Nitrogen‐doped tin oxide electron transport layer for stable perovskite solar cells with efficiency over 23%

Tin oxide has made a major breakthrough in high‐efficiency perovskite solar cells (PSCs) as an efficient electron transport layer by the low‐temperature chemical bath deposition method. However, tin oxide often contains pernicious defects, resulting in unsatisfactory performance. Herein, we develop high‐quality tin oxide films via a nitrogen‐doping strategy for high‐efficiency and stable planar PSCs. The aligned energy level at the interface of doped SnO2/perovskite, more excellent charge extraction and reduced nonradiative recombination contribute to the enhanced efficiency and stability. Correspondingly, the power conversion efficiency of the devices based on N‐SnO2 film increases to 23.41% from 20.55% of the devices based on the pristine SnO2. The N‐SnO2 devices show an outstanding stability retaining 97.8% of the initial efficiency after steady‐state output at a maximum power point for 600 s under standard AM1.5G continuous illumination without encapsulation, while less than 50% efficiency remains for the devices based on pristine SnO2. This simple scalable strategy has shown great promise toward highly efficient and stable PSCs.


| INTRODUCTION
Halide perovskite solar cells (PSCs) have attracted great attention in the last decade; the power conversion efficiency (PCE) has increased rapidly from 3.8% to 25.5%. [1][2][3][4] As a key component, electron transport layer (ETL) plays an important role in the performance of devices, which extracts electrons from the perovskite light-absorbing layer to transfer them to the cathode, prevents hole from transferring to the cathode, and reduces the interface charge accumulation and nonradiation recombination. [5][6][7] Meanwhile, the long-term stability of PSCs is strongly affected by ETLs. [8,9] Various metal oxide semiconductors have been employed as ETLs, such as TiO 2 , [10] ZnO, [11] WO 3 , [12] and SnO 2 . [13,14] Especially, the latter has been investigated extensively in PSCs due to its high transmittance, wide bandgap, and high electron mobility. SnO 2 thin films can be fabricated by electrochemical deposition, [15] atomic layer deposition (ALD), [16] sol−gel methods, [17][18][19] and so forth. Among them, chemical bath deposition (CBD) method is the most promising method to prepare the SnO 2 ETLs, as it is a simple and low-temperature process and easy to scale up. However, SnO 2 crystals made by CBD naturally contain defects with undesirable charge-trap states, resulting in nonnegligible hysteresis, low fill factor (FF), and poor stability for PSCs. [2,20] Element doping can effectively regulate defects in SnO 2 nanocrystals, as well as the conductivity and band structure. The SnO 2 nanocrystals have been doped with metal elements, such as Sb, [21] Mo, [22] Al, [23] Ga, [24] Zr, [25] and La, [26] and nonmetal elements, such as F, [27] P, [28] and Cl. [29] Bu et al. demonstrated that element doping improved the conductivity of SnO 2 ETLs, decreased the energy barrier between SnO 2 /perovskite interface, and enabled an increase in the PCE from 19.01% to 20.69%. [30] Recently, cobalt chloride hexahydrate (CoCl 2 ·6H 2 O) was introduced into SnO 2 ETLs to manage the detrimental defects and band structure. With the favorable energy level alignment and charge extraction, an enhanced open-circuit photovoltage (V oc ) up to 1.20 V was achieved along with an improved stability. [31] Although most of the research was focused on the ETLs prepared by spin-coating method that is unscalable, the doping technique has shown its superiority in improving the performance of SnO 2 ETLs.
Here, we report highly efficient and stable planar PSCs based on N-doped SnO 2 ETLs through the introduction of NH 4 Cl in a CBD method, reaching a PCE over 23% with a high FF of 0.82 and a V oc of 1.155 V, which benefit from favorable energy level alignment between ETL and perovskite, accelerated extraction and transport of electrons, and suppressed charge recombination. The N doping also improves the stability due to the reduced defects, retaining 97.8% of the initial efficiency after steady-state output at a maximum power point for 600 s. This strategy represents a breakthrough as it provided a simple and scalable way to deposit high-quality SnO 2 ETL for efficient and stable PSCs with efficiencies over 23%.

| RESULTS AND DISCUSSION
The SnO 2 film was fabricated by the CBD method; that is, soaking FTO glass substrates in an SnCl 4 aqueous solution bath at 90°C for 2 h. [32] The properties of the SnO 2 film are extremely important, directly affecting the final performance of prepared PSCs. Doping is a powerful technique to tune the properties of semiconductors. Here NH 4 Cl was chosen as the N-doping source, and the SnO 2 films prepared by the addition of 250 mg NH 4 Cl in bath solution are labeled as N-SnO 2 . Figure 1A,B shows the scanning electron microscope (SEM) images of the SnO 2 films on FTO prepared by CBD. Both pristine SnO 2 and N-SnO 2 films are quite uniform and dense, indicating that the addition of NH 4 Cl does not alter the morphology of SnO 2 films, so it is not surprising to see that the transmittance of the SnO 2 films is almost the same, as clearly shown in Figure S1.
To check whether N can be doped into SnO 2 by adding NH 4 Cl, the x-ray photoemission spectroscopy (XPS) measurement was carried out. Figure 1C,D summarizes the XPS results on the pristine SnO 2 and the N-SnO 2 . The corresponding Sn 3d 5/2 peaks appear at 487.06 and 486.89 eV, respectively ( Figure 1C). [33] Sn 3d 5/2 peak shifts to lower binding energy with the N doping due to the lower electronegativity of N than O. [28] Meanwhile, the N-doped SnO 2 has a much stronger N signal than the pristine SnO 2 ( Figure 1D), confirming the doping of N. A very weak peak observed in the undoped sample is due to the contamination from the ambient; that is, molecularly chemisorbed γnitrogen, which is confirmed by other researchers. [34,35] The N 1s high-resolution spectrum could be deconvoluted into two peaks, one peak at 399.6 eV attributed to anionic N incorporated in SnO 2 forming O-Sn-N linkages, and the other peak at 400.4 eV attributed to oxidized nitrogen. Quantitative estimates of the N/Sn ratios in N-SnO 2 film is 2.0 at.%. The position of O 1s peak exhibits no significant change as shown in Figure S2, implying that the nature of oxygen is similar. As it is well known, proper band alignment is crucial for efficient ETLs in PSCs. Ultraviolet photoelectron spectroscopy measurement was performed to investigate the conduction band minimum (CBM) and valence band maximum (VBM) positions of the SnO 2 films and perovskite as displayed in Figure 2A It is found the CBM of the pristine SnO 2 is slightly higher than that of the perovskite, which will somehow impede the extraction of electrons in the perovskite to the ETL. With the doping of N, the CBM (−4.32 eV) shifts downward a little lower than that of the perovskite (−4.28 eV), as clearly shown in the energy level diagram ( Figure 2C) that is drawn according to the E g of the N-SnO 2 (3.87 eV) and perovskite (1.57 eV) extracted from Tauc plots ( Figure S3). Apparently, the N-SnO 2 would show a better performance than the pristine SnO 2 ETL, with a reduced interface energy barrier and suppressed nonradiative recombination.
Another important property of ETLs is electron mobility. To check the effect of the N doping on the electron mobility of the SnO 2 films, the space charge-limited current measurement ( Figure 2D) was conducted, [37] showing that electron mobility of the N-SnO 2 is 7.153 × 10 −3 cm 2 V −1 s −1 , a little higher than that (6.990 × 10 −3 cm 2 V −1 s −1 ) of the pristine SnO 2 . ETLs with higher electron mobility could effectively transfer electrons, reduce the charge accumulation at the interface of ETL/perovskite, and then improve the FF of the device. Therefore, the N-SnO 2 is a good candidate for ETLs.
The quality of the ETL, including hydrophilicity, morphology, and attached chemical groups, directly determines the quality of the perovskite layer. [7] The UV-Vis spectra ( Figure S4a) show a slightly stronger absorption of the perovskite deposited on the N-SnO 2 ETLs. Similarly, the XRD spectra in Figure S4b shows that doping slightly enhances the crystallinity of the perovskites. The SEM images of the perovskites (Figure 3A,B) and the corresponding statistics ( Figure S5) show that the grain size of perovskites on the N-SnO 2 (1 μm) is larger than that on the pristine SnO 2 (0.8 μm). In general, a hydrophobic surface will result in enlarged grains of the deposited perovskites. [38] The contact angle images ( Figure S6) clearly show a higher hydrophobicity of the N-SnO 2 surface.
To further examine the electron extraction ability of the ETLs, the steady-state photoluminescence (PL) spectra ( Figure 3C) and normalized time-resolved PL (TRPL) spectra ( Figure 3D) of the perovskite deposited on bare glass, SnO 2 , and N-SnO 2 were measured. The strong steady PL of the perovskite on the bare glass indicates the high quality of the perovskite crystals. With the introduction of the SnO 2 ELTs, the PL is obviously suppressed. Compared with the pristine SnO 2 , the N-SnO 2 has a stronger PL quench effect due to the better electron extraction and transportability. For the TRPL, the perovskite deposited on the bare glass has a slight decay process and a long lifetime, with an average lifetime of~4.0 µs, while the average lifetime (Table S1) of the perovskites on ETLs was reduced by an order of magnitude, 0.313 µs (the pristine) and 0.271 µs (the doped), respectively. This further indicates that the N-SnO 2 is a much better electron extraction layer compared with the pristine SnO 2 . Then the SnO 2 ETLs were used to fabricate the complete devices, the cross-sectional SEM images of which are shown in Figure 4A,B. The related photovoltaic characteristics are listed in Table S2 and displayed in Figure S7. Figure 4C shows the champion J-V curves in forward and reverse scans under the illumination of AM1.5G 1 sun, and the corresponding photovoltaic characteristics are listed in Table 1. It is found that the pristine SnO 2 devices achieve a PCE of 20.55%, a V oc of 1.081 V, a short-circuit photocurrent (J sc ) of 24.42 mA/cm 2 , and a FF of 77.8% for the reverse scan. Interestingly, for the N-SnO 2 devices, the performance is significantly improved, achieving a high V oc of 1.155 V, a J sc of 24.81, a FF of 81.7%, and a PCE of 23.41% for the reverse scan. This is mainly due to the suitable interface contact and energy level alignment for the N-SnO 2 and perovskite as discussed, which reduces the interface defect density, accelerates the extraction of electrons, and reduces the energy loss caused by nonradiative recombination. The corresponding external quantum efficiency spectra are shown in Figure S8, and the integrated current density of devices based on the pristine SnO 2 and the N-SnO 2 is 23.93 and 24.10 mA/cm 2 , respectively, consistent with J sc from J-V curves.
To further understand the photovoltaic response and extract transfer resistance in the complete devices, electrical impedance spectroscopy was employed. The arc in the Nyquist plot ( Figure 4D) of the devices measured at 0.8 V under dark conditions from 2 MHz to 0.5 Hz represents the charge recombination at the interface. [7] The N-SnO 2 devices have a recombination resistance (R rec ) of 3597 Ω, much larger than that (1122 Ω) of the pristine SnO 2 devices, revealing that the N-SnO 2 devices have a faster electron extraction and lower nonradiative recombination at the interface between the perovskite film and ETL.
The stability of the pristine SnO 2 and N-SnO 2 devices was also evaluated. The steady-state power output at a maximum power point was measured under standard AM1.5G 1 sun illumination for the unencapsulated devices, as shown in Figure 4E  97.8% of the initial efficiency after 600 s, while the pristine SnO 2 device can only remain less than 50% of the initial efficiency. In addition, the devices were taken 20 cycles of J-V scan continuously under one sun illumination. The N-SnO 2 device shows a good stability, retaining~98% of the initial efficiency, much better than the pristine SnO 2 with only 79% remaining ( Figure S9). Impressively, the N-SnO 2 devices have good storage stability in the dark without any decrease in efficiency after 500 h storage, as shown in Figure 4F. These results demonstrate that the doping of N greatly improves the stability of SnO 2 deposited by CBD, and would be a useful strategy for scaling up PSCs.

| CONCLUSION
In summary, an effective strategy by employing NH 4 Cl as N source doping into CBD SnO 2 has been developed for efficient and stable ETLs. The optimized SnO 2 film shows better electron mobility and electron extraction ability, and significantly increases the voltage and FF of the PSCs by effectively reducing interfacial charge accumulation and nonradiative recombination.
Remarkably, the PCE of the champion device based on the N-SnO 2 ETL has been improved to 23.41% from 20.55%. The stability of the N-SnO 2 device without encapsulation is enhanced to retain 97.8% of the initial efficiency after steady-state output at a maximum power point for 600 s, much better than the pristine device that has only 50% efficiency left. This study provides a promising strategy toward efficient and stable PSCs with high-quality scalable ETLs.