Interfacial Voids Trigger Carbon-Based, All-Inorganic CsPbIBr2 Perovskite Solar Cells with Photovoltage Exceeding 1.33 V

Highlights A novel interface design of producing interfacial voids is proposed for CsPbIBr2 perovskite solar cells (PSCs), which is free of any extra modification layer. Interfacial voids improve absorption of CsPbIBr2 film, reduce saturation current density, and enlarge built-in potential of the PSCs. The PSC yields a superior efficiency of 10.20% with a record-high photovoltage of 1.338 V. Electronic supplementary material The online version of this article (10.1007/s40820-020-00425-1) contains supplementary material, which is available to authorized users.


Introduction
Organic-inorganic hybrid perovskite semiconductors have emerged as promising absorber materials for solar cells that are known as perovskite solar cells (PSCs), owing to their excellent optoelectronic properties, low cost, and versatile low-temperature processability of thin films [1][2][3][4]. Furthermore, the certified power conversion efficiency (PCE) of PSCs has reached up to 25.2% [5], approaching that of commercial crystalline silicon solar cells. However, the poor stability of organic-inorganic hybrid perovskite films under thermal, humidity, and light stresses brings a severe reliability issue to PSCs [3,4,6]. In addition, most of highefficiency PSCs generally contain expensive and unstable organic hole transport layers (HTLs) coupled with noble metal electrodes. These functional films not only aggravate the instability problem but also burden the production cost of corresponding PSCs [3,6,7]. Hence, there is a growing attention to carbon-based, all-inorganic PSCs that avoid use of any HTLs and noble metal electrodes, in view of their multiply superiorities including upgraded stability, costeffectiveness, and facile fabrication [7][8][9][10][11].
To solve the above obstacles, great efforts have been dedicated in improving the quality of CsPbIBr 2 films, targeting to suppress non-radiative recombination of charge carriers, by crystallization engineering or heteroatom replacement on Cs site or Pb site [19,20,25,31]. For example, our previous work [19] demonstrated that modification of general one-step method with an intermolecular exchange reaction could produce a pinhole-free CsPbIBr 2 film composed of micro-sized, [100] preferential grains, which finally enabled an optimized PCE of 9.16%. More recently, it was reported that Ba(II) doping could boost the grain sizes and crystallinity of CsPbIBr 2 film, bringing out reduced trap states and suppressed non-radiative recombination. As a result, the PCE of corresponding PSC was boosted from 8.40 to 10.51% [31].
Interfacial modification with extra functional materials has been attempted to weaken interfacial recombination of charge carriers and therefore improve the PCEs of CsPbIBr 2 PSCs [17,18,30,32]. For example, the innovative modification of TiO 2 electron transport layer (ETL) with CsBr or SmBr 3 could strikingly improve the PSC's PCE to over 10% coupled with desirable FF of ~ 0.70, which represents the highest PCE among CsPbIBr 2 PSCs reported to date [18,30]. Guo et al. [17] reported that surficial passivation of SnO 2 ETL could enable the CsPbIBr 2 PSCs with a recordhigh V oc of 1.31 V in spite of its low PCE of 2.06%. Clearly, these progresses verify the considerable feasibility of interfacial modification strategy in overcoming the obstacle faced by carbon-based, all-inorganic CsPbIBr 2 PSCs. However, the extra interfacial modifiers inevitably complex the configuration of CsPbIBr 2 PSCs and thus bring some new parasitic issues. For example, the modifiers multiply the interfaces of CsPbIBr 2 PSCs, leading to an increased degradation risk [3,33,34]. Meanwhile, they will also alter the crystallization dynamics of atop CsPbIBr 2 films [35,36], which brings a new challenge to prepare high-quality CsPbIBr 2 film. Thus, it is highly desired to develop more feasible interfacial engineering strategy to boost interfacial kinetics of charge carriers in carbon-based, all-inorganic CsPbIBr 2 PSCs, thereby boosting its V oc and FF as well as PCE.
In this study, we propose an interfacial engineering strategy for carbon-based, all-inorganic CsPbIBr 2 PSCs by making interfacial voids between TiO 2 ETL and CsPbIBr 2 film. This strategy avoids introducing any extra modification layer in ultimate PSCs compared with the conventional ones. Further, we reveal that the interfacial voids can stimulate enhanced absorption, decreased interfacial recombination of charge carriers, as well as enlarged built-in potential (V bi ) in resulting CsPbIBr 2 PSCs. Hence, the optimized one yields the outstanding PCE of 10.20% coupled with a high V oc of 1.338 V, a FF of 0.65, and a short-circuit current density (J sc ) of 11.73 mA cm −2 . The PCE is beyond nearly most of 1 3 CsPbIBr 2 -based cells reported earlier, and particularly the V oc represents the highest level among them. Moreover, the PSC exhibits excellent photo-stability as well as long-term stability against thermal and humidity stresses.

Preparation of PEAI-Loaded FTO/TiO 2 Substrate
Firstly, patterned FTO glass with size 2 × 2.5 cm 2 was washed with detergent, ultrapure water, acetone, and ethanol in an ultrasonic cleaner successively. Next, TiO 2 sol was spin coated onto the FTO glass at 3500 rpm for 30 s. After being annealed at 500 °C for 60 min in ambient air, the FTO/TiO 2 substrate was achieved. Finally, 100 μL 3 mg mL −1 PEAI in isometric deionized water and isopropanol was spin-coated on the substrate at 3500 rpm for 60 s and baked at 80 °C for 5 min in ambient air in order to load PEAI species onto FTO/TiO 2 substrate.

Deposition of CsPbIBr 2 Film
CsPbIBr 2 precursor (1.0 M) was firstly prepared by dissolving 370.0 mg PbBr 2 and 260.0 mg CsI in 1 mL DMSO at room temperature with vigorously stirring. Then, the PEAIloaded FTO/TiO 2 substrate was transferred into a N 2 -filled glovebox. And, 70 μL CsPbIBr 2 precursor was spin-coated the substrate at 1500 rpm for 20 s and 5000 rpm for 60 s. After the spin-coating process, the sample was annealed at 280 °C for 15 min to form the crystallized CsPbIBr 2 film. For comparison, a CsPbIBr 2 film on FTO/TiO 2 substrate without loading of PEAI species was also prepared according to the same procedures above.

Fabrication of Carbon-Based, All-Inorganic PSC
Carbon paste with the area of 0.10 cm 2 was firstly coated on CsPbIBr 2 film by screen-printing method. Then, the sample was heated at 120 °C for 15 min to promote evaporation of solvent of carbon paste. After cooling down naturally to room temperature, the carbon-based, all-inorganic PSC was achieved for further characterizations.

Formation Mechanism of CsPbIBr 2 Films with Interfacial Voids
The main preparation procedures of CsPbIBr 2 film with interfacial voids are shown in Scheme 1. To be brief, FTO/ TiO 2 substrate was firstly prepared according to the recipe described in our previous work [18,19]. Then, 2-phenylethylammonium iodide (PEAI, C 6 H 5 CH 2 CH 2 NH 3 I) in isometric deionized water and isopropanol was deposited on  Fig. S1. Next, CsPbIBr 2 precursor was spin-coated onto the substrate by one-step spincoating method. Finally, the sample was annealed at 280 °C for 15 min to promote the crystallization of CsPbIBr 2 grains. It is noted that the release of PEAI species happened simultaneously during this high-temperature annealing stage, as discussed below.
We conducted X-ray photoelectron spectroscopy (XPS) measurement to examine the constituents of obtained CsPbIBr 2 film. For comparison, a same CsPbIBr 2 film that was prepared at a low annealing temperature of 100 °C was also investigated. The reason for adopting this low annealing temperature to prepare the control sample is that it cannot cause the sublimation or decomposition of PEAI molecules, as demonstrated previously [37,38]. Figure 1a manifests that clear XPS peaks corresponding to Cs, Pb, I, and Br elements appear for the films annealed at either 280 or 100 °C, indicating that they are composed of CsPbIBr 2 materials. However, no any signal corresponding to PEAI species can be measured from them, even though the FTO/TiO 2 substrate contains PEAI species. The core-level N 1s XPS spectra in Fig. 1b further verify this fact, since the typical N 1s peak of PEAI species is absent [38].
In general, XPS measurement can only reflect surficial composition information of samples. To insight bulk composition of the concerned CsPbIBr 2 films, they were redissolved in situ by N,N-dimethylformamide (DMF) solvent and annealed again at 90 °C for 10 min to evaporate the solvent away. As shown in Fig. 1a, after such treatment the additional Sn 3d peaks that originate from FTO/TiO 2 substrate can be detected from both films, indicating the full dissolution of CsPbIBr 2 films by DMF solvent and therefore the possible non-uniform composition distribution in the films or their interfaces is eliminated. In this manner, XPS measurement can be used to study bulk composition of CsPbIBr 2 films reasonably. Once again, as exhibited in Fig. 1b, the N 1s peak located at 401.8 eV can be detected from CsPbIBr 2 film prepared at 100 °C, manifesting the existence of PEAI species in it [37,38]. By contrast, the peak is still missing in CsPbIBr 2 film prepared at 280 °C, indicating that there is no detectable PEAI species in it indeed. This can  be also supported by the energy-dispersive spectrometer (EDS) result in Fig. S2. We attribute such phenomenon to the sublimation of PEAI species during the high-temperature annealing stage of CsPbIBr 2 film.
To support our standpoint, thermal property of PEAI powder was investigated by thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) [39,40]. The TGA curve in Fig. 1c indicates that PEAI powder undergoes 100% weight loss in one step with the onset temperature of ~ 185 °C, suggesting that it is sublimated without complex decomposition procedure. This can be supported by the sharp transition step in corresponding differential TGA curve in Fig. S3. Moreover, the endothermic peak at ~ 283 °C in the DSC curve in Fig. 1d also suggests that the release of PEAI is related to a sublimation process. Thus, the typical sublimation temperature of PEAI powder is estimated to be ~ 283 °C, which is nearly the same as the annealing temperature of CsPbIBr 2 film adopted in this work. Additionally, it is noted that the DSC measurement was performed on PEAI powder. In the case of within CsPbIBr 2 thin film, a much larger surface-to-volume ratio is provided; therefore, the sublimation temperature could be lowered considerably to below ~ 283 °C for PEAI. Thus far, all the above observations provide conclusive evidences that in terms of CsPbIBr 2 film prepared at 280 °C the original PEAI species on FTO/TiO 2 substrate do not reside in its bulk or at its interface, but are sublimated during the thermal annealing stage.

Morphology Characterization
With this fact in mind, we turned to study the microstructure of CsPbIBr 2 films deposited on FTO/TiO 2 substrates without and with loading of PEAI species. Their top-view SEM images are provided in Fig. 2a, b. Clearly, both films are composed of closely packed crystalline grains without any discernible pinholes. Their grains are featured by dominant triple-junctions grain boundaries with equal vertex angle of 120°, implying that they are in equilibrium state with low internal stress [41,42]. By contrast, the CsPbIBr 2 film deposited on FTO/TiO 2 substrate loaded with PEAI species exhibits larger grain sizes. The statistical results in Fig. S4 indicate its average grain size of ~ 576 nm, which is slightly larger than the one deposited on FTO/TiO 2 substrate without loading of PEAI species (~ 475 nm). It should be noted that the formation of coarsened grains is mainly attributed to the decreased nucleation density of CsPbIBr 2 grains, as a result of improved non-wetting of CsPbIBr 2 precursor to FTO/TiO 2 substrate induced by PEAI species (Fig. S5) [36,43], along with the templating effect of PEAI species during the crystallization of CsPbIBr 2 grains.
Figures 2c, d and S6 provide cross-sectional SEM images of the concerned CsPbIBr 2 films. They reveal that the films have a similar thickness of ~ 400 nm. And, each of individual grains in the films is able to vertically penetrate entire film thickness. Such feature maybe endow excellent transport behavior of charge carriers along them, since the transport and transfer of charge carriers do so without traversing any grain boundaries in this case [4,41]. By comparison, interfacial microstructure of these two films differs obviously. For the CsPbIBr 2 film deposited on FTO/TiO 2 substrate without loading PEAI species, the interface between CsPbIBr 2 film and FTO/TiO 2 substrate is contacted seamlessly. However, there are many voids that have the typical size of ~ 100 nm in the same interface, in terms of CsPbIBr 2 film deposited on FTO/TiO 2 substrate loaded with PEAI species. These voids give rise to the rough contact interface between CsPbIBr 2 film and FTO/TiO 2 substrate. It is also noted that because PEAI species are sublimated during the thermal annealing stage of CsPbIBr 2 film, we conclude that the pre-deposited PEAI species likely act as pore-forming agent in the present case. The strategy of using pore-forming agent has been widely used to create pores in traditional ceramic materials [44,45]. Our viewpoint can be further supported by the cross section SEM image of the same CsPbIBr 2 film prepared with large concentration of PEAI solution (10 mg mL −1 ), as provided in Fig. S7. It is clear that the voids with much larger sizes are produced when the more PEAI species were predeposited. Overall, the PEAI species on FTO/TiO 2 substrate, serving as pore-forming agent, can promote the formation of large-sized grains in ultimate CsPbIBr 2 film as well as some voids at the interface between CsPbIBr 2 film and FTO/ TiO 2 substrate.

Crystalline and Optical Property Analysis
Next, the crystalline structure and optical properties of asprepared CsPbIBr 2 films are investigated. For clarification, the CsPbIBr 2 film as well the cell deposited on FTO/TiO 2 substrate without loading of PEAI species are labeled as "without void," and the ones on FTO/TiO 2 substrate with loading of PEAI species are named as "with voids" hereinafter. Figure 3a presents X-ray diffraction (XRD) patterns of CsPbIBr 2 films without and with interfacial voids. Both samples exhibit the pure cubic perovskite phase, since the diffraction peaks at 15.1°, 21.4°, 26.6°, 30.4° 33.8°, and 37.8° are well consistent with those of (100), (110), (111), (200), (210), and (211) planes of CsPbIBr 2 crystals reported previously [10,26]. Meanwhile, the XRD peaks of (100) and (200) planes are dominantly intense for both samples, indicating that the grains in them have (100)-preferential orientation. For cubic perovskite polycrystalline films, {100} grains facets are generally along the direction perpendicular to FTO/TiO 2 substrate, which well coincides the vertically penetrated grains morphology observed in the cros section SEM images in Figs. 2c, d and S6 for the concerned CsPbIBr 2 films. In addition, the dominant diffraction peaks of CsPbIBr 2 film with interfacial voids are more intense than the one without voids, and the full-width-at-half-maximum (FWHM) values of its (100) and (200) XRD peaks are smaller (Table S1). These results reveal the better crystallinity with fewer intragranular defects and less scattering of grain boundaries for the CsPbIBr 2 film with interfacial voids, in consideration of the similar thickness of the two films [22,41,43]. Such features are in accordance with the larger grains of CsPbIBr 2 film with interfacial voids. Figure 3b presents the ultraviolet-visible (UV-vis) spectra of as-prepared CsPbIBr 2 films. Both of them exhibit the typical absorption features of CsPbIBr 2 materials including the exciton absorption peak at ~ 560 nm and absorption edge of ~ 599 nm that is in accordance with the bandgap of CsPbIBr 2 (2.07 eV) [20,25,27]. In comparison, the CsPbIBr 2 film with interfacial voids shows higher absorption intensities in the whole absorption ranges, revealing its better light absorption capacity. In principle, this desired feature mainly originates from either improved quality of CsPbIBr 2 film or scattering effect of interfacial voids in it, in view of the similar thickness of the two films. To exclude the effect of the former, UV-vis characters of the concerned CsPbIBr 2 films were collected from the side of FTO glass, wherein the detection beam streams from the top surface of CsPbIBr 2 films. The results in Fig. S8 reveal that the differences of their absorption intensities decrease accordingly. That is to say, the better absorption ability of CsPbIBr 2 film with interfacial voids observed in Fig. 3b largely comes from the beneficial scattering effect of interfacial voids, which may make great contribution to J sc of ultimate PSC. Figure 3c, d gives steady-state photoluminescence (PL) spectra and time-resolved PL (TRPL) curves of CsPbIBr 2 films without and with interfacial voids grown on insulating glass substrates instead of FTO/TiO 2 substrates, respectively. As shown in Fig. 3c, CsPbIBr 2 films exhibit the dominant PL peaks at ~ 600 nm, in well agreement with their bandgaps. The shoulder peaks can be also observed for both samples, as indicated by the logarithmic-scale PL plots in Fig. S9. This phenomenon is mainly caused by light-induced halide phase segregation in CsPbIBr 2 films, wherein some I-rich domains with low bandgap and Br-rich domains with wide bandgap are formed under illumination, as demonstrated in previous works [19,[46][47][48]. In addition, the CsPbIBr 2 film with interfacial voids yields the stronger PL peak intensity, revealing the weaker non-radiative recombination of charge carriers in it. Figure 3d shows the corresponding TRPL results, which can provide semi-quantitative information about non-radiative recombination of charge carriers [2,4,9,48]. The curves are fitted by a bi-exponential decay function of time (t) [2,4]: F(t) = ∑ A i e ∕ i ; i = 1, 2 , in which A i and t i are weight factor and time constant, respectively. In this manner, the average carrier lifetime (t ave ) can be estimated The CsPbIBr 2 film with interfacial voids exhibits the τ ave of 1.91 ns, which is higher than the one without interfacial voids (1.26 ns). The longer τ ave of the former indicates the fewer defects in it, which is originally attributed to the larger grains and better crystallinity of CsPbIBr 2 film with interfacial voids.

Photovoltaic Performance
Hitherto, we reveal that the CsPbIBr 2 film with interfacial voids has slightly larger grains and higher crystallinity, which result in the weaker non-radiative recombination of charge carriers in it. Such feature is expected to improve the V oc and FF of ultimate PSC [18,30,32,49]. Meanwhile, the interfacial voids trigger intense light scattering effect [50], which gives rise to the much higher absorption of the CsPbIBr 2 film and hence maybe boost the J sc of corresponding PSC. It is generally recognized that interfacial voids are harmful to the performance of planar PSC, since

Fig. 3 a XRD patterns and b UV-vis absorption spectra of CsPbIBr 2 films without and with interfacial voids, c steady-state PL spectra and d TRPL curves of CsPbIBr 2 films without and with interfacial voids on insulating glass substrates instead of FTO/TiO 2 substrates
they inevitably reduce the contact area between absorber film and charge transporting layer, thus increasing the transporting resistance of charge carriers. Even so, in the present case, we speculate that such negative effect can be fully counteracted by the multiply beneficial contributions of the interfacial voids described above. To verify our inference, the CsPbIBr 2 films without and with interfacial voids are used to construct carbon-based, all-inorganic PSCs with the structure of FTO/TiO 2 /CsPbIBr 2 /Carbon as illustrated in Fig. 4a. Firstly, the PEAI solution concentration was optimized according to the performance of resulting PSCs. As shown in Fig. 4b, the PSC yields the optimized PCE when 3 mg mL −1 PEAI solution was used. So, this solution is adopted to fabricate all the PSCs hereinafter. Figure 4c provides the statistical PCEs of 20 independent PSCs based on the CsPbIBr 2 films without and with interfacial voids, which were measured under simulated AM 1.5 G illumination (100 mW cm −2 ) and a scan rate of 10 mV s −1 . A black mask containing a circular aperture with specific area of 0.085 cm 2 was used to confine the active area of the PSCs. Accordingly, the average PCEs are estimated to be 5.22 ± 0.54% and 9.54 ± 0.59% for the PSCs fabricated with CsPbIBr 2 films without and with interfacial voids, respectively. It is extremely surprised that the CsPbIBr 2 films with interfacial voids can enable over 85% enhancement in average PCE of carbon-based, all-inorganic PSCs. Figure 4d shows the statistical V oc values of as-fabricated PSCs, from which the average V oc values are estimated to be 1.230 ± 0.023 and 1.297 ± 0.018 V for the PSCs based on CsPbIBr 2 films without and with interfacial voids, respectively. Thus, we can conclude that the V oc values of the PSCs based on CsPbIBr 2 films with interfacial voids are larger consistently. It is worth noting that the superior V oc is not only vital to overcome the PCE obstacle facing current CsPbIBr 2 PSCs, but also can promote their applications in other energy conversion systems, such as water splitting cells [11,17,28]. Figure 4e displays the current density versus voltages (J-V) curves of the best-performing PSCs employed CsPbIBr 2 films without and with interfacial voids, respectively. The typical photovoltaic parameters including V oc , FF, J sc , and PCE are provided in Table 1. The PSC fabricated with CsPbIBr 2 film with interfacial voids delivers the V oc of 1.338 V, FF of 0.65, and J sc of 11.73 mA cm −2 , leading to a much higher PCE of 10.20%. Such PCE is beyond nearly all the CsPbIBr 2 PSCs reported previously. More importantly, as shown in Table 2, the V oc achieved herein stands the highest value among them [8, 17-20, 25-27, 30, 31, 35, 47, 48, 51-56].
External quantum efficiency (EQE) spectra of the champion PSCs are given in Fig. 4f. Photocurrent response of each PSC is consistent with the absorption feature of CsPbIBr 2 film. By contrast, the PSC based on CsPbIBr 2 film with interfacial voids exhibits much higher EQEs in the wavelength regions from 400 to 580 nm, disclosing its better photoelectric conversion capability as a result of higher light absorption and more energetic extraction of charge carriers. Moreover, a remarkable EQE peak of 86% can be realized in the PSC. In addition, the integrated current densities from the EQE spectra (Fig. S10) are basically consistent with the J sc provided in Table 1, wherein the discrepancies may be resulted from the absence of light soaking during EQE tests and the spectral mismatch [30,31,52]. We also tested the steady-state photocurrent outputs of the champion PSCs under maximum power-point conditions (Fig. 4g). They exhibit fast light-response, indicating smooth charge extraction. This desired character is mainly relied to large energy difference of conduction band minimums (CBMs) for CsPbIBr 2 absorber and TiO 2 ETL [18,51]. One can see that the PSC based on CsPbIBr 2 film without interfacial voids yields the stabilized current density of ~ 6.18 mA cm −2 under a constant bias voltage of 0.880 V, which corresponds to a stabilized PCE of 5.44%. And, the one without interfacial voids exhibits the stabilized PCE of 8.99% under 1.040 V, which supports the better performance of the PSC fabricated with CsPbIBr 2 film with interfacial voids. However, both steady-state PCEs are smaller than those of extracted from J-V curves, indicative of inevitable anomalous J-V hysteresis. The J-V hysteresis is frequently observed in previous CsPbIBr 2 PSCs and primarily related to light-induced phase segregation in CsPbIBr 2 films [8, 46-48, 51, 52, 54, 55]. In addition, one can notice that during the test period the photocurrent outputs of both PSCs are stable in ambient atmosphere, revealing their excellent ability against humidity and light attacking.
Finally, thermal stability and long-term stability of the champion PSC prepared with CsPbIBr 2 film with interfacial voids were studied. For thermal stability test, the un-encapsulated PSC was continuously heated at 85 °C on a hotplate in N 2 -filled glovebox. And, for long-term stability test the same PSC was stored at room temperature in ambient atmosphere with relative humidity of 30-40%. As shown in Fig. 4h, the PSC can remain ~ 97% of its initial PCE after being tested for 12 h, indicative of its superior thermal stability. We attribute this favored aspect to the excellent resistance of CsPbIBr 2 film to high temperature, as it is prepared at a higher annealing temperature of 280 °C. In general, the annealing temperatures for its analogs including CsPbI 3 , CsPbI 2 Br, and CsPbBr 3 are 200, 160, and 250 °C, respectively. Figure 4i provides the result of long-term stability test. The same PSC can sustain ~ 94% of its initial PCE after storing for 60 days, making clear its superior long-term stability against humidity and interface degradation. Overall, one can conclude that the CsPbIBr 2 film with interfacial voids can enable the carbon-based, all-inorganic PSC with outstanding PCE and record-high V oc , along with excellent operation stability under light, thermal, or humidity stress.

Carrier Dynamics
To gain insight into the relationship between the carrier dynamics and performance of PSCs prepared with CsPbIBr 2 films without and with interfacial voids, we explored the transport mechanisms in the fabricated PSCs by means of various characterization techniques, including dark J-V, light-intensity-dependent V oc , transient photocurrent (TPC), transient photovoltage (TPV), electrochemical impedance spectroscopy (EIS), and Mott-Schottky (M-S) measurements. Figure 5a gives the dark J-V curves of investigated PSCs. In general, the dark J-V curve of a solar cell is dominated by the diode law. Based on it, the J 0 can be estimated from the intercept at zero voltage of the linear regression from semi-logarithmic J-V plot in the region close to the diode turn-on voltage [19,23,57]. In general, J 0 is governed by recombination of charge carriers in the cell, wherein a low J 0 is always companied by the suppressed recombination. Herein, we can see that the PSC based on CsPbIBr 2 film with interfacial voids exhibits much smaller J 0 (1.14 × 10 −7 mA cm −2 ) than the one without interfacial voids (1.37 × 10 −5 mA cm −2 ), indicating the weaker recombination of charge carriers in it. It should be noted that although the larger grains and higher crystallinity can partly decrease J 0 of the PSC based on CsPbIBr 2 film with interfacial voids, we believe that the interfacial voids may dominantly contribute to its much smaller J 0 . This is because the PSC exhibits the similarly lower dark current under high bias voltages such as 1.5 V. Figure 5b presents the plots of V oc as a function of incident light intensity for the concerned PSCs. One can see that the V oc improves with the logarithmic light intensity monotonically, revealing the dominant trap-assisted Shockley-Read-Hall recombination in both PSCs [9,23,30,31]. In addition, the PSC based on CsPbIBr 2 film without interfacial void shows a slope of 2.08 k B T/q, while the one with voids yields a smaller slope of 1.70 k B T/q that is more close to 1 k B T/q, where q is the electric charge, T is temperature, and k B is the Boltzmann constant. These results further confirm that the recombination of charge carries has been greatly suppressed in the PSC based on CsPbIBr 2 film with interfacial voids. Figure 5c gives the TPC curves of the corresponding PSCs, which can reflect the extraction and transport properties of charge carriers. The photocurrent decay constants are estimated to be 2.64 and 1.83 μs for the PSCs based on CsPbIBr 2 film without and with interfacial voids, respectively. These results mean that the interfacial voids do not damage the extraction and transport of charge carriers [18,19,51], even though they induce significantly decreased dark current density. Instead, the extraction and transport of charge carriers are improved slightly, which probably benefit from the larger grains and higher crystallinity of CsPbIBr 2 film with interfacial voids that endow its better transport ability for photon-generated carriers. At the same time, the recombination of charge carriers in the PSCs is monitored by TPV measurements [18,19,51]. The results in Fig. 5d indicate that the PSC based on CsPbIBr 2 film with interfacial voids has the slower photovoltage decay (1.06 ms) than the one without interfacial void (0.79 ms), which supports the suppressed recombination of charge carriers in the former again.
Furthermore, EIS tests were performed on the fabricated PSCs to estimate recombination resistance (R rec ) of charge carriers in them [14,17,25]. The Nyquist plots that were recorded at a forward bias voltage of 1.0 V under dark condition are given in Fig. 5e. The clear arc is involved to carrier recombination, reflecting the recombination resistance R rec . The larger R rec of the PSC based on CsPbIBr 2 film with interfacial voids is well accordance with the suppressed recombination of charge carriers in it. On the other hand, the M-S results in Fig. 5f indicate that the built-in potential (V bi ) of the PSC based on CsPbIBr 2 film with interfacial voids is 1. 56 V, which is much higher than that of the one without interfacial void (V bi = 1.30 V). A larger V bi not only provide higher driving force to dissociate photo-generated carriers, but also contribute to form a wider depletion region for suppressing recombination of charge carriers in the PSC, as verified by the dark J-V, TPV, and EIS results above [15,18,30,31]. As a whole, the higher V bi and weaker recombination of charge carriers directly contributes to the record-high V oc as well as superior FF of the PSC based on CsPbIBr 2 film with interfacial voids. And, the higher absorption of CsPbIBr 2 film as a result of intense light scattering effect of interfacial voids gives rise to its higher J sc . Therefore, the PCE of the PSC based on CsPbIBr 2 film with interfacial voids is improved significantly.

Conclusions
We demonstrate a new interfacial engineering for carbonbased, inorganic CsPbIBr 2 PSC by manufacturing interfacial voids between TiO 2 ETL and CsPbIBr 2 film. The interfacial voids contribute to better absorption, higher V bi , and smaller J 0 of the PSC, as results of enhanced light scattering in CsPbIBr 2 film, boosted crystal quality of CsPbIBr 2 grains, and significant suppressed recombination of charge carriers. Consequently, the optimized PSC achieves the PCE of as high as 10.20% coupled with the V oc of 1.338 V. Such V oc represents the best value among CsPbIBr 2 PSCs reported earlier. More promisingly, the non-encapsulated PSC can remain ~ 97% and ~ 94% of its initial PCE after being heated at 85 °C for 12 h or storing in ambient atmosphere with relative humidity of 30-40% for 60 days, respectively, suggesting its excellent stability against light, thermal, and humidity stresses.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.