Effect of high-temperature post-deposition annealing on cesium lead bromide thin films deposited by vacuum evaporation

Optical wireless power transmission (OWPT) is one of the wireless power transmission techniques that uses a light source and an optical power converter.¹ Semiconductor lasers or light emitting diodes (LEDs) are used for the light source, and solar cells are used for the optical power converter.² In this application, the efficiency of the OWPT system is mainly determined by the product of the conversion efficiency of the light source and the conversion efficiency of the optical power converter. Therefore, the development of a high efficiency solar cell for monochromatic light is important. When a solar cell is illuminated with monochromatic light with a photon energy of E_ph, the conversion efficiency η(E_ph) can be expressed as follows:

where F is the incident photon flux, V_OC is the open circuit voltage of the solar cell, J_SC is the short-circuit current density of the solar cell, FF is the fill factor of the solar cell, and e is the elementary charge. The first term on the right-hand side is referred to as voltage efficiency and the second term on the right-hand side is referred to as current efficiency that is equal to the external quantum efficiency (EQE) of the solar cell. According to the above equation, the upper limit of the voltage efficiency is obtained when E_ph is equal to the bandgap (E_g) of the light absorption layer of the solar cell. The detailed balance theory reported by Shockley and Queisser indicates that V_OC of the solar cell is an almost linear function of the E_g and V_OC deficit (E_g − V_OC) for an ideal solar cell that is about 0.25–0.3 V under the standard sunlight.^3,4 Therefore, the maximum voltage efficiency, V_OC/E_g, increases with an increase in E_g. This means that a solar cell using a wide-gap light absorbing material has a high potential for a high efficiency optical power converter. For wide-gap optical power converters, it is important to use short-wavelength monochromatic light. It is well known that blue light emitting diodes (LEDs) show a high wall-plug efficiency of above 80%.⁵ The wall-plug efficiency of blue lasers is limited to be about 40% at present, but there is room for improvement of the efficiency.⁶ This suggests that blue LEDs and lasers are some of the good candidates for the light source of OWPT. Therefore, it is important to develop an optical power converter using a wide bandgap material for a high efficiency OWPT system.

Cesium lead bromide (CsPbBr₃) is a promising candidate for a light absorbing material for an optical power converter for OWPT. This material has been investigated as the top cell material of a four-terminal tandem solar cell since its large bandgap of approximately 2.3 eV.⁷ A further increase in the bandgap is possible by replacing some Br with Cl to obtain CsPbBr_xCl_3−x.⁸ CsPbBr₃ is known as a higher stability material against moisture compared to other lead halide perovskite materials.^9–12 Kulbak et al. demonstrated that a CsPbBr₃ perovskite based solar cell has long-term stability under the relative humidity (RH) of 60%–70%.¹³ Solution-based processes by using lead bromide (PbBr₂) and cesium bromide (CsBr) are widely used for CsPbBr₃ film deposition¹⁴ Zhao et al. developed a CsPbBr₃ solar cell with a conversion efficiency of 10.6%.¹⁵ Compared with the solution-based process, thermal evaporation is advantageous to obtain good film morphology.¹⁶ Post-deposition annealing is one of the key factors to obtain good CsPbBr₃ films by using thermal evaporation of PbBr₂ and CsBr.¹⁷ Zhang et al. obtained high quality CsPbBr₃ films with grain size exceeding 2 μm for post-deposition annealing in RH = 30%.¹⁸ In previous reports, the temperature of the post-deposition annealing was less than 250 °C.^14,19 According to Ref. 13, the sublimation temperatures of PbBr₂, CsBr, and CsPbBr₃ can be estimated to be approximately 500 °C, 700 °C, and 600 °C, respectively. Therefore, higher temperature annealing at about 500 °C is possible to apply for CsPbBr₃. Higher annealing temperature (500 °C) has potential to improve the solar cell performance in terms of improvement of crystallinity.¹⁷ However, the effect of higher temperature annealing on the optoelectronic properties of CsPbBr₃ films has not been investigated in detail. In this study, we investigated the effect of high temperature post-deposition annealing (PDA) on the properties of CsPbBr₃ films deposited by the thermal evaporation method. We also discuss the potential of the CsPbBr₃ solar cell as an optical power converter for OWPT.

CsPbBr₃ films were deposited on a Corning 7059 glass substrate by sequential evaporation of PbBr₂ and CsBr. The distance between the crucibles for the source materials and the substrate was approximately 35 cm. The back pressure was about 3 × 10⁻³ Pa. The Cs/Pb ratio of the films was controlled by changing the weight ratio of PbBr₂ and CsBr sources that were placed in each crucible. The PbBr₂ and CsBr at a stoichiometric molar ratio of 1:1 were used in this study. After film deposition, thermal annealing was carried out at 500 °C for 1–120 min in nitrogen (N₂) ambient by using a hot plate. A clean glass sheet was placed on the hot plate, and the samples were placed with the film side down on the glass sheet to prevent sublimation of the deposited films. This annealing process is referred to as PDA. The total thickness of the PbBr₂/CsBr stack layer was about 320 nm. The crystal structure and grain size were evaluated by X-ray diffraction (XRD) and atomic force microscopy (AFM), respectively. Steady-State Photo-Conductivity (SSPC) measurements were carried out by using the films with coplanar Au electrodes. The width and gap of the electrodes were 3 mm and 0.2 mm, respectively. The carrier diffusion length of the CsPbBr₃ films was estimated from SSPC measurements under monochromatic light with a wavelength of 400 nm. The chopped monochromatic light with a power density of 0.52 mW/cm² was irradiated between the gap of the coplanar Au electrodes. Photoconductivity (σ_ph) was measured by using a lock-in measurement system. The measured photoconductivity can be described by

where e is the elementary charge (1.60 × 10⁻¹⁹ C), μ is the mobility, τ is the excited carrier lifetime, F is photon flux (1.0 × 10¹⁵ cm⁻² s⁻¹), A is the absorbance of the films, and d is the thickness of the films. The value of A was estimated by transmittance and reflectance measurements. By applying Einstein’s relationship, Eq. (1) can be rearranged as follows:

where L is the carrier diffusion length, k_B is the Boltzmann constant, and T is the temperature (298 K).

We also fabricated CsPbBr₃ solar cells by using a compact titanium oxide (c-TiO₂) electron transport layer and a Poly(3-hexylthiophene-2,5-diyl) (P3HT) hole transport layer. The structure of the solar cell was glass/fluorine-doped tin oxide (FTO)/c-TiO₂/CsPbBr₃ (330 nm, 680 nm, 1020 nm, 1360 nm)/P3HT/Au. The sheet resistance of the glass/FTO substrate was 10 Ω⁻². The c-TiO₂ layer was spin-coated (4000 rpm for 20 s) by using a mixture of titanium butoxide (150 μl)/ethanol (3 ml)/HNO₃ (0.1M, 150 μl). After spin-coating, the sample was annealed at 500 °C for 1 h in air. Subsequently, the substrate was immersed in an aqueous solution of 50 mM titanium chloride at 70 °C for 30 min and then annealed at 500 °C for 30 min in air. After the deposition and PDA of the CsPbBr₃ layer, a P3HT layer was prepared by spin coating of a mixture of P3HT (5 mg)/chloroform (1 ml) and subsequently annealed at 180 °C for 10 min in N₂ ambient. Finally, Au electrodes with an area of 0.09 cm² were thermally evaporated. The fabricated solar cells were characterized by the illuminated current–voltage (I–V) measurements under AM1.5 spectrum and external quantum efficiency (EQE) measurements.

Figure 1 shows the XRD patterns of the CsPbBr₃ films on glass without and with PDA at 500 °C. The typical diffraction peaks of CsPbBr₃ located at 2θ = 15.2° and 30.7° were observed. The peak intensity of the CsPbBr₃ film with PDA at 500 °C is markedly larger than that of the CsPbBr₃ film without PDA. The peak intensity reaches its maximum at 90 min and slightly decreased at 120 min. This suggests that longer time annealing enhances the crystallization, but simultaneously sublimation of CsPbBr₃ occurred. The effect of PDA on the surface morphology of the CsPbBr₃ film on glass was investigated by AFM, as shown in Fig. 2. The as-deposited CsPbBr₃ film has a small grain size of approximately 200 nm. The PDA for 5 min increased the grain size from sub-micrometer order to micrometer order. Longer time PDA up to over 90 min further increases the grain size. In general, a large grain size is preferable for high-performance optoelectronic devices due to the reduction of the defective grain boundaries.

Figure 3 shows the carrier diffusion length of the CsPbBr₃ films. The carrier diffusion length increased with an increase in the PDA time up to 90 min. The PDA for 90 min leads to the highest carrier diffusion length of 0.5 μm. According to the surface morphology shown in Fig. 2, this increase in the carrier diffusion length is attributed to the carrier mobility and/or suppression of the carrier recombination at the grain boundaries.

Figure 4 shows the illuminated I–V characteristics of the CsPbBr₃ solar cells. To emphasize the effect of the PDA, we only show reverse scan curves of the champion cells for each annealing condition. Forward scan curves showed slightly worse photovoltaic performance than the corresponding reserve scan curves. The solar cell using CsPbBr₃ without PDA showed very poor conversion efficiency less than 0.01%. On the other hand, the solar cells using CsPbBr₃ with PDA clearly showed a substantial improvement of all solar cell parameters. These results demonstrate that the PDA can enhance the solar cell performance by increasing the carrier diffusion length. The solar cell with PDA for 60 min has a V_OC of 1.388 V, a J_SC of 5.52 mA/cm², and a FF of 0.706, yielding a conversion efficiency of 5.41%. We also fabricated the solar cell with 90-min-PDA, but no photovoltaic performance was observed due to the many pinholes on the solar cell. When we characterized the electrical properties of CsPbBr₃ films on a glass substrate, no pinholes were observed from the 90-min-PDA films. This suggests that the formation of pinholes was influenced by the substrate. In the case of PDA for 60 min, the use of the thicker CsPbBr₃ layer (680 nm) improved the solar cell performance and we obtained a conversion efficiency of 6.62% (V_OC = 1.465 V, J_SC = 6.57 mA/cm², and FF = 0.688). We also fabricated CsPbBr₃ solar cells with thicker CsPbBr₃ layers (1020 nm and 1360 nm). The conversion efficiency for the solar cell with a 1020-nm-thick CsPbBr₃ layer showed a similar conversion efficiency of 6.10% (V_OC = 1.428 V, J_SC = 6.40 mA/cm², and FF = 0.667); however, the conversion efficiency of the solar cell with a 1360-nm-thick CsPbBr₃ layer deteriorated to 5.48% (V_OC = 1.425 V, J_SC = 5.73 mA/cm², and FF = 0.671). The deterioration of the J_SC for the solar cell with a thicker CsPbBr₃ layer is attributed to the smaller carrier diffusion length. The diffusion length of the CsPbBr₃ layer in the solar cell structure is discussed below.

The EQE spectra of the champion cells plotted in Fig. 5 also indicate that the PDA drastically enhanced the carrier diffusion length in the CsPbBr₃ layer. The EQE spectra were analyzed using e-ARC software version 2.0.^20,21 We calculated EQE spectra of the CsPbBr₃ solar cell by varying the effective carrier collection length (L_c)²¹ in the CsPbBr₃ layer from 0.3 μm to 1.0 μm and infinite. The structure of the calculated device is the same as that of the fabricated devices. The optical properties of glass, FTO, c-TiO₂, and Au layers were taken from the database of e-ARC. For CsPbBr₃, we used the optical properties estimated by spectroscopic ellipsometry measurements. The optical properties of P3HT were taken from Ref. 22. The calculated EQE spectra are also shown in Fig. 5 using solid lines and broken lines. As indicated in Fig. 5, EQE in the range of 350 nm to 550 nm increases with increasing PDA time in the case of the solar cell using a 340-nm-thick CsPbBr₃ layer. The calculated EQE using L_c of 0.3 µm and 0.4 μm well reproduces the measured EQE for the 30-min-PDA device and the 60-min-PDA device, respectively. These carrier collection lengths are very close to the measured carrier diffusion length of CsPbBr₃ films. The calculated EQE for the devices with a carrier collection length of 1.0 μm and infinite suggest that further improvement of EQE is possible and a carrier collection length of about 1.0 μm is enough to obtain good EQE. According to these analyses, the J_SC of our device is smaller than other reports,^17,23,24 even though the V_OC and FF are almost comparable. This small J_SC is caused by an insufficient carrier diffusion length and a relatively thin CsPbBr₃ layer thickness. The 60-min-PDA device with a 680-nm-thick CsPbBr₃ layer showed improvement of EQE, as shown in Fig. 5(b), and this measured EQE was well reproduced by the calculated EQE using a L_c of 0.45 μm. This analysis also indicates that further improvement of EQE is possible to use the CsPbBr₃ layer with a L_c of 1.0 μm. Therefore, optimization of the annealing condition for a larger carrier diffusion length would enhance solar cell performance, especially J_SC.

We also measured the I–V characteristics of CsPbBr₃ solar cells under monochromatic light with a peak wavelength of 453 nm by using blue LED illumination. Due to the shunt path formation during the I–V measurements under AM1.5 illumination before the measurements under the blue LED, the best CsPbBr₃ solar cell with a 680-nm-thick CsPbBr₃ layer showed deterioration of the V_OC, as shown in Fig. 6. This problem of shunt path formation was not observed in solar cells with thicker CsPbBr₃ layers, as shown in Fig. 6. The CsPbBr₃ solar cell with a 1020-nm-thick CsPbBr₃ layer showed a conversion efficiency of 22.5% (V_OC = 1.502 V, J_SC = 53.7 mA/cm², and FF = 0.574) under blue LED light with an intensity of 206 mW/cm². Higher light intensity of the blue LED illumination improved the V_OC; however, FF deteriorated due to the effect of series resistance. Our theoretical calculation indicated that an ideal CsPbBr₃ solar cell shows a conversion efficiency of about 61% for monochromatic light with a wavelength of 450 nm. Further improvements of device performance, especially V_OC and FF by improving the electron and hole transport layer and reducing the series resistance, are important to improve the device performance.

In conclusion, we investigated the effect of high-temperature (500 °C) PDA on the properties of CsPbBr₃ thin films prepared by sequential PbBr₂ and CsBr evaporation. We found that the grain size of CsPbBr₃ films increased with an increase in the PDA time. A large grain with a size of above 2 μm was observed in the film with PDA at 500 °C for over 60 min. The carrier diffusion length of CsPbBr₃ films was also increased by the PDA, and a carrier diffusion length of 0.5 μm was obtained under the best PDA condition. The fabricated CsPbBr₃ solar cell using CsPbBr₃ with PDA showed a conversion efficiency of 6.62% (V_OC = 1.46 V, J_SC = 6.57 mA/cm², and FF = 0.688). The conversion efficiency for monochromatic light with a wavelength of 450 nm was estimated to be 29.1%. These results indicate that the present device does not have sufficient conversion efficiency for OWPT using blue laser or LED as a light source; however, there is plenty of room for improvement of the device performance.

This work was supported by the JSPS Grants-in-Aid for Scientific Research (KAKENHI) (Grant No. 17H03532).