Nanocrystalline gallium nitride electron transport layer for cesium lead bromide photovoltaic power converter in blue light optical wireless power transmission

Nanocrystalline gallium nitride (nc-GaN) layers were deposited by RF magnetron sputtering for the electron transport layer of the cesium lead bromide (CsPbBr3) photovoltaic power converter. We investigated the structural and electrical properties of the nc-GaN layers and found that substrate heater temperature is a key factor to determine the electrical conductivity of the nc-GaN layers. CsPbBr3 photovoltaic power converters with nc-GaN electron transport layers show good photovoltaic performance. The best performance was obtained at the substrate heater temperature of 550 °C and a conversion efficiency of 5.56% (V OC = 1.24 V, J SC = 6.68 mA cm−2, FF = 0.66) under AM1.5 G illumination with a light intensity of 100 mW cm−2. The estimated conversion efficiency under blue light with a wavelength of 450 nm is 28.8%.


Introduction
Perovskite solar cells are known as high-efficiency thin-film solar cell. 1,2) For photovoltaic power generation, perovskite materials with a bandgap of around 1.5 eV are intensively investigated since the optimum bandgap of a single junction solar cell is about 1.4 eV under natural sunlight. [3][4][5] Perovskite materials with wider and narrower bandgaps are also investigated for multijunction solar cell applications. 6,7) Apart from photovoltaic power generation, optical wireless power transmission (OWPT) [8][9][10][11] is an important application field for perovskite-based solar cells with a wide bandgap light-absorbing layer.
OWPT is a wireless power transmission technique using a light source and photovoltaic power converter. Normally, the light source is a laser or a LED and a photovoltaic power converter is a solar cell. From the viewpoint of the conversion efficiency of the light source, the wavelength range of around 790-1080 nm (IR light) 12) or 440 nm (blue light) 13) is the candidate of the light source of the OWPT system. From the viewpoint of the photovoltaic power converter, a solar cell with wider bandgap light-absorbing layer is theoretically more promising since the theoretical conversion efficiency is higher when the photovoltaic power converter is illuminated with a light source with slightly higher photon energy than the bandgap of light-absorbing layer. Therefore, it is very important to develop a photovoltaic power converter with a suitable bandgap for blue light. High-efficiency optical power converter for the IR light source has been developed and its conversion efficiency reaches to 68.9%. 14) However, there are few reports of photovoltaic power converters for the blue light source and their conversion efficiency is about 20%. 10,11) Theoretically, OWPT using blue light has a higher potential than that using the IR light source and the blue light source has the additional advantage of being suitable for underwater applications which are impossible for OWPT using the IR light source. Therefore, it is important to improve the conversion efficiency of the photovoltaic power converter which is suitable for the blue light source to fully use the potential of OWPT using a blue light source.
In our previous report, we developed a photovoltaic power converter using a cesium lead bromide (CsPbBr 3 ) lightabsorbing layer. 11) In this previous study, we investigated CsPbBr 3 photovoltaic power converter with the structure of glass/fluorine-doped tin oxide (FTO)/TiO 2 /CsPbBr 3 /P3HT/ Au. A conversion efficiency of 6.62% under AM1.5 G, 100 mW cm −2 illumination (sunlight), and a conversion efficiency of 22.5% under a blue LED light source with a wavelength of 453 nm were obtained by using a hightemperature device fabrication process. Some other groups also investigated CsPbBr 3 solar cells for photovoltaic power generation. They achieved a conversion efficiency of 10.79% under AM1.5 G, 100 mW cm −2 illumination (sunlight) 15) and an estimated conversion efficiency under the blue light source is about 43.5% [estimated from solar cell performance under AM1.5 G illumination and External quantum efficiency (EQE)]. Several reasons exist for this lower conversion efficiency than that of the photovoltaic power converter for the IR light source. The quality of the CsPbBr 3 layer, the quality of the electron transport layer (ETL) and hole transport layer (HTL), and the interface of each layer are possible reasons. In this study, we focus on the ETL/CsPbBr 3 interface. Figure 1 shows the band alignment of CsPbBr 3 , TiO 2 (standard ETL material for perovskite solar cell), gallium nitride (GaN), P3HT (HTL used in our study), 16) FTO, and Au. For the standard perovskite material, such as CH 3 NH 3 PbI 3 , with relatively large electron affinity near 3.9-4.1 eV, [17][18][19][20] the standard TiO 2 ETL is suitable since the electron affinity of TiO 2 is also about 4.1 eV 20,21) and small conduction band offset (ΔE c ) is formed. On the other hand, larger ΔE c at the interface between TiO 2 and CsPbBr 3 is expected since the electron affinity of CsPbBr 3 is about 3.3-3.4 eV. 6,22) For smooth electron transport, it is important to use an ETL material with a smaller electron affinity to reduce the ΔE c . One of the candidates is GaN. Our recent investigation of nc-GaN suggested that the electron affinity of nc-GaN is about 3.5 eV. 23) Therefore, we investigated the electrical properties of nc-GaN to realize the highly conductive nc-GaN layer. We also fabricated CsPbBr 3 photovoltaic power converters with nc-GaN ETL and characterized their performance.

Experimental methods
Nanocrystalline GaN (nc-GaN) layers were deposited on Eagle-XG glass and crystalline silicon (c-Si) substrates by RF magnetron sputtering using an undoped GaN target for structural and electrical characterizations. The base pressure of the sputtering chamber is about 5 × 10 −5 Pa. The RF and RF power were 13.56 MHz and 0.8 Wcm 2 , respectively. A gas mixture of N 2 (3.0 sccm) and Ar (7.0 sccm) was used as the sputtering gas. The deposition time ranged from 50 to 70 min to deposit approximately 45 nm-thick nc-GaN layers. In this study, we investigated the effect of substrate heater temperature on the properties of the deposited nc-GaN layers by changing the substrate heater temperature from 200°C to 600°C. X-ray diffraction (XRD) and Fourier transform IR absorption spectroscopy (FTIR) was used to investigate the structural properties of the deposited nc-GaN layers. Transmission microscopy (TEM) observations were also performed for a selected nc-GaN layer to analyze the microstructure of the sample in more detail. Electrical properties of the deposited nc-GaN layers were investigated by Hall measurement. Secondary ion mass spectroscopy (SIMS) was used to investigate impurity concentrations in selected nc-GaN layers. The Eagle-XG glass substrate was used for the XRD, TEM, SIMS, and electrical characterizations. The c-Si substrate was used for the FTIR measurements.
We also fabricated CsPbBr 3 photovoltaic power converters with nc-GaN ETL. The structure of the devices is antireflection (AR) film/Glass/FTO/nc-GaN/CsPbBr 3 /P3HT/Au. The FTO glass substrate (sheet resistance: 10 Ω sq. −1 ) was purchased from Furuuchi Chemical Co. After ultrasonic cleaning with acetone and 2-propanol and ozone cleaning of FTO glasses, nc-GaN layers were deposited at different substrate heater temperatures. On the FTO glasses with the nc-GaN layers, CsPbBr 3 was thermally evaporated from CsPbBr 3 power purchased from TCI (low water content grade). The average deposition rate is about 50 nm min -1 . After the thermal evaporation, the samples were annealed at 300°C on a hot plate in a N 2 atmosphere for 3 min. We also applied a 2-propanol treatment to improve the electrical properties of the CsPbBr 3 layers. The 2-propanol treatment is the process in which immersion of the sample in 2-propanol solution for short time (a few seconds) and drying with N 2 is repeated 100 times in an atmosphere of 30% humidity. After the 2-propanol treatment, the P3HT layer was prepared on the CsPbBr 3 layer by spin coating of a mixture of P3HT (10 mg)/chloroform (1 ml) and subsequently annealed at 180°C for 10 min in N 2 ambient. The Au back electrode with a size of 3 mm square was thermally evaporated on the P3HT layer. Finally, the AR film (Geomatec g.moth 5021) was attached to the glass surface. The fabricated CsPbBr 3  photovoltaic power converters were characterized by I-V measurements under a solar simulator with an AM1.5 G spectrum with a light intensity of 100 mW cm −2 . The voltage was scanned from −0.3 V to 2.0 V in the 0.01 V step, and the measurement time at each voltage was 0.01 s. EQE measurements without white bias were also applied for a more detailed analysis of the devices. Both I-V and EQE measurements were performed at RT (25°C), but the device temperature during the I-V measurements was slightly higher than 25°C because a device cooling equipment was not installed in our measurement system. The performance of the devices under blue light with a wavelength of 450 nm was also estimated from the device performance under solar simulators and EQE.  (112). 24) This clearly indicates that the deposited GaN layers have a polycrystalline structure. When the substrate heater temperature is below 450°C, the (100) peak intensity is almost negligible, but the (100) peak intensity clearly increased at the substrate heater temperature above 450°C. This suggests that the structure of the GaN layers changed at a substrate heater temperature of 450°C. In Fig. 2(b), θ−2θ scan, only the (002) peak was observed. This means the preferential orientation of GaN is (002). Figure 3 shows the FTIR spectra of the GaN layers deposited on c-Si substrates. All GaN layers show an absorption peak at a wavenumber of around 555 cm −1 which corresponds to the E1(TO) mode of GaN. 25) The shape of the peak was influenced by the substrate heater temperature. For substrate heater temperatures below 300°C the shape of the peak is almost Gaussian, but the peak shape is Lorentzian for the substrate heater temperatures above 400°C. This indicates that the crystallinity of the GaN layer was improved by substrate heating and this result is in good agreement with the crystallinity shown in Fig. 2(b). 23) The cross-sectional TEM image of the GaN layer deposited at a substrate heater temperature of 550°C is shown in Fig. 4. Columnar grains with a lateral size of about 10 nm were clearly observed in the TEM image. In addition, the crystalline phase was observed at the Eagle-XG glass/GaN layer interface, indicating that the GaN layer deposited at 550°C is a highly crystallized nc-GaN layer. According to the XRD and FTIR characterization, it was found that higher substrate heater temperature leads to higher crystallinity. All deposited GaN layers are nc-GaN. We also investigated the impurities in the nc-GaN layers by SIMS measurements. Figure 5 shows the SIMS depth profiles of oxygen, carbon, hydrogen, and silicon in the nc-GaN layers deposited at 400°C and 550°C. The nc-GaN layers were deposited on Eagle-XG glass substrates. In both layers, a high concentration of silicon of about 2 × 10 20 cm −3 was observed although the sputtering target used in this study was undoped GaN. One of the reasons for this unintentional doping of silicon is cross contamination since the deposition of amorphous silicon layers was also performed in the same sputtering chamber. Oxygen concentration in the nc-GaN layer was almost a similar value of about 2 × 10 21 cm −3 for both layers. Carbon and hydrogen concentrations in the nc-GaN layer deposited at 550°C are much smaller than those in the nc-GaN layer deposited at 400°C. The candidates of hydrogen and carbon sources in the sputtering process are the remaining H 2 O vapor and hydrocarbon in the sputtering chamber. Higher substrate temperature prevents the absorption of these molecules or radicals generated by the decomposition of these molecules, leading to the small concentration of carbon and hydrogen in the nc-GaN layer deposited at 550°C.
The electrical properties of the nc-GaN layers were investigated by AC Hall measurements. The dependence of conductivity, carrier concentration, and Hall mobility of the nc-GaN layers on the substrate heater temperature is summarized in Fig. 6. The conductivity of the nc-GaN layers deposited at substrate heater temperatures below 400°C is about 10 −2 S cm −1 . This small conductivity is mainly due to the small Hall mobility of the order of 10 −3 to 10 −2 cm −2 /V/s. When the substrate heater temperature increased above 450°C, the Hall  The Japan Society of Applied Physics by IOP Publishing Ltd mobility increased up to 0.3 cm −2 V -1 s -1 and gradually increased with increasing the substrate heater temperature. According to the XRD results shown in Fig. 2(b), the tendency of the Hall mobility is very similar to the tendency of the peak intensity of GaN (002). Therefore, the improvement of the Hall mobility is mainly due to the improvement of the crystallinity of the nc-GaN layers. The Hall measurements also showed that all nc-GaN layers showed n-type conductivity. The carrier concentration (electron concentration) is relatively high values above 2 × 10 18 cm −3 although we used an undoped GaN sputtering target for the sputtering. At the optimum substrate heater temperature (500°C-550°C), the carrier concentration reaches about 8 × 10 19 cm −3 . This very high carrier concentration is probably due to the unintentional doping of silicon and oxygen as mentioned in the explanations of the SIMS measurements. Silicon and oxygen are well-known n-type dopants in GaN. [26][27][28] As shown in Fig. 5, the concentration of the unintentionally doped silicon is about 2 × 10 20 cm −3 , therefore, the experimentally obtained carrier concentration of about 8 × 10 19 cm −3 can be realized if the unintentionally doped silicon and oxygen act as donors. The smaller carrier concentration for the nc-GaN layer deposited below 450°C is probably due to the high carbon concentration in the nc-GaN layer. It is known that carbon creates substitutional deep acceptor (C N ) in GaN codoped with silicon and carbon by substituting nitrogen site in GaN. The carbon addition into silicon-doped GaN leads to a decrease in the electron concentration. 29) We also observed that the reduction of carrier concentration at a substrate heater temperature of 400°C. At present, it is difficult to explain this behavior clearly, but there is a possibility that this is affected by the change in the crystal orientation of nc-GaN. As shown in Fig. 2(b), GIXRD patterns and crystal orientation were changed at the substrate heater temperature between 400°C to 450°C. More detailed analysis is required to clarify this behavior.

Characterization of CsPbBr 3 photovoltaic power converter
CsPbBr 3 photovoltaic power converters were fabricated to investigate the potential of nc-GaN ETL. Figure 7 shows the I-V curves of the devices with nc-GaN ETL deposited at different substrate heater temperatures. The device parameters are summarized in Table I. The devices with nc-GaN ETL deposited at 550°C and 600°C showed good photovoltaic performance. For comparison, we also fabricated a CsPbBr 3 photovoltaic power converter with TiO 2 ETL (without AR film on the glass). The device with TiO 2 ETL showed better  The Japan Society of Applied Physics by IOP Publishing Ltd performance than that with nc-GaN ETL. The reason for the poor performance compared with the device with TiO 2 ETL and performance improvement using nc-GaN deposited at high substrate heater temperatures can be explained by the effect of the FTO/nc-GaN interface. As shown in Fig. 1, the use of nc-GaN can reduce the ΔE c at ETL/CsPbBr 3 interface, however, ΔE c at FTO/ETL interface is significantly increased. In this case, the electron transport barrier at the FTO/ETL interface strongly affects the device performance. Therefore, the device performance for nc-GaN ETL is inferior compared with TiO 2 ETL. When the carrier concentration in the nc-GaN ETL increases, the energy barrier width at the FTO/nc-GaN interface decreases, leading to the tunneling carrier transport being accelerated. This is a possible explanation for the improvement of the device performance when we used the nc-GaN deposited at high substrate heater temperatures. Finally, we would like to discuss the potential of the blue light photovoltaic power converter. Figure 8(a) shows the EQE spectra of the CsPbBr 3 photovoltaic power converters. It should be noted that the EQE spectra were corrected to match the J SC obtained from the I-V measurements since the calculated J SC from the raw EQE spectra and the AM1.5 G spectrum differed from the J SC obtained from the I-V measurement. This difference between the two J SC was probably due to the EQE measurement without white bias light. The difference between the device with nc-GaN ETL and TiO 2 ETL in the short wavelength region (300-400 nm) is the existence of the AR film on the glass surface. The AR film was only applied for the devices with nc-GaN ETL. The AR film absorbs short wavelength light, but it reduces the reflectance in the visible light region as shown in Fig. 8(b). Therefore, the devices with nc-GaN ETL showed higher EQE than that of the device with TiO 2 ETL for the wavelength region from 400 to 500 nm. As explained in our previous publication, 11) the conversion efficiency (η) of a photovoltaic power converter (solar cell) under monochromatic illumination can be calculated by using the following equation.  Table I. The best-estimated conversion efficiency for blue light is 28.8% for the device with nc-GaN deposited at 550°C. A slightly higher conversion efficiency for blue light of 29.7% was also obtained from the device with TiO 2 ETL. Although the band alignment between ETL and CsPbBr 3 is better for nc-GaN ETL, the obtained device performance is still inferior to the reference device with TiO 2 ETL. Further investigation of the FTO/ETL interface, e. g. insertion of a buffer layer or replacement of FTO with different transparent conductive oxide with a smaller work function, [30][31][32] is required to demonstrate the full potential of the nc-GaN ETL. The estimated conversion efficiency under blue light is about 30% is very promising. According to the device performance of the best GaAs solar cell (V OC = 1.1272 V, FF = 0.867) 33) and EQE at 450 nm of 0.9, 34) we can also estimate the conversion efficiency of about 31% (under the same photon flux as AM1.5 G illumination with 100 mW cm −2 ) for GaAs photovoltaic power converter under the blue light. Although the blue light conversion efficiency of the GaAs device is still slightly higher than our CsPbBr 3 devices, there is a large room for improvement of our CsPbBr 3 devices different from the GaAs device. Therefore, our results show the potential of CsPbBr 3 photovoltaic power converters.

Conclusions
We investigated the properties of nc-GaN deposited by RF magnetron sputtering for ETL application of CsPbBr 3 photovoltaic power converters. Highly crystallized nc-GaN layers were obtained at substrate heater temperatures above 450°C. These highly crystallized nc-GaN layers showed good . It should be noted that the AR film was used only for the devices with nc-GaN ETL. These I-V curves correspond to the reverse scan. The V OC and FF values for the forward scan were approximately 80% and 60% of the V OC and FF values for the reverse scan. electrical conductivity of about 10 S cm −1 with an electron concentration of 7.55 × 10 19 cm −3 and Hall mobility of 0.8 cm 2 V -1 s -1 . The nc-GaN layers were applied to the ETL in CsPbBr 3 photovoltaic power converters. The device with the nc-GaN ETL deposited at a substrate heater temperature of 550°C shows the best conversion efficiency of 5.56% under AM1.5 G illumination with irradiation of 100 mW cm −2 . The conversion efficiency under blue light (450 nm) was also estimated to be 28.8% by using device performance under AM1.5 G illumination and EQE at 450 nm. This value is still slightly smaller than that of the reference device with TiO 2 ETL. This is due to the conduction band offset between FTO/ nc-GaN. For more improvement in the performance of CsPbBr 3 photovoltaic power converters, it is important to optimize the FTO/ETL interface as well as the ETL/CsPbBr 3 interface.