DOI: 10.1002/ente.
201500045
Outdoor Performance and Stability under Elevated Temperatures and
Long-Term Light Soaking of Triple-Layer Mesoporous Perovskite
Photovoltaics
Xiong Li,[a] Manuel Tschumi,[a] Hongwei Han,[e] Saeed Salem Babkair,[c] Raysah Ali Alzubaydi,[c] Azhar
Ahmad Ansari,[c] Sami S. Habib,[d] Mohammad Khaja Nazeeruddin,[a] Shaik M. Zakeeruddin,[a] and
Michael Grtzel*[a, b]
Lack of proven stability has become a major obstacle on the deployed on a large scale along with conventional silicon-
path of metal halide perovskite solar cells (PSCs), in particu- based solar cells.
lar methylammonium lead triiodide (MAPbI3), towards com- However, the best performing devices all use gold or silver
mercial viability. This correlates with the intrinsic affinity of as the back contact in conjunction with hole-transporting ma-
MAPbI3 towards moisture and ambient air in particular, terials (HTMs) acting as electron-blocking layers (denoted as
leading to its degradation in ambient conditions. We per- standard PSCs in the following). The costs of HTMs such as
formed extensive stability tests to prove the durability of 2,2’,7,7’-tetrakis(N,N-di-p-methoxyphenylamine)-9,9’-spirobi-
hole-conductor-free PSCs based on a triple-layer architecture fluorene(spiro-OMeTAD) are highly limiting for wide-
employing carbon as a back contact, including outdoor tests spread applications. Besides, the vacuum deposition process
in the hot desert climate and indoor long-term light soaking for the noble-metal-based counter electrodes (CEs) con-
as well as heat exposure during 3 months at 80–85 8C. These sumes large amounts of energy. Obviously, it is worth devel-
results show no evidence for device degradation under the oping HTM-free mesoscopic perovskite solar cells and re-
test conditions, confirming that the triple-layer device archi- placing the noble-metal-based CEs with inexpensive and
tecture provides a promising path towards realizing efficient abundantly available materials. Previously, we have reported
and stable perovskite photovoltaics. a fully printable PSC employing a mesoscopic CH3NH3PbI3/
TiO2 heterojunction and carbon as counter electrode.[3] A
schematic presentation of this embodiment is shown in
Organic–inorganic hybrid solar cells using metal-halide per- Figure 1 (denoted in the following as triple-layer PSCs). The
ovskites such as CH3NH3PbI3 as light harvesters exhibit sev-
eral appealing features such as high optical cross section, ex-
cellent ambipolar charge transport, small exciton binding
energy, tunable band gaps, and low-cost fabrication.[1] Impor-
tantly, the CH3NH3PbI3 perovskite solar cells (PSCs) are so-
lution-processible, which is beneficial to inexpensive large-
scale commercialization. Over the past three years several
groups have reported high power conversion efficiencies
(PCEs) of over 15 %[2] raising expectations for PSCs to be
[a] X. Li, M. Tschumi, M. K. Nazeeruddin, S. M. Zakeeruddin,
Prof. Dr. M. Grtzel
Laboratory of Photonics and Interfaces (LPI), Faculty of Basic Science, Ecole
Polytechnique F¦d¦rale de Lausanne (Switzerland)
E-mail: michael.graetzel@epfl.ch
[b] Prof. Dr. M. Grtzel Figure 1. Schematic cross section of the triple-layer perovskite-based fully
Center of Nanotechnology, King Abdulaziz University, Jeddah, 21589 (Saudi printable mesoscopic solar cell.
Arabia)
[c] S. S. Babkair, R. A. Alzubaydi, A. A. Ansari
Center of Nanotechnology, Department of Physics, Faculty of Science, King
Abdulaziz, University, Jeddah, 21589 (Saudi Arabia) TiO2 film and the CE are separated by a mesoporous ZrO2
[d] S. S. Habib layer, preventing the back flow of photogenerated electrons
Center of Nanotechnology, Department of Aeronautical Engineering, King Ab- from the photoanode to the CE.[4] Screen-printing allows the
dulaziz University, Jeddah, 21589 (Saudi Arabia) precise control of the thickness of the porous ZrO2 spacer
[e] Prof. H. Han layer, which in turn affects the series resistance of the device.
Michael Grtzel Center for Mesoscopic Solar Cells, Wuhan National Laborato-
By employing a drop-casting method to infiltrate the perov-
ry for Optoelectronics, Huazhong University of Science and Technology,
Wuhan 430074, Hubei (PR China) skite CH3NH3PbI3 into the porous triple layers, we achieved
Supporting information for this article is available on the WWW under a certified power conversion efficiency of 12.8 % and stable
http://dx.doi.org/10.1002/ente.201500045. performance over > 1000 h in ambient air under full sunlight,
Energy Technol. 2015, 3, 551 – 555 Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 551
which shows promise for the development of low-cost and
stable photovoltaics.[3b]
It is well known that CH3NH3PbI3 degrades under humid
conditions and forms PbI2 at higher temperatures due to the
loss of CH3NH3I.[5] These instabilities could hamper outdoor
applications. Surprisingly few stability studies have been per-
formed so far with PSCs and more extensive tests to probe
their stability are urgently warranted to ascertain that PSCs
can meet the stringent international norms for outdoor pho-
tovoltaic applications. Here, we present the first outdoor
measurements of PSCs in a hot desert climate and compare
these to indoor tests at elevated temperatures or under long-
term light-soaking conditions.
Figure 3. J–V curves for the best performing device under simulated standard
We infiltrated a chemically modified CH3NH3PbI3 material
AM1.5 solar irradiation measured at room temperature.
into the porous TiO2/ZrO2 scaffold by drop-casting a solution
containing PbI2, CH3NH3I (MAI), and 5-ammoniumvaleric
acid (5-AVA) iodide in g-butyrolactone through the carbon that is, the short-circuit photocurrent (Jsc), open-circuit volt-
layer.[3b] The 5-AVA molecules are proposed to form linear age (Voc), fill factor (FF), and power conversion efficiency
hydrogen-bonded chains between their COOH and NH3 + (PCE) of 22.7 mA cm¢2, 0.85 V, 0.66, and 12.9 %, respective-
groups and I¢ ions from the PbI6 octahedra; this acts as ly.
a templating agent to create mixed-cation perovskite (5- We tested the stability behavior of the PSC by continuous-
AVA)x(MA)1-xPbI3 crystals with lower defect concentrations ly exposing the device outdoors in Jeddah, Saudi Arabia,
and better pore filling as well as more complete contact with during one week from September 7–14, 2014. The J–V curve
the TiO2 scaffold. Figure 2 a shows a cross-sectional scanning of the solar cell was measured around noontime on every
day. Devices were masked with a black tape having a round
aperture area of 0.283 cm2. J–V curves were recorded to de-
termine the time evolution of the photovoltaic metrics. As
displayed is Figure 4, the Voc measured outdoors is nearly
100 mV higher than that of the indoor test shown in
Figure 3. A bigger size of the mask aperture (0.28 cm2) em-
ployed for the outdoor test than that used for the indoor
measurement (0.16 cm2) would increase the photocurrent by
a factor of 1.8 and accordingly account for a Voc increase of
46 mV. In addition, we ascribe the remaining approximately
Figure 2. Cross-sectional SEM images of the complete device: a) high mag-
nification, b) low magnification.
electron microscopy (SEM) image presenting the morpholo-
gy of the complete PSC, and Figure 2 b zooms in on the mes-
oporous TiO2/ZrO2 oxide double layer portion. The latter
shows complete infiltration of the (5-AVA)x(MA)1-xPbI3 per-
ovskite within the mesopores TiO2/ZrO2 film, where the per-
ovskite domains show a highly connected network and con-
tinuous paths within the mesoporous triple-layer architec-
ture. This not only helps to improve the charge transport of
the CH3NH3PbI3 inserted into the porous metal oxides, but
also enhances the light-harvesting ability of the device.[3b]
We measured devices employing (5-AVA)x(MA)1-xPbI3 as
light harvester at standard reporting conditions, that is,
AM1.5 G solar light at 100 mW cm¢2 intensity and room tem-
perature. The data shown in Figure 3 refer to the best per-
forming cell. From the photocurrent–voltage curve we ex- Figure 4. Time evolution of the encapsulated PSC solar cell metrics during
tract values for the key photovoltaic performance metrics; outdoor aging in Jeddah, Saudi Arabia.
Energy Technol. 2015, 3, 551 – 555 Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 552
54 mV enhancement to cell aging as the outdoor measure-
ments were performed several weeks later after the indoor
standard test as well as to the higher intensity of the solar ra-
diation in Jeddah, Saudi Arabia, which exceeds that of the
AM1.5 G standard. Figure 4 shows that these parameters re-
mained remarkably stable over the 7 day period and their
final values were even slightly above the initial ones. The
short-circuit photocurrent showed the largest fluctuations, re-
sponding to variations in the incident light intensity due to
cloudiness or hazy weather conditions. These results bode
well for the outdoor deployment of triple-layer PSCs in the
hot climate prevailing in Saudi Arabia as there is no indica-
tion for device degradation under the hot and humid climate
conditions for a week compared to the initial device per-
formance.
We also recorded the outdoor performance of a PSC over
one day on September 7, 2014 from 8 a.m. to 5 p.m. Photo-
voltaic performance metrics are listed in Table S1 (Support-
ing Information). The PCE of the devices reached a maxi-
mum value of 11.4 % at noontime when the temperature
reached 42 8C. Please note the high Voc value of 0.972 V mea-
sured at 1 p.m. when the outside temperature reached
43.3 8C. The open-circuit voltage of a solar cell is given by
the expression:[6]
V oc ¼ ðRT=FÞlnðJ sc =J 0 Þ ð1Þ
Figure 5. Outdoor cell measurements in Jeddah, Saudi Arabia at various
times on September 7, 2014. a) Output electric power density as a function
in which R is the ideal gas constant, T the absolute tempera-
of applied voltage and b) outdoor irradiance as well as PCE measured at the
ture, F is FaradayÏs constant, and J0 the density of the dark different hours of the day.
current. RT/F has the value 25.852 mV at 300 K. As raising
the temperature increases the dark current, the Voc decreases
with temperature. The temperature derivative of Voc is: strong performance of the PSC under the local climate con-
ditions prevailing in Jeddah, which peak at the hottest time
ðdV oc =dTÞ ¼ ðV oc ¢Eg,0K Þ=T¢ð3 R=FÞ ð2Þ of the day. This appears to run counter to the decrease of the
PCE with increasing temperature that is commonly observed
in which Eg,0K is the band gap of the semiconductor at T = with conventional silicon cells, where the PCE declines by
0 K.[7] Using Eg,0K = 1.6 eV for CH3NH3PbI3 one obtains at approximately 0.5 % for each 1 8C increase in temperature.
43 8C a temperature coefficient (dVoc/dT) of ¢2.2 mV/degree. However, it should be noted that the data in Figure 5 were
By linear extrapolation, the Voc is expected to attain 993 mV obtained under real outdoor conditions where the light inten-
at 34 8C, provided the photocurrent remains the same at the sity and spectral distribution of the photon flux changed
two temperatures. However, in Table S1 (Supporting Infor- during the day, thereby affecting the photovoltaic per-
mation) the Jsc value at 8 a.m. was only 10.5 mA cm¢2 due to formance and particularly the short-circuit photocurrent. At
morning clouds or haze compared to 18.4 mA cm¢2 measured 5 p.m. the air mass number (AM) had certainly increased
at 1 p.m. in full sunlight. According to Equation (1), the well beyond 2.5. Due to the longer path in the atmosphere at
lower photocurrent causes a drop in Voc, which compensates higher AM number, the blue component of the solar emis-
the gain from the decrease in the dark current J0. The non- sion was attenuated by light scattering. This decreased the
monotonic trend of Voc during the day is therefore the result solar photon flux absorbed by the perovskite cell as the peak
of the cumulative effects of temperature on the dark current emission shifted towards the near-infrared region, which re-
and Jsc. sulted in the observed reduction in the photocurrent and
Figure 5 shows plots of the photovoltaic power density conversion efficiencies.
generated by the PSC during a one-day outdoor exposure of To examine the stability of the device during intense and
the PCS in Jeddah as a function of voltage. The power was prolonged heat stress, we kept encapsulated triple-layer
measured point by point under stationary conditions to elim- PSCs in an oven at 80–85 8C for 3 months interrupting only
inate any possible hysteresis effects that are often observed a few times to monitor their PV performance under standard
when the J–V curves of PSCs are recorded by scanning the operating conditions. The conditions of this heat stress test
voltage backward from Voc to short-circuit conditions, which are close to the ISOS-D-2 protocol.[9] The J–V curves were
results in inflated efficiency values.[8] These data reflect the recorded under full solar AM1.5 light exposure at room tem-
Energy Technol. 2015, 3, 551 – 555 Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 553
Figure 6. Indoor heat stress test of a triple-layer PSCs. The device was encap-
sulated and kept for 3 months in a normal oven filled with ambient air at 80– Figure 7. Evolution of relative device performance parameters for triple-layer
85 8C. They were removed at several intervals from the oven and cooled over- PSCs aged under Ar at 45 8C and at maximum power point tracking (MPPT)
night to equilibrate at ambient temperature before recording the PV device conditions.
performance metrics. Measurements employed simulated full solar AM1.5
light at room temperature.
Experimental Section
Materials synthesis: CH3NH3I (MAI) and HOOC(CH2)4NH3I (5-
perature, and the time evolution of the performance matrix AVAI) were synthesized as follows: Hydroiodic acid together
for one PV device derived from the J–V curves is shown in with methylamine or 5-aminopentanoic acid were added with
Figure 6; the statistical information on the photovoltaic pa- a mole ratio of 1:1 into round-bottom flask and stirred in the ice
rameters of a batch of four devices subjected to the heat bath for 2 h. Then, the remaining liquid was evaporated in a rota-
vapor at 50 8C. The precipitate was washed three times with di-
stress test are provides in Table S2 (Supporting Information).
ethyl ether and dried in a vacuum drying oven.
Remarkably, all the PV parameters were stable within a few
Fabrication of perovskite based mesoscopic solar cells: As pre-
percent, showing a strong resilience of the perovskite solar sented in Figure 1 a, the fluorinated tin oxide (FTO) glass was
cells towards thermal degradation. first etched to form two separated electrodes before being
To further ascertain the long-term photostability of the cleaned ultrasonically with ethanol. Then, the patterned sub-
triple-layer PSC, we performed indoor light-soaking tests strates were coated with a compact TiO2 layer by aerosol spray
under continuous illumination with a white light light-emit- pyrolysis, and a 1 mm nanoporous TiO2 layer was deposited by
ting diode (LED) array, emitting visible light at an intensity screen printing of a TiO2 slurry, which was prepared as reported
previously.[2c] After being sintered at 450 8C for 30 min, a 2 mm
of 100 mW cm¢2. We used unsealed devices that were kept
ZrO2 space layer was printed on top of the nanoporous TiO2
under an argon atmosphere at 45 8C during the test. The cells
layer using a ZrO2 slurry, which acts as an insulating layer to pre-
were held at their maximum power point over 1056 h and vent electrons from reaching the back contact. Finally, a carbon
the photovoltaic metrics were recorded every 6 h by comput- black/graphite counter electrode with a thickness of approxi-
er-controlled measurements of the J–V curve. Figure 7 shows mately 10 mm was coated on the top of the ZrO2 layer by print-
the temporal evolution of the device performance parame- ing carbon black/graphite composite slurry and sintering at
ters. At the outset of the long-term light-soaking test, the 400 8C for 30 min. After cooling down to room temperature,
photovoltaic metrics determined using the simulated AM1.5 a 40 wt % perovskite precursor solution was infiltrated by drop
casting using the top of the carbon counter electrode. After
solar irradiation at an intensity of 100 mW cm¢2 were Jsc =
drying at 50 8C for one hour, the mesoscopic solar cells contain-
15.5 mA cm¢2, Voc = 790 mV, FF = 0.652, and PCE = 8.2 %. All
ing perovskite were obtained.
parameters remained remarkably stable during long-term The 40 wt % perovskite precursor solution was prepared as fol-
light-soaking at 45 8C, showing no evidence for any signifi- lows: for the MAPbI3 precursor solution, 0.395 g MAI and
cant performance degradation under these conditions. 1.146 g PbI2 were dissolved in 2 mL g-butyrolactone and then
In conclusion, we have investigated for the first time the stirred at 60 8C overnight. The (5-AVA)x(MA)1-xPbI3 precursor
stability of metal-halide perovskite photovoltaics under out- solution was prepared in the same manner except that a mixture
door conditions in a hot and humid climate and subjected of 5-AVAI and MAI with the mole ratio between 1:20 and 1:30
was used.
the cells indoors to prolonged light soaking and heat stress.
Characterization: To simulate the real conditions of the perov-
A hole-conductor-free cell architecture employing carbon as
skite in a working device, the 40 wt % perovskite precursor solu-
a back contact material achieved excellent stability even tion was infiltrated by drop casting into a ZrO2/TiO2 double
under prolonged exposure to temperatures above 80 8C. layer supported on the FTO glass substrate and then dried at
These promising results bode well for future practical appli- 50 8C. Cross-sectional SEM pictures of the devices and images
cations of perovskite solar cells from the top surface of the mesoporous ZrO2/TiO2 films infiltrat-
Energy Technol. 2015, 3, 551 – 555 Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 554
ed with perovskites were obtained by using a field-emission scan- Acknowledgements
ning electron microscope (FE-SEM). Photocurrent density–volt-
age (J–V) curves were characterized using a Keithley 2400
Financial and technical support of this work by King Abdula-
source/meter and a Newport solar simulator (model 91192) to
ziz University (KAU) Jeddah, Saudi Arabia under grant
produce light with AM1.5 G spectral distribution. A black mask
with a circular aperture (0.16 cm2) smaller than the active area of number 1-20-1432/HiCi is gratefully acknowledged. M.G.
the square solar cell (0.5 cm2) was applied on top of the cell. The thanks the European Research Council for an Advanced Re-
incident photon conversion efficiency (IPCE) was measured search Grant (ARG 247404) funded under his “Mesolight”
using a 150 W xenon lamp (Oriel) fitted with a monochromator project. X.L. and MKN thank CTI Switzerland for financial
(Cornerstone 74004) as a monochromatic light source. These ex- support.
periments employed TiO2 and ZrO2 mesoporous films filled with
the respective perovskite. The UV-vis spectra were measured
with the perovskite-infiltrated mesoscopic TiO2 films supported
by FTO glass using a PerkinElmer Lambda 950 spectrophotome- Keywords: outdoor performance · perovskites ·
ter. photovoltaics · solar cells · stability
Ourdoor Stability measurements in Jeddah, Saudi Arabia: The
carbon film that served as back contact was covered with a thin
glass sheet. The latter was separated from the TCO front glass
by a Surlyn spacer. The complete device was sealed with epoxy
raisin to avoid penetration of moisture into its interior. The cells
were not equipped with a UV filter. Outdoor measurements [1] a) P.-W. Liang, C.-Y. Liao, C.-C. Chueh, F. Zuo, S. T. Williams, X.-K.
were taken over a period of two weeks every day at around Xin, J. Lin, A. K. Y. Jen, Adv. Mater. 2014, 26, 3748 – 3754; b) K.-C.
1 p.m., at which time the air mass number at Jeddah was 1.07 Wang, J.-Y. Jeng, P.-S. Shen, Y.-C. Chang, E. W.-G. Diau, C.-H. Tsai,
under clear sky conditions. The conditions of these outdoor sta- T.-Y. Chao, H.-C. Hsu, P.-Y. Lin, P. Chen, T.-F. Guo, T.-C. Wen, Sci.
bility measurements are close to the ISOS-O-1 protocol.[9] Rep. 2014, 4, 4756; c) Q. Wang, Y. Shao, Q. Dong, Z. Xiao, Y. Yuan, J.
Huang, Energy Environ. Sci. 2014, 7, 2359 – 2365; d) S. Aharon, S.
Indoor stability measurements: PSC devices were maintained in
Gamliel, B. El Cohen, L. Etgar, Phys. Chem. Chem. Phys. 2014, 16,
a box with a glass cover that was flushed with a flow of argon at 10512 – 10518; e) H.-B. Kim, H. Choi, J. Jeong, S. Kim, B. Walker, S.
a rate of 20 mL min¢1. The goal was to simulate an anaerobic en- Song, J. Y. Kim, Nanoscale 2014, 6, 6679 – 6683.
vironment, without water, and to increase the reproducibility of [2] a) J. T.-W. Wang, J. M. Ball, E. M. Barea, A. Abate, J. A. Alexander-
the test conditions. This box was placed inside a larger housing Webber, J. Huang, M. Saliba, I. Mora-Sero, J. Bisquert, H. J. Snaith,
thatcontained a temperature controller and a white LED light R. J. Nicholas, Nano Lett. 2014, 14, 724 – 730; b) J. Burschka, N. Pellet,
source. The spectrum distribution of the LED light is exhibited S.-J. Moon, R. Humphry-Baker, P. Gao, M. K. Nazeeruddin, M. Grat-
in Figure S1 (Supporting Information). J–V curves were recorded zel, Nature 2013, 499, 316 – 319; c) M. Liu, M. B. Johnston, H. J.
by using a computer-controlled electronic circuit system featur- Snaith, Nature 2013, 501, 395 – 398; d) R. F. Service, Science 2014, 344,
458; e) S. Ryu, J. H. Noh, N. J. Jeon, Y. C. Kim, W. S. Yang, J. W. Seo,
ing a delta-sigma analog-to-digital converter. The electronic
S. I. Seok, Energy Environ. Sci. 2014, 7, 2614 – 2618.
system was responsible for the measurements and maintained [3] a) Z. Ku, Y. Rong, M. Xu, T. Liu, H. Han, Sci. Rep. 2013, 3, 3132;
the cells at the maximum point between measurements. The cur- b) A. Mei, X. Li, L. Liu, Z. Ku, T. Liu, Y. Rong, M. Xu, M. Hu, J.
rent noise of the measurement was 300 nA, 50 mV, and the gain Chen, Y. Yang, M. Grtzel, H. Han, Science 2014, 345, 295 – 298.
error was < 0.2 %. The J–V curves were recorded at a scan rate [4] a) H. Wang, G. Liu, X. Li, P. Xiang, Z. Ku, Y. Rong, M. Xu, L. Liu,
of 25 mV s¢1 using steps of 5 mV. M. Hu, Y. Yang, H. Han, Energy Environ. Sci. 2011, 4, 2025 – 2029;
Long-term light-soaking test: The devices were subjected to b) M. Xu, G. Liu, X. Li, H. Wang, Y. Rong, Z. Ku, M. Hu, Y. Yang, L.
a white light LED array with an intensity of 100 mW cm¢2 to sim- Liu, T. Liu, H. Han, Org. Electron. 2013, 14, 628 – 634.
ulate the solar intensity, and the cell temperature was maintained [5] A. Dualeh, P. Gao, S. Il Seok, M. K. Nazeeruddin, M. Grtzel, Chem.
Mater. 2014, 26, 6160 – 6164.
at 45 8C. Data were collected automatically on the cells at inter-
[6] W. Tress, N. Marinova, O. Ingans, M. K. Nazeeruddin, S. M. Zakeer-
vals of 6 h. Each measurement comprises a set of three J–V uddin, M. Graetzel, Adv. Energy Mater. 2015, 5, 1400812.
curves, that is, in the dark, at 10 mW cm¢2, and at 100 mW cm¢2 [7] http://pveducation.org/pvcdrom/solar-cell-operation/effect-of-tempera-
light intensity Between measurements, the cells were maintained ture.
at the maximum power point using a maximum power point [8] M. Grtzel, Nat. Mater. 2014, 13, 838 – 842.
tracking (MPPT) algorithm. The light intensity was [9] M. O. Reese, Sol. Energy Mater. Sol. Cells 2011, 95, 1253 – 1267.
100 mW cm¢2, and a mask of 0.36 cm2 aperture area was placed
on top of the cell. The conditions of this long-term light-soaking Received: February 5, 2015
test are close to the ISOS-L-1 protocol.[9] Published online on May 1, 2015
Energy Technol. 2015, 3, 551 – 555 Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 555
