Controllable perovskite crystallization at a gas-solid interface for hole conductor-free solar cells with steady power conversion efficiency over 10%

Article abstract of DOI:10.1021/ja509245x

Depositing a pinhole-free perovskite film is of paramount importance to achieve high performance perovskite solar cells, especially in a heterojunction device format that is free of hole transport material (HTM). Here, we report that high-quality pinhole-free CH₃NH₃PbI₃ perovskite film can be controllably deposited via a facile low-temperature (<150 °C) gas-solid crystallization process. The crystallite formation process was compared with respect to the conventional solution approach, in which the needle-shaped solvation intermediates (CH₃NH₃PbI₃·DMF and CH₃NH₃PbI₃·H₂O) have been recognized as the main cause for the incomplete coverage of the resultant film. By avoiding these intermediates, the films crystallized at the gas-solid interface offer several beneficial features for device performance including high surface coverage, small surface roughness, as well as controllable grain size. Highly efficient HTM-free perovskite solar cells were constructed with these pinhole-free CH₃NH₃PbI₃ films, exhibiting significant enhancement of the light harvesting in the long wavelength regime with respect to the conventional solution processed one. Overall, the gas-solid method yields devices with an impressive power conversion efficiency of 10.6% with high reproducibility displaying a negligible deviation of 0.1% for a total of 30 cells.

Full text of DOI:10.1021/ja509245x

Article
pubs.acs.org/JACS
Controllable Perovskite Crystallization at a Gas−Solid Interface for
Hole Conductor-Free Solar Cells with Steady Power Conversion
Efficiency over 10%
Feng Hao,† Constantinos C. Stoumpos,† Zhao Liu,‡ Robert P. H. Chang,‡ and Mercouri G. Kanatzidis*,†
‡
†Department of Chemistry, and Department of Materials Science and Engineering, and Argonne-Northwestern Solar Energy
Research (ANSER) Center, Northwestern University, 2145 Sheridan Road, Evanston, Illinois 60208, United States
S
* Supporting Information
ABSTRACT: Depositing a pinhole-free perovskite film is of paramount importance
to achieve high performance perovskite solar cells, especially in a heterojunction
device format that is free of hole transport material (HTM). Here, we report that
high-quality pinhole-free CH3NH3PbI3 perovskite film can be controllably deposited
via a facile low-temperature (<150 °C) gas−solid crystallization process. The
crystallite formation process was compared with respect to the conventional solution
approach, in which the needle-shaped solvation intermediates (CH3NH3PbI3·DMF
and CH3NH3PbI3·H2O) have been recognized as the main cause for the incomplete
coverage of the resultant film. By avoiding these intermediates, the films crystallized
at the gas−solid interface offer several beneficial features for device performance
including high surface coverage, small surface roughness, as well as controllable grain
size. Highly efficient HTM-free perovskite solar cells were constructed with these
pinhole-free CH3NH3PbI3 films, exhibiting significant enhancement of the light
harvesting in the long wavelength regime with respect to the conventional solution processed one. Overall, the gas−solid method
yields devices with an impressive power conversion efficiency of 10.6% with high reproducibility displaying a negligible deviation
of 0.1% for a total of 30 cells.
also are responsible for accepting electrons from the perovskite
and transporting them to the collecting electrode.4,13,18,19
Second is the meso-superstructured architecture, where an inert
scaffold of alumina is used instead of the aforementioned
semiconductor layer.3,20 In these device architectures, excited
electrons should be only transported through the perovskite
due to the insulating nature of the alumina scaffold. Third is a
planar thin-film heterojunction architecture lacking the
mesoporous scaffold layer that has recently reported to have
the highest conversion efficiencies, where the perovskite layer is
utilized both as light absorber and as long-range transporter of
both charge species.15,16,21 The final is a hole-conductor free
architecture, where the removal of a separate hole transporter
material (HTM) layer means that all hole transport occurs
through the perovskite itself.6,7 The thin-film planar and
mesostructured approaches should ultimately provide the
highest efficiencies, because the interfacial area at these
heterojunctions, where nonradiative recombination losses are
most likely to occur, is significantly reduced.22 While most
research attention has been drawn into pursuing higher
conversion efficiency of these perovskite solar cells via judicious
interfacial engineering,23−25 the material cost should also be
considered as an important factor governing the practical
application of this promising technology, where the HTM-free
INTRODUCTION
■
Organic−inorganic hybrid perovskites have gained considerable
research attention during the past two years and have
revolutionized the prospects of emerging photovoltaic tech-
nologies, in forms of both light harvesters1−4 and hole transport
materials.5−7 These organic−inorganic hybrid perovskite
compounds adopt the ABX3 perovskite structure, which
consists of a network of corner-sharing BX6 octahedra, where
the B atom is a divalent metal cation (typically Ge2+, Sn2+, or
Pb2+) and X is a monovalent anion (typically F−, Cl−, Br−, or
I−); the A cation is selected to balance the total charge, and it
can be a Cs+ or a small molecular species.8−10 Such perovskites
afford several important features including excellent optical
properties that are tunable by controlling the chemical
compositions,11,12 ambipolar charge transport,13 and long
electron and hole diffusion lengths.14,15 Solar cells based on
these perovskites have gained enormous significance and
reached a power conversion efficiency close to 20%,1,3,4,16
approaching the efficiency of commercialized c-Si solar cells
and thin film photovoltaic solar cells such as CdTe and
Cu2ZnSn(Se,S)4.17
Currently, four main device architectures have been explored
in these perovskite solar cells. First is the so-called “sensitizing”
architecture, where the perovskite simply acts as a light
absorber coating a semiconducting mesoporous layer of TiO2
or ZnO nanoparticles. In addition to acting as a structure-
directing support, these semiconducting mesoporous scaffolds
Received: September 8, 2014
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja509245x | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 1. (a) Schematic gas−solid crystallization procedure for the pinhole-free perovskite film. (b) Crystal formation scheme for the perovskite
material via conventional solution process, where the solvated CH3NH3PbI3·DMF intermediate phase dominated the needle-shaped and incomplete
coverage of the obtained film with pinholes. As indicated by the red arrow, our low-temperature gas−solid crystallization process effectively addresses
this problem, leading to a high-quality pinhole-free perovskite film. (C) X-ray diffraction (XRD) patterns of the gas−solid crystallized perovskite
films under 150 °C with a different reaction time. Patterns of a sequential solution processed perovskite film and the simulated standard one are also
presented for comparison.
devices possess a substantial advantage by eliminating the need
to employ an additional organic or inorganic HTM layer for the
purpose of hole extraction.
the PbI2 is first spin-coated from solution in dimethylforma-
mide (DMF) or dimethyl sulfoxide (DMSO) onto the
mesoporous titania film and subsequently transformed into
the perovskite by dipping into a solution of methylammonium
iodide (MAI), typically in isopropanol. Formation of perovskite
is instantaneous within the mesoporous host upon contacting
the two components.26,27 By following this fabrication protocol
for perovskite solution coating on mesoporous TiO2 substrates,
power conversion efficiencies up to 10% have been realized,
reported by several independent research groups.28−30
However, two major problems were found for sequential
deposition on planar substrate: one is incomplete conversion of
PbI2 to perovskite, and the other is uncontrolled perovskite
crystal growth as well as surface morphology,31,32 thus leading
to the poor reproducibility of the perovskite films as well as a
large variation on the photovoltaic response. A vapor
deposition method was also proposed to deposit extremely
uniform perovskite film with crystalline platelets at nanometer
scale.33,34 However, this is a high energy-consuming process
with respect to the cost-effective solution process and also
limits the substrate choice.35 More recently, a vapor-assisted
solution process (VASP) has been developed for the deposition
of perovskite absorbing layer, which is like a combination of the
In this regard, Etgar et al. first reported that CH3NH3PbI3
perovskite nanocrystals could act both as a light harvester as
well as a hole transporter in solar cells comprising a mesoscopic
CH3NH3PbI3/TiO2 heterojunction.6 This HTM-free hetero-
junction architecture bears a much simpler structure beneficial
for reducing the cost and lowering the charge recombination
possibility by minimizing the interfacial area. With efforts on
the perovskite layer and back contact interfacial modification, a
power conversion efficiency of 8.04% has been achieved by
developing an efficient perovskite deposition method with a
combination of different solvents.7 It has been proposed that
the HTM-free heterojunction architecture’s lower performance
most likely arises from pinhole formation, incomplete perov-
skite film coverage resulting in low-resistance shunting paths, as
well as reduced light absorption in the solar cell. Any pinholes
formed in the perovskite layer can compromise the efficiency
by causing shunt paths in the device. Currently, the most
commonly applied deposition methods for the perovskites layer
are the sequential deposition from solution4 and dual-source
vapor evaporation.21 In a typical sequential deposition process,
B
dx.doi.org/10.1021/ja509245x | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
aforementioned two methods.31 Conversion efficiency over
12% has been obtained in a planar architecture with Spiro-
OMeTAD as HTM. It has also been observed that PbI2
impurity easily formed in the adjoining perovskite grains in
this VASP process, which was probably due to the weakly
orientation polarization between the organic and inorganic
species in the hybrid perovskite.36 Obviously, it is worth
developing more efficient and reproducible HTM-free meso-
scopic heterojunction perovskite solar cells using more
controllable and versatile deposition methods for the perovskite
layer.
The method employed here takes advantage of the fact that
CH3NH3I is a stable adduct of two gases, CH3NH2 and HI.
Despite the fact that the total decomposition temperature of
the CH3NH3I salt lies in the 280−350 °C range, as determined
by thermogravimetric analysis (see Figure S1 in the Supporting
Information), the vapor pressure of the individual gases is
sufficient to generate enough molecules of CH3NH2 and HI in
the gaseous phase for the PbI2 conversion to CH3NH3PbI3 to
proceed at a controlled rate. This can be further tuned by
careful control of the decomposition temperature (100−150 °C
was used in this report). The method offers several positive
features; because the reaction proceeds at the solid−gas
interface, the presence of adventitious solvent molecules is
strictly avoided. Moreover, the method promises control over
the grain size of the crystallites during the crystal growth
process with particle size and film surface coverage being
tunable parameters based on the evaporation time and
temperature.
The composition and morphology evolution of the perov-
skite film along the gas−solid crystallization process are first
investigated by a series of X-ray diffraction (XRD) and scanning
electron microscopy (SEM) measurements with different
reaction time. All of the films were deposited on mesoporous
TiO2 electrodes to ensure the condition comparable to that of
the working device. As shown in Figure 1c, the strong Bragg
peaks at 14.08°, 28.41°, 31.85°, and 43.19° can be assigned to
(110), (220), (310), and (330) of the CH3NH3PbI3 crystal,
corresponding to a tetragonal I4cm crystal structure of halide
perovskite with high crystallinity.2,37 The XRD pattern of a
sequential solution processed perovskite film is also presented
for comparison. Notably, a signature peak at 12.65° indicates
the incompletely converted PbI2.4 Such PbI2 impurity has been
widely observed even in the highly efficient devices from the
sequential deposition method.4,38 The absence of this peak in
the present gas−solid crystallized perovskite film suggests
complete conversion of the PbI2 precursor. Interestingly, the
XRD pattern of the as-deposited PbI2 films from DMF solution
suggested the presence of a slightly modified PbI2 phase with
larger lattice parameter (2θ = 13.24°) as compared to the
calculated pattern of PbI2 (2θ = 12.65°).39 A possible
explanation is that PbI2 is composed of planes of Pb2+ ions
sandwiched between adjacent layers of hexagonally arranged
iodide ions forming extended 2D layers held together by van
der Waals interactions. The interlayer space of ∼2.66 Å is large
enough to host a variety of small molecules such as the remnant
DMF molecules from the solution processing, and it can even
form stable adducts such as PbI2·DMF.39 The ease of
intercalation of different guest molecules can result in a lattice
expansion along the c axis, and it is also responsible for the high
reactivity of PbI2 upon addition of the CH3NH3I solutions.40
Very recently, it has been shown that the DMSO solvent can
form stable solvates with CH3NH3PbI3, possibly CH3NH3PbI3·
DMSO, thus facilitating the crystallization process of the
perovskite layer.41 The crystal structure of CH3NH3PbI3·
DMSO is expected to be similar to those of CH3NH3PbI3·
DMF and CH3NH3PbI3·H2O, which are discussed below.
These intermediate phases are more reactive than PbI2 itself
and facilitate the subsequent crystallization process of the
corresponding perovskite film.
Throughout the recent literature, it is apparent that the
morphology of CH3NH3PbI3 crystallites significantly impacts
the film surface coverage and coloration of the devices as well as
the electrical properties. A major issue in the solution-based
fabricated devices, which currently hold the record efficiency
value of 19.3%,16 is the uncontrolled reproducibility of the
device performance, which affords a wide range of efficiency
distribution among devices prepared with identical fabrication
protocols. In this study, we demonstrate that high-quality
pinhole-free methylammomium lead triiodide (CH3NH3PbI3)
perovskite films can be deposited via a facile low-temperature
(<150 °C) gas−solid crystallization reaction under inert
atmosphere, but with better control on the film morphology
and chemical composition than the previously reported VASP
method. We show here that the perovskite film quality such as
the particle size and surface coverage could be easily controlled
by the film growth process. We show that the perovskite thin
films undergo a grain coarsening process that leads to a
controllable grain size. With respect to the conventional
solution process, our gas−solid crystallization process effec-
tively avoids the solvation and hydration processing or
undesirable complex structural reorganization transitions. This
enabled us to construct highly efficient and reproducible HTM-
free perovskite solar cells with these pinhole-free CH3NH3PbI3
films, exhibiting significant enhancement of the light harvesting
in the long wavelength regime, in comparison to the devices
with the solution deposited perovskite films, thus yielding a
notable conversion efficiency of 10.6% along with a negligible
standard deviation of 0.1% for a total of 30 cells.
RESULTS AND DISCUSSION
■
Perovskite Film Deposition. As shown in Figure 1a, the
CH3NH3PbI3 perovskite thin films were fabricated via a
straightforward gas−solid crystallization reaction between the
two components in a nitrogen filled glovebox, in which the PbI2
framework was first spin-cast from a DMF solution on the
mesoporous TiO2 layer, and subsequently converted to
perovskite by a gas−solid crystallization process in a closed
container. This process directly avoids the formation of the
needle-forming solvated CH3NH3PbI3·DMF intermediate
phase, as indicated by the crystal formation scheme in Figure
1b. Unlike the previously reported VASP process, direct contact
between the CH3NH3I powder and the PbI2 framework is
avoided by facing down the PbI2-infiltrated TiO2 electrodes at a
constant height of around 2 in. against the CH3NH3I powder,
which is more similar to the setup of commercial vapor
evaporation equipment.21 Our design provides excellent control
of the perovskite crystal growth rate and film coverage
including the CH3NH3I precursor loading amounts, the
atmosphere, and the reaction temperature and time. The
detailed film deposition and device fabrication procedure are
described in the Supporting Information.
The importance of controlling the interfacial layers in the
solar cell devices is apparent. First, a local interfacial order at
the titania/perovskite interface promoted by specific local
interactions has been demonstrated.42 Second, a built-in electric
C
dx.doi.org/10.1021/ja509245x | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 2. (a) Scanning electron microscope (SEM) topography, (c) cross-sectional SEM morphology, (e) atomic force microscopy (AFM)
topography of the surface of sequential solution deposited perovskite film. (b) SEM topography, (d) cross-sectional SEM, and (f) AFM morphology
of a gas−solid generated pinhole-free perovskite film.
Figure 3. SEM topography of the perovskite films with a different reaction time of (a) 0 min, (b) 15 min, (c) 30 min, (d) 45 min, and (e) 60 min,
and (f) the calculated particle size distributions. All scale bars represent 500 nm.
field generated in the ferroelectric perovskite layer has
highlighted the importance of ferroelectric domains acting as
“highways” for the transport of charge carriers.43 Under-
standably, controlling the perovskite crystal growth will have a
huge impact on controlling the interfaces between the
subsequently deposited perovskite layers, in that the roughness
of the hole conductor layer will depend on that of the
perovskite layer. This is particularly true for planar hetero-
junction devices with no hole conductor layer, where a
nonuniform perovskite layer would dramatically increase the
D
dx.doi.org/10.1021/ja509245x | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
chance of charge carrier recombination. In our gas−solid
crystallization process, the as-formed CH3NH3PbI3 film exhibits
full surface coverage and is composed of micrometer-sized
grains as shown in Figure 2b. For comparison, a sequentially
deposited CH3NH3PbI3 film was also fabricated following the
reported procedure.4 Obviously, the crystal size of the
sequential deposited CH3NH3PbI3 is roughly estimated to be
promotes a facile rearrangement of PbI2 and/or reorganization
of PbI2 framework via enhanced diffusion during film
growth.26,45 We observed that as the reaction time increases
to 30, 45, and 60 min, micrometer-scale grains are observed in
the perovskite films. In comparison to the original PbI2
intermediate phase, the perovskite film differs in both
morphology and size. The in-plane grain size of perovskite
crystals is 3−4 times the film thickness, indicating that the
growth of perovskite polycrystalline films mainly follows
normal grain growth mode.46 More interestingly, during
crystallization of the perovskite film, a grain coarsening process
is observed through motion of grain boundaries resulting in the
shrinkage and elimination of small grains, which, in turn, results
in an increase in the average size of the remaining grains
(Figure 3f). Such grain coarsening is mainly driven by the
reduction in the total grain boundary area and in the
corresponding reduction in the total grain boundary energy
that accompanies the area decrease.46 It should pointed out that
the calculated grain sizes are in fair agreement with those
obtained from the XRD data using the peak width analysis. This
analysis was carried out for the (022) reflection at 24.54°,
which together with the (121) reflection are the only ones that
are not split in the I4cm symmetry setting. Detailed analysis and
tabulated data are provided in Supporting Information Figure
S2 and Table S1, respectively.
Crystal Structures of the Intermediate Phases. To gain
further insight into the perovskite film quality, we discuss the
crystallite formation process with respect to different fabrication
techniques. In the commonly applied one-step solution
approach, despite the fact that several solvents have been
employed for the dissolution of the CH3NH3PbI3 perovskite
such as dimethylformamide (DMF),3 γ-butyrolactone (GBL),13
and dimethyl sulfoxide (DMSO),32 a common feature is always
present: a needle-shaped morphology (Figure 4a) of the
crystallites is preceded by the formation of an intermediate
yellow phase with needle-shaped morphology. This phase is
distinctively characterized by low angle X-ray diffraction
reflections (typically 2θ < 10°), but the crystal structure has
not been elucidated. In an attempt to contribute to the better
understanding of these intermediates, we present the crystal
structures of two such compounds in the context of this work.
The first one, CH3NH3PbI3·H2O, was obtained as thin,
elongated, pale yellow needles from saturated CH3NH3PbI3
solutions in concentrated hydriodic acid (HI, 57%) at ambient
conditions. Similarly, CH3NH3PbI3·DMF was obtained as thin,
elongated, pale yellow needles from saturated solutions of
CH3NH3PbI3 in DMF, also at ambient conditions. The former
compound, CH3NH3PbI3·H2O, first mentioned by Weber,47 is
metastable and spontaneously loses its crystalline water
molecule converting to black, polycrystalline CH3NH3PbI3
upon removal of the crystals from the mother liquor. Because
of this behavior, we determined the crystal structure of the
compound at −153 °C to prevent this structural trans-
formation. The latter compound, CH3NH3PbI3·DMF, is
relatively stable in air and converts only very slowly to black
CH3NH3PbI3 upon removal of the crystals from the mother
liquor. For comparison, the DMSO−solvate variant is a
completely stable intermediate up to 100 °C.41 To highlight
the universality of the tendency of the CH3NH3PbI3 perovskite
to incorporate a neutral small molecule accompanied by a
profound change in its crystal structure, we mention the
“ammonia bleaching” process, where the black perovskite films
200
50 nm, which is much smaller as compared to that
derived from gas−solid crystallization process. Meanwhile, bare
TiO2 surfaces can be observed in the sequential deposited
perovskite films (Figure 2a), manifesting the film coverage is
not complete.
The surface roughness of the film was further monitored by
high-resolution atomic force microscopy (AFM) (Figure 2e and
f). The size of the AFM images is 3 × 3 μm2. The measured
average roughness (Ra) and root-mean-square roughness (Rq)
are 75.1 and 91.7 nm for the sequential deposited CH3NH3PbI3
film and 26.3 and 32.8 nm for the gas−solid crystallized film
regardless of its microscale grain size. Cross-sectional SEM
images (Figure 2c and d) indicate both of the two resulting
films have a comparable thickness of around 250 nm, with well-
defined grains across the film thickness. Therefore, the full
surface coverage, microscale grain size, and uniform grain
structure of the as-formed perovskite film are highly suitable for
planar heterojunction photovoltaic devices. These attractive
characteristics arise from the combination of the relative
smoothness of the preformed PbI2 film, and the effective
intercalation of CH3NH3I (or CH3NH2 + HI) vapor in it.
The crystal growth of the perovskite film is of major
importance to fabricate efficient planar heterojunction devices,
especially in the absence of the hole conductor layer. It is thus
necessary to monitor the underlying thermodynamic mecha-
nism of the crystal growth in perovskite thin-film though this
gas−solid crystallization process. The perovskite film morphol-
ogy evolution was investigated with an interval crystallization
time of 15 min at a temperature of 150 °C in the nitrogen
glovebox. As depicted in Figure 3a, the initial PbI2/DMF
intermediate film exhibits uniform polygon-shaped grains of a
few hundred nanometers, as well as scattered voids among
adjacent grains. The thickness of the initial film is around 200
nm. The particle size distributions (PSD) of the as-obtained
perovskite films were further analyzed on the basis of the
method introduced by Holzer et al.44 More details can be found
in the Supporting Information. After exposure to the CH3NH3I
gas for 15 min, it converts to CH3NH3PbI3 perovskite (without
any PbI2) from the evidence of XRD data. This indicates the
formation of CH3NH3PbI3 perovskite is greatly accelerated in
the gas−solid interface as compared to the sequential solution
deposition process, in which the unreacted PbI2 peaks can still
be seen for 1 h of reaction in isopropanol solution.4 The
accelerated crystal formation and growth rate may originate
from the minimization of the diffusion limitation of the
methylammonium iodide into the mesoporous TiO2 pores at a
gas−solid interface.
Notably, the CH3NH3PbI3 films exhibit a full coverage on the
mesoporous TiO2 layer with a median grain size of 330 nm
based on PSD analysis. The significantly enhanced grain size
with respect to the PbI2 precursor could be ascribed to volume
expansion during the crystal growth from PbI2 to perovskite by
the intercalation of CH3NH3I, as well as rearrangement of the
aggregated structure of PbI2 driven by minimizing the grain
boundary energy.31 The absence of voids in the perovskite film
is striking and suggests that the gas−solid crystallization
E
dx.doi.org/10.1021/ja509245x | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
APbI3 compounds (A = NH4, or an alkali metal). For example,
this is the stable configuration of unsolvated CsPbI3 and
RbPbI349 and the solvated forms of KPbI3·2H2O and NH4PbI3·
2H2O at ambient conditions.50 Interestingly, the latter, unlike
the CH3NH3PbI3 solvates described above, adopt the dihydrate
form rather than the monohydrate one. From the above crystal
structures, a clear cation size effect is apparent with the largest
cations stabilizing an anhydrous structure and the smallest ones
a hydrated one. The size effect is also detrimental with respect
to the formation of the NH4CdCl3-type or perovskites crystal
structure. The borderline for the two structures lies between Rb
and Cs because CsPbI3 is able to adopt the perovskite at high
temperature (∼300 °C), whereas RbPbI3 does not convert to a
perovskite throughout the temperature range, below its melting
point.49 The above discussion on the structural correlations in
APbI3 highlights why the gas−solid crystallization method is
very appealing and provides a consistent approach for the
fabrication of robust perovskite layers in solid-state solar cells as
we will show below. It avoids the complications associated with
solvation and hydration processing and undesirable structural
transitions.
Device Fabrication and Photovoltaic Performance.
For the device preparation, a 30 nm-thick TiO2 blocking layer
was first deposited on the prepatterned FTO substrates by
atomic layer deposition system. The mesoporous TiO2 layer
with a thickness of 250 nm was then deposited using a diluted
TiO2 paste in ethanol. Deposition of the CH3NH3PbI3
perovskite layer was performed in two steps under nitrogen
atmosphere. PbI2 was first spin coated on the mesoporous TiO2
layer from a 1 M DMF solution. The gas−solid crystallization
process was carried at a temperature of 150 °C in a closed
container with different reaction time to compare the grain size
effect on the final device photovoltaic performance. Finally, the
devices were completed by evaporating a thin gold layer (80
nm) on them as the back contact. For comparison, devices were
also fabricated following the reported sequential deposition
method with a reaction time of 5 min,4 as well as the one-step
deposition method from a mixed DMF solution of PbI2 and
CH3NH3I at a molar ratio of 1:1.3
Figure 5a shows the schematic image of the planar device
structure (FTO glass/blocking TiO2 layer/mesoporous TiO2
layer/perovskite/Au) without the hole conductor layer. The
light absorbance of these perovskite films was first probed by
optical diffuse-reflectance measurements prior to the photo-
voltaic characterization as shown in Figure 5b. Impressively, the
gas−solid crystallized film shows remarkably enhanced light
harvesting in the long wavelength regime (650−800 nm), as
evidenced by a much steeper absorption edge. Such an
enhancement is attributed to the combined effect of high
crystalline quality, well-defined crystal morphology, and
excellent film coverage in the gas−solid interface crystallization
process.
Figure 4. (a) SEM morphology of the perovskite film from one-step
solution method. Scale bar represents 5 μm. (b) Corresponding
current−voltage plots of the heterojunction devices in the absence of
HTM. (c) The crystal structure of CH3NH3PbI3·DMF highlighting
+
the strong DMF−CH3NH3 interaction and the NH4CdCl3-type slab
−
of the PbI3 chains. The boxed fragment indicates the relationship
−
between the 2D layers of neat PbI2 and the 1D PbI3 fragments of
CH3NH3PbI3·DMF.
can be reversibly converted to yellow upon exposure in NH3
vapors.48
The crystal structure of the H2O solvate consists of one-
dimensional (1D) {PbI3}− inorganic chains isolated from one
another by CH3NH3+ cations and water molecules crystallizing
in the monoclinic P21/m space group. The {PbI3}− fragments
adopt the “double-chain” assembly, which is reminiscent of the
NH4CdCl3 structure-type characterized by three fused [PbI6]4−
octahedra sharing a common corner. In essence, these
fragments are ribbons sliced from the layered PbI2 structure
+
(CdI2 structure-type). The inserted CH3NH3 cations stabilize
the negatively charged {PbI3}− chains through electrostatic
interactions. The guest H2O molecules provide extra
stabilization to the crystal structure by means of a strong
symmetrically bifurcated H-bond (d(N−O) = 2.829 Å). In the
case of CH3NH3PbI3·DMF, the double-chain motif still
dominates the configuration of the inorganic {PbI3}− chain,
+
+
which is surrounded by the CH3NH3 cations. The CH3NH3
cations are strongly H-bonded to a crystalline DMF molecule
forming asymmetric bifurcated bonds with amide oxygen atom
d(N−O) = 2.775 Å, d(N′−O) = 2.926 Å. The difference in the
relative strength of the H-bonds in CH3NH3PbI3·H2O and
CH3NH3PbI3·DMF appears as a sensible explanation for the
relative stability of the compounds at ambient conditions.
Another difference between the H2O and the DMF solvate is
the doubling of the unit cell in the case of the bulkier, less
symmetric DMF molecule, which possibly causes the length-
ening of the unit cell along the crystallographic c-axis.
Figure 5c shows the current−voltage (J−V) characteristics of
heterojunction perovskite solar cells based on the simple
structure FTO/blocking TiO2 layer/mesoporous TiO2 layer/
perovskite/Au with different grain size from the gas−solid
crystallization. A reference cell fabricated with the one-step
solution method was prepared for comparison, which displayed
a short-circuit current density (Jsc) of 6.28 mA/cm2, a Voc of
0.59 V, and fill factor (FF) of 54%, resulting in a poor power
conversion efficiency (PCE) of 2%, as shown in Figure 4b. As
compared to the efficient perovskite solar cells, the significant
drop in these parameters may be a result of poor coverage of
The structural motif adopted by the {PbI3}− chains is very
common in iodoplumbate chemistry, and it is observed in many
F
dx.doi.org/10.1021/ja509245x | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
With the superior perovskite film coverage of our process, all
of the devices with gas−solid crystallized films show significant
enhancement in the photovoltaic parameters. For instance, with
an average gain size of 300 nm, device A delivers a Jsc of 15.75
mA/cm2, a Voc of 0.758 V, a FF of 68%, and a total PCE of
8.7%. We also notice a dependence of the perovskite grain size
on the final photovoltaic performance. Both the Jsc and the FF
slightly increased with the grain size growing from 300 to 600
nm, whereas the Voc only marginally differed. Further increase
of the grain size to approximately 600 nm increased the
performance with a PCE of 10.6% under standard global AM
1.5 illumination. This enhancement originates from the
enhanced crystalline orientation accompanying our crystal
growth process. We also noticed that a further grain size
increase to approximately 800 nm induced a slightly declined Jsc
and Voc. As recent studies have revealed the electron−hole
diffusion length for CH3NH3PbI3 perovskite is less than 1
μm,14,15 a possible explanation for this decrease might be the
limited charge collection efficiency in the cross-film direction of
the perovskite particles.
Figure 6 shows the photovoltaic performance of the best-
performing devices with these films. Devices with the gas−solid
crystallized CH3NH3PbI3 perovskite layer with a grain size of
550 nm deliver a Jsc of 18.33 mA/cm2, a Voc of 0.818 V, a FF of
71%, corresponding to an overall PCE of 10.6% under standard
global AM 1.5 illumination, which is significantly higher than
that from a sequential solution deposited perovskite film, with a
Jsc of 15.48 mA/cm2, a Voc of 0.684 V, a FF of 60%, and a PCE
of 6.4%. Notably, improvements are due to increased Jsc, Voc,
and FF, and are likely to stem from two effects resulting from
improved perovskite film quality. First is the higher collection
capability of a higher fraction of incident photons with higher
surface coverage, increasing the current generated. Second, the
increased coverage has reduced the contact area between the
back electrode and the n-type TiO2 layer, which removes a
Figure 5. (a) Device architecture of the depleted heterojunction
perovskite solar cell (FTO glass/blocking TiO2 layer/mesoporous
TiO2 layer/perovskite/Au). (b) Normalized absorbance spectra of the
perovskite films with different grain size. Absorbance spectrum of the
sequential solution processed perovskite film is also presented for
comparison. (c) Characteristic current−voltage plots of the hetero-
junction devices with different grain sized perovskite layers.
perovskite films. The effects of poor coverage are 2-fold: first, if
there are regions with no perovskite coverage, light will pass
through without absorption, decreasing the available photo-
current; second, any insufficient coverage results in a high
frequency of “shunt paths” allowing contact between back
electrode and the TiO2 layer. Any such contact will act as a
parallel diode in the solar cell equivalent circuit, causing a drop
in Voc and FF, and accordingly power conversion efficiency.51,52
Figure 6. Comparison of the photovoltaic performance of the devices with the gas−solid generated and sequential solution deposited perovskite
films. (a) Characteristic photocurrent−voltage and dark current−voltage plots. (b) The incident photon-to-current efficiency (IPCE) spectra for the
best performing devices with the gas−solid crystallized and sequential solution deposited perovskite films. (c) Histogram of the short-circuit
photocurrent density for a bath of 30 devices with the pinhole-free perovskite films (red) and the conventional sequential deposited films (black).
(d) Comparison of the photocurrent density as a function of time for the cells held at a forward bias of maximum output power point (0.50 V for the
device with a solution processed perovskite film and 0.62 V for that with a gas−solid crystallized perovskite film). The cell was placed in the dark
prior to the start of the measurement.
G
dx.doi.org/10.1021/ja509245x | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
shunt path previously leading to leakage currents.51,52
Reduction of these shunt paths would therefore be expected
to enhance the fill factor and Voc. As evidently demonstrated by
the dark current−voltage curves in Figure 6a, devices with
perovskite films from our gas−solid crystallization process
showed much higher dark current onset as compared to devices
fabricated with the solution processed perovskite films. This is
indicative of an effectively inhibited charge recombination in
the former case.
interface states associated with solution process.57 Therefore,
we believe that the pinhole-free perovskite films from the gas−
solid crystallization process provide a more stable stage for
further efficiency improvement.
CONCLUSION
■
A fast, low temperature, gas−solid crystallization method has
been described that enables the reproducible fabrication of
high-quality perovskite films without pinholes. This allows
better control over the dynamics of nucleation and crystal grain
growth of CH3NH3PbI3. Surface morphological analysis
indicated that the perovskite films undergo a grain coarsening
process, which is mainly driven by the reduction in the total
grain boundary energy. This grain coarsening leads to a
controllable grain size with full surface coverage of the resulting
perovskite film. With respect to the conventional solution
process, our gas−solid crystallization process effectively avoids
the solvation and hydration processing or undesirable structural
transitions. Taking advantage of these high-quality pinhole-free
perovskite thin films, efficient planar heterojunction perovskite
solar cells with no hole conductor materials can be fabricated.
These films exhibit significant light harvesting enhancement in
the long wavelength region and a considerably improved
photocurrent density, as compared to conventional solution
deposited films. Improvements in terms of open-circuit voltage
and fill factor have also been achieved using these high-quality
perovskite films, primarily due to the reduction of interfacial
charge recombination with a fully covered perovskite film. A
power conversion efficiency of over 10.6% has been realized
with a negligible standard deviation. The present findings
demonstrate that this gas−solid crystallization process is
suitable for planar heterojunction perovskite solar cells, not
only for the hole conductor free devices demonstrated in the
present work, but also for other device architectures. It is
anticipated that with further interfacial engineering, such as
further facilitating the electron transport in TiO2 media and
hole extraction in perovskite solar cells, additional improve-
ments in solar cell performance up to 20% are possible in
HTM-free cells in the near future.
Figure 6b shows the incident-photon-to-current conversion
efficiency (IPCE) spectra for the perovskite cells from the gas−
solid crystallization and the sequential deposition. The
photocurrent generation starts at around 800 nm, in agreement
with the observed band gap (Eg = 1.55 eV) of CH3NH3PbI3,53
and reaches a peak IPCE value of around 80% at 502 nm for
the device with a sequential deposited film. The use of gas−
solid crystallized perovskite film gave a noticeable improvement
of the photocurrent over the whole region, especially in the
long wavelength region between 600 and 800 nm due to more
effective charge extraction and/or light harvesting. This IPCE
enhancement correlates well with the higher light absorption in
our thin films (Figure 5b). Integrating the overlap of the IPCE
spectrum with the AM 1.5G solar photon flux yielded a
photocurrent density of 17.98 and 15.17 mA/cm2 for the gas−
solid crystallization and sequential deposited perovskite film,
which are in excellent agreement with the measured photo-
current density. This confirms that any mismatch between the
simulated sunlight and the AM1.5G standard is negligibly small.
To further check the reproducibility of the high efficient
perovskite device without hole conductor layer, Figure 6c
shows the histogram of the Jsc for a batch of 30 devices with the
pinhole-free gas−solid crystallized perovskite films and the
sequential deposited films. Obviously, the average Jsc value for
the devices with gas−solid crystallized perovskite film is 18.14
0.12 mA/cm2, which is much higher than that of the
sequential solution-deposited one at 15.07
0.15 mA/cm2.
The small standard deviation indicates the good device
reproducibility.
Recently, an anomalous hysteresis in the J−V curves of
perovskite solar cells has been widely observed.54−56 The
hysteresis predominantly arises from the presence of the
perovskite absorber in the solar cell, which is also dependent on
the contact material, including p- and n-type contacts, and
device architecture.56 Monitoring the stabilized current density
at the maximum output power point was widely recommended
to provide more reliable photovoltaic performance alongside
the current−voltage scan derived power conversion efficiency.
We recorded the photocurrent density of two representative
cells held at a forward bias near their maximum output power
point as a function of time to monitor the stabilized power
output under working conditions (Figure 6d). For the device
with high-quality gas−solid crystallized perovskite film, the
photocurrent density stabilizes within seconds to approximately
16.8 mA cm−2, yielding a stabilized power conversion efficiency
around 10.5%, measured after 600 s. This indicates that out test
condition (from forward bias to short circuit scans with a
scanning rate of 0.15 V s−1) provides an accurate representation
of the cells photovoltaic performance. However, a significant
decay of the photocurrent from approximately 12.8 to 10.3 mA
cm−2 was observed with the first hundred seconds under
illumination, which then stabilized for the entire testing period.
Such a photocurrent density decay could originate from the
defects near the surface of perovskite or specifically generated
ASSOCIATED CONTENT
■
S
* Supporting Information
Detailed experimental procedures and crystalline structure
information on the salvation intermediates (CH3NH3PbI3·
DMF and CH3NH3PbI3·H2O), Figures S1−S7, and Tables S1−
S11. This material is available free of charge via the Internet at
http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
m-kanatzidis@northwestern.edu
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Electron microscopy analyses were done at the Electron Probe
Instrumentation Center (EPIC) at Northwestern University.
This research was supported as part of the ANSER Center, an
Energy Frontier Research Center funded by the U.S.
Department of Energy, Office of Science, Office of Basic
Energy Sciences, under award no. DE-SC0001059.
H
dx.doi.org/10.1021/ja509245x | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(33) Era, M.; Hattori, T.; Taira, T.; Tsutsui, T. Chem. Mater. 1997, 9,
8.
(34) Matsushima, T.; Fujita, K.; Tsutsui, T. Jpn. J. Appl. Phys. 2006,
45, 523.
(35) Graetzel, M.; Janssen, R. A. J.; Mitzi, D. B.; Sargent, E. H. Nature
2012, 488, 304.
(36) Chen, Q.; Zhou, H.; Song, T.-B.; Luo, S.; Hong, Z.; Duan, H.-S.;
Dou, L.; Liu, Y.; Yang, Y. Nano Lett. 2014, 14, 4158.
(37) Stoumpos, C. C.; Malliakas, C. D.; Kanatzidis, M. G. Inorg.
Chem. 2013, 52, 9019.
(38) Cao, D. H.; Stoumpos, C. C.; Malliakas, C. D.; Katz, M. J.;
Farha, O. K.; Hupp, J. T.; Kanatzidis, M. G. APL Mater. 2014, 2,
091101.
REFERENCES
■
(1) Kojima, A.; Teshima, K.; Shirai, Y.; Miyasaka, T. J. Am. Chem. Soc.
2009, 131, 6050.
(2) Kim, H. S.; Lee, C. R.; Im, J. H.; Lee, K. B.; Moehl, T.;
Marchioro, A.; Moon, S. J.; Humphry-Baker, R.; Yum, J. H.; Moser, J.
E.; Gratzel, M.; Park, N. G. Sci. Rep. 2012, 2, 591.
(3) Lee, M. M.; Teuscher, J.; Miyasaka, T.; Murakami, T. N.; Snaith,
H. J. Science 2012, 338, 643.
(4) Burschka, J.; Pellet, N.; Moon, S. J.; Humphry-Baker, R.; Gao, P.;
Nazeeruddin, M. K.; Gratzel, M. Nature 2013, 499, 316.
(5) Chung, I.; Lee, B.; He, J. Q.; Chang, R. P. H.; Kanatzidis, M. G.
Nature 2012, 485, 486.
(6) Etgar, L.; Gao, P.; Xue, Z. S.; Peng, Q.; Chandiran, A. K.; Liu, B.;
Nazeeruddin, M. K.; Gratzel, M. J. Am. Chem. Soc. 2012, 134, 17396.
(7) Abu Laban, W.; Etgar, L. Energy Environ. Sci. 2013, 6, 3249.
(8) Borriello, I.; Cantele, G.; Ninno, D. Phys. Rev. B 2008, 77,
235214.
(9) Mitzi, D. B. Prog. Inorg. Chem. 1999, 48, 1−121.
(10) Kagan, C. R.; Mitzi, D. B.; Dimitrakopoulos, C. D. Science 1999,
286, 945.
(39) Wakamiya, A.; Endo, M.; Sasamori, T.; Tokitoh, N.; Ogomi, Y.;
Hayase, S.; Murata, Y. Chem. Lett. 2014, 43, 711.
(40) Beckmann, P. A. Cryst. Res. Technol. 2010, 45, 455.
(41) Jeon, N. J.; Noh, J. H.; Kim, Y. C.; Yang, W. S.; Ryu, S.; Seok, S.
I. Nat. Mater. 2014, 13, 897.
(42) Roiati, V.; Mosconi, E.; Listorti, A.; Colella, S.; Gigli, G.; De
Angelis, F. Nano Lett. 2014, 14, 2168.
(43) Frost, J. M.; Butler, K. T.; Brivio, F.; Hendon, C. H.; van
Schilfgaarde, M.; Walsh, A. Nano Lett. 2014, 14, 2584.
(44) Munch, B.; Holzer, L. J. Am. Ceram. Soc. 2008, 91, 4059.
(45) Mitzi, D. B.; Prikas, M. T.; Chondroudis, K. Chem. Mater. 1999,
11, 542.
(46) Thompson, C. V. Annu. Rev. Mater. Sci. 2000, 30, 159.
(47) Poglitsch, A.; Weber, D. J. Chem. Phys. 1987, 87, 6373.
(48) Zhao, Y. X.; Zhu, K. Chem. Commun. 2014, 50, 1605.
(49) Trots, D. M.; Myagkota, S. V. J. Phys. Chem. Solids 2008, 69,
2520.
(50) Bedlivy, D.; Mereiter, K. Acta Crystallogr., Sect. B 1980, 36, 782.
(51) Docampo, P.; Snaith, H. J. Nanotechnology 2011, 22, 225403.
(52) Han, L. Y.; Koide, N.; Chiba, Y.; Mitate, T. Appl. Phys. Lett.
2004, 84, 2433.
(53) Hao, F.; Stoumpos, C. C.; Chang, R. P. H.; Kanatzidis, M. G. J.
Am. Chem. Soc. 2014, 136, 8094.
(11) Noh, J. H.; Im, S. H.; Heo, J. H.; Mandal, T. N.; Seok, S. I. Nano
Lett. 2013, 13, 1764.
(12) (a) Hao, F.; Stoumpos, C. C.; Cao, D. H.; Chang, R. P. H.;
Kanatzidis, M. G. Nat. Photonics 2014, 8, 489. (b) Lee, B. Y.;
Stoumpos, C. C.; Zhou, N. J.; Hao, F.; Malliakas, C. M.; Yeh, C. Y.;
Marks, T. J.; Kanatzidis, M. G.; Chang, R. P. H. J. Am. Chem. Soc.
2014, 136, 15379−15385.
(13) Heo, J. H.; Im, S. H.; Noh, J. H.; Mandal, T. N.; Lim, C. S.;
Chang, J. A.; Lee, Y. H.; Kim, H. J.; Sarkar, A.; Nazeeruddin, M. K.;
Gratzel, M.; Seok, S. I. Nat. Photonics 2013, 7, 487.
(14) Xing, G. C.; Mathews, N.; Sun, S. Y.; Lim, S. S.; Lam, Y. M.;
Gratzel, M.; Mhaisalkar, S.; Sum, T. C. Science 2013, 342, 344.
(15) Stranks, S. D.; Eperon, G. E.; Grancini, G.; Menelaou, C.;
Alcocer, M. J. P.; Leijtens, T.; Herz, L. M.; Petrozza, A.; Snaith, H. J.
Science 2013, 342, 341.
(54) Kim, H. S.; Park, N. G. J. Phys. Chem. Lett. 2014, 5, 2927.
(55) Sanchez, R. S.; Gonzalez-Pedro, V.; Lee, J. W.; Park, N. G.;
Kang, Y. S.; Mora-Sero, I.; Bisquert, J. J. Phys. Chem. Lett. 2014, 5,
2357.
(56) Snaith, H. J.; Abate, A.; Ball, J. M.; Eperon, G. E.; Leijtens, T.;
Noel, N. K.; Stranks, S. D.; Wang, J. T. W.; Wojciechowski, K.; Zhang,
W. J. Phys. Chem. Lett. 2014, 5, 1511.
(57) Kim, H. S.; Mora-Sero, I.; Gonzalez-Pedro, V.; Fabregat-
Santiago, F.; Juarez-Perez, E. J.; Park, N. G.; Bisquert, J. Nat. Commun.
2013, 4, 2242.
(16) Zhou, H.; Chen, Q.; Li, G.; Luo, S.; Song, T.-b.; Duan, H.-S.;
Hong, Z.; You, J.; Liu, Y.; Yang, Y. Science 2014, 345, 542.
(17) Green, M. A.; Emery, K.; Hishikawa, Y.; Warta, W.; Dunlop, E.
D. Prog. Photovoltaics 2014, 22, 701.
(18) Kim, H. S.; Lee, J. W.; Yantara, N.; Boix, P. P.; Kulkarni, S. A.;
Mhaisalkar, S.; Gratzel, M.; Park, N. G. Nano Lett. 2013, 13, 2412.
(19) Liu, D. Y.; Kelly, T. L. Nat. Photonics 2014, 8, 133.
(20) Ball, J. M.; Lee, M. M.; Hey, A.; Snaith, H. J. Energy Environ. Sci.
2013, 6, 1739.
(21) Liu, M. Z.; Johnston, M. B.; Snaith, H. J. Nature 2013, 501, 395.
(22) Eperon, G. E.; Stranks, S. D.; Menelaou, C.; Johnston, M. B.;
Herz, L. M.; Snaith, H. J. Energy Environ. Sci. 2014, 7, 982.
(23) Nayak, P. K.; Cahen, D. Adv. Mater. 2014, 26, 1622.
(24) Schulz, P.; Edri, E.; Kirmayer, S.; Hodes, G.; Cahen, D.; Kahn,
A. Energy Environ. Sci. 2014, 7, 1377.
(25) Ponseca, C. S.; Savenije, T. J.; Abdellah, M.; Zheng, K. B.;
Yartsev, A.; Pascher, T.; Harlang, T.; Chabera, P.; Pullerits, T.;
Stepanov, A.; Wolf, J. P.; Sundstrom, V. J. Am. Chem. Soc. 2014, 136,
5189.
(26) Mitzi, D. B. Chem. Mater. 2001, 13, 3283.
(27) Liang, K. N.; Mitzi, D. B.; Prikas, M. T. Chem. Mater. 1998, 10,
403.
(28) Shi, J. J.; Dong, J.; Lv, S. T.; Xu, Y. Z.; Zhu, L. F.; Xiao, J. Y.; Xu,
X.; Wu, H. J.; Li, D. M.; Luo, Y. H.; Meng, Q. B. Appl. Phys. Lett. 2014,
104, 063901.
(29) Rong, Y. G.; Ku, Z. L.; Mei, A. Y.; Liu, T. F.; Xu, M.; Ko, S. G.;
Li, X.; Han, H. W. J. Phys. Chem. Lett. 2014, 5, 2160.
(30) Aharon, S.; Gamliel, S.; El Cohen, B.; Etgar, L. Phys. Chem.
Chem. Phys. 2014, 16, 10512.
(31) Chen, Q.; Zhou, H. P.; Hong, Z. R.; Luo, S.; Duan, H. S.; Wang,
H. H.; Liu, Y. S.; Li, G.; Yang, Y. J. Am. Chem. Soc. 2014, 136, 622.
(32) Wu, Y.; Islam, A.; Yang, X.; Qin, C.; Liu, J.; Zhang, K.; Peng, W.;
Han, L. Energy Environ. Sci. 2014, 7, 2934.
I
dx.doi.org/10.1021/ja509245x | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX