Air-assisted flow and two-step spin-coating for highly efficient CH3NH3PbI3 perovskite solar cells

Vincent Obiozo Eze; Binglong Lei; Tatsuo Mori

doi:10.7567/JJAP.55.02BF08

1. Introduction

Organic–inorganic hybrid methylammonium lead halide perovskites (e.g., CH₃NH₃PbX₃, X = Cl, Br, I) have recently attracted considerable attention for use in thin-film photovoltaics.¹^–⁶⁾ Their excellent photovoltaic properties, such as broad spectral absorption,⁷⁾ high charge-carrier mobility,⁸⁾ low exciton binding energy (∼50 meV),⁹⁾ and large exciton diffusion length,¹⁰^,¹¹⁾ make them an ideal material for photovoltaic devices. Kojima et al. were the first to employ a lead halide perovskite light absorber in a dye-sensitized solar cell, demonstrating a power conversion efficiency (PCE) of 3.8%.¹²⁾ Following this pioneering work, the PCE of perovskite solar cells (PSCs) increased dramatically to over 15%,¹^,¹³^,¹⁴⁾ which is close to that of single-crystalline silicon solar cells.

Several methods of fabricating uniform and homogeneous perovskite films in both mesoscopic and planar architectures have been reported. Era et al.,¹⁵⁾ Lui et al.,¹⁶⁾ Mitzi et al.,¹⁷⁾ and Salau¹⁸⁾ adopted vapor deposition in vacuum via a dual or single source to deposit the thin layer of mixed halide perovskite employed in planar-structured solar cells and achieved a PCE of over 15%. However, vapor deposition significantly raises the cost of fabrication and consequently poses a potential hurdle to large-scale production.

The perovskite layer can be easily prepared by a solution-processed method. In previous studies, the one-step spin-coating method has been used to form the perovskite layer. This process resulted in nonuniform films that permitted direct contact between the electron-transporting layer (ETL) and the hole-transporting layer (HTL).¹⁹^,²⁰⁾ In contrast, a two-step sequential deposition method that involves the spin-coating of the hot lead(II) iodide (PbI₂) in N,N-dimethylformamide (DMF) onto mesoscopic or planar-structured substrates, followed by dipping into methylammonium iodide–isopropanol (CH₃NH₃I–IPA) solution, has enabled the successful construction of uniform and fully covered CH₃NH₃PbI₃ films.¹^,²¹^–²³⁾ However, depositing perovskite films prepared by two-step dipping on a planar substrate has been reported to be associated with problems such as the incomplete reaction of PbI₂ crystals with organic ammonium and uncontrolled perovskite crystal size as well as rough surface morphology.¹^,²⁴^,²⁵⁾ These reports demonstrated that the resultant perovskite films are more suitable for use in mesoporous-structured PSCs.²²⁾ Therefore, a faster and easier method in which perovskite formation and crystal growth can be controlled, resulting in high-quality films with homogeneous morphology, is exceedingly desirable for the preparation of planar heterojunction (PHJ) devices with superior performance. PSCs based on an air-assisted flow (AAF) PbI₂ blade-coating deposition process and gas-assisted preparation of perovskite films consisting of a monolayer of single-crystalline grains prepared by the conventional one-step spin-coating deposition process were previously reported.¹³^,²⁶⁾

Herein, we prepared CH₃NH₃PbI₃ films via a two-step spin-coating method.¹⁴^,²⁷⁾ The uniformity and coverage of a perovskite light-absorbing layer over the underlying compact-TiO₂ ETL were improved by incorporating AAF during spin-coating. In addition, we meticulously investigated the effect of AAF on (i) PbI₂ crystallization, (ii) perovskite morphology, and (iii) device performance. The results of the device constructed with AAF (w/ AAF) and a control device without AAF (w/o AAF) were compared. The photovoltaic performance characteristics of both devices are reproducible, but with evident disparities in their photovoltaic parameters; this result was found to correlate with their CH₃NH₃PbI₃ morphological difference.

2. Experimental methods

2.1. Materials and reagents

CH₃NH₃I was synthesized according to the reported procedures.¹⁾ Fluorine-doped tin oxide (FTO) glass substrates (sheet resistance: 12 Ω/sq) were purchased from Asahi Glass. Semico-Clean, acetone, IPA, titanium(IV) isopropoxide (99.0%), ethanol, acetonitrile, Spiro-OMeTAD(2,2',7,7'-tetrakis(N,N-di-p-methoxyphenylamine)-9,9'-spirobifluorene), chlorobenzene, DMF (Sigma-Aldrich, 99.8%), and PbI₂ (Sigma-Aldrich, 99.0%) were used without further treatment.

2.2. Device fabrication

FTO glass substrates were sequentially cleaned with Semico-Clean, ultrahigh-purification water, acetone, and IPA. The substrates were then treated with ultraviolet ozone cleaner for 15 min. The compact-TiO₂ layer was subsequently deposited onto the cleaned FTO substrates by spin-coating a solution of titanium(IV) isopropoxide (1.5 ml) in ethanol (10 ml) and hydrochloric acid (0.1 ml) to form an approximately 50-nm-thick layer. All the samples were moved onto a hotplate and kept at 125 °C for 20 min and then sintered at 500 °C for 30 min. For the perovskite layer, a 1.1 M PbI₂/DMF solution was consecutively spin-coated onto the TiO₂ layer at 2000 rpm for 5 s and 6000 rpm for 10 s. After a delay time of 2 s, clean airflow (0.2 MPa) was blown over the surface of the PbI₂ during the spin-coating process. The height from the substrate to the outlet is controlled to be 8–10 cm. Next, the PbI₂ film was dried at 100 °C for 10 min, then 10 mg/ml CH₃NH₃I/IPA solution was loaded onto the PbI₂ layer with a waiting time of 30 s before spin-coating at 4000 rpm for 30 s to form a 350-nm-thick layer. Dry air was blown over the surface of the perovskite solution for the duration of spin-coating. After spin-coating, the samples were set on the hotplate for crystallization at 100 °C for 10 min. For w/o AAF cells used as a control group, other than omitting the airflow, every other condition was kept identical. A solution of Spiro-OMeTAD was spin-coated on top of the perovskite film at 2000 rpm for 60 s to form an approximately 250-nm-thick layer. The Spiro-OMeTAD solution was prepared by mixing 17.5 µl of lithium bis(trifluoromethanesulfonyl)imide (Li-TFSI) solution (520 mg of Li-TSFI in 1 ml of acetonitrile) and 28.8 µl of 4-tert-butylpyridine with 72.3 mg of Spiro-OMeTAD in 1 ml of chlorobenzene solution. The final devices were completed by vacuum evaporation of Au back contacts with a thickness of 80 nm at 3.4 × 10⁻⁴ Pa. A metal mask was used to define the active area at 0.06 cm².

2.3. Characterization

The morphology of the perovskite layer was observed using field-emission scanning electron microscopy (FE-SEM; JEOL JSM-6335FM, acceleration voltage of 10 kV). The atomic force microscopy (AFM) images were obtained using a Keyence VN-8000 viewer and analyzer. The current–voltage (J–V) measurement was recorded by applying external potential biases to the cells and recording the output photocurrent with a digital source meter (Agilent B2901A). A 150 W xenon lamp (Bunkoukeiki Otento-SUN3 Xe-S150) was applied as the light source and the output irradiation intensity was adjusted to the AM 1.5G condition (100 mW/cm²). Before each measurement, the light intensity was calibrated using a silicon reference cell (Bunkoukeiki). The step voltage and delay time were 10 mV and 10 ms, respectively. Incident photon conversion efficiency (IPCE) measurement was conducted in the wavelength range of 300–1100 nm with a 300 W xenon light source and a monochromator (Asahi Spectra PVL 3300). Ultraviolet–visible (UV–vis) absorption spectra were measured using a UV–vis spectrophotometer (Shimadzu UV 2450). X-ray diffraction (XRD) patterns were recorded in the 2θ range of 5–65° using an X-ray diffractometer (Rigaku RINT2500V/PC) with Cu Kα radiation (40 kV, 100 mA). Photoluminescence (PL) spectra were measured using a Jasco NRS-5100PL laser Raman spectrophotometer. The excitation wavelength was 325.29 nm and the detection range was 600 to 900 nm. For thickness testing, we applied a thickness tester (Kosaka Laboratory Surfcorder ET200).

3. Results and discussion

All our fabrications were conducted in a clean room in ambient air with approximately ∼35% relative humidity, except for the vacuum deposition of the Au back contact (see Sect. 2 for more details). The schematic presentation of the device layout is shown in Fig. 1(a). Figure 1(b) shows the energy level alignment of the materials used in the solar cell construction. Figure 1(c) presents the schematic of AAF in the two-step spin-coating process for preparing PbI₂ and CH₃NH₃PbI₃ films. The w/o AAF devices, which were fabricated under identical conditions, were compared with the w/ AAF devices.

**Fig. 1.** (a) Device structure. (b) Approximate energy alignment of the materials used for the planar structure. (c) Schematic of procedure for air-assisted flow in the two-step spin-coating method progressing from left to right.
Download figure:
Standard image High-resolution image

The top-view and cross-sectional SEM images of PbI₂ and CH₃NH₃PbI₃ films prepared by w/ AAF and w/o AAF methods are shown in Figs. 2 and 3. The thicknesses of perovskite films prepared by the two methods were both about 350 nm. Under the w/o AAF condition, PbI₂ and CH₃NH₃PbI₃ films did not sufficiently cover the underlying compact-TiO₂ layer, as shown in Figs. 2(a) and 2(b). Rough PbI₂ and CH₃NH₃PbI₃ layers with several grooves that could lead to direct contact between the compact-TiO₂ layer and HTL [Figs. 3(a) and 3(b), red arrows] were also confirmed. However, there were significant improvements when the w/ AAF method was incorporated [Figs. 2(c) and 2(d)]. The compact-TiO₂ underlying layer was completely covered by PbI₂ and CH₃NH₃PbI₃ crystal particles, as shown in Figs. 3(c) and 3(d).

**Fig. 2.** SEM images of the PbI₂ and CH₃NH₃PbI₃ films prepared by (a, b) w/o AAF method and (c, d) w/ AAF method.
Download figure:
Standard image High-resolution image

**Fig. 3.** Cross-sectional SEM images of PbI₂ and CH₃NH₃PbI₃ films prepared by (a, b) w/o AAF method and (c, d) w/ AAF method.
Download figure:
Standard image High-resolution image

The sizes of the perovskite crystals obtained w/ AAF and w/o AAF were analyzed by Nano Measure, as shown in Figs. 4(a) and 4(b). The perovskite film prepared by the w/ AAF method had particle sizes ranging from 100 to 300 nm with a mean particle size of 196.4 ± 21 nm. In contrast, the w/o-AAF-based perovskite film had particle sizes ranging from 130 to 400 nm, with a mean particle size of 262.5 ± 51 nm. The smaller variation in particle size distribution exhibited by the w/ AAF perovskite film could be attributed to both the fast evaporation of DMF solvent in PbI₂ and the controlled reaction kinetics between PbI₂ and CH₃NH₃I by AAF during spin-coating. The morphology of perovskite films with relatively uniform distributions of crystal sizes is expected to be beneficial in enhancing the device PCE.

Fig. 4. — Download figure:
Standard image High-resolution image

Notably, we have also employed the AAF route to prepare PbI₂ films, contrary to what had been previously reported.¹³^,²⁸⁾ We found that a high-quality CH₃NH₃PbI₃ film is strongly influenced by the morphology of PbI₂ in two-step spin-coating or sequential deposition, and by controlling the crystallization of PbI₂ with AAF, a highly compact layer could be readily obtained.²⁶⁾ In our case, the PbI₂ films prepared by the w/o AAF route are usually made up of large crystals and have a rough surface, whereas the w/ AAF method yielded a much smoother surface [Figs. 5(a) and 5(b)]. The rough surface of the w/o AAF films could be attributed to the ultralow evaporation rate of DMF, which has a boiling point of ∼153 °C and a low vapor pressure of about 2.6 mmHg at 20 °C. These solvent properties result in low degrees of supersaturation and precipitation,¹³⁾ and thereby promote the formation of nonconstant perovskite morphology in the underlying compact-TiO₂ layer.

Fig. 5. — Download figure:
Standard image High-resolution image

Typical XRD patterns for w/ AAF and w/o AAF CH₃NH₃PbI₃ crystals coated on FTO glass/compact-TiO₂ substrates are presented in Fig. 6. Strong diffraction peaks located at 2θ of 14.08°, 23.48°, 28.40°, and 31.86° can be assigned to (110), (211), (220), and (310) planes of the tetragonal CH₃NH₃PbI₃ phase, respectively. Notably, the XRD patterns for the two films are the same, and they reveal traces of remnant PbI₂ in both w/ AAF and w/o AAF CH₃NH₃PbI₃ films. We inferred that this might be the result of an incomplete reaction of PbI₂ and CH₃NH₃I owing to the absence of a mesoporous scaffold.²⁹⁾ We made effort to extend the loading time to 180 s, but the PbI₂ phase could not be entirely removed from the perovskite film. Previous reports have suggested that residual unreacted PbI₂ may act as a built-in hole-blocking layer, and may be advantageous for the solar cell light conversion.²³^,³⁰⁾

The J–V curves of forward and reverse scans of our best-performance solar cells are shown in Fig. 7. The photovoltaic parameters of the best performance, average values, and corresponding standard deviation of w/ AAF and w/o AAF planar PSCs are summarized in Table I. All the parameters of the best-performance w/ AAF solar cell are superior to those of the w/o AAF solar cell. For the w/o AAF solar cell, the forward potential scan (from short circuit to open circuit) revealed values of 17.91 mA/cm², 0.864 V, 0.56, and 8.67% for the short-circuit density (J_sc), open-circuit voltage (V_oc), fill factor (FF), and PCE, respectively. Similarly, the reverse potential scan (from open circuit to short circuit) resulted in J_sc of 17.91 mA/cm², V_oc of 0.881 V, FF of 0.62, and PCE of 9.78%. Conversely, for the w/ AAF device, higher values of J_sc of 21.52 mA/cm², V_oc of 0.98 V, FF of 0.63, and PCE of 13.28% were obtained from the forward potential scan, and 21.63 mA/cm², 1.02 V, 0.714, and 15.75% respectively, from the reverse potential scan. The calculated i.e., average PCE values were 14.52 and 9.22% for the best-performance w/ AAF and w/o AAF devices, respectively.

Table I. Photovoltaic parameters of PSCs prepared by w/ AAF and w/o AAF methods, measured under AM 1.5G 1 sun illumination conditions (100 mW/cm²). The values in brackets correspond to the average values from 10 planar cells and the corresponding standard deviation.

Solar cells	Scan direction	J_sc (mA/cm²)	V_oc (V)	FF	PCE (%)	Average PCE (%)
(Best cells)	Forward	21.52	0.980	0.630	13.28
w/ AAF	Reverse	21.63	1.020	0.714	15.75	14.52
w/ AAF	Forward	(21.46 ± 0.20)	(0.980 ± 0.003)	(0.620 ± 0.01)	(13.00 ± 0.15)
(10 Cells)	Reverse	(21.61 ± 0.56)	(1.000 ± 0.006)	(0.710 ± 0.01)	(15.34 ± 0.20)
(Best cells)	Forward	17.91	0.864	0.560	8.67
w/o AAF	Reverse	17.91	0.881	0.620	9.78	9.22
w/o AAF	Forward	(17.45 ± 0.21)	(0.894 ± 0.031)	(0.529 ± 0.02)	(8.25 ± 0.30)
(10 Cells)	Reverse	(17.43 ± 0.21)	(0.900 ± 0.030)	(0.560 ± 0.05)	(8.77 ± 0.70)

Hysteresis was observed in the J–V characterization of both cells, but was not pronounced. The results of the statistical analysis of photovoltaic parameters obtained from 10 planar PSCs are shown in Fig. 8. Our planar cells showed average efficiencies with small standard deviation (Table I). The w/o AAF and w/ AAF devices have average efficiencies and standard deviations of 8.25 ± 0.30% and 13.00 ± 0.15% for the forward potential scan and 8.77 ± 0.70% and 15.34 ± 0.20% for the reverse potential scan, respectively. The small standard deviations obtained from the devices indicate that both types of devices have good reproducibility. The evident disparities between the photovoltaic parameters of the two types of solar cells are attributed to their CH₃NH₃PbI₃ layer morphological differences. Note that a nonuniform perovskite morphology will cause (1) light to pass straight through the perovskite film without absorption, thereby decreasing the available photocurrent, and (2) a high frequency of shunt paths, which will enable contact between the HTM layer and the compact-TiO₂ layer.⁸^,³¹⁾ Both of these effects can unavoidably cause a decrease in V_oc and FF and consequently PCE.

**Fig. 8.** Statistical analysis of forward potential scan (F-scan) and reverse potential scan (R-scan) results of J_sc, V_oc, FF, and PCE of the perovskite solar cells fabricated by w/ AAF and w/o AAF methods. The data from 10 cells were statistically analyzed.
Download figure:
Standard image High-resolution image

To gain insight into the influence of perovskite morphology on J_sc, the IPCE spectra, which specify the ratio of extracted electrons to incident photons at a given wavelength, were measured for the best-performance w/o AAF and w/ AAF solar cells (Fig. 9). The IPCE spectra of both devices show photoresponse in a wide range from 350 to 800 nm, with a maximum value close to ∼80% for the w/ AAF solar cell. The relatively high maximum efficiency (∼80%) obtained by the w/ AAF solar cell is indicative of excellent light harvesting efficiency due to the homogeneous and well-controlled perovskite film. The integrated J_sc values from IPCE curves for w/o AAF (17.07 mA/cm²) and w/ AAF (21.32 mA/cm²) solar cells were found to be consistent with the experimentally measured J_sc value under simulated AM 1.5 illumination of 100 mW/cm². Mismatch between the simulated sunlight and the AM 1.5G standard is assumed to be negligible.

**Fig. 9.** IPCE spectra and integrated photocurrent curves for the best w/ AAF (blue circles) and w/o AAF (red squares) solar cells.
Download figure:
Standard image High-resolution image

Absorption spectroscopy was carried out to study the optical properties of CH₃NH₃PbI₃ prepared by w/ AAF and w/o AAF methods. As shown in Fig. 10, the UV–vis spectra of PbI₂ and CH₃NH₃PbI₃ films prepared by the two methods span 400–800 nm. The w/ AAF PbI₂ and CH₃NH₃PbI₃ films comparatively show an increased absorption intensity from the UV-to-near-infrared region. This is consistent with the increased performance of IPCE, as shown in Fig. 9. The increased absorbance exhibited by the AAF-based films should be attributed to the improved surface coverage of the underlying compact-TiO₂ layer.

We further characterized the PbI₂ and CH₃NH₃PbI₃ films derived by the two methods by AFM, as shown in Fig. 11. The calculated root-mean-squared roughnesses of PbI₂ and perovskite films made by the w/ AAF method [Figs. 11(a) and 11(b)] are 21.9 and 8.7 nm, respectively. For the films made by the w/o AAF method [Figs. 11(c) and 11(d)], they are 67.1 and 62.8 nm, respectively. The roughness of films fabricated by the w/ AAF route is significantly reduced compared with that of w/o AAF films, as is evident in the AFM images. The significant disparities in the calculated RMS values between the w/o AAF and w/ AAF films show that the AAF route is more effective for fabricating very smooth and homogeneous perovskite films for reproducible and high-efficiency solar cells.

The PL measurements were performed to gain further insight into the relevance of the compact-TiO₂/perovskite and perovskite/HTM interface for the two types of preparation methods. Figure 12 shows the PL spectra for compact-TiO₂/CH₃NH₃PbI₃ and CH₃NH₃PbI₃/spiro-OMeTAD samples prepared by the w/ AAF and w/o AAF methods. The emission peaks observed at around 760 nm resulted from the PL of CH₃NH₃PbI₃ which is similar to results indicated in other reports.¹⁾ In general, samples prepared with AAF exhibit lower PL emission. However, the PL quenching (Fig. 12) is stronger when the samples are excited through the Spiro-OMeTAD film. When considering the sufficient coverage of CH₃NH₃PbI₃ for the w/ AAF samples compared with the w/o AAF samples, the lower PL emission and significant quenching indicate that better electron and hole collections are achieved when the electron–hole pair is generated close to the compact-TiO₂/perovskite and perovskite/spiro-OMeTAD interfaces, respectively. We then inferred that the improved PV performance of the w/ AAF solar cells is mainly due to the efficient charge separation at the two interfaces. However, further studies are in progress.

**Fig. 12.** PL spectra of compact-TiO₂/CH₃NH₃PbI₃ and compact-TiO₂/CH₃NH₃PbI₃/spiro-OMeTAD on FTO glass substrate prepared by w/ AAF and w/o AAF methods.
Download figure:
Standard image High-resolution image

4. Conclusions

We demonstrated the importance of using AAF in the two-step spin-coating method for preparing CH₃NH₃PbI₃ perovskite films for highly efficient solar cells. Two-step spin-coating deposition is found to be more precise than the two-step dipping method because of its quantitatively managed procedure. Improving the CH₃NH₃PbI₃ coverage of the underlying compact-TiO₂ layer was vital to achieving high light harvesting and efficient charge separation for the w/ AAF devices. The application of AAF in the fabrication of both PbI₂ and perovskite films for PSCs led to a PCE of 13.28% for the forward potential scan and 15.75% for the reverse potential scan. The correlations between the efficiencies and film morphologies of our solar cells were systematically studied using absorption spectra, AFM images, and SEM images. Our studies provide a promising route for fabricating low-cost high-performance PHJ solar cells by an easy method.

Acknowledgments

This research is financially supported by the MEXT-Supported Program for Strategic Research Foundation at Private Universities (S1001033, 2010–2014), an AIT special grant for Education and Research, and Hibi Science Foundation. We thank Professor Wakamiya and Dr. Endo of Kyoto University for assistance in the fabrication of the perovskite solar cells and Professor Kobayashi for the measurement of XRD at the Research Institute of Industrial Technology, AIT.

Air-assisted flow and two-step spin-coating for highly efficient CH₃NH₃PbI₃ perovskite solar cells

Article metrics

Permissions

Author e-mails

Author affiliations

Dates

Abstract

1. Introduction