once phase of the experiment, we propose that the perovskite itself
the LED has been switched off, the solar cell retains an excess has now undergone material degradation, more severe than just
of trapped charge that continues to modify its internal electric inducing defects. This damage will act to reduce the efficiency
field. In order for the device to be brought back to equilibrium, of light harvesting, and/or trigger sub-microsecond recombina-
new charge must be injected into the cell to neutralize the tion of charge carriers that is otherwise not resolved during our
trapped charge that is otherwise not extracted – a “compensation TPC measurement. As a result of these specific losses, the mag-
current”, which will appears as negative here.[52] We propose nitude of the compensation current required to neutralize the
that the underlying mechanism for the formation of charge bar- (now fewer) trapped charges after LED illumination decreases,
riers is an increase in the density of mobile defects (or ionic spe- as evidenced in Figure 2d.
cies) and electronic traps within the perovskite layer.[43,55] For the To explore the idea of a two-stage degradation process, and
mobile defects, their activation energy is low enough to enable determine the permanency of any efficiency loss in PSCs caused
them to screen the built-in field on the timescale of hundreds by operating with exposure to oxygen, we continued to monitor
of microseconds to seconds whilst the device is under illumi- the device that had been aged under 1% O2 (which underwent
nation. This behavior is analogous to the way in which mobile the smallest reduction in PCE after 24 h) after the oxygen
defects are thought to affect JV scans, where holding at high volt- supply to the sample chamber had been switched off. As shown
ages permits movement of defects in such a way as to enhance in Figure 3, extended operation under dry nitrogen results in
the field in the device and assist charge extraction (resulting the partial recovery of PCE (≈30% of the initial value) that had
in the enhanced reverse JV scan).[18,38] The compensation cur- otherwise been lost during operation with oxygen present. Data
rent is then related to the defects diffusing away from the elec- in Figure 3a show the close correlation between PCE evolution
trodes and neutralization of the now built-up charge. These and changes in photocurrent, indicating that further insight
processes continue until, following extended O2 exposure (after can be provided by our in situ TPC measurements. TPC traces
≈12 h), a loss in transient photocurrent is eventually recorded at distinct stages of this experiment are presented in Figure 3b,
at the beginning of the LED pulse. We use this relatively fast where it can be seen that recovery modifies the PSC photo-
measurement as a proxy for photocurrent efficiency before current over hundreds of microseconds, this timescale being
poor extraction effects dominate the TPC dynamics. Because commensurate with the proposed screening process of the PSC
we observe lower instantaneous photocurrent in this second built-in field discussed earlier. In addition, the recovery process

1600014 (4 of 10) wileyonlinelibrary.com © 2016 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Energy Mater. 2016, 6, 1600014
 16146840, 2016, 13, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.201600014 by Universitatsbibliothek, Wiley Online Library on [30/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advenergymat.de
www.MaterialsViews.com

FULL PAPER
(a) 1.0 10000
(b) 0.06
Stabilized
0.05 24 hours
0.8 48 hours

Metric / Normalized
1000 0.04

Current / mA
0.6

O2 / ppm
0.03
100 0.02
0.4
0.01
10
0.2
PCE 0.00
JSC
0.0 1 -0.01
0 8 16 24 32 40 48 -0.2 0.0 0.2 0.4 0.6 0.8 1.0
Ageing time / hours Time / ms

Figure 3. Aging data for a reference solar cell exposed to 1% O2 during operation. After 24 h, the O2 supply to the sample testing chamber is switched
off. a) The relative solar cell PCE and JSC is plotted against aging time (left axis), alongside the atmospheric concentration of O2 (right axis). b) Transient
photocurrent data before O2 aging (stabilized), after 24 h and after 48 h.

“switches” the transient decay current from being negative (cor- mesoporous alumina scaffold, however several studies have
responding to charge injection) to positive (corresponding to considered photovoltage decay behavior in CH3NH3PbI3 PSCs
charge extraction). Note the marginal changes in TPC current employing a mesoporous TiO2 scaffold.[39,56,57] Critically, TPV
shortly after the LED switches on. Repeating this experiment decays in these device architectures were strongly biexponen-
on PSCs aged under 5% O2 for 24 h did not observe a recovery tial, in contrast to the monoexponential behavior apparently
in device performance during subsequent operation under N2 observed here. O’Regan et al. proposed that under 1 sun con-
(see the Supporting Information, Figure S6). ditions, photovoltage decays with a lifetime of around 1 μs
Further insight into the evolution of the solar cell during correspond to the dominant electron/hole recombination
accelerated aging can be obtained through transient photo- channel.[39] The physical location of this recombination pro-
voltage measurements. From Figure 1b it was shown that cess was not confirmed, however related work suggested that
during aging under 5% O2, the solar cell VOC underwent an the long-lifetime decay component relates to charge recombi-
approximately linear decrease over 24 h. Figure 4 shows photo- nation within the TiO2 scaffold[56,57] with the shorter compo-
voltage decay kinetics, as measured through TPV, at distinct nent (up to several microseconds) quantifying carrier lifetime
points in the experiment. It can be seen (Figure 4b) that the within the perovskite layer. Although we caution against the
photovoltage decay time increases over the course of the experi- comparison of absolute TPV decay times between independent
ment, from within the sub-microsecond resolution of our setup studies, due to likely variations in device preparation and meas-
up to tens of microseconds. This trend has previously been urement conditions, in our solar cells the volume fraction of
observed in PBDTTT-EFT:PC71BM organic photovoltaic devices TiO2 is relatively low because the metal oxide exists as a com-
during burn-in, where in both cases the TPV measurements pact cathode interface layer only. We therefore suggest that our
were performed under constant white light bias.[53] Because TPV measurements are primarily sensitive to charge recombi-
this does not necessarily allow for the internal conditions of nation within the perovskite layer. When VOC is matched close
the solar cell to be matched (e.g., due to differences in charge to 1 sun conditions, the measured faster decays in devices aged
density) we also consider TPV kinetics under varying incident for 4 h are likely a result of more recombination sites being
white light bias at distinct stages of the aging process. These present, a consequence of the increased number of electronic
specific results are presented in Figure 4c,d. traps and mobile defects that account for the proposed loss
From Figure 4c it can be seen that upon adjusting the bias in quasi-steady state photocurrent.[58–60] The extent to which
light intensity to set an open-circuit voltage of 0.71 V, the solar mobile defects and electronics traps are linked is presently
cell in its stabilized state exhibits a slower photovoltage decay unknown, and whilst we acknowledge that our TPC/TPV meas-
compared against its aged state. Characteristic decay times – urements are primarily sensitive to processes that take place
assuming monoexpontential kinetics – are 11 and 5 μs, respec- on the microsecond timescale, different kinetics could well be
tively. We have repeated this experiment on an additional device observed if these measurements are repeated on a timescale
from the same production batch and observed a similar trend that is commensurate with defect mobility.[43–47]
when matching VOC at 0.81 V (i.e., close to 1 sun conditions, see To identify the underlying mechanisms behind the irrevers-
the Supporting Information, Figure S7). In Figure 4d, extracted ible loss in PCE of the reference cells, we note that recent work
photovoltage decay times are presented at different VOC values on aged CH3NH3PbI3 films identified a detrimental photochem-
at distinct stages in the aging experiment. Specifically, upon ical reaction involving oxygen.[61,62] Specifically, it was proposed
aging in 5% O2 for 4 h, the relative loss in steady-state solar cell that photogenerated electrons in the perovskite permit a reac-
photocurrent is 50%. After 24 h, this loss now amounts to 90%, tion between the methylammonium organic cation and super-
with the VOC also being significantly reduced (see Figure 4b). oxide, a process that results in decomposition of CH3NH3PbI3
To interpret our TPV data, we first consider an appropriate with methylamine, PbI2, I2, and H2O as byproducts:
mechanism which can be used to describe photovoltage decay
in our solar cells. To the best of our knowledge, TPV data have
Deprotonation
not been published on CH3NH3PbI3−XClX PSCs employing a CH 3NH 3PbI3 + O •−
2 ⎯⎯⎯⎯⎯⎯
→ CH 3NH 2 + PbI 2 + 1/2 I 2 + H 2O (1)

Adv. Energy Mater. 2016, 6, 1600014 © 2016 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com (5 of 10) 1600014
 16146840, 2016, 13, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.201600014 by Universitatsbibliothek, Wiley Online Library on [30/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advenergymat.de
www.MaterialsViews.com
FULL PAPER

(a) Stabilized (b) 0.80

4 hours
8 hours
1E-3 12 hours 0.75
16 hours
Voltage / V

20 hours

VOC / V
24 hours 0.70

0.65
1E-4

0.60
0 20 40 60 80 100 0 4 8 12 16 20 24
Time / μ s Ageing time / hours

(c) Stabilized (d) 1E-4

Aged (4 hours)

0.01

Decay time / s
Voltage / V

1E-5
1E-3

Stabilized
Aged (4 hours)
Aged (24 hours)
1E-4 1E-6
-50 -25 0 25 50 75 100 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85
Time / μ s VOC / V

Figure 4. a) In situ transient photovoltage decay kinetics for a reference solar cell at different aging times under 5% O2, as measured under constant
white light bias (1 sun). For ease of comparison the decay profiles are presented after subtraction of VOC. b) Absolute VOC values of the solar cell
during aging. c) Transient photovoltage decay kinetics for a nominally identical device aged under 5% O2, with neutral density filters used to match
device VOC at 0.71 V. In this figure, 0 µs corresponds to the end of the LED perturbation pulse. The relative voltage perturbations for the device in its
stabilized and aged state are 5% and 1%, respectively. d) Photovoltage decay times across a range of VOC values for the same device in its stabilized
and aged (5% O2) states.

The damage caused to the perovskite layer by this pro- observe a striking difference in discoloration for devices held
cess was shown to be linked to the efficiency of electron at the two conditions.
extraction; decomposition via photooxidation was slower when For the solar cell held continuously at open circuit, a significant
CH3NH3PbI3 was interfaced to a mesoporous TiO2 scaffold discoloration is observed relative to the control (c.f. Figure 5a,c),
when compared against an equivalent perovskite film without with the underlying Au electrode more visible. Regions with
an electron acceptor present,[62] indicating the importance of a characteristic yellow hue are present across the substrate
electron density in the perovskite when discussing the effects (photographs of the devices aged under 5% O2 are presented in
of photostability. We can understand this process on complete the Supporting Information, Figure S8). In contrast, the color
PSCs using CH3NH3PbI3−XClX by undertaking the following changes between the control and the solar cell held at short
straightforward experiment: two devices
were stabilized and subsequently aged under
10% O2 whilst being held at either short cir-
cuit or open circuit. The cell held at short cir-
cuit will have a relatively low average charge
density since all photogenerated charges
will be extracted rapidly. When the device
is held at open-circuit, similar to an isolated
film, the photogenerated charges cannot be
extracted and so the charge density will be Figure 5. Optical microscopy images of reference solar cells aged for 24 h under 10% O . a)
2
much higher. After aging, the devices were Control device which was not subject to aging, b) device held at short circuit throughout aging,
visually examined. As shown in Figure 5, we and c) device held at open circuit throughout aging. Scale bar represents 0.5 mm.

1600014 (6 of 10) wileyonlinelibrary.com © 2016 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Energy Mater. 2016, 6, 1600014
 16146840, 2016, 13, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.201600014 by Universitatsbibliothek, Wiley Online Library on [30/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advenergymat.de
www.MaterialsViews.com

of the tetragonal perovskite phase.[2] The relatively low-intensity

FULL PAPER
circuit (Figure 5b) are less apparent. This straightforward test
supports the literature discussed earlier and confirms that the feature at 2θ = 12.67° can be attributed to PbI2,[3] its presence
oxygen tolerance of PSCs is strongly related to charge density in the XRD pattern of the fresh device suggesting incomplete
within the perovskite layer. As a consequence, efforts to develop conversion of the perovskite precursor during device fabrica-
PSCs with enhanced O2 stability should therefore aim to main- tion. For the aged samples, an 85% reduction in X-ray scat-
tain a high efficiency of charge extraction, either through the tering intensity of the 110 perovskite reflection (at 2θ = 14.08°)
use of conducting scaffolds or high performance compact is evidenced after 24 h relative to the fresh device. In contrast,
electrode interface layers. However, clearly there is a conun- the changes in X-ray scattering intensity after 4 h are minimal
drum here, since to obtain the maximum operating voltages and (approximately 1%), providing further support to our hypoth-
hence efficiency significant splitting of the quasi-Fermi levels esis that device degradation occurs over two distinct phases.
for electrons and holes needs to be occurring during operation Namely, phase (1) is the initial formation of mobile defects
at maximum power point. Hence to reach the ultimate high within the device, which reduce the efficiency of charge extrac-
efficiencies while maintaining high stability, it is likely that tion in the cell and lead to a build-up of charge. Over time and
the intrinsic sensitivity of the perovskite to this reaction with under conditions of continued device illumination and oxygen
oxygen will need to be reduced. This may be achieved via the exposure, these blocked charges will contribute toward the oxi-
reduction of residual defects in the perovskite layer that form dation of the perovskite material, resulting in its decomposition
during sample preparation,[63] or through chemical substitution (phase (2)).
(for example, by replacement of methylammonium with forma- In general, the XRD measurements of the aged devices show
midinium or cesium).[64,65] Alongside these expected develop- no clear correlation with either the level of amorphous content
ments, we anticipate that related studies will also seek to clarify (that would likely modify the background level of X-ray scat-
the role of non-perovskite phases in governing sample sensi- tering) or the amount of crystalline PbI2, the latter inferred from
tivity. This would be particularly relevant when these phases are the intensity of X-ray scatter at 2θ = 12.6° (see Figure 6b). We
present by design, as can be the case with, for example, excess suggest therefore that prolonged aging primarily induces the
PbI2.[66,67] In terms of module design, device stability will also loss of mass content from the sample, with CH3NH3PbI3−XClX
be linked to the ability of oxygen to efficiently diffuse laterally breaking down – and the PSC undergoing irreversible degra-
and vertically through each layer in the solar cell. For example, dation – due to the loss of methylamine and iodine from the
for PSCs employing Spiro-MeOTAD, such as those investigated sample during the second degradation phase.[61] In making this
in this study, thin films of this material are known to form pin- statement we acknowledge that the X-ray beam in these meas-
holes through which oxygen molecules may efficiently travel. urements samples an area of several mm2 across the device,
Recent work has however demonstrated that the number of therefore providing an insight into average crystallinity only.
pinholes may be reduced by preparing the hole transport layer To complement this dataset we also performed energy-disper-
from a volatile solvent,[68] suggesting a straightforward strategy sive X-ray spectroscopy (EDS) measurements on similarly aged
for mitigating this specific degradation pathway. devices to characterize their chemical composition (note that
Lastly, to confirm that decomposition of the perovskite takes EDS was performed after removal of the Au electrode). The
place within our reference solar cells, and relate these changes results of this experiment are summarized in Figure 7. Here, it
to our transient data, new PSCs were aged under 5% O2 for 4 or can be seen that aging induces an increase in Pb content within
24 h. These devices were subsequently measured using XRD to the sample relative to the concentration of both C and I, with
provide information on the crystallinity and composition of the the largest increase taking place between 4 and 24 h aging. In
perovskite layer. Data are presented in Figure 6. addition, the absolute wt% of C and I decrease with aging (full
We attribute the relatively narrow peaks located at 2θ = 14.08° EDS survey data are presented in the Supporting Information,
and 28.35° to X-ray scattering from the (110) and (220) planes Table S1), supporting our interpretation of the XRD data. When

(a) 100000 w/o perovskite

(b) 80000
w/o perovskite
Fresh Fresh
Aged 4 hours Aged 4 hours
10000 Aged 24 hours 60000 Aged 24 hours
Counts (Offset)

Counts

1000
40000

100

20000
10

1 0
10 20 30 40 50 60 12.0 12.5 13.0 13.5 14.0 14.5 15.0
2θ / ° 2θ / °

Figure 6. X-ray diffraction data for reference perovskite solar cells aged under 5% O2. a) The XRD patterns have been offset for clarity. b) Raw data
highlighting features over the 2θ range 12°–15° are presented. Note that data corresponding to “w/o perovskite” correspond to an equivalent solar
cell structure without any perovskite material present (i.e., glass/FTO/TiO2/Al2O3/Spiro-MeOTAD/Au).

Adv. Energy Mater. 2016, 6, 1600014 © 2016 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com (7 of 10) 1600014
 16146840, 2016, 13, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.201600014 by Universitatsbibliothek, Wiley Online Library on [30/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advenergymat.de
www.MaterialsViews.com
FULL PAPER

50 4. Experimental Section
C
Materials and Solvents: Unless otherwise stated, all materials were
I purchased from Sigma-Aldrich or Alfa Aesar and used as received. Spiro-
Relative concentration / %

40 Pb OMeTAD was purchased from Borun Chemicals.

Sample Preparation: For mesosuperstructured device fabrication,
fluorine-doped tin oxide (FTO)-coated glass substrates (Pilkington,
30 7 Ω 䊐−1) were first cleaned with detergent and deionized water, then
with acetone and isopropanol, followed by oxygen plasma treatment. The
electron-accepting TiO2 compact layer was spin-coated (2000 rpm for 60 s)
from a precursor solution prepared by adding titanium isopropoxide
20 into a mixture of ethanol and diluted HCl. The relative amounts used
were 70 µL titanium isopropoxide, 1 mL ethanol, and 7 µL 2 M HCl.
After spin-coating, the sample was sintered at 500 °C. An alumina
10 mesoporous scaffold was then deposited by spin-coating (2500 rpm for
60 s) a colloidal dispersion of 20 nm Al2O3 nanoparticles in isopropanol
followed by drying at 150 °C. The perovskite layer was then spin-coated
in air in a humidity-controlled chamber with 15–30% relative humidity
0 (2000 rpm for 45 s) and annealed at 100 °C for 120 min. Spiro-OMeTAD
Fresh Aged 4 hours Aged 24 hours hole-transport material was then deposited via spin-coating (2000 rpm
for 45 s) a 0.0788 M solution in chlorobenzene, with additives of
Device
0.0184 M lithium bis(trifluoromethanesulfonyl)imide (added in 0.61 M
Figure 7. Composition analysis for perovskite solar cells aged under acetonitrile solution) and 0.0659M 4-tert-butylpyridine. Gold electrodes
5% O2, as determined from EDS measurements. Here, the data have were deposited via thermal evaporation to complete the devices.
been normalized to consider the relative fractions of C, I, and Pb in each Sample Characterization–PCE: The current density–voltage (J–V)
sample only. curves were measured (2400 Series SourceMeter, Keithley Instruments)
under simulated AM 1.5 sunlight at 100 mW cm−2 irradiance generated
by an Abet Class AAB Sun 2000 simulator, with the intensity calibrated
with an NREL calibrated KG5 filtered Si reference cell. The mismatch
combining these results with the data presented in Figures S6 factor was calculated to be less than 1%. The solar cells were masked
and S8 (in the Supporting Information) we therefore conclude with a metal aperture to define the active area of 0.0625 cm−2 and
measured in a light-tight sample holder to minimize any edge effects
that prolonged PSC operation in the presence of oxygen results
and ensure that the reference cell and test cell are located during
in the severe breakdown of the perovskite phase, with the measurement in the same spot under the solar simulator. For the fast JV
spatially resolved level of degradation strongly influenced by the scans, cells were scanned from forward bias to short-circuit at a rate of
local ability to extract photogenerated charge carriers. Work is 0.38 V s−1 after holding under illumination at 1.4 V for 5 s. The maximum
currently underway to measure the photovoltaic and composi- power point was determined from these fast JV scans and the current
tion characteristics of a PSC in situ to provide a quantifiable link was measured holding at this voltage for the stabilized power output
between these properties. measurements.
Sample Characterization–In situ IV, TPC, TPV: To minimize device
degradation, solar cells were stored under vacuum or glovebox
conditions before transfer to the aging setup. During each experiment
the chamber was held at a slight overpressure, with the N2 supply
3. Conclusions prefiltered (SGT Super Clean) to minimize residual oxygen, moisture
and hydrocarbon content. Oxygen and water content was monitored
Transient photocurrent and photovoltage measurements in situ using a gas analyzer (Rapidox 3100D – Cambridge Sensotec)
have been used to give insight into the behavior of model connected to the gas exit line of the sample chamber. A Newport Solar
Simulator with equivalent AM 1.5 G 1-sun output was used to illuminate
CH3NH3PbI3−XClX perovskite solar cells operating in the pres-
the entire device substrate; solar cell current–voltage characteristics
ence of oxygen. Stress testing under these conditions evidences were collected using a Keithley 2636 SMU. A 465 nm LED (LED465E,
a reduction in solar cell efficiency that is dominated by a loss in Thor Labs) was used as the light source for transient experiments,
photocurrent. This process occurs over two distinct phases: an connected to an Agilent 33500B wavefunction generator and a purpose-
initial semi-reversible screening of the built-in field of the solar built low-noise power supply. Solar cell transients were recorded by
cell that results from an increased defect density and, following connecting the device to a Tektronix DPO 3032 oscilloscope. For TPC
extended O2 exposure, breakdown of the light-harvesting per- measurements, the device was connected to the 50 Ω input of the
oscilloscope via a custom transimpedance amplifier. The 1 MΩ input
ovskite layer itself. In addition to the relative concentration of of the oscilloscope was used for TPV measurements. White-light bias
oxygen present during testing (that is, the external conditions), for TPV measurements was provided by the solar simulator. A custom-
the extent to which CH3NH3PbI3−XClX undergoes photooxida- written LabView VI was used for instrument control and data acquisition
tion is strongly influenced by charge density and charge-extrac- (a looped sequence of IV, TPC, TPV measurements).
tion efficiency, in agreement with recent work on CH3NH3PbI3 Sample Characterization–X-Ray Scattering: Coupled 2θ–θ scans of
films. Our findings provide an important contribution toward X-ray diffraction from perovskite devices were obtained using a Bruker
D8 X-ray Diffractometer (step size 0.01°, scan time 40 min). Samples
understanding the operation of perovskite solar cells under
were measured under ambient conditions and low light to minimize
working conditions, as well as identifying a rational framework sample degradation during measurement.
toward mitigating oxygen degradation, helping to realize these Sample Characterization–EDS: Measurements were performed using
devices as an efficient and stable PV technology. an FEI Philips XL30 sFEG with an EDS (EDAX) detector. An accelerating

1600014 (8 of 10) wileyonlinelibrary.com © 2016 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Energy Mater. 2016, 6, 1600014
 16146840, 2016, 13, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.201600014 by Universitatsbibliothek, Wiley Online Library on [30/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advenergymat.de
www.MaterialsViews.com

FULL PAPER
voltage of 15 kV was used throughout, with EDS spectra acquired from a [18] T. Leijtens, G. E. Eperon, N. K. Noel, S. Habisreutinger, P. Petrozza,
series of maps (50 × 50 µm2) on each sample. H. J. Snaith, Adv. Energy. Mater. 2015, 5, 1500963.
[19] K. Leo, Nat. Nanotechnol. 2015, 10, 574.
[20] A. Y. Mei, X. Li, L. F. Liu, Z. L. Ku, T. F. Liu, Y. G. Rong, M. Xu,
Supporting Information M. Hu, J. Z. Chen, Y. Yang, M. Gratzel, H. W. Han, Science 2014,
345, 295.
Supporting Information is available from the Wiley Online Library or [21] X. Li, M. Tschumi, H. W. Han, S. S. Babkair, R. A. Alzubaydi,
from the author. A. A. Ansari, S. S. Habib, M. K. Nazeeruddin, S. M. Zakeeruddin,
M. Gratzel, Energy Technol. 2015, 3, 551.
[22] W. S. Yang, J. H. Noh, N. J. Jeon, Y. C. Kim, S. Ryu, J. Seo, S. I. Seok,
Science 2015, 348, 1234.
Acknowledgements [23] T. Baikie, Y. N. Fang, J. M. Kadro, M. Schreyer, F. X. Wei, S. G. Mhaisalkar,
This work was supported by SABIC and by the EPSRC, including by the M. Graetzel, T. J. White, J. Mater. Chem. A 2013, 1, 5628.
Supergen Supersolar Consortium (EP/J017361/1) and the European Union [24] J. L. Yang, B. D. Siempelkamp, D. Y. Liu, T. L. Kelly, ACS Nano 2015,
Seventh Framework Program [FP7 2007-2003] under grant agreement 9, 1955.
604032 of the MESO project. G.E.E. was supported by the EPSRC and [25] J. A. Christians, P. A. M. Herrera, P. V. Kamat, J. Am. Chem. Soc.
Oxford Photovoltaics Ltd. through a Nanotechnology KTN CASE award. 2015, 137, 1530.
J.T.-W.W. acknowledges the Swire Educational Trust for supporting his [26] A. M. A. Leguy, Y. Hu, M. Campoy-Quiles, M. I. Alonso, O. J. Weber,
D.Phil. study at Oxford. The authors thank Dr Sian Dutton (University of P. Azarhoosh, M. van Schilfgaarde, M. T. Weller, T. Bein, J. Nelson,
Cambridge) for access to XRD facilities and Dr Felix Deschler (University P. Docampo, P. R. F. Barnes, Chem. Mater. 2015, 27, 3397.
of Cambridge) for helpful discussions. The data underlying this publication [27] S. N. Habisreutinger, T. Leijtens, G. E. Eperon, S. D. Stranks,
are available at https://www.repository.cam.ac.uk/handle/1810/254858. R. J. Nicholas, H. J. Snaith, Nano Lett. 2014, 14, 5561.
[28] J. W. Liu, S. Pathak, T. Stergiopoulos, T. Leijtens, K. Wojciechowski,
Received: January 4, 2016
S. Schumann, N. Kausch-Busies, H. J. Snaith, J. Phys. Chem. Lett.
Revised: March 24, 2016
2015, 6, 1666.
Published online: April 26, 2016
[29] G. D. Niu, W. Z. Li, F. Q. Meng, L. D. Wang, H. P. Dong, Y. Qiu,
J. Mater. Chem. A 2014, 2, 705.
[30] H. P. Zhou, Q. Chen, G. Li, S. Luo, T. B. Song, H. S. Duan,
[1] A. Kojima, K. Teshima, Y. Shirai, T. Miyasaka, J. Am. Chem. Soc. Z. R. Hong, J. B. You, Y. S. Liu, Y. Yang, Science 2014, 345, 542.
2009, 131, 6050. [31] G. E. Eperon, S. N. Habisreutinger, T. Leijtens, B. J. Bruijnaers,
[2] M. M. Lee, J. Teuscher, T. Miyasaka, T. N. Murakami, H. J. Snaith, J. J. van Franeker, D. W. de Quilettes, S. Pathak, R. J. Sutton,
Science 2012, 338, 643. G. Grancini, D. S. Ginger, R. A. Janssen, A. Petrozza, H. J. Snaith,
[3] J. Burschka, N. Pellet, S. J. Moon, R. Humphry-Baker, P. Gao, ACS Nano 2015, 9, 9380.
M. K. Nazeeruddin, M. Gratzel, Nature 2013, 499, 316. [32] J. M. Frost, K. T. Butler, F. Brivio, C. H. Hendon, M. van Schilfgaarde,
[4] M. A. Green, A. Ho-Baillie, H. J. Snaith, Nat. Photonics 2014, 8, 506. A. Walsh, Nano Lett. 2014, 14, 2584.
[5] H. J. Snaith, J. Phys. Chem. Lett. 2013, 4, 3623. [33] I. C. Smith, E. T. Hoke, D. Solis-Ibarra, M. D. McGehee,
[6] G. Hodes, Science 2013, 342, 317. H. I. Karunadasa, Angew. Chem. Int. Ed. 2014, 53, 11232.
[7] J. H. Noh, S. H. Im, J. H. Heo, T. N. Mandal, S. I. Seok, Nano Lett. [34] J. Liu, Y. Z. Wu, C. J. Qin, X. D. Yang, T. Yasuda, A. Islam, K. Zhang,
2013, 13, 1764. W. Q. Peng, W. Chen, L. Y. Han, Energy Environ. Sci. 2014, 7, 2963.
[8] G. E. Eperon, S. D. Stranks, C. Menelaou, M. B. Johnston, [35] H. S. Kim, C. R. Lee, J. H. Im, K. B. Lee, T. Moehl, A. Marchioro,
L. M. Herz, H. J. Snaith, Energy Environ. Sci. 2014, 7, 982. S. J. Moon, R. Humphry-Baker, J. H. Yum, J. E. Moser, M. Gratzel,
[9] L. Protesescu, S. Yakunin, M. I. Bodnarchuk, F. Krieg, R. Caputo, N. G. Park, Sci. Rep. 2012, 2, 591.
C. H. Hendon, R. X. Yang, A. Walsh, M. V. Kovalenko, Nano Lett. [36] J. M. Ball, M. M. Lee, A. Hey, H. J. Snaith, Energy Environ. Sci. 2013,
2015, 15, 3692. 6, 1739.
[10] C. C. Stoumpos, C. D. Malliakas, M. G. Kanatzidis, Inorg. Chem. [37] T. Leijtens, E. T. Hoke, G. Grancini, D. J. Slotcavage,
2013, 52, 9019. G. E. Eperon, J. M. Ball, M. De Bastiani, A. R. Bowring, N. Martino,
[11] F. Deschler, M. Price, S. Pathak, L. E. Klintberg, D. D. Jarausch, K. Wojciechowski, M. D. McGehee, H. J. Snaith, A. Petrozza,
R. Higler, S. Huttner, T. Leijtens, S. D. Stranks, H. J. Snaith, Adv. Energy. Mater. 2015, 5, 1500962.
M. Atature, R. T. Phillips, R. H. Friend, J. Phys. Chem. Lett. 2014, 5, [38] H. J. Snaith, A. Abate, J. M. Ball, G. E. Eperon, T. Leijtens,
1421. N. K. Noel, S. D. Stranks, J. T. W. Wang, K. Wojciechowski,
[12] Z. K. Tan, R. S. Moghaddam, M. L. Lai, P. Docampo, R. Higler, W. Zhang, J. Phys. Chem. Lett. 2014, 5, 1511.
F. Deschler, M. Price, A. Sadhanala, L. M. Pazos, D. Credgington, [39] B. C. O’Regan, P. R. F. Barnes, X. E. Li, C. Law, E. Paomares,
F. Hanusch, T. Bein, H. J. Snaith, R. H. Friend, Nat. Nanotechnol. J. M. Marin-Beloqui, J. Am. Chem. Soc. 2015, 137, 5087.
2014, 9, 687. [40] R. S. Sanchez, V. Gonzalez-Pedro, J. W. Lee, N. G. Park, Y. S. Kang,
[13] S. D. Stranks, H. J. Snaith, Nat. Nanotechnol. 2015, 10, 391. I. Mora-Sero, J. Bisquert, J. Phys. Chem. Lett. 2014, 5, 2357.
[14] J. S. Luo, J. H. Im, M. T. Mayer, M. Schreier, M. K. Nazeeruddin, [41] W. Tress, N. Marinova, T. Moehl, S. M. Zakeeruddin,
N. G. Park, S. D. Tilley, H. J. Fan, M. Gratzel, Science 2014, 345, 1593. M. K. Nazeeruddin, M. Gratzel, Energy Environ. Sci. 2015, 8, 995.
[15] L. T. Dou, Y. Yang, J. B. You, Z. R. Hong, W. H. Chang, G. Li, Y. Yang, [42] J. M. Frost, K. T. Butler, A. Walsh, APL Mater. 2014, 2, 081506.
Nat. Commun. 2014, 5, 5404. [43] C. Eames, J. M. Frost, P. R. F. Barnes, B. C. O’Regan, A. Walsh,
[16] S. Yakunin, M. Sytnyk, D. Kriegner, S. Shrestha, M. Richter, M. S. Islam, Nat. Commun. 2015, 6. 7497.
G. J. Matt, H. Azimi, C. J. Brabec, J. Stangl, M. V. Kovalenko, [44] Y. Zhang, M. Z. Liu, G. E. Eperon, T. C. Leijtens, D. McMeekin,
W. Heiss, Nat. Photonics 2015, 9, 444. M. Saliba, W. Zhang, M. de Bastiani, A. Petrozza, L. M. Herz,
[17] G. D. Niu, X. D. Guo, L. D. Wang, J. Mater. Chem. A 2015, 3, 8970. M. B. Johnston, H. Lin, H. J. Snaith, Mater. Horiz. 2015, 2, 315.

Adv. Energy Mater. 2016, 6, 1600014 © 2016 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com (9 of 10) 1600014
 16146840, 2016, 13, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.201600014 by Universitatsbibliothek, Wiley Online Library on [30/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advenergymat.de
www.MaterialsViews.com
FULL PAPER

[45] Y. Zhao, C. J. Liang, H. M. Zhang, D. Li, D. Tian, G. B. Li, X. P. Jing, [60] J. Xu, A. Buin, A. H. Ip, W. Li, O. Voznyy, R. Comin, M. Yuan,
W. G. Zhang, W. K. Xiao, Q. Liu, F. J. Zhang, Z. Q. He, Energy S. Jeon, Z. Ning, J. J. McDowell, P. Kanjanaboos, J. P. Sun, X. Lan,
Environ. Sci. 2015, 8, 1256. L. N. Quan, D. H. Kim, I. G. Hill, P. Maksymovych, E. H. Sargent,
[46] Z. G. Xiao, Y. B. Yuan, Y. C. Shao, Q. Wang, Q. F. Dong, C. Bi, Nat. Commun. 2015, 6, 7081.
P. Sharma, A. Gruverman, J. S. Huang, Nat. Mater. 2015, 14, 193. [61] N. Aristidou, I. Sanchez-Molina, T. Chotchuangchutchaval,
[47] S. van Reenen, M. Kemerink, H. J. Snaith, J. Phys. Chem. Lett. 2015, M. Brown, L. Martinez, T. Rath, S. A. Haque, Angew. Chem. Int. Ed.
6, 3808. 2015, 54, 8208.
[48] H. S. Kim, N. G. Park, J. Phys. Chem. Lett. 2014, 5, 2927. [62] F. T. F. O’Mahony, Y. H. Lee, C. Jellett, S. Dmitrov, D. T. J. Bryant,
[49] L. K. Ono, S. R. Raga, S. H. Wang, Y. Kato, Y. B. Qi, J. Mater. Chem. J. R. Durrant, B. C. O’Regan, M. Graetzel, M. K. Nazeeruddin,
A 2015, 3, 9074. S. A. Haque, J. Mater. Chem. A 2015, 3, 7219.
[50] D. Bryant, S. Wheeler, B. C. O’Regan, T. Watson, P. R. F. Barnes, [63] W. Zhang, M. Saliba, D. T. Moore, S. K. Pathak, M. T. Hörantner,
D. Worsley, J. Durrant, J. Phys. Chem. Lett. 2015, 6, 3190. T. Stergiopoulos, S. D. Stranks, G. E. Eperon, J. A. Alexander-Webber,
[51] N. K. Noel, A. Abate, S. D. Stranks, E. S. Parrott, V. M. Burlakov, A. Abate, A. Sadhanala, S. Yao, Y. Chen, R. H. Friend,
A. Goriely, H. J. Snaith, ACS Nano 2014, 8, 9815. L. A. Estroff, U. Wiesner, H. J. Snaith, Nat. Commun. 2015, 6,
[52] Z. Li, F. Gao, N. C. Greenham, C. R. McNeill, Adv. Funct. Mater. 6142.
2011, 21, 1419. [64] G. E. Eperon, S. D. Stranks, C. Menelaou, M. B. Johnston,
[53] I. Hwang, C. R. McNeill, N. C. Greenham, J. Appl. Phys. 2009, 106, L. M. Herz, H. J. Snaith, Energy Environ. Sci. 2014, 7, 982.
094506. [65] J. Lee, D. Kim, H. Kim, S. Seo, S. M. Cho, N. Park, Adv. Energy
[54] A. J. Pearson, P. E. Hopkinson, E. Couderc, K. Domanski, Mater. 2015, 5, 1501310.
M. Abdi-Jalebia, N. C. Greenham, Org. Electron. 2016, 30, 225. [66] F. Liu, Q. Dong, M. K. Wong, A. B. Djurišić, A. Ng, Z. Ren,
[55] A. Baumann, S. Väth, P. Rieder, M. C. Heiber, K. Tvingsted, Q. Shen, C. Surya, W. K. Chan, J. Wang, A. M. C. Ng, C. Liao,
V. Dyakonov, J. Phys. Chem. Lett. 2015, 6, 2350. H. Li, K. Shih, C. Wei, H. Su, J. Dai, Adv. Energy. Mater. 2016, 6,
[56] J. W. Lee, T. Y. Lee, P. J. Yoo, M. Gratzel, S. Mhaisalkar, N. G. Park, 1502206.
J. Mater. Chem. A 2014, 2, 9251. [67] D. Bi, W. Tress, M. I. Dar, P. Gao, J. Luo, C. Renevier, K. Schenk,
[57] V. Roiati, S. Colella, G. Lerario, L. De Marco, A. Rizzo, A. Listorti, A. Abate, F. Giordano, J. C. Baena, J. Decoppet, S. M. Zakeeruddin,
G. Gigli, Energy Environ. Sci. 2014, 7, 1889. M. K. Nazeeruddin, M. Grätzel, A. Hagfeldt, Sci. Adv. 2016, 2,
[58] J. H. Im, H. S. Kim, N. G. Park, APL Mater. 2014, 2. 081510. 1501170.
[59] Z. G. Xiao, Q. F. Dong, C. Bi, Y. C. Shao, Y. B. Yuan, J. S. Huang, [68] L. K. Ono, S. R. Raga, M. Remeika, A. J. Winchester, A. Gabea, Y. Qi,
Adv. Mater. 2014, 26, 6503. J. Mater. Chem. A 2015, 3, 15451.

1600014 (10 of 10) wileyonlinelibrary.com © 2016 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Energy Mater. 2016, 6, 1600014