Sie sind auf Seite 1von 27

Review

Perovskite Solar Cells www.advenergymat.de

Capturing the Sun: A Review of the Challenges and


Perspectives of Perovskite Solar Cells
Michiel L. Petrus, Johannes Schlipf, Cheng Li, Tanaji P. Gujar, Nadja Giesbrecht,
Peter Müller-Buschbaum, Mukundan Thelakkat, Thomas Bein, Sven Hüttner,*
and Pablo Docampo*

following statement: These materials


Hybrid metal halide perovskites have become one of the hottest topics in can be processed from solution and yet
optoelectronic materials research in recent years. Not only have they sur- exhibit optoelectronic materials proper-
passed everyone’s expectations and achieved similar performance as tried ties described in classic solid-state physics
textbooks. This unprecedented combina-
and true polycrystalline silicon photovoltaic devices, but they are also finding tion has led to photovoltaic devices that
applications in a variety of different fields, including lighting. The main advan- can be processed at room temperature
tages of hybrid metal halide perovskites are simple processability, compatible and achieve performance levels similar to
with large-scale solution processing such as roll-to-roll printing, and abun- industry giant polycrystalline silicon, from
dance of ingredients, all coupled to materials properties reminiscent of GaAs. a starting point of 3.8% in 2009.[1–4] The
first light-emitting electrochemical cells[5]
On the road to this remarkable success, a series of challenges have been
and diodes[6–8] have also been introduced
overcome, while some still remain. In this review, some of these challenges recently, leading to tunable light emission
and possible solutions are described. In particular, understanding of the per- spectra and internal quantum efficiencies
ovskite crystallization process and how this knowledge can be harnessed to exceeding 15%.
enable better performing devices, how to overcome reproducibility issues and The road towards these achievements has
mitigate hysteresis, and the long-term prospects of the technology in terms of been marked by a constant improvement of
perovskite deposition techniques fueled by
stability and sustainability will all be discussed. our increasing understanding of the crys-
tallization processes. The choice of depo-
sition technique, either from solution or
1. Introduction vapor, and the composition and order in which precursor species
are applied has a great influence on the crystallization kinetics.
Recently, hybrid metal halide perovskite materials have become Here, one can distinguish one-step methods where the perovs-
one of the hottest topics in optoelectronic materials research. kite is formed from a precursor mixture dissolved in a common
The excitement of the community can be summarized in the solution, and two-step methods where the inorganic compound
is deposited first and transformed to the perovskite phase later
Dr. M. L. Petrus, N. Giesbrecht, Prof. T. Bein, Dr. P. Docampo by addition of the organic constituents.[9–11] Furthermore, many
Department of Chemistry and Center for NanoScience (CeNS) parameters during processing can have an effect on the crystal-
University of Munich (LMU) lization mechanism and consequently on film morphology, such
Butenandtstr. 11, 81377 Munich, Germany
as the choice of solvents, concentrations and processing additives
E-mail: pablo.docampo@newcastle.ac.uk
that are often not incorporated into the final product.[11–13] We dis-
J. Schlipf, Prof. P. Müller-Buschbaum
Lehrstuhl für Funktionelle Materialien cuss this aspect in Section 2, where we focus our attention on the
Physik-Department most commonly employed solution-based techniques. Addition-
TU München ally, as for any new technology trying to displace an incumbent
James-Franck-Str. 1, 85748 Garching, Germany technology, higher efficiencies, lower costs, reproducibility, sta-
Dr. C. Li, Dr. T. P. Gujar, Prof. S. Hüttner, Prof. M. Thelakkat bility, and sustainability are paramount. In particular, reproduc-
Macromolecular Chemistry I
University of Bayreuth ibility has been a major challenge in perovskite solar cells from
University Str-30, Germany the beginning: small variations in morphology, like crystal size,
E-mail: sven.huettner@uni-bayreuth.de film roughness, and pinholes can have detrimental effects on
Dr. P. Docampo photovoltaic performance.[14] On the other hand, current–voltage
School of Electrical and Electronic Engineering measurements often showed a hysteresis that is rarely seen in
Newcastle University
other photovoltaic technologies.[15] These topics are highlighted
Merz Court, NE1 7RU Newcastle upon Tyne, UK
in Sections 3 and 4. Finally, in Section 5 we discuss the frame-
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/aenm.201700264. work for market introduction, such as cost reduction by choosing
low-cost charge transport layers, or how the lifetime of perovskite
DOI: 10.1002/aenm.201700264 absorbers can be extended by the choice of materials composition.

Adv. Energy Mater. 2017, 7, 1700264 1700264  (1 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

2. Crystallization of Perovskites
Michiel Petrus is a
In order to tune the morphology of perovskite films and thus Postdoctoral Researcher in
the performance of the resulting solar cells, understanding the the Department of Chemistry
crystallization process is crucial. While developing a deposi- at the Ludwig Maximilians
tion technique that allows for full surface coverage of the sub- University (LMU), Munich.
strate initially led to large leaps in efficiency, this is only the He received his Ph.D. degree
first step towards pushing the performance to the limit. Various from Delft University of
studies additionally showed that it is important to maximize Technology in 2014, for his
the crystalline domains in the resulting perovskite solar cells work on donor materials
by introducing different crystallization strategies, which ena- for organic solar cells, and
bled longer crystal growth times.[16] Here we will elaborate the was a visiting scholar at the
different strategies employed so far, summarized in Figure 1. University of Cambridge.
In principle, there are many different pathways for crystalliza- His current research interests include the development
tion in solution, as discussed in great detail by Zhou et al.[17] of perovskite solar cells, with a focus on low-cost hole
While homogeneous nucleation is likely relevant for the forma- transporting materials, environmental impact, and stability
tion of large single crystals, e.g., by inverse temperature crys- of the perovskite.
tallization,[18] heterogeneous crystallization is considered the
dominant mechanism for thin film synthesis. It is induced by Sven Huettner currently
growth of particles or domains from the nucleation sites, e.g., works as an Assistant
the solution/substrate interface by precipitation or supersatura- Professor (Juniorprofessor)
tion, which might be induced by an antisolvent or by attach- at the University of Bayreuth.
ment to a colloid, like a solid–liquid complex.[17] He received his diploma in
physics from the University
of Bayreuth in 2005, and
2.1. One-Step (Solvent Engineering) worked on his Ph.D. degree
in a joint collaboration
Initial studies focused on achieving sufficient surface coverage with Prof. Ullrich Steiner
of the perovskite film by tuning the stoichiometry of the pre- (University of Cambridge) and
cursors. Thus, applying a 3:1 molar ratio of methylammonium Prof. Thelakkat (University of
iodide (MAI) and PbCl2 in dimethylformamide (DMF), Lee Bayreuth). From 2010–2013 he worked as a postdoc in the
et al. were able to fabricate the first perovskite solar cell (PSC) group of Prof. Sir Richard Friend (University of Cambridge)
with power conversion efficiency (PCE) exceeding 10%.[9] Sub- in the field of optoelectronics. His current research is con-
sequently, the influence of processing parameters was further cerned with investigating structure–function relations of
investigated with the focus shifting towards different solvent novel organic and hybrid semiconductors toward applica-
media, like dimethyl sulfoxide (DMSO) or N-methylpyrrolidone tions in photovoltaics, transistors, and sensors.
(NMP), which resulted in the first PSC without a mesoporous
scaffold and topping the 10% efficiency mark.[14] However, Pablo Docampo is currently
simple one-step fabrication methods seem to be more difficult a Senior Lecturer in Physics
to control than sequential deposition on both mesostructured at Newcastle University. He
and planar TiO2 substrates.[10,11] obtained his Ph.D. from the
Jeon et al. and Xiao et al. almost simultaneously reported University of Oxford in 2012,
that adding an antisolvent on top of the wet precursor film for his work on the electronic
during spin-coating induces fast crystallization.[12,21] In a basic properties of metal oxides
approach, Xiao et al. used an equimolar solution of MAI and in solid-state dye-sensitized
PbI2 in DMF, and tested the influence of a variety of antisol- solar cells. He was previ-
vents like chlorobenzene (CB), benzene, xylene, toluene, iso- ously a Marie Curie Fellow
propylalcohol (IPA), and chloroform. The use of the antisolvent and Assistant Professor at
promotes fast nucleation, which results in micron-sized crystals the LMU Munich, where he
and PCEs close to 14% using a planar device architecture.[21] started his group developing novel hybrid halide perovskite
Jeon et al., on the other hand, dissolved their precursors in a materials for optoelectronic applications.
mixture of γ-butyrolactone (GBL) and DMSO, and used toluene
as the antisolvent, and thereby achieved a certified efficiency
of 16.2% for an optimized mixed-MAPb(I1−xBrx)3 perovskite
on mesoporous TiO2, which was a record PCE at that time.[12] The influence of precursor stoichiometry was investigated by
Besides the influence of the solvent on the crystallization, the Nenon et al. using in situ XRD during thermal annealing.[30]
choice of the precursor species and their concentration has Depending on the composition, the film undergoes several
shown a strong impact. Several groups studied the role of lead phase regimes during annealing. Additionally, this has impli-
precursors, like PbI2, PbCl2 and PbAc2.[19,27–29] cations for the thermal stability of the film. Especially in the

Adv. Energy Mater. 2017, 7, 1700264 1700264  (2 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

Figure 1.  Overview of the relations between perovskite thin film deposition methods from solution, starting from simple one-step, where often non-
stoichiometric precursor mixtures are used to achieve homogeneous films a).[14,19] Reproduced with permission.[19] Copyright 2015, Nature Publishing
Group, and two-step methods, where a predeposited metal halide film is converted to perovskite g).[11,20] Reproduced with permission.[20] Copyright
2016, American Chemical Society. The first was improved by either antisolvent drop to induce fast crystallization b,c). (b) Reproduced with permis-
sion.[21] Copright 2014, John Wiley and Sons. (c) Reproduced with permission.[22] Copyright 2016, Royal Society of Chemistry, or by application of
complex-forming solvents like DMSO in a mixed solvent approach e). Reproduced with permission.[12] Copyright 2014, Nature Publishing Group,
or by post-processing f). Reproduced with permission.[23] Copyright 2015, American Chemical Society. Two-step methods also use DMSO to form a
complex d). Reproduced with permission.[24] Copyright 2014, Royal Society of Chemistry, but also different conversion methods were explored either
by spin-coating and interdiffusion h). Reproduced with permission.[25] Copyright 2014, Royal Society of Chemistry, or organic vapor i). Reproduced with
permission.[26] Copyright 2013, American Chemical Society.

presence of chloride, either through methylammonium chlo- the lead precursor and DMSO molecules leads to decelerated
ride (MACl) or PbCl2, the crystallization kinetics seem to reaction kinetics in the conversion to the perovskite phase
change dramatically. This is further discussed in Section 2.3. (cf. Figure 2a). As a result, crystallization becomes more con-
trollable, which leads to uniform crystals, often with increased
sizes.[23,24,33,34] Chemically, this complex can be described as the
2.1.1. Crystallization through Complex Formation solvent molecules donating a free electron pair to the Pb2+ ions.
These so-called Lewis adducts are comparable to the lead-halide
As mentioned before, crossing the 10% PCE mark for planar cage in the final perovskite structure.[35] Thus, complex-forming
perovskite solar cells was possible by choosing appropriate solvents can be used to “pre-stretch” the PbI2 lattice for the
solvents.[24] DMSO is frequently used as a solvent for fabrica- organic compounds that are incorporated into the structure via
tion of perovskite films and has been described as leading a molecule exchange reaction.[36] An example of this is depicted
to the formation of an amorphous PbI2 layer in the first step for DMSO and NMP in Figure 2b.
of sequential film deposition.[24] As a further advantage, the Subsequently, research into the ability of DMSO to form lead
authors of this publication stated that PbI2 has a higher solu- complexes has been extended to other solvents as well.[36,38,39]
bility in DMSO (2 m) as compared to, e.g., DMF (1 m), due to So-called polar aprotic solvents, implying solvents with func-
a strong coordination at a ratio of 1:2. This is apparent from tional groups containing oxygen (O), sulphur (S), or nitrogen
the PbO bond length of 2.38(6) Å in DMSO as opposed to (N), can act as Lewis bases in the vicinity of the Lewis acid PbI2.
2.431(4) Å in DMF.[31,32] The strong complex bond between They can be classified into O-donors, like DMSO, NMP, or

Adv. Energy Mater. 2017, 7, 1700264 1700264  (3 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

Figure 2.  Various crystallization routes suggested in the literature for different deposition methods. a) Principle of crystallization through complex for-
mation. Reproduced with permission.[24] Copright 2014, Royal Society of Chemistry b) PbI2 structure intercalated with complex-forming solvents DMSO
and NMP. Reproduced with permission.[36] Copyright 2016, John Wiley and Sons, c) timescales of one-step methods employing different lead sources.
Reproduced with permission.[27] Copyright 2015, American Chemical Society, d) the influence of chlorine showing an intermediate phase. Reproduced
with permission.[30] Copyright 2016, Royal Society of Chemistry, e) mechanistic reorganization in a two-step method as a consequence of confined
growth close to the substrate and Ostwald ripening on the film surface. Reproduced with permission.[13] Copyright 2015, American Chemical Society,
f) in situ conversion of the precursor preserving crystal orientation and dissolution-recrystallization mechanism. Reproduced with permission.[37]
Copyright 2016, John Wiley and Sons.

GBL, S-donors, like thiourea or thioacetamide, and N-donors, orthogonal solvents are used to induce fast crystallization with
like pyridine or aniline. However, it is crucial to take the stoi- a high density of nucleation sites.[21] The result of the retarded
chiometry of the reaction into account and to select solvents crystallization is a film of highly oriented perovskite crystals
rationally. Foley et al. used computational screening to select that is formed directly from the stoichiometric precursor solu-
tetrahydrothiophene-1-oxide (THTO) as a solvent having a tion, as proven by in situ grazing incidence wide angle X-ray
stronger solvation efficacy than the commonly used solvents scattering (GIWAXS). Although the photovoltaic performance
as a result of its sulfoxide (SO) functional group that has a of these films was moderate, this approach nevertheless shows
coordination ratio of 3:1 with lead.[38] By suppressing the for- the ability to tune the crystal orientation and film morphology
mation of intermediate structures, crystallization is retarded by rational selection of solvent media.[38] Examples from lit-
significantly as the density of nucleation sites is reduced. Thus, erature show that the highest efficiencies so far were achieved
this strategy runs counter to the antisolvent approach, where with mixed solvent approaches, generally employing DMF and

Adv. Energy Mater. 2017, 7, 1700264 1700264  (4 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

DMSO, where the stoichiometry of the precursor solution was authors maintained a solvent-rich environment during the
precisely controlled.[4,40] The latest records that have surpassed crystallization process that retarded solvent evaporation and
the 20% PCE barrier followed an approach using a mixture of extended the crystallization time. This resulted in dense layers
solvents (4:1 DMF:DMSO, which would result in a molar ratio with large grains of up to 10 µm in size, which are also per-
between DMSO:PbI2 of ≈2:1) and the CB antisolvent method, fectly oriented with the z-axis perpendicular to the substrate.
while mixing three to four different cations and two halides.[4,41] A high degree of crystal orientation can lead to enhanced
The impact of solvents on the perovskite film morphology was solar cell performance when directly compared to samples
further investigated by post-deposition solvent annealing.[23] exhibiting a lower degree of orientation.[47–49] For example,
Giesbrecht et al. correlated the photocurrent to the degree of
order in the system and showed internal quantum efficiencies
2.1.2. Crystallization from Lead-Acetate Precursors approaching unity for the perfectly oriented samples. These
findings are supported by the work of Cho et al. who showed
Besides the interactions of the solvents with lead, the lead(II) anisotropic charge transport in MAPbI3 films with a different
salt composition also has a strong impact on the perovskite crys- degree of oriented crystallites in the perovskite film.[50]
tallization. Computational studies by Buin et al. suggested that
iodine-rich conditions result in high trap densities. In contrast,
iodine-poor conditions obtained through the use of lead acetate 2.2. Two-Step (Sequential Deposition)
as the precursor result in fewer defects and a longer charge dif-
fusion length.[42] In a comparative study with other lead precur- Burschka et al.[10] adapted a sequential deposition method intro-
sors, Aldibaja et al. showed that films derived from lead acetate duced by Liang et al.[51] for the preparation of perovskite solar
resulted in the most homogeneous perovskite films, and a cells. The authors achieved PCEs of 15% by spin-coating a PbI2
photovoltaic performance comparable to the state-of-the-art.[43] solution onto mesoporous TiO2 and subsequent immersion
Zhang et al. further optimized the use of lead acetate as a pre- in a MAI-containing solution. This method was transferred
cursor and showed that ultrasmooth and pinhole-free perovs- to the planar solar cell architecture by Liu et al.[52] Subse-
kite films could be obtained with this method. With the aid of quently, Docampo et al. adopted the method and used a mixed
in situ GIWAXS, it was later found that the activation energies MAI:MACl solution for conversion.[11] This resulted in a PCE
for crystallization are dependent on the lead salt, and that they of 15.7% and showed the great potential of the much simpler,
influence the optimum annealing time and temperature.[19,27] planar design. Soon thereafter, many possibilities were explored
In addition, the authors also determined that the primary in order to convert a predeposited PbI2 film into perovskite,
step in perovskite formation is the removal of excess organic e.g., by spin-coating a more concentrated MAI solution on top
salt from the precursor,[27] while the fast reaction mechanism, of a PbI2 film[25] or by exposing the PbI2 film to an MAI vapor
compared to precursors such as PbI2 and PbCl2, leads to homo- atmosphere (VASP).[26] Other groups modified the PbI2 film
geneous nucleation and thus results in the observed smooth itself, e.g., by solvent annealing[53] or by replacing DMF with
perovskite film morphology.[27] How the reaction times depend DMSO as the solvent.[24] In the latter case, a complex between
on the employed lead sources is shown in Figure 2c. Here, the lead and DMSO molecules is formed, preventing the PbI2 from
most significant aspect is the low activation barrier for perovs- crystallizing, and thereby slowing down the transformation to
kite crystallization with the lead acetate precursor, which results the perovskite phase during the subsequent immersion. The
in full crystallization while residues of the DMF solvent are still role of the complex was further discussed in section 2.1.1.
in the film.[27] Yang et al. proposed two mechanisms for the transformation
Other factors that affect the growth include the presence of of PbI2 to MAPbI3 for free-standing crystals that enabled them
water in the crystallization environment. Ling et al. suggested to tune their shape: By so-called in situ conversion, the original
that a controlled amount of water in the lead acetate precursor morphology of the PbI2 crystal is preserved while it is converted
helps the formation of an intermediate phase, which in turn to MAPbI3 via incorporation of MAI ions. In the dissolution–
leads to more uniform perovskite films. The authors found recrystallization mechanism, on the other hand, the PbI2 crystal
an optimum by using PbAc2 × 1.5H2O, which resulted in the dissolves and the perovskite recrystallizes in a thermodynami-
highest PCEs.[44] Several studies already showed that exposure cally favored shape. These two processes compete during crys-
of the films to ambient conditions can have a positive effect tallization, while in situ conversion is favored in environments
on perovskite film formation,[45] i.e., it was shown to trigger with a low MAI concentration.[54] Shortly after, this concept was
the crystallization[46] or partly dissolve the reactant species and transferred to thin films as well.[13,37,55]
accelerate mass transport within the film.[3] Schlipf et al. used GISAXS to gain quantitative information
In a recent study, Giesbrecht et al. used lead acetate as the on crystal-size distribution in perovskite thin films fabricated by
precursor for the crystallization of methylammonium lead a sequential deposition method.[11,13] They found that starting
bromide (MAPbBr3) perovskite films.[47] The authors reported from a film of crystalline PbI2, where the crystals are arranged
a new synthesis approach based on the control of the deposi- in a pancake-like stack, the MAI diffuses from the top down
tion environment while the film is still wet. In contrast to lead through the stack and begins the conversion into the perovskite
halide precursors, lead acetate allows for the crystallization phase. As the structure expands upon the inclusion of MAI,
of the perovskite when the film still contains solvent.[27] This growth is constrained in the lateral direction, leading to pre-
is mainly due to the much faster growth kinetics and easily dominant growth perpendicular to the substrate. Additionally,
removed organic byproducts, which drive the reaction. The lead iodide sheets below the top-most layer experience stronger

Adv. Energy Mater. 2017, 7, 1700264 1700264  (5 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

confinement effects and begin to crack, leaving smaller grains employed to achieve high-quality perovskite films. One of the
of perovskite. On the film surface, however, larger crystals are most successful early reports by Liu et al.[71] includes the growth
formed by Ostwald ripening in a dissolution–recrystallization of extremely uniform perovskite films by co-evaporation of the
process promoted by the higher MAI concentration.[13] This lead precursor and the organic salt. This approach is particu-
process is depicted in Figure 2e. Further evidence that both pro- larly interesting with a view on introducing perovskite devices
cesses are contributing to the morphological features of these to the market: Today’s semiconductor industry employs many
films was later given by Oesinghaus et al., where GIWAXS vapor deposition techniques, which could take advantage of the
measurements suggested that every process results in a unique more energy-efficient processing of perovskites. For instance,
crystal orientation. In particular, the in situ conversion process this could lead to the straightforward realization of tandem
is closely connected to the PbI2 morphology and results in crys- devices with conventional semiconductors (like silicon), or for
tals being tilted at an angle of about 55°, which corresponds energy level alignment by gradually controlling the doping level
to (202) planes being parallel to the substrate.[37] This phenom- in the perovskite. The process is also compatible with flexible
enon was reported in several other publications dealing with device applications, as demonstrated by Bolink and cow-orkers,
different fabrication methods.[37,56–59] In their study, Oesing- who showed the remarkable mechanical properties of sublimed
haus et al. further suggested that the presence of Cl– promotes perovskite films in flexible devices.[72,73] An initial setback to the
the dissolution of the PbI2 layer, leading to more crystals ori- process is that the perovskite crystallization during evaporation
ented in (002) direction (cf. Figure 2f). This might be due to the seems to be affected by the choice of substrate. In particular,
lower activation energy for the formation of methylammonium MAI seems to have a varying affinity to physisorb on certain
lead chloride (MAPbCl3).[27] substrates and might even decompose, which hinders the for-
mation of the perovskite phase in the first few nanometers.[74,75]
Nevertheless, this issue can be overcome by employing a
2.3. Role of Chloride Ions fullerene intermediate layer, allowing the fabrication of per-
ovskite solar cells with a high stabilized efficiency.[76] For more
A particularly intriguing crystallization process is achieved details on vapor-based techniques, we refer the reader to the
when incorporating chloride precursors into the precursor recently published review by Ono et al.[77]
mixture, which is beneficial for the optoelectronic properties of Another emerging research topic concerns lower-dimen-
the resulting perovskite film.[11,60–62] Yet the exact role chloride sional perovskites that form by substitution of the organic
actually plays has been rather puzzling: while earlier reports compound with bulkier molecules, generally employing
suggested doping effects[63] that could explain the longer charge longer alkylammonium chains. This approach is particularly
carrier diffusion lengths, finding chloride in the final perovs- interesting in terms of enhancing the stability of the struc-
kite films has been rather challenging.[11,62,64,65] ture against humidity, as these compounds generally exhibit
Chloride seems to have mainly an effect on the crystal growth, a higher resistance against the incorporation of water into the
as was suspected previously,[11,63] and it has since been shown that crystal. Initial results showed rather poor device efficiencies,
it is released together with excess MA during film annealing, or, mostly due to the preferential growth direction that results in
in general, that it can facilitate the release of excess organic com- 2D perovskite nanoplatelets aligned parallel to the substrate,
pounds.[62,66] Binek et al.[67] showed that MAPbCl3 crystallizes thus inhibiting charge transport.[78] A recent study, therefore,
directly after the deposition of the starting solution, thus acting as combined these 2D perovskites with 3D perovskites in a sand-
a template for the formation of the desired iodide-based perovskite wich approach, by converting the upmost layer of their more
via a slow halide exchange from the iodide-rich environment. They efficient 3D counterparts into lower-dimensional species, to
also showed that extending the time for the template formation form a multifunctional heterojunction, which exhibited a
resulted in more beneficial crystal morphologies, resulting in per- higher open-circuit voltage and PCE, and longer stability in a
ovskite films with enhanced optoelectronic properties. Additional humid atmosphere.[79] The reverse approach, converting a 2D
in situ studies following alternative strategies suggest that the perovskite into a three-dimensional form by interdiffusion of
presence of chloride in the precursor solution promotes the for- simple MA+ cations, also showed very promising results.[80]
mation of intermediate phases (cf. Figure 2d).[27,30,68] This, in turn, Currently, the highest efficiencies of lower dimensional per-
changes the kinetics of the reactions leading to the formation of ovskites have been claimed by Tsai et al. using a hot-casting
more uniform films. In the case of two-step deposition methods, method.[81]
chloride promotes the dissolution of the crystalline precursor pos- Recent efforts have been focussed on single crystal studies.
sibly by forming thermodynamically favored intermediates, which Although photovoltaic devices based on single crystals cannot
has an impact on the conversion mechanism and resulting film compete with the standards set by thin films so far, the mate-
morphologies.[37] When using a combination of Cl− additives with rials have proven themselves to be useful as photodetectors,
complex-forming solvents such as DMSO, the timescales, as well and are, furthermore, very interesting for fundamental mate-
as the pathways of crystallization, are highly tunable.[37,38,69,70] rials studies. For instance, the dependence of the absorption
onset on the crystal size is apparent by the red-shift observed for
larger single crystals.[82] The best results are currently achieved
2.4. Alternatives to Solution-Processed Thin Films by taking advantage of the inverse solubility effect observed for
perovskites in certain solvents, like GBL for MAPbI3 and DMF
While solution-based techniques have dominated the scene, for MAPbBr3.[18,83] Recently, very thin perovskite wafers could
a plethora of alternative deposition techniques can also be be realized by confined growth within a microreactor created

Adv. Energy Mater. 2017, 7, 1700264 1700264  (6 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

from two glass slides with appropriate spacers, paving the way itself. Initial results with this technique were reported by Bur-
for mass production of solution-processed wafers.[84] For fur- schka et al.[94] for mesoscopic solar cells with an average PCE
ther information on the growth control of perovskite films and of 12.0% and the highest PCE of 12.9%. A similar procedure,
single crystals by many different methods that could also be rel- with the addition of a pre-wetting step of the PbI2 layer in IPA
evant to industry, we refer the reader to the recently published prior to its immersion in the MAI solution, led to a significant
review by Chen et al.[85] increase in the PCE and its reproducibility.[10] A maximum PCE
of 15.0% and a certified PCE of 14.1% was achieved using the
two-step method, but data regarding average PCEs were not
3. Reproducibility of Perovskite Materials published. As noted before, film crystallization can inherently
be different on planar substrates. Docampo et al.[11] studied the
Although reproducibility is an important factor for the techno- influence of the addition of chloride additives in the immersion
logical realization of PSCs, most authors mention reproduc- solution as well as the temperature to optimize this process for
ibility only in passing. In order to determine the reproducibility planar substrates.[48] These results pointed towards an increase
and real device performance, device-to-device and batch-to- in device reproducibility at higher temperatures due to a
batch statistics are absolutely necessary.[86] Histograms of a decrease in the orientational disorder of the perovskite crystals
large number of devices (ideally 30+) with batch-to-batch repro- due to a dissolution–recrystallization mechanism, described in
ducibility, should be a requirement to prove the desired device Section 2.2. and 2.3. An additional aspect of their method is
reproducibility and narrow device fluctuation statistics. the spin-coating of PbI2 at an elevated temperature, which con-
The main challenge lies in processing high-quality perovs- trols the evaporation rate of the solvent (DMF) and leads to an
kite films along with crystallization at low temperatures, which optimized film porosity to accommodate the volume expansion
is different from most other inorganic and organic semiconduc- during the transformation to the perovskite phase.[37]
tors. This can lead to substantial differences in the film quality Jiang et al.[95] and Ko et al.[96] studied the effect of PbI2 film
and device performance, depending on which method and morphology on device performance and reproducibility and
temperature were used to crystallize the perovskite.[87–89] While found that it strongly affects these parameters. In order to
solar cells with record PCEs can be achieved without good con- overcome large deviations in film morphology and composi-
trol of the perovskite crystallization, this generally leads to films tion, these authors proposed an alternative protocol where
with large morphological deviations, leading to large PCE fluc- the MAI solution with high ion concentration was spin-coated
tuations in the resulting devices.[90] onto the PbI2 film. This, in turn, creates a nanocubic perovs-
In this part of the review, we will discuss the reports on kite morphology with a high degree of control on the crystal
reproducible PCE and solar cell parameters. It should be noted size, dependent on the concentration of the MAI solution and
that the reproducibility we are referring to is considered as a waiting time.[91] In Figure 3a–c, we show that a small change in
narrow scatter of PCE and solar cell parameters of PSCs pro- fabrication parameters can change the complete morphology of
cessed under the same conditions. As outlined in the previous perovskite films.
section, thin film perovskites can be prepared by a variety of Working with the complex-forming solvent DMSO, Wu
methods. SEM images of films prepared by different typical et al.[24] gained better control of the conversion step in the afore-
methods are shown in Figure 3, in detail: a) two-step sequen- mentioned dipping method as the intercalation of MA ions was
tial deposition with dipping, b) two-step sequential deposition slowed down (cf. Section 2.1.1.) As a result, the crystal size dis-
with a hot MAI solution containing 5% MACl, c) sequential tribution is very narrow, centered around 200 ± 30 nm; whereas
spin-coating of PbI2 and MAI, d) spin-coating a single mixture when the deposition is done using DMF, the crystal size distri-
precursor solution of both components, e) spin-coating a mixed bution ranges from 50 to 400 nm. The narrow crystal size distri-
halide single mixture precursor, f) antisolvent engineering, g) bution leads to very reproducible PCE values, with a maximum
fast deposition–crystallization, h) solvent–solvent extraction, i) efficiency of 14.2%, and a standard deviation of only 0.6%.
full vacuum deposition method, and j) MAI-vapor-assisted solu-
tion process, respectively. In the following, we compare these
methods regarding their capability to enable reproducible thin 3.2. One-Step Deposition
film morphologies and device parameters.
One-step methods seem to be favorable in terms of mini-
mizing the uncertainties in each processing step. In most one-
3.1. Two-Step Methods step techniques, the precursors of MAI and PbI2 or PbCl2 are
simply dissolved in DMF or DMSO in a specific ratio, i.e., stoi-
Although more versatile, in principle, two-step methods gen- chiometric or with an excess of one of the components,[97–101]
erally suffer from less reproducible devices, as there are many and are then spin-coated onto the substrate and annealed at
more parameters that have to be controlled. In terms of film a particular temperature and time to obtain crystalline layers
quality, the key parameters for two-step sequential deposition of perovskite (Figure 3d,e). As mentioned before, simple one-
and perovskite conversion by dipping are immersion time, tem- step methods often lead to inhomogeneous coverage of the
perature of the conversion solution, and precursor concentra- substrates.[14,102–104] In particular, Eperon et al. studied the
tion. In addition, the morphology not only depends upon the influence of the TiO2 underlayer and its impact on perovs-
immersion conditions of the PbI2 film, but it also depends on kite-crystal formation when employing a non-stoichiometric
small changes in the processing conditions of the PbI2 film mixture of MAI and PbCl2. They concluded that there is an

Adv. Energy Mater. 2017, 7, 1700264 1700264  (7 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

Figure 3.  Top-view SEM micrographs of perovskite thin films deposited by a) two-step sequential deposition. Reproduced with permission.[52] Copy-
right 2014, Nature Publishing Group, b) two-step sequential deposition with 60 °C hot solution of CH3NH3I with 5% CH3NH3Cl. Reproduced with
permission.[11] Copyright 204, John Wiley and Sons, c) sequential spin coating of PbI2 and MAI method. Reproduced with permission.[91] Copyright
2014, Nature Publishing Group d) coating a stoichiometric precursor solution. Reproduced with permission.[21] Copyright 2014, John Wiley and Sons,
e) mixed halide single precursor. Reproduced with permission.[3] Copyright 2014, Science Publishing Group f) antisolvent engineering of mixed halide.
Reproduced with permission.[12] Copyright 2014, Nature Publishing Group, g) antisolvent method with pure MAPbI3. Reproduced with permission.[21]
Copyright 2014, John Wiley and Sons, h) solvent–solvent extraction. Reproduced with permission.[92] Copyright 2015, Royal Society of Chemistry,
i) vacuum deposition method. Reproduced with permission.[71] Copyright 2013, Nature Publishing Group, and j) MAI-vapor assisted solution process.
Reproduced with permission.[93] Copyright 2016, John Wiley ans Sons.

unfavorable interaction between these layers, which leads to approach leads to both the highest efficiencies reported, as well
the growth of perovskite “pores”, largely determined by the as the narrowest device parameter distributions, as shown in
thickness of the perovskite film.[14,105,106] Needless to say, the Figure 4.
reproducibility of the resulting devices suffers as a result. Sev-
eral improvements to the device performance can be achieved
by introducing a templating step, as described above,[67] or via 3.3. Solvent–Solvent Extraction Method
flash-annealing of the films;[49] however, the overall reproduc-
ibility of this deposition method still lags behind other one- In the solvent–solvent extraction process, a mixture of a perovs-
step approaches. kite precursor solution is spin-coated onto a substrate, and the
A more successful strategy involves the fast deposition–crys- coated substrate is immediately immersed into a bath of another
tallization of the perovskite film via the addition of an antisol- (diethyl ether) solvent, in which the precursors and the final
vent during the spin-coating protocol. This leads to uniform product are insoluble. This results in efficient extraction of the
and smooth perovskite films (Figure 3f,g),[12,21] and thus more precursor solvent, and induces rapid crystallization of uniform,
reproducible devices. This approach really shines when com- ultra-smooth perovskite thin films over large areas (Figure 3h).
bined with the crystallization of an intermediate complex with a PSCs obtained by this method deliver PCEs up to 15.2%, and
solvent, typically DMSO.[2,22,98,107,108] In this case, device perfor- an average PCE of above 13.0%.[92] When the solvent-extracted
mance with a standard deviation below 1% can be achieved,[21,40] films were annealed at 150 °C in air for 15 min, the average PCE
while the PCE reaches over 21% for mixed perovskites.[22] This was improved to 17.0%, with the highest value of 17.8%.[110]

Adv. Energy Mater. 2017, 7, 1700264 1700264  (8 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

Figure 4.  PCE histograms of perovskite solar cells prepared by different methods (TSSD: two-step sequential deposition method,[109] MH: mixed halide
method,[3] SSC: sequential spin coating of PbI2 and MAI method,[91] ASE: antisolvent engineering,[12] FDC: fast deposition crystallization,[21] VASP: MAI-
vapor assisted solution processing)[93] (No. of devices used for histogram is given in bracket).

3.4. Vacuum Deposition Method and MAI-Vapor Assisted by evaporation of PbI2, followed by MAI spin-coating with mod-
Solution Process ification of TiO2/perovskite interface) exhibit PCE ≈16% with
fluctuations of PCE from 12 to 17.9%.[76] As a compromise, an
A complete vacuum deposition process is favorable for smooth, MAI- VASP was proposed to fabricate perovskite thin films,
uniform, pin-hole free perovskite formation. Liu et al. reported where the solution-processed PbI2 film was treated with MAI
planar heterojunction PSCs obtained by a vacuum-deposition vapor.[26,93,113,114] VASP produces films with well-defined grain
process, with an average PCE of 12.3% and highest PCE of structure, with grain sizes up to the microscale and small sur-
15.4%, and noted that the vapor-deposition process was better face roughness (Figure 3j). The as-deposited PbI2 films were
than the solution process in planar PSCs because of the smooth annealed in MAI vapor at 120, 150, or 165 °C in vacuum or
film formation in the former (Figure 3i).[71] Bolink and co- N2 atmosphere, to form the uniform and large-grain perovskite
authors have been working on entirely vacuum-evaporation- films.
based methods for complete solar cells, and they could improve Recently, Gujar et al. have demonstrated relatively high-per-
the PCE from 12 to 19.1% with a narrow distribution of solar formance PSCs with excellent reproducibility using VASP.[93]
cell parameters.[73,111,112] However, in this process it is difficult Figure 5a shows the J–V characteristics of a set of 4 solar cells
to simultaneously control the deposition rate of PbI2 or PbCl2 fabricated on one FTO substrate, illustrating that a set of four
and MAI, respectively. Chen et al. reported that most of the solar cells on one substrate exhibits high reproducibility with
devices fabricated with this method (perovskite layer deposited all the cells having the same solar cell parameters, leading to

Figure 5.  MAI-Vapor assisted solution processed perovskite device performance, a) J–V curves of four different devices on the same substrate, b) his-
togram of highest PCEs from four different batches, c–f) histograms of the solar cell parameters JSC, VOC, FF, and PCE for the 35 devices from four
batches. Reproduced with permission.[93] Copyright 2016, John Wiley & Sons, Inc.

Adv. Energy Mater. 2017, 7, 1700264 1700264  (9 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

a constant PCE of 13.8%. To verify the batch-to-batch repro- scale of hysteresis is in the range of ≈10 to 100 s,[117,118] which
ducibility, keeping all the fabrication conditions constant, is in great contrast to the typical charge generation/recombi-
four different batches of 36 devices were prepared, and solar nation processes (≈ns) in PSCs.[119] d) The hysteresis strongly
cell performance was measured. A histogram of 35 working depends on the external scanning parameters,[120–122] such
devices out of 36 devices (yield = 97%) shows a very small as scan rate, amplitude of external electrical field, scanning
fluctuation in the solar cell parameters. Figure 5b reveals that direction, pre-scanning conditions, etc. During the last years,
the four different batches with the same device-preparation different mechanisms have been proposed to be responsible
conditions show a negligible variation for the highest PCE including 1) ferroelectricity, 2) charge trapping/detrapping, and
value. 3) ionic migration. Hereby, ionic migration seems to account
for the majority of the effects, as shown by the latest studies.

4. Hysteresis
4.2. Ferroelectricity
Despite the tremendous progress in crystallization and pro-
cessing, the challenge of the anomalous hysteretic current– Ferroelectricity relies on the hysteretic switching of ferroelec-
voltage (J–V) characteristic in PSCs, as shown in Figure 6, still tric domains in an electric field. Even though the crystal struc-
needs to be resolved. Hysteretic J–V characteristics, noted as ture of MAPbI3 belongs to the centrosymmetric tetragonal
hysteresis in this review, imply that the photocurrent response space group I4/mcm, the reorientation of the organic groups
exhibits a discrepancy between two sweeping directions and the distortion of the PbI6 cages can result in a polariza-
(from short-circuit towards open-circuit, and vice versa). In tion, which is one of the prerequisites for any ferroelectric
this section, we focus on examining the microscopic models behavior.[123,124] A second criterion is the ability to switch this
behind hysteresis, and present some of the approaches devel- polarization by an external field. Several groups, such as Chen
oped to suppress hysteresis and the potential applications in et al.,[125] Coll et al.[126] and Kutes et al.,[127] have demonstrated
photovoltaics. switching of spontaneous polarization, that is a piezoresponse
hysteresis loop in both amplitude and phase by using piezo-
electric force microscopy (PFM). Wei et al.[128] also presented
4.1. Origin of Hysteresis ferroelectric properties by employing traditional E–P measure-
ments on the basis of a classical Sawyer–Tower circuitry. Even
Recently Snaith et al.[15] compiled several features related to though ferroelectricity has been used to model and explain
the hysteretic behavior: a) hysteresis is dominantly associated hysteresis,[124,125,128,129] there are strong arguments that view
with the perovskite material itself. b) Selective contact mate- this interpretation to be quite controversial. These encompass
rials,[76,116] i.e., hole and electron transport layers (TLs), play i) overall effects that are too small,[128] ii) the fast decay of the
crucial roles in this behavior, including the material and the polarized switching behavior,[127] and iii) disagreement between
morphology (mesoporous or planar) of the TLs. c) The time the timescale of polarization switching and the one observed in
J–V curve hysteresis. Based on first principle simulations (e.g.,
DFT), the time interval for the polarization switching was in the
nanosecond region.[130] Leguy et al.[131] estimated the expected
time interval within a millisecond range, and noted that ferro-
electric materials require an electric field (larger than the coer-
cive field) to switch ferroelectric domains, which contradicts the
observations in step-wise J–V curve measurements, exhibiting
always a transition-like behavior.[132] Hence, although the con-
tribution of ferroelectricity to hysteresis is still under debate,[133]
the above discussion suggests that ferroelectric polarization
may not play a dominant role in the J–V hysteretic behavior. A
recent study employing PFM found periodic twin domains that
point towards the direction of ferroelasticity and would require
a noncentrosymmetric polar space group, I4 cm, that is also
often assigned to MAPbI3 in the literature.[134]

4.3. Charge Trapping/Detrapping


Figure 6. A typical J–V curve hysteresis behavior in a perovskite solar
cell, with a scanning speed of ≈0.9 V s−1. The arrows exhibit the direc- The possibility for low-temperature solution-processing of
tions of voltage scanning. Inset indicates the kink in the J–V curve. The perovskite films inevitably involves the trapping of a certain
dashed red line is the extension of the J–V curve below ≈0.75 V based on
amount of defects, which potentially impact the charge sepa-
a standard single-junction diode equation, and the blue ellipse indicates
the horizontal shift of the curves near the flat-band condition. This shift ration/recombination and charge transport.[135] These defects
will be discussed in the following part. Reproduced with permission.[115] can be 1) defects located energetically at the CB/VB band
Copyright 2016, John Wiley & Sons, Inc. edge following an exponential decrease distribution, and

Adv. Energy Mater. 2017, 7, 1700264 1700264  (10 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

2) deep-level defects within the bandgap following a Gaussian


distribution.[136] For the first type, the evidence suggests that
the energetic width of the defect states near the band edge, i.e.,
Urbach energy, ranges between 20 and 40 meV, depending on
the fabrication methods.[137] This value is much smaller than
in many typical inorganic semiconductor materials, e.g., Si,
copper indium gallium (di)selenide (CIGS), CdS, etc.[138] For
the second type, these defects lay deep within the bandgap and
can serve as potential trap sites for the charge carriers, being
responsible for non-radiative (Shockley–Read–Hall) recombi-
nation.[139,140] Filling and releasing charges at these trap sites
could modulate the charge transport of photogenerated car-
riers, which ultimately leads to hysteretic behavior.
Figure 7 depicts the possible defect states that are prevalent
in these perovskite films, including vacancies, antisites, and
interstitials.[141] Atomistic simulations reveal their respective
energetic distribution. With the help of spectroscopic charac-
terization techniques, it could be shown that these traps are Figure 7.  Energy levels associated with the defect states in MAPbI3 per-
dominantly accumulated at the interfaces or surface, where ovskites corresponding to neutral and charged vacancies (VPb, VI, VMA),
the periodic crystalline structure is most susceptible to being neutral and charged interstitials (Pbi, Ii, MAi), and neutral and charged
deformed.[142–144] Temperature-dependent admittance spectros- states associated with antisites (PbI and IPb). Reproduced with permis-
copy enables one to further identify the energetic distribution sion.[42] Copyright 2014, American Chemical Society.
of these defects in the bandgap.[145,146] Furthermore, calcula-
tions showed that shallow level defects would only require low which is much shorter than the typical time for the hysteretic
formation energies, whereas deep-level defects require higher behavior.[159] The slow processes, such as the giant switchable
formation energies. The latter would be responsible for nonra- photovoltaic effect and the very slow (seconds range) increase
diative recombination acting as Shockley-Read-Hall nonradia- in photocurrent after poling, cannot be explained by a charge
tive recombination centers.[147] This explains the origin of the trapping mechanism alone.[131] Van Reenen et al.[157] recently
excellent defect tolerance of these perovskite materials. proposed that rather than having a single origin, the hyster-
Furthermore, methods such as Kelvin probe force micro­ esis has to be interpreted in terms of a combination of factors,
scopy (KPFM)[148,149] and electron-beam-induced current (EBIC) including 1) charge trapping/detrapping and 2) the ionic migra-
measurements[150,151] allowed for the measurement of the tion, which will be discussed in the following section.
potential and current distribution within a device, providing
more direct evidence for a charge trapping and detrapping
process. In these experiments, the charge-carrier-density dis- 4.4. Ion Migration
tribution across the cross-section of the device was measured,
allowing one to track the evolution of accumulated charges Recently, a series of studies have presented direct and indirect
after illumination. Illumination leads to an accumulation of evidence that points to ionic migration as the dominant factor
hole carriers at the interface between the perovskite and hole in the origin of hysteresis. Examples are the estimation of
transporting layer (HTL), which leads to a reduction of the respective activation energies, the observation of band-bending
effective charge extraction through this interface and an unbal- effects, and capacitance measurements, only to name a few.
anced charge transport. Mobile ions not only influence the hysteresis in J–V curves[160]
Of course, the question arises if those trap states could be but also the emission properties of the perovskite (e.g., pho-
filled and passivated to avoid charge trapping. By employing toluminescence, electroluminescence, blinking), they induce
luminescence characterization, Motti et al.[152] demonstrated capacitive effects,[161–165] and affect transient measurements,
that an exposure to small amounts of oxygen led to a significant respectively.[166]
decrease of trap state density in the perovskite material. Noel In general, an external electric field drives ions towards
et al.[153] and Dane et al.[154] used Lewis bases such as pyridine the opposite interfaces (perovskite/ETL and perovskite/HTL,
to treat the perovskite films in order to reduce recombination respectively). These accumulated ions result in both a change
centers, enhancing the photoluminescence intensity and life- of the internal field and a concomitant modulation of interfacial
time. In addition, it has been widely observed that [6,6]-phenyl- barriers, which ultimately results in the hysteretic behavior.[121]
C61-butyric acid methyl ester (PCBM) molecules,[155,156] either This model can be simply depicted as an interfacial mecha-
as an interlayer or mixed as bulk heterojunction structure, nism, as shown in Figure 8.
are able to passivate defects and alleviate or even eliminate
hysteresis.[76]
Although the defect passivation process significantly 4.4.1. Migrating Species
enhances the device performance and decreases the hysteresis,
it is found that the time interval for charge trapping and detrap- First, it is important to understand which species are actually
ping ranges within milliseconds,[157] or even nanoseconds,[158] mobile and can drift within an electric field in MAPbI3 PSCs.

Adv. Energy Mater. 2017, 7, 1700264 1700264  (11 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

experiment, MAPbI3 first decomposed into PbI2, which could


be driven by the external field. In this context Li et al.[115] carried
out X-ray photoemission spectroscopy (XPS) measurements
Figure 8.  Schematic diagram of the ionic migration under external elec- of long-term biased perovskite films with laterally-configured
trical fields in a PSC. Positive and negative ions drift towards opposite electrodes. After 30 min of applying a voltage, the perovskite
electrodes and accumulate at these interfaces. Reproduced with permis- film was measured by XPS, which quantitatively characterized
sion.[167] Copyright 2015, The Royal Society of Chemistry. the elemental distribution between the two lateral electrodes.
As shown in Figure 9f, it was demonstrated that at the positive
There are several candidates for the migrating ions: a) proton, electrode the I/Pb ratio was 5.57, whereas the ratio was only
or hydrogen ions, b) methylammonium, or MA+ ions, and 2.59 at the ground electrode. This result showed the significant
c) iodide ions.[168,169] In detail, these ions originate from defect iodine redistribution, in which the ratio I/Pb was stoichiometric
states during low-temperature fabrication and include vacan- (≈3) in the pristine device. In addition, as shown in Figure 9a,
cies (e.g., VMA, VPb, and VI) (i.e., Schottky defects), interstitial analyzing the binding energy of Pb in the XPS signal revealed
defects (e.g., MAi, Pbi, Ii), i.e., Frenkel defects, and anti-site the decomposition of MAPbI3 at the negative electrode, which
substitutions (e.g., MAI, PbI),[167,170] as previously mentioned in was consistent with the observed degradation characterized by
Figure 7. Note that due to the high activation energy (explained Xiao et al.[173]
in the following part),[167,171] migration of Pb2+ ions can be
excluded in this discussion.
For a) hydrogen ions, on the basis of first principle calcula- 4.4.2. Ion Migration Process
tions (DFT method), Egger et al.[172] proposed that hydrogen-
ion diffusion is enabled through the collective displacement of The detailed migration process of ions in PSCs can be theo-
iodide ions. However, it is rather difficult to maintain a high retically treated using dynamics of defect states. Due to the
concentration of H+ in PSCs due to the weak acidity of methyl- difficulty of directly characterizing the ionic movement,
ammonium.[169] Hence, even though H+ ions might exist, they most of the description is on the basis of theoretical DFT
should rather play a minor role compared with other defects. calculations.[130,171,177]
The migration of b) MA+ ions was proposed to be one of Herein, the ionic transport is described as a hopping mecha-
the reasons for the observed giant field-switchable photovoltaic nism, i.e., jumping between neighboring sites.[178] As shown in
effect in PSCs.[173] Employing the recently invented technique Figure 10a,b, i) MA+ ions migrate into a nearby vacant cage,
of photothermal induced resonance microscopy (PTIR), Yuan ii) Pb2+ ions migrate via the diagonal direction, whilst iii) iodide
et al.[174] observed a MA+ ion concentration contrast between ions move along an octahedron in the Pb–I plane.[171] The
before and after poling the device. This redistribution of MA+ migration of these ions requires an energy to open the PbX3
ions induced by an external electric field clearly indicated the framework, which acts as a barrier for the ionic migration. This
migration of MA+ ions and Azpiroz et al.[167] proposed the same barrier results in an activation energy, Ea. Detailed DFT calcu-
on the basis of computational studies. However, the extent of lations have been provided by Mosconi et al.[177] and Meloni
the contribution of MA+ ions to the hysteresis is still under et al.[130] The latter calculated the migrating pathways for each
debate. Van Reenen et al.[157] estimated the diffusion coefficient defect, including iodide vacancies, MA vacancies, Pb vacancies,
of mobile ions, without identifying the species with around and iodide interstitials, as shown in Figure 10c–e, respectively.
2 × 10−11 cm2 s−1. Eames et al.[171] estimated the diffusion of I− The migration pathways of individual ions/defects are con-
ions with a coefficient of 10−12 cm2 s−1, whereas their estimated sistent with the ones proposed by Eames et al.[171]
value for MA+ ions was only 10−16 cm2 s−1,[171] which was four Apart from hopping through point defects (Schottky and
to five orders of magnitude lower than the diffusion coefficient Frenkel defects) within the crystal lattice,[178] as shown in
of I− ions. These findings suggest that iodide ions play a more Figure 10f,g, a range of other possible ion migration chan-
important role in hysteresis effects compared to MA+ ions.[171] nels can exist. Yuan et al. suggest local lattice distortion,
Recent evidence consistently suggests that c) the migration induced through charges or additional impurity atoms, light-
of iodide ions plays the major role in the hysteretic behavior of induced softening, and piezoelectric effects, as shown in
PSCs.[121] Xiao et al.[173] proposed that positively charged iodide Figure 10h–l.[168] Furthermore, grain boundaries (GB) that pos-
vacancies, VI, were responsible for the n-type doping in PSC sess a large density of defects may play an even more important
under an external electrical field. Furthermore, long-term elec- role, illustrated in Figure 10k. The influence of GBs has been
trical biasing in devices with lateral electrodes led to an obvious demonstrated by different groups. For example, Xiao et al.[173]
degradation at the anode side. To elucidate the species of ions, found that, compared with large crystal size devices, films com-
Yang et al.[175] conducted long-term high temperature biasing posed of smaller size grains can be more easily switched under
experiments using a device structure of Pb/MAPbI3/AgI/ external electric fields. By carrying out conducting atomic force
Ag. The formation of PbI2 at the Pb electrode could be char- microscopy (c-AFM), and scanning electron microscopy (SEM),
acterized by energy dispersive X-ray spectroscopy (EDX) and Shao et al.[179] concluded that GBs where these defects resided
X-ray diffraction (XRD), indicating that migrated iodide ions were the fast moving channel for ion migration. Moreover, Yun
caused this observed reaction. Similarly, by using X-ray diffrac- et al.[180] conducted Kelvin probe force microscopy (c-KFM) to
tion (XRD) and energy-dispersive X-ray spectroscopy (EDX), reveal that ion migration near GBs was much faster than inside
Yuan et al.[176] directly observed the migration of iodide ions at the grains. At the GB, ions exhibited higher diffusivity and
higher temperatures (330 K) under an electric field. During the more effective charge separation. Xing et al.[181] also observed

Adv. Energy Mater. 2017, 7, 1700264 1700264  (12 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

Figure 9.  XPS results of a CH3NH3PbI3−xClx device with lateral electrodes, a) and b) are XPS spectra of Pb and I elements near the ground electrode,
respectively, d) and e) are the XPS spectra of Pb and I elements near the positive electrode, respectively. c) Schematic illustration of the set-up for
the measurement of elemental distribution. The applied DC voltage is 1 V for 30 min. The lateral electrode distance is 200 µm. f) The ratio of I/Pb
distribution across the channel after applying a bias between two lateral electrodes. The dashed line shows the stoichiometry of perovskite, I/Pb ≈ 3.
Reproduced with permission.[115] Copyright 2016, John Wiley & Sons, Inc.

that devices with larger crystal size exhibited a higher activation importance of GBs, further studies are expected to be dedicated
energy, that is, ions had to overcome higher barriers to migrate. towards their detailed investigation.
MacDonald et al.[182] revealed that GBs demonstrated a depth- Furthermore, ionic migration/transport can also take place
dependent resistivity, and that this inhomogeneity was one of within the bulk of solids. This phenomenon has been inten-
the limits for the improvement of PSC efficiency. In view of the sively investigated, especially in traditional perovskite materials

Figure 10.  Schematic diagram for ionic migration on microscopic and macroscopic scales, respectively. a) Iodide ions move along the octahedron edge,
Pb2+ ions move along the diagonal direction. b) MA+ ions move via neighboring vacancy sites. Reproduced with permission.[171] Copyright 2015, Nature
Publishing Group. c–e) Detailed view of migration events from blue (initial) to green (intermediate) to red (end) for c) I−, d) MA+, and e) Pb2+. Adapted
with permission.[130] Copyright 2016, Nature Publishing Group. Macroscopic schematic of ionic migration channels. f) Schottky defects or vacancies,
g) Frenkel or interstitial defects. Distortion of lattice due to h) accumulation of charges, i) impurities, j) light-illumination-induced lattice softening,
k) ion migration at the ground boundaries, and l) piezoelectric effect. Reproduced with permission.[168] Copyright 2016, American Chemical Society.

Adv. Energy Mater. 2017, 7, 1700264 1700264  (13 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

(ABO3).[178] The movement of ions is facilitated by a hopping 0.27 to 0.08 eV in polycrystalline MAPbI3, and from 1.05 to
mechanism in the atomic lattices via defect states (i.e., vacan- 0.47 eV in a single crystal, compared to dark conditions, specu-
cies, interstitials, etc.).[183] This hopping process to the neigh- lating that illumination may promote the decomposition of
boring sites requires energy, i.e., an activation energy, Ea, to MAPbI3 causing a larger number of vacancies and interstitials,
overcome the barriers. This hopping process usually depends making it easier for ions to move. In addition, we note that
strongly on temperature and is described as a thermally-acti- different contact materials, i.e., interfaces with the perovskite
vated point-defect movement. The migration can be empirically layer, can also affect the ionic migration rates and effective acti-
derived by a thermally activated process in form of an Arrhe- vation energies, as reported by Levine et al.[116]
nius relation resulting in a hopping rate, rm.[115,130,167,184]

 E  4.4.3. Modulation of the Energetic Environment at Interfaces


rm ∝ exp  − a  (1)
 kBT 
To fully understand the influence of ionic migration on the
where kB and T are Boltzmann constant and absolute device performance, an energy level diagram is a useful sche-
temperature, respectively. Since Ea strongly depends on matic approach to illustrate the detailed process, not only the
the type of ionic species, determining the value of Ea can field distribution inside, but also the interfacial barriers.[76,121]
enable us to identify which ions are moving. Several Several groups have made efforts regarding the construction of
groups have determined the involved activation energies, a band diagram, based on indirect evidence, such as transient
for example, Meloni et al.,[130] Eames et al.,[171] Li et al.,[115] photocurrent[171] and surface potential characterization.[180]
Yu et al.,[184] and Levine et al.[116] extracted Ea by fitting On the other hand, electroabsorption spectroscopy[185,186] has
temperature-dependent transient J–V curve measurements. proven to be a powerful approach to study the internal elec-
Yang et al.[175] obtained the Ea by carrying out temperature- tric field and to characterize the interfacial barrier modulation
dependent conductivity and dielectric constant measure- under working conditions.[187] Li et al.[115] found that there is
ments. Mosconi et al.[177] obtained the Ea using the fitted a shift of the built-in potential during the device scanning. In
slope of temperature-dependent PL dynamic measurements. addition, the shift of the potential ξ is equivalent to the change
The values for Ea range between 0.2 and 0.5 eV for iodide, of the open circuit voltage, Voc. This indicates that the modula-
and 0.5 and 0.8 eV for MA+ ions. The latest measurements tion of Voc can be attributed to a change in the internal field,
for iodide rather suggest values at the lower end of the caused by the ion migration.
quoted range. Theoretical calculations correlate these values Based on the above evidence, a band diagram modulated by
with the respective ionic species.[167] The calculations also the ionic migration is presented in Figure 11. To simplify the
show that much higher activation energies would have to be discussion, the work functions of both electrodes are the same.
expected for Pb ions (interstitials),[171] thus excluding these Without an external electric field, a classical metal–insulator–
species as migrating ions. The calculated value of H+ ion metal (MIM) junction can be assumed, and the voltage drops
activation energy, i.e., the barrier for the proton migration across the whole film. When a field is applied to the device,
of ≈0.29 eV[172] currently does not allow to rule out the con- ions, i.e., positive and negative defects are driven towards
tribution of hydrogen ions to the ionic migration in PSCs, opposite selective electrodes, with concentrations of the order
however, it is likely to play only a minor role due to their low of over 1017 cm−3 (Figure 11).[173,188] These ions accumulate at
concentration, as explained earlier. the selective contacts, causing: a) changes in the internal field
Interestingly, illumination of the perovskite can significantly of the device due to the screening of the external field, and b)
decrease the barrier for hopping and facilitate ion migration. modulation of the perovskite/electrode interfacial barriers. In
In this regard, Xing et al.[181] observed that under 0.25 sun the first case (Figure 11b), the electric field causes negative ions
illumination conditions, the activation energy decreased from to travel towards the HTL and positive species to accumulate at

Figure 11.  Schematic diagrams of band bending in perovskite solar cells under electrical fields. The red dashed arrows indicate the direction of the
effective work function movement. E presents the direction of the external electrical field. a) Before migration, the voltage drops uniformly across the
perovskite film. b,c) After the redistribution of ions, the internal field is shielded by the accumulated ions at the interfaces. In the meantime, interfacial
dipole layers emerge, influencing charge injection/extraction. Adapted with permission.[115] Copyright 2016, John Wiley & Sons, Inc.

Adv. Energy Mater. 2017, 7, 1700264 1700264  (14 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

the TiO2 interface, which alters the built-in field. In the second not mean that the device is free of ion migration.[195,196] As pre-
case (Figure 11c), the electric field drives the ions towards the viously discussed, the accumulation of ions/defects leads 1) to
other electrode. These charges are compensated by opposite a high local p- and n-doping, effectively enhancing/decreasing
charges within the electrodes, causing a change of the effective the built-in potential, and 2) to a change in the interfacial bar-
band level. Therefore, the effective built-in field will be modu- riers in the device. Hence, this significant reduction of hyster-
lated and either enhance (Figure 11b) or decrease (Figure 11c) esis is interpreted in terms of these two origins.
the original field. As a consequence, the separation of photo­ It is important to note that in perovskite solar cells that incor-
generated charges is either improved or deteriorated.[171] Apart porate a PCBM layer or a mixed perovskite/PCBM layer, PCBM
from the influence of an internal field, accumulated charges molecules will distribute within the bulk of the perovskite
also can modulate the interfacial barriers. The drifted ions film. Thermally activated PCBM molecules are easily driven
located at the interface between the perovskite and the selec- into the perovskite film by diffusion during thermal annealing
tive electrode result in a modification of the charge injection/ procedures. As shown in Figure 12a, PCBM absorbed at per-
extraction barriers. As a result, the effective charge-transfer pro- ovskite defect sites, i.e., iodide interstitials or iodide vacancies,
cess through these interfaces can be changed by the distribu- passivate those and decrease or even completely impede ion
tion of ions, which are driven by the external electric field.[189] migration. Xu et al.[197] have demonstrated the formation of
Li et al.[115] also mentioned the often-observed kinks illus- PCBM–halide-radical species in perovskite–PCBM mixed sys-
trated in Figure 6 in the J–V curve measurement. Several tems using UV–Vis absorption spectroscopy. Specifically, iodide
groups have observed a similar behavior,[15,117,128,190,191] and this ions form a strong bond with PCBM molecules by a direct elec-
phenomenon especially prevails when using fast scan rates. In tron transfer from the anions (iodide ions here) to PCBM or
the context of the ion-migration model, the authors found an C60, as shown in Figure 12c.[198] The reaction between PCBM
explanation for this kink in the J–V curve. When scanning from and I3 trimers associated with the PbI antisite defects leads to
a negative bias (short-circuit to open-circuit), the flat band con- the creation of anionic fullerene derivatives.[168] In this respect,
dition is reached at some point. During the flat band condition, the iodide ions/defects are immobilized by combining with the
where there is no electrical field inside, ion diffusion occurs, PCBM molecules in the bulk of the perovskite, thus impeding
driven through the concentration gradient (without ion drift) their further motion towards the perovskite/TL interfaces.
and rearranges the internal field, which leads to the shift of Voc. Following this argument, PCBM is able to passivate surface
The prevalence of this kink is another evidence for the ionic states and also to promote the charge transfer process through
migration during electric field sweeping. the perovskite/ETL interface.[76] Wojciechowski et al.[200] pro-
vided spectroscopic evidence that C60 buckyballs formed a self-
assembled monolayer and passivated or inhibited the traps
4.4.4. Suppression of Hysteresis states at the interface between TiO2 and the perovskite layer.
Xing et al.[199] observed an obvious reduction of the perovskite/
From a technological perspective, it is very desirable to avoid TiO2 interfacial barrier by inserting a PCBM layer on top of the
hysteresis. In particular, the finding that mobile/drifting ions TiO2 electrode, as shown in Figure 12e–g. Shao et al.[146] attrib-
play a major role makes it obvious that it would be very benefi- uted the elimination of hysteresis to the passivation of charge
cial to not only reduce the hysteresis, but also to increase the trap states in the bulk of the perovskite film during the thermal
long-term stability of photovoltaic devices. It seems promising annealing process. The interdiffused PCBM molecules pas-
to suppress the hysteresis via three approaches: 1) reducing the sivate traps at the surface and the GBs of the perovskite, and
concentration of defects/ions, 2) hindering the motion of these therefore enhance the charge transport and charge extraction.
ions, and 3) avoiding interfacial barriers and enhancing the A further indication along this line of reasoning is given by a
interfacial charge-transfer process. study where TiO2 was treated with an ionic liquid that appar-
It has been demonstrated that processing perovskite films on ently decreases trap state density and led to a certified PCE of
mesoporous TiO2 layers helps to create uniformly and densely- over 19%.[201] The suppressed hysteresis was associated with
packed large perovskite grains, leading to negligible hyster- both the reduction of ionic motion and the improvement of the
esis.[12] The careful exposure to additional humidity during the interfacial charge-transfer process.
growth of perovskite crystal films can enhance their quality
leading to a reduction in the hysteresis, too.[192] In this case, the
authors attributed this significantly decreased hysteresis to the 5. Long-Term Viability: Stability and Sustainability
realization of large-sized crystal grains that possess 1) much
fewer defects that are the source of mobile ions, 2) fewer grain In order for perovskite materials to be viable in long-term solu-
boundaries serving as fast channels for ionic migration, and tions, they must be stable under operational conditions, com-
3) better charge transfer at the interface. pete with incumbent technologies on price, and address their
It has been widely shown that the incorporation of PCBM toxicity concerns. In this section, we review recent contribu-
molecules, either used as a selective ETL[193] or intermixed tions towards fulfilling this rather tall order.
within the bulk of the active layer,[194] significantly reduces Photovoltaic technologies currently on the market, i.e., poly-Si
the hysteresis and leads to a higher device stability. We note, or CdTe, reach a remarkable raw device cost around $0.30 per peak
however, that even devices that appear to be hysteresis-free at Watt.[202] Current record-breaking perovskite solar cells employ
room temperature still show the effect at lower temperatures. expensive organic layers for optimal charge extraction, leading
Moreover, just because there is no observable hysteresis does to a projected cost just from the hole transporter, assuming

Adv. Energy Mater. 2017, 7, 1700264 1700264  (15 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

(Spiro-OMeTAD) requires several steps with


a poor overall yield (15 to 30% depending on
the synthesis route)[203] and is therefore often
replaced by synthetically more accessible
cores. Simple moieties such as phenyl,[206]
biphenyl,[207] carbazole,[207,208] triphe-
nylamine, [209] or triptycene[210] can be used to
reduce the cost, and spiro-based alternatives
such as spiro[fluorene-9,9′-xanthene],[211,212]
spiro[cyclopenta[2,1-b:3,4-b′]dithiophene-4,9′-
fluorene] and spiro[cyclopenta[2,1-b:3,4-b′]
dithiophene-4,9′-fluorene] have been intro-
duced.[213] Li et al. introduced heteroatoms in
the core by using the simple 3,4-ethylenedi-
oxythiophene (EDOT) moiety. Additionally,
they were able to perform their reaction in
one pot to further reduce the cost.[214] Many
of these materials have shown a photovoltaic
performance in perovskite-based devices that
is comparable to, or even better than, state-of-
the-art Spiro-OMeTAD.
Another route to reduce the cost is
by employing low-cost chemistries and
reducing tedious product purification. Petrus
et al. introduced Schiff base condensation
chemistry to prepare low-cost azomethine-
based HTMs[203] that has several advantages:
it can be performed under ambient condi-
tions, does not require metal catalysts, and
water is the only side-product (Scheme 1).
Additionally, this chemistry greatly simpli-
fies the product purification up to the point
where it might not be necessary at all.[215] In
particular, the authors have designed a new
Figure 12.  Proposed mechanism for the reduction of hysteresis by using PCBM. a) Passivation small molecule, termed “EDOT-OMeTPA”
process via PCBM in perovskite through interdiffusion. PCBM absorbs on the antisite defects based on previous azomethine work.[215–218]
located at the GB. b) Thermal admittance investigation of trap density of states before and after
As a hole-transporting layer in perovskite
the annealing process. Adapted with permission.[146] Copyright 2014, Nature Publishing Group.
c) Electron transfer occurs between the perovskite anions and PCBM, forming PCBM–halide- solar cells, this material achieved device effi-
radical. d) Theoretical calculations (DFT) indicate that charge transfer leads to the transition ciencies comparable to Spiro-OMeTAD. The
of trap states, from deep traps towards shallow traps. Adapted with permission.[197] Copyright good photovoltaic performance has been
2015, Nature Publishing Group. Schematic energy diagram of e) pure perovskite material, attributed to the fast injection of holes in
f) perovskite/PCBM interface layer, and g) perovskite/TiO2 interface layer. Reproduced with the material, which outcompetes second-
permission.[199] Copyright 2015, John Wiley & Sons, Inc.
order recombination.[219] Cost analysis for
the developed EDOT-OMeTPA molecule
20% power conversion efficiencies, of ≈$0.30 per peak Watt.[203] shows a total estimated material cost of only $10 per gram,
Clearly, for perovskite photovoltaics to enter the market, new low- which results in a negligible material cost contribution of
cost charge-extraction layers are necessary. The commonly used ≈$0.004 Wp−1 in the final device. Condensation chemistry was
organic materials are expensive as a result of their multistep syn- recently also used by Daskeviciene et al. to synthesize the low-
thesis that requires stringent reaction conditions and transition cost enamine-based HTM “V950” (Scheme 1), which resulted
metal catalysts. The side products and catalyst residues need to be in power conversion efficiencies close to 18%.[220] The costs of
removed via purification methods like column chromatography these materials are comparable to those of the highly conduc-
or sublimation, which are time-consuming and expensive.[204,205] tive polymer PEDOT (not to be confused with PEDOT:PSS),
This generally leads to commercial prices exceeding $200 g−1, and which was used as the hole transport material on top of the
thus reaching lower module target costs is challenging. perovskite.[221]
In the past years, several research groups have put their The combination of clean and low-cost chemistry, in combi-
focus on synthesizing low-cost hole-transporting mate- nation with cheap starting materials, has been shown to result
rials (HTMs) by reducing the number of synthetic steps. in a dramatic reduction in the cost contribution of the HTM.[203]
Especially, the synthesis of the Spiro core in 2,2′,7,7′- With these materials in hand, commercial applications of per-
tetrakis(N,N-di-p-methoxyphenylamine)-9,9′-spirobifluorene ovskite photovoltaics can be targeted.

Adv. Energy Mater. 2017, 7, 1700264 1700264  (16 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

Scheme 1.  Synthetic schemes for HTMs prepared using low-cost condensation chemistry.[203,220] For this synthesis, metal-free catalysts para-tolue-
nesolfonic acid (pTSA) and camphor sulfonic acid (CSA) were used, where water is the only side-product. The estimated material cost is displayed in
green.

5.1. Stability performance of “degraded” devices was recovered with a simple


drying step under a dry nitrogen stream for a few hours.
5.1.1. Moisture Moisture during processing can also affect the final perovs-
kite morphology and stability.[45,223] Petrus et al. have recently
Water interacts strongly with commonly used perovskite mate- investigated the influence of moisture on MAPbI3 films and
rials for solar cells, such as MAPbI3.[90,222] As the structure is solar cells derived from nonstoichiometric precursor mix-
soluble in water, the presence of humidity during film pro- tures.[101] They followed both the structural changes under
cessing can significantly influence the thin film morphology controlled air humidity (Figure 13) through in situ X-ray dif-
and can lead to improved solar cell performance.[3,46,223] How- fraction, and the electronic behavior of devices prepared from
ever, when fully-formed perovskite solar cells are exposed to air these films. The results showed that a small excess of PbI2 in
with a water-vapor content above 50% at room temperature, a the films improved the stability of the perovskite compared
loss in performance is incurred.[224] Understanding the effect of to stoichiometric samples. This was attributed to excess PbI2
moisture on perovskite materials is thus rather important with layers at the perovskite grain boundaries or to the termination
a view towards achieving stable performance over a decade. of the perovskite crystals with Pb and I, which is supported
Leguy et al. have identified two distinct degradation path- by theoretical calculations from Mosconi et al.[233] In contrast,
ways depending on whether water condensation occurs on the the films with an excess of MAI composed of smaller perovs-
surface of the film or not.[225] If the films are exposed to warm kite crystals showed increased electronic disorder and reduced
humid air with a water vapor content above 50% at room tem- device performance, owing to poor charge collection. Upon
perature, water condensation occurs, resulting in decomposi- exposure to moisture (similar to solvent annealing) followed
tion into HI (soluble in water), solid PbI2, and CH3NH2, either by dehydration, these films recrystallized to form larger, highly
released as gas or dissolved in water.[32,226,227] In this instance, oriented crystals, with fewer electronic defects and a remark-
the process is irreversible and losses to device performance are able improvement in photocurrent, photovoltaic efficiency, and
observed. stabilized power output.
When the films are exposed to cool humid air, and no water Thus, exposure to moisture can be beneficial to device perfor-
condensation occurs, then water is slowly incorporated into mance, but ultimately will degrade the device. For this reason,
the crystal to form a new crystal structure with isolated [PbI6]4− strategies that protect the perovskite material from moisture
octahedra.[228–232] The crystal structure formed depends on the but still achieve high efficiencies have been a well-explored
humidity content, temperature, and exposure time, with the research topic. An example was shown by Hu et al., who
initial formation of a monohydrated phase, followed by the for- employed moisture-resistant layered perovskite (LPK) struc-
mation of a dihydrated crystal phase. The water incorporation tures (Figure 14a).[79] Here, they demonstrated a perovskite/
process appears to be isotropic, meaning that it occurs more perovskite heterojunction solar cell obtained via a facile solu-
or less homogeneously throughout the sample, rather than tion-based cation infiltration process, to deposit layered perovs-
crystallization beginning at the outside and moving inwards, kite structures onto methylammonium lead iodide (MAPbI3)
suggesting that water is highly mobile within the lattice and/ films. This led to an enhancement of the open-circuit voltage
or along grain boundaries.[225] In this case, the process is and power conversion efficiency due to reduced recombination
reversible in films and in solar cells, where the photovoltaic losses, as well as improved moisture stability in the resulting

Adv. Energy Mater. 2017, 7, 1700264 1700264  (17 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

Figure 13.  SEM cross-section of photovoltaic devices with a MAPbI3 perovskite layer prepared with different precursor ratios. Top figures show SEM
cross-sections of freshly prepared devices and the figures below show the samples after exposure to 90% RH for 45 min before applying spiro-OMeTAD.
While the sample with an MAI excess recrystallizes, resulting in large crystals, the other two samples appear to degrade. The bottom images depict a
schematic representation of the morphology of the perovskite film before hydration with an inset showing the hypothesized crystal surface termination
for the perovskite grains at an atomic level. Reproduced with permission.[101] Copyright 2016, John Wiley & Sons, Inc.

photovoltaic devices, due to the hydrophobic nature of the LPK the major points of concern. In spite of the increased under-
layer.[234] A similar strategy was employed by Yang et al.,[235] standing and improvements made regarding the moisture
who functionalized the surface of the perovskite with hydro- stability, encapsulation of the devices is still key for long-term
phobic tertiary and quaternary alkyl ammonium (TAE) cations, stability.[241]
greatly enhancing the stability of perovskite films under high
relative humidity (R.H. 90 ± 5%) to over 30 days (Figure 14b).
Alternative approaches include employing other layers in the 5.1.2. Heat
solar cell to function as a barrier against moisture-induced deg-
radation. In particular, Spiro-OMeTAD, which is currently the Besides the impact of moisture, heating MAPbI3 also results
material used in state-of-the-art devices, forms layers that con- in structural changes. This material undergoes a reversible
tain many pinholes that poorly protect the perovskite against crystal-phase transition between tetragonal and cubic sym-
the ingress of water,[236] while additionally also promoting the metry in the temperature range of 54 to 57 °C,[228,242] which
formation of PbIOH as a degradation product.[237] As result, is in the range of common solar cell operating temperatures
several groups have focused on the synthesis of hydrophobic during summer.[243] Changes to the electronic band structure
HTMs that form pinhole-free films.[208,238,239] This generally can potentially reduce the photovoltaic performance to a large
results in a reasonable improvement of the stability of the extent, as high-efficiency solar cells employ fine-tuned charge-
device, with only minor losses of the performance. A particu- extraction layers. Additionally, cycling between the two crystal
larly interesting strategy has been developed by Habisreutinger phases during the day and night cycles is likely to lead to mate-
et al. who employed highly conductive carbon nanotubes encap- rial fatigue and shorter device lifetimes.
sulated in a polymer matrix as the charge extraction layer.[240] Several related perovskite structures, such as MAPbBr3 or
This approach leads to devices which can actually be placed caesium lead iodide (CsPbI3), are available in order to avoid
under a stream of water and suffer no degradation to the device phase transitions at solar cell operating temperatures. How-
performance as a result. ever, exchanging the iodide for bromide leads to a sub-optimal
Despite the development of these different strategies to pre- band gap of 2.3 eV,[47] and a correspondingly lower power con-
vent moisture-induced degradation of PSCs, this is still one of version efficiency. Similarly, the all-inorganic caesium-based

Adv. Energy Mater. 2017, 7, 1700264 1700264  (18 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

Figure 14.  a) Schematic illustration of the crystal structures of methylammonium lead iodide (CH3NH3PbI3) and a layered perovskite (LPK), forming
a moisture barrier. The inset shows the solar cells after 19 days storage at 75% RH, with clear degradation of the perovskite without the LPK. Adapted
with permission.[79] Copyright 2016, American Chemical Society. b) The possible interaction pathway between the perovskite and water using MA- and
TAE-terminated surfaces. Reproduced with permission.[235] Copyright 2016, Nature Publishing Group.

perovskite CsPbI3 has a large bandgap of 2.8 eV at room tem- incorporate a 3 or 4 cation mixture,[4] although their stability
perature, while the desired cubic phase, with a bandgap of still needs to be studied in detail.
1.7 eV, is only stable at high temperatures.[244] Currently, the Besides the perovskite, the thermal stability of the other
best option is to exchange the organic cation from MA to for- layers and interfaces in the device play an important role in
mamidinium (FA) in iodide-based perovskites.[245] This leads to the thermal stability. While inorganic blocking layers such as
a slight narrowing of the bandgap to 1.53 eV for the trigonal titanium dioxide (TiO2) and tin oxide (SnO2) are generally con-
α-phase, which is near the optimum for single junction solar sidered stable under operating conditions, organic blocking
cells, and a phase transition at 125 °C,[246] well beyond cell layers and their interfaces are much more sensitive. The
operating temperature. hole-transporting material Spiro-OMeTAD has been shown to
Unfortunately for formamidinium lead iodide (FAPbI3), crystallize over time, and this process is accelerated at higher
the thermodynamically most stable structure at room tem- temperatures, even though the glass transition temperature
perature is not a “3D structure”. This material crystallizes (125 °C) is well above the operating conditions.[236,253,254] Sev-
in a hexagonal structure, similar to the PbI2 lattice, with the eral novel hole transporting materials have been introduced
lead halide octahedra sharing edges, in the so-called δ-phase. in the last years, of which many show an improved stability
This disrupts charge transport and results in a wide bandgap of the device under operating conditions, generally achieved
of approximately 2.2 eV, clearly unfavourable for solar cell by either increasing the glass transition temperature or by
applications. However, depending on the synthesis tempera- cross-linking the HTM.[255–258] We further note that the loss
ture, FAPbI3 can also be crystallized in the ‘α-phase’, which in performance is often related to the use of additives like Li-
is a pseudocubic structure, i.e., the lead halide octahedra TFSI, tert-butylpyridine (tBP), and Co-complexes, necessary to
share the corner atoms, with trigonal symmetry. Thus, the oxidize the HTM and achieve high enough conductivities to
challenge is to find a synthesis route that leads to FAPbI3 ensure balanced charge extraction and maximum device per-
in the dark phase, which is stable at solar cell operating formance.[259–261] Therefore, several HTMs have been devel-
conditions. oped that do not require these additives in order to obtain
A popular strategy to overcome this issue is to mix the FA good performance. However, in general, devices without addi-
cation with other cations.[247–252] For instance, the exchange of tives show lower PCEs than their counterparts containing the
approximately 15% of the formamidinium cation with meth- additives.
ylammonium leads to the crystallization of a material with Surprisingly, recent studies have highlighted the point
the same crystal structure as FAPbI3.[247,249,252] Moreover, this that the metal contact can also influence the device degrada-
desired phase is stable between room temperature and 220 °C tion.[262,263] In particular, MAPbI3 has been shown to form HI
(Figure 15).[247] The resulting material exhibits the same lattice and CH3NH2 in an equilibrium reaction, even in encapsulated
parameters, photoluminescence emission peak wavelength, devices. The HI can then react with the silver top electrode,
and bandgap as those found for neat FAPbI3, meaning that which would further shift the degradation reaction of the per-
no significant lattice contraction occurred after the inclusion ovskite to the right.[262] This effect is enhanced in the absence
of the smaller MA cation.[247] Combining caesium, FA, and a of the hole transporting layer, or upon illumination/heating. In
mixture of halides, additionally allowed for bandgap tuning, contrast, gold is inert and in principle does not react with the
which is necessary for inclusion in tandem architectures with a perovskite. However, it has also been shown that gold can dif-
low-bandgap bottom cell,[248] while also enhancing the stability fuse through the hole-transporting layer under operating condi-
of the system. Finally, the best performing solar cells currently tions, and severely affect the device performance.[263]

Adv. Energy Mater. 2017, 7, 1700264 1700264  (19 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

Figure 15.  XRD pattern for a range of temperatures for neat FAPbI3 (left) and FAMAPbI3 (right), where 13% of the FA cations are exchanged with MA.
The two FAPbI3 crystal structures are depicted above the XRD patterns, and the characteristic reflection positions are marked with an arrow. Adapted
with permission.[247] Copyright 2015, American Chemical Society.

5.1.3. Mixed-Halide, Mixed-Cation Perovskites stability and possible degradation mechanisms of these multi-
cation, mixed-halide perovskites are essential for long-term
The intrinsic stability of the MAPbX3 (X = Cl, Br, I) perovskites stability and potential commercialization of these complex
has been studied with density functional theory (DFT) calcula- perovskites.
tions,[264,265] and experimentally by measuring the formation
enthalpies.[266] Despite the differences between these studies,
all report that MAPbI3 was found to be the least stable, while 5.2. Sustainability
MAPbCl3 is the most stable. Partially substituting the iodide
with bromide or chloride would, therefore, be expected to One final challenge that needs to be overcome is that of the
improve the thermodynamic stability of the perovskite com- toxic nature of the current state-of-the-art perovskite structures.
pared to the pure iodide material, while maintaining the good Although a recent study has shown that the concentration of
photovoltaic performance. However, MAPb(BrxI1−x)3 mixed- lead in the soil would not dramatically increase, even in the
halide perovskites undergo a reversible phase separation into worst-case scenario of complete dissolution of a solar panel and
iodide- and bromide-rich domains upon illumination.[267,268] dispersion in the ground,[270] it is essential to take care of the
The de-mixing is expected to result in a large increase of recom- lead. The most elegant route would be to substitute the toxic
bination centres, which limits the open-circuit voltage, leading lead with another less toxic metal. Looking at the periodic table,
to a rather poor photovoltaic performance, combined with the tin (Sn) based systems would, in principle, be the most viable
poor long-term stability of these materials. alternative. These materials have shown good charge mobili-
Recent reports have shown that the unwanted halide segre- ties and efficiencies,[271] although these values still trail behind
gation can be suppressed by preparing a mixed-cation, mixed- the state-of-the-art lead system.[272,273] Unfortunately, by moving
halide perovskite (CsyFA1−yPb(BrxIx−1)3).[248,269] These perovs- up in the periodic table, the stability of the 2+ oxidation state is
kites with a caesium content between 0.10 < y < 0.30 show a decreased, which causes the perovskite to degrade very quickly
high crystallinity and good optoelectronic properties, and are in the presence of oxygen and moisture. The high sensitivity of
especially interesting for tandem applications. Upon the addi- the material not only decreases the lifetime, but also reduces
tion of a third cation to these mixtures (Cs/MA/FA), both the reproducibility of the system. As an alternative, researchers
highly efficient solar cells and good long-term stability under have looked into the possibility of incorporating bismuth
continuous illumination were demonstrated. Although theoret- instead, which forms air-stable structures.[274–276] However, pre-
ical considerations would rule out its use, rubidium was added paring high-quality films has been challenging for these mate-
as a fourth cation for the preparation of solar cells, further rials, and therefore the resulting device performance is still
increasing the photovoltaic performance.[4] Although very stable lower than for their lead- and tin-based counterparts. Recently,
devices were reported, introducing several cations and halide silver/bismuth-based elpasolites (more commonly known as
anions could open the way for more degradation and demixing double perovskites) with the general formula A2AgBiX6, (A =
pathways. Once one of the cations or halides segregates, it MA, FA or Cs; X = Cl, Br, I, Figure 16) have been suggested as a
could also influence the stability of the rest of the perovs- promising lead-free and air-stable alternative.[277,278] Single crys-
kite,[269] making stability studies on these materials very com- tals of these double perovskites have shown good optoelectronic
plex. Therefore, further studies regarding the thermodynamic properties and long charge carrier lifetimes. However, the

Adv. Energy Mater. 2017, 7, 1700264 1700264  (20 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

to enhance the potential for the commerciali-


zation of the technology:
i. Perovskite crystal growth
ii. Device reproducibility
iii. Hysteresis
iv. Stability and sustainability

The first challenges are two sides of the


same coin; in order to avoid poor device repro-
ducibility and large scattering of the solar
cell parameters, we must learn about and
carefully control the crystallization process.
Figure 16.  Structure of the double perovskite Cs2AgBiBr6, time-resolved photoluminescence In particular, forming highly crystalline, pin-
and a fit for the PL decay. The inset shows a photo of a single crystal. Reproduced with permis- hole free, surface-smooth perovskite films is
sion.[279] Copyright 2016, American Chemical Society. required for highly efficient and reproducible
PSCs. These parameters depend on the initial
preparation of high-quality films has shown to be challenging, nucleation and growth mechanism, so it is important to develop
and therefore photovoltaic devices have not been published so new handles to control them. While several strategies have been
far.[279] As lead-free alternatives are relatively early in develop- put forward, i.e., one-step, two-step, etc., the more successful pro-
ment, the general research focus of the community has been cesses are based on slowing down the crystallization by inducing
on preventing exposure of the lead to the environment. the formation of an intermediate phase that is then converted
Encapsulation of the device prevents the contamination of into the final perovskite film. This strategy offers the greatest con-
the environment with lead and, additionally, it is necessary to trol over the crystallization process and leads to a smoother and
reduce the degradation of the perovskite.[280] However, encapsu- more homogeneous film, and thus more reproducible solar cells.
lation is not the holy grail, as devices may still degrade, as was Most notably, the technique leading to the highest-efficiency solar
recently shown by Han et al.[262] Furthermore, the lead will still cells employs the formation of an intermediate precursor–DMSO
be present at the end of the lifetime of the devices. In an effort complex, with a mixture of cations that is then converted into the
to ensure the viability of perovskite solar cell parks, Binek et al. perovskite phase upon heating on a hotplate.
have recently followed an alternative route, whereby the solar Secondly, we have covered the possible mechanisms for hyster-
cells are recycled by collecting and reusing the toxic lead in the etic behavior, including ferroelectricity, charge trapping/detrapping,
structure.[281] This is not only necessary to avoid lead waste, but
is also required, according to international electronic-waste-dis-
posal laws. The reported procedure allows for the removal of
every layer of the solar cells separately (Figure 17), which gives
the possibility to selectively isolate the different materials and,
in particular, allows for reusing the expensive conductive glass
substrate without a loss of performance. Recycled perovskite
solar cells with a performance above 16% have been demon-
strated in this manner.[282]
Furthermore, the energy payback time (EBT) of perovskite
solar cells needs to be assessed. Recycling of individual parts of
PSC, as shown by Binek et al., can serve as an effective strategy
to reduce energy-intensive steps that are especially needed for
the electrodes.[283] As mentioned before, new HTMs can be
developed, not only to reduce costs, but also to limit environ-
mental impact.[203] Finally, low temperature processed ETLs,
such as PCBM or alternative TiO2 fabrication techniques, can
be employed to reduce energy consumption in this step.[283–285]

6. Conclusions
Hybrid halide perovskite materials are a very promising pho- Figure 17.  Recycling procedure for perovskite solar cells. I) Removal of
tovoltaic technology for low-cost solar energy generation, and Au electrode. II) Removal of the HTM. III) Transformation of the perov-
skite into MAI and PbI2 and extraction of MAI in water. IV,V) Removal of
represent an extremely rich research platform to develop high-
PbI2 and TiO2. VI) Preparation of a new TiO2 film. VII) Formation of the
quality, solution-processable optoelectronic materials. In this perovskite film on recycled FTO from recycled PbI2. VIII) Preparation of
review, we have identified several key research areas and chal- the HTM layer. IX) Deposition of the Au top electrode. Reproduced with
lenges that need to be expanded upon and addressed, in order permission.[281] Copyright 2016, American Chemical Society.

Adv. Energy Mater. 2017, 7, 1700264 1700264  (21 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

as well as ionic migration under an external electric field. Recent [1] A. Kojima, K. Teshima, Y. Shirai, T. Miyasaka, J. Am. Chem. Soc.
experimental and theoretical results consistently suggest a com- 2009, 131, 6050.
bined role of ion migration and charge trapping/detrapping, [2] N. J. Jeon, J. H. Noh, W. S. Yang, Y. C. Kim, S. Ryu, J. Seo,
whereby, considering the timescales for each of the processes, S. I. Seok, Nature 2015, 517, 476.
[3] H. Zhou, Q. Chen, G. Li, S. Luo, T.-b. Song, H.-S. Duan, Z. Hong,
the migration of ions appears to be a major factor causing the
J. You, Y. Liu, Y. Yang, Science 2014, 345, 542.
hysteresis. We have presented, in detail, the different aspects of [4] M. Saliba, T. Matsui, K. Domanski, J.-Y. Seo, A. Ummadisingu,
ionic migration, i.e., the ionic species, migration channels, activa- S. M. Zakeeruddin, J.-P. Correa-Baena, W. R. Tress, A. Abate,
tion energies, and their influence on the band structure. We also A. Hagfeldt, M. Grätzel, Science 2016, 354, 206.
have presented the strategies developed to eliminate or alleviate [5] M. F. Aygüler, M. D. Weber, B. M. D. Puscher, D. D. Medina,
hysteresis, including increasing crystallite-domain size, improving P. Docampo, R. D. Costa, J. Phys. Chem. C 2015, 119, 12047.
crystal quality, and employing PCBM molecules for passivation. [6] Z.-K. Tan, R. S. Moghaddam, M. L. Lai, P. Docampo, R. Higler,
Finally, in order to achieve the ultimate goal of employing F. Deschler, M. Price, A. Sadhanala, L. M. Pazos, D. Credgington,
hybrid halide perovskites in commercial applications, the systems F. Hanusch, T. Bein, H. J. Snaith, R. H. Friend, Nat. Nano 2014, 9,
must fulfil the rather tall order of being stable, sustainable, and 687.
[7] J. Wang, N. Wang, Y. Jin, J. Si, Z.-K. Tan, H. Du, L. Cheng, X. Dai,
able to compete with existing technologies on price. Cost-wise,
S. Bai, H. He, Z. Ye, M. L. Lai, R. H. Friend, W. Huang, Adv. Mater.
the devices currently employ rather expensive small molecules as 2015, 27, 2311.
the HTM, which has driven research to push for materials with [8] A. Sadhanala, S. Ahmad, B. Zhao, N. Giesbrecht, P. M. Pearce,
a lower number of synthetic steps or alternative, simpler chemis- F. Deschler, R. L. Z. Hoye, K. C. Gödel, T. Bein, P. Docampo,
tries. Chief among these, condensation chemistry promises a par- S. E. Dutton, M. F. L. De Volder, R. H. Friend, Nano Lett. 2015, 9,
ticularly attractive approach, as it both lowers the number of neces- 6095.
sary synthetic steps and greatly simplifies the purification of the [9] M. M. Lee, J. Teuscher, T. Miyasaka, T. N. Murakami, H. J. Snaith,
compounds. Stability-wise, the recently-developed mixed-cation Science 2012, 338, 643.
systems have greatly enhanced this parameter, leading to dem- [10] J. Burschka, N. Pellet, S.-J. Moon, R. Humphry-Baker, P. Gao,
onstrated working lifetimes in the range of thousands of hours. M. K. Nazeeruddin, M. Grätzel, Nature 2013, 499, 316.
[11] P. Docampo, F. C. Hanusch, S. D. Stranks, M. Döblinger,
Approaches to encapsulate the perovskite material employing
J. M. Feckl, M. Ehrensperger, N. K. Minar, M. B. Johnston,
hydrophobic molecules as HTM, or directly incorporating a hydro- H. J. Snaith, T. Bein, Adv. Energy Mater. 2014, 4, 1400355.
phobic element in the device stack, have provided real progress [12] N. J. Jeon, J. H. Noh, Y. C. Kim, W. S. Yang, S. Ryu, S. I. Seok, Nat.
towards protecting the materials from moisture, while main- Mater. 2014, 13, 897.
taining the high efficiency of the devices. Lastly, addressing toxicity [13] J. Schlipf, P. Docampo, C. J. Schaffer, V. Körstgens,
concerns is of chief importance in a technology that aims to be L. Bießmann, F. Hanusch, N. Giesbrecht, S. Bernstorff, T. Bein,
widespread and possibly incorporated into buildings. While sev- P. Müller-Buschbaum, J. Phys. Chem. Lett. 2015, 6, 1265.
eral lead-free alternatives have been developed, they are neither [14] G. E. Eperon, V. M. Burlakov, P. Docampo, A. Goriely, H. J. Snaith,
as efficient nor as stable as the lead systems. This has pushed Adv. Funct. Mater. 2014, 24, 151.
researchers to develop recycling schemes to extract the lead from [15] H. J. Snaith, A. Abate, J. M. Ball, G. E. Eperon, T. Leijtens,
N. K. Noel, S. D. Stranks, J. T.-W. Wang, K. Wojciechowski,
the system at the end of the life cycle of the device.
W. Zhang, J. Phys. Chem. Lett. 2014, 5, 1511.
[16] F. Hanusch, M. Petrus, P. Docampo, in Unconventional Thin Film
Photovoltaics 2016, 32.
Acknowledgements [17] Y. Zhou, O. S. Game, S. Pang, N. P. Padture, J. Phys. Chem. Lett.
2015, 6, 4827.
M.L.P. and J.S. contributed equally to the work. The authors [18] M. I. Saidaminov, A. L. Abdelhady, B. Murali, E. Alarousu,
acknowledge funding from the Bavarian Collaborative Research V. M. Burlakov, W. Peng, I. Dursun, L. Wang, Y. He, G. Maculan,
Project “Solar Technologies Go Hybrid” (SolTech), the Excellence A. Goriely, T. Wu, O. F. Mohammed, O. M. Bakr, Nat. Commun.
Cluster “Nanosystems Initiative Munich” (NIM), and the Center for
2015, 6, 7586.
NanoScience (CeNS). The authors acknowledge funding from the
[19] W. Zhang, M. Saliba, D. T. Moore, S. K. Pathak, M. T. Horantner,
German Federal Ministry of Education and Research (BMBF) under the
T. Stergiopoulos, S. D. Stranks, G. E. Eperon, J. A. Alexander-Webber,
agreement number 03SF0516B.
A. Abate, A. Sadhanala, S. Yao, Y. Chen, R. H. Friend,
L. A. Estroff, U. Wiesner, H. J. Snaith, Nat. Commun. 2015, 6,
6142.
Conflict of Interest [20] P. Docampo, T. Bein, Acc. Chem. Res. 2016, 49, 339.
[21] M. Xiao, F. Huang, W. Huang, Y. Dkhissi, Y. Zhu, J. Etheridge,
The authors declare no conflict of interest. A. Gray-Weale, U. Bach, Y.-B. Cheng, L. Spiccia, Angew. Chem.
2014, 126, 10056.
[22] M. Saliba, T. Matsui, J.-Y. Seo, K. Domanski, J.-P. Correa-Baena,
Keywords M. K. Nazeeruddin, S. M. Zakeeruddin, W. Tress, A. Abate,
A. Hagfeldt, M. Grätzel, Energy Environ. Sci. 2016, 9, 1989.
crystallization, hybrid metal halide perovskites, hysteresis, solar cells, [23] J. Liu, C. Gao, X. He, Q. Ye, L. Ouyang, D. Zhuang, C. Liao, J. Mei,
stability W. Lau, ACS Appl. Mater. Interfaces 2015, 7, 24008.
[24] Y. Wu, A. Islam, X. Yang, C. Qin, J. Liu, K. Zhang, W. Peng, L. Han,
Received: January 27, 2017 Energy Environ. Sci. 2014, 7, 2934.
Revised: March 28, 2017 [25] Z. Xiao, C. Bi, Y. Shao, Q. Dong, Q. Wang, Y. Yuan, C. Wang,
Published online: June 28, 2017 Y. Gao, J. Huang, Energy Environ. Sci. 2014, 7, 2619.

Adv. Energy Mater. 2017, 7, 1700264 1700264  (22 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

[26] Q. Chen, H. Zhou, Z. Hong, S. Luo, H.-S. Duan, H.-H. Wang, [55] Y. Fu, F. Meng, M. B. Rowley, B. J. Thompson, M. J. Shearer,
Y. Liu, G. Li, Y. Yang, J. Am. Chem. Soc. 2013, 136, 622. D. Ma, R. J. Hamers, J. C. Wright, S. Jin, J. Am. Chem. Soc. 2015,
[27] D. T. Moore, H. Sai, K. W. Tan, D.-M. Smilgies, W. Zhang, 137, 5810.
H. J. Snaith, U. Wiesner, L. A. Estroff, J. Am. Chem. Soc. 2015, 137, [56] T. M. Brenner, Y. Rakita, Y. Orr, E. Klein, I. Feldman, M. Elbaum,
2350. D. Cahen, G. Hodes, Chem. Mater. 2016, 28, 6501.
[28] D. T. Moore, H. Sai, K. Wee Tan, L. A. Estroff, U. Wiesner, APL [57] B. Jeong, S. M. Cho, S. H. Cho, J. H. Lee, I. Hwang, S. K. Hwang,
Mater. 2014, 2, 081802. J. Cho, T.-W. Lee, C. Park, Phys. Status Solidi RRL 2016, 10, 381.
[29] R. Munir, A. D. Sheikh, M. Abdelsamie, H. Hu, L. Yu, K. Zhao, [58] R. Kang, J.-S. Yeo, H. J. Lee, S. Lee, M. Kang, N. Myoung, S.-Y. Yim,
T. Kim, O. E. Tall, R. Li, D. M. Smilgies, A. Amassian, Adv. Mater. S.-H. Oh, D.-Y. Kim, Nano Energy 2016, 27, 175.
2017, 29, 1604113. [59] T. Miyadera, Y. Shibata, T. Koganezawa, T. N. Murakami, T. Sugita,
[30] D. P. Nenon, J. A. Christians, L. M. Wheeler, J. L. Blackburn, N. Tanigaki, M. Chikamatsu, Nano Lett. 2015, 15, 5630.
E. M. Sanehira, B. Dou, M. L. Olsen, K. Zhu, J. J. Berry, J. M. Luther, [60] S. Colella, E. Mosconi, P. Fedeli, A. Listorti, F. Gazza, F. Orlandi,
Energy Environ. Sci. 2016, 9, 2072. P. Ferro, T. Besagni, A. Rizzo, G. Calestani, G. Gigli, F. d. Angelis,
[31] H. Miyamae, Y. Numahata, M. Nagata, Chem. Lett. 1980, 9, 663. R. Mosca, Chem. Mater. 2013, 25, 4613.
[32] A. Wakamiya, M. Endo, T. Sasamori, N. Tokitoh, Y. Ogomi, [61] V. D’Innocenzo, A. R. Srimath Kandada, M. d. Bastiani, M. gandini,
S. Hayase, Y. Murata, Chem. Lett. 2014, 43, 711. A. Petrozza, J. Am. Chem. Soc. 2014, 136, 17730.
[33] W. Fu, J. Yan, Z. Zhang, T. Ye, Y. Liu, J. Wu, J. Yao, C.-Z. Li, H. Li, [62] Q. Chen, H. Zhou, Y. Fang, A. Z. Stieg, T.-B. Song, H.-H. Wang,
H. Chen, Sol. Energy Mater. Sol. Cells 2016, 155, 331. X. Xu, Y. Liu, S. Lu, J. You, P. Sun, J. McKay, M. S. Goorsky, Y. Yang,
[34] J. S. Manser, B. Reid, P. V. Kamat, J. Phys. Chem. C 2015, 119, Nat. Commun. 2015, 6, 7269.
17065. [63] S. Colella, E. Mosconi, G. Pellegrino, A. Alberti, V. L. P. Guerra,
[35] S. J. Yoon, K. G. Stamplecoskie, P. V. Kamat, J. Phys. Chem. Lett. S. Masi, A. Listorti, A. Rizzo, G. G. Condorelli, F. d. Angelis,
2016, 7, 1368. G. Gigli, J. Phys. Chem. Lett. 2014, 5, 3532.
[36] Y. Jo, K. S. Oh, M. Kim, K.-H. Kim, H. Lee, C.-W. Lee, D. S. Kim, [64] M. I. Dar, N. Arora, P. Gao, S. Ahmad, M. Gratzel,
Adv. Mater. Interfaces 2016, 3, 1500768. M. K. Nazeeruddin, Nano Lett. 2014, 14, 6991.
[37] L. Oesinghaus, J. Schlipf, N. Giesbrecht, L. Song, Y. Hu, T. Bein, [65] S. T. Williams, F. Zuo, C.-C. Chueh, C.-Y. Liao, P.-W. Liang,
P. Docampo, P. Müller-Buschbaum, Adv. Mater. Interfaces 2016, 3, A. K. Y. Jen, ACS Nano 2014, 8, 10640.
1600403. [66] H. Yu, F. Wang, F. Xie, W. Li, J. Chen, N. Zhao, Adv. Funct. Mater.
[38] B. J. Foley, J. Girard, B. A. Sorenson, A. Z. Chen, J. Scott Niezgoda, 2014, 24, 7102.
M. R. Alpert, A. F. Harper, D.-M. Smilgies, P. Clancy, W. A. Saidi, [67] A. Binek, I. Grill, N. Huber, K. Peters, A. G. Hufnagel,
J. J. Choi, J. Mater. Chem. A 2017, 5, 113. M. Handloser, P. Docampo, A. Hartschuh, T. Bein, Chem. Asian J.
[39] J.-W. Lee, H.-S. Kim, N.-G. Park, Acc. Chem. Res. 2016, 49, 311. 2016, 11, 1199.
[40] N. Ahn, D.-Y. Son, I.-H. Jang, S. M. Kang, M. Choi, N.-G. Park, [68] E. L. Unger, A. R. Bowring, C. J. Tassone, V. L. Pool, A. Gold-Parker,
J. Am. Chem. Soc. 2015, 137, 8696. R. Cheacharoen, K. H. Stone, E. T. Hoke, M. F. Toney,
[41] D. Bi, W. Tress, M. I. Dar, P. Gao, J. Luo, C. Renevier, K. Schenk, M. D. McGehee, Chem. Mater. 2014, 26, 7158.
A. Abate, F. Giordano, J.-P. Correa Baena, J.-D. Decoppet, [69] Z. Zhang, X. Yue, D. Wei, M. Li, P. Fu, B. Xie, D. Song, Y. Li, RSC
S. M. Zakeeruddin, M. K. Nazeeruddin, M. Gratzel, A. Hagfeldt, Adv 2015, 5, 104606.
Sci. Adv. 2016, 2, e1501170. [70] V. Murugan, Y. Ogomi, S. S. Pandey, T. Toyoda, Q. Shen,
[42] A. Buin, P. Pietsch, J. Xu, O. Voznyy, A. H. Ip, R. Comin, S. Hayase, Appl. Phys. Express 2015, 8, 125501.
E. H. Sargent, Nano Lett. 2014, 14, 6281. [71] M. Liu, M. B. Johnston, H. J. Snaith, Nature 2013, 501, 395.
[43] F. K. Aldibaja, L. Badia, E. Mas-Marzá, R. S. Sánchez, E. M. Barea, [72] C. Roldan-Carmona, O. Malinkiewicz, A. Soriano,
I. Mora-Sero, J. Mater. Chem. A 2015, 3, 9194. G. Minguez Espallargas, A. Garcia, P. Reinecke, T. Kroyer,
[44] L. Ling, S. Yuan, P. Wang, H. Zhang, L. Tu, J. Wang, Y. Zhan, M. I. Dar, M. K. Nazeeruddin, H. J. Bolink, Energy Environ. Sci.
L. Zheng, Adv. Funct. Mater. 2016, 26, 5028. 2014, 7, 994.
[45] J. You, Y. Yang, Z. Hong, T.-B. Song, L. Meng, Y. Liu, C. Jiang, [73] O. Malinkiewicz, A. Yella, Y. H. Lee, G. M. Espallargas, M. Grätzel,
H. Zhou, W.-H. Chang, G. Li, Y. Yang, Appl. Phys. Lett. 2014, 105, M. K. Nazeeruddin, H. J. Bolink, Nat. Photon. 2014, 8, 128.
183902. [74] S. Olthof, K. Meerholz, Sci. Rep. 2017, 7, 40267.
[46] K. K. Bass, R. E. McAnally, S. Zhou, P. I. Djurovich, [75] H. Xu, Y. Wu, J. Cui, C. Ni, F. Xu, J. Cai, F. Hong, Z. Fang,
M. E. Thompson, B. C. Melot, Chem. Commun. 2014, 50, 15819. W. Wang, J. Zhu, L. Wang, R. Xu, F. Xu, Phys. Chem. Chem. Phys.
[47] N. Giesbrecht, J. Schlipf, L. Oesinghaus, A. Binek, T. Bein, 2016, 18, 18607.
P. Müller-Buschbaum, P. Docampo, ACS Energy Lett. 2016, 1, 150. [76] C. Tao, S. Neutzner, L. Colella, S. Marras, A. R. Srimath Kandada,
[48] P. Docampo, F. C. Hanusch, N. Giesbrecht, P. Angloher, M. Gandini, M. D. Bastiani, G. Pace, L. Manna, M. Caironi,
A. Ivanova, T. Bein, APL Mater. 2014, 2, 081508. C. Bertarelli, A. Petrozza, Energy Environ. Sci. 2015, 8, 2365.
[49] M. Saliba, K. W. Tan, H. Sai, D. T. Moore, T. Scott, W. Zhang, [77] L. K. Ono, M. R. Leyden, S. Wang, Y. Qi, J. Mater. Chem. A 2016,
L. A. Estroff, U. Wiesner, H. J. Snaith, J. Phys. Chem. C 2014, 118, 4, 6693.
17171. [78] D. H. Cao, C. C. Stoumpos, O. K. Farha, J. T. Hupp,
[50] N. Cho, F. Li, B. Turedi, L. Sinatra, S. P. Sarmah, M. R. Parida, M. G. Kanatzidis, J. Am. Chem. Soc. 2015, 137, 7843.
M. I. Saidaminov, B. Murali, V. M. Burlakov, A. Goriely, [79] Y. Hu, J. Schlipf, M. Wussler, M. L. Petrus, W. Jaegermann,
O. F. Mohammed, T. Wu, O. M. Bakr, Nat. Commun. 2016, 7, T. Bein, P. Müller-Buschbaum, P. Docampo, ACS Nano 2016, 10,
13407. 5999.
[51] K. Liang, D. B. Mitzi, M. T. Prikas, Chem. Mater. 1998, 10, 403. [80] T. M. Koh, V. Shanmugam, J. Schlipf, L. Oesinghaus,
[52] D. Liu, T. L. Kelly, Nat. Photon 2014, 8, 133. P. Müller-Buschbaum, N. Ramakrishnan, V. Swamy, N. Mathews,
[53] Y. Wang, S. Li, P. Zhang, D. Liu, X. Gu, H. Sarvari, Z. Ye, J. Wu, P. P. Boix, S. G. Mhaisalkar, Adv. Mater. 2016, 28, 3653.
Z. Wang, Z. D. Chen, Nanoscale 2016, 8, 19654. [81] H. Tsai, W. Nie, J.-C. Blancon, C. C. Stoumpos, R. Asadpour,
[54] S. Yang, Y. C. Zheng, Y. Hou, X. Chen, Y. Chen, Y. Wang, H. Zhao, B. Harutyunyan, A. J. Neukirch, R. Verduzco, J. J. Crochet,
H. G. Yang, Chem. Mater. 2014, 26, 6705. S. Tretiak, L. Pedesseau, J. Even, M. A. Alam, G. Gupta, J. Lou,

Adv. Energy Mater. 2017, 7, 1700264 1700264  (23 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

P. M. Ajayan, M. J. Bedzyk, M. G. Kanatzidis, A. D. Mohite, Nature [114] Y. Li, J. K. Cooper, R. Buonsanti, C. Giannini, Y. Liu, F. M. Toma,
2016, 536, 312. I. D. Sharp, J. Phys. Chem. Lett. 2015, 6, 493.
[82] Q. Dong, Y. Fang, Y. Shao, P. Mulligan, J. Qiu, L. Cao, J. Huang, [115] C. Li, S. Tscheuschner, F. Paulus, P. E. Hopkinson, J. Kießling,
Science 2015, 347, 967. A. Köhler, Y. Vaynzof, S. Huettner, Adv. Mater. 2016, 28, 2446.
[83] G. Maculan, A. D. Sheikh, A. L. Abdelhady, M. I. Saidaminov, [116] I. Levine, P. K. Nayak, J. T.-W. Wang, N. Sakai, S. Van Reenen,
M. A. Haque, B. Murali, E. Alarousu, O. F. Mohammed, T. Wu, T. M. Brenner, S. Mukhopadhyay, H. J. Snaith, G. Hodes, D. Cahen,
O. M. Bakr, J. Phys. Chem. Lett. 2015, 6, 3781. J. Phys. Chem. C 2016, 120, 16399.
[84] Y. Liu, Y. Zhang, Z. Yang, D. Yang, X. Ren, L. Pang, S. Liu, Adv. [117] R. S. Sanchez, V. Gonzalez-Pedro, J.-W. Lee, N.-G. Park, Y. S. Kang,
Mater. 2016, 28, 9204. I. Mora-Sero, J. Bisquert, J. Phys. Chem. Lett. 2014, 5, 2357.
[85] Y. Chen, M. He, J. Peng, Y. Sun, Z. Liang, Adv. Sci. 2016, 3, [118] R. Gottesman, E. Haltzi, L. Gouda, S. Tirosh, Y. Bouhadana,
1500392. A. Zaban, E. Mosconi, F. De Angelis, J. Phys. Chem. Lett. 2014, 5,
[86] E. J. Luber, J. M. Buriak, ACS Nano 2013, 7, 4708. 2662.
[87] D. B. Mitzi, Prog. Inorg. Chem. 2007, 48, 1. [119] S. D. Stranks, G. E. Eperon, G. Grancini, C. Menelaou,
[88] Z. Cheng, J. Lin, CrystEngComm 2010, 12, 2646. M. J. P. Alcocer, T. Leijtens, L. M. Herz, A. Petrozza, H. J. Snaith,
[89] F. Brivio, A. B. Walker, A. Walsh, APL Mater. 2013, 1, 042111. Science 2013, 342, 341.
[90] T.-B. Song, Q. Chen, H. Zhou, C. Jiang, H.-H. Wang, Y. M. Yang, [120] H. S. Kim, N.-G. Park, J. Phys. Chem. Lett. 2014, 5, 2927.
Y. Liu, J. You, Y. Yang, J. Mater. Chem. A 2015, 3, 9032. [121] M. De Bastiani, G. Dell’Erba, M. Gandini, V. D’Innocenzo,
[91] J.-H. Im, I.-H. Jang, N. Pellet, M. Grätzel, N.-G. Park, Nat. Nano- S. Neutzner, A. R. S. Kandada, G. Grancini, M. Binda, M. Prato,
technol. 2014, 9, 927. J. M. Ball, M. Caironi, A. Petrozza, Adv. Energy Mater. 2016, 6,
[92] Y. Y. Zhou, M. J. Yang, W. W. Wu, A. L. Vasiliev, K. Zhu, 1501453.
N. P. Padture, J. Mater. Chem. A 2015, 3, 8178. [122] E. L. Unger, E. T. Hoke, C. D. Bailie, W. H. Nguyen, A. R. Bowring,
[93] T. P. Gujar, M. Thelakkat, Energy Technol. 2016, 4, 449. T. Heumueller, M. G. Christoforo, M. D. McGehee, Energy Environ.
[94] P. Qin, S. Tanaka, S. Ito, N. Tetreault, K. Manabe, H. Nishino, Sci. 2014, 7, 3690.
M. K. Nazeeruddin, M. Grätzel, Nat. Commun. 2014, 5, 3834. [123] D. Litvin, Acta Crystallogr. 1986, A42, 44.
[95] C. Jiang, S. L. Lim, W. P. Goh, F. X. Wei, J. Zhang, ACS Appl. Mater. [124] T. S. Sherkar, L. Jan Anton Koster, Phys. Chem. Chem. Phys. 2016,
Interfaces 2015, 7, 24726. 18, 331.
[96] H.-S. Ko, J.-W. Lee, N.-G. Park, J. Mater. Chem. A 2015, 3, [125] B. Chen, J. Shi, X. Zheng, Y. Zhou, K. Zhu, S. Priya, J. Mater. Chem.
8808. A 2015, 3, 7699.
[97] F. Liu, Q. Dong, M. K. Wong, A. B. Djurišić, A. Ng, Z. Ren, [126] M. Coll, A. Gomez, E. Mas-Marza, O. Almora, G. Garcia-Belmonte,
Q. Shen, C. Surya, W. K. Chan, J. Wang, Adv. Energy Mater. 2016, M. Campoy-Quiles, J. Bisquert, J. Phys. Chem. Lett. 2015,
6, 1502206. 6, 1408.
[98] Y. C. Kim, N. J. Jeon, J. H. Noh, W. S. Yang, J. Seo, J. S. Yun, [127] Y. Kutes, L. Ye, Y. Zhou, S. Pang, B. D. Huey, N. P. Padture, J. Phys.
A. Ho-Baillie, S. Huang, M. A. Green, J. Seidel, Adv. Energy Mater. Chem. Lett. 2014, 5, 3335.
2015, 6, 1502104. [128] J. Wei, Y. Zhao, H. Li, G. Li, J. Pan, D. Xu, Q. Zhao, D. Yu, J. Phys.
[99] D.-Y. Son, J.-W. Lee, Y. J. Choi, I.-H. Jang, S. Lee, P. J. Yoo, H. Shin, Chem. Lett. 2014, 5, 3937.
N. Ahn, M. Choi, D. Kim, Nat. Energy 2016, 1, 16081. [129] H.-W. Chen, N. Sakai, M. Ikegami, T. Miyasaka, J. Phys. Chem. Lett.
[100] T. J. Jacobsson, J.-P. Correa-Baena, E. Halvani Anaraki, B. Philippe, 2015, 6, 164.
S. D. Stranks, M. E. F. Bouduban, W. Tress, K. Schenk, J. Teuscher, [130] S. Meloni, T. Moehl, W. Tress, M. Franckevicius, M. Saliba,
J.-E. Moser, J. Am. Chem. Soc. 2016, 138, 10331. Y. H. Lee, P. Gao, M. K. Nazeeruddin, S. M. Zakeeruddin,
[101] M. L. Petrus, Y. Hu, D. Moia, P. Calado, A. M. A. Leguy, U. Rothlisberger, M. Graetzel, Nat. Commun. 2016, 7, 10334.
P. R. F. Barnes, P. Docampo, ChemSusChem 2016, 9, 2699. [131] A. M. A. Leguy, J. M. Frost, A. P. McMahon, V. G. Sakai,
[102] N.-G. Park, Nano Convergence 2016, 3, 1. W. Kockelmann, C. Law, X. Li, F. Foglia, A. Walsh, B. C. O’Regan,
[103] J.-H. Im, H.-S. Kim, N.-G. Park, APL Mater. 2014, 2, 081510. J. Nelson, J. T. Cabral, P. R. F. Barnes, Nat. Commun. 2015, 6, 7124.
[104] B. Conings, L. Baeten, C. De Dobbelaere, J. D’Haen, J. Manca, [132] B. Chen, M. Yang, S. Priya, K. Zhu, J. Phys. Chem. Lett. 2016, 7,
H. G. Boyen, Adv. Mater. 2014, 26, 2041. 905.
[105] G. E. Eperon, V. M. Burlakov, A. Goriely, H. J. Snaith, ACS Nano [133] J. Beilsten-Edmands, G. E. Eperon, R. D. Johnson, H. J. Snaith,
2013, 8, 591. P. G. Radaelli, Appl. Phys. Lett. 2015, 106, 173502.
[106] V. M. Burlakov, G. E. Eperon, H. J. Snaith, S. J. Chapman, [134] I. M. Hermes, S. A. Bretschneider, V. W. Bergmann, D. Li,
A. Goriely, Appl. Phys. Lett. 2014, 104, 091602. A. Klasen, J. Mars, W. Tremel, F. Laquai, H.-J. Butt, M. Mezger,
[107] W. Chen, Y. Wu, Y. Yue, J. Liu, W. Zhang, X. Yang, H. Chen, E. Bi, J. Phys. Chem. C 2016, 120, 5724.
I. Ashraful, M. Grätzel, Science 2015, 350, 944. [135] L. K. Ono, Y. Qi, J. Phys. Chem. Lett. 2016, 7, 4764.
[108] Y. Rong, S. Venkatesan, R. Guo, Y. Wang, J. Bao, W. Li, Z. Fan, [136] D. W. Miller, G. E. Eperon, E. T. Roe, C. W. Warren, H. J. Snaith,
Y. Yao, Nanoscale 2016, 8, 12892. M. C. Lonergan, Appl. Phys. Lett. 2016, 109, 153902.
[109] F. Wang, M. Endo, S. Mouri, Y. Miyauchi, Y. Ohno, A. Wakamiya, [137] S. Singh, C. Li, F. Panzer, K. L. Narasimhan, A. Graeser, T. P. Gujar,
Y. Murata, K. Matsuda, Nanoscale 2016, 8, 11882. A. Köhler, M. Thelakkat, S. Huettner, D. Kabra, J. Phys. Chem. Lett.
[110] M. Yang, Y. Zhou, Y. Zeng, C. S. Jiang, N. P. Padture, K. Zhu, Adv. 2016, 7, 3014.
Mater. 2015, 27, 6363. [138] M. A. Green, A. Ho-Baillie, H. J. Snaith, Nat. Photon. 2014, 8, 506.
[111] C. Momblona, O. Malinkiewicz, C. Roldan-Carmona, A. Soriano, [139] G.-J. A. H. Wetzelaer, M. Scheepers, A. M. Sempere, C. Momblona,
L. Gil-Escrig, E. Bandiello, M. Scheepers, E. Edri, H. J. Bolink, APL J. Ávila, H. J. Bolink, Adv. Mater. 2015, 27, 1837.
Mater. 2014, 2, 081504. [140] L. M. Herz, Annu. Rev. Phys. Chem. 2015, 67, 65.
[112] D. Forgács, L. Gil-Escrig, D. Pérez-Del-Rey, C. Momblona, [141] P. Delugas, C. Caddeo, A. Filippetti, A. Mattoni, J. Phys. Chem.
J. Werner, B. Niesen, C. Ballif, M. Sessolo, H. J. Bolink, Adv. Energy Lett. 2016, 7, 2356.
Mater. 2016, 7, 1602121. [142] X. Wu, M. T. Trinh, D. Niesner, H. Zhu, Z. Norman, J. S. Owen,
[113] Q. Chen, H. Zhou, T.-B. Song, S. Luo, Z. Hong, H.-S. Duan, O. Yaffe, B. J. Kudisch, X.-Y. Zhu, J. Am. Chem. Soc. 2015, 137,
L. Dou, Y. Liu, Y. Yang, Nano Lett. 2014, 14, 4158. 2089.

Adv. Energy Mater. 2017, 7, 1700264 1700264  (24 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

[143] V. D’Innocenzo, G. Grancini, M. J. P. Alcocer, A. R. S. Kandada, [173] Z. Xiao, Y. Yuan, Y. Shao, Q. Wang, Q. Dong, C. Bi, P. Sharma,
S. D. Stranks, M. M. Lee, G. Lanzani, H. J. Snaith, A. Petrozza, A. Gruverman, J. Huang, Nat. Mater. 2015, 14, 193.
Nat. Commun. 2014, 5, 3586. [174] Y. Yuan, J. Chae, Y. Shao, Q. Wang, Z. Xiao, A. Centrone, J. Huang,
[144] Y. Yamada, T. Nakamura, M. Endo, A. Wakamiya, Y. Kanemitsu, Adv. Energy Mater. 2015, 5, 1500615.
J. Am. Chem. Soc. 2014, 136, 11610. [175] T.-Y. Yang, G. Gregori, N. Pellet, M. Grätzel, J. Maier, Angew.
[145] H.-S. Duan, H. Zhou, Q. Chen, P. Sun, S. Luo, T.-B. Song, B. Bob, Chem., Int. Ed. 2015, 54, 7905.
Y. Yang, Phys. Chem. Chem. Phys. 2015, 17, 112. [176] Y. Yuan, Q. Wang, Y. Shao, H. Lu, T. Li, A. Gruverman, J. Huang,
[146] Y. Shao, Z. Xiao, C. Bi, Y. Yuan, J. Huang, Nat. Commun. 2014, 5, Adv. Energy Mater. 2016, 6, 1501803.
5784. [177] E. Mosconi, D. Meggiolaro, H. J. Snaith, S. D. Stranks,
[147] W.-J. Yin, T. Shi, Y. Yan, Appl. Phys. Lett. 2014, 104, 063903. F. De Angelis, Energy Environ. Sci. 2016, 9, 3180.
[148] V. W. Bergmann, Y. Guo, H. Tanaka, I. M. Hermes, D. Li, A. Klasen, [178] G. C. Farrington, J. L. Briant, Science 1979, 204, 1371.
S. A. Bretschneider, E. Nakamura, R. Berger, S. A. L. Weber, ACS [179] Y. Shao, Y. Fang, T. Li, Q. Wang, Q. Dong, Y. Deng, Y. Yuan, H. Wei,
Appl. Mater. Inerfaces 2016, 8, 19402. M. Wang, A. Gruverman, J. Shield, J. Huang, Energy Environ. Sci.
[149] V. W. Bergmann, S. A. L. Weber, F. Javier Ramos, 2016, 9, 1752.
M. K. Nazeeruddin, M. Grätzel, D. Li, A. L. Domanski, [180] J. S. Yun, J. Seidel, J. Kim, A. M. Soufiani, S. Huang, J. Lau,
I. Lieberwirth, S. Ahmad, R. Berger, Nat. Commun. 2014, 5, 5001. N. J. Jeon, S. I. Seok, M. A. Green, A. Ho-Baillie, Adv. Energy Mater.
[150] E. Edri, S. Kirmayer, A. Henning, S. Mukhopadhyay, K. Gartsman, 2016, 6, 1600330.
Y. Rosenwaks, G. Hodes, D. Cahen, Nano Lett. 2014, 14, 1000. [181] J. Xing, Q. Wang, Q. Dong, Y. Yuan, Y. Fang, J. Huang, Phys. Chem.
[151] E. Edri, S. Kirmayer, S. Mukhopadhyay, K. Gartsman, G. Hodes, Chem. Phys. 2016, 18, 30484.
D. Cahen, Nat. Commun. 2014, 5, 3461. [182] G. A. MacDonald, M. Yang, S. Berweger, J. P. Killgore, P. Kabos,
[152] S. G. Motti, M. Gandini, A. J. Barker, J. M. Ball, J. J. Berry, K. Zhu, F. W. DelRio, Energy Environ. Sci. 2016, 9, 3642.
A. R. Srimath Kandada, A. Petrozza, ACS Energy Lett. 2016, 1, 726. [183] G. H. Vineyard, J. Phys. Chem. Solids 1957, 3, 121.
[153] N. K. Noel, A. Abate, S. D. Stranks, E. S. Parrott, V. M. Burlakov, [184] H. Yu, H. Lu, F. Xie, S. Zhou, N. Zhao, Adv. Funct. Mater. 2016, 26,
A. Goriely, H. J. Snaith, ACS Nano 2014, 8, 9815. 1411.
[154] W. d. Dane, S. M. Vorpahl, S. D. Stranks, H. Nagaoka, [185] T. M. Brown, R. H. Friend, I. S. Millard, D. J. Lacey, T. Butler,
G. E. Eperon, M. E. Ziffer, H. J. Snaith, D. S. Ginger, Science 2015, J. H. Burroughes, F. Cacialli, J. Appl. Phys. 2003, 93, 6159.
348, 683. [186] I. H. Campbell, T. W. Hagler, D. I. Smith, J. P. Ferraris, Phys. Rev.
[155] L. Meng, J. You, T.-F. Guo, Y. Yang, Acc. Chem. Res. 2016, 49, 155. Lett. 1996, 76, 1900.
[156] J. H. Heo, H. J. Han, D. Kim, T. K. Ahn, S. H. Im, Energy Environ. [187] C. Li, G. J. Beirne, G. Kamita, G. Lakhwani, J. Wang,
Sci. 2015, 8, 1602. N. C. Greenham, J. Appl. Phys. 2014, 116, 114501.
[157] S. van Reenen, M. Kemerink, H. J. Snaith, J. Phys. Chem. Lett. [188] R. A. Belisle, W. H. Nguyen, A. R. Bowring, P. Calado, X. Li,
2015, 6, 3808. S. J. C. Irvine, M. D. McGehee, P. R. F. Barnes, B. C. O’Regan,
[158] T. Leijtens, G. E. Eperon, A. J. Barker, G. Grancini, W. Zhang, Energy Environ. Sci. 2017, 10, 192.
J. M. Ball, A. R. S. Kandada, H. J. Snaith, A. Petrozza, Energy [189] Y. Zhang, M. Liu, G. E. Eperon, T. C. Leijtens, D. McMeekin,
Environ. Sci. 2016, 9, 3472. M. Saliba, W. Zhang, M. de Bastiani, A. Petrozza, L. M. Herz,
[159] W. Tress, N. Marinova, T. Moehl, S. M. Zakeeruddin, M. B. Johnston, H. Lin, H. J. Snaith, Mater. Horiz. 2015, 2, 315.
M. K. Nazeeruddin, M. Grätzel, Energy Environ. Sci. 2015, 8, [190] D. A. Egger, E. Edri, D. Cahen, G. Hodes, J. Phys. Chem. Lett. 2015,
995. 6, 279.
[160] G. Richardson, S. E. J. O’Kane, R. G. Niemann, T. A. Peltola, [191] B. Wu, K. Fu, N. Yantara, G. Xing, S. Sun, T. C. Sum, N. Mathews,
J. M. Foster, P. J. Cameron, A. B. Walker, Energy Environ. Sci. 2016, Adv. Energy. Mater. 2015, 5, 1500829.
9, 1476. [192] M. K. Gangishetty, R. W. J. Scott, T. L. Kelly, Nanoscale 2016, 8,
[161] Y. Tian, A. Merdasa, M. Peter, M. Abdellah, K. Zheng, 6300.
C. S. Ponseca, T. Pullerits, A. Yartsev, V. Sundström, [193] Y. Bai, H. Yu, Z. Zhu, K. Jiang, T. Zhang, N. Zhao, S. Yang, H. Yan,
I. G. Scheblykin, Nano Lett. 2015, 15, 1603. J. Mater. Chem. A 2015, 3, 9098.
[162] T. Zhao, C.-C. Chueh, Q. Chen, A. Rajagopal, A. K. Y. Jen, ACS [194] C.-H. Chiang, C.-G. Wu, Nat. Photon. 2016, 10, 196.
Energy Lett. 2016, 1, 757. [195] P. Calado, A. M. Telford, D. Bryant, X. Li, J. Nelson, B. C. O’Regan,
[163] D. W. deQuilettes, W. Zhang, V. M. Burlakov, D. J. Graham, P. R. F. Barnes, Nat. Commun. 2016, 7, 13831.
T. Leijtens, A. Osherov, V. Bulović, H. J. Snaith, D. S. Ginger, [196] D. Bryant, S. Wheeler, B. C. O’Regan, T. Watson, P. R. F. Barnes,
S. D. Stranks, Nat. Commun. 2016, 7, 11683. D. Worsley, J. Durrant, J. Phys. Chem. Lett. 2015, 6, 3190.
[164] S. Chen, X. Wen, S. Huang, F. Huang, Y.-B. Cheng, M. Green, [197] J. Xu, A. Buin, A. H. Ip, W. Li, O. Voznyy, R. Comin, M. Yuan,
A. Ho-Baillie, Solar RRL 2017, 1, 1600001. S. Jeon, Z. Ning, J. J. McDowell, P. Kanjanaboos, J.-P. Sun, X. Lan,
[165] C. Li, Y. Zhong, C. Luna, T. Unger, K. Deichsel, A. Gräser, J. Köhler, L. N. Quan, D. H. Kim, I. G. Hill, P. Maksymovych, E. H. Sargent,
A. Köhler, R. Hildner, S. Huettner, Molecules 2016, 21, 1081. Nat. Commun. 2015, 6, 7081.
[166] B. C. O’Regan, P. R. F. Barnes, X. Li, C. Law, E. Palomares, [198] C. D. Weber, C. Bradley, M. C. Lonergan, J. Mater. Chem. A 2014,
J. M. Marin-Beloqui, J. Am. Chem. Soc. 2015, 137, 5087. 2, 303.
[167] J. M. Azpiroz, E. Mosconi, J. Bisquert, F. De Angelis, Energy [199] G. Xing, B. Wu, S. Chen, J. Chua, N. Yantara, S. Mhaisalkar,
Environ. Sci. 2015, 8, 2118. N. Mathews, T. C. Sum, Small 2015, 11, 3606.
[168] Y. Yuan, J. Huang, Acc. Chem. Res. 2016, 49, 286. [200] K. Wojciechowski, S. D. Stranks, A. Abate, G. Sadoughi,
[169] J. M. Frost, A. Walsh, Acc. Chem. Res. 2016, 49, 528. A. Sadhanala, N. Kopidakis, G. Rumbles, C.-Z. Li, R. H. Friend,
[170] N. K. Elumalai, A. Uddin, Sol. Energy Mater. Sol. Cells 2016, 157, A. K. Y. Jen, H. J. Snaith, ACS Nano 2014, 8, 12701.
476. [201] D. Yang, X. Zhou, R. Yang, Z. Yang, W. Yu, X. Wang, C. Li, S. F. Liu,
[171] C. Eames, J. M. Frost, P. R. F. Barnes, B. C. O’Regan, A. Walsh, R. P. H. Chang, Energy Environ. Sci. 2016, 9, 3071.
M. S. Islam, Nat. Commun. 2015, 6, 7497. [202] EnergyTrend, http://pv.energytrend.com (accessed: March 2017).
[172] D. A. Egger, L. Kronik, A. M. Rappe, Angew. Chem., Int. Ed. 2015, [203] M. L. Petrus, T. Bein, T. J. Dingemans, P. Docampo, J. Mater.
54, 12437. Chem. A 2015, 3, 12159.

Adv. Energy Mater. 2017, 7, 1700264 1700264  (25 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

[204] K. H. Hendriks, W. Li, G. H. L. Heintges, G. W. P. van Pruissen, [231] J. Yang, B. D. Siempelkamp, D. Liu, T. L. Kelly, ACS Nano 2015, 9,
M. M. Wienk, R. A. J. Janssen, J. Am. Chem. Soc. 2014, 136, 11128. 1955.
[205] M. P. Nikiforov, B. Lai, W. Chen, S. Chen, R. D. Schaller, J. Strzalka, [232] J. A. Christians, P. A. Miranda Herrera, P. V. Kamat, J. Am. Chem.
J. Maser, S. B. Darling, Energy Environ. Sci. 2013, 6, 1513. Soc. 2015, 137, 1530.
[206] J. Wang, Y. Chen, M. Liang, G. Ge, R. Zhou, Z. Sun, S. Xue, Dyes [233] E. Mosconi, J. M. Azpiroz, F. De Angelis, Chem. Mater. 2015, 27,
Pigm. 2016, 125, 399. 4885.
[207] J.-Y. Shao, D. Li, K. Tang, Y.-W. Zhong, Q. Meng, RSC Adv. 2016, 6, [234] I. C. Smith, E. T. Hoke, D. Solis-Ibarra, M. D. McGehee,
92213. H. I. Karunadasa, Angew. Chem. 2014, 126, 11414.
[208] T. Leijtens, T. Giovenzana, S. N. Habisreutinger, J. S. Tinkham, [235] S. Yang, Y. Wang, P. Liu, Y.-B. Cheng, H. J. Zhao, H. G. Yang, Nat.
N. K. Noel, B. A. Kamino, G. Sadoughi, A. Sellinger, H. J. Snaith, Energy 2016, 1, 15016.
ACS Appl. Mater. Interfaces 2016, 8, 5981. [236] L. K. Ono, S. R. Raga, M. Remeika, A. J. Winchester, A. Gabe, Y. Qi,
[209] H. Choi, S. Park, S. Paek, P. Ekanayake, N. Mohammad K, J. Ko, J. Mater. Chem. A 2015, 3, 15451.
J. Mater. Chem. A 2014, 2, 19136. [237] B.-A. Chen, J.-T. Lin, N.-T. Suen, C.-W. Tsao, T.-C. Chu, Y.-Y. Hsu,
[210] A. Krishna, D. Sabba, H. Li, J. Yin, P. P. Boix, C. Soci, T.-S. Chan, Y.-T. Chan, J.-S. Yang, C.-W. Chiu, H. M. Chen, ACS
S. G. Mhaisalkar, A. C. Grimsdale, Chem. Sci. 2014, 5, 2702. Energy Lett. 2017, 2, 342.
[211] M. Maciejczyk, A. Ivaturi, N. Robertson, J. Mater. Chem. A 2016, [238] L. Zheng, Y.-H. Chung, Y. Ma, L. Zhang, L. Xiao, Z. Chen, S. Wang,
4, 4855. B. Qu, Q. Gong, Chem. Commun. 2014, 50, 11196.
[212] B. Xu, D. Bi, Y. Hua, P. Liu, M. Cheng, M. Grätzel, L. Kloo, [239] Y. S. Kwon, J. Lim, H.-J. Yun, Y.-H. Kim, T. Park, Energy Environ. Sci.
A. Hagfeldt, L. Sun, Energy Environ. Sci. 2016, 9, 873. 2014, 7, 1454.
[213] M. Saliba, S. Orlandi, T. Matsui, S. Aghazada, M. Cavazzini, [240] S. N. Habisreutinger, T. Leijtens, G. E. Eperon, S. D. Stranks,
J.-P. Correa-Baena, P. Gao, R. Scopelliti, E. Mosconi, R. J. Nicholas, H. J. Snaith, Nano Lett. 2014, 14, 5561.
K.-H. Dahmen, F. De Angelis, A. Abate, A. Hagfeldt, G. Pozzi, [241] C. Law, L. Miseikis, S. Dimitrov, P. Shakya-Tuladhar, X. Li,
M. Grätzel, M. K. Nazeeruddin, Nat. Energy 2016, 1, 15017. P. R. F. Barnes, J. Durrant, B. C. O’Regan, Adv. Mater. 2014, 26,
[214] H. Li, K. Fu, A. Hagfeldt, M. Grätzel, S. G. Mhaisalkar, 6268.
A. C. Grimsdale, Angew. Chem. 2014, 126, 4169. [242] T. Baikie, Y. N. Fang, J. M. Kadro, M. Schreyer, F. X. Wei,
[215] M. L. Petrus, R. K. M. Bouwer, U. Lafont, S. Athanasopoulos, S. G. Mhaisalkar, M. Grätzel, T. J. White, J. Mater. Chem. A 2013,
N. C. Greenham, T. J. Dingemans, J. Mater. Chem. A 2014, 2, 9474. 1, 5628.
[216] M. L. Petrus, F. S. F. Morgenstern, A. Sadhanala, R. H. Friend, [243] E. Skoplaki, A. G. Boudouvis, J. A. Palyvos, Sol. Energy Mater. Sol.
N. C. Greenham, T. J. Dingemans, Chem. Mater. 2015, 27, 2990. Cells 2008, 92, 1393.
[217] M. Koole, R. Frisenda, M. L. Petrus, M. L. Perrin, [244] G. E. Eperon, G. M. Paterno, R. J. Sutton, A. Zampetti,
H. S. J. van der Zant, T. J. Dingemans, Org. Electron. 2016, 34, 38. A. A. Haghighirad, F. Cacialli, H. J. Snaith, J. Mater. Chem. A 2015,
[218] M. L. Petrus, R. K. M. Bouwer, U. Lafont, D. H. K. Murthy, 3, 19688.
R. J. P. Kist, M. L. Bohm, Y. Olivier, T. J. Savenije, L. D. A. Siebbeles, [245] J.-S. Yeo, R. Kang, S. Lee, Y.-J. Jeon, N. Myoung, C.-L. Lee,
N. C. Greenham, T. J. Dingemans, Polymer Chemistry 2013, 4, D.-Y. Kim, J.-M. Yun, Y.-H. Seo, S.-S. Kim, S.-I. Na, Nano Energy
4182. 2015, 12, 96.
[219] E. M. Hutter, J.-J. Hofman, M. L. Petrus, M. Moes, R. D. Abellon, [246] C. C. Stoumpos, C. D. Malliakas, M. G. Kanatzidis, Inorg. Chem.
P. Docampo, T. J. Savenije, Adv. Energy Mater. 2017, 7, 1602349. 2013, 52, 9019.
[220] M. Daskeviciene, S. Paek, Z. Wang, T. Malinauskas, [247] A. Binek, F. C. Hanusch, P. Docampo, T. Bein, J. Phys. Chem. Lett.
G. Jokubauskaite, K. Rakstys, K. T. Cho, A. Magomedov, 2015, 6, 1249.
V. Jankauskas, S. Ahmad, H. J. Snaith, V. Getautis, [248] D. P. McMeekin, G. Sadoughi, W. Rehman, G. E. Eperon,
M. K. Nazeeruddin, Nano Energy 2017, 32, 551. M. Saliba, M. T. Hörantner, A. Haghighirad, N. Sakai, L. Korte,
[221] J. Liu, S. Pathak, T. Stergiopoulos, T. Leijtens, K. Wojciechowski, B. Rech, M. B. Johnston, L. M. Herz, H. J. Snaith, Science 2016,
S. Schumann, N. Kausch-Busies, H. J. Snaith, J. Phys. Chem. Lett. 351, 151.
2015, 6, 1666. [249] N. Pellet, P. Gao, G. Gregori, T.-Y. Yang, M. K. Nazeeruddin,
[222] T. A. Berhe, W.-N. Su, C.-H. Chen, C.-J. Pan, J.-H. Cheng, J. Maier, M. Grätzel, Angew. Chem., Int. Ed. 2014, 53, 3151.
H.-M. Chen, M.-C. Tsai, L.-Y. Chen, A. A. Dubale, B.-J. Hwang, [250] Y. Jiang, M. A. Green, R. Sheng, A. Ho-Baillie, Data in Brief 2015,
Energy Environ. Sci. 2016, 9, 323. 3, 201.
[223] G. E. Eperon, S. N. Habisreutinger, T. Leijtens, B. J. Bruijnaers, [251] S. Luo, P. You, G. Cai, H. Zhou, F. Yan, W. A. Daoud, Mater. Lett.
J. J. van Franeker, D. W. deQuilettes, S. Pathak, R. J. Sutton, 2016, 169, 236.
G. Grancini, D. S. Ginger, R. A. J. Janssen, A. Petrozza, H. J. Snaith, [252] M. Bag, Z. Jiang, L. A. Renna, S. P. Jeong, V. M. Rotello,
ACS Nano 2015, 9, 9380. D. Venkataraman, Mater. Lett. 2016, 164, 472.
[224] J. H. Noh, S. H. Im, J. H. Heo, T. N. Mandal, S. I. Seok, Nano Lett. [253] Y. Fang, X. Wang, Q. Wang, J. Huang, T. Wu, Phys. Status Solidi A
2013, 13, 1764. 2014, 211, 2809.
[225] A. Leguy, Y. Hu, M. Campoy-Quiles, M. I. Alonso, O. J. Weber, [254] T. Leijtens, I. K. Ding, T. Giovenzana, J. T. Bloking, M. D. McGehee,
P. Azarhoosh, M. van Schilfgaarde, M. T. Weller, T. Bein, J. Nelson, A. Sellinger, ACS Nano 2012, 6, 1455.
P. Docampo, P. R. F. Barnes, Chem. Mater. 2015, 27, 3397. [255] A. Abate, S. Paek, F. Giordano, J. P. Correa Baena, M. Saliba,
[226] J. M. Frost, K. T. Butler, F. Brivio, C. H. Hendon, P. Gao, T. Matsui, J. Ko, S. M. Zakeeruddin, K. H. Dahmen,
M. Van Schilfgaarde, A. Walsh, Nano Lett. 2014, 14, 2584. A. Hagfeldt, M. Grätzel, N. Mohammad K, Energy Environ. Sci.
[227] G. Niu, W. Li, F. Meng, L. Wang, H. Dong, Y. Qiu, J. Mater. Chem. 2015, 8, 2946.
A 2014, 2, 705. [256] Z. Li, Z. Zhu, C.-C. Chueh, J. Luo, A. K. Y. Jen, Adv. Energy Mater.
[228] A. Poglitsch, D. Weber, J. Chem. Phys. 1987, 87, 6373. 2016, 6, 1601165.
[229] F. Hao, C. C. Stoumpos, Z. Liu, R. P. H. Chang, M. G. Kanatzidis, [257] J. Xu, O. Voznyy, R. Comin, X. Gong, G. Walters, M. Liu,
J. Am. Chem. Soc. 2014, 136, 16411. P. Kanjanaboos, X. Lan, E. H. Sargent, Adv. Mater. 2016, 28, 2807.
[230] B. R. Vincent, K. N. Robertson, T. S. Cameron, O. Knop, Can. J. [258] Y.-D. Lin, B.-Y. Ke, K.-M. Lee, S. H. Chang, K.-H. Wang,
Chem. 1987, 65, 1042. S.-H. Huang, C.-G. Wu, P.-T. Chou, S. Jhulki, J. N. Moorthy,

Adv. Energy Mater. 2017, 7, 1700264 1700264  (26 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

Y. J. Chang, K.-L. Liau, H.-C. Chung, C.-Y. Liu, S.-S. Sun, T. J. Chow, M. B. Johnston, A. petrozza, L. Herz, H. Snaith, Energy Environ.
ChemSusChem 2016, 9, 274. Sci. 2014, 7, 3061.
[259] J. Liu, Y. Wu, C. Qin, X. Yang, T. Yasuda, A. Islam, K. Zhang, [273] F. Hao, C. C. Stoumpos, D. H. Cao, R. P. H. Chang,
W. Peng, W. Chen, L. Han, Energy Environ. Sci. 2014, 7, 2963. M. G. Kanatzidis, Nat. Photon. 2014, 8, 489.
[260] F. Zhang, X. Liu, C. Yi, D. Bi, J. Luo, S. Wang, X. Li, Y. Xiao, [274] B.-W. Park, B. Philippe, X. Zhang, H. Rensmo, G. Boschloo,
S. M. Zakeeruddin, M. Grätzel, ChemSusChem 2016, 9, 2578. E. M. J. Johansson, Adv. Mater. 2015, 27, 6806.
[261] C. Huang, W. Fu, C.-Z. Li, Z. Zhang, W. Qiu, M. Shi, P. Heremans, [275] Y. Kim, Z. Yang, A. Jain, O. Voznyy, G.-H. Kim, M. Liu, L. N. Quan,
A. K. Y. Jen, H. Chen, J. Am. Chem. Soc. 2016, 138, 2528. F. P. García de Arquer, R. Comin, J. Z. Fan, E. H. Sargent, Angew.
[262] Y. Han, S. Meyer, Y. Dkhissi, K. Weber, J. M. Pringle, U. Bach, Chem., Int. Ed. 2016, 55, 9586.
L. Spiccia, Y.-B. Cheng, J. Mater. Chem. A 2015, 3, 8139. [276] A. J. Lehner, H. Wang, D. H. Fabini, C. D. Liman, C.-A. Hébert,
[263] K. Domanski, J.-P. Correa-Baena, N. Mine, M. K. Nazeeruddin, E. E. Perry, M. Wang, G. C. Bazan, M. L. Chabinyc, R. Seshadri,
A. Abate, M. Saliba, W. Tress, A. Hagfeldt, M. Grätzel, ACS Nano Appl. Phys. Lett. 2015, 107, 131109.
2016, 10, 6306. [277] F. Wei, Z. Deng, S. Sun, F. Zhang, D. M. Evans, G. Kieslich,
[264] Y.-Y. Zhang, S. Chen, P. Xu, H. Xiang, X.-G. Gong, A. Walsh, S. Tominaka, M. A. Carpenter, J. Zhang, P. D. Bristowe,
S.-H. Wei, arXiv:1506.01301 [cond-mat.mtrl-sci] 2015. A. K. Cheetham, Chem. Mater. 2017, 29, 1089.
[265] A. Buin, R. Comin, J. Xu, A. H. Ip, E. H. Sargent, Chem. Mater. [278] C. N. Savory, A. Walsh, D. O. Scanlon, ACS Energy Lett. 2016, 1,
2015, 27, 4405. 949.
[266] G. P. Nagabhushana, R. Shivaramaiah, A. Navrotsky, Proc. Natl. [279] A. H. Slavney, T. Hu, A. M. Lindenberg, H. I. Karunadasa, J. Am.
Acad. Sci. U.S.A. 2016, 113, 7717. Chem. Soc. 2016, 138, 2138.
[267] E. T. Hoke, D. J. Slotcavage, E. R. Dohner, A. R. Bowring, [280] H. C. Weerasinghe, Y. Dkhissi, A. D. Scully, R. A. Caruso,
H. I. Karunadasa, M. D. McGehee, Chem. Sci. 2015, 6, 613. Y.-B. Cheng, Nano Energy 2015, 18, 118.
[268] C. G. Bischak, C. L. Hetherington, H. Wu, S. Aloni, D. F. Ogletree, [281] A. Binek, M. L. Petrus, N. Huber, H. Bristow, Y. Hu, T. Bein,
D. T. Limmer, N. S. Ginsberg, Nano Lett. 2017, 17, 1028. P. Docampo, ACS Appl. Mater. Interfaces 2016, 8, 12881.
[269] W. Rehman, D. P. McMeekin, J. B. Patel, R. L. Milot, [282] J. M. Kadro, N. Pellet, F. Giordano, A. Ulianov, O. Müntener,
M. B. Johnston, H. J. Snaith, L. M. Herz, Energy Environ. Sci. 2017, J. Maier, M. Grätzel, A. Hagfeldt, Energy Environ. Sci. 2016, 9, 3172.
10, 361. [283] J. Gong, S. B. Darling, F. You, Energy Environ. Sci. 2015, 8, 1953.
[270] B. Hailegnaw, S. Kirmayer, E. Edri, G. Hodes, D. Cahen, J. Phys. [284] J. M. Ball, M. M. Lee, A. Hey, H. J. Snaith, Energy Environ. Sci.
Chem. Lett. 2015, 6, 1543. 2013, 6, 1739.
[271] C. Kagan, D. Mitzi, C. Dimitrakopoulos, Science 1999, 286, 945. [285] L. Song, A. Abdelsamie, C. J. Schaffer, V. Körstgens, W. Wang,
[272] N. K. Noel, S. D. Stranks, A. Abate, C. Wehrenfennig, S. Guarnera, T. Wang, E. D. Indari, T. Fröschl, N. Hüsing, T. Haeberle, Adv.
A. Haghighirad, A. Sadhanala, G. E. Eperon, S. K. Pathak, Funct. Mater. 2016, 26, 7084.

Adv. Energy Mater. 2017, 7, 1700264 1700264  (27 of 27) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Das könnte Ihnen auch gefallen