Atmospheric pressure chemical vapor deposition of methylammonium bismuth iodide thin films

Article abstract of DOI:10.1039/c7ta06578g

We demonstrate the atmospheric pressure chemical vapor deposition of methyl ammonium bismuth iodide ((CH₃NH₃)₃Bi₂I₉ or MA₃Bi₂I₉) films. MA₃Bi₂I₉ possesses an indirect optical bandgap of 1.80 eV and a room temperature excitonic peak at 511 nm. In contrast to recent reports, the films are n-type semiconductors with a room temperature carrier concentration of 3.36 × 10¹⁸ cm^-3 and a Hall mobility of 18 cm² V^-1 s^-1, which are superior to those of solution-processed, undoped films. The precursors used for the deposition are methylammonium iodide and bismuth iodide which are co-sublimated at 199 °C and 230 °C, respectively, in an Ar flow inside a tube furnace with a variable temperature profile. The substrate temperature is set at 160 °C, and dense polycrystalline films (～775 nm thick) are deposited. Extensive characterization combined with first-principles density functional theory calculations unravels the synthesis-structure-property relationship in these films. The degradation of properties under ambient conditions results from film oxidation with a characteristic bi-exponential decay in resistivity, signifying a fast surface oxidation followed by a slower oxidation of the bulk.

Full text of DOI:10.1039/c7ta06578g

View Article Online
View Journal
Journal of
Materials Chemistry A
Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: X. Chen, Y.
Myung, A. S. Thind, Z. Gao, B. Yin, M. Shen, S. B. Cho, P. Cheng, B. Sadtler, R. Mishra and P. Banerjee, J.
Mater. Chem. A, 2017, DOI: 10.1039/C7TA06578G.
This is an Accepted Manuscript, which has been through the
Volume
4 Number 1 7 January 2016 Pages 1–330
Royal Society of Chemistry peer review process and has been
accepted for publication.
Journal of
Materials Chemistry A
Accepted Manuscripts are published online shortly after
Materials for energy and sustainability
www.rsc.org/MaterialsA
acceptance, before technical editing, formatting and proof reading.
Using this free service, authors can make their results available
to the community, in citable form, before we publish the edited
article. We will replace this Accepted Manuscript with the edited
and formatted Advance Article as soon as it is available.
You can find more information about Accepted Manuscripts in the
author guidelines.
Please note that technical editing may introduce minor changes
to the text and/or graphics, which may alter content. The journal’s
standard Terms & Conditions and the ethical guidelines, outlined
in our author and reviewer resource centre, still apply. In no
event shall the Royal Society of Chemistry be held responsible
for any errors or omissions in this Accepted Manuscript or any
consequences arising from the use of any information it contains.
ISSN 2050-7488
PAPER
Kun Chang, Zhaorong Chang et al.
Bubble-template-assisted synthesis of hollow fullerene-like
MoS2 nanocages as a lithium ion battery anode material
rsc.li/materials-a

Page 1 of 39
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C7TA06578G
Atmospheric pressure chemical vapor deposition of methylammonium
bismuth iodide thin films
Xiao Chen1, Yoon Myung1,2, Arashdeep Thind3, Zhengning Gao3, Bo Yin3, Meikun
Shen4, Sung Beom Cho1, Peifu Cheng1, Bryce Sadtler3,4, Rohan Mishra1,3, Parag
Banerjee1,3,*
1Department of Mechanical Engineering and Materials Science, Washington University in St.
Louis, St. Louis, MO – 63130, USA
2Department of Nanotechnology and Advanced Materials Engineering, Sejong University, Seoul,
05006, Korea
3Institute of Materials Science & Engineering, Washington University in St. Louis, St. Louis,
MO – 63130, USA
4Department of Chemistry, Washington University in St. Louis, St. Louis, MO – 63130, USA
*Email ꢀ parag.banerjee@wustl.edu
Keywords: Methylammonium bismuth iodide, chemical vapor deposition (CVD), Hall
measurements, leadꢀfree photovoltaics
Abstract:
We demonstrate the atmospheric pressure chemical vapor deposition of methyl ammonium
bismuth iodide ((CH3NH3)3Bi2I9 or MA3Bi2I9) films. MA3Bi2I9 possesses an indirect optical
bandgap of 1.80 eV and a room temperature excitonic peak at 511 nm. In contrast to recent
reports, the films are nꢀtype semiconductors with a room temperature carrier concentration of
3.36 x 1018 cmꢀ3 and Hall mobility of 18 cm2/V.s; values superior to solutionꢀprocessed, undoped
1

Journal of Materials Chemistry A
Page 2 of 39
View Article Online
DOI: 10.1039/C7TA06578G
films. The precursors used for the deposition are methylammonium iodide and bismuth iodide
which are coꢀsublimated at 199 oC and 230 oC, respectively in Ar flow inside a tube furnace with
o
a variable temperature profile. The substrate temperature is set at 160 C, and dense
polycrystalline films ~ 775 nm thick are deposited. Extensive characterization combined with
firstꢀprinciples density functional theory calculations unravel the synthesisꢀstructureꢀproperty
relationship in these films. Degradation of properties in ambient results from film oxidation with
a characteristic biꢀexponential decay in resistivity, signifying a fast surface oxidation followed
by a slower oxidation of the bulk.
1
Introduction
Lead halide perovskites (CH3NH3PbI3 or MAPbI3 and its variants) are promising solarꢀcell
absorber materials with reported power conversion efficiencies that have rapidly increased from1
3.8% in 2009, to an impressive2 22.1% in 2017. The fundamental material properties in lead
halide perovskites that drive such progress are its high chargeꢀcarrier mobilities3, high optical
absorption coefficient,4 ideal and tunable bandgap5 and ultraꢀlong carrier diffusion lengths up to
a few microns.3, 6, 7
Though the reported power conversion efficiencies of lead halide perovskite solar cells
are now competitive with commercial silicon solar cells, there are two limitations preventing the
commercialization of such solar cells.8 These are (1) the environmental, device and
thermodynamic instability9, 10 of the perovskites in the presence of air, humidity, light and
electric fields; and (2), the toxicity11 due to the presence of lead. Attempts to address stability
+
limitations have included replacing the Aꢀsite organic component (CH3NH3 ) with inorganic
cations12, 13, such as Cs+ and Rb+ or by providing protective overlayers14 that prevent exposure of
2

Page 3 of 39
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C7TA06578G
lead halide perovskite films to ambient conditions. However, the toxicity due to the presence of
lead in these perovskites presents a challenge to further scaleꢀup and eventual commercialization.
Recently, bismuth15ꢀ17 (Bi3+) has drawn attention18 as a substituent to the Bꢀsite occupied by Pb2+
for ‘leadꢀfree’ perovskites,19 since it is a nonꢀtoxic 6pꢀblock element that is isoelectronic20, 21 22
with Pb2+. Reports on promising optoelectronic and photovoltaic properties of methylammonium
bismuth iodide (with formula unit (CH3NH3)3Bi2I9 and henceforth called MA3Bi2I9) with high
absorption coefficient23 and long term stability24 have been published. However, MA3Bi2I9 based
solar cells suffer from low efficiency.25 Besides the subꢀoptimal electronic structure of MA3Bi2I9
compared to MAPbI3,17 26 the diminished performance of the solarꢀcells is also attributed to the
commonly used solution deposition techniques that result in poor morphology, purity and
coverage of the films on substrate. 27, 28
Thus in this Article, we demonstrate a gasꢀphase deposition technique for MA3Bi2I9 thin
o
films using a low temperature (160 C), atmospheric pressure chemical vapor deposition
(APCVD) process. We obtain films with better control over morphology, purity and coverage
with the added advantage of rapid scaleꢀup and integration with other solar manufacturing
processes. The APCVD process occurs in a tube furnace by taking advantage of the variable
temperature gradient along the length of the tube. Specific temperature zones within the tube are
utilized to coꢀsublimate methylammonium iodide (with chemical formula unit CH3NH3I and
referred henceforth as MAI) and bismuth iodide (BiI3) into vapor form, which condenses and
reacts on the cooler, downstream side of the tube furnace. The result is an nꢀtype, dense and
polycrystalline MA3Bi2I9 film with an optical bandgap of 2.08 eV, a roomꢀtemperature free
electron carrier density of 3.36 x 1018 cmꢀ3 and Hall mobility of 18 cm2/V.s. We discuss the
3

Journal of Materials Chemistry A
Page 4 of 39
View Article Online
DOI: 10.1039/C7TA06578G
synthesisꢀstructureꢀproperty interrelationship in MA3Bi2I9 films obtained from a suite of
characterization techniques and firstꢀprinciples density functional theory (DFT) calculations.
2
Experiment
2.1 Methylammonium iodide preparation
Methylammonium iodide (MAI) was synthesized using minor modifications to a previously
published procedure29. First, 25 mL of methylamine (33 wt% in ethanol, SigmaꢀAldrich) was
added to a 250 mL, threeꢀneck flask maintained at a temperature lower than 7°C. 10 mL of
hydroiodic acid (HI, 57 wt% in water, SigmaꢀAldrich) was added dropwise to the flask at a speed
of approximately 6 drops/10secs to MAI solution while the solution was stirred at 300 rpm. The
white MAI precipitate was recovered from the solution using rotary evaporation at a pressure of
60 Torr with the flask in a water bath at 50 °C.
The collected MAI powder was dissolved in 10 ml absolute ethanol while stirring and
sonicating for 10 minutes and then precipitated by adding 50ꢀ55 ml of diethyl ether to the
solution and refrigerating it for 20 minutes. After filtration, the process was repeated 3 times, and
finally a white powder was obtained and dried at 60 °C overnight under vacuum using a Schlenk
line. The phase purity of the compound was confirmed by a Bruker D8 Advance powder Xꢀray
diffractometer (supporting information S1).
2.2 Perovskite synthesis via APCVD
The MAI powder prepared above was used as a first precursor for the APCVD. BiI3 powder
(99%, Aldrich) was used as a second precursor for the APCVD and used without further
purification. A quartz tube mounted on a single zone furnace (Compact Split Tube Furnace with
1” TubeꢀOTFꢀ1200XꢀS, MTI Corporation.) was used for all APCVD studies. Silicon wafers
(Test Grade, University Wafers Inc.) with or without a 100 nm thermal oxide layer, glass slides
4

Page 5 of 39
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C7TA06578G
(48300ꢀ025, VWR), preꢀcut 5 mm x 5 mm high purity quartz substrates (MTI Corporation) and 2
cm x 2 cm Willow® flexible glass from Corning were used as substrates. The silicon and glass
substrates were cut to approximately 1 cm x 2 cm size, ultrasonically cleaned in DI water,
acetone (99.9% Aldrich) and ethanol (99.9% Aldrich) mixture for 10 minutes, washed by IPA
and DI water, dried in compressed air, cleaned in UV Ozone (Ossila E511) for 10 minutes, and
then immediately placed in the quartz tube, 23 cm away from the upstream end.
The schematic of the experimental setꢀup is shown in Figure 1a, while a picture of the
furnace after successful MA3Bi2I9 deposition is shown in Figure 1b. The deposition is marked
by the characteristic orange color of the MA3Bi2I9 film on the downstream side.15 Initially, 50
mg of MAI and 60 mg of BiI3 were used as sources and placed at 2.5 cm and 13 cm from the
upstream end of the quartz tube in alumina boat crucibles. First, the quartz tube was purged with
280 standard cubic centimeters per minute (sccm) flow of ultraꢀhigh purity Ar gas (UHP 300,
Airgas) for 10 minutes. The temperature in the quartz tube was then raised at the rate of 1.8 oC /
o
minute until the temperature at the center of the tube reached 230 C. The furnace was
maintained at this temperature for various times of 15, 30, 45, 60, 90, 180, 240, 360 and 480
minutes of deposition. After deposition, the furnace was naturally allowed to cool down to
ambient temperature. This usually took 3 hours.
The key aspect of an APCVD process lies in the precise placement of the MAI and BiI3
sources along the length of the quartz furnace reactor. This allows us to coꢀsublimate both the
sources together. Figure 1c shows the temperature profile of the furnace along the length of the
CVD quartz tube. This calibration measurement was done prior to the deposition by sliding a
thermocouple probe inserted inside the quartz tube with Ar flowing at the rate of 280 sccm. The
o
temperature varies from 199 C from the upstream end (labeled as 2.5 cm on the xꢀaxis) and
5

Journal of Materials Chemistry A
Page 6 of 39
View Article Online
DOI: 10.1039/C7TA06578G
peaks to the desired temperature setꢀpoint of the furnace (230 oC) at the center of the tube (13 cm
o
from the upstream end). The temperature then drops off on the downstream side to 160 C, 23
cm from the upstream end.
The variability of the temperature along the length of the quartz furnace reactor is
exploited in the APCVD process. The MAI boat is placed at 2.5 cm from the upstream side
o
where the temperature is 199 C and allows the MAI to sublimate. We note that MAI has a
o
melting point of 270 C. The BiI3 is kept in the middle of the quartz reactor where the
o
o
temperature is 230 C. BiI3 has a melting point30 of 402 C but is known to sublimate as well.
Thus, the Ar flow causes the simultaneous transport of MAI and BiI3 vapor to the cooler zone in
the downstream region. Condensation and reaction of the MAI and BiI3 vapors take place to
produce the MA3Bi2I9 film at around 23 cm from the upstream side, where the temperature is
maintained at 160 oC.
Thermogravimetric measurements have been used to calculate vapor pressure of the two
precursors, MAI and BiI3. The details of this analysis are provided in supporting information
S2. The enthalpy of sublimation of MAI is obtained from Dualeh et al.,31 to be 105 ± 5 kJ/mol,
and for BiI3 is estimated in this work to be 128 ± 2 kJ/mol. The sublimation temperatures at
atmospheric pressure for MAI and BiI3 are found to be 247 ± 26 oC and 316 ± 2 oC, respectively.
Based on this analysis, the partial pressure of MAI and BiI3 when the total pressure is 760 Torr,
at the points of sublimation and deposition has been calculated. The vapor pressure as a function
of temperature is provided in Table 1. At the points of sublimation (199 oC for MAI and 230 oC
for BiI3), it can be seen that the partial pressure of MAI is higher than BiI3 by 4.7x. As the vapor
o
moves downstream, the temperature drops and at 160 C (i.e., the deposition temperature), the
MAI vapor pressure is 87x higher than BiI3. Further the thermodynamic driving force for
6

Page 7 of 39
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C7TA06578G
(T ꢀ T)
condensation which is given as,32
is found to ꢀ17.57 kJ/mol for
sub
ꢁG|gas→solid =ꢀ(ꢁHsub ) x
T
sub
MAI but ꢀ33.9 kJ/mol for BiI3 using data from supporting information S2. This implies that the
BiI3 should readily condense on the substrate, nucleate and grow. On the other hand, MAI
condensation followed by a solidꢀstate reaction with BiI3 to form MA3Bi2I9 may be a kinetically
rate determining step for MA3Bi2I9 film formation. Evidence to this effect will be presented and
discussed in the Results & Discussion section.
2.3 Film characterization
Film morphology was characterized by a JEOL JSMꢀ7001 LVF Field Emission SEM under an
accelerating voltage of 10 kV. Xꢀray diffraction (2θ scans) of deposited MA3Bi2I9 films were
obtained by Rigaku IV Xꢀray diffractometer using the Cu Kα Xꢀray source (λ = 1.5405 Å) within
a diffraction angle (2θ) from 5° to 60°. Xꢀray photoelectron spectroscopy (XPS) was measured
using the PHI Versa Probe II spectrometer (Physical Electronics) with a photon energy of 1486.6
eV (Al Kα). The absorption spectrum of a MA3Bi2I9 film deposited onto an ITO substrate was
measured using a PerkinꢀElmer Lambda 950 spectrometer equipped with a 150 mm integrating
sphere, a PMT detector, a tungsten halogen lamp for the visible region (350–800 nm), and a
deuterium lamp for the ultraviolet region (250–350 nm). The scan rate for the measurement was
0.4 nm/s, and the data interval was 1.0 nm. The film was placed in the center of the integrating
sphere to collect both the transmitted, T, and the reflected, R, (including the spectrally reflected
and diffusely scattered) light with the PMT detector. The transflectance measurement allowed
for the absorbance, A, of the film to be determined by A = − log(T + R).
The electrochemical analysis was performed with a standard threeꢀelectrode cell using
the cyclic voltammetry (CV) mode in a Biologic potentiostat (SPꢀ200, BioꢀLogic SAS) with a
scan rate of 20 mV/s. A glassy carbon electrode was used as the working electrode to measure
7

Journal of Materials Chemistry A
Page 8 of 39
View Article Online
DOI: 10.1039/C7TA06578G
the formal potential of ferrocene/ferrocenium (Fc/Fc+), whereas the Pt foil acted as the counter
electrode. Silver (Ag) wire in 0.01 M AgNO3 + 0.1 M tetrabutylammonium hexafluorophosphate
(TBAPF6) (98%, Aldrich) in acetonitrile (99.9%, SigmaꢀAldrich) was used as the reference
electrode (BAS Inc.). The conduction band (CB) energy was calculated from the onset of
reduction potential (Ered) values, assuming the energy level of Fc/Fc+ to be −4.8 eV below the
vacuum level. The formal potential of Fc/Fc+ was measured to be 0.075 V against a Ag/Ag+
reference electrode in an electrolyte consisting of 0.001 M ferrocene in 15 mL of 0.1 M TBAPF6
in acetonitrile solution . Therefore, from equation (1)
ECB (ELUMO) = − (Ered + 4.725) eV,
[1]
where, the onset potential values (Ered) are relative to the Ag/Ag+ reference electrode. The
valence band (VB) energy, EVB (EHOMO) was calculated based on the band gap value determined
from UV−vis spectra, EVB (EHOMO) = ECB −Eg (indirect). MA3Bi2I9 films deposited on glass
slides were dissolved in a solution of 0.1 M tetrabutylammonium hexafluorophosphate
(TBAPF6), acetonitrile for conducting these tests.
All electronic transport property measurements were carried out in a commercial probe
station (Janis ST500ꢀ1ꢀ2CX) with Cu–Be probe tips having 50 ꢂm tip diameter. A Keithley 2400
source meter was used for resistivity measurements on MA3Bi2I9 films deposited on 5 mm × 5
mm highꢀpurity quartz substrates (MTI Corp). Van der Pauw structures were created by first
sputtering 50 nm Pt on the four corners of the quartz substrates prior to MA3Bi2I9 deposition.
After MA3Bi2I9 deposition, indium dots were attached to these four corners and served as the
contact electrodes for the four probe tips. Corresponding Hall measurements were done by
placing an FeꢀNd ring magnet around the sample. The magnetic field strength at the center of the
ring magnet was 2135 G. The pressure in the chamber was maintained at or below 1x10ꢀ4 Torr.
8

Page 9 of 39
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C7TA06578G
2.4 DFT calculation details
DFT calculations were performed using the Vienna Abꢀinitio Simulation Package (VASP).33 We
employed the PerdewꢀBurkeꢀErnzerhof exchangeꢀcorrelation (XC) functional revised for solids
(PBEsol).34 The core and valence electrons were modeled using the projectorꢀaugmentedꢀwave
(PAW) method.35 We used a planeꢀwave basis set with a cutoff of 500 eV and performed
relaxation until the Hellmann–Feynman forces on the atoms were less than 0.001 eV/Å. The
Brillouin zone was sampled using a Gammaꢀcentered MonkhorstꢀPack36 kꢀpoints mesh while
keeping the kꢀpoints density (
nk
×a ), where
is the number of kꢀpoints and is the lattice
nk a
parameter, equal to ~26 for structural relaxation and ~52 for electronic structure calculations.
The spinꢀorbit coupling (SOC) effects were included for calculating the electronic band structure
and the absorption spectra. The charge partitioning for the Bader charge analysis was carried out
using the approach developed by Henkelman et al.37ꢀ39 The elemental charges, used for Bader
charge analysis, are the difference between the total valence electrons considered in the PAW
potential and the total electrons assigned to an element after the charge density partitioning. The
bandꢀedge alignment for MA3Bi2I9 was performed using a surface slab with material thickness of
~55 Å and a vacuum of ~50 Å. The valence band edge (EVB) is calculated using the expression,
VB , where Φ
− ε
is the vacuum electrostatic potential and εVB is the eigenvalue for
E
VB
=
φ
VAC
VAC
the valence band maxima. The conduction band edge (ECB) is calculated using the expression,
E
CB = EVB + E
G
, where EG is the band gap calculated using PBEsol with SOC.
9

Journal of Materials Chemistry A
Page 10 of 39
View Article Online
DOI: 10.1039/C7TA06578G
3
Results & Discussion
3.1 Materials characterization
Figure 2a shows the color of the films on glass substrates as a function of deposition time.
Initially, for short times (< 60 minutes), there is hardly any deposition observed. After 90
minutes of reaction, a visible orange color is seen on the glass substrate. We note that there is a
stronger color close to the upstream side of the sample, indicating that the deposition is
determined by the Ar flow. For films ≥ 180 minutes, a continuous film is observed on the glass
substrate. The orange color is characteristic of MA3Bi2I9.15 Successfully deposition of MA3Bi2I9
films on Willow® glass ꢀ Corning’s flexible glass substrate is shown in Figure 2b and Figure
2c. These flexible glass substrates can withstand temperatures up to 500 oC and thus, can be used
for the APCVD of MA3Bi2I9 films.
Highꢀmagnification SEM images are shown for the same substrates in Figure 3.
Mechanistic aspects of the deposition process can be garnered from this sequence of images.
First, Figure 3a, b and c (i.e., for deposition times 15, 30 and 45 minutes) show isolated
hexagonal crystals growing in size from 0.5 ꢀm for 15 minutes, to 1.0 ꢀm for 45 minutes of
deposition. The crystal shape is characteristic of BiI340 and has been independently confirmed by
through Raman (see supporting information S3) and XRD measurements, as will be shown
below. Thus, in the initial part of the deposition scheme, isolated BiI3 crystals 0.5 – 1.0 ꢀm in
diameter are deposited on the substrate surface.
The BiI3 crystals show signs of MAI incorporation starting at 60 minutes (Figure 3d).
While the crystal shapes remain intact, the surface of the crystal shows marked roughness. For
90 minutes (Figure 3e), the deposited structure becomes interconnected. The interconnected
10

Page 11 of 39
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C7TA06578G
structure shows signs of growth for 180 minutes (Figure 3f) while the surface structure is rough
and full of pinholes.
The surface structure further changes for the 240ꢀminute sample (Figure 3g), where wellꢀ
crystallized MA3Bi2I9 grains are visible, though evidence of MAI incorporation can still be seen
at the grain boundaries. For the 360ꢀminute sample (Figure 3h), the MA3Bi2I9 polycrystalline
grains are well developed, indicating complete incorporation of the MAI. However, there are
intergranular gaps observed, and such morphology is not ideal for thin film photovoltaic
application. Finally, for the 480ꢀminute sample (Figure 3i), we see secondary nucleation of
grains in between larger polycrystalline grains.
The secondary nucleation of the MA3Bi2I9 crystals observed in the intergranular regions
of the film has been used to grow a dense, compact and polycrystalline film. In Figure 4a a lowꢀ
magnification image of the 360ꢀminute sample is shown. The intergranular regions are clearly
observed and, as indicated before, are sites for secondary nucleation after 480 minutes of
deposition. Therefore, repeating the process twice (labeled henceforth as, 360 minute (2x)), can
eliminate these gaps and produce a film with dense morphology. This is shown in Figure 4b.
While such an approach may see excessive long deposition times, its need is dictated primarily
by the limited mass loading of the MAI and BiI3 precursors in our sublimation crucibles. Larger
furnace reactors may easily overcome such hardware limitations and yield dense films in a single
process run. The ‘twice deposited’ films were subsequently used to conduct all optical and
electrical measurements.
XRD data of MA3Bi2I9 samples deposited for times varying from 15 minutes to 60
minutes are shown in Figure 5a. Initially a diffraction peak corresponding to single crystal BiI3
with violet symbol ‘■’ can be seen. The peak at 2θ = 52.9° can be indexed41 to BiI3 (0012). The
11

Journal of Materials Chemistry A
Page 12 of 39
View Article Online
DOI: 10.1039/C7TA06578G
other peaks represent MAI, marked by the red symbol ‘•’, (JCPDS ref no. 000ꢀ10ꢀ0737) at
54.6o corresponding to (222), and a slightly displaced peak from polycrystalline BiI3 (JCPDS 48ꢀ
1795) given as ‘□’ at 56.2o corresponding to (226). As deposition time increases, it can be seen
that the presence of MAI (JCPDS 10ꢀ0737) becomes prominent and the single crystal and
polycrystalline BiI3 signal diminishes.
Figure 5b shows the XRD for deposition times varying from 90 minutes to 480 minutes.
After 90 minutes’ of deposition, the presence of MA3Bi2I9 is detectable. The MA3Bi2I9 phase is
denoted by the cyan ‘♦’ symbol. Here, we note that there are no JCPDS files available for
MA3Bi2I9, and we use the patterns generated by Abulikemu et al.,42 to index our peaks. Initially
at 90 minutes, peaks at 29.07o (024), 31.6o (025), 32.38o (116) and 44.7o (029) are observed. A
strong peak at 24.6o corresponds to MAI (110). Though unlikely at this early stage of deposition,
this peak could also represent the MA3Bi2I9 (006) peak at 24.5o. At 180 minutes, the primary
peak is still assigned to MAI (110) at 24.6o. Beyond 180 minutes, multiple peaks of MA3Bi2I9
start to emerge. The peak at 24.6o can be now assigned to MA3Bi2I9 (006), given that there are
associated peaks, all corresponding to MA3Bi2I9 phase emerge. Other strong peaks corresponding
to MA3Bi2I9 observed are at 23.6o (015), 25.3o (022), 32.2o (116), 41.07o (028), 44.7o (029) and
50.1o (136). The characteristically polycrystalline nature of the MA3Bi2I9 films is inꢀline with
SEM data since polycrystallinity is clearly shown in SEM images for 240ꢀ, 360ꢀ and 480ꢀminute
samples. Finally, we note that there appears to be a weak peak observed for the 240 minute
sample at 35.2o corresponding to (116) plane in polycrystalline BiI3 which could indicate
concurrent deposition of BiI3 during MA3Bi2I9 formation, but this peak is not detected for the
360 and 480 minute samples.
12

Page 13 of 39
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C7TA06578G
Next, we compare the lattice parameters from our XRD results to the crystal structure
data obtained from firstꢀprinciples DFT calculations. MA3Bi2I9 exhibits hexagonal crystal
43ꢀ45
P6 / mmc
3
symmetry at room temperature with
space group.
The methylammonium cation
+
(CH3NH3 (MA+)) shows no preferential ordering in the hexagonal phase.43 Figure 5c shows the
DFTꢀoptimized crystal structure of MA3Bi2I9 oriented along [100] and [001] directions. The
geometry optimization has been performed using the PBEsol functional,34 which gives good
agreement of theoretical lattice parameters with the experimental values. The calculated lattice
parameters from DFT are 8.72 Å, 8.27 Å and 21.76 Å, which are, respectively, within 1.77%,
3.47% and 0.01% of the experimental43 lattice parameters of the hexagonal phase. Since, we use
a static groundꢀstate orientation of the MA cation in our DFT calculations, the ideal hexagonal
P6
relaxed structure. Instead the DFT relaxed structure exhibits a monoclinic symmetry with space
Cc P6 / mmc
3
/ mmc
symmetry of the lattice is broken, which results in lowering of the symmetry for the
3
group symmetry
, which is a subgroup of
.
The MA3Bi2I9 structure can be thought of as a derivative of the perovskite (ABX3)
structure. In both cases the Bꢀsite cation (Bi in case of MA3Bi2I9) is octahedrally coordinated by
Xꢀsite anions (I in case of MA3Bi2I9). However, as compared to the ideal perovskite structure,
only 2/3rd of the total octahedral sites are occupied by the Bi cation in A3Bi2I9 compounds.16
Additionally, as opposed to the cornerꢀconnected octahedral network of BX6 octahedra in
perovskites, MA3Bi2I9 forms a layered structure consisting of isolated Bi2I93− bioctahedra. These
bioctahedra are formed by two faceꢀshared BiI6 octahedra, which share three I anions Figure 5c).
Figure 6 shows the XPS fine spectra of the 15ꢀ, 180ꢀ and 480ꢀminute MA3Bi2I9 films
whereas the full survey spectra of these samples are provided in supporting information S4. In
Figure 6a, the C 1s fine spectrum is shown. For the 15ꢀminute sample, the primary peak is at
13

Journal of Materials Chemistry A
Page 14 of 39
View Article Online
DOI: 10.1039/C7TA06578G
284.5 eV corresponding to the CꢀC bond related to adventitious carbon. There is a slight
shoulder detected at 282.6 eV, which is related to the CꢀN bond46 originating from MAI. The Cꢀ
N bond increases in strength for the 180ꢀminute sample and dominates the C 1s signal for the
480ꢀminute sample. These data suggest that MAI incorporation in the BiI3 is a slower process
than the BiI3 condensation, nucleation and growth process. This conclusion is in line with earlier
observations made from vapor pressure calculations in Table 1 as well as SEM images in Figure
3.
N 1s, Bi 4f and I 3d XPS fine spectra are provided in Figure 6b, c and d, respectively
and the peak positions are shown in Table 2. For the N 1s, the 15ꢀminute sample shows a weak
signal but progressively grows stronger for the 180ꢀ and 480ꢀminute samples indicating MA+
incorporation. The N 1s peak position shifts to lower binding energy from 402.0 eV to 401.5 eV.
The Bi 4f and I 3d peaks are strong even at 15 minutes and indicate the presence of BiI3. MA+
incorporation produces a shift to lower binding energy of 0.7 eV and 0.5 eV in the Bi 4f and I 3d
peaks, respectively. The peakꢀshifts indicate electron transfer to the N, Bi and I during the
transformation of BiI3 to MA3Bi2I9.
To explain the shifts in the XPS spectra, Bader charge analysis has been performed for
each element during the chemical reaction leading up to the formation of MA3Bi2I9. The average
Bader charges for each elemental species in MA3Bi2I9, MAI and BiI3 are shown in Table 3.
Recall that in Figure 6a, the intensity of the deꢀconvoluted CꢀN peak increases as the reaction
proceeds, indicating increased presence of CꢀN bond with increasing reaction time. The CꢀN
bond occurs at a higher binding energy compared to the CꢀC bond, signifying loss of electrons in
the C atom upon N addition. From the Bader analysis, one finds loss of electrons on the C atom
confirmed by the higher atomic charge on C in MA3Bi2I9 (+0.453 ± 0.015)e than in MAI
14

Page 15 of 39
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C7TA06578G
(+0.445e). However, for N there is disagreement between the calculated charge transfer and the
shift in XPS peak presented in Figure 6b. According to calculated Bader charges, the average
charge on N is (−2.982 ± 0.009)e and (−3.029e) for MA3Bi2I9 and MAI respectively, which
means N anion is being oxidized while going from MAI to MA3Bi2I9. However, this charge
transfer is not supported by the shift in the XPS peak towards lower binding energy with
increasing reaction time, which points towards the reduction (gain in electrons) of N anion. This
disagreement could be attributed to the weak XPS intensity for N at reaction times of 15 and 180
minutes, which makes the determination of charge transfer difficult with the N 1s fine spectra.
The Bi cations in MA3Bi2I9 have an average charge of (+1.049 ± 0.002)e compared to
+1.061e in the reactant BiI3 and are, therefore, in a reduced state in MA3Bi2I9 as compared to
BiI3. This reduction of the Bi cation is also supported by the shift in experimental XPS peak
towards lower binding energy as shown in Figure 6c. On comparing the average Bader charge of
I anions in the primary reactants, i.e. the weighted average of the Bader charge of I in MAI and
BiI3 (−0.459e) to that in MA3Bi2I9 (−0.472e), we find that the I anions are also reduced. This
gain of electrons is also confirmed by the shift of the XPS peak for I anion towards lower
binding energy as shown in Figure 6d. Moreover, the reduction of Bi and I ions in MA3Bi2I9
with respect to the reactants (MAI and BiI3) is accompanied by a simultaneous oxidation of the
MA cations in MA3Bi2I9 with respect to MAI, which balances out the net charge transfer. The
net reduction of Bi and I requires a total charge transfer of 0.141e per formula unit of MA3Bi2I9
to Bi and I (0.024e for Bi and 0.117e for I). This charge transfer is achieved by a transfer of
electrons from MA cations to Bi and I as the reaction proceeds. The net average charge on MA
cation is (+0.668e) and (+0.715e) in MAI and MA3Bi2I9 respectively, which amounts to a net
charge transfer of 0.141e per formula unit of MA3Bi2I9 from MA to Bi and I.
15

Journal of Materials Chemistry A
Page 16 of 39
View Article Online
DOI: 10.1039/C7TA06578G
Finally, the charge transfer in H can be categorized into two types, based on the position
of H in the MA cations. H atoms directly bonded to C and those directly bonded to N lead to two
different bonding environments. The H atoms bonded to N show an atomic charge of +1 for both
MA3Bi2I9 and MAI. Whereas, the H atoms bonded to C show an average atomic charge of
+0.081 and +0.084 in MA3Bi2I9 and MAI respectively. This difference in the atomic charge is
due to the higher electronegativity of N than C, which renders the HꢀN bonds more ionic than the
H–C bonds.
3.2 Optical characterization
Figure 7a shows the UVꢀvis absorption spectra of the 360 minute (2x) MA3Bi2I9 film at
wavelengths between 400 nm and 800 nm. The absorption onset of MA3Bi2I9 was observed at
approximately 600 nm. The presence of a peak at 511 nm before the onset of continuous
absorption is attributed to excitons.42, 44, 47, 48 We have verified separately the MAI and BiI3
absorption spectra do not contain an exciton peak. This exciton peak has also been observed for
single crystals of MA3Bi2I9 from temperatures between 78 K to 301 K, and it starts to broaden as
the temperature increases.47 The thickness dependence of the exciton levels is shown supporting
information S5 where a stronger and narrower exciton peak is observed for a thinner film (~ 480
nm) whereas, a broader exciton peak is observed for a thicker (~ 1000 nm) film.
The bandgap of MA3Bi2I9 films was calculated first by establishing a baseline and removing the
exciton peak from the absorption spectrum. Then by plotting the baseline as a Tauć plot and
using the equation49:
α
h
υ
=
β
(hυ
− Eg )n
υ
is the photon energy, β is a constant, and Eg is the
[2]
h
where α is the absorption coefficient,
bandgap; a linear fit is used to extract the bandgap. Assuming the lowest energy optical transition
16

Page 17 of 39
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C7TA06578G
in MA3Bi2I9 is indirect (see DFT calculations below), n=1/2 is used in the evaluation. The inset
of Figure 7a shows the Tauć plot, and the extrapolated bandgap is 1.80 eV. The value of the
MA3Bi2I9 bandgap is in line with those reported by Lyu et al.20
Figure 7b shows the electronic band structure of MA3Bi2I9 calculated using PBEsol with
spinꢀorbit coupling (SOC) along the high symmetry points in the Brillouin zone. Using the
PBEsol functional and SOC effects, we find MA3Bi2I9 has an indirect band gap of 1.77 eV from
K (1/3, 1/3, 0) in the valence band to (0.05, 0.05, 0) in the conduction band as shown in Figure
7b. The calculated direct band gap is 1.89 eV and occurs at (0.075, 0.075, 0). The indirect and
direct transitions are shown as red and black arrows respectively in Figure 7b. MA3Bi2I9 exhibits
relatively flat bands as compared to the more promising organohalide perovskite solar cell
50, 51
absorbers such as CH3NH3PbI3,
which is typical for A3Bi2I9 compounds17 because of the
disrupted octahedraꢀconnectivity along all three crystallographic directions.
We also calculated the band structure using the hybrid HeydꢀScuseriaꢀErnzerhof (HSE06)
functional with SOC effects,52 which is known to result in better band gap estimation for a
variety of semiconductors, as shown in supporting information S6. We find HSE06
overestimates the bandgap with an indirect band gap of 2.41 eV from K (1/3, 1/3, 0) in the
valence band to (0.05, 0.05, 0) in the conduction band, and a direct band gap of 2.53 eV at K
(1/3, 1/3, 0). Both the indirect and direct transitions are shown as red and black arrows
respectively in the Figure S6.
To extract the experimental band edge positions of the MA3Bi2I9, CV measurements are
shown in Figure 7c. MA3Bi2I9 has the reduction peak edge at, Ered = ꢀ1.22 V. Referring to
equation [1], the MA3Bi2I9 conduction band edge position is therefore calculated as, ECB = ꢀ3.505
eV with respect to vacuum. Using the bandgap obtained from UVꢀvis (1.80 eV), the valence
17

Journal of Materials Chemistry A
Page 18 of 39
View Article Online
DOI: 10.1039/C7TA06578G
band edge position is calculated as EVB= ꢀ5.30 eV. The data from UVꢀvis + CV measurements
and DFT calculations are summarized in Figure 7d.
3.3 Electrical characterization
For the MA3Bi2I9 film deposited twice (360 minute (2x)) on the quartz substrate, conductivity
was measured using a van der Pauw fourꢀpoint configuration. The conductivity is 9.7 S·cm–1
(i.e., resistivity 0.103 ꢃ·cm) at room temperature for a film with a thickness of 775 nm (see
supporting information S7 for thickness measurement using crossꢀsection SEM image). The
detailed result is shown in Table 4 along with comparison of electrical data from other
reports.15,44 This conductivity is 1168x better than the reported conductivity (0.0083 S·cmꢀ1) of a
solutionꢀprocessed,20 500 nm MA3Bi2I9 film highlighting the importance of the type of synthesis
in determining film properties.
Room temperature Hall measurement was performed to determine conductivity type,
carrier mobility and concentration of the MA3Bi2I9 film. According to the measurements, the
Hall coefficient is negative, which indicates the carriers to be nꢀtype for the CVD MA3Bi2I9 film.
This result is in contrast to the solution processed MA3Bi2I9 film which shows pꢀtype
conductivity.20, 53 Further, the calculated carrier concentration of the CVD MA3Bi2I9 film is 3.36
x 1018 cmꢀ3, which is 2 orders higher in magnitude than solutionꢀbased, undoped MA3Bi2I9
films.20, 53 Finally, the mobility was estimated to be 18 cm2/V.s. This is higher compared to the
mobility reported by Lyu et al.,20 of 1 cm2/V.s and Vigneshwaran et al.,53 of 2.28 cm2/V.s. The
improved carrier concentration and mobility highlight typical advantages that vapor phase
processing of MA3Bi2I9 films have over solutionꢀbased techniques including, minimal impurities,
better crystallinity and film morphology.
18

Page 19 of 39
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C7TA06578G
3.4 Film stability
It has been reported that the MA3Bi2I9 is stable in ambient, thus presenting a significant
advantage over other organic halide perovskite candidates.54 However, in our work we find that
MA3Bi2I9 film exposed to ambient (room temperature, 45% humidity and 1 atm) undergoes
changes to its composition and degradation to its optical and electrical properties. In Figure 8a, a
loss of N is detected in the MA3Bi2I9 film exposed to the ambient in just 5 days. The absence of
N has been observed by Li et al.,55 for degraded Pbꢀbased perovskite films as well. Further, in
supporting information S8, we provide the XPS 1s fine spectra of O and 4f fine spectra of Bi in
the MA3Bi2I9 film. The O 1s shows the characteristic O peak associated with surface adsorbed
hydroxyl groups with an additional shoulder corresponding to BiꢀO bond formation after 5 days
and growing stronger after 14 days of exposure to the ambient. Similarly, the Bi 4f shows signs
of oxidation as well after 5 days. Figure 8b shows XPS fine spectra of I 3d7/2 and 3d5/2 for fresh
films and for films exposed 5 and 14 days to the ambient. Unlike N and O, there is no noticeable
change observed to the I 3d fine spectra after 5 days. However, shoulders appear for the I 3d7/2
and 3d5/2 peaks on the higher binding energy side after 14 days, indicating oxidation56 of I. Thus,
from XPS data it can be concluded that, while the loss of surface N and oxygen incorporation
(most likely associated with Bi) is a rapid process, iodine oxidation is a slower process.
The compositional changes in the MA3Bi2I9 are accompanied by changes to optical and
electrical properties. Figure 8c shows the UVꢀvis absorption spectra of the MA3Bi2I9 film
exposed to ambient over a period of 5 days. The characteristic 510 nm exciton peak disappears
on day 5. Correspondingly, the insets in Figure 8c show the film changes color from bright
orange to a pale shade of yellow. In Figure 8d, the excitonic intensity at 510 nm (black, left
axis) and the 4ꢀwire resistivity (red, right axis) are plotted as a function of time over a five day
19

Journal of Materials Chemistry A
Page 20 of 39
View Article Online
DOI: 10.1039/C7TA06578G
period. A biꢀexponential model involving two time constants is used to fit the increase in
resistivity over time. Two time constants, τ1 = 0.1 day and τ2 = 1.7 days are obtained with a
model fit that has an adjusted r2 ~ 0.99.
The XPS analysis along with the biꢀexponential decay of resistance suggests two decay
mechanisms might be operative in degrading the MA3Bi2I9 films. First, the time constant
τ1 = 0.1 day could indicate surface oxidation of MA3Bi2I9 where the Bi may be oxidized to
Bi2O3. In line with this assumption, it can be seen that the exciton peak intensity only slightly
degrades in the initial period. This is because the surface Bi2O3 layer may provide temporary
protection24 from further attack by ambient O2, keeping the excitonic modes active in the bulk.
On the other handτ2 = 1.7 days indicates the slower diffusionꢀlimited oxidation of the bulk
MA3Bi2I9, and as the film gets oxidized after day 4, the exciton peak intensity rapidly degrades.
4
Conclusions
In conclusion, we have demonstrated an atmospheric pressure chemical vapor deposition process
for the deposition of MA3Bi2I9 perovskite film. The film is deposited using coꢀsublimation of
o
o
methylammonium iodide (kept at 199 C) and BiI3 (kept at 230 C) as precursors in a flow of
o
inert Argon. The temperature of deposition is 160 C. The key advantages of this synthesis
approach is that the deposition is at atmospheric pressure and at low temperatures, which opens
the possibility for depositing MA3Bi2I9 films in high volume, cleanroom compatible,
manufacturing environments. The deposition on flexible Willow® glass has been demonstrated.
Wellꢀdeveloped crystals are obtained after 360 minutes of deposition. Repeating the
process twice leads to dense, high quality polycrystalline films. Optical bandgap is measured to
be 1.80 eV. Room temperature Hall measurements on 775 nm thick MA3Bi2I9 films indicate
20

Page 21 of 39
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C7TA06578G
them to be nꢀtype with a carrier concentration of 3.36 x 1018 cmꢀ3 and a Hall mobility of 18
cm2/V.s; values superior to solution processed, undoped films. The MA3Bi2I9 films oxidize under
ambient conditions in a matter of ~ 5 days and in the process, 1) N is lost, 2) excitonic peaks are
quenched and, 3) resistivity increases with a biꢀexponential decay with time constants related to
a fast surface oxidation mechanism, followed by a slower oxidation of the bulk film.
5
Acknowledgements
Partial support for this work was provided under the U.S.ꢀIndia Partnership to Advance Clean
EnergyꢀResearch (PACEꢀR) for the Solar Energy Research Institute for India and the United
States (SERIIUS), funded jointly by the U.S. Department of Energy ( Office of Science, Office
of Basic Energy Sciences, and Energy Efficiency and Renewable Energy, Solar Energy
Technology Program, under subcontract DEꢀAC36ꢀ08GO28308 to the national Renewable
Energy Laboratory, Golden, Colorado) and the Government of India, through the Department of
Science and Technology under Subcontract IUSSTF/JCERDCꢀSERIIUS/2012 is acknowledged.
Willow® glass substrates were kindly provided by Dr. Sean Garner from Corning. Partial
support from U.S. Army RDECOM Acquisition Grant W911NFꢀ15ꢀ1ꢀ0178, Subgrant RSC15032
is acknowledged. R.M. acknowledges support from Oak Ridge Associated Universities through a
Ralph E. Powe Junior Faculty Enhancement award. The XPS measurements were made possible
through the support from NSF MRI Grant 1337374. The microscopy facility and support from
Dr. Huafang Li at the Institute of Materials Science and Engineering are acknowledged. The
authors thank R. A. Loomis for use of his spectrometer and integrating sphere. Help from Dr.
D.C. Osborn at the XRD Facilities at University of Missouri St. Louis is acknowledged. This
work used computational resources of the Extreme Science and Engineering Discovery
21

Journal of Materials Chemistry A
Page 22 of 39
View Article Online
DOI: 10.1039/C7TA06578G
Environment (XSEDE), which is supported by National Science Foundation grant number ACIꢀ
1053575.
22

Page 23 of 39
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C7TA06578G
References:
1.
2.
3.
4.
5.
6.
7.
A. Kojima, K. Teshima, Y. Shirai and T. Miyasaka, J. Am. Chem. Soc., 2009, 131, 6050ꢀ
6051.
NREL,
Best
ResearchꢀCell
Efficiencies,
https://www.nrel.gov/pv/assets/images/efficiencyꢀchart.png).
C. Wehrenfennig, G. E. Eperon, M. B. Johnston, H. J. Snaith and L. M. Herz, Adv.
Mater., 2014, 26, 1584ꢀ1589.
J.ꢀH. Im, C.ꢀR. Lee, J.ꢀW. Lee, S.ꢀW. Park and N.ꢀG. Park, Nanoscale, 2011, 3, 4088ꢀ
4093.
J. H. Noh, S. H. Im, J. H. Heo, T. N. Mandal and S. I. Seok, Nano Lett., 2013, 13, 1764ꢀ
1769.
S. D. Stranks, G. E. Eperon, G. Grancini, C. Menelaou, M. J. Alcocer, T. Leijtens, L. M.
Herz, A. Petrozza and H. J. Snaith, Science, 2013, 342, 341ꢀ344.
G. Xing, N. Mathews, S. Sun, S. S. Lim, Y. M. Lam, M. Grätzel, S. Mhaisalkar and T. C.
Sum, Science, 2013, 342, 344ꢀ347.
8.
9.
S. F. Hoefler, G. Trimmel and T. Rath, Monatsh. Chem., 2017, 148, 795ꢀ826.
B. Hailegnaw, S. Kirmayer, E. Edri, G. Hodes and D. Cahen, The journal of physical
chemistry letters, 2015, 6, 1543ꢀ1547.
10.
G. Nagabhushana, R. Shivaramaiah and A. Navrotsky, Proceedings of the National
Academy of Sciences, 2016, 113, 7717ꢀ7721.
11.
12.
A. Babayigit, A. Ethirajan, M. Muller and B. Conings, Nat. Mater., 2016, 15, 247ꢀ251.
M. Saliba, T. Matsui, J.ꢀY. Seo, K. Domanski, J.ꢀP. CorreaꢀBaena, M. K. Nazeeruddin, S.
M. Zakeeruddin, W. Tress, A. Abate and A. Hagfeldt, Energy Environ. Sci., 2016, 9,
1989ꢀ1997.
13.
14.
15.
Y. Zhou, Z. Zhou, M. Chen, Y. Zong, J. Huang, S. Pang and N. P. Padture, J. Mater.
Chem. A, 2016, 4, 17623ꢀ17635.
N. Rajamanickam, S. Kumari, V. K. Vendra, B. W. Lavery, J. Spurgeon, T. Druffel and
M. K. Sunkara, Nanotechnology, 2016, 27, 235404.
B. W. Park, B. Philippe, X. Zhang, H. Rensmo, G. Boschloo and E. M. Johansson, Adv.
Mater., 2015, 27, 6806ꢀ6813.
23

Journal of Materials Chemistry A
Page 24 of 39
View Article Online
DOI: 10.1039/C7TA06578G
16.
17.
A. J. Lehner, D. H. Fabini, H. A. Evans, C.ꢀA. Hébert, S. R. Smock, J. Hu, H. Wang, J.
W. Zwanziger, M. L. Chabinyc and R. Seshadri, Chem. Mater., 2015, 27, 7137ꢀ7148.
X. Huang, S. Huang, P. Biswas and R. Mishra, J. Phys. Chem. C, 2016, 120, 28924ꢀ
28932.
18.
19.
A. Kulkarni, T. Singh, M. Ikegami and T. Miyasaka, RSC Adv., 2017, 7, 9456ꢀ9460.
C. T. Zuo and L. M. Ding, Angewandte ChemieꢀInternational Edition, 2017, 56, 6528ꢀ
6532.
20.
21.
22.
23.
24.
M. Lyu, J.ꢀH. Yun, M. Cai, Y. Jiao, P. V. Bernhardt, M. Zhang, Q. Wang, A. Du, H.
Wang, G. Liu and L. Wang, Nano Res., 2016, 9, 692ꢀ702.
A. M. Ganose, C. N. Savory and D. O. Scanlon, Chem Commun (Camb), 2016, 53, 20ꢀ
44.
R. E. Brandt, V. Stevanović, D. S. Ginley and T. Buonassisi, Mrs Communications, 2015,
5, 265ꢀ275.
K. Eckhardt, V. Bon, J. Getzschmann, J. Grothe, F. M. Wisser and S. Kaskel, Chem.
Commun., 2016, 52, 3058ꢀ3060.
R. L. Hoye, R. E. Brandt, A. Osherov, V. Stevanović, S. D. Stranks, M. W. Wilson, H.
Kim, A. J. Akey, J. D. Perkins and R. C. Kurchin, Chem. Eur. J., 2016, 22, 2605ꢀ2610.
T. Okano and Y. Suzuki, Mater. Lett., 2017, 191, 77ꢀ79.
Z. Xiao, W. Meng, J. Wang, D. B. Mitzi and Y. Yan, Materials Horizons, 2017, 4, 206ꢀ
216.
25.
26.
27.
28.
29.
M. Liu, M. B. Johnston and H. J. Snaith, Nature, 2013, 501, 395ꢀ398.
Y. X. Zhao and K. Zhu, J. Phys. Chem. Lett., 2014, 5, 4175ꢀ4186.
M. M. Tavakoli, L. L. Gu, Y. Gao, C. Reckmeier, J. He, A. L. Rogach, Y. Yao and Z. Y.
Fan, Scientific Reports, 2015, 5.
30.
31.
A. Cuña, I. Aguiar, A. Gancharov, M. Pérez and L. Fornaro, Cryst. Res. Technol., 2004,
39, 899ꢀ905.
A. Dualeh, P. Gao, S. I. Seok, M. K. Nazeeruddin and M. Grätzel, Chem. Mater., 2014,
26, 6160ꢀ6164.
32.
33.
D. V. Ragone, Thermodynamics of materials ꢀ I, Wiley, New York, 1995.
G. Kresse and J. Furthmuller, Phys. Rev. B, 1996, 54, 11169ꢀ11186.
24

Page 25 of 39
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C7TA06578G
34.
J. P. Perdew, A. Ruzsinszky, G. I. Csonka, O. A. Vydrov, G. E. Scuseria, L. A.
Constantin, X. Zhou and K. Burke, Phys Rev Lett, 2008, 100, 136406.
P. E. Blöchl, Phys. Rev. B, 1994, 50, 17953.
35.
36.
37.
38.
H. J. Monkhorst and J. D. Pack, Phys. Rev. B, 1976, 13, 5188ꢀ5192.
G. Henkelman, A. Arnaldsson and H. Jónsson, Comput. Mater. Sci., 2006, 36, 354ꢀ360.
E. Sanville, S. D. Kenny, R. Smith and G. Henkelman, J. Comput. Chem., 2007, 28, 899ꢀ
908.
39.
40.
41.
W. Tang, E. Sanville and G. Henkelman, J. Phys. Condens. Matter, 2009, 21, 084204.
D. Nason and L. Keller, J. Cryst. Growth, 1995, 156, 221ꢀ226.
K. M. Boopathi, S. Raman, R. Mohanraman, F.ꢀC. Chou, Y.ꢀY. Chen, C.ꢀH. Lee, F.ꢀC.
Chang and C.ꢀW. Chu, Sol. Energy Mater. Sol. Cells, 2014, 121, 35ꢀ41.
M. Abulikemu, S. OuldꢀChikh, X. Miao, E. Alarousu, B. Murali, G. O. N. Ndjawa, J.
Barbé, A. El Labban, A. Amassian and S. Del Gobbo, J. Mater. Chem. A, 2016, 4, 12504ꢀ
12515.
42.
43.
44.
45.
46.
47.
48.
M. E. Kamminga, A. Stroppa, S. Picozzi, M. Chislov, I. A. Zvereva, J. Baas, A.
Meetsma, G. R. Blake and T. T. Palstra, Inorg Chem, 2017, 56, 33ꢀ41.
M. Scholz, O. Flender, K. Oum and T. Lenzer, J. Phys. Chem. C, 2017, 121, 12110ꢀ
12116.
C. Ran, Z. Wu, J. Xi, F. Yuan, H. Dong, T. Lei, X. He and X. Hou, J Phys Chem Lett,
2017, 8, 394ꢀ400.
L. Liu, J. A. McLeod, R. Wang, P. Shen and S. Duhm, Appl. Phys. Lett., 2015, 107,
061904.
T. Kawai, A. Ishii, T. Kitamura, S. Shimanuki, M. Iwata and Y. Ishibashi, J. Phys. Soc.
Jpn., 1996, 65, 1464ꢀ1468.
S. Öz, J.ꢀC. Hebig, E. Jung, T. Singh, A. Lepcha, S. Olthof, F. Jan, Y. Gao, R. German,
P. H. M. van Loosdrecht, K. Meerholz, T. Kirchartz and S. Mathur, Sol. Energy Mater.
Sol. Cells, 2016, 158, 195ꢀ201.
49.
50.
51.
J. I. Pankove, Optical processes in semiconductors, Courier Corporation, 2012.
P. Umari, E. Mosconi and F. De Angelis, Scientific Reports, 2014, 4, 4467.
A. S. Thind, X. Huang, J. Sun and R. Mishra, Chem. Mater., 2017, DOI:
10.1021/acs.chemmater.7b01781.
25

Journal of Materials Chemistry A
Page 26 of 39
View Article Online
DOI: 10.1039/C7TA06578G
52.
53.
J. Heyd, G. E. Scuseria and M. Ernzerhof, J. Chem. Phys., 2003, 118, 8207ꢀ8215.
M. Vigneshwaran, T. Ohta, S. Iikubo, G. Kapil, T. S. Ripolles, Y. Ogomi, T. Ma, S. S.
Pandey, Q. Shen and T. Toyoda, Chem. Mater., 2016, 28, 6436ꢀ6440.
54.
R. L. Hoye, R. E. Brandt, A. Osherov, V. Stevanović, S. D. Stranks, M. W. Wilson, H.
Kim, A. J. Akey, J. D. Perkins and R. C. Kurchin, Chemistry–A European Journal, 2016.
Y. Li, X. Xu, C. Wang, C. Wang, F. Xie, J. Yang and Y. Gao, 2015.
55.
56.
J. Chastain, R. C. King and J. Moulder, Handbook of Xꢀray photoelectron spectroscopy:
a reference book of standard spectra for identification and interpretation of XPS data,
Physical Electronics Division, PerkinꢀElmer Corporation Eden Prairie, Minnesota, 1992.
26

Page 27 of 39
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C7TA06578G
List of Figures
Figure 1: (a) Schematic of APCVD for MA3Bi2I9 thin films, (b) Image of MAI and BiI3 boats
inside the tube furnace and (c), temperature profile of the tube furnace showing the MAI and BiI3
sublimation temperatures and MA3Bi2I9 film deposition temperature.
27

Journal of Materials Chemistry A
Page 28 of 39
View Article Online
DOI: 10.1039/C7TA06578G
Figure 2: (a) Optical images of APCVD MA3Bi2I9 films on glass substrate for varying times of
deposition. Arrow represents direction of Ar flow in the furnace. Scale bar = 1 cm. (b) MA3Bi2I9
deposited for 360 minutes on Willow® flexible glass of size 2 cm x 2cm and (c) substrate held in
hand.
28

Page 29 of 39
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C7TA06578G
Figure 3: SEM images of APCVD films for varying times of deposition. Scale bar = 500 nm.
29