“high-performance ch nh pbi perovskite solar cells ......on account of the facts that air flow can...
TRANSCRIPT
Supporting information for
“High-performance CH3NH3PbI3 perovskite solar cells fabricated
under ambient conditions with high relative humidity”
Binglong Lei, Vincent Obiozo Eze, and Tatsuo Mori*
Department of Electrical and Electronics Engineering, Aichi Institute of Technology,
Yakusa-cho, Toyota, Aichi, 470-0392, Japan.
E-mail: [email protected]
1 Film fluctuation:
Fig. S1 One-dimensional linear scanning of the GFA and w/o-GFA perovskite
surfaces with a TiO2 background dense layer of 50 nm in thickness (a) and the fluctuation
distributions thereof (b)
By virtue of the best GFA perovskite, we approximately calculated absorption
coefficient for the CH3NH3PbI3 perovskite by the formula of T=I/I0=Exp(-αd), where T, I, I0,
α and d represent transmittance, transmitted light intensity, incident light intensity,
absorption coefficient and layer thickness (300 nm), respectively. It is indicated in Fig. S2
that CH3NH3PbI3 exhibits high absorption coefficient of 5103-5104 cm-1 in the range of
500-800 nm and a specific value of 1.48104 cm-1 at 550 nm, as is much consistent with
previously reported data collected by the integrating sphere technique.1) This has laterally
proved that the amenable quality of the GFA perovskite.
Fig. S2 Absorption spectra of two kinds of perovskite layers constructing GFA and
w/o-GFA solar cells and an approximate calculation curve of absorption coefficient based
on the GFA perovskite of 300 nm in thickness.
2 Perovskite thickness of GFA and w/o-GFA films
Table S1 Correlation of perovskite thickness via spin velocity
Spin velocity (rpm/s) 6000 4000 3000 2000
GFA films (nm) 272±18 a 326±23 387±35 461±69
w/o-GFA films (nm) 285±58 335±67 406±73 494±125
(a: the minus-plus values represent film peak and valley fluctuation in a typical sample
instead of thickness variation from multiple samples. They can roughly reflect the film
quality as the RMS values)
Fig. S3 Correlation and data fitting of perovskite thickness via spin velocity
(Approximately, R2=0.9940, Chi2/DOF=45.69 for thickness fitting to GFA and w/o-GFA
films)
On account of the facts that air flow can significantly enhance film thinning2) and that
for perovskite solar cells, sufficient light-absorbing film thickness must be guaranteed (i.e.
for an absorption coefficient of 1.48104 cm-1 at 550 nm, 300-400 nm as half
incident-reflection absorption thickness is best required), we did not impose the GFA
immediately before or after the spinning was commenced but 2 s later after it. This
technique enabled us to produce the film thickness in the range of 250-500 nm, from which
we chose an ideal value of 4000 rpm for final perovskite fabrication when the surface
roughness was also taken into account.
According to the following empirical Formula (S1),3) we fitted the data sets of both
GFA and w/o-GFA perovskite film thickness versus spin velocity. It is demonstrated in
Table S1 and Fig. S3 that a thickness reduction of only 10-25 nm was successfully
guaranteed. This also indicates that the film thickness can also be expressed by the Formula
(1) as shown in the main manuscript, where m is approximately equal to 10-25 nm denoting
the function of thickness reduction produced by GFA. Therefore it is safe to conclude that
the gas flowing does not significantly enhance the film thinning by radial convection or
mass transfer but expedited solvent evaporation to generate more homogeneous films.
kh (S1)
where h is the film thickness, ω is the angular velocity, k denotes an empirical constant
relating to solution concentration, while we applied an α as -0.5.
3.1 Materials and Reagents
Methylammonium iodide (CH3NH3I) was synthesized according to the reported
procedures.4) Fluorine-doped tin oxide (FTO) glass substrates (sheet resistance 12 Ω /sq)
were purchased from Asahi Glass Co. Ltd. The Semicon detergent, acetone, isopropanol,
titanium diisopropoxide bis(acetylacetonate), absolute ethanol, acetonitrile, methyl
ammonium, hydroiodic acid, Spiro-OMeTAD (2,2’, 7,7’-tetrakis (N, N-di-p-methoxy-
phenyalmine) -9,9’ -spirobifuorene), dimethyl formamide (DMF) and lead iodide (99.99%,
Aldrich) were used without further treatment.
3.2 Semi-product CH3NH3I Synthesis
CH3NH3I was synthesized by reacting 27.8 ml methylamine (about 0.273 mol, 40% in
methanol) and 30.0 ml HI (about 51.03 g, 0.228 mol, 57 wt% in water) in a 500 ml round
bottomed flask at 0 ℃ for 2 h while stirring. The latter solution was added into the former
one within 5 min (the addition was required to be finished in 10 min). Then the solution was
vaporized to receive the precipitate at 60 ℃ for about 1 hour and subsequently
recrystallized for two times in ethanol and washed by diethyl ether for another two times.
The semi product was dried at 60 ℃ in a vacuum oven for 2-4 hours. The synthesized
CH3NH3I weighed up to 34.3 g, with a recovery rate of about 94.62%, and it was kept in
2-propanol as stock solution (10 mg/ml) or stored in vacuum.
3.3 Perovskite solar cell fabrication
GFA and w/o-GFA cells are hereafter referred to the cells fabricated with and without
Gas Flowing Assisting, respectively.
Solar cells with a geometry of FTO glass /TiO2 dense-layer/ CH3NH3PbI3/ HTM/Au
have been applied by us. Fig. S4 demonstrates the schematic fabrication procedures for the
GFA and w/o-GFA cells. Complete conductions in detail will be listed as following seven
sections and all steps were conducted in air (room temperature, 23-25 ℃; RH, 42-48%)
without the assistance of glove box and the step for perovskite layer should be quickly
finished in 30 min.
Fig. S4 Schematic fabrication procedures for perovskite solar cells. DPS denotes
deposition; HTM, Spiro-OMeTAD.
(1) FTO pre-treating and cleaning
FTO substrates with a size of 23 cm2 were obtained by cutting from the original
3035 cm2 and then etched and treated by ultrasonication in Semicon cleaning solution,
deionized water, ethanol and acetone for one or two times. After drying in air, they were
treated in the O3/ultraviolet condition for 1.5 min and 15 min respectively.
(2) TiO2 dense layer
TiO2 dense blocking layers were completed by commonly used spin coating after the
anode part of about 0.7 cm was masked by a 3M scotch tape. A solution of 0.15 ml HCl
(36.5%)/2.25 ml Titanium diisopropoxide bis(acetylacetonate)/15 ml ethanol was prepared
and then spin-coated on the substrate samples at 2000 rpm/10s for the dense-layer with a
thickness of about 50 nm. All samples were moved onto a hotplate and kept at 125 ℃ for
20 min and then sintered at 500℃ for 30 min in a closed furnace.
(3) Treatment with TiCl4
Masking the anode part with tape was conducted beforehand, and then the substrates
obtained above with TiO2 dense layer were immersedly treated in a 0.04 M TiCl4 solution (3
ml/72 ml in water) at 70 ℃ for 20 min. After that the substrates were sufficiently rinsed
with purified water and then heated at 70 ℃ until they were dried and then they were again
heated at 500℃ for 30 min.
(4) PbI2 precursor layer by spin-coating
In the two consequential steps for light absorbing layer, the precursor PbI2 has a
decisive impact upon final perovskite layers and therefore delicate control must be
guaranteed. PbI2 (99.99%) was first dissolved in DMF at a concentration of 500 mg/ml at 70
℃ while stirring. While kept at this temperature, the PbI2 solution was spin-coated on the
substrates with speeds of 4000 rpm/20s for thickness of about 250 nm and no loading time
was applied. A vertical gas flowing assisting (GFA, with air employed) was additionally
applied at 2 s just after the spinning had been commenced. After the spin coating, the
samples were set on the hotplate for crystallization at 100 ℃ for 15 min. These samples
were labeled as GFA cells (semi), while for w-GAF (semi) cells as a control group, no gas
flowing had been implied but with the other conditions kept identical.
As for the flowing gas, we employed the high RH room air (in the clean room), which
possessed the same RH (42-48%) with the ambient conditions but much less dusts in its
content if compared with the outdoor unclear air. The outlet nozzle of the air flower was
about 2.52.5 cm2 and conveyed air with a rate of approximately 3.6 L/s (about 5.76 m/s).
The distance from the substrate to the outlet was controlled at 8-10 cm. According to the our
final results, we speculate that water or moisture from the high RH air had positively
functioned for mass transfer and crystallization of perovskite, since the existence of solvent
during crystallization is somewhat beneficial for perovskite quality.5)
It is worth a mention that the GFA PbI2 films, including our final GFA perovskite
layers, were all conducted at an optimized air flowing rate of approximately 3.6 L/s (about
5.76 m/s). When the rate we applied was lower than about 2.0 L/s, it was found that the
flowing impact upon PbI2 films was not very pronounced due to the high RH values (final
cells possessed adjacent PCEs to w/o-GFA cells but with larger variations), while when the
rate was elevated up to 3.2 L/s, the positive effect thereafter can not significantly increase
(Fig. S5). Therefore a rate of 3.6 L/s as preference were employed by us and also directly
used for the following transformation step from PbI2 to final perovskite transformation to
expedite evaporation of isopropanol as solvent, which is mainly due to the consideration that
the quality of PbI2 films dominates the final perovskite layers in the two-step sequential
procedures.
Fig. S5 Performance of solar cells fabricated by the GFA route with varying air flowing
rates (reverse scanned PCE are statistically shown from 10 solar cells, while 15 or 20 cells
are calculated for 0 L/s (w/o-GFA) and the best GFA cells at 3.6 L/s)
(5) Transformation of PbI2 into CH3NH3PbI3
For the semi GAF samples fabricated above, they were endowed with a structure of
FTO glass /TiO2 dense-layer/PbI2, and the current step will transform the PbI2 layer into
perovskite by the following way: spin coating of a thermal CH3NH3I solution (70 ℃) with
GFA assisting as it is mentioned above. The CH3NH3I solution of 200 µl, for every single
sample, was loaded on the substrates with a duration of 20 s before the spin coating was
conducted at 4000 rpm/30 s (air flowing started 2 s later after spinning) and then desiccation
and crystallization were done at 100 ℃ for 15 min. For the w-GAF samples, identical
conductions had been finished but without gas flowing assisting.
It is worth a mention that attempts had been made to extend the soaking duration to 120
s, however the remnant PbI2 still can not be completely eliminated in all perovskite samples.
Further observation reveals that some PbI2 substrates with a slight hazy surface fabricated
by a spin-coating rate of less than 3000 rpm (without GFA assisting) can be more quickly
converted into perovskite than the GFA-based PbI2 layers. This auxiliarily demonstrates that
GFA can impair preferential crystallization and enhance the compactness of PbI2 layer,
which to some extent has also hindered the transformation in terms of dynamics.
(6) Formation of HTM layer
The samples achieved above were deposited by the Spiro-OMeTAD solution (the
composition was list in Table S2) by spin coating the solution at 2000 rpm for 60s for
thickness of about 200 nm. After that, they were set in air for 2-4 hours for HTM oxidation
before the back gold deposition was conducted in the next step.
Table S2 Composition of Spiro-OMeTAD solution for CH3NH3PbI3 solar cells
Ingredients Contents
Spiro-MeOTAD (98%) 144.6 mg
4-tert-butylpridine (TBP) 57.6 ul
520 mg/ml LiTFSI in acetonitrile 35 ul
Chlorobenzene 2 ml
(7) Gold back contact deposition in vacuum
The final solar cell devices were completed by thermal evaporation of gold with a
thickness of 100 nm at 3.4 10-4 Pa. A metal mask has been used to define the active area as
0.15 cm2.
3 Testing and characterization
The morphology of perovskite layer was observed by a field emission scanning
electron microscope (FE-SEM, JSM-6335FM, JEOL, accelerating voltage of 10 KV). The
voltage-current J–V measurement was recorded by applying external potential biases to the
cells and recording the output photocurrent with a digital source meter (Model B2901A,
Agilent Corp.). A 150 W xenon lamp (Model Otento-SUN3, Xe-S150, Bunkoukeiki) was
applied as light source and the output irradiation intensity was adjusted to the AM 1.5G
condition (100 mW/cm2). The voltage step length and delay time were set at 2-40 mV and
0.5-1000 ms, respectively. Incident photon conversion efficiency (IPCE) measurement was
conducted in the wavelength range of 350-830 nm with a 300 W xenon light adjusted by a
monochromator (PVL 3300, Asahi Spectra). X-Ray diffraction (XRD) patterns were
recorded by in the 2-theta range of 5°-65° on an X-ray diffractometer (RINT2500V/PC,
Rigaku) by Cu Kα radiation (40 kV, 100 mA). The absorption spectra of our solar devices
were measured with (Model UV-2450, Shimadzu) in the wavelength of 400 nm to 900 nm.
Surface morphologies and roughness demonstration were scanned using an atomic force
microscopy (AFM, Keyence VN-HIV8). For thickness testing and surface one-dimensional
scanning and comparison, we applied a thickness tester (Surfcorder ET200, Kosaka
Labortory Ltd.) with a step length of 0.03 um.
For the J–V scanning, we had tried series of step delays from 0.1 ms to 1000 ms and
found that it was most favorable at 50 ms for forward-reverse balance to drastically suppress
the hysteresis effect. A step delay of 50 ms was therefore applied for testing all our cells.
Reference
1) H. S. Jung and N. G. Park, Small 11, 10 (2015).
2) S. Middleman, J. Appl. Phys. 62, 2530 (1987).
3) K. Norrman, A. Ghanbari-Siahkali, and N. Larsen, Ann. Rep. Sect. C, 101, 174 (2005).
4) M. Liu, M. B. Johnston, and H. J. Snaith, Nature 501, 395 (2013).
5) W. Nie, H. Tsai, R. Asadpour, J.-C. Blancon, A. J. Neukirch, G. Gupta, J. J. Crochet, M. Chhowalla, S. Tretiak, and M. A. Alam, Science 347, 522 (2015).