Interfacial stabilization for inverted perovskite solar cells with long-term stability

: Perovskite solar cells (PSCs) commonly exhibit significant performance degradation due to ion migration through the top charge transport layer and ultimately metal electrode corrosion. Here, we demonstrate an interfacial management strategy using a boron chloride subphthalocyanine (Cl 6 SubPc)/fullerene electron-transport layer, which not only passivates the interfacial defects in the perovskite, but also suppresses halide diffusion as evidenced by multiple techniques, including visual element mapping by electron energy loss spectroscopy. As a result, we obtain inverted PSCs with an efficiency of 22.0% (21.3% certified), shelf life of 7000 hours, T 80 of 816 h under damp heat stress (compared to less than 20 h without Cl 6 SubPc), and initial performance retention of 98% after 2000 hours at 80 o C in inert environment, 90% after 2034 h of illumination and MPP tracking in ambient for encapsulated devices and 95% after 1272 h outdoor testing ISOS-O-1, which is among the top device performance for the inverted PSCs.


Introduction
Metal halide perovskite solar cells (PSCs) have exhibited significant progress in terms of both conversion efficiency and stability in recent years. [1][2][3][4][5][6] However, the device stability is not sufficient for the commercialization, and, hence, is more crucial than conversion efficiency at present. 7 Whether conventional n-i-p or inverted p-i-n devices are used, PSCs commonly exhibit degradation when exposed to moisture, ambient atmosphere, heat, and electric bias. 4,7,8 The degradation initiates from defect sites near the surfaces unpon exposure to light, moisture, oxygen, and heat, and these defects also initiate ion migration, resulting in reactions of the perovskite (PVK) at the charge transporting interfaces and the electrodes. 8 Of the different device architectures, inverted PSCs are of particular interest, since the absence of doped organic charge transport layer on top of the perovskite enables superior thermal stability in comparison to conventional devices, 9 and high stability under combined light and heat stressing 10 and thermal cycling. 11 In addition, they also exhibit improved stability under reverse bias compared to conventional devices. 3,9,12 It is well known that the PVK/charge transport layer, namely hole transport layer (HTL) or electron transport layer (ETL) dominate conversion efficiency and stability of devices mainly by affecting interfacial defect density and ions diffusion. 3,13 In an inverted device structure, the top PVK/ETL interface is expected to significantly influence the device stability due to its effect on moisture and oxygen penetration into the perovskite and halide ion diffusion to the electrode. It is well recognized that inverted PSCs with commonly used fullerene-based ETLs not only exhibit significant interfacial recombination losses, 6,13,14 but also have susceptibility to the oxygen and moisture ingress into the devices. 4,15 In addition, their thermal stability is limited by the aggregation of fullerene acceptor, which leads to the deterioration of contact between PVK and ETL. 4 More importantly, volatile perovskite decomposition products readily diffuse through the fullerene-based layer, 15 exacerbating PVK and electrode degradation, since almost all metals react with PVK decomposition products, 4 and leading to shunting at low reverse bias voltages. 16 Therefore, interfacial stabilization at the interface of PVK/ETL is crucial for achieving both high efficiencies and long-term stabilities in inverted p-i-n PSCs. While the performance improvements have been demonstrated by various approaches, 3,6,,13,15,17-22 including the use of different inorganic 13,17 or organic 15,[18][19][20] interfacial layers between PVK and C60, further performance improvements are still needed to bring these devices closer to commercialization. 7 Here we use a boron chloride subphthalocyanine (Cl6SubPc)/fullerene ETL to simultaneously reduce the interfacial defect density and hinder ion migration, resulting in power conversion efficiencies of 22 Phthalocyanine and porphyrin have excellent thermal and photochemical stability and have been used in different types of solar cells, including PSCs 23 and OPVs. 24,25 Phthalocyanine-and porphyrin-related molecules exhibit one intriguing aspect, namely their high iodine adsorption capacity and general capability of adsorption of various guest molecules owing to their cloud of π electrons. 26 This aspect of phthalocyaninerelated molecules has been little explored in PSCs, but potentially offers a route to inhibit ion migration and/or diffusion of perovskite decomposition products through the ETL, protecting the electrode from corrosion. Thus, we investigated the use of a chlorinated macrocyclic molecule, Cl6SubPc, which has been previously used as an acceptor in OPVs. 24,25 The chemical structure of Cl6SubPc and the device architecture of p-i-n planar PSCs are shown in Figures 1a and 1b. We use inorganic Cu doped NiOx nanoparticles as the HTL and C60 as the ETL, 27 and a Cs + , formamidinium (FA + ) and methylammonium (MA + ) (CsFAMA) cations mixed perovskite 28,29 as the active layer.
We investigated both CsFAMA perovskites, labeled as 3D PVK, and CsFAMA perovskites with the surface treated with phenyl ethyl-ammonium iodide (PEAI) solution, labeled as 3D/2D PVK (for experimental details, see Methods, Supplementary Information). The use of a PEAI solution treatment to form a 2D capping layer on top of the perovskite has been investigated, since a low dimensional perovskite capping layers passivates surface defects and improves the perovskite stability. 30 The Cl6SubPc and C60 bilayer ETL was then thermally evaporated on top of the perovskites.
The performances of different devices are shown in Figure 1 c-h and performance of 3D PVK devices for both ETLs is inferior to that of 3D/2D devices, in agreement with literature reports. 30 We also observe that CsFAMA 3D PVK devices with thin Cl6SubPc (20 nm) ETLs exhibited a relatively low efficiency of 19.3%, due to the low Jsc and Voc. This may arise from the mismatch of the energy levels at the interface which hinders electron transfer, due to the relatively high LUMO level (~3.8 eV) 24   d-e) J-V curves of the optimal 3D/2D/C60 (d) and 3D/2D/Cl6SubPc/C60 (e) devices under reverse and forward scan directions; f) Device performance statistics for 3D/2D/C60 and 3D/2D/Cl6SubPc/C60 devices; g) Steady power output (SPO) of the optimal 3D/2D/C60 and 3D/2D/Cl6SubPc/C60 devices test at the bias of maximum power point; h) EQE spectra for the optimal 3D/2D/C60 and 3D/2D/Cl6SubPc/C60 devices.
Since non-fullerene ETLs offer a compromise between efficiency and stability when employed as interlayer between the perovskite and fullerene, 20 we also investigated the performance of devices with Cl6SubPc/C60 bilayer ETL for different show that Cl6SubPc is amorphous with no obvious diffraction features of Cl6SubPc for the 3D/2D PVK/Cl6SubPc sample. We note that the arc shape at q～0.9 Å -1 suggests a PbX2 (PbI2 and/or PbBr2) residue. 1D azimuthal integrated scattering profiles are shown in Supplementary Figure 8. We can observe a decrease in PbX2 content (peak ratios of signature peaks at q～0.91 Å -1 and q～1.0 Å -1 are assigned to the (001) plane of PbX2 crystal and (110) plane of 3D PVK) in 2D/3D PVK compared to 3D PVK. The improvement in crystallinity after PEAI treatment is in agreement with the surface induced secondary grain growth observed upon treating halide perovskite surfaces with organic ammonium solution. 33 Surprisingly, the deposition of Cl6SubPc also induced further changes in the crystal structure. We observe reduced FWHMs (full width at half maximum) of the reflection at q ~ 0.28 Å -1 and q ~ 0.38 Å -1 (OOP) of the 2D perovskites Cl6SubPc, increased ratio of the areas of the reflection at q ~ 0.77 Å -1 for 2D PVSK to the reflection at q ～ 1.0 Å -1 for CsFAMA PVK, as well as reduced peaks of the hexagonal non-PVK phase (δ phase) at q ~ 0.85 Å -1 (OOP) and PbX2 at q ~ 0.91 Å -1 , indicating improved crystallinity. One possible mechanism behind the observed phenomenon is that Cl6SubPc caused surface-induced secondary grain growth after the deposition of organic molecules on the surface of the perovskite to minimize the interfacial energy, facilitated by the low activation energies for ion diffusion, low elastic modulus and consequent liquid-like behavior of the soft perovskite lattice. 33   showing clearly the layered structure of the 2D perovskite with interlayer distance of 7.1Å, which is consistent with the values from the single crystal structure; g) Electron energy loss spectroscopy (EELS) mapping of the fresh devices with 3D/2D perovskite and Cl6SubPc/C60 ETLs. The Cl and N signal demonstrate that the Cl6SubPc is mixed with the C60 film.
The interactions between Cl6SubPc and perovskite could be expected from findings in a previous report on the strong chemical interactions between the copper phthalocyanine and the perovskite, 37 and are consistent with observed solar cell performance improvements and improved crystallinity. Possible mechanisms of defect passivation include the formation of Pb-Cl bonds, 38 and the interaction between the perovskite and the pyrrole ring. 37 We investigated these possibilities using DFT and molecular dynamics simulations, as described in the Supplementary Note 1. We found that two Cl ions bond to Pb in the PbI2-terminated perovskite (001)  respectively. Thus, the use of Cl6SubPc is expected to significantly increase the strength of interaction between the perovskite and ETL, due to hydrogen bonding (2.7 times stronger bonding with Cl6SubPc compared to hydrogen bonding between PEA molecules), Pb-Cl interactions, and the strong bonding of iodine to Cl6SubPc (more than 20 times stronger bonding with iodine compared to C60). The formation of strong bonds between the perovskite and charge transport layer are beneficial for device performance because they inhibit ion migration, 1 and increase resistance to degradation due to oxygen and moisture. 17 In addition, the existence of strong interactions between the perovskite and interfacial layer on one side, and interfacial layer and fullerene derivative on the other side was found to result in defect passivation and substantial improvement in stability. 18 Interfacial bonding was confirmed to play a role in stability improvements for different molecules, 37,39 including copper phthalocyanine. 37 Thus, the stronger bonding achieved by using Cl6SubPc interfacial layer can contribute to defect passivation 22 and to increase the device stability. 4,19,37,39 Therefore, we performed comprehensive investigation of the effects of the incorporation of Cl6SubPc on the charge recombination dynamics, the photoluminescence (PL) decay dynamics were measured to probe the interfacial recombination at the 3D/2D PVK/ETLs interfaces, followed by comprehensive stability testing. As shown in Figure 3  where Vbi is given by the intersection on the bias axis. 41 The trap density (Figure 3e) obtained is in the range expected for a polycrystalline halide perovskite film. 42 We observe a reduction in the trap state density for both trap states with depths ~0.30-0.42 eV and 0.50-0.60 eV in the Cl6SubPc/C60 device in comparison to the C60 control one by approximately one order of magnitude (for example, from 1.5310 17 cm -3 eV -1 to 3.4310 16 cm -3 eV -1 for the trap state ~0.30-0.42 eV). We attribute these to the decrease of traps in grain boundaries and n-type interfaces, respectively, 29 indicating that the PVK/ETL interface quality has a significant influence on the formation of defects in the planar PSCs. The dependence of the Voc on light intensity is shown in Figure 3f.
The diode ideality factor (Nd) can be calculated from the Voc dependence on illumination intensity. 43 Nd is reduced from 1.33 KT/q for the C60 ETLs device to 1.12

KT/q for the Cl6SubPc/C60 device. A lower ideality factor indicates lower trap-assisted
Shockley-Read-Hall monomolecular recombination, 44,45 which is consistent with the increased Voc, 6 lower trap densities, and the suppression of interfacial recombination for devices with Cl6SubPc/C60 ETLs. From the energy levels of different materials in the devices, shown in Figure 3, we can observe that the electron collection would be more favorable when using C60 ETL, which likely accounts for observed higher efficiency of the devices containing only C60 ETL compared to only Cl6SubPc ETL. Figure 3. a-c) PL decay dynamics for the 3D/2D perovskites with different interlayers as noted in the figures. The excitation from glass and perovskites sides were recorded for comparison. "PVK side" indicates the excitation is from perovskite side and "Glass side" means the excitation is from glass side. d) UPS spectra for the 3D and 3D/2D perovskites as well as C60 and Cl6SubPc ETLs prepared on Si substrates; e) Energy level alignments of the various layers, the VBM, CBM and EF values were calculated from the UPS results, data for BCP/Ag were cited from literature; f) t-DOS characteristics for C60 control and Cl6SubPc/C60 ETLs based PSCs; g) Open circuit voltage (Voc) as function of illumination intensity for the C60 control and Cl6SubPc/C60 ETLs based PSCs.
After examining the charge recombination dynamics in detail, we performed comprehensive stability tests since interface degradation is a major contributor to the short and long-term PSC stability. 46 We performed stability tests in an inert environment to obtain information on the intrinsic stability independent of the encapsulant used. 7 During continuous one-sun illumination, 95% of the initial PCE is retained after 1200 hours, as shown in Figure 4a  We also performed outdoor testing since outdoor stability studies of PSCs in general have been scarce, 7 and there have been no outdoor tests of inverted devices to date.
Superior stability of the devices with Cl6SubPc/C60 ETLs compared to C60 ETLs is also confirmed in outdoor stability tests following ISOS-O1 protocol, where T95 of 1272h for device with Cl6SubPc is obtained, as shown in Figure 4d.  It should be noted that all the stability tests were performed on devices with 3D/2D PVK layers. Since it has been reported that bulky organic cations in the 2D PVK structure of the 3D/2D PVK capping layers improves device stability by inhibiting ion migration, 8,30,47,48 we also compared the stability of devices with 3D and 3D/2D PVK and C60 ETL, as shown in Supplementary Figure 19. Thus, we can observe that while 3D/2D PVK results in improved stability compared to 3D PVK active layer, as expected, 8,30,35,47,48 it is not sufficient to enable significant performance improvement under realistic operating conditions. From the comparison of the stability of the devices with 3D PVK active layer and Cl6SubPc/C60 and C60-only ETLs, as shown in Supplementary Figure 20. We see that the stability of the devices with 3D PVK and Cl6SubPc/C60 is clearly better than that of 3D PVK and C60-only. Thus, we focus on the effect of the PVK/ETL interface on device stability in devices with 3D/2D perovskite and Cl6SubPc/C60 bilayer ETL.
In addition to stability tests at elevated temperature, humidity and/or illumination, we performed stability testing under reverse bias, since this type of test is a strong indicator of susceptibility to ion migration and electrode corrosion. 16,17 To further enhance susceptibility to ion migration, the devices were illuminated during testing, since illumination and electrical bias both promote ion and defect migration, and accelerate harmful chemical reactions. 7 In the device without Cl6SubPc (Supplementary Figure 21) we can see clear damage to the Ag electrode, and also Ag deposition on the ITO side, in agreement with the literature, 16 while in the devices with Cl6SubPc/C60 (Supplementary Figure 22) no obvious electrode damage can be seen, confirming further that Cl6SubPc is capable of blocking ion migration.
To investigate the mechanism behind the effect of Cl6SubPc in device stability tested above, we performed ToF-SIMS, EELS mapping, and XPS measurements on devices with different ETLs. From obtained SIMS profiles ( Figure 5 a and b, and Supplementary Figure 23) of the C60-only and Cl6SubPc/C60 devices before and after aging, we observe the iodine at the initial sputtering atmosphere of the aged C60only device in contrast to the fresh one, along with the variation of silver information.
In contrast, the Cl6SubPc/C60 device shows negligible change in those profiles.
Information on other key elements or species involved in the devices before and after aging are also shown in Supplementary Figure 23.  This shift is consistent with electron transfer, which likely occurs due to high electron affinity of boron, similar to a previous report of electron transfer between SubPc and C60, which also resulted in a similar shift of the B 1s peak. 50      image of the marked area in (d), showing clearly the layered structure of the 2D perovskite with interlayer distance of 7.1Å, which is consistent with the values from the single crystal structure; g) Electron energy loss spectroscopy (EELS) mapping of the fresh devices with 3D/2D perovskite and Cl6SubPc/C60 ETLs.
The Cl and N signal demonstrate that the Cl6SubPc is mixed with the C60 lm. Figure 3 a-c) PL decay dynamics for the 3D/2D perovskites with different interlayers as noted in the gures. The excitation from glass and perovskites sides were recorded for comparison. "PVK side" indicates the excitation is from perovskite side and "Glass side" means the excitation is from glass side. d) UPS spectra for the 3D and 3D/2D perovskites as well as C60 and Cl6SubPc ETLs prepared on Si substrates; e) Energy level alignments of the various layers, the VBM, CBM and EF values were calculated from the UPS results, data for BCP/Ag were cited from literature; f) t-DOS characteristics for C60 control and Cl6SubPc/C60 ETLs based PSCs; g) Open circuit voltage (Voc) as function of illumination intensity for the C60 control and Cl6SubPc/C60 ETLs based PSCs.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Supportinginformation.pdf