Microstructures and photovoltaic properties of perovskite-type CH₃NH₃PbI₃ compounds

Recently, organic–inorganic hybrid solar cells with perovskite-type pigments have been heavily studied.^1–5) Organic solar cells with a CH₃NH₃PbI₃ compound that has a perovskite structure have high conversion efficiencies and stability.^6–8) Since achieving a photoconversion efficiency of 15%,⁹⁾ efficiencies as high as 19.3%¹⁰⁾ have been reported for various device structures and processes.^11–14) The photovoltaic properties of solar cells are strongly dependent on the fabrication process, hole transport layers, electron transport layers, nanoporous layers, interfacial microstructures, and crystal structures of the perovskite compounds. In particular, the crystal structures of the perovskite-type compounds strongly affect the electronic structures, such as energy band gaps and carrier transport, and a detailed analysis of them is imperative.

The purpose of the present work is to investigate microstructures and photovoltaic properties of photovoltaic devices with perovskite-type CH₃NH₃PbI₃ compounds, TiO₂ electron transport layers, and 2,2',7,7'-tetrakis(N,N-di-pmethoxyphenylamine)-9,9'-spirobifluorene (spiro-OMeTAD) hole-transport layers (HTL), which were prepared by a simple spin-coating technique. Effects on the microstructures and photovoltaic properties of annealing and multiple spin-coating of perovskite compounds were investigated by X-ray diffraction (XRD), transmission electron microscopy (TEM), current density–voltage (J–V) curves, and incident photon-to-current conversion efficiency (IPCE). TEM is a powerful tool for structure analysis of solar cells¹⁵⁾ and perovskite-type structures at the atomic scale.^16,17)

A schematic illustration of the fabrication of the present TiO₂/CH₃NH₃PbI₃ photovoltaic cells is shown in Fig. 1. The details of the fabrication process are described in the reported paper^9,18) except for the mesoporous TiO₂ layer.¹⁹⁾ The 0.30 M TiO_x precursor solution was prepared from titanium diisopropoxide bis(acetyl acetonate) (Sigma Aldrich, 0.55 µL, 0.11 mL) with 1-butanol (1 mL), then spin-coated on the F-doped tin oxide (FTO) substrate and annealed at 125 °C for 5 min. This process was performed two times, and the FTO substrate was sintered at 500 °C for 30 min to form the compact TiO₂ layer. After that, mesoporous TiO₂ paste was coated on the substrate by spin-coating. For the mesoporous TiO₂ layer, the TiO₂ paste was prepared with TiO₂ powder (Aerosil P-25) with poly(ethylene glycol) (Nacalai tesque, PEG #20000) in ultrapure water. The solution was mixed with acetylacetone (10.0 µL) and triton X-100 (Sigma Aldrich, 5 µL) for 30 min. The cells were annealed at 120 °C for 5 min and at 500 °C for 30 min. CH₃NH₃I was synthesized by reacting 23.2 mL of methylamine (CH₃NH₂, Tokyo Chemical Industry) and 25.0 mL of hydroiodic acid at 0 °C for 2 h with stirring. Then dried using a rotary evaporator at 60 °C for 3 h. The obtained CH₃NH₃I was then dried in a vacuum. For the preparation of pigment with a perovskite structure, a solution of CH₃NH₃I (98.8 mg) and PbI₂ (Sigma Aldrich, 289.3 mg) with a mole ratio of $1:1$ in γ-butyrolactone (Nacalai Tesque, 0.5 mL) was mixed at 60 °C. The solution of CH₃NH₃PbI₃ was then introduced into the TiO₂ mesopores by a multiple spin-coating method and annealed at 100 °C for 15 min. The specimen was dried and cooled in air for a few seconds between the multiple spin-coatings to avoid the re-dissolution of the pre-coated layer. Then, the HTL was prepared by spin-coating. To create the HTLs, a solution of spiro-OMeTAD (Wako Pure Chemical Industries, 36.1 mg) in chlorobenzene (0.5 mL) was mixed with a solution of lithium bis(trifluoromethylsulfonyl) imide (Li-TFSI, Tokyo Chemical Industry, 260 mg) in acetonitrile (0.5 mL) for 12 h. The former solution with 4-tert-butylpyridine (14.4 µL) was mixed with the Li-TFSI solution (8.8 µL) for 30 min at 70 °C. All procedures were carried out in air. Finally, gold (Au) metal contacts were evaporated to form top electrodes. Layered structures of the present photovoltaic cells were denoted as FTO/TiO₂/CH₃NH₃PbI₃/spiro-OMeTAD/Au, as shown in a schematic illustration in Fig. 1.

Zoom In Zoom Out Reset image size

The J–V characteristics of the photovoltaic cells were measured under illumination at 100 mW cm⁻² by using an AM 1.5 solar simulator. The solar cells were illuminated through the side of the FTO substrates, and the illuminated area was 0.16 cm². The IPCE of the cells was also investigated (Enli Technology QE-R). The microstructures of the thin films were investigated by using an X-ray diffractometer (Bruker D2 PHASER) and a transmission electron microscope (Hitachi H-8100) operated at 200 kV.

The J–V characteristics of the TiO₂/CH₃NH₃PbI₃/spiro-OMeTAD photovoltaic cells under illumination are shown in Fig. 2(a), which indicates an annealing effect of the CH₃NH₃PbI₃ layer. The as-deposited CH₃NH₃PbI₃ cell provided a conversion efficiency of 2.83%. The CH₃NH₃PbI₃ cell annealed at 100 °C for 15 min provided better photovoltaic properties compared with the as-deposited one, as listed in Table I. The highest efficiency was obtained for the annealed CH₃NH₃PbI₃ cell, which provided a power conversion efficiency (η) of 5.16%, a fill factor (FF) of 0.486, a short-circuit current density (J_SC) of 12.9 mA cm⁻², and an open-circuit voltage (V_OC) of 0.827 V. Figure 2(b) shows results of multiple spin-coating of CH₃NH₃PbI₃, which will be described later.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

XRD patterns of CH₃NH₃PbI₃ thin films on the glass substrate before and after annealing are shown in Fig. 3(a). The diffraction peaks can be indexed by tetragonal and cubic crystal systems for as-deposited and annealed films, respectively. Although the as-deposited film is a single perovskite phase, broader diffraction peaks due to a PbI₂ compound appeared after annealing, as shown in Fig. 3(a). Figures 3(b) and 3(c) are enlarged XRD patterns at 2θ of ∼14 and ∼28°, respectively. Split diffraction peaks of 002–110 and 004–220 for the as-deposited sample changed into diffraction peaks of 100 and 200 after annealing, which indicates the structural transformation from the tetragonal to cubic crystal system.

Zoom In Zoom Out Reset image size

The CH₃NH₃PbI₃ crystals have perovskite structures and provide structural transitions from a tetragonal to a cubic system upon heating at ∼330 K,^20–22) as shown in the structure models of Figs. 4(a) and 4(b). Although the CH₃NH₃PbI₃ perovskite crystal has a cubic symmetry for the highest temperature phase, the CH₃NH₃ ion is polar and has C_3v symmetry, which results in a disordered cubic phase.^23,24) In addition to the disordering of the CH₃NH₃ ion, the iodine (I) ions are also disordered in the cubic phase, as shown in Fig. 4(a). Site occupancies were set as 1/4 for I and 1/12 for C and N. The CH₃NH₃ ion occupies 12 equivalent orientations of the C₂ axis, and hydrogen atoms have two kinds of configurations on the C₂ axis.²²⁾ As the temperature decreased, the cubic phase was transformed into the tetragonal phase, and I ions became ordered, which resulted in a lower symmetry than in the cubic phase, as shown in Fig. 4(b). Site occupancies were set as 1/4 for C and N for the tetragonal CH₃NH₃PbI₃. For the high temperature phase, unit cell volume of the cubic structure is 261.0 ų, which is bigger than that of the tetragonal structure (245.6 ų), as listed in Table II. This would be due to both thermal expansion of the unit cell and atomic disordering of I in the cubic phase. As the temperature lowers, the tetragonal phase is transformed into the orthorhombic systems because of the ordering of CH₃NH₃ ions in the unit cell.²⁵⁾ Calculated XRD patterns of the CH₃NH₃PbI₃, with cubic and tetragonal structures based on the structure models of Figs. 4(a) and 4(b), are shown in Fig. 4(c), and the full width at half maximum was set as 0.10. The calculated XRD pattern indicates splitting of diffraction reflections of 100 to 002/110 by reduction of crystal symmetry from the cubic to the tetragonal phase.

Zoom In Zoom Out Reset image size

The XRD results in Fig. 3 indicated phase transformation of the CH₃NH₃PbI₃ perovskite structure from the tetragonal to the cubic system by partial separation of PbI₂ from the CH₃NH₃PbI₃ phase during annealing, which corresponds to the decrease of the unit cell volume of the perovskite phase from 248.27 to 246.78 ų, as listed in Table II. In addition to the I atoms, Pb atoms might also be disordered, and site occupancies of Pb atoms would be less than 1. It should be noted that the structure transition from the tetragonal to the cubic structure in the present work is due to the partial separation of PbI₂ from the CH₃NH₃PbI₃ phase, which is different from the ordinary tetragonal–cubic transition at 330 K.^20,21)

Figures 5(a) and 5(b) are a TEM image and an electron diffraction pattern of TiO₂/CH₃NH₃PbI₃, respectively. The TEM image shows TiO₂ nanoparticles with sizes of ∼50 nm, and the polycrystalline CH₃NH₃PbI₃ phase shows dark contrast, which is due to Pb having the largest atomic number in the present materials. The electron diffraction pattern of Fig. 5(b) shows the Debye–Scherrer rings from the nanocrystalline TiO₂ particles, which are indexed as the 101, 004, and 200 spacings of the anatase TiO₂. The thickness of the mesoporous TiO₂ layer was measured to be ∼300 nm by using atomic force microscopy. In addition to the Debye–Scherrer rings, diffraction spots corresponding to the CH₃NH₃PbI₃ crystal with a cubic structure²³⁾ were observed and indexed, as shown in Fig. 5(b). Other diffraction spots, except for the Debye–Scherrer rings of TiO₂, are also from the CH₃NH₃PbI₃ nanoparticles. A structure model and its calculated electron diffraction pattern of a cubic CH₃NH₃PbI₃ phase projected along the [210] direction are shown in Figs. 5(c) and 5(d), respectively. The calculated electron diffraction pattern agrees well with the observed pattern of Fig. 5(b).

Zoom In Zoom Out Reset image size

The J–V characteristics of the TiO₂/CH₃NH₃PbI₃/spiro-OMeTAD photovoltaic cells prepared by multiple spin-coating of CH₃NH₃PbI₃ are shown in Fig. 2(b). Figure 2(b) indicates the effect of spin-coating times of CH₃NH₃PbI₃ on the photovoltaic properties, and the detailed parameters of the best device are listed in Table III. The highest efficiency was obtained for the cell that was spin-coated for four times, which provided an η of 6.96%, an FF of 0.496, a J_SC of 16.5 mA cm⁻², and a V_OC of 0.848 V. Further spin-coating lowered the efficiencies of the cells, as shown in Fig. 2(b). Although 2-step deposition⁶⁾ (spin-coating PbI₂ and dipping in the CH₃NH₃I solution) was also performed in air, the efficiency was lower compared with that by multiple spin-coating, as observed in Fig. 2(b) and Table III.

Figure 6(a) shows XRD patterns of TiO₂/CH₃NH₃PbI₃ cells that were spin-coated with CH₃NH₃PbI₃ layers for one or four times. Enlarged XRD patterns of the 100 reflection of the cubic structure are shown in Fig. 6(b). The 100 reflection of 4-layers was shifted to a lower angle and broadened compared with the cell prepared by 1-time spin coating. The lattice parameters of the CH₃NH₃PbI₃ for 1 and 4 layers are 6.3049 and 6.3222 Å, respectively, which indicates lattice expansion of the unit cell. On the other hand, the crystallite sizes for 1 and 4 layers are 52.0 and 43.3 nm, respectively. It is believed that the CH₃NH₃PbI₃ phase was stuffed in pores inside of the mesoporous TiO₂ layer during the 1- or 2-times spin-coating. After the pores inside of the mesoporous TiO₂ were fully filled with the perovskite phase, the perovskite-only layer would be formed on the mesoporous TiO₂ layer for 3- or 4-times spin-coating, which provided the highest efficiency. A Decrease in crystallite size due to spin-coating indicates that the grain growth of the perovskite phase might be enhanced in the mesoporous TiO₂.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The IPCE spectrum of the photovoltaic cell with the TiO₂/CH₃NH₃PbI₃/spiro-OMeTAD structure is shown in Fig. 7(a). The measurement region is in the range of 300 to 900 nm, and the perovskite CH₃NH₃PbI₃ structure exhibits photo-conversion efficiencies between 300 and 800 nm, which almost agrees with the reported energy gaps of 1.51²⁵⁾ and 1.61 eV²⁶⁾ (corresponding to 821 and 770 nm, respectively) for the CH₃NH₃PbI₃ phase. This indicates that excitons and/or free charges would be effectively generated in the perovskite layers upon illumination by light.

Zoom In Zoom Out Reset image size

From the TEM results, size distributions of TiO₂ nanoparticles were observed, indicating a microcrystalline structure, as shown in Fig. 7(b), and there seems to be no special crystallographic relation at the interface. The interface between the TiO₂ and CH₃NH₃PbI₃ phases would not be perfectly connected over the large area. The cell prepared by 4-times spin-coating provided the highest efficiency, which would have an interfacial microstructure as shown in Fig. 7(b). For the cell prepared by 10-times spin coating, layer thickness of the CH₃NH₃PbI₃ phase was too large, which resulted in an increase in the inner electronic resistance and decrease in the efficiency, as listed in Table III.

An energy level diagram of TiO₂/CH₃NH₃PbI₃ photovoltaic cells is summarized as shown in Fig. 7(a). Previously reported values were also used for the energy levels of the figures by adjusting them to the present work.^1,15) An energy barrier would exist at the semiconductor metal interface.²⁷⁾ The electronic charge generation is caused by light irradiation from the FTO substrate side. The TiO₂ layer receives the electrons from the CH₃NH₃PbI₃ crystal, and the electrons are transported to the FTO. The holes are transported to an Au electrode through spiro-OMeTAD. Experimental evidence for photogenerated free carriers, which would enhance the carrier transport, was reported.²⁸⁾ In the present work, the samples were prepared in air, which may result in the reduction of the photo- and thermal stability. Perovskite crystals with higher quality should be prepared in further works.

In conclusion, the structure analysis of TiO₂/CH₃NH₃PbI₃ indicated phase transformation of the perovskite structure from the tetragonal to the cubic system by partial separation of PbI₂ from the CH₃NH₃PbI₃ phase during annealing, which was confirmed by reduction of the unit cell volume of the perovskite phase and resulted in the improvement of photovoltaic properties of the devices. Effects of a multiple spin-coating method were also investigated, which increased the efficiency to 6.96% for the 4-times spin-coating. This improvement would be due to the perfect coverage and optimum thickness of the perovskite phase on the porous TiO₂ layer. Moreover, the lattice constant and crystallite size of the CH₃NH₃PbI₃ increased and decreased, respectively, which indicates the microstructural difference of the perovskite phase between the inside of and above the porous TiO₂.