Interfacial Contact Passivation for Efficient and Stable Cesium-Formamidinium Double-Cation Lead Halide Perovskite Solar Cells.

Perovskite solar cells (PSCs) have achieved extremely high power conversion efficiencies (PCEs) of over 25%, but practical application still requires further improvement in the long-term stability of the device. Herein, we present an in situ interfacial contact passivation strategy to reduce the interfacial defects and extraction losses between the hole transporting layer and perovskite. The existence of PbS promotes the crystallization of perovskite, passivates the interface and grain boundary defects, and reduces the nonradiation recombination, thereby leading to a champion PCE of 21.07% with reduced hysteresis, which is one of the best results for the methylammonium (MA)-free, cesium formamidinium double-cation lead-based PSCs. Moreover, the unencapsulated device retains more than 93% and 82% of its original efficiencies after 1 year's storage under ambient conditions and thermal aging at 85°C for 1,000 h in a nitrogen atmosphere, likely due to the usage of MA-free perovskite and the enhanced surface hydrophobicity.

Introduction

Perovskite solar cells (PSCs) have drawn intense research interest due to their excellent optoelectronic properties and ease of fabrication (Kojima et al., 2009, Zhang et al., 2016, Brenner et al., 2016, Yang et al., 2018). Tremendous efforts from different groups have boosted the power conversion efficiencies (PCEs) of PSCs from 3.8% to over 25% in just a few years, making perovskite one of the most promising photovoltaic (PV) materials (NREL Chart.). However, long-term stability is still limiting the practical application of PSCs, which is partly related to the usage of methylammonium (MA)-containing perovskite (Wang et al., 2019). The instability of MAPbI3 is ascribed to the degradation of CH3NH3I into CH3I and NH3 at temperature as low as 80°C (Turren-Cruz et al., 2018). Fortunately, the pure formamidinium (FA)-based perovskite exhibits higher thermal stability and narrower band gaps (1.48 eV) than the MA-containing perovskite (Lee et al., 2015), showing great application potentials in PV area. Nevertheless, the pure FA-based perovskite is traditionally prone to form photo-inactive phase (yellow phase) at room temperature; the photo-active phase (black phase) can only be obtained at high temperature (more than 150°C). Although many works have been performed on avoiding the yellow phase of FA-based perovskite by doping with small amount of Cs+ and/or Rb+, the PCEs of the MA-free PSCs are still not satisfactory compared with the PCEs of MA-containing PSCs (Turren-Cruz et al., 2018, Wu et al., 2018, Chen et al., 2018, Chen et al., 2019a). Much effort should thus be devoted to elevate the performance of MA-free, FA-based PSCs.

In addition, the device stability is also reported to be related to the imperfect contact and carrier recombination at the interfaces and grain boundaries (GBs). Owing to the low formation energy and poor thermal stability, presence of a large number of defects at the interfaces and GBs within the polycrystalline perovskite is unavoidable (Zheng et al., 2017, Niu et al., 2018). In fact, defects can enhance non-radiative recombination, seriously reducing charge-carrier lifetime and photoluminescence (PL) yield. In addition, these defects form convenient channels for moisture or oxygen diffusion and ion migration, lead to the grain degradation, and decrease the stability of PSCs (Liu et al., 2018, Zheng et al., 2018). Therefore, it is highly crucial to passivate the defects at the surface and the GB to further boost the PCE, and to prolong the durability of PSCs. Interfacial engineering of perovskite film by inserting an interfacial layer, such as methimazole (Liu et al., 2018), quaternary ammonium halide (Zheng et al., 2017), π-conjugated Lewis base (Lin et al., 2017, Qin et al., 2018), organic semiconductor polymer (Wu et al., 2016, Wu et al., 2018), organic halide salt (Jiang et al., 2019), wide band-gap perovskite (Cho et al., 2017), or two-dimensional perovskite materials (Cho et al., 2018), between the perovskite and the charge transport layers has been the widely used method to eliminate the interfacial trap states and improve the device stability. These modifications of the perovskite interfaces not only serve as encapsulants against the environment but also play multifunctional roles, acting as a semiconductor or charge tunneling layer, reducing charge-carrier recombination at the interface, and protecting the perovskite from degradation.

Similar interfacial passivation strategy has also been employed in other PV systems, such as quantum dots (QDs) solar cells. It is amazing that the perovskite materials can be used as a passivant in QD solar cells. Sargent et al. (Yang et al., 2015a) and Liang et al. (Peng et al., 2016) have used the MAPbI3 perovskite to passivate inorganic PbS colloidal QDs itself and the interface of PbS QDs film, respectively. Significantly improved device performance has been achieved for inorganic PbS QDs solar cells. It was reported that the perovskite and PbS/PbSe can be well connected in their lattice fringes to form a heterocrystal, i.e., QDs-in-matrix, owing to the extremely minimal lattice mismatch (Ning et al., 2015, Yang et al., 2015b), which might be the reason for the improved device performance of PbS QDs solar cells. PbS QDs have also been employed as the hole transporter and co-sensitizer in PSCs (Li et al., 2015, Hu et al., 2015, Etgar et al., 2014); however, the performance is still not satisfactory. Besides PbS, numerous QDs, such as PbSe (Zhang et al., 2017), SnS (Han et al., 2017), CdSe (Qi et al., 2018), and Cu(In,Ga)(S,Se)2 (Khanzada et al., 2016), have also been introduced into PSCs to be used as a dopant, passivant, interfacial layer, or charge transporting layer. The detailed PV performances are summarized in Table S1. These results indicate that the QDs exhibit great application potential in PV area. Recently, Lin and co-workers (Han et al., 2018) have fabricated a perovskite-QDs hybrid film (bulk heterojunction [BHJ]) by adding the PbS QDs into the perovskite precursor solution. Improved film quality combined with a PCE of 18.6% has been achieved due to the formation of QDs-in-perovskite structure. However, the BHJ structure has the drawback of uncontrolled distribution of QDs in the perovskite layer, which can lead to unfavorable contact between electron- and hole-extracting materials (Wu et al., 2016).

Here, we report an interfacial heterojunction engineering method to passivate the defects at the interface and GBs of MA-free perovskite. A modified green anti-solvent-assisted approach was adopted to fabricate the PbS QDs containing (Cs,FA)-based perovskite films (the formula of perovskite was Cs0.15FA0.85Pb(I0.9Br0.1)3) (Chen et al., 2018). The interfacial trap states have been well passivated by the PbS interfacial layer along with the significantly reduced non-radiation interfacial recombination, which is favorable for the charge-carrier transfer and collection. As a result, the device with PbS QDs passivation layer yielded an open-circuit voltage (VOC) as high as 1.146 V and a PCE of 21.07% and showed excellent long-term ambient stability, demonstrating the effectiveness of defects passivation to achieve efficient and stable PSCs.

Results and Discussion

The PbS QDs were synthesized according to a widely adopted hot-injection colloidal solution approach (Barkhouse et al., 2011). The crystal phase and morphology of the synthesized QDs were disclosed to be cubic structure with an average size of 3.35 nm by using X-ray diffraction (XRD) patterns and transmission electron microscopy, respectively (as shown in Figure S1). The PbS-passivated perovskite film was fabricated by using the PbS-containing green antisolvent (PbS/anisole), as shown in Figure 1A. Figures 1B, 1C, and S2 present the top-view scanning electron microscopic images of the perovskite films without and with PbS. The pristine perovskite film is fully covered without pinholes, and the average crystal size of the perovskite is 200–300 nm. After the introduction of PbS QDs, slightly enlarged perovskite crystal size can be found, indicating that the quality of perovskite film has been improved. Similar results can be reached by the XRD patterns, as shown in Figure 1D. Both perovskite films made without and with PbS in the anti-solvent show the same crystal structure without detectable impurities, except that the film with PbS exhibited significantly increased intensity of the diffraction peaks located at 14.06° and 28.31°, which were assigned to the (110) and (220) crystal planes, respectively, indicative of the better crystallinity and preferable crystal growth along (110) facet. These results indicated that existence of PbS can promote and control the crystallization of perovskite, which agrees well with the previously reported results (Han et al., 2018). The values of the root-mean-square roughness were identified to be 16.1 and 14.5 nm for the pristine and passivated films, respectively, by atomic force microscopy, as shown in Figure S3. It is notable that the smoother surface is usually indicative of higher film quality. However, the existence of small amount of PbS at the interface and grain does not change the light absorption properties of perovskite films (Figure 1E). Larger grains and smoother surface are beneficial for the improvement of PV performance.

Figure 1

The Film Preparation Process and Characterization

(A–E) (A) Schematic illustration of the film preparation process; scanning electron microscopic images of the perovskite films without (B) and with PbS passivation layer (C); XRD patterns (D) and UV-visible absorption spectra (E) of the perovskite films with and without passivation layer.

Figure S4 shows the elemental mapping of the perovskite film after passivation. The result confirmed that element introduced by PbS QDs is uniformly distributed on the surface of the perovskite film. To gain insight into the distribution of PbS in the perovskite film, we then employed time-of-flight secondary-ion mass spectrometry to disclose the depth profile of PbS in the perovskite layer. In a negative mass detection mode, I127 and S32 were used as indicators of the perovskite and PbS, respectively. As shown in Figure 2A, a continuous and gradual increase with a significant spatial overlap on the I127 profile was revealed. In addition, the S32 depth profile shows a gradual decline in the very shallow perovskite layer (below 10 nm). These results demonstrate that a certain amount of PbS incorporated into the shallow perovskite with a gradient distribution near the interface; however, most of the PbS distributed at the perovskite surface. Given the fact that the synthesized PbS shows p-type character and higher the valence band maximum (VBM) (−5.1 eV) (Hyun et al., 2008) than perovskite, the interfacial carrier transfer and collection should be well improved.

Figure 2

PSCs and their Photovoltaic Performance

(A) Secondary-ion mass spectrometry profiles from the top to the bottom of the PbS-passivated perovskite film. See also Figure S4.

(B) Statistics of the PCEs for devices with different concentration of PbS passivant.

(C) J-V curves of the devices based on the pristine and PbS-passivated perovskite (inset: the device architecture). See also Figure S6 and Table 1.

(D) External quantum efficiency curves and the corresponding integrated photocurrent for both devices.

Table 1

The Key Photovoltaic Parameters of the PSCs with and without Passivation Layer

	Scan direction	VOC (mV)	JSC (mA cm−2)	FF (%)	PCE (%)	HEI (%)	Rs (Ω cm2)	Rsh (Ω cm2)
Pristine	Reverse	1,109	22.93	76.47	19.35	11.26	5.29	2,413.8
	Forward	1,075	22.97	70.01	17.21
PbS	Reverse	1,146	23.06	79.82	21.07	1.65	4.29	4,186.1
	Forward	1,124	23.07	77.55	20.07

We then fabricated planar perovskite devices to evaluate the efficacy of the passivation layer, the structure is FTO/c-TiO2/perovskite/spiro-OMeTAD/Au (FTO: fluorine-doped tin oxide, spiro-OMeTAD: 2,2′,7,7′-tetrakis(N,N-dipmethoxyphenylamine)-9,9′-spirobifluorene). The thicknesses of the compact TiO2, perovskite layer, and hole transporting layer (HTL) were about 30, 700, and 250 nm, respectively (Figure S5). The concentration of PbS in anisole was first investigated. As shown in Figure 2B, the device performance highly depends on the concentration of PbS. The average PCE was improved from 18.9% for the pristine device to 20.5% for the passivated device with 0.1 mg/mL of PbS/anisole. Figure 2C presents the J-V characteristics of the best devices with and without PbS passivation layer under standard AM 1.5 illumination (at 100 mW cm−2). The main PV parameters, including short-circuit current density (JSC), VOC, fill factor (FF), and PCE, are summarized in Table 1. The control device based on the pristine perovskite exhibits a PCE of 19.35% with JSC = 22.93 mA cm−2, VOC = 1,109 mV, and FF = 76.47%, whereas the PCE for the champion device with PbS passivation layer jumped to 21.07% with JSC = 23.06 mA cm−2, VOC = 1,146 mV, and FF = 79.82%. The improvement of the device efficiency is mainly attributed to the significantly enhanced Voc and FF for the passivated device. The enhancement of VOC should be related to the reduced interfacial defects (Wu et al., 2018). High FF (>75%) usually means the absence of large carrier extraction/injection barrier for the interface between perovskite and hole transporting materials/electronic transporting materials (HTM/ETM) (Hou et al., 2017). The FFs of the present devices are between 76% and 80%, indicating that the interfacial passivation can effectively improve carrier transfer and collection. Moreover, a modified J-V hysteresis effect index (HEI) is defined (as shown in Figure S6 and Table 1) by the following equation (Rong et al., 2017):where J and J represent the current density at 80% of V for F-R and R-F scans. The resulting HEI of 1.65% for the PbS QDs-passivated devices is much lower than that of the pristine one (11.26%), which indicated that the hysteresis can be effectively suppressed with PbS QDs as the interfacial layer.

The integrated photocurrent densities are calculated to be 22.13 and 22.29 mA cm−2 from the corresponding external quantum efficiency spectra for the control and passivated devices (Figure 2D), respectively, which are comparable to the JSC values extracted from the corresponding J-V curves. Steady-state output efficiency measured at the maximum power point voltage (Vmpp) was then studied to accurately assess the performance of the device with and without passivation layer, as shown in Figure S7. The PCE of the PbS-passivated device stabilizes at 20.3% (Vmpp = 0.96 V), which is among the highest efficiency reported till date for the (Cs,FA)-based perovskite. In contrast, the control device presents a steady output efficiency of 18.1% (Vmpp = 0.92 V). These results demonstrate the superiority of the interfacial contact passivation in achieving high-efficiency PSC devices with high VOC. To assess the reproducibility of our devices, 60 individual cells for the control and passivated devices were fabricated and the detailed parameters are shown in Figures 3A–3D. It is clear that all the key J-V parameters of the passivated device are better, and the corresponding distributions are narrower when compared with the control ones, confirming the reliability of our method to fabricate high-efficiency PSCs.

Figure 3

Statistics of PV Parameters

(A–D) Statistics of PV parameters for the devices based on the pristine and PbS-passivated perovskites: (A) JSC, (B) VOC, (C) FF, and (D) PCE.

To understand the large enhancement of the PV performance, the trap density of states (tDOS) was measured to analyze the passivation effect of PbS QDs by using the thermal admittance spectroscopy (TAS), which has been widely used to determine the trap states in thin-film (Cu-In-Ga-Se) and organic solar cells (Boix et al., 2009, Walter et al., 1996). The density of shallower trap states (0.30–0.40 eV) was assigned to the trap at the GB and that of the deeper trap states (0.40–0.50 eV) was identified to the defects at the film surface (Zheng et al., 2017). As shown in Figure 4A, the device with PbS passivation layer had lower tDOS almost over the whole trap depth region, indicating that both GB and interfacial defects for the perovskite films were effectively passivated by PbS. It is reported that the VOC of solar cells is strongly related to the recombination of defects in the bulk or on the surface of the absorber layer (Wetzelaer et al., 2015, Buin et al., 2014). Therefore the lower trap states should be responsible for the higher VOC (1.146 V) for the passivated device in the present work. To deeply understand the reason of the defects passivation, the defect activation energy (Ea) was measured by TAS accordingly. The relationship between Ea and ω0 follows the equation: ω0 = βT2exp(−Ea/kT), where the characteristic transition angular frequency (ω0) was extracted from the capacitance-frequency spectrum (as shown in Figures S8A and S8B). As reported (Chen et al., 2019b), Ea is approximately equal to the depth of the trap state energy (ET) relative to the VBM energy level (EVBM) of perovskite (Ea = ET − EVBM). The Ea values for the pristine and the PbS-passivated devices were calculated to be 0.2698 eV and 0.2329 eV, respectively, from the Arrhenius plot as shown in Figure S8C (which derived from the temperature-dependent capacitance-frequency spectra). This result indicated that the device with passivation layer has a lower trap state energy. In other words, there were fewer vacancies or interstitial traps for the passivated films. Therefore the performance of the modified device has been effectively improved by PbS passivation, especially in terms of the V. We further tested the dark J-V curves for both devices to understand the recombination mechanism in the devices (Figure S9A). It is well known that less defect in the perovskite film can decrease the recombination process and the corresponding dark saturation current density (J0) of the device (Zhang et al., 2018). J0 is an extremely important parameter and is traditionally used as an indicator of the recombination in a device. The values of J0 can be extracted from the linear fit for the semilogarithmic plot in a range near VOC from the dark J-V plots (Figure S9B). The J0 decreased from 3.86×10−8 mA/cm2 for the control device to 5.25×10−9 mA/cm2 for the device with passivation layer, indicative of a considerably lower recombination rate (Lv et al., 2018). Owing to this the VOC increases on decreasing the value of J0; significantly enhanced VOC (from 1.109 to 1.146 V) was achieved for the passivated device in the present work. The ideality factor (n) is also an important parameter to reflect the trap-assisted recombination and is commonly used to describe the dependence of applied light intensity and voltage, which can be derived from the relationship of the light intensity versus VOC (as shown in Figure 4B) (Bernechea et al., 2016). The n decreased from 1.80 for the pristine device to 1.53 for the passivated device, indicating that the trap-assisted (non-radiation) recombination process was suppressed due to the reduction of the interfacial and GB defect states of the perovskite film (Yang et al., 2016), which agrees well with the results of TAS.

Figure 4

Dynamics Analysis of the Devices

(A) Trap density in pristine and PbS-passivated perovskite.

(B) VOC versus light intensity for the control and passivated devices.

(C) TRPL decay profiles of the perovskites with and without passivation layer.

(D) Electrochemical impedance spectroscopy spectra for the devices based on the pristine and passivated perovskites.

PL spectroscopy is an important means to disclose the interfacial charge carrier transfer and extraction. Figure S10 shows that the HTM (spiro-OMeTAD) can quench the PL of the passivated perovskite film more effectively than the pristine perovskite film, which could be ascribed to the efficient charge transfer at the interface between HTM and perovskite (Tan et al., 2019). Similar results were further provided by the time-resolved PL spectroscopy (shown in Figure 4C). The fitted data are summarized in Table S2. The average PL life time (τave) for the pristine perovskite sample is 368.1 ns, indicative of its high-quality film (Zheng et al., 2017, Yang et al., 2017, Wu et al., 2017). The average PL decay time is reduced to 20.97 ns upon introducing an HTL, indicating that the HTL can effectively extract the holes from the perovskite film. Moreover, PL decay time was further decreased to 9.44 ns for the sample with passivation layer, meaning that the holes transfer faster for the passivated sample than the control one, which should be attributed to the reduced interfacial trap state and the hole transfer ability of the passivant. Similar dual function effect has been achieved for the device passivated with the PTAA (poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine]) and poly-TPD (poly(4-butylphenyldiphenylamine)) (Wu et al., 2018, Tan et al., 2019). It is known that the reduced PL decay time is beneficial for suppressing the charge recombination at the interface and for improving device performance effectively. We then carried out electrochemical impedance spectroscopy to study the interfacial carrier transfer and transport kinetics. Figure 4D shows the Nyquist plots of the control and target devices. The arcs at high frequency were ascribed to carrier transport features in bulk layer and the incomplete semicircles in low frequency represent carrier recombination process at interfaces (Zhang et al., 2018, Yang et al., 2016). The corresponding values of Rtr (transport resistance) and Rrec (recombination resistance) were fitted by the Z-view software with the given equivalent circuit and summarized in Table S3. The device with passivation layer displays a lower Rtr and higher Rrec compared with the pristine one, indicating that a faster charge-carrier transfer and lower recombination rate is obtained for the passivated device, in good agreement with the previous results.

The stability of PSCs in ambient conditions is challenged by their sensitivity to moisture and oxygen due to the hygroscopic nature of the perovskite films. We thus evaluated the stability of both devices under ambient conditions at room temperature and relative humidity of 40%–70%. Figure 5A and Table S4 show the main parameters for the unencapsulated control and passivated devices. The PCE retains more than 93% of its initial efficiency even after 1 year's storage under the ambient conditions for both devices. To obtain an in-depth understanding of the reasons for stability enhancement, the surface water contact angle (CA) of perovskite film was measured accordingly, as shown in Figure 5B. The CA for PbS-passivated perovskite film was 68.3°, much larger than 55.1° for the pristine film, indicating that the passivated film exhibits better hydrophobicity. This result is in good agreement with recently reported results that the increased CA was beneficial for inhibiting moisture penetration into the film and enhancing the overall device stability in humid environments (Li et al., 2016). Moreover, we further evaluated the thermal stability of the devices without and with the passivation layer by replacing the spiro-OMeTAD with PTAA. As shown in Figure 5C, more than 82% of its initial efficiency can be retained even after 1,000 hours of thermal aging at 85°C, demonstrating its superior thermal stability. The excellent stability of the present PSCs could be attributed to two factors: (1) the highly stable system of (Cs,FA) double-cation perovskite adopted in this work when compared with the MA-containing perovskites (Lee et al., 2015, Zheng et al., 2019, Park and Seok, 2019) and (2) the hydrophobicity of the perovskite film surface derived from PbS QDs.

Figure 5

Stability of the Perovskite Solar Cells

(A) Long-term stability of the unsealed devices based on the pristine and PbS-passivated perovskite stored in dark under ambient conditions with humidity of 40%–70%.

(B) The water contact angle on the perovskite films with and without the PbS passivation layer.

(C) Normalized efficiencies of the pristine and the PbS-passivated PSCs under continuous heating at 85°C in N2-filled glove box.

Finally, to verify the universality of our interfacial contact passivation approach, perovskites with different component have been studied (as shown in Figure S11 and Table S5). It is clear that all the key J-V parameters for the PbS QD-passivated PSCs are better than those for the pristine one independent of the perovskite component. The significantly improved device performance should be attributed to the more superior carrier transport performance and the reduced non-radiation interfacial recombination after PbS passivation. These results demonstrated the effectiveness of interfacial contact passivation by using QDs in achieving highly efficient PSCs.

Conclusion

In summary, we have developed a universal interfacial contact passivation strategy to improve both efficiency and stability of planar PSCs by introducing a semiconductor passivation layer. The interfacial charge-carrier transfer was significantly improved due to the reduced trap states and suppressed non-radiation recombination at the interface of HTL/perovskite, resulting in a VOC of 1.146 V and a stabilized efficiency of 20.3%. Moreover, the (Cs,FA) double-cation perovskite-based device exhibits excellent long-term ambient and thermal stability, showing great application promise for the future mass production of PSCs.

Limitations of the Study

In this work, we introduced PbS QDs into the PSCs to in situ passivate the interfacial defects between perovskite and hole transporting layer. The optimal device shows excellent PCE with great long-term stability compared with the control one due to the reduced non-radiative recombination. However, the devices with PbS passivation layer show decreased device performance when the concentration of PbS exceeds 0.1 mg/mL. It would be more interesting if the crystallization process of perovskite films were studied through the grazing incidence wide-angle X-ray scattering measurement, to further investigate the effects of PbS-like QDs on the crystal orientations of MA-free, cesium/FA-double-cation-based perovskite films.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.