Article

pubs.acs.org/JACS

Crystallization of Methyl Ammonium Lead Halide Perovskites:

Implications for Photovoltaic Applications
Yaron Tidhar,† Eran Edri,‡ Haim Weissman,† Dorin Zohar,† Gary Hodes,*,‡ David Cahen,*,‡
Boris Rybtchinski,*,† and Saar Kirmayer*,‡
Departments of †Organic Chemistry and ‡Materials and Interfaces, Weizmann Institute of Science, Rehovot 76100, Israel
*
S Supporting Information

ABSTRACT: Hybrid organic/lead halide perovskites are promising

materials for solar cell fabrication, resulting in efficiencies up to 18%.
Downloaded from pubs.acs.org by UNIV OF ROCHESTER on 08/13/18. For personal use only.

The most commonly studied perovskites are CH3NH3PbI3 and

CH3NH3PbI3−xClx where x is small. Importantly, in the latter system,
the presence of chloride ion source in the starting solutions used for the
perovskite deposition results in a strong increase in the overall charge
diffusion length. In this work we investigate the crystallization
parameters relevant to fabrication of perovskite materials based on
CH3NH3PbI3 and CH3NH3PbBr3. We find that the addition of PbCl2 to
J. Am. Chem. Soc. 2014.136:13249-13256.

the solutions used in the perovskite synthesis has a remarkable effect on

the end product, because PbCl2 nanocrystals are present during the fabrication process, acting as heterogeneous nucleation sites
for the formation of perovskite crystals in solution. We base this conclusion on SEM studies, synthesis of perovskite single
crystals, and on cryo-TEM imaging of the frozen mother liquid. Our studies also included the effect of different substrates and
substrate temperatures on the perovskite nucleation efficiency. In view of our findings, we optimized the procedures for solar cells
based on lead bromide perovskite, resulting in 5.4% efficiency and Voc of 1.24 V, improving the performance in this class of
devices. Insights gained from understanding the hybrid perovskite crystallization process can aid in rational design of the
polycrystalline absorber films, leading to their enhanced performance.

■ INTRODUCTION
Organic−inorganic solar cells based on lead halide perovskites
to be possible due to the very wide alumina bandgap and low
alumina electron affinity, so that the electrons need to move
are drawing growing interest recently due to their high through the perovskite to reach the electron conductor, TiO2.
efficiency, low fabrication cost (solution processing), and From these reports it is clear that the film deposition, involving
tunability of their optical properties.1−4 Since the introduction perovskite crystallization, has a critical influence over the film
in 2009,5 devices using these materials have evolved rapidly to morphology and, consequently, on the cell performance.15
yield efficiencies of about 18%6 (at the time of writing) for Crystallization is a complicated process and is of key
small contact area devices. The most commonly used importance in many fields.16,17 It is widely recognized that
perovskite was CH3NH3PbI3 (MAPbI3). It was found that crystallization takes place via a stepwise route, which starts from
the presence of PbCl2 in the deposition solution improved the a relatively slow nucleation step, followed by a faster growth
photovoltaic properties of the perovskite, mainly by increasing process.18−20 The nucleation can occur in the crystallizing
the electron diffusion length.7,8 Remarkably, no significant phase with no external interactions;16 this is known as
chloride concentrations are found in the end product. In almost homogeneous nucleation. Heterogeneous nucleation occurs
all studies, a mixture of PbCl2 and excess CH3NH3I (MAI) in a when the critical nucleus is formed on an impurity, an interface,
1:3 molar ratio in DMF solution was used, which yielded a very or other inhomogeneity present in the system.16,21 Heteroge-
significant improvement in the quality of the perovskite neous nucleation strongly influences the formation kinetics of
polycrystalline films and orientation of the crystallites.9 crystals and polycrystalline films22 and is in general more
The two materials systems, i.e., with and without PbCl2 efficient than homogeneous nucleation. Increasing the avail-
addition, have continued to show improved power conversion ability of heterogeneous nucleation sites interacting with a
efficiencies through various film preparation techniques and supersaturated solution can greatly increase the rate at which
solar cell designs (e.g., two-step liquid deposition on mp- crystals nucleate.22
TiO2,10 two-step vapor deposition of a thin film,11 and thermal While crystallization is very important in general, it seems to
coevaporation12,13). Furthermore, the perovskite-based solar be particularly so for the perovskite cells, where the deposition
devices performed more efficiently if the mesoporous (mp) from solution results in heterogeneous polycrystalline films,
TiO2 scaffold was replaced by mp alumina (on a compact TiO2
layer).14 This is despite the fact that injection of the Received: June 7, 2014
photogenerated electrons into the alumina is not considered Published: August 29, 2014

© 2014 American Chemical Society 13249 dx.doi.org/10.1021/ja505556s | J. Am. Chem. Soc. 2014, 136, 13249−13256
Journal of the American Chemical Society Article

Figure 1. SEM images of perovskite films deposited on dense TiO2 from DMF: (A) MAPbI3 deposited at 70 °C substrate temperature (1:1
MAI:PbI2), showing isolated crystals. (B) MAPbI3−xClx (3:1 MAI:PbCl2), deposited at 20 °C substrate temperature and placed under vacuum
overnight, showing a dense layer of smaller nonoriented crystals. (C) MAPbI3−xClx (3:1 MAI: PbCl2) deposited at 70 °C substrate temperature and
annealed at 120 °C for 45 min, showing large fused crystalline islands.

whose morphology determines the solar device performance. In placed on a hot plate (100−120 °C; annealing, customarily
this study, we aimed at elucidating the factors that define employed in perovskite cell fabrication), the film changes color
crystalline morphology of the solution-deposited perovskites, to red (occasionally), then to a deep yellow, followed by deep
especially the role of Cl ions. The crystallization of two black (over a period of minutes). The annealing process will be
perovskite systems has been studied, namely methylammonium discussed below. SEM images of such layers show that the
lead bromide (MAPbBr3) and methylammonium lead iodide morphology is dramatically different from that obtained when
(MAPbI3). We find that most of the morphological differences PbCl2 is not involved in the fabrication (Figure 1). While using
encountered during fabrication of perovskite polycrystalline PbCl2, large, interconnected domains are formed (Figure 1C),
films can be attributed to the perovskite nucleation process. needle-like (room temperature substrate, Figure S1) or mainly
Thus, controlling the extent as well as the location (i.e., in isolated separate crystals (70 °C substrate, Figure 1A) are
solution or on a surface) of the nucleation can help control the formed in the PbCl2-free synthesis. In addition, although the
morphology and uniformity of the formed layers, resulting in precursor solution contained more than 20 mol % chloride,
improved photovoltaic device performances. only traces could be detected in the fabricated film. Energy

■ RESULTS AND DISCUSSION

The Role of PbCl2 in the Nucleation and Growth
dispersive spectroscopy in the SEM and X-ray photoelectron
spectroscopy (XPS) showed that the amount of chloride in the
film is minute and usually below the sensitivity threshold of the
Processes. As previously reported,7,14,15,23 the addition of measuring technique/instrument. Thus, it appears that in the
PbCl2 to the precursor solution has a profound effect on the case of MAPbI3, the PbCl2 addition affects the crystal formation
morphology, leading to improved performance of the perov- kinetics as it is not a part of the final phase, except, possibly, at
skite photovoltaic devices. Adding PbCl2 appears to increase trace concentrations. While the diffractograms of the MAPbI3
the crystal domain size, induce a preferred orientation in the and MAPbI3−xClx films are similar (Figure S2), their
deposited film, and increase the electron diffusion length.7,24 comparison clearly shows that the “mixed halide” films exhibit
These effects can also be achieved by introducing chloride as preferred orientation at the (110) direction with the c-axis
MACl instead of PbCl2 in the precursor solution.9 The question normal to the substrate, as has been reported before.12,14 Since
remains, though, what is the role that the PbCl2 plays in the the annealing process dramatically changes the crystal
synthesis. To answer this question, we compared the morphology, we need to separate the effects of the thermal
crystallization processes with and without added PbCl2. treatment from those of the solution deposition. Figure 1C
Equimolar Iodide Case. Spin-casting an equimolar shows an SEM image of a spin-coated film using 3:1
solution of MAI and PbI2 in DMF on a flat TiO2 substrate MAI:PbCl2 DMF solution annealed at 120 °C for 45 min;
(i.e., without any mesoporous scaffold), initially yields a Figure 1B shows a spin-coated 3:1 MAI:PbCl2 film deposited
transparent film with a yellowish tint. Heating the substrate from DMF that was dried under vacuum overnight at room
to 60 °C for a few minutes transforms the film color to a dark temperature. The vacuum-dried sample was transparent at
brown (almost black). Scanning electron microscopy (SEM) deposition and changed color to black within a few minutes,
imaging reveals the formation of micrometer-long needles which agrees with its transformation to the perovskite structure.
ranging from hundreds of nanometers to several micrometers Removing the solvent from the film resulted in the formation of
in diameter (Figure S1 in the Supporting Information). The many small crystals scattered in a disordered polycrystalline
density and coverage in this system is a function of the layer. Examining the annealed film (Figure 1C) indicates that
substrate roughness and temperature, as will be discussed the small crystals formed upon deposition coalesce during the
below. A 3:1 MAI and PbI2 case was studied as well (see annealing process to yield large interconnected crystalline
below). domains. Annealing of the lead iodide perovskite films in the
Mixed Iodide-Chloride Case. A mixed halide solution is absence of PbCl2 lead to rather isolated crystalline domains
prepared by mixing a 3:1 molar ratio of MAI to PbCl2 in DMF, (Figures S1 and S5C)
using conditions typically employed in solar cell fabrication.15 Equimolar Bromide Case. While MAPbI3 perovskite can
For concentrations used in device fabrication (30−40% by contain only very small amounts of chloride in the crystal
weight), a MAI:PbCl2 ratio lower than 3:1 results in incomplete structure,25 the bromide analog has been reported to form a
dissolution of the lead salt. Upon deposition of the 3:1 solution solid solution across a wide range of chloride concentrations.26
on a substrate, a transparent yellowish film is formed. When Methylammonium lead bromide (MAPbBr3) and bromide-
13250 dx.doi.org/10.1021/ja505556s | J. Am. Chem. Soc. 2014, 136, 13249−13256
Journal of the American Chemical Society Article

Figure 2. SEM images of perovskite films deposited on dense TiO2 from DMF at 80 °C substrate temperature: (A) MAPbBr3 showing very large
isolated crystals, exhibiting surface coverage of <20%. (B) MAPbBr3−xClx showing smaller crystals covered in part by excess organic material that did
not evaporate under low pressure. (C) MAPbBr3−xClx annealed at 160 °C for 45 min, showing smaller, close-spaced crystals exhibiting surface
coverage of close to 100%.

chloride (MAPbBr3−xClx) are materials of particular interest for MAPbBr3 crystals were grown from equimolar solutions in
high open circuit voltage solar cells.27,28 Spin-casting an DMF (similar to the ones deposited on substrates for the solar
equimolar solution of MABr and PbBr2 in DMF on a flat cell production) by slow vapor mixing of alcohols into the
TiO2 substrate at room temperature, and heating to 80 °C to DMF solution. The MAPbI3 crystals were grown by a modified
remove the solvent, results in the formation of an orange- version of a published procedure.29 In the modified procedure,
colored layer. SEM images show large, tens of micrometer-sized the concentrated solution was cooled to room temperature and
cuboids with truncated facets (Figure 2A). These are scattered was placed in a close vial under isopropanol vapor. Here, the
over the surface, exhibiting a very low surface coverage slow exchange of the solvent (accompanied by decrease in
(between 10 and 50%, depending on substrate type and polarity) induced the crystallization rather than a temperature
temperature, see below). change, as in the original method (see Supporting Information
Mixed Chloride-Bromide Case. In contrast to the for details). The PbCl2 content in these solutions was gradually
“Equimolar Bromide Case” (Figure 2A) if PbCl2 is used, increased from 1 to 10 mol % by mixing an equimolar solution
mixed with MABr in a 1:3 molar ratio in DMF and deposited with a second mixed halide solution (the two solutions were
on a substrate, at first a transparent pale yellow layer is formed mixed to give the desired PbCl2 concentration, prior to the
(Figure 2B). After annealing at 150−160 °C, the transparent vapor mixing process). These experiments showed that
layer becomes deep lemon-yellow and then (in 30−60 min) increasing amounts of PbCl2 in the precursor solution for
vivid orange (Figure 2C). While XRD confirms that the both types of crystals induces the growth of a larger number of
material is MAPbBr3 (see Supporting Information), the crystals (Figures 3 and S4), and the first visual appearance of
morphology and the surface coverage of this layer are very these crystals occurs faster under what are, except for the PbCl2
different from that obtained in the pure bromide case (Figure concentration, otherwise identical conditions.
2A). SEM images show that instead of micron-sized cubes,
hundreds of nanometer-sized cubes are formed in a densely
packed manner, resulting in a better surface coverage (Figure
2C). The transmission spectra of the two films on a flat
substrate corroborate higher optical density of films formed
from PbCl2 and MABr (Figure S3). Therefore, although the
two materials, MAPbI3 and MAPbBr3, have different crystal
growth properties and different intermixing properties with Cl,
PbCl2 appears to play a similar role of increasing the surface
coverage.
To verify that the observed effects are due to the presence of
PbCl2 and are not the result of the excess methylammonium
Figure 3. Crystallization product of MAPbBr3 in vials containing
salts in the precursor solution, the deposition processes were increasing amounts of the mixed halide precursor solution. The
repeated using 3:1 MAI:PbI2 and 3:1 MABr:PbBr2. SEM number of the crystals increases with PbCl2 concentration and the size
images of the resulting films (Figure S5) reveal that the increase of the crystals decreases.
in organic salt content results in morphological changes (in
comparison to equimolar) systems less significant than in the
presence of PbCl2.
Overall, the PbCl2 addition leads to a much more Table 1. Summary of the Results of Crystallization
pronounced decrease in crystal size, the increase in number Experiments, In Which Different Amounts of PbCl2 Were
of crystals, and, ultimately, in better surface coverage. We Added to a Concentrated Perovskite Precursor Solution
anticipated that this may be due to heterogeneous nucleation
induced by the less soluble PbCl2, remaining as particles in percent of PbCl2 out of all Pb 0% 1% 2% 5% 10%
precursor in the solution.
solution. To test this hypothesis, we studied the effect of PbCl2 number of single crystals formed: 1 6 10 10 16
addition on the growth of perovskite single crystals in systems MAPbBr3
that allow the effect to be observed visually (the procedures are number of single crystals formed: 1 1 3 4 5
described in detail in the Supporting Information). In short, MAPbI3

13251 dx.doi.org/10.1021/ja505556s | J. Am. Chem. Soc. 2014, 136, 13249−13256

Journal of the American Chemical Society Article

These observations are consistent with the idea that PbCl2 under high vacuum, the solvent sublimed and nanoparticles
provides heterogeneous nucleation sites for the crystallization ranging from 3 to 20 nm in diameter were observed to be
of perovskite crystals. The solubility of lead halides in organic uniformly spread within amorphous material over the entire
solvents increases with increase in the ionic radius of the halide, imaged areas (Figure 4A,B). No such particles were detected in
due to the decrease in Pb-halide bond strength (based on lead- the 1:1 MAI:PbI2 case. Electron diffraction patterns (Figure 4C
halide, Pb-X, bond strength; Table 2) and better solvation of and Table 3) indicated that these are PbCl2 nanoparticles,

Table 2. Lead-Halide Bond Strengths (values from ref 30) Table 3. Comparison of the Measured d-Spacings (in Å) to
and Solubilities (at room temperature) the ICSD (X-ray) Diffraction Data for PbCl2 and PbI2
bond D (kJ/mol) solubility in DMF (g/mL) PbI2 PbCl2 e-diffraction
Pb−Cl 243 0.08 ± 0.01 6.89 6.16 6.21
Pb−Br 201 0.17 ± 0.02 3.94 3.75 3.42
Pb−I 142 0.36 ± 0.01 3.49 2.92 3.04
2.27 2.64
2.20 2.20 2.25
larger ions with diffuse charge density. Therefore, PbCl2 is the 2.00 2.10 2.12
least soluble Pb-halide in the DMF solutions. We note that
there could be also a kinetic effect, i.e., slower dissolution of
PbCl2 in DMF that can lead to the presence of PbCl2 particles. supporting the occurrence of heterogeneous nucleation induced
However, in all experiments described here, a visibly clear by PbCl2. These observations were also corroborated using
solution was used for spin-coating. Furthermore, using a 2-week dynamic light scattering (DLS), revealing presence of ca. 10 nm
aged solution yielded similar film morphology as a fresh particles in the filtered precursor solutions containing PbCl2.
solution. Also, filtering the solution (through a 0.22 μm filter) We note that it is possible that the initial PbCl2 nuclei grow to
did not affect the morphology. We envisaged that due to its some extent (nucleating their own growth). However, these
lower (or slower) solubility, small PbCl2 nanoparticles are crystals would continue to react with the MAI, also leading to
present in the DMF precursor solutions, inducing the enhanced growth of MAPbI3−xClx phase due to heterogeneous
nucleation of perovskite crystals. Larger number of nucleation nucleation.
sites eventually leads to smaller crystals and better coverage The Effect of Substrate Temperature and Texture.
upon spin-coating. Surface roughness and temperature of the substrate are also
Direct evidence for the involvement of PbCl2 nanoparticles important in determining the efficiency and rate of the
in the crystallization process comes from cryo-TEM studies of nucleation and growth processes. The effect of the substrate
the precursor solution in DMF. In a typical experiment, 5 μL temperature as well as of the annealing temperature on the
3:1 MAI:PbCl2 DMF solution or 1:1 MAI:PbI2 DMF solution morphology of the crystalline films were reported before,
were spread over TEM grids and vitrified by plunging the grid although these effects were not correlated with the crystal-
into liquid nitrogen. Importantly, in cryo-TEM sample lization pathway.15,33,34 While examining the formation of
preparations, vitrification, taking place upon instantaneous MAPbBr3 on mesoporous Al2O3, we noted that the substrate
cooling, is known to capture the solution-phase state, temperature is directly correlated to the surface coverage of the
preventing processes occurring upon slower cooling or polycrystalline film (better coverage at higher temperatures).
drying.31,32 At first, no indication of nanoparticles was observed, To study the thermal effect, the surface temperature of the
possibly due to low contrast difference between the solution substrate was monitored using a noncontact thermometer (see
and the particles. However, in the 3:1 MAI: PbCl2 case, when Supporting Information) immediately before spin-coating.
the sample was allowed to warm up to 153 K (from 77 K) Figure 5 shows SEM images of MAPbBr3 after deposition on

Figure 4. (A) TEM image of freeze-dried perovskite MAPbI3−xClx precursor solution, showing an amorphous film (probably a mixture of the
starting materials) with embedded PbCl2 nanocrystals. (B) Higher magnification TEM image of the same freeze-dried perovskite precursor solution
used for the e-diffraction shown in (C).

13252 dx.doi.org/10.1021/ja505556s | J. Am. Chem. Soc. 2014, 136, 13249−13256

Journal of the American Chemical Society Article

Figure 5. SEM images of MAPbBr3 films formed on the surface of (A) mesoporous Al2O3 at room temperature; covered area 25% and (B)
mesoporous Al2O3 heated to 60 °C; covered area 41%, (C) on the surface of dense TiO2 heated to 60 °C; covered area 10%. All samples were
annealed at 100 °C for 30 min after coating.

substrates at different temperatures ranging from 25 to 60 °C, morphology, however the cubic crystals return to the tetragonal
followed by annealing at 100 °C for 30 min. The latter is phase upon cooling.29,35
commonly applied during solar cell fabrication in order to When analyzing the combined effects of the process
remove volatiles; it does not significantly influence the parameters on the dynamics of crystal growth, one should
morphology of MAPbBr3. The size of the crystals decreases also take into account the type of surface on which the material
and the number density increases as the substrate temperature is deposited. The roughness and inhomogeneity of the
increases. Thus, the substrate surface temperature during the substrate may allow for lower energy pathways for nucleation,
casting of the film affects the degree of surface-induced thus increasing the probability to overcome the energy barrier
nucleation: as the temperature is increased, so does the number for nucleation and resulting in a better surface coverage. Thus,
of nucleation events,16,22 resulting in a larger number of smaller it is not surprising that changing the substrate from the dense,
crystals and better surface coverage. It is typical that the relatively flat TiO2, to a mesoporous one (either mp-TiO2 or
nucleation process has a higher activation energy and therefore mp-Al2O3) results in a dramatic increase in the number of
slower rate than the following growth process. Temperature crystallites formed on the surface (Figure 5A,B vs C) and better
increase leads to faster nucleation and, hence, higher number of surface coverage. We note that apart from the surface area and
crystals, resulting in denser, more uniform films. morphology, surface−solution interactions and surface wetting
A similar experiment was carried out with MAPbI3 films, can also influence the nucleation process. Addressing these
where the surface temperature at the casting stage was varied interactions is a subject of ongoing research.
between 20 and 60 °C. However, in this case, the nucleation To further address the influence of the annealing on the film
and growth had to be separated (see below). Since the surface- morphology and composition (loss of Cl), EDS analysis of
induced nucleation is temperature sensitive, it is restricted to MAPbI3−xClx films annealed at 120 °C for 30, 45, and 60 min
the initial deposition time (from the moment of solution was performed (Table S1). It shows that the Cl content in the
film is reduced with annealing time, which is also supported by
casting until the substrate cools down). Fast growth is achieved
time-dependent XRD data of the annealing process (Figure
by placing the samples under vacuum for the removal of the
S9). Furthermore, EDS analysis of the evaporated organic salt,
solvent, without further annealing (to avoid coarsening and
captured in an in situ study of the annealing process by air-SEM
additional nucleation). This allows us to decouple the surface
(Figure S8), identifies it as MACl. Thus, the annealing process
nucleation from any other form of nucleation and from the appears to remove the excess organic precursor in the form of
crystal fusion (coalescence) that takes place during the MACl (leading to loss of Cl) and brings about crystal
annealing. SEM images of the resulting films are presented in coalescence, forming large crystalline domains. We believe
Figure 6. As before, the number of the crystallites increased that these two processes may lead to the improved charge
dramatically with increase in surface temperature. However, in carrier diffusion in the annealed material, as observed for
addition to the increase in crystallite density, the increase in the devices based on MAPbI3−xClx.7
substrate temperature also affects the directionality of the To demonstrate how understanding of the parameters
crystal growth. Casting the precursor solution on a substrate regulating the film formation is implemented in a solar cell
that is at room temperature and removing the solvent by low fabrication, high-voltage devices were fabricated from
pressure result in the crystals growing with the c-axis parallel to MAPbBr 3−xCl x using the optimal (i.e., best coverage)
the surface. When a similar experiment is done on a hot surface, crystallization conditions as described above. The full details
more nucleation sites are formed on the surface, which results of the device fabrication are given in the Supporting
in a large number of crystallites with different orientation. This Information. In short, a 40.5 wt % solution of MAPbBr3−xClx
effect can be attributed to the lower symmetry (tetragonal I4/ was prepared by mixing a 1:3 molar ratio of PbCl2 and MABr in
mcm) of the MAPbI3 crystals29,35 that may allow for a preferred DMF. Planar TiO2 substrates were dried on a hot plate set to
growth facet, as opposed to the high symmetry cubic MAPbBr3 110 °C for 10 min. The temperature of the surface was
that will grow homogeneously regardless of the nucleation monitored by a noncontact thermometer to be 80 °C at the
origin. Furthermore, MAPbI3 has a reversible tetragonal-to- time of perovskite solution-casting that involves spin-coating.
cubic phase transition at 54 °C35 that can affect the growth The deposited layers were then annealed in air at 150 °C for 50
pattern of the crystal. This may influence the overall min. The devices were completed by depositing an organic hole
13253 dx.doi.org/10.1021/ja505556s | J. Am. Chem. Soc. 2014, 136, 13249−13256
Journal of the American Chemical Society Article

Figure 6. SEM images of MAPbI3 films, spin-coated on the surface of dense TiO2 at different substrate temperatures (solvent removed by low
pressure, no annealing): substrate at (A) 20 °C; (B) 40 °C; (C) 50 °C; and (D) 60 °C. (E) Optical transmission measurements of MAPbI3
perovskite films on dense TiO2 layer, prepared by spin-coating on a substrate at different temperatures. The transmission spectra are reflection
corrected and normalized to 100%.

conductor (CBP/LiTFSI),24,27 and subsequent keeping in dry

air for 12 h, followed by evaporating 80 nm thick gold layer as a
■ CONCLUSIONS
Understanding crystallization processes pertinent to the
back contact. The solar cell device characteristics are given in formation of polycrystalline perovskite films allowed for a
Figure 7. Controlling the layer morphology and improving its rationalization of the perovskite film preparation. By examining
quality enabled a device exhibiting an open circuit voltage of the effect of the different deposition processes in the context of
1.24 V, short circuit current of 7.8 mA cm−2, FF of 56%, and crystal nucleation, we determined the role of each step in the
5.4% efficiency. This device exhibited nearly double the current preparation. It is apparent that mesoporous scaffolds offer more
density and substantially higher fill factor then previously nucleation sites on the surface and hence contribute to a more
reported MAPbBr3−xClx high-voltage cells,28 although at the uniform layer than flat substrates. Elevated temperature is
cost of a reduced open circuit voltage. Importantly, a needed to overcome the energy barrier for nucleation on any
mesoporous (mp) scaffold was not required since high quality given substrate. We conclude from synthesis of single crystals of
of the perovskite film was ensured by the proper crystallization MAPbI3 and MAPbBr3 that PbCl2 plays a key structural role in
conditions. Very recently a MAPbBr3 solar cell with VOC up to the crystallization process of these materials. We have
1.40 V and 6.7% efficiency has been reported (based on a mp- established that due the low solubility of PbCl2 in DMF,
TiO2 scaffold).36 some fraction of it remains as inhomogeneities (nanoparticles)
13254 dx.doi.org/10.1021/ja505556s | J. Am. Chem. Soc. 2014, 136, 13249−13256
Journal of the American Chemical Society

■
Article

ACKNOWLEDGMENTS
We thank the Leona M. and Harry B. Helmsley Charitable
Trust, the Israel Ministry of Science’s “Tashtiot”, and the
Weizmann-U.K. Joint Research Program (D.C., G.H.) for
partial support. D.C. holds the Sylvia and Rowland Schaefer
Chair in Energy Research. We also wish to thank the team at B-
nano: Yaniv Brami, Yonat Milstein, and Rafi de Picciotto for
their help with the in situ study of the MAPbX3 annealing. We
thank Tatyana Bendikov and Hagai Cohen for XPS analyses,
and Ronit Popovitz-Biro for electron diffraction studies.

■ REFERENCES
(1) Nazeeruddin, M. K.; Gao, P.; Grätzel, M. Energy Environ. Sci.
2014, 7, 2448−2463.
(2) Park, N.-G. J. Phys. Chem. Lett. 2013, 4, 2423−2429.
(3) Hodes, G. Science 2013, 342, 317−318.
(4) Snaith, H. J. J. Phys. Chem. Lett. 2013, 4, 3623−3630.
(5) Kojima, A.; Teshima, K.; Shirai, Y.; Miyasaka, T. J. Am. Chem. Soc.
2009, 131, 6050−6051.
(6) National Center for Photovoltaics. Best Research-cell Efficiencies
Figure 7. J−V characteristics for planar MAPbBr3−xClx-based cells Chart; National Renewable Energy Labratory: Boulder, Colorado,
under standard AM1.5G illumination (100 mW cm−2) (solid curves) USA, 2014; http://www.nrel.gov/ncpv/images/efficiency_chart.jpg.
and in the dark (dotted curves) for cell area of 0.04 cm2. The (7) Stranks, S. D.; Eperon, G. E.; Grancini, G.; Menelaou, C.;
measurements were made from −0.5 to 1.5 V in steps of 0.0 2 V with a Alcocer, M. J. P.; Leijtens, T.; Herz, L. M.; Petrozza, A.; Snaith, H. J.
waiting time of 200 ms between points. Black line: MAPbBr3−xClx Science 2013, 342, 341−344.
deposited on dense a TiO2 substrate heated to 80 °C, followed by (8) Edri, E.; Kirmayer, S.; Henning, A.; Mukhopadhyay, S.;
annealing at 150 °C for 45 min. Red line: MAPbBr3−xClx deposited on Gartsman, K.; Rosenwaks, Y.; Hodes, G.; Cahen, D. Nano Lett.
a dense TiO2 substrate heated to 30 °C, followed by annealing at 150 2014, 14, 1000−1004.
°C for 45 min. Blue line: MAPbBr3 deposited on a dense TiO2 (9) Zhao, Y.; Zhu, K. J. Phys. Chem. C 2014, 118, 9412−9418.
substrate heated to 80 °C, followed by annealing at 150 °C for 45 min. (10) Burschka, J.; Pellet, N.; Moon, S.-J.; Humphry-Baker, R.; Gao,
P.; Nazeeruddin, M. K.; Grätzel, M. Nature 2013, 499, 316−319.
(11) Chen, Q.; Zhou, H.; Hong, Z.; Luo, S.; Duan, H.-S.; Wang, H.-
H.; Liu, Y.; Li, G.; Yang, Y. J. Am. Chem. Soc. 2014, 136, 622−625.
in the solution as shown by cryo-TEM. These nanoparticles act (12) Liu, M.; Johnston, M. B.; Snaith, H. J. Nature 2013, 501, 395−
as heterogeneous nucleation sites for the perovskite crystals. 398.
This insight can explain the structural phenomena observed (13) Docampo, P.; Ball, J. M.; Darwich, M.; Eperon, G. E.; Snaith, H.
under the different conditions used to cast the films from J. Nat. Commun. 2013, 4, 1−6.
solution during the fabrication of perovskite-based photovoltaic (14) Lee, M. M.; Teuscher, J.; Miyasaka, T.; Murakami, T. N.; Snaith,
devices. Furthermore, we demonstrated that in MAPbI3−xClx H. J. Science 2012, 338, 643−647.
the annealing process removes the excess organic precursor in (15) Eperon, G. E.; Burlakov, V. M.; Docampo, P.; Goriely, A.;
the form of MACl and drives the well-dispersed crystals to Snaith, H. J. Adv. Funct. Mater. 2014, 24, 151−157.
(16) Dirksen, J. A.; Ring, T. A. Chem. Eng. Sci. 1991, 46, 2389−2427.
coalesce and form large oriented crystalline domains. These
(17) Desiraju, G. R. J. Chem. Sci. 2010, 122, 667−675.
two processes lead, in our view, to the improved charge carrier (18) Oxtoby, D. W. Acc. Chem. Res. 1998, 31, 91−97.
diffusion. We also demonstrated the added value of a rational (19) Weissbuch, I.; Lahav, M.; Leiserowitz, L. Cryst. Growth Des.
fabrication process by producing MAPbBr3−xClx cells, which 2003, 3, 125−150.
yielded 5.4% efficiency, roughly double our previously (20) Chung, S.-Y.; Kim, Y.-M.; Kim, J.-G.; Kim, Y.-J. Nat. Phys. 2009,
published efficiency for this material, although at a somewhat 5, 68−73.
reduced VOC. (21) The Theory of Transformations in Metals and Alloys; Christian, J.

■
*
ASSOCIATED CONTENT
S Supporting Information
W., Ed.; Pergamon: Oxford, 2002.
(22) Cacciuto, A.; Auer, S.; Frenkel, D. Nature 2004, 428, 404−406.
(23) Wojciechowski, K.; Saliba, M.; Leijtens, T.; Abate, A.; Snaith, H.
J. Energy Environ. Sci. 2014, 7, 1142−1147.
Additional experimental procedures, electron microscopic (24) Edri, E.; Kirmayer, S.; Mukhopadhyay, S.; Gartsman, K.; Hodes,
images, spectroscopic data, device fabrication procedures, and G.; Cahen, D. Nat. Commun. 2014, 5, 3461.
structural analysis data. This material is available free of charge (25) Mosconi, E.; Amat, A.; Nazeeruddin, M. K.; Grätzel, M.; De
via the Internet at http://pubs.acs.org.

■
Angelis, F. J. Phys. Chem. C 2013, 117, 13902−13913.
(26) Kitazawa, N.; Watanabe, Y.; Nakamura, Y. J. Mater. Sci. 2002,
AUTHOR INFORMATION 37, 3585−3587.
Corresponding Authors (27) Edri, E.; Kirmayer, S.; Cahen, D.; Hodes, G. J. Phys. Chem. Lett.
[email protected] 2013, 897−902.
[email protected] (28) Edri, E.; Kirmayer, S.; Kulbak, M.; Hodes, G.; Cahen, D. J. Phys.
Chem. Lett. 2014, 429−433.
[email protected]
(29) Stoumpos, C. C.; Malliakas, C. D.; Kanatzidis, M. G. Inorg.
[email protected] Chem. 2013, 52, 9019−9038.
Notes (30) Huheey, J. Inorganic Chemistry Principles of Structure and
The authors declare no competing financial interest. Reactivity; 4th ed.; Prentice Hall: Upper Saddle River, NJ, 1997.

13255 dx.doi.org/10.1021/ja505556s | J. Am. Chem. Soc. 2014, 136, 13249−13256

Journal of the American Chemical Society Article

(31) Cui, H.; Hodgdon, T. K.; Kaler, E. W.; Abezgauz, L.; Danino,
D.; Lubovsky, M.; Talmon, Y.; Pochan, D. J. Soft Matter 2007, 3, 945.
(32) Friedrich, H.; Frederik, P. M.; de With, G.; Sommerdijk, N. A. J.
M. Angew. Chem., Int. Ed. 2010, 49, 7850−7858.
(33) Ball, J. M.; Lee, M. M.; Hey, A.; Snaith, H. J. Energy Environ. Sci.
2013, 6, 1739−1743.
(34) Tan, K. W.; Moore, D. T.; Saliba, M.; Sai, H.; Estroff, L. A.;
Hanrath, T.; Snaith, H. J.; Wiesner, U. ACS Nano 2014, 8, 4730−4739.
(35) Baikie, T.; Fang, Y.; Kadro, J. M.; Schreyer, M. K.; Wei, F.;
Mhaisalkar, S. G.; Gratzel, M.; White, T. J. Mater. Chem. A 2013, 1,
5628−5641.
(36) Ryu, S.; Noh, J. H.; Jeon, N. J.; Kim, Y. C.; Yang, W. S.; Seo, J.
W.; Seok, S. I. Energy Environ. Sci. 2014, 7, 2614−2618.

13256 dx.doi.org/10.1021/ja505556s | J. Am. Chem. Soc. 2014, 136, 13249−13256