1 Introduction

Hybrid organic–inorganic halide perovskite solar cells have been considered as a promising photovoltaic technology with a rapid rise in power conversion efficiency (PCE) from 3.8% to 22.7% and a low cost for fabrication by solution process [1,2,3,4,5]. A major obstacle to commercialization is their instability. In particular, they subject to compositional degradation at high temperature [6]. It has been reported that when MAPbI3 is annealed above 85 °C, it can be significantly decomposed into PbI2 and MAI [7]. One promising way to enhance their thermal stability is to substitute the organic component (CH3NH3+, NH2CH=NH2+) with an inorganic component such as Cs [8,9,10]. Accordingly, CsPbI3, due a suitable band gap of 1.73 eV, has been developed for photovoltaic application and the based solar cells have been fabricated with a PCE of 2.9% by Snaith and coworkers [11]. Unfortunately, CsPbI3 is unstable in the black cubic phase at room temperature which will quickly convert to the yellow nonperovskite phase, especially in ambient atmosphere. Meanwhile, CsPbBr3 has also been used as absorber in solar cells and a PCE of 6% has been achieved by Kulbak et al. [12, 13]. However, the band gap of CsPbBr3 is 2.3 eV, which is too wide to be used even in multi-junction tandem solar cells [14]. Hence, a series of CsPbI3−xBrx perovskite has been developed in order to obtain both suitable band gap and high phase stability simultaneously [15,16,17,18,19,20,21]. Among these, CsPbI2Br, with a band gap of ~1.9 eV, is suitable for a top block in a triple-junction device and accordingly has been concerned by many researchers [22,23,24,25,26,27,28]. Up to now, CsPbI2Br-based regular solar cells have achieved PCE as high as 13.47% and show excellent stability at both room and elevated temperatures when prevented from exposure to water vapor [29]. However, there is still large energy loss, reflected by the large difference between band gap (Eg) and open-circuit voltage (VOC), especially for devices with inverted structures. Furthermore, the bad humidity stability puts forward high requirement for encapsulation, brings large inconvenience and increased cost for study and hinders its further development and potential application in the future.

In this paper, we incorporated a small amount of hydrophobic material of CH3NH3Cl (MACl) to CsPbI2Br precursor solution. The influences of the MACl content on the morphology and defect density of the perovskite film as well as the final device performance were studied carefully. Ultraviolet–visible (UV–Vis) spectra and X-ray photoelectron spectroscopy (XPS) characterization were also conducted to analyze the band gap and composition changes. Furthermore, the stability of CsPbI2Br solar cells based on MACl additive was tested, which is much better than that of the pristine device. This work provides useful information about perovskite solar cells and is an advance of practical applications to cater the current energy need of the world.

2 Experimental

2.1 Device fabrication

Glass substrates (TEC-15, NSG Pilkington) with the etched fluorine-doped tin oxide (FTO) coating were first ultrasonically cleaned with detergent solution, Milli-Q water, ethyl alcohol and acetone in sequence. After drying with clean dry air, a p-type NiMgLiO film serving as hole extraction layer was deposited onto FTO glass by spray pyrolysis at 550 °C according to our previous work [30]. Then, the NiMgLiO-coated FTO glass substrates were transferred to a N2-filled glove box. The anti-solvent-assisted spin-coating technology was used for the deposition of CsPbI2Br(MACl)x-based perovskite layers: 0.85 mol·L−1 dimethylformamide (DMF)/dimethyl sulfoxide (DMSO) (4:1 by volume ratio) mixture solution of PbI2/CsBr/MACl (1:1: x by molar ratio, x = 0, 0.01, 0.03, 0.05, 0.08) was spin-coated at 4000 r·min−1 for 45 s, followed by rapidly drop-casting diethyl ether (1 ml) as anti-solution, and then annealed at 240 °C for 20 s. The formed perovskites were denoted as MACl 0, MACl 0.01, MACl 0.03, MACl 0.05, and MACl 0.08, respectively. After the inorganic perovskite films were prepared, a chlorobenzol solution of [6]-phenyl-C61-butyric acid methyl ester (PCBM) (20 mg·ml−1) was spin-coated on top of them at the rotation speed of 2000 r·min−1 for 30 s. Subsequently, a 5-nm-thick buffer layer was fabricated by spin-coating saturated methanol solution of BCP at the rotation speed of 6000 r·min−1 for 30 s. Finally, 120-nm-thick Ag electrodes were deposited under high vacuum (< 5 × 10−4 Pa) in evaporation chamber.

2.2 Characterization

Scanning electron microscopy (SEM) images were obtained via a Nova Nano 450 SEM (FEI Co., the Netherlands) at a 5 kV accelerating voltage. X-ray diffraction (XRD) characterization was performed on a Philips X-ray diffractometer with Cu Kα radiation. X-ray photoelectron spectroscopy (XPS) measurements were carried out on an AXIS-ULTRA DLD-600W Ultra spectrometer (Kratos Co., Japan). The ultraviolet–visible (UV–Vis) spectra were obtained from a Lambda 950 spectrophotometer (PerkinElmer Co., USA). The PL spectra were performed on an Edinburgh FLS920 fluorescence spectrometer (Edinburgh Co., UK). The current density–voltage (J–V) curves were measured via a Keithley 2400 source meter. A solar simulator (Oriel, model 9119) with AM 1.5G filter (Oriel, model 91192) was used to provide an irradiance of 100 mW·cm−2, and the light intensity of the simulated solar light was precisely calibrated with a standard Si photodiode detector. The effective area of the solar cell was defined to be 0.09 cm2 with a black metal mask. The incident photo-to-electron conversion efficiency (IPCE) was measured on a Newport IPCE system (Newport, USA). The Mott–Schottky plots were obtained via an electrochemical workstation (Zahner Zennium, Germany).

3 Results and discussion

Based on anti-solvent method, CsPbI2Br films with varying MACl contents were fabricated. To investigate the effect of different MACl additions on the surface morphology of perovskite films, top-view SEM images were processed and the results are shown in Fig. 1. It is found that the changes are not manifest among the perovskite films when the MACl content varies from 0 to 0.03, while if the MACl content is further increased to more than 0.05, the CsPbI2Br film becomes inhomogeneous and lots of small grains emerge (Fig. 1d, e). Such results indicate that moderate MACl additive is beneficial for the interface passivation and assists in achieving high-quality inorganic perovskite films.

Fig. 1
figure 1

Surface SEM images of CsPbI2Br(MACl)x films: ax = 0, bx = 0.01, cx = 0.03, dx = 0.05 and ex = 0.08

Full size image

XRD measurements are provided to quantify the effect of MACl additive on the crystallinity of CsPbI2Br film, and the results are depicted in Fig. 2. It is noted that all the MACl-doped films show a typical perovskite phase with the dominant peaks at 14.6° and 29.5°, assigned to (100) and (200) planes, respectively, which is well consistent with previous studies [28, 31]. Upon MACl doping, all the inorganic perovskites demonstrate similar XRD patterns, indicating that small amount of MACl doping does not change the growth direction of CsPbI2Br crystals. In addition, it is found that the absolute intensities of both the two peaks for all those patterns decrease apparently with the increase in MACl content. These results coherently indicate that MACl additive has a significant influence on the grain growth and litter size grains emerge with the MACl additive increasing, which are confirmed by the SEM images in Fig. 1.

Fig. 2
figure 2

XRD patterns of CsPbI2Br(MACl)x films with varying MACl contents

Full size image

For the purpose of understanding the effect of MACl additive on the elemental composition at the surface of inorganic perovskite films, XPS measurements were carried out on the CsPbI2Br and CsPbI2Br(MACl)0.03 films. The results are shown in Fig. S1. Obvious peak of Cl element centered at 198.2 eV is observed for the CsPbI2Br(MACl)0.03 film. Furthermore, there is a little shift for Pd 4f peaks toward lower binding energy, indicating that Pb–Cl bonds may be formed in the lattice. Atomic ratios are listed in Table S1. As seen, the Cl/Pb atomic ratio is about 3.2%, close to the molar content of MACl additive. Meanwhile, the reduced I/Pb atomic ratios are estimated to be 221% for the CsPbI2Br(MACl)0.03 film, far less than 247% for the CsPbI2Br film, which means that partial I− has been substituted by Cl−. Based on XPS results within the typical XPS detection depth of 10 nm, it is noted that the content of C increases sharply after incorporating 0.03 mol% MACl into CsPbI2Br film, which means that a great deal of MA+ stays at the surface of inorganic film as a passivator. The results indicate that MA+ is enriched at MACl-doped film surfaces, which can passivate defect states at the surface, quite similar to our previous study on Ca2+-doped perovskite films [32].

Optical and PL measurements were also conducted on the CsPbI2Br(MACl)x films, and the results are presented in Fig. 3. Figure 3a demonstrates that with the increase in MACl content from 0 to 0.03, the corresponding UV–Vis spectroscopy shows an apparent blueshift and their absorption onsets shift gradually from 656 nm (~ 1.9 eV) to 636 nm (~ 1.95 eV). In contrast, if the MACl content further increases to 0.08, the corresponding absorption peak positions keep almost constant, suggesting that it has been saturated for MACl doping just with a little content of 0.03. At the same time, with the MACl content further increasing, the absorption of the inorganic film slightly degrades which may be attributed to large number of small grains demonstrated by SEM.

Fig. 3
figure 3

a UV–Vis absorption spectra; b steady-state PL spectra; and c normalized time-resolved PL spectra of CsPbI2Br(MACl)x films with varying MACl contents on glass substrates

Full size image

Steady-state photoluminescence (PL) spectra for CsPbI2Br(MACl)x films on glass substrates are given in Fig. 3b. They exhibit apparent blueshift with MACl content increasing from 0 to 0.08, attributed to insertion of Cl− to the lattice of CsPbI2Br perovskite, which is in good agreement with the UV–Vis results. At the same time, their peak intensity increases first and then decreases with the increase in MACl content and achieves the maximum intensity for the 0.03 MACl additive. These results suggest that 0.03 MACl-doped film has the least defect states, which is consistent with their improved morphology and abrupt slope of UV–Vis absorption.

The improvement in the steady PL intensity usually relates to prolonged carrier lifetime, which can be verified by time-resolved PL spectra depicted in Fig. 3c. Herein, the time-resolved PL decay curve can be fitted with following bi-exponential decay function [33]:

$$F(t) = A_{1} \exp \left( { - \frac{t}{{\tau_{1} }}} \right) + A_{2} \exp \left( { - \frac{t}{{\tau_{2} }}} \right)$$
(1)

where F is the normalized intensity; t is the time; τ1, τ2 are lifetimes related to two kinds of recombination; and A1 and A2 are the related weight contents. The average decay time (mean PL average lifetime) can be obtained by the following equation:

$$\tau_{\text{avg}} = \frac{{A_{1} \tau_{1} + A_{2} \tau_{2} }}{{A_{1} + A_{2} }}$$
(2)

where τavg denotes the average decay time. With variation in the content of MACl additive, the fitting parameters (τ1, τ2, A1, A2) and PL average lifetime are summarized in Table S2. Obviously, the PL lifetime for MACl 0.03 additive film is 20.6 ns, much larger than that for pristine CsPbI2Br film (8.84 ns) and MACl 0.08 additive film (1.11 ns). This result indicates that moderate MACl doping may be useful for improving optoelectronic properties of inorganic CsPbI2Br perovskite film.

Based on MACl-doped films, perovskite solar cells (PSCs) with inverted architecture of “FTO/NiMgLiO/CsPbI2Br(MACl)x/PCBM/BCP/Ag” (Fig. 4a) have been fabricated. Their J–V characteristic curves of the champion samples under AM 1.5 G irradiation at 100 mW·cm−2 are demonstrated in Fig. 4b, and the resultant performance parameters are listed in Table S3. At the same time, the statistic distribution of performance parameters depending on MACl additive content is presented in Fig. 4c–f. In this study, the champion cell based on 0.03 MACl additive exhibits the highest power conversion efficiency (PCE) of 12.9%, with an open-circuit voltage (VOC) of 1.22 V, a short-circuit current density (JSC) of 13.8 mA·cm−2 and a fill factor (FF) of 0.76. However, the CsPbI2Br-based cell only obtains a PCE of 10.6%, with VOC of 1.08 V, JSC of 14.4 mA·cm−2 and FF of 0.68. Obviously, the main enhancement in PCE of 0.03 MACl device comes from enhanced VOC and FF due to better morphology and wider band gap. It is noteworthy that if the MACl content further increases from 0.03 to 0.08, VOC, JSC and FF decrease significantly to 0.95 V, 12.6 mA·cm−2 and 0.62, respectively, resulting in a much degraded PCE of 7.4%. Such result may be attributed to poor morphology with excessive MACl additive. Furthermore, it is found that the device based on 0.03 MACl content exhibits little hysteresis compared to pristine CsPbI2Br-based device (Fig. S2), which is attributed to the excellent carrier extraction capability of surface and trap passivation by MACl.

Fig. 4
figure 4

a Device architecture of FTO/NiMgLiO/CsPbI2Br(MACl)x/PCBM/BCP/Ag; bJ–V characteristics of champion devices based on CsPbI2Br(MACl)x under forward scan direction; photovoltaic parameters of perovskite devices as a function of MACl additive content: cJSC, dVOC, e FF, f PCE (16 pieces of solar cells included)

Full size image

It is obviously found that the JSC value for 0.03 MACl-doped device is 13.8 mA·cm−2, which is smaller than that for the pristine CsPbI2Br-based device (14.4 mA·cm−2), matching well with the integrated JSC values from the IPCE data presented in Fig. 5a. This result may be ascribed to the blueshift of absorption spectra, leading to a smaller cutoff wavelength.

Fig. 5
figure 5

a Incident photo-to-electron conversion efficiency (IPCE) curves (solid lines) with integrated photocurrents (dashed lines); b Mott–Schottky plots for champion devices based on CsPbI2Br and CsPbI2Br(MACl)0.03 films; c bilogarithmic diagram of I–V curves in dark for devices with architecture of FTO/perovskite/Au

Full size image

Besides, VOC of CsPbI2Br(MACl)0.03-based cell achieves a value of 1.22 V, much higher than that of pristine CsPbI2Br-based cell (1.08 V) and CsPbI2Br(MACl)0.08-based cell (0.95 V). The higher Voc is partly attributed to the wider band gap of ~ 1.95 eV, but more importantly, the much-reduced defect density and more effective charge extraction at the surface may play a great role. To elucidate the origin of enhanced VOC, Mott–Schottky analysis has been performed on the two photovoltaic devices based on CsPbI2Br and CsPbI2Br(MACl)0.03. Figure 5b presents the capacitance–voltage (1/C2 − V) plots for the corresponding cells, and the built-in potentials (Vbi) can be acquired with the following Mott–Schottky equation [28]:

$$\frac{1}{{^{{\mathop C\nolimits^{2} }} }} = \frac{2}{{\varepsilon \varepsilon_{0} qA^{2} N}}(V_{\text{bi}} - V)$$
(3)

where V is the applied bias and the parameters ε, ε0, q, A and N represent mean relative permittivity, vacuum permittivity, elementary charge, active area and free carrier concentration, respectively. Based on the method reported in our previous literature [27], the values of Vbi are equal to 0.64 and 0.94 V for the cells based on CsPbI2Br and CsPbI2Br(MACl)0.03, respectively (Fig. 5b). This result is consistent with the trend of VOC values extracted from their J–V curves. A higher Vbi means an improved driving force for the separation of photo-generated carriers and an extended depletion region for efficient suppression of electron–hole recombination. As a result, the incorporation of small amount of MACl is beneficial to the increase in the output voltage of CsPbI2Br PSCs.

To further elucidate the impact of MACl addition on perovskite trap density, the I–V responses of two samples based on CsPbI2Br and CsPbI2Br(MACl)0.03, with the architecture of FTO/perovskite/Au, have been performed [34, 35]. Their dark current–voltage (I–V) characteristic curves are presented in Fig. 5c. Herein, the trap-filled limit voltage (VTFL) can be used to calculate the trap state density (Ntrap) with the equation of Ntrap = VTFL (2εε0)/eL2, where e is the elementary charge (1.6 × 10−19 C), L is the perovskite film thickness (~ 300 nm), the vacuum permittivity (ε0) is equal to 8.854 × 10−12 F·m−1 and the relative permittivity (ε) for CsPbI2Br perovskite is ~ 8.6 [36]. From Fig. 5c, it can be determined that VTFL for samples based on CsPbI2Br and CsPbI2Br(MACl)0.03 is 2.95 and 1.41 V, respectively. The resultant Ntrap is calculated to be 3.12 × 1016 and 1.49 × 1016 cm−3, respectively. Obviously, the trap state densities are decreased after the CsPbI2Br film is passivated by MACl. It is noted that the enhanced quality of perovskite films is consistent with the corresponding steady PL and time-resolved PL spectra depicted in Fig. 3b, c.

In addition to device efficiency, the ambient stability of devices based on CsPbI2Br and CsPbI2Br(MACl)0.03 films is also examined. Figure 6a shows the normalized PCEs of unencapsulated devices under continuous illumination (simulated solar light, 100 mW·cm−2) in ambient atmosphere (T = ~25 °C and relative humidity (RH) = ~30%). It is found that the CsPbI2Br(MACl)0.03 device retains 80% of its initial efficiency for ~ 8.1 h, nearly 1.8 times that for CsPbI2Br device (~ 4.4 h). The enhanced ambient stability may be partially attributed to interface passivation; moreover, the incorporated MACl additive as a hydrophobic material may play an important role. Herein, two different devices, namely FTO/NiMgLiO/CsPbI2Br and FTO/NiMgLiO/CsPbI2Br(MACl)0.03, were fabricated and compared under their optimal interfacial condition. As demonstrated in Fig. 6b, when the two devices were placed in ambient atmosphere, the pristine CsPbI2Br perovskite layer was corroded rapidly within ~ 3 min, while the CsPbI2Br(MACl)0.03 layer was kept nearly undamaged for ~ 10 min. Thus, the much stable CsPbI2Br(MACl)0.03 layer gives the device higher stability in ambient atmosphere.

Fig. 6
figure 6

a Normalized PCEs of unencapsulated photovoltaic devices based on CsPbI2Br and CsPbI2Br(MACl)0.03 films under continuous illumination (simulated solar light, 100 mW·cm−2), placed in ambient atmosphere with a humidity of ~ 30%; b comparison of degradation speeds from two different samples: CsPbI2Br film and CsPbI2Br(MACl)0.03 film, two films were all placed in ambient atmosphere for 20 min

Full size image

4 Conclusion

In summary, the effects of MACl on CsPbI2Br perovskite films and the based devices were investigated. With the incorporation of a small amount of MACl (≤ 0.03), CsPbI2Br perovskite films demonstrate denser and uniform morphology and achieve the best state when the MACl content is equal to 0.03. Being attributed to a good passivation effect, the defect density decreases dramatically and the average PL lifetime increases to 20.6 ns. Based on such optimized MACl-doped films, the champion cell achieves a PCE of 12.9% with much higher Voc of 1.22 V compared to the pristine CsPbI2Br cells, with the Voc of 1.08 V. In addition, compared with the CsPbI2Br-based device, the CsPbI2Br(MACl)0.03-based cell reveals a superior ambient stability, which is attributed to the improved perovskite quality and better hydrophobic characteristics of CsPbI2Br(MACl)0.03 layers. This work provides a clue for improving ambient stability of inorganic perovskite cells, which is of particular importance for practical application.