www.advenergymat.

de
www.MaterialsViews.com

Few-Layer MoS2 Flakes as Active Buffer Layer for Stable

FULL PAPER
Perovskite Solar Cells
Andrea Capasso, Fabio Matteocci, Leyla Najafi, Mirko Prato, Joka Buha, Lucio Cinà,
Vittorio Pellegrini, Aldo Di Carlo,* and Francesco Bonaccorso*

and the possibility of both vacuum[3] and

Solution-processed few-layer MoS2 flakes are exploited as an active buffer solution-processing.[7] Hybrid organometal
layer in hybrid lead–halide perovskite solar cells (PSCs). Glass/FTO/compact- halide perovskites were proposed in 2009
TiO2/mesoporous-TiO2/CH3NH3PbI3/MoS2/Spiro-OMeTAD/Au solar cells are as light absorbers in dye-sensitized solar
cells (DSSCs) with a power conversion
realized with the MoS2 flakes having a twofold function, acting both as a pro-
efficiency (η) ≈4%.[4] Several reports fol-
tective layer, by preventing the formation of shunt contacts between the per- lowed this seminal work, demonstrating
ovskite and the Au electrode, and as a hole transport layer from the perovskite that perovskite materials can be used to
to the Spiro-OMeTAD. As prepared PSC demonstrates a power conversion engineer highly efficient solar cells, with η
efficiency (η) of 13.3%, along with a higher lifetime stability over 550 h with up to 22.1%.[8]
Perovskites have the general for-
respect to reference PSC without MoS2 (Δη/η = −7% vs. Δη/η = −34%).
mula ABX3, where X is an anion while
Large-area PSCs (1.05 cm2 active area) are also fabricated to demonstrate the A and B are cations.[5] In the organo-
scalability of this approach, achieving η of 11.5%. Our results pave the way metal halide perovskites, the anion X is a
toward the implementation of MoS2 as a material able to boost the shelf life of halogen, generally I, although Br and Cl
large-area perovskite solar cells in view of their commercialization. are also used.[5] The A cation is organic,
usually methylammonium (CH3NH3+),
or less frequently ethylammonium
(CH3CH2NH3+) and formamidinium
+ [5]
(NH2CH NH2 ). The B cation is predominantly Pb in the
1. Introduction
PSCs with the highest reported η = 22.1%.[8,9] Sn was also
Perovskite solar cells (PSCs) are emerging as the most prom- used[10] to increase photon absorption.[11] However, Sn suffers
ising solution-processed photovoltaic technology.[1–3] The oxidation issues strongly undermining the cell stability.[10] Cur-
interest stems from several key features shown by the hybrid rently, the standard compound for PSCs is methylammonium
organometal halide perovskites used as photoactive materials, lead triiodide (CH3NH3PbI3).[5] For the realization of PSCs, two
such as intense light absorption coefficient (1.5 × 104 cm−1 different approaches have been proposed: the so-called meso-
at 550 nm),[4,5] long electron and hole diffusion length (from structured[1,2,7] and planar-heterojunction[3,12,13] PSCs. Meso-
100 nm to 1 mm),[4] high carrier mobility (1–10 cm2 V−1 s−1),[6] structured PSCs rely on a mesoporous layer (typically made of
nanocrystalline TiO2[1,2,7]) acting as scaffold for the perovskite
growth.[14] Planar-heterojunction PSCs are instead made of
Dr. A. Capasso, L. Najafi, Dr. V. Pellegrini, a compact perovskite layer, which is grown without any sup-
Dr. F. Bonaccorso porting structure.[13,15]
Istituto Italiano di Tecnologia Electron (ETL) and hole (HTL) transporting layers are added
Graphene Labs in PSCs between the photoactive layer and the electrodes
Via Morego 30, Genova 16163, Italy
E-mail: francesco.bonaccorso@iit.it in order to separate the generated charges and collect them
Dr. F. Matteocci, Dr. L. Cinà, Prof. A. D. Carlo (holes at the photoanode and electrons at the back electrode)
Department of Electronic Engineering efficiently.[16] For what regards the ETL, compact layers of TiO2
University of Roma “Tor Vergata” (50–80 nm)[17] or ZnO (25 nm)[18] are usually chosen to prevent
Center for Hybrid and Organic Solar Energy (CHOSE) holes recombination at the photo-anode. Concerning the HTL,
Rome 00173, Italy
the standard de facto material for high performance (i.e., high
E-mail: aldo.dicarlo@uniroma2.it
η) PSCs is 2,2′,7,7′-tetrakis-(N,N-di-4-methoxyphenylamino)-
L. Najafi
Dipartimento di Chimica e Chimica Industriale 9,9′-spirobifluorene (Spiro-OMeTAD).[1,3,7,19,20] Spiro-OMeTAD
Università degli Studi di Genova can be easily integrated in PSCs since it can be solution-pro-
Via Dodecaneso, 31, Genova 16146, Italy cessed in water-based solvents, having also a high glass-tran-
Dr. M. Prato, Dr. J. Buha sition temperature (120 °C).[19,21] Thanks to the highest-occu-
Istituto Italiano di Tecnologia pied molecular orbital (HOMO) energy level of 5.03 eV,[22]
Nanochemistry Department
Via Morego, 30, Genova 16163, Italy which is a value approaching the valence band maximum
(VBM) of CH3NH3PbI3 (5.4 eV)[23] and CH3NH3PbI3−xClx
DOI: 10.1002/aenm.201600920 (5.3 eV),[24] Spiro-OMeTAD effectively transfers holes from the

Adv. Energy Mater. 2016, 1600920 © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com (1 of 12)  1600920
 www.advenergymat.de
CH3NH3PbI3 to the photoanode.[7] However, Spiro-OMeTAD
FULL PAPER
is expensive (Solaronix, ≈565 = C g−1),[25] and for this reason
many other materials were proposed and tested as alterna-
tives.[9,26–36] The long list of materials used as HTL instead
of Spiro-OMeTAD includes: inorganic compounds such as
copper(I) iodide (CuI),[26] copper(I) thiocyanate (CuSCN),[27]
and nickel(II) oxide (NiO);[28] organic polymers such as poly(3-
hexylthiophene) P3HT,[29] poly(3,4-ethylenedioxythiophene)
polystyrene sulfonate (PEDOT:PSS),[30] poly[2,5-bis(2-decyl-
dodecyl)pyrrolo[3,4-c]pyrrole-1,4(2H,5H)-dione-(E)-1,2-di(2,2′-
bithiophen-5-yl) ethene] (PDPPDBTE),[31] and poly(triaryl
amine) (PTAA),[9,32,33] modified fluorene–dithiophene (FDT)[37]
as well as carbon-based nanomaterials such as graphene oxide
(GO)[34] and reduced graphene oxide (RGO).[34–36] In particular,
PSCs based on PTAA[9] and FDT[37] HTLs have reached η = 20%
(comparable to Spiro-OMeTAD), motivating further research
and optimization on these materials. Nevertheless, despite
these recent results,[9,37] Spiro-OMeTAD still represents the ref-
erence HTL material used for the fabrication of PSCs.
A more significant limitation that poses severe concerns in
view of the possible commercialization of PSCs is their lifetime
stability,[38] which is adversely affected by the presence of the
Spiro-OMeTAD.[39] This is due to the fact that Spiro-OMeTAD,
in its pristine form (amorphous), has low electrical conduc-
tivity (σ),[40] i.e., ≈10−8 S cm−1.[41] In order to increase the σ with
the target to reach high device η,[8,9] Spiro-OMeTAD is usu-
ally doped with a combination of tert-butylpyridine (TBP)[42,43]
and lithium bis(trifluoromethanesulfonyl)imide (Li-TFSI).[44]
Such doping process successfully increases its hole mobility
(from 10−5 to 10−3 cm2 V−1 s−1)[45] and hence the σ (from 10−8
to 10−3 S cm−1).[41] The doping process has also an effect on the
HOMO level, which increases from 5.03 to 5.22 eV.[46] Unfor-
tunately, the downside of such doping process is the reduced
stability of the PSCs, which is currently the main drawback of
such technology.[38,39] The Li-TFSI tends to desorb during cell
operation, reducing the cell η (a 20% decrease in the first 120 h
at room temperature with a humidity of ≈40%);[47] TBP instead
causes the corrosion of the perovskite layer strongly affecting
the η of the cells over time.[43,48]
Another factor limiting the device stability regards the
interface created by the Spiro-OMeTAD and the perovskite
layer.[49,50] Recent analysis using in-situ TEM characterization of
PSCs operating under illumination reported evidence of iodine
migration from the perovskite layer to the Spiro-OMeTAD,
with consequent degradation of the cell η.[49] Furthermore, it is
known[51] that a direct contact between the perovskite and the
metal electrode results in high recombination losses, which
results in a decrease of VOC and FF.[50] Such losses can occur
in solution-processed PSCs due to the incomplete coverage
of the perovskite by the HTL.[52] Another source of loss is the
metal electrode that can also get in contact with the perovskite
by migrating through the HTL, as demonstrated in organic
solar cells (OSCs) (i.e., metal migration through the photoac-
tive organic layer during device operation).[53] One strategy to
minimize such detrimental effect could be the increase of the
HTL thickness, which is however associated with an increase
of the PSCs series resistance, determining a reduction of the
solar cell η.[54] Guarnera et al. proposed the addition of an insu-
lating mesoporous Al2O3 layer between the perovskite and the
1600920  (2 of 12) wileyonlinelibrary.com © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.MaterialsViews.com
Spiro-OMeTAD layer to prevent shunting pathways, reporting
an increase in the lifetime stability of the devices over 350 h of
operation (starting η of 13% that decreased by 5% after 350 h,
for the best cell).[55] Such Al2O3 layer has a “passive” protective
function, because it physically prevents the contact between the
Au electrode and the perovskite layer.[50,52,55]
Here we successfully implement liquid-phase exfoliated few-
layer MoS2 flakes in solution-processed PSCs with glass/FTO/
compact-TiO 2/mesoporous-TiO 2/CH 3NH 3PbI 3/MoS 2/Spiro-
OMeTAD/Au structure to create an “active” buffer layer (ABL)
between the Spiro-OMeTAD and the perovskite layer with a dual
function: (i) barrier toward the metal electrode migration, and
(ii) additional energy-matching layer to further ease the hole
­collection of the Spiro-OMeTAD. The presented approach allow
us to build metal–organic PSCs with η matching that of the
Spiro-OMeTAD reference PSCs, but with superior lifetime sta-
bility over a period of 550 h. In fact, MoS2-based PSCs (0.1 cm2
active area), show η = 13.3% and higher lifetime stability after 550
h with respect to reference cells (Δη/η = −7% vs. Δη/η = −34%).
We have also fabricated large-area cells (1.05 cm2 active area)
obtaining a η = 11.5%, demonstrating the versatility of the ABL
toward the scaling up of the cell size.
2. Results and Discussion
The interest in MoS2 flakes for photovoltaic applications[56]
stems originally from a high charge carriers mobility (up to
≈470 cm2 V−1 s−1 for electrons, and ≈480 cm2 V−1 s−1 for holes)[57]
and chemical inertness of the basal-plane flake surface.[58] The
solution-processed MoS2 flakes were already exploited as HTL
in organic solar cells (OSCs),[59,60] achieving η comparable
with the ones based on standard HTLs (i.e., MoO3[61] and
PEDOT:PSS[62]). In particular, Gu et al. fabricated OSCs based
on PTB7:PB71BM using MoS2 HTL obtaining η = 8.1%, which
is higher than the reference cell with thermally evaporated
MoO3 (η = 7.5%).[60] Kelvin Probe microscopy analysis indicated
that MoS2 flakes have a surface potential analogous to that
of MoO3 (the energy offset between the two materials is only
0.02 eV), thus working efficiently as hole collector in OSCs.[60]
We carried out the liquid-phase exfoliation (LPE) of MoS2,
see Experimental Section for details, in N-methyl-2-pyrrolidone
(NMP) solvent, since it has a surface tension matching the sur-
face energy of the exfoliated flakes, which thus tend to be sta-
bilized against re-aggregation.[63,64] NMP was found to provide
higher concentration [0.023 g L−1] of exfoliated flakes (single or
few layer) in the processed dispersions than low boiling point
solvents such as isopropyl alcohol (IPA, 0.009 g L−1) and ace-
tone (0.007).[65] Although NMP is the most effective solvent for
the MoS2 exfoliation, in the case of PSCs its use is problematic
because it dissolves the perovskite crystals during the HTL pro-
cessing.[37,52] Moreover, NMP has a high boiling point (≈202 °C)
and post-processing treatments are then required to remove its
residues on the flakes after the process.[66] Additionally, the exfo-
liation of MoS2 in NMP results in the presence of superficial
oxidized Mo species,[67] which have to be carefully quantified
and removed before device fabrication. To address these limita-
tions and implement MoS2 few-layer flakes in the fabrication
of a buffer layer between Spiro-OMeTAD and perovskite layer
Adv. Energy Mater. 2016, 1600920
 www.advenergymat.de
www.MaterialsViews.com

FULL PAPER
Figure 1.  Characterization of the as-exfoliated MoS2 sample in NMP and IPA. a) Absorption spectra of the dispersion of MoS2 flakes in NMP and in
IPA. b) Raman spectra at 532 nm excitation wavelength of bulk and exfoliated MoS2 cast on Si/SiO2. c) Representative AFM image of MoS2 flakes
deposited on Si/SiO2 with the corresponding d) thickness statistics (calculated on 30 flakes measured).

in PSCs, we have developed an ad-hoc preparation procedure.
After the LPE of MoS2 carried out in NMP solvent we carried
out a solvent-exchange process in isopropyl alcohol (IPA), see
Experimental Section for details. The compatibility of the IPA
solvent with the perovskite layer was preliminarily assessed in
previous experiments.[68]
Optical absorption spectroscopy was carried out on both the
MoS2 samples in NMP and IPA. The UV–vis spectra of the two
samples (Figure 1a) show the main features typical of exfoliated
MoS2 flakes:[69] (i) a first absorption onset at ≈700 nm due to a
direct transition at the K point; (ii) two peaks at ≈682 nm and
at ≈622 nm due to A and B excitonic transitions, respectively;[65]
(iii) a second absorption onset at ≈500 nm, due to excitonic transi-
tions from the valence band to the conduction band.[58] The con-
centration of the MoS2 samples in NMP and IPA is estimated
to be 0.04 and 0.02 g L−1, respectively, see Experimental Section
for details. Raman spectroscopy was performed on bulk MoS2 as
well as on the e­ xfoliated flakes in NMP and IPA. In Figure 1b,
the Raman spectra of the NMP- and IPA-based samples are com-
pared to the one of bulk MoS2. The bulk MoS2 spectrum shows
the two dominant peaks centered at ≈375 cm−1 and ≈402 cm−1
corresponding to the E12g and A1g MoS2 modes, respectively.[70]
For the exfoliated MoS2 flakes in NMP, the E12g and A1g modes
appear at ≈379 cm−1 and 405 cm−1, respectively. The MoS2 flakes
in IPA sample show the two peaks at ≈377 cm−1 and 403 cm−1.
The Raman analysis demonstrates that the solvent-exchange
process does not affect the structure of the LPE MoS2 flakes.
The thickness of the exfoliated MoS2 flakes is evaluated by
atomic force microscopy (AFM) measurements. The MoS2
flakes shown in Figure 1c have a thickness up to 3 nm. The
thickness statistics reported in Figure 1d confirms that the
flakes’ thickness is in the 1–5 nm range, indicating the pres-
ence of few-layer MoS2 flakes.
To further characterize the material before and after the sol-
vent exchange process, the MoS2 flakes in NMP and IPA were
analyzed by transmission electron microscopy (TEM). The TEM
observations indicate that the MoS2 flakes in both samples have
lateral sizes in the 200–600 nm range. A detailed TEM charac-
terization of the materials present in NMP is summarized in
Figure 2.
The sample in NMP exhibits a mixture of exfoliated MoS2
flakes and amorphous oxygen-rich phases (i.e., Mo–O), which
are formed as by-products during the exfoliation of bulk MoS2
in NMP. The aforementioned analysis was repeated in various
areas of the TEM grid confirming this result. Within aggregates
of crystalline MoS2 flakes, such as that shown in Figure 2a (the
corresponding polycrystalline electron diffraction pattern is
shown in Figure 2b)[71] and Figure 2c, the energy-filtered trans-
mission electron micro­scopy (EFTEM) mapping revealed also
the presence of a Mo–O phase (Figure 2c,g). Unlike the fully
crystalline MoS2 flake outlined by a triangle in Figure 2c,h,
the Mo–O phase, outlined by a circle in the same Figures, was
found to be amorphous. Oxygen was detected in the crystal-
line MoS2 flakes (Figure 2f), while the nitrogen signal in the

Adv. Energy Mater. 2016, 1600920 © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com (3 of 12)  1600920
 www.advenergymat.de
www.MaterialsViews.com
FULL PAPER

Figure 2.  Transmission electron microscopy characterization of the as-exfoliated sample in NMP. a) An aggregate of MoS2 flakes and b) the corre-
sponding polycrystalline electron diffraction pattern indexed according to the 2H-MoS2 structure. c) Another aggregate of flakes with the corresponding
d) Mo, e) S, f) O, and g) N EFTEM maps. A Mo- and S-rich flake outlined by a triangle in c)–h) with the corresponding Fourier transform (FT) in
(i) structurally matches the MoS2 phase in its [001] zone axis. A Mo- and O-rich flake outlined by a circle in c)–h) is amorphous as determined by
(h) high-resolution TEM (HRTEM) and (j) the corresponding FT. k) HRTEM image from one of the MoS2 flakes shown in (c) oriented in the [001]
zone axis having a 2H structure (a sketch of the arrangement of Mo and S atoms in this structure and orientation is superimposed on the HRTEM
image). l) Intensity profile taken along the [010] direction in (k) (indicated by a dashed line) exhibits the periodicity of Mo and S atomic columns
characteristic of the 2H structure. On the other hand, a periodicity of four atomic columns along (m) the [010] direction, characteristic of the 1T
structure, was observed in a region of another flake (outlined by a yellow dashed line) shown in (n), while the surrounding region of the same flake
exhibited the 2H structure.

EFTEM (Figure 2g) is expected to originate from the NMP res- transform (FT) in Figure 2i), which is an indication of the exfo-
idue on the surface of the flakes. Most of the MoS2 flakes were liation process occurring along their basal planes.
found be oriented in their [001] zone axis (e.g., a flake outlined The TEM observations of the purified MoS2 sample dispersed
by a triangle in Figure 2h with the corresponding Fourier in IPA (used in the following for the device fabrication) indicate

1600920  (4 of 12) wileyonlinelibrary.com © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Energy Mater. 2016, 1600920
 www.advenergymat.de
www.MaterialsViews.com

FULL PAPER
Figure 3.  Transmission electron microscopy characterization of the purified sample in IPA used for the device fabrication. a) MoS2 flakes aggregate
and b) the corresponding electron diffraction pattern. c) HRTEM and d) scanning TEM (STEM) image of a flake consisting of several layers and more
than one flake overlapped with each other (based on the Moiré fringes visible in (c), with the corresponding e) Mo, f) S, g) O, and h) N EFTEM maps.
A region outlined by a rectangle in (c) is shown enlarged in (i) with the corresponding FT in the inset. j) Close-up view of the atomic arrangement in
this region. Sketch of the arrangement of the Mo and S atoms viewed from [001] direction is superimposed on the HRTEM image. The blue triangles
in (j) indicate the distance between Mo atoms along the [010] direction of MoS2, while the red triangles indicate two atomic columns in between, cor-
responding to S columns (note that the Mo and S atoms are overlapped along the [001] direction for two or more layers thick crystal and not clearly
distinguishable by HRTEM).

the successful removal of solvent residues as compared to the
starting sample, i.e., as-exfoliated in NMP. A typical aggregate
of MoS2 flakes is shown in Figure 3a with the corresponding
electron diffraction pattern given in Figure 3b. The EFTEM
mapping of a several layers thick flake, shown in Figure 3c,
confirms the presence of Mo and S, without clear indication
of strong preferential O and N presence associated with the
flakes. No amorphous Mo–O aggregated with MoS2 flakes has
been observed either. The crystalline structure of the MoS2 flake
shown in Figure 3c was further analyzed by HRTEM (the region
outlined by the rectangle in Figure 3c is enlarged in Figure 3i,j).
To further clarify the nature of chemical bonding in the sam-
ples before and after the solvent exchange process, we acquired
­ igh-resolution X-ray photoelectron spectroscopy (XPS) data on
h
both samples in NMP and IPA, focusing on the energy regions typ-
ical for N, Mo, and S peaks (see Experimental Section for details).
The XPS spectra collected over these energy regions are reported
in Figure 4 together with the results of the data fitting procedure.
The XPS signals collected on the NMP-based sample are
reported in Figure 4a,b, for different energy ranges, named
Region 1 and Region 2, respectively (see Experimental Section
for further details). Region 1 is dominated by an intense and
narrow peak centered at binding energy of (400.5 ± 0.2) eV,
corresponding to the N 1s peak due to NMP residuals.[72] By
quantitative analysis of the XPS data over Regions 1 and 2, we
estimated a N:Mo ratio close to 4:1 (calculated considering the

Adv. Energy Mater. 2016, 1600920 © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com (5 of 12)  1600920
 www.advenergymat.de
www.MaterialsViews.com
FULL PAPER

Figure 4.  X-ray photoelectron spectroscopy spectra collected on the MoS2 flakes cast on Si from the two dispersions (in NMP and IPA) over a,c) Region 1
and b,d) Region 2, as defined in the Experimental Section. In order to simplify the comparison between the two samples, the data are shown over intensity
ranges such to keep constant the intensity ratio between the Mo 3p3/2 and the Mo 3d5/2 components for MoS2. See text for details.

areas of N 1s and Mo 3d peaks). This result indicates that, after
the exfoliation step, the sample still contains NMP residuals.
Analysis of the data collected in Region 2 indicates that Mo
oxides are present, together with MoS2, thus confirming the
results obtained by the TEM analysis. We obtained a MoS2:MoOx
ratio of 0.6 (MoOx being the sum of the MoO3 and Mo2O5 con-
tributions), confirming that the exfoliation process produced Mo
oxides as by-products, as reported in ref. [67]. The XPS signals
acquired after the solvent-exchange process to IPA are reported
in Figure 4c,d, for Region 1 and Region 2, respectively.
Noteworthy, the solvent-exchange process allowed to signifi-
cantly increasing sample purity. Indeed, the N content (i.e., the
amount of residual NMP molecules) is now negligible, as wit-
nessed by the absence of N signal in the (400.5 ± 0.2) eV range.
Concerning the presence on Mo oxides, the MoS2:MoOx ratio is
now increased to ≈8.5:1. Therefore, the solvent-exchange pro-
cess has proven to be efficient in removing the NMP molecules
adsorbed on the MoS2 flakes. By exploiting this procedure it is
thus possible to obtain almost pristine MoS2 flakes, having only
traces of residual Mo oxides.
The MoS2 flakes in IPA were spin-coated onto the perovskite
layer to create the ABL. Scanning electron microscopy (SEM)
images were taken after the deposition to evaluate the surface
coverage (Figure 5). MoS2 flakes coating the perovskite layer

Figure 5.  Scanning electron microscopy images of the MoS2 flakes spin-coated onto the perovskite layer of the cell. a) Image in false colors (dark
cyan for MoS2 flakes and purple for the underlying perovskite layer) highlighting the coverage provided by the spin-coating process of the MoS2 flakes.
b) MoS2 flakes on top of the perovskite crystals.

1600920  (6 of 12) wileyonlinelibrary.com © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Energy Mater. 2016, 1600920

www.MaterialsViews.com
Figure 6.  I/V curves of the cells fabricated with MoS2 HTL and without
HTL (0.1 cm2 active area).
are observed. By systematically exploring all the parameters
involved in the MoS2 deposition process (i.e., spin coating time,
speed, and number of coatings, post-annealing steps), we were
able to form a planar layer of flakes that covers the ­perovskite
crystals. Although some of the perovskite crystals remain
uncovered, we chose to adopt these deposition parameters as a
tradeoff between uniform and planar deposition and minimal
buffer layer thickness.
The efficiency of the MoS2 layer for the holes collection is
expected to reside in the alignment between the VBM of MoS2
and MAPbI3 (i.e., a small energy offset between the two bands
results in an efficient hole collection at the Au electrode).[5,73]
We measured the VBM of the MoS2 layer deposited on a
Si substrate and found a value of −5.1 eV (Figure S3, Sup-
porting Information), confirming this small energy offset
(−5.4 eV for MAPbI3).[23] To prove further this point we tested
the MoS2 as HTL in cell composed by glass/FTO/compact-
TiO2/mesoporous-TiO2/CH3NH3PbI3/MoS2/Au structure, i.e.,
without the Spiro-OMeTAD layer. As shown in Figure 6, such
cells revealed a significant increase in η with respect to cells
with no HTL (4.5% vs. 1.5%), confirming that the MoS2 flakes
ease the hole transport and collection at the Au electrode. This
result is in line with our previous experiments concerning
MoS2 flakes directly exfoliated in IPA and deposited as HTL.[68]
After this preliminary test, we fabricated PSCs with glass/
FTO/compact-TiO 2/mesoporous-TiO 2/CH 3NH 3PbI 3/MoS 2/
Spiro-OMeTAD/Au structure to fully assess the performance
of MoS2 as ABL (see the Experimental Section for experi-
mental details). The results of the I/V measurements under
illumination are reported in Figure 7a. In comparison with
the reference PSCs, the cells with MoS2 buffer layer show
a decrease in VOC (0.93V vs. 1.01 V) and fill factor (66.7% vs.
74.6%). However, the JSC is found to significantly increase
(−21.5 vs. −18.8 mA cm−2) bringing the η to 13.3% (η = 14.2%
for the reference PSCs). After an ageing test of 170 h, the PSCs
were measured again under illumination (Figure 7b and S6,
Supporting Information). Remarkably, the PSCs with MoS2
buffer layer have a η of 13.5% vs. 10.9% of the reference cell.
The I/V characteristics of the PSCs were further measured
Adv. Energy Mater. 2016, 1600920 © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advenergymat.de
FULL PAPER
Figure 7. a) I/V curves of the cells measured after fabrication (0.1 cm2 active
area). b) η trend of cells with (blue) and without (red) MoS2 over 550 h.
after 550 h from fabrication (Figure 7b and S5, Supporting
Information). These results show better stability of the MoS2-
based sample with respect to the reference, i.e., a η of 12.4%
vs. 9.3%. The MoS2–based PSC shows a decrease in η of only
7%, i.e., a much lower value than the reference PSC (−34%).
The improved stability of the MoS2-based PSC with respect to
the reference cell is ascribed to the surface passivation of the
perovskite layer provided by the MoS2 ABL, which prevents the
iodine migration from the perovskite into the Spiro-OMeTAD[49]
and the formation of Au pathways from the metal electrode to
the perovskite layer.[74]
To verify whether the deposition of the MoS2 ABL can be a
viable approach in view of up scaling of the cells size,[75] we also
fabricated PSCs with an active area of 1.05 cm2 (Figure 8). The
I/V characteristics large-area cells are reported in Figure 8a, while
Figure 8b displays a photograph of the as-prepared large-area cell.
Remarkably, the η of the cell with the MoS2 ABL achieved
now the same value as the reference cell (11.5% vs. 11.4%),
having a higher JSC with respect to the reference PSC (−18.5 vs.
−17.5 mA cm−2), confirming the results obtained for small-area
cells, reported in Figure 7.
A statistical analysis of the I/V results is reported in Table S1
(Supporting Information) for both small and large area devices.
wileyonlinelibrary.com (7 of 12)  1600920
 www.advenergymat.de
www.MaterialsViews.com
FULL PAPER

Figure 8. a) I/V curves of the large-area cells (1.05 cm2 active area) with (blue curve) and without MoS2 (red curve) measured under illumination
(AM1.5G 100 mW cm−2) using class A sun simulator. Mean photovoltaic parameters for a batch of six large-area PSCs made with the MoS2 ABL: VOC =
0.943 ± 0.026 V, JSC = −18.11 ± 0.65 mA cm−2, FF = 62.8 ± 3.4%, η = 10.7 ± 0.4%. b) Photograph of a large-area cell with the MoS2 ABL.

The increase in JSC is further testified by the incident photon-

to-electron conversion efficiency (IPCE) measurements of
the two cells (Figure 9). The IPCE graphs of the two cells in
Figure 9 have analogous spectral trends. However, the curve
associated with the MoS2-based cell shows a higher IPCE value
of around 10% in the 350-750 nm range, with respect to the
reference. These IPCE trends are in agreement with the inte-
grated current density values calculated from 300 to 850 nm at
AM1.5G condition, which are 18.03 and 16.37 mA cm−2 for the
PCSs with MoS2 and the reference, respectively.
In order to investigate the influence of the MoS2 ABL on the
regeneration efficiency of the perovskite and on the recombina-
tion pathways of the TiO2-free carriers, we tested the dynamic
performance under pulsed light condition by means of tran-
sient photovoltage measurements (TPV), reported in Figure 10.
The photo-voltage rise test (Figure 10a) is carried out on the
PSCs in steady state operating conditions, i.e., under dark at
open circuit, and switching on the light source to monitor the

Figure 10.  Transient photovoltage measurements of the cells with (blue

curve) and without (red curve) MoS2 ABL made with a switchable LED
Figure 9.  Left axis: IPCE spectra of the cells with (blue curve) and without lamp. a) Transient VOC rise, (from dark to light condition). b) Transient
(red curve) MoS2 ABL. Right axis: integrated current densities in the two VOC decay (from light to dark condition). The curves were fitted indicating
cases (AM1.5G). the constant times (τ).

1600920  (8 of 12) wileyonlinelibrary.com © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Energy Mater. 2016, 1600920

www.MaterialsViews.com
subsequent rise in photovoltage. The transient rise time of the
VOC is generally correlated to two main processes: the elec-
tron transfer from the perovskite to TiO2, and the hole transfer
from the perovskite to the HTL.[76] The PSCs with MoS2 ABL
show faster VOC rise time with respect to the reference cell
(Figure 10a), indicating a more efficient hole transfer at the per-
ovskite/MoS2 interface than at the perovskite/SpiroOMeTAD
one.[77]
Finally, to account for the difference in VOC observed in
Figure 7a between cells with and without MoS2 ABL (0.93 V vs.
1.01 V), the transient decay curves of the two PSCs were meas-
ured (see Figure 10b, where a tri-exponential fitting of the
curves is also reported). The reference PSC is characterized by a
slow Voc decay trend with a τ3 time constant of around 22 s, one
order of magnitude higher than that of the cell with the MoS2
ABL (τ3 = 2 s). Considering that the Voc decay trend is given
by the recombination of the free charges at the perovskite/HTL
interface,[36] this indicates that the presence of the MoS2 ABL
can activate additional recombination mechanisms with respect
to the configuration where only the bare Spiro-OMeTAD is
present. In fact, the MoS2 ABL device shows a sudden voltage
drop from +0.6 to 0 of normalized Voc in comparison with the
reference PSC with Spiro-OMeTAD (Figure 10b).
These results are in agreement with the I/V characteristic
under dark conditions (Figure S7, Supporting Information),
where the PSC with the MoS2 ABL has higher dark current
values than the reference PSC in the same voltage range from
0 to +0.6V. As already shown however, this VOC decrease is
compensated by the increase in JSC (Figure 7a) provided by the
MoS2 ABL, which overall also stabilize the PSCs.
3. Conclusion
We demonstrated the use of MoS2 flakes as an active buffer
layer in hybrid lead halide perovskite solar cells with glass/
FTO/compact-TiO 2/mesoporous-TiO 2/CH 3NH 3PbI 3/MoS 2/
Spiro-OMeTAD/Au structure. The MoS2 flakes are first exfoli-
ated in NMP and successively re-dispersed in IPA by using a
solvent-exchange process. The MoS2 flakes are then spin-coated
onto the perovskite layer before the deposition of the Spiro-
OMeTAD HTL and the Au electrode. Reference PSCs without
MoS2 have shown a η of 14.2% diminishing to η = 9.3%
after an endurance test of 550 h. MoS2-based PSCs instead
reached a η of 13.3% after device fabrication which decreased
to η = 12.4% after 550 h, proving the enhanced stability of
the cell due to the presence of MoS2 layer acting as an active
buffer layer. We have also demonstrated the scalability of our
approach by fabricating large-area cells (1.05 cm2), obtaining
η = 11.5%. Our results pave the way towards the implementa-
tion of MoS2 to boost the shelf life of PSCs in view of their
commercialization.
4. Experimental Section
Materials: MoS2 powder, NMP (99% purity), and IPA (absolute
alcohol, without additive, ≥99.8%) were purchased from Sigma–Aldrich
and used without further purification.
Adv. Energy Mater. 2016, 1600920 © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advenergymat.de
Liquid-Phase Exfoliation of MoS2: Bulk MoS2 was exfoliated in NMP,
FULL PAPER
obtaining dispersions of MoS2 flakes. 100 mg of MoS2 was dispersed in
10 mL of NMP and ultrasonicated (Branson 5800) for 6 h. The obtained
dispersion was ultracentrifuged at 3000 g (Beckman Coulter Optima
XE-90 with a SW41Ti rotor) for 20 min at 15 °C, to remove thick and
un-exfoliated MoS2 flakes. After the ultracentrifugation process, 90% of
the supernatant was collected by pipetting.
Solvent-Exchange Process: After the MoS2 exfoliation in NMP, we used
a solvent-exchange process to re-disperse the MoS2 flakes in IPA. The
NMP solvent was successively removed from the supernatant dispersion
with a rotary evaporator (70 °C at 1 mPa). The precipitate was collected
and washed with IPA using an ultracentrifuge at 77 000 g for 10 min for
three times to optimize the NMP removal. After the last washing step,
the IPA solvent was evaporated at 60 °C to collect the MoS2 powder.
The flakes (27 mg) were dried in vacuum at 200 °C for 2 h to further
purify the material, which was then re-dispersed in 1 mL of IPA. This
solvent-exchange process has a dual function: i) it removes the NMP,
detrimental for the perovskite layer; ii) it reduces the presence of Mo
oxide. Finally, the purified MoS2 flakes in IPA are deposited on the
perovskite layer of the cell by spin coating.
Characterization: The MoS2 dispersions were characterized by optical
absorption spectroscopy in the range 300–900 nm with a Cary Varian
6000i UV–vis–NIR spectrometer. The absorption spectra were acquired
using a 1 mL quartz glass cuvette. The solvent baseline was subtracted
to the recorded spectrum. The concentration of MoS2 flakes in the
dispersions can be estimated from the optical absorption coefficient at
672 nm, using A = αlc where l [m] is the light path length, c [g L−1] is the
concentration of the dispersed flakes, and α [L g−1 m−1] is the absorption
coefficient, with α ≈ 3400 L g−1 m−1 at 672 nm.[65] Raman spectroscopy
(Renishaw inVia confocal Raman microscope) was performed on the
MoS2 flakes collected from the dispersions. The samples were drop-
cast on a Si/SiO2 (300 nm SiO2) substrate and dried under vacuum. An
excitation wavelength of 532 nm was used with a 50× objective (incident
power of ≈1 mW on the sample). The MoS2 flakes, drop-cast on Si/SiO2
substrate, were characterized morphologically by AFM with an Innova
AFM (Bruker, Santa Barbara, CA). The measurements were taken in
tapping mode with a NTESPA 3.75 mm cantilever (Bruker, 300 kHz k:
40 N m−1), in air at room temperature, with a relative humidity less
than 30%.
The crystal structure, composition, and chemical bonding of the
samples in NMP and IPA were analyzed by means of TEM and XPS.
The samples for the TEM analysis were prepared by drop-casting the
MoS2 dispersion samples onto ultrathin carbon on holey carbon-coated
copper TEM grids. A low-resolution TEM characterization was carried
out using a JOEL JEM 1011 TEM operated at 100 kV. The HRTEM, STEM,
EFTEM, and energy-dispersive X-ray spectroscopy (EDX) mapping were
performed using a JEOL JEM 2200FS microscope operated at 200 kV
and equipped with a field emission gun, in-column Omega energy filter
and CEOS image Cs corrector and Bruker Quantax EDS system with a
XFlash 6T-60 silicon drift detector. A FEI Tecnai F20 operated at 200 kV
and equipped with Gatan Enfinum SE spectrometer was used for the
compositional analysis by means of electron energy loss spectroscopy
(EELS). The EELS data were collected in the TEM mode with a collection
semi-angle of 100 mrad.
The XPS analysis was carried out using a Kratos Axis Ultra
spectrometer on samples drop cast onto silicon wafers. The XPS spectra
were acquired using a monochromatic Al Kα source operated at 20 mA
and 15 kV. The analyses were carried out on a 300 μm × 700 μm area.
Wide scans were collected with pass energy of 160 eV and energy step
of 1 eV; high-resolution spectra were instead acquired at pass energy of
10 eV and energy step of 0.1 eV. The Kratos charge neutralizer system
was used on all specimens. Spectra have been charge corrected to
the main line of the C 1s spectrum set to 284.8 eV, and analyzed with
CasaXPS software (version 2.3.17). We acquired high-resolution XPS
data focusing on two binding energy regions: the first (named Region
1) is comprised between 388 and 425 eV, and it is typical for Mo 3p
and N 1s peaks;[78] the second (Region 2) is comprised between 222 and
243 eV, and is typical for Mo 3d and S 2s peaks.[78] The analysis of
wileyonlinelibrary.com (9 of 12)  1600920
 www.advenergymat.de
FULL PAPER
Region 1 allows us to identify possible residues of NMP in the samples
after each preparation step, by analyzing the N 1s peak. The analysis
of Region 2 is instead used to identify the evolution of the oxidation
state of Mo after exfoliation in NMP and after the solvent exchange in
IPA. In particular, fitting of the data in Region 2 is used to determine
the ratio between MoS2 and Mo oxides. The fitting has been performed
considering that, while N 1s and S 2s signals can be fitted with single
peaks, both Mo 3p and Mo 3d signals are actually doublets, due to spin–
orbit coupling.[79] Mo 3p doublets were fitted to pairs of peaks with the
constraints of having the same full width at half maximum (FWHM), a
spin–orbit splitting of 17.4 eV and a branching ratio of 1/2.[78] Similarly,
the Mo 3d doublets were fitted to pairs of peaks with the constraints
of having the same FWHM, a spin–orbit splitting of 3.14 eV and
a branching ratio of 2/3.[78] The position of each doublet has been
identified by the position of the component centered at lower binding
energy, i.e., Mo 3p3/2 for Mo 3p and Mo 3d5/2 for Mo 3d. The best fit of
the data over Region 1 has been obtained considering, together with the
N 1s peak, also two Mo 3p doublets, having their Mo 3p3/2 components
centered at (394.7 ± 0.3) eV and at (397.9 ± 0.3) eV. The observed
values are in good agreement with those reported in ref. [80] for MoS2
and MoO3, respectively. Region 2 is characterized by the presence
of four peaks at ≈226 eV (peak A), ≈229 eV (peak B), ≈232 (peak C),
and ≈236 eV (peak D). By fitting the data, we were able to assign peak
A to S 2s (peak centered at (226.0 ± 0.3) eV), peak B to the Mo 3d5/2
component for Mo(IV) as observed in MoS2 (peak centered at (228.8 ±
0.3) eV), and peak D to the Mo 3d3/2 component for Mo(VI) in MoO3
(peak centered at (235.9 ± 0.3) eV). The broad peak C is the results of
the overlapping of the Mo 3d3/2 component for Mo(IV) in MoS2 (peak
centered at (231.9 ± 0.3) eV) with the Mo 3d5/2 component for Mo(VI)
in MoO3 (peak centered at (232.7 ± 0.3) eV).[67] A further Mo 3d doublet
was needed for the best fitting, having its Mo 3d5/2 component centered
at (231.3 ± 0.3) eV. This value is in good agreement with the presence of
Mo2O5 species,[81] which could appear only in presence of MoO3 as the
result of partial de-oxidation. The positions of the aforementioned peaks
closely match with the data recently reported in ref. [67].
Ultraviolet photoelectron spectroscopy (UPS) analysis was performed
on drop-cast films of MoS2 flakes on Si, to estimate the position of the
valence band maximum (VBM) of the materials under investigation and
the effect of illumination on it. The measurements were carried out with
a Kratos Axis UltraDLD spectrometer using an He I (21.22 eV) discharge
lamp. The analyses were carried out on an area of 55 μm in diameter, at
pass energy of 5 eV and with a dwell time of 100 ms. The work function
(i.e., the position of the Fermi level with respect to vacuum level) was
measured from the threshold energy for the emission of secondary
electrons during He I excitation. A −9.0 V bias was applied to the
sample in order to precisely determine the low kinetic energy cut-off,
as discussed in ref. [82]. Then, the position of the VBM vs. vacuum
level was estimated by measuring their distance from the Fermi level,
according to the graphical method used in refs. [73,83].
Solar Cells Fabrication and Morphological Characterization: A raster
scanning laser (Nd:YVO4 pulsed at 30 kHz average output power
P = 10 W) was used to etch the FTO/glass substrates (Pilkington,
8 Ω cm−1, 25 mm × 25 mm), used as transparent conductive window.
Small area cell (0.1 cm2) and large area cell (1.05 cm2) were realized
using a specific pattern of the FTO substrates (the cell active area was
calculated from the overlap area between FTO and gold electrodes). The
patterned substrates were cleaned in an ultrasonic bath, using detergent
with de-ionized water, acetone and IPA (10 min for each cleaning
step). A 50 nm thick blocking TiO2 (BL-TiO2) layer was deposited
onto the patterned FTO by spray pyrolysis deposition.[17,77] A 150 nm
layer nanocrystalline mesoporous TiO2 (18NR-Tpaste Dyesol diluted
with terpineol and ethylcellulose) was screen-printed onto the BL-TiO2
surface and sintered at 480 °C for 30 min. The mesoporous-TiO2
thickness was measured using the profilometer Dektak Veeco 150. The
perovskite layer was deposited by solvent-engineering method using
spin-coating technique, which permits a complete perovskite coverage
of the mesoporous TiO2, avoiding recombination of the charge carriers
at the HTM/TiO2 interface.[32] To obtain the perovskite dispersion,
1600920  (10 of 12) wileyonlinelibrary.com © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.MaterialsViews.com
1.1 m PbI2 powder (Sigma Aldrich) and 1.1 m of CH3NH3I (Dyesol) were
dispersed in dimethyl sulfoxide and stirred at 70 °C overnight. 110 μL of
this dispersion was spin coated on the TiO2 substrate at 1000 rpm for
10 s and 5000 rpm 30 s. 200 μL of toluene was poured on the substrate
10 s prior the end of the second ramp. Then, the perovskite layer growth
was obtained after annealing step at 100 °C for 1 h. After the perovskite
growth, a layer of (1) doped Spiro-OMeTAD or (2) doped MoS2 exfoliated
flakes+Spiro-OMeTAD was deposited, as follows: (1) Spiro-OMeTAD
was spin coated at 2000 rpm for 20 s on the perovskite layer from a
dispersion in chlorobenzene (74 mg mL−1) doped by the addition of
27 μL of TBP, 16 μL of Li-TFSI dispersion (520 mg in 1 mL of acetonitrile
solvent) and 7 μL of cobalt additive (FK209, 0.25 m in acetronitrile).
(2) MoS2 exfoliated flakes in IPA were spin-coated onto the perovskite
layer at 1700 rpm for 45 s. Then, the samples were annealed for one
minute at 70 °C. After the MoS2 deposition, the perovskite surface of
test samples was analyzed by SEM to evaluate the uniformity of the
deposition. Spiro-OMeTAD was then deposited on the MoS2 flakes as
reported in (1). The Spiro-OMeTAD and MoS2 + Spiro-OMeTAD PSCs
were at last introduced into a high vacuum chamber (10−6 mbar) to
thermally evaporate Au back contacts (thickness 100 nm). No sealing
was applied to the fabricated cells.
Solar Cells Photovoltaic Characterization: The PSCs were tested under a
solar simulator (ABET Sun 2000, class A) at AM1.5G and 100 mW cm−2
illumination conditions (calibrated with a certified reference Si Cell,
RERA Solutions RR-1002). Incident light power was measured with a
Skye SKS 1110 sensor. The endurance test was carried out monitoring
the devices for the period of time specified in the text (when not under
measurement, the devices were kept unsealed into a desiccator in
dark conditions). Incident photon-to-current conversion efficiency was
measured using an apparatus made of an amperometer (Keithley 2612)
and a monochromator (Newport Mod. 74000). The cells were masked
with a aperture of 0.25 cm2. Dark I/V and transient measurements were
performed with a measurement system based on a high speed source
meter (0.6 MS s−1) and a white LED (Arkeo—Cicci research srl).
Supporting Information
Supporting Information is available from the Wiley Online Library or
from the author.
Acknowledgements
This project has received funding from the European Union’s Horizon
2020 research and innovation program under grant agreement No.
696656—GrapheneCore1.
Received: May 2, 2016
Revised: May 31, 2016
Published online:
[1] J. Burschka, N. Pellet, S.-J. Moon, R. Humphry-Baker, P. Gao,
M. K. Nazeeruddin, M. Graetzel, Nature 2013, 499, 316.
[2] N. J. Jeon, J. H. Noh, W. S. Yang, Y. C. Kim, S. Ryu, J. Seo, S. I. Seok,
Nature 2015, 517, 476.
[3] M. Liu, M. B. Johnston, H. J. Snaith, Nature 2013, 501, 395.
[4] A. Kojima, K. Teshima, Y. Shirai, T. Miyasaka, J. Am. Chem. Soc.
2009, 131, 6050.
[5] M. A. Green, A. Ho-Baillie, H. J. Snaith, Nat. Photonics 2014, 8, 506.
[6] C. Motta, F. El-Mellouhi, S. Sanvito, Sci. Rep. 2015, 5, 12746.
[7] K. Wojciechowski, M. Saliba, T. Leijtens, A. Abate, H. J. Snaith,
Energy Environ. Sci. 2014, 7, 1142.
[8] http://www.nrel.gov/ncpv/images/efficiency_chart.jpg, accessed:
April, 2016.
Adv. Energy Mater. 2016, 1600920

www.MaterialsViews.com
[9] W. S. Yang, J. H. Noh, N. J. Jeon, Y. C. Kim, S. Ryu, J. Seo, S. I. Seok,
Science 2015, 348, 1234.
[10] N. K. Noel, S. D. Stranks, A. Abate, C. Wehrenfennig, S. Guarnera,
A.-A. Haghighirad, A. Sadhanala, G. E. Eperon, S. K. Pathak,
M. B. Johnston, A. Petrozza, L. M. Herz, H. J. Snaith, Energy
Environ. Sci. 2014, 7, 3061.
[11] P. Umari, E. Mosconi, F. De Angelis, Sci. Rep. 2014, 4, 4467.
[12] a) O. Malinkiewicz, A. Yella, Y. H. Lee, G. Mınguez Espallargas,
M. Graetzel, M. K. Nazeeruddin, H. J. Bolink, Nat. Photonics
2014, 8, 128; b) H. Zhou, Q. Chen, G. Li, S. Luo, T.-B. Song,
H.-S. Duan, Z. Hong, J. You, Y. Liu, Y. Yang, Science 2014, 345, 542;
c) S. Casaluci, L. Cinà, A. Pockett, P. S. Kubiak, R. G. Niemann,
A. Reale, A. Di Carlo, P. J. Cameron, J. Power Sources 2015, 297, 504.
[13] C.-G. Wu, C.-H. Chiang, Z.-L. Tseng, M. K. Nazeeruddin,
A. Hagfeldt, M. Grätzel, Energy Environ. Sci. 2015, 8, 2725.
[14] a) N. Li, H. Dong, H. Dong, J. Li, W. Li, G. Niu, X. Guo, Z. Wu,
L. Wang, J. Mater. Chem. A 2014, 2, 14973; b) Y. Zhao, A. M. Nardes,
K. Zhu, Faraday Discuss. 2015, 176, 301.
[15] W. Wang, J. Yuan, G. Shi, X. Zhu, S. Shi, Z. Liu, L. Han, H.-Q. Wang,
W. Ma, ACS Appl. Mater. Interfaces 2015, 7, 3994.
[16] S. Chatterjee, A. Bera, A. J. Pal, ACS Appl. Mater. Interfaces 2014, 6,
20479.
[17] F. Matteocci, G. Mincuzzi, F. Giordano, A. Capasso, E. Artuso,
C. Barolo, G. Viscardi, T. M. Brown, A. Reale, A. Di Carlo, Org. Elec-
tron. 2013, 14, 1882.
[18] D. Liu, T. L. Kelly, Nat. Photonics 2014, 8, 133.
[19] U. Bach, D. Lupo, P. Comte, J. E. Moser, F. Weissörtel, J. Salbeck,
H. Spreitzer, M. Grätzel, Nature 1998, 395, 583.
[20] F. Di Giacomo, V. Zardetto, A. D’Epifanio, S. Pescetelli,
F. Matteocci, S. Razza, A. Di Carlo, S. Licoccia, W. M. Kessels,
M. Creatore, Adv. Energy Mater. 2015, 5, 1401808.
[21] a) S. A. Haque, T. Park, A. B. Holmes, J. R. Durrant, ChemPhy-
sChem 2003, 4, 89; b) J. E. Kroeze, N. Hirata, L. Schmidt-Mende,
C. Orizu, S. D. Ogier, K. Carr, M. Grätzel, J. R. Durrant, Adv. Funct.
Mater. 2006, 16, 1832; c) P. Zacharias, M. C. Gather, M. Rojahn,
O. Nuyken, K. Meerholz, Angew. Chem. Int. Ed. 2007, 46, 4388.
[22] T. Leijtens, I.-K. Ding, T. Giovenzana, J. T. Bloking, M. D. McGehee,
A. Sellinger, ACS Nano 2012, 6, 1455.
[23] J. H. Noh, S. H. Im, J. H. Heo, T. N. Mandal, S. I. Seok, Nano Lett.
2013, 13, 1764.
[24] A. Abrusci, S. D. Stranks, P. Docampo, H.-L. Yip, A. K.-Y. Jen,
H. J. Snaith, Nano Lett. 2013, 13, 3124.
[25] http://shop.solaronix.com/spiro-ometad.html, accessed: May, 2016.
[26] J. A. Christians, R. C. M. Fung, P. V. Kamat, J. Am. Chem. Soc. 2014,
136, 758.
[27] P. Qin, S. Tanaka, S. Ito, N. Tetreault, K. Manabe, H. Nishino,
M. K. Nazeeruddin, M. Grätzel, Nat. Commun. 2014, 5, 3834.
[28] K.-C. Wang, P.-S. Shen, M.-H. Li, S. Chen, M.-W. Lin, P. Chen,
T.-F. Guo, ACS Appl. Mater. Interfaces 2014, 6, 11851.
[29] a) H. Chen, X. Pan, W. Liu, M. Cai, D. Kou, Z. Huo, X. Fanga,
S. Dai, Chem. Commun. 2013, 49, 7277; b) F. Di Giacomo, S. Razza,
F. Matteocci, A. D’Epifanio, S. Licoccia, T. M. Brown, A. Di Carlo, J.
Power Sources 2014, 251, 152; c) Y. Guo, C. Liu, K. Inoue, K. Harano,
H. Tanaka, E. Nakamura, J. Mater. Chem. A 2014, 2, 13827.
[30] a) O. Malinkiewicz, C. Roldán-Carmona, A. Soriano, E. Bandiello,
L. Camacho, M. K. Nazeeruddin, H. J. Bolink, Adv. Energy
Mater. 2014, 4, 1400345; b) J. Seo, S. Park, Y. C. Kim, N. J. Jeon,
J. H. Noh, S. C. Yoon, S. I. Seok, Energy Environ. Sci. 2014, 7,
2642.
[31] Y. S. Kwon, J. Lim, H.-J. Yun, K. Yun-Hi, T. Park, Energy Environ. Sci.
2014, 7, 1454.
[32] N. J. Jeon, J. H. Noh, Y. C. Kim, W. S. Yang, S. Ryu, S. I. Seok, Nat.
Mater. 2014, 13, 897.
[33] S. Ryu, J. H. Noh, N. J. Jeon, Y. C. Kim, W. S. Yang, J. Seoa,
S. I. Seok, Energy Environ. Sci. 2014, 7, 2614.
Adv. Energy Mater. 2016, 1600920 © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advenergymat.de
[34] Z. Wu, S. Bai, J. Xiang, Z. Yuan, Y. Yang, W. Cui, X. Gao, Z. Liu,
FULL PAPER
Y. Jin, B. Sun, Nanoscale 2014, 6, 10505.
[35] a) J.-S. Yeo, R. Kang, S. Lee, Y.-J. Jeon, N. Myoung, C.-L. Lee,
D.-Y. Kim, J.-M. Yun, Y.-H. Seo, S.-S. Kim, S.-I. Na, Nano Energy
2015, 12, 96; b) K. Yan, Z. Wei, J. Li, H. Chen, Y. Yi, X. Zheng,
X. Long, Z. Wang, J. Wang, J. Xu, Small 2015, 11, 2269; c) T. Liu,
D. Kim, H. Han, A. R. Bin Mohd Yusoff, J. Jang, Nanoscale 2015, 7,
10708; d) Q. Luo, Y. Zhang, C. Liu, J. Li, N. Wang, H. Lin, J. Mater.
Chem. A 2015, 3, 15996.
[36] A. L. Palma, L. Cinà, S. Pescetelli, A. Agresti, M. Raggio,
R. Paolesse, F. Bonaccorso, A. Di Carlo, Nano Energy 2016, 22,
349.
[37] W. Nie, H. Tsai, R. Asadpour, J.-C. Blancon, A. J. Neukirch, G. Gupta,
J. J. Crochet, M. Chhowalla, S. Tretiak, M. A. Alam, H.-L. Wang,
A. D. Mohite, Science 2015, 347, 522.
[38] J. Seo, J. H. Noh, S. I. Seok, Acc. Chem. Res. 2016, 49, 562.
[39] T. A. Berhe, W.-N. Su, C.-H. Chen, C.-J. Pan, J.-H. Cheng,
H.-M. Chen, M.-C. Tsai, L.-Y. Chen, A. A. Dubale, B.-J. Hwang,
Energy Environ. Sci. 2016, 9, 323.
[40] J. Burschka, A. Dualeh, F. Kessler, E. Baranoff, N.-L. Cevey-Ha,
C. Yi, M. K. Nazeeruddin, M. Graetzel, J. Am. Chem. Soc. 2011, 133,
18042.
[41] W. H. Nguyen, C. D. Bailie, E. L. Unger, M. D. McGehee, J. Am.
Chem. Soc. 2014, 136, 10996.
[42] K. Walzer, B. Maennig, M. Pfeiffer, K. Leo, Chem. Rev. 2007, 107,
1233.
[43] W. Li, H. Dong, L. Wang, N. Li, X. Guo, J. Li, Y. Qiu, J. Mater. Chem.
A 2014, 2, 13587.
[44] a) A. Abate, T. Leijtens, S. Pathak, J. Teuscher, R. Avolio,
M. E. Errico, J. Kirkpatrik, J. M. Ball, P. Docampo, I. McPherson,
H. J. Snaith, Phys. Chem. Chem. Phys. 2013, 15, 2572;
b) R. I. Masel, Chemical Kinetics and Catalysis, Wiley-Interscience,
New York 2001.
[45] T. Leijtens, J. Lim, J. Teuscher, T. Park, H. J. Snaith, Adv. Mater. 2013,
25, 3227.
[46] T. Salim, S. Sun, Y. Abe, A. Krishna, A. C. Grimsdale, Y. M. Lam,
J. Mater. Chem. A 2015, 3, 8943.
[47] J. Liu, Y. Wu, C. Qin, X. Yang, T. Yasuda, A. Islam, K. Zhang,
W. Peng, W. Chen, L. Han, Energy Environ. Sci. 2014, 7, 2963.
[48] a) J. Krüger, R. Plass, L. Cevey, M. Piccirelli, M. Grätzel, U. Bach,
Appl. Phys. Lett. 2001, 79, 2085; b) W. Howie, J. Harris, J. Jennings,
L. Peter, Solar Energy Mater. Solar Cells 2007, 91, 424.
[49] G. Divitini, S. Cacovich, F. Matteocci, L. Cinà, A. Di Carlo, C. Ducati,
Nat. Energy 2016, 1, 15012.
[50] E. J. Juarez-Perez, M. Wußler, F. Fabregat-Santiago, K. Lakus-Wollny,
E. Mankel, T. Mayer, W. Jaegermann, I. Mora-Sero, J. Phys. Chem.
Lett. 2014, 5, 680.
[51] a) L. Etgar, P. Gao, Z. Xue, Q. Peng, A. K. Chandiran, B. Liu,
M. K. Nazeeruddin, M. Grätzel, J. Am. Chem. Soc. 2012, 134, 17396;
b) W. A. Laban, L. Etgar, Energy Environ. Sci. 2013, 6, 3249; c) J. Shi,
J. Dong, S. Lv, Y. Xu, L. Zhu, J. Xiao, X. Xu, H. Wu, D. Li, Y. Luo,
Q. Meng, Appl. Phys. Lett. 2014, 104, 063901.
[52] G. E. Eperon, V. M. Burlakov, P. Docampo, A. Goriely, H. J. Snaith,
Adv. Funct. Mater. 2014, 24, 151.
[53] a) R. Roesch, K. R. Eberhardt, S. Engmann, G. Gobsch, H. Hoppe,
Solar Energy Mater. Solar Cells 2013, 117, 59; b) H. M. Grandin,
S. M. Tadayyon, W. N. Lennard, K. Griffiths, L. L. Coatsworth,
P. R. Norton, Z. D. Popovic, H. Aziz, N. X. Hu, Org. Electron.:
Phys. Mater. Appl. 2003, 4, 9; c) Z. Y. Yuan, W. Zhou, Z. L. Zhang,
J. Q. Liu, J. Z. Wang, H. X. Li, L. M. Peng, Surf. Interface Anal. 2001,
32, 102; d) R. Rösch, D. M. Tanenbaum, M. Jørgensen, M. Seeland,
M. Bärenklau, M. Hermenau, E. Voroshazi, M. T. Lloyd, Y. Galagan,
B. Zimmermann, U. Würfel, M. Hösel, H. F. Dam, S. A. Gevorgyan,
S. Kudret, W. Maes, L. Lutsen, D. Vanderzande, R. Andriessen,
G. Teran-Escobar, M. Lira-Cantu, A. Rivaton, G. Y. Uzunoğlu,
wileyonlinelibrary.com (11 of 12)  1600920
 www.advenergymat.de
FULL PAPER
D. Germack, B. Andreasen, M. V. Madsen, K. Norrman, H. Hoppe,
F. C. Krebs, Energy Environ. Sci. 2012, 5, 6521; e) N. Grossiord,
J. M. Kroon, R. Andriessen, P. W. M. Blom, Org. Electron.: Phys.
Mater. Appl. 2012, 13, 432.
[54] a) A. Dualeh, T. Moehl, N. Tétreault, J. Teuscher, P. Gao,
M. K. Nazeeruddin, M. Grätzel, ACS Nano 2014, 8, 362;
b) I. K. Ding, N. Tétreault, J. Brillet, B. E. Hardin, E. H. Smith,
S. J. Rosenthal, F. Sauvage, M. Grätzel, M. D. McGehee, Adv. Funct.
Mater. 2009, 19, 2431.
[55] S. Guarnera, A. Abate, W. Zhang, J. M. Foster, G. Richardson,
A. Petrozza, H. J. Snaith, J. Phys. Chem. Lett. 2015, 6, 432.
[56] F. Bonaccorso, L. Colombo, G. Yu, M. Stoller, V. Tozzini,
A. C. Ferrari, R. S. Ruoff, V. Pellegrini, Science 2015, 347, 1246501.
[57] a) W. Bao, X. Cai, D. Kim, K. Sridhara, M. S. Fuhrer, Appl. Phys. Lett.
2013, 102, 042104; b) S. Kim, A. Konar, W.-S. Hwang, J. H. Lee,
J. Lee, J. Yang, C. Jung, H. Kim, J.-B. Yoo, J.-Y. Choi, Y. W. Jin,
S. Y. Lee, D. Jena, W. Choi, K. Kim, Nat. Commun. 2012, 3, 1011.
[58] B. Radisavljevic, A. Radenovic, J. Brivio, V. Giacometti, A. Kis, Nat.
Nano 2011, 6, 147.
[59] a) N. Balis, E. Stratakis, E. Kymakis, Mater. Today, 2016, 10.1016/j.
mattod.2016.03.018; b) J.-M. Yun, Y.-J. Noh, J.-S. Yeo, Y.-J. Go,
S.-I. Na, H.-G. Jeong, J. Kim, S. Lee, S.-S. Kim, H. Y. Koo, T.-W. Kim,
D.-Y. Kim, J. Mater. Chem. C 2013, 1, 3777.
[60] X. Gu, W. Cui, H. Li, Z. Wu, Z. Zeng, S.-T. Lee, H. Zhang, B. Sun,
Adv. Energy Mater. 2013, 3, 1262.
[61] X. Fan, C. Cui, G. Fang, J. Wang, S. Li, F. Cheng, H. Long, Y. Li, Adv.
Funct. Mater. 2012, 22, 585.
[62] Z. He, B. Xiao, F. Liu, H. Wu, Y. Yang, S. Xiao, C. Wang, T. P. Russell,
Y. Cao, Nat. Photonics 2015, 9, 174.
[63] a) A. O’Neill, U. Khan, J. N. Coleman, Chem. Mater. 2012, 24,
2414; b) A. Capasso, A. E. Del Rio Castillo, H. Sun, A. Ansaldo,
V. Pellegrini, F. Bonaccorso, Solid State Commun. 2015, 224, 53;
c) Y. Hernandez, V. Nicolosi, M. Lotya, F. M. Blighe, Z. Sun, S. De,
I. T. McGovern, B. Holland, M. Byrne, Y. K. Gun’Ko, J. J. Boland,
P. Niraj, G. Duesberg, S. Krishnamurthy, R. Goodhue, J. Hutchison,
V. Scardaci, A. C. Ferrari, J. N. Coleman, Nat. Nano 2008, 3, 563.
[64] F. Bonaccorso, A. Bartolotta, J. N. Coleman, C. Backes, Adv. Mater.
2016, DOI: 10.1002/adma.201506410.
[65] J. N. Coleman, M. Lotya, A. O’Neill, S. D. Bergin, P. J. King, U. Khan,
K. Young, A. Gaucher, S. De, R. J. Smith, I. V. Shvets, S. K. Arora,
G. Stanton, H.-Y. Kim, K. Lee, G. T. Kim, G. S. Duesberg, T. Hallam,
J. J. Boland, J. J. Wang, J. F. Donegan, J. C. Grunlan, G. Moriarty,
A. Shmeliov, R. J. Nicholls, J. M. Perkins, E. M. Grieveson,
K. Theuwissen, D. W. McComb, P. D. Nellist, V. Nicolosi, Science
2011, 331, 568.
[66] U. Halim, C. R. Zheng, Y. Chen, Z. Lin, S. Jiang, R. Cheng, Y. Huang,
X. Duan, Nat. Commun. 2013, 4, 2213.
1600920  (12 of 12) wileyonlinelibrary.com © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.MaterialsViews.com
[67] A. Jawaid, D. Nepal, K. Park, M. Jespersen, A. Qualley, P. Mirau,
L. F. Drummy, R. A. Vaia, Chem. Mater. 2016, 28, 337.
[68] A. Capasso, A. E. D. R. Castillo, L. Najafi, V. Pellegrini,
F. Bonaccorso, F. Matteocci, L. Cinà, A. D. Carlo, presented at 2015
IEEE 15th Int. Conf. on Nanotechnology (IEEE-NANO), 27–30 July 2015.
[69] J. P. Wilcoxon, G. A. Samara, Phys. Rev. B 1995, 51, 7299.
[70] C. Lee, H. Yan, L. E. Brus, T. F. Heinz, J. Hone, S. Ryu, ACS Nano
2010, 4, 2695.
[71] S. C. Moss, J. J. Huang, B. Schoenfeld, Izvestiya Akademii Nauk
SSSR, Neorg. Mater. 1983, 19, 957.
[72] H. Sun, A. E. Del Rio Castillo, S. Monaco, A. Capasso, A. Ansaldo,
M. Prato, D. A. Dinh, V. Pellegrini, B. Scrosati, L. Manna,
F. Bonaccorso, J. Mater. Chem. A 2016, 4, 6886.
[73] P. Schulz, E. Edri, S. Kirmayer, G. Hodes, D. Cahen, A. Kahn, Energy
Environ. Sci. 2014, 7, 1377.
[74] F. Matteocci, Y. Busby, J.-J. Pireaux, G. Divitini, S. Cacovich,
C. Ducati, A. Di Carlo, ACS Appl. Mater. Interfaces 2015, 7, 26176.
[75] A. C. Ferrari, F. Bonaccorso, V. Fal’ko, K. S. Novoselov, S. Roche,
P. Boggild, S. Borini, F. H. L. Koppens, V. Palermo, N. Pugno,
J. A. Garrido, R. Sordan, A. Bianco, L. Ballerini, M. Prato, E. Lidorikis,
J. Kivioja, C. Marinelli, T. Ryhanen, A. Morpurgo, J. N. Coleman,
V. Nicolosi, L. Colombo, A. Fert, M. Garcia-Hernandez,
A. Bachtold, G. F. Schneider, F. Guinea, C. Dekker, M. Barbone,
Z. Sun, C. Galiotis, A. N. Grigorenko, G. Konstantatos, A. Kis,
M. Katsnelson, L. Vandersypen, A. Loiseau, V. Morandi,
D. Neumaier, E. Treossi, V. Pellegrini, M. Polini, A. Tredicucci,
G. M. Williams, B. H. Hong, J.-H. Ahn, J. Min Kim, H. Zirath,
B. J. van Wees, H. van der Zant, L. Occhipinti, A. Di Matteo,
I. A. Kinloch, T. Seyller, E. Quesnel, X. Feng, K. Teo, N. Rupesinghe,
P. Hakonen, S. R. T. Neil, Q. Tannock, T. Lofwander, J. Kinaret,
Nanoscale 2015, 7, 4598.
[76] J. Halme, P. Vahermaa, K. Miettunen, P. Lund, Adv. Mater. 2010, 22, E210.
[77] F. Matteocci, S. Razza, F. Di Giacomo, S. Casaluci, G. Mincuzzi,
T. M. Brown, A. D’Epifanio, S. Licoccia, A. Di Carlo, Phys. Chem.
Chem. Phys. 2014, 16, 3918.
[78] NIST X-ray Photoelectron Spectroscopy Database, Version 4.1,
National Institute of Standards and Technology, Gaithersburg 2012,
http://srdata.nist.gov/xps/, accessed: October, 2015.
[79] J. F. Watts, J. Wolstenholme, An Introduction to Surface Analysis
by XPS and AES, John Wiley & Sons Ltd., Chichester, UK
2003.
[80] D. Ganta, S. Sinha, R. T. Haasch, Surf. Sci. Spectra 2014, 21, 19.
[81] P. A. Spevack, N. S. McIntyre, J. Phys. Chem. 1992, 96, 9029.
[82] M. G. Helander, M. T. Greiner, Z. B. Wang, Z. H. Lu, Appl. Surf. Sci.
2010, 256, 2602.
[83] A. Calloni, A. Abate, G. Bussetti, G. Berti, R. Yivlialin, F. Ciccacci,
L. Duò, J. Phys. Chem. C 2015, 119, 21329.
Adv. Energy Mater. 2016, 1600920