All-Inorganic Perovskite Solar Cells Based on CsPbIBr2 and Metal Oxide Transport Layers with Improved Stability

Despite the successful improvement in the power conversion efficiency (PCE) of perovskite solar cells (PSCs), the issue of instability is still a serious challenge for their commercial application. The issue of the PSCs mainly originates from the decomposition of the organic–inorganic hybrid perovskite materials, which will degrade upon humidity and suffer from the thermal environment. In addition, the charge transport layers also influence the stability of the whole devices. In this study, inorganic transport layers are utilized in an inverted structure of PSCs employing CsPbIBr2 as light absorbent layer, in which nickel oxide (NiOx) and cerium oxide (CeOx) films are applied as the hole transport layer (HTL) and the electron transport layer (ETL), respectively. The inorganic transport layers are expected to protect the CsPbIBr2 film from the contact of moisture and react with the metal electrode, thus preventing degradation. The PSC with all inorganic components, inorganic perovskite and inorganic transport layers demonstrates an initial PCE of 5.60% and retains 5.56% after 600 s in ambient air at maximum power point tracking.


Introduction
The solar cells adopting organic-inorganic hybrid perovskite as absorber have obtained significant attention in the field of photovoltaics, with an unprecedented rise in power conversion efficiency (PCE) from 3.8% to 25.2% in a few years [1]. Although the highest certified efficiency of the PSCs is comparable to that of commercialized silicon photovoltaic devices, the issue of instability of PSCs has prevented its practical application. In particular, the volatile and hygroscopic nature of the organic cations such as methyl ammonium and formamidinium makes the organic-inorganic hybrid perovskite materials become more unstable under illumination, thermal, moisture and stresses [2,3].
In recent years, cesium lead halide or mixed halide perovskites CsPbX 3 have attracted much attention as light absorbent layer materials due to their excellent thermal stability as well as suitable photophysical properties [4][5][6]. In particular, the CsPbIBr 2 have emerged as a better candidate among its cousins, because of its capacity to balance the bandgap and its thermal stability [7], which has a bandgap of 2.05 eV, promising a potential highest PCE about 17.5% according to the Shockley-Queisser limit [8] and the material is unexpectedly stable with a melting point more than 460 • C under ambient atmosphere [9]. The higher performance PSCs usually have a sandwich structure, in which the perovskite active layer is placed between a hole transport layer (HTL) and an electron transport layer (ETL). The transport layers not only promote the transportation of the photogenerated carriers produced in the perovskite active layer and prevent carrier recombination at the interfaces, but also form ohmic contact with the electrode. Typically, in a n-i-p structure, the perovskite materials usually are deposited on a compact TiO 2 layer and an optional mesoporous TiO 2 scaffold layer [10], and covered with a layer of doping 2,2 ,7,7 -tetrakis(N,N-p-dimethoxyphenylamino)-9,9 -spirobifluorene (Spiro-OMeTAD) sequentially. As a counterpart, the p-i-n structure, perovskite materials are deposited onto the film of poly(3,4-ethylene dioxythiophene):polystyrene sulfonic acid (PEDOT:PSS) or poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine] (PTAA ) and then covered with a layer of [6,6]-phenyl-C61-butyric acid methyl ester (PCBM) or C 60 . However, PCBM and Spiro-OMeTAD are expensive and the doping agents of Spiro-OMeTAD could break down the electrode and perovskite layers gradually. To achieve novel PSCs with long-term stability and low-cost of fabrication, the strategies of replacing the organic transport layers with inorganic materials such as CuSCN [11], CuI [12], NiO x [13] as the hole transport layers and ZnO [14], CdS [15], SnO 2 [16] as the electron transport layers has been extensively explored. The inorganic metal oxides have much higher carrier mobility and superior stability as well as easy-processing from corresponding precursors. Herein, we use cerium oxide (CeO x ) as electron transport layers with in situ decomposition from its precursor at low temperature (100 • C) and nickel oxide (NiO x ) as hole transport layers from pre-prepared NiO x nanoparticles solution as well as CsPbIBr 2 perovskite as the photoactive layer in the device. The prepared perovskite solar cells with a configuration of ITO/NiO x /CsPbIBr 2 /CeO x /Ag based on an all-metal-oxide charge transport layer and the all-inorganic perovskite material shows 5.60% efficiency and significantly improved stability compared to those reported [17,18].

Preparation of NiO x Nanoparticles
The NiO x nanoparticles were prepared according reported lecture with a little modification [19]. To be specific, A clear green solution of nickel (II) salt was obtained by dissolving 14.55 g (0.05 mol) Ni(NO 3 ) 2 ·6H 2 O in 100 mL deionized water with vigorous stirring. Then a NaOH aqueous solution of 10 mol L −1 was added to the solution until the pH value reached 10. After another 10 min' stirring, the obtained green precipitants through suction filtration were transferred to vacuum oven and dried 6 hours at 80 • C under reduced pressure. The NiO x nanoparticles were prepared as dark gray powders by calcining the dried precipitants at 270 • C for 2 h.
The perovskite (CsPbIBr 2 ) precursor solution was prepared by mixing 2.34 g cesium iodide (CsI) and 3.30 g lead bromide (PbBr 2 ) with equal molar ratio in 10 mL dimethyl sulfoxide (DMSO) with stirring at 60 • C until all the solid dissolved completely.

Device Fabrication
The pre-patterned indium tin oxide (ITO) glass was cleaned in ultrasonic sequential with detergent, deionized water, acetone, and isopropanol for 15 min, dried in N 2 , followed by treatment of plasma for 10 min. The film of NiO x was deposited on the ITO substrate using 0.5 mg mL −1 NiO x nanoparticles aqueous solution by spin coating at 2500 r.p.m. for 60 s, and annealed at 150 C for 30 min in air. Then, the perovskite precursor solution was spin-coated on the NiO x film at 2500 r.p.m. for 53 s and annealed on a hot plate at 100 • C for 20 min. The ETL of CeO x was fabricated on the top of the perovskite layer by spin-coating the precursor at 2000 r.p.m. for 30 s and annealed in N 2 at 100 • C for 10 min. Finally, 100 nm Ag was thermally evaporated on the ETL as the top metal electrode under a pressure less than 5 × 10 −4 Pa.

Characterization
The top-view and cross-section morphologies of perovskite films of PSC was characterized using a field-emission scanning electron microscope (FESEM, Quanta 200 FEG, FEI Co.). Ultraviolet photoelectron spectroscopy (UPS) measurements were measured on a ThermoFisher ESCALAB 250Xi instrument using the HeI (21.22 eV) emission line. Water contact angle on these films were recorded with JC000DI contact angle measuring instrument (Zhong Chen. Shanghai, China). Atomic force microscopy (AFM) images were obtained using a Veeco MultimodeV instrument to evaluate the surface morphology of films in the tapping and intelligent mode. The steady state photoluminescence spectra (PL) and time-resolved photoluminescence spectra (TRPL) were acquired on an FS5 spectrometer from Edinburgh Instruments excited with 500 W Xenon lamp and 5 mW picoseconds pulsed diode laser at 410 ± 10 nm. The solar cells were measured using simulated air mass 1.5 global sunlight (AM 1.5G) conditions (100 mW cm −2 ) with 2400 Source Meter, Keithley Instruments. The external quantum efficiency (EQE) of the solar cells were measured using a combined system, including a xenon lamp, monochromator, chopper and lock-in amplifier together with a calibrated silicon photodetector. Perovskite films' ultraviolet-visible (UV-Vis) absorption measurements were performed through a Shimadzu UV-3600 spectrometer.

Results
NiO x was applied in this work as HTL. The optical and electric properties were measured with UV-Vis spectra and ultraviolet photoelectron spectroscopy (UPS) (Figure 1). Nanomaterials 2019, 9, x FOR PEER REVIEW 3 of 10 The pre-patterned indium tin oxide (ITO) glass was cleaned in ultrasonic sequential with detergent, deionized water, acetone, and isopropanol for 15 min, dried in N2, followed by treatment of plasma for 10 min. The film of NiOx was deposited on the ITO substrate using 0.5 mg mL −1 NiOx nanoparticles aqueous solution by spin coating at 2500 r.p.m. for 60 s, and annealed at 150 C for 30 min in air. Then, the perovskite precursor solution was spin-coated on the NiOx film at 2500 r.p.m. for 53 s and annealed on a hot plate at 100 °C for 20 min. The ETL of CeOx was fabricated on the top of the perovskite layer by spin-coating the precursor at 2000 r.p.m. for 30 s and annealed in N2 at 100 °C for 10 min. Finally, 100 nm Ag was thermally evaporated on the ETL as the top metal electrode under a pressure less than 5 × 10 −4 Pa.

Characterization
The top-view and cross-section morphologies of perovskite films of PSC was characterized using a field-emission scanning electron microscope (FESEM, Quanta 200 FEG, FEI Co.). Ultraviolet photoelectron spectroscopy (UPS) measurements were measured on a ThermoFisher ESCALAB 250Xi instrument using the HeI (21.22 eV) emission line. Water contact angle on these films were recorded with JC000DI contact angle measuring instrument (Zhong Chen. Shanghai, China). Atomic force microscopy (AFM) images were obtained using a Veeco MultimodeV instrument to evaluate the surface morphology of films in the tapping and intelligent mode. The steady state photoluminescence spectra (PL) and time-resolved photoluminescence spectra (TRPL) were acquired on an FS5 spectrometer from Edinburgh Instruments excited with 500 W Xenon lamp and 5 mW picoseconds pulsed diode laser at 410 ± 10 nm. The solar cells were measured using simulated air mass 1.5 global sunlight (AM 1.5G) conditions (100 mW cm −2 ) with 2400 Source Meter, Keithley Instruments. The external quantum efficiency (EQE) of the solar cells were measured using a combined system, including a xenon lamp, monochromator, chopper and lock-in amplifier together with a calibrated silicon photodetector. Perovskite films' ultraviolet-visible (UV-Vis) absorption measurements were performed through a Shimadzu UV-3600 spectrometer.

Results
NiOx was applied in this work as HTL. The optical and electric properties were measured with UV-Vis spectra and ultraviolet photoelectron spectroscopy (UPS) (Figure1). As shown in Figure 1a, the bandgap of NiOx film could be evaluated from the UV-Vis absorbance spectrometry. The direct bandgap energy (Eg) can be determined using the direct transition equation [22] ( ℎ ) 2 = (ℎ − ). The plot of (αhv) 2 versus hv is given in the inset of Figure 1a, and the curve is obtained by the M-K equation. Eg for the NiOx film is determined to be 3.7 eV. The UPS of the NiOx film is shown in Figure 1b, indicating the work function of the film. The calculated results of the valence band energy level as shown in Figure 1c is 5.42 eV, facilitating the extraction of holes from perovskite, while the conductive band energy level could be calculated as As shown in Figure 1a, the bandgap of NiO x film could be evaluated from the UV-Vis absorbance spectrometry. The direct bandgap energy (E g ) can be determined using the direct transition equation [22] (αhν) 2 = A hν − E g . The plot of (αhv) 2 versus hv is given in the inset of Figure 1a, and the curve is obtained by the M-K equation. E g for the NiO x film is determined to be 3.7 eV. The UPS of the NiO x film is shown in Figure 1b, indicating the work function of the film. The calculated results of the valence band energy level as shown in Figure 1c is 5.42 eV, facilitating the extraction of holes from perovskite, while the conductive band energy level could be calculated as 1.72 eV through the optical bandgap and the valence band energy level, which is higher than that of CsPbIBr 2 as shown in Figure 1c, blocking the injection of electrons.
Generally, the morphology of perovskite film can influence the performance of perovskite solar cells. The morphology of the CsPbIBr 2 films is characterized by the scanning electron microscopy (SEM) images ( Figure 2). The image of the top-view SEM of CsPbIBr 2 film on the top of the NiO x in Figure 2a exhibits a compact and smooth film on the top of the NiO x without any pin holes. Furthermore, the energy dispersive spectrometer EDS mapping of corresponding element Cs, Pb, I and Br are homogeneously distributed throughout the perovskite film without phase separation shown in Figure 2b, confirming the full coverage of CsPbIBr 2 film on the NiO x film, partially guaranteeing the high performance of the perovskite solar cells. X-ray diffraction (XRD) measurements were carried out to reveal the formation of CsPbIBr 2 perovskite (Figure 2c). The diffraction peaks in the XRD patterns were consistent with those of [23]. Nanomaterials 2019, 9, x FOR PEER REVIEW 4 of 10 1.72 eV through the optical bandgap and the valence band energy level, which is higher than that of CsPbIBr2 as shown in Figure 1c, blocking the injection of electrons. Generally, the morphology of perovskite film can influence the performance of perovskite solar cells. The morphology of the CsPbIBr2 films is characterized by the scanning electron microscopy (SEM) images ( Figure 2). The image of the top-view SEM of CsPbIBr2 film on the top of the NiOx in Figure 2a exhibits a compact and smooth film on the top of the NiOx without any pin holes. Furthermore, the energy dispersive spectrometer EDS mapping of corresponding element Cs, Pb, I and Br are homogeneously distributed throughout the perovskite film without phase separation shown in Figure 2b, confirming the full coverage of CsPbIBr2 film on the NiOx film, partially guaranteeing the high performance of the perovskite solar cells. X-ray diffraction (XRD) measurements were carried out to reveal the formation of CsPbIBr2 perovskite (Figure 2c). The diffraction peaks in the XRD patterns were consistent with those of [23]. The performance of device depends not only on the morphology of the film but also on the separation of exciton and the charge carrier's mobility through the interfaces. The charge transport layers widely are employed to enhance carrier injection, decrease trap-states, and reduce contact resistance [24]. In this work, another inorganic transport material, CeOx is used as the electron transport layer (ETL) in inverted CsPbIBr2 PSCs. In order to research how the energy level align, we measured the bandgap of the CeOx layer. The Eg of CeOx film could be obtained from the UV-Vis absorbance spectrometry in Figure 3a, and which is determined to be 3.5 eV. The UPS of the CeOx film is shown in Figure 3b, and the valence band (EVB) of the CeOx film is calculated to be 7.52 eV. According to the relationship of CB = VB + , the conductive band energy level (ECB) of CeOx is calculated as 4.02 eV. Compared to the conventional ETL (PCBM), CeOx shows a better energy level match with CsPbIBr2 due to it has a lower valence band energy level, the larger driving force will facilitate the election injection and blocks the holes. At the same time, the CeOx energy level of the conductive band is slightly lower than CsPbIBr2, proving CeOx can be used as a suitable ETL for CsPbIBr2 PSCs (Figure 3c).
The conductive band energy level of CsPbIBr2 and CeOx discontinuity (ΔEc) is 0.1 eV, the valence band energy level discontinuity is 1.52 eV, which means CeOx layer can actually be an electronic transfer tunnel when the energy bands bend downwards, decreasing the energy barrier from the conductive band of perovskite absorber (Figure 3d). To further demonstrate the electron extraction mechanism, the steady-state photoluminescence (PL) and TRPL spectra were measured and are shown in Figure 3e,f This shows that the CsPbIBr2/CeOx exhibit more efficient PL quenching ability, the significantly lower PL intensity indicates a stronger interfacial electron extraction The performance of device depends not only on the morphology of the film but also on the separation of exciton and the charge carrier's mobility through the interfaces. The charge transport layers widely are employed to enhance carrier injection, decrease trap-states, and reduce contact resistance [24]. In this work, another inorganic transport material, CeO x is used as the electron transport layer (ETL) in inverted CsPbIBr 2 PSCs. In order to research how the energy level align, we measured the bandgap of the CeO x layer. The E g of CeO x film could be obtained from the UV-Vis absorbance spectrometry in Figure 3a, and which is determined to be 3.5 eV. The UPS of the CeO x film is shown in Figure 3b, and the valence band (E VB ) of the CeO x film is calculated to be 7.52 eV. According to the relationship of E CB = E VB + E g , the conductive band energy level (E CB ) of CeO x is calculated as 4.02 eV. Compared to the conventional ETL (PCBM), CeO x shows a better energy level match with CsPbIBr 2 due to it has a lower valence band energy level, the larger driving force will facilitate the election injection and blocks the holes. At the same time, the CeO x energy level of the conductive band is slightly lower than CsPbIBr 2 , proving CeO x can be used as a suitable ETL for CsPbIBr 2 PSCs (Figure 3c).
bi-exponential decay function [27], and the PL decay time of glass/CsPbIBr2 are τ1 = 0.53 ns, τ2 = 3.07 ns. While that of glass/CsPbIBr2/CeOx are τ1 = 0.38 ns, τ2 = 2.58 ns. The reduction of the PL decay time indicates a more efficient electron extraction capability. Furthermore, the presence of CeOx can speed up the electron extract and increase the photovoltaic performance of the inorganic PSCs. The contact resistance strongly depends on adhesion of contact interface and the flatness of the contact interface. To estimate the roughness of the contact interface with and without depositing CeOx film, the AFM measurements were tested (Figure 4a,b). The root-mean-square (RMS) roughness of the pure perovskite was 19.17 nm, the RMS roughness of the CsPbIBr2/CeOx film was 11.86 nm, the reduced RMS roughness would be beneficial in decreasing the contact resistance after deposited CeOx layer. The water contact angle on these films is also tested (Figure 4c). The contact angle is 63.1° for the pure perovskite film, while 34.8° for the perovskite/CeOx film. A relatively good wetting property can have a better atomic adsorption compared to the non-wetting surfaces, it can also contribute to low electrical contact resistance [28]. The conductive band energy level of CsPbIBr 2 and CeO x discontinuity (∆E c ) is 0.1 eV, the valence band energy level discontinuity is 1.52 eV, which means CeO x layer can actually be an electronic transfer tunnel when the energy bands bend downwards, decreasing the energy barrier from the conductive band of perovskite absorber (Figure 3d). To further demonstrate the electron extraction mechanism, the steady-state photoluminescence (PL) and TRPL spectra were measured and are shown in Figure 3e,f This shows that the CsPbIBr 2 /CeO x exhibit more efficient PL quenching ability, the significantly lower PL intensity indicates a stronger interfacial electron extraction efficiency from perovskite absorber layer [25,26]. The two TRPL spectra are fitted with a bi-exponential decay function [27], and the PL decay time of glass/CsPbIBr 2 are τ 1 = 0.53 ns, τ 2 = 3.07 ns. While that of glass/CsPbIBr 2 /CeO x are τ 1 = 0.38 ns, τ 2 = 2.58 ns. The reduction of the PL decay time indicates a more efficient electron extraction capability. Furthermore, the presence of CeO x can speed up the electron extract and increase the photovoltaic performance of the inorganic PSCs.
The contact resistance strongly depends on adhesion of contact interface and the flatness of the contact interface. To estimate the roughness of the contact interface with and without depositing CeO x film, the AFM measurements were tested (Figure 4a,b). The root-mean-square (RMS) roughness of the pure perovskite was 19.17 nm, the RMS roughness of the CsPbIBr 2 /CeO x film was 11.86 nm, the reduced RMS roughness would be beneficial in decreasing the contact resistance after deposited CeO x layer. The water contact angle on these films is also tested (Figure 4c). The contact angle is 63.1 • for the pure perovskite film, while 34.8 • for the perovskite/CeO x film. A relatively good wetting property can have a better atomic adsorption compared to the non-wetting surfaces, it can also contribute to low electrical contact resistance [28].  To evaluate the influence of the metal oxide transport layers on the device performance, an inverted planar CsPbIBr2 PSCs with a configuration of ITO/NiOx/CsPbIBr2/CeOx/Ag were fabricated. The current-voltage (J-V) curve of the inorganic PSCs is shown in Figure 5a. The best power conversion efficiency (PCE) of 5.60% was obtained in the device with an open-circuit (Voc) of 1.01 V, a short-circuit current density (Jsc) of 8.76 mA cm −2 , a fill factor (FF) of 63.35%. The device also shows a small J-V hysteresis (Figure 5a). Figure 5b gives statistical PCEs of 24 independent cells. Average PCE is calculated to be 5.56% for all the inorganic PSCs.
The external quantum efficiency (EQE) spectrum of the device was measured and is shown in Figure 5c. The corresponding integrated Jsc of 8.71 mA cm −2 , which is close to the Jsc derived from the J-V curve. This result indicates that the device has a better photoelectric conversion property. The stabilized power output (SPO) of the PSCs is also shown in Figure 5d, and the stabilized efficiency output is estimated to 5.54% for the solar cell after being measured for over 600 s. In addition, the PCE output is much more stable during long time measurement, indicating the inorganic materials can have a better stability. To evaluate the influence of the metal oxide transport layers on the device performance, an inverted planar CsPbIBr 2 PSCs with a configuration of ITO/NiO x /CsPbIBr 2 /CeO x /Ag were fabricated. The current-voltage (J-V) curve of the inorganic PSCs is shown in Figure 5a. The best power conversion efficiency (PCE) of 5.60% was obtained in the device with an open-circuit (V oc ) of 1.01 V, a short-circuit current density (J sc ) of 8.76 mA cm −2 , a fill factor (FF) of 63.35%. The device also shows a small J-V hysteresis (Figure 5a). Figure 5b gives statistical PCEs of 24 independent cells. Average PCE is calculated to be 5.56% for all the inorganic PSCs.
The external quantum efficiency (EQE) spectrum of the device was measured and is shown in Figure 5c. The corresponding integrated J sc of 8.71 mA cm −2 , which is close to the J sc derived from the J-V curve. This result indicates that the device has a better photoelectric conversion property. The stabilized power output (SPO) of the PSCs is also shown in Figure 5d, and the stabilized efficiency output is estimated to 5.54% for the solar cell after being measured for over 600 s. In addition, the PCE output is much more stable during long time measurement, indicating the inorganic materials can have a better stability.
The electrical property of these inorganic materials is investigated to determine they are better adapted to organic materials used in the PSCs. The space-charge-limited current (SCLC) measurement has been tested. We measured the electron-only devices and these are shown in the Figure 6a,b. The SCLC can be divided into three sections: ohmic regime, trap-filling regime and trap-free SCLC regime. The trap-state density (N trap ) can be determined by the trap-filled limit voltage equation [29,30], and the N trap of the PCBM and CeO x ETL are 6.95 × 10 16 and 3.89 × 10 16 cm −3 , respectively, indicating the CeO x as the ETL can reduce the trap density. The mobility of the devices is calculated using Equation [31] J = 9ε 0 ε r µV 2 /8L 2 , the mobility values of these devices are 0.39 cm 2 V −1 S −1 and 1.26 cm 2 V −1 S −1 , respectively. This result implies the device with the CeO x exhibits more efficient electron extraction from the perovskite layer. The electrical property of these inorganic materials is investigated to determine they are better adapted to organic materials used in the PSCs. The space-charge-limited current (SCLC) measurement has been tested. We measured the electron-only devices and these are shown in the Figure 6a and b. The SCLC can be divided into three sections: ohmic regime, trap-filling regime and trap-free SCLC regime. The trap-state density (Ntrap) can be determined by the trap-filled limit voltage equation [29,30], and the Ntrap of the PCBM and CeOx ETL are 6.95 × 10 16 and 3.89 × 10 16 cm −3 , respectively, indicating the CeOx as the ETL can reduce the trap density. The mobility of the devices is calculated using Equation [31] = 9 0 2 8 2 ⁄ , the mobility values of these devices are 0.39 cm 2 V −1 S −1 and 1.26 cm 2 V −1 S −1 , respectively. This result implies the device with the CeOx exhibits more efficient electron extraction from the perovskite layer.
To further investigate the carrier transfer and recombination in the inorganic PSC, the transient photovoltage (TPV) and transient photocurrent (TPC) measurements are carried out. Figure 6d shows the photocurrent decay of the different ETL devices measured at short current condition. The CeOx-based device has faster decay than the PCBM-based device, suggesting that the CeOx-based device has a better carrier extraction and transport. The TPV is carried out to measure the charge recombination lifetime (Figure 6c). The devices with CeOx ETL demonstrates a slower photovoltage decay, indicating the charge recombination can be suppressed and lower trap density. This situation is consistent with the PL and SCLC results that improve the charge transfer and decrease the trap states in the all inorganic PSCs. Finally, the stabilities of the devices are examined. All the devices are evaluated and the results are shown in Figure 7. These devices are stored in an atmospheric condition at 40-45 °C and humidity in the 45-50% without encapsulation. To further investigate the carrier transfer and recombination in the inorganic PSC, the transient photovoltage (TPV) and transient photocurrent (TPC) measurements are carried out. Figure 6d shows the photocurrent decay of the different ETL devices measured at short current condition. The CeO x -based device has faster decay than the PCBM-based device, suggesting that the CeO x -based device has a better carrier extraction and transport. The TPV is carried out to measure the charge recombination lifetime (Figure 6c). The devices with CeO x ETL demonstrates a slower photovoltage decay, indicating the charge recombination can be suppressed and lower trap density. This situation is consistent with the PL and SCLC results that improve the charge transfer and decrease the trap states in the all inorganic PSCs.
Finally, the stabilities of the devices are examined. All the devices are evaluated and the results are shown in Figure 7. These devices are stored in an atmospheric condition at 40-45 • C and humidity in the 45-50% without encapsulation.  It can be seen that the PCE of the device with the CeOx ETL can retain 90% of its initial efficiency after 500 h. Generally, the water can damage the perovskite layer and cause decomposition [32]. However, the CeOx film can protect the perovskite layer from water and has more stability in air. According to the previous report, the organic ETL is not dense enough against the water and makes the perovskite layer degenerate. In contract, the inorganic ETL can prevent the perovskite layer from water and act as a diffusion barrier to prevent the metal electrode decomposing. It can be seen that the PCE of the device with the CeO x ETL can retain 90% of its initial efficiency after 500 h. Generally, the water can damage the perovskite layer and cause decomposition [32]. However, the CeO x film can protect the perovskite layer from water and has more stability in air. According to the previous report, the organic ETL is not dense enough against the water and makes the perovskite layer degenerate. In contract, the inorganic ETL can prevent the perovskite layer from water and act as a diffusion barrier to prevent the metal electrode decomposing.

Conclusions
In summary, cost-effective, efficient p-i-n PSCs were demonstrated with an inorganic perovskite layer sandwiched between two metal oxides transport layers. The inorganic transport layers retarded the degradation of the perovskite film and decreased the contact resistance, trap density between the interfaces of the CsPbIBr 2 film and the Ag electrode effectively. Furthermore, the suitable energy alignments of the layers increased the performance of the PSCs with a V oc of 1.01 V and J sc of 8.76 mA cm −2 . The best PCE of the devices is of 5.60% with a stabilized value of 5.56%. In addition, the high-quality all inorganic PSC show a long-term endurance against the influences of heat and humidity. Therefore, the all inorganic materials involved in PSCs provide a novel approach to fabricate future cheap and stable perovskite solar cells with an expectation of practicability.