Spontaneous Formation of 1D/3D Perovskite Heterojunctions for Efficient Inverted Perovskite Solar Cells

Interfacial modification is a key strategy for improving the performance of perovskite photovoltaic devices. While the modification of the top surface of the perovskite active layer is well established, engineering of the buried interface is highly challenging. Here, the spontaneous formation of a 1D/3D perovskite heterojunction at the buried interface of a perovskite active layer by incorporating choline acetate alongside the perovskite precursors is reported. Importantly, extensive spectroscopic and microscopic characterization and solid‐state nuclear magnetic resonance experiments demonstrate the formation of phase‐pure 1D and 3D domains. The 1D/3D junction results in a suppression of the defect states and an improved energetic level alignment at the buried interface, leading to a maximum power conversion efficiency of >24% when incorporated in inverted architecture perovskite solar cells. This work introduces a versatile approach to the modification of the buried interface of the perovskite active layer.


Introduction
Metal halide perovskites have emerged as one of the most promising contenders for future photovoltaic technologies. [1][4][5][6][7] This is related to the observation that the surface of the perovskite layer contains up to DOI: 10.1002/aenm.202304126 100 times more defects than its bulk. [8]pecifically, energetically deep-level traps lead to increased non-radiative recombination, thus reducing the open-circuit voltage (V OC ) and hence the power conversion efficiency (PCE). [9,10]A mismatch of the energy bands at the interfaces may also decrease device performance. [11,12]hese loss mechanisms have motivated the development of a vast array of surface modification strategies.These include, for example, the treatment of the perovskite surface with a passivating agent, [5,13] mechanical polishing of the perovskite surface, [8,14] modification of the surface energetics, [15,16] formation of a low dimensional (LD) perovskite layer [17] and many others. [17,18]mong these strategies, the formation of LD/3D heterostructures is a particularly powerful approach, since it not only makes it possible to passivate point defects, but also reconstruct the 3D perovskite surface by eliminating unwanted surface crystal phases and modifying morphological defects such as pinholes and cracks.Furthermore, a rational design of the LD structure (e.g., composition and thickness) enables control over the interfacial charge carrier extraction kinetics.[21] However, the use of 2D cations may lead to the formation of quasi-2D phases with various n values when combined with 3D cations in different proportions. [17,22]This increases the process complexity for obtaining a dimensional junction with a certain, predetermined composition and introduces difficulties in forming a junction with stable phases, leading to uncertainty in the electrical properties of the LD layer and potentially, device instability. [23]This challenge is especially significant when attempting to construct heterojunctions at the buried interface of the 3D perovskite layer.As a consequence, both the LD/3D formation strategy as well as other modification approaches focus on engineering the top surface of the perovskite layer, while methods to modify the buried perovskite interface remain scarce.This is partly related to the limitations imposed by the solution processing of perovskite layers, by which most high-efficiency PSCs are fabricated.The bottom interface in PSCs is difficult to alter after film deposition, making it necessary to modify it prior to deposition, while still maintaining a suitable surface energy in order to ensure a high degree of wettability for the subsequent formation of a compact perovskite layer.Another contributing factor is the fact that while the top perovskite surface is easily accessible for microscopic and spectroscopic characterization, the buried interface is challenging to probe.Consequently, it is difficult to explore the internal mechanisms of buried interface modification and to directly correlate it to changes in device performance.26][27][28] In this work, we focus on the development of a strategy for forming an LD/3D junction as a modification route for engineering the buried surface of 3D perovskite layer.We demonstrate the spontaneous formation of a 1D/3D heterojunction perovskite layer fabricated using a two-step deposition method by adding choline acetate (CA) into the lead iodide (PbI 2 ) precursor deposited in the first step.[31] Here, we found that the 1D choline perovskite structure is stable and is phase-separated to the 3D perovskite.The 1D phase formed at the buried interface not only serves as a passivating layer, but also introduces an interfacial dipole at the 1D/3D interface promoting the extraction of the holes from 3D perovskite to the hole transporting layer (HTL).Consequently, the 1D/3D devices reach a maximum PCE of 24.03% with negligible hysteresis, significantly surpassing the performance of reference devices (22.3%) of the same device architecture.This strategy can be applied to different compositions of perovskite photovoltaic devices and motivates the development of new strategies for the design of efficient inverted PSCs.

Spontaneous Formation of 1D/3D Perovskite Heterojunctions
Figure 1a illustrates the fabrication procedure of the perovskite active layer based on the previously reported "two-step" method.
In short, in the first step of the fabrication, a layer of PbI 2 is spincoated and converted into a perovskite layer in the second step by spin-coating a solution of formamidinium (FA) dominated organic halide precursors.In this work, CA is added into the PbI 2 precursor solution with different concentrations (10, 20, and 30 mg mL −1 ).In the following, the corresponding CA-including PbI 2 films and perovskite films are abbreviated as "PbI 2 -CAconcentration" and "CA-concentration", respectively.The chemical structure of CA is shown in Figure 1b.The deposited perovskite layers are then integrated into photovoltaic devices with an inverted architecture (Figure 1c).We utilized a mixed selfassembled monolayer-based hole transporting layer on the anode which was recently shown to enhance device performance, [32,33] and a fullerene-based electron extraction at the cathode similar to previous works. [34,35]o understand the impact of CA on the formation process of the perovskite layer, we first characterized the surface morphology of the PbI 2 layer by scanning electron microscopy (SEM).We found that with increasing amounts of CA, the surface morphology of the PbI 2 films changed significantly (Figure S1, Supporting Information).To explore these changes, we also separately introduced choline iodide and lead acetate into the lead iodide layers and examined the surface morphology by SEM (Figure S2, Supporting Information).The results suggest that at lower CA concentrations, the change in morphology originates from the introduction of acetate, while at higher concentrations, it is due to the combined effect of choline and acetate.[38] This is consistent with the fact that the PbI 2 film was formed without thermal annealing, resulting in PbI 2 mainly existing in the form of complexes with DMSO or as amorphous PbI 2 . [36]On the other hand, the introduction of CA into the PbI 2 film led to the emergence of a prominent diffraction peak at 2 ≈10.1°.To determine the origin of this peak, we investigated the formation of LD CA-based perovskite phases by mixing CA with PbI 2 at a molar ratio of 1:1 in a DMF solution with the addition of HI.This is a common method for synthesizing LD perovskite powders. [39]The photographic images of the resultant LD powder, as well as the precursors (CA and PbI 2 ), are shown in Figure S4 (Supporting Information).The X-ray diffraction pattern (XRD) pattern of the resulting powder exhibited a clear diffraction peak at 2 ≈10.1°, coinciding with the reflections observed in the CA-PbI 2 films (Figure S3a, Supporting Information).The XRD pattern of the powder is consistent with a hexagonal 1D choline-based perovskite ChPbI 3 (Ch = choline) reported in literature. [30]The result suggests that the introduction of CA into the PbI 2 solution leads to the formation of ChPbI 3 even without annealing.
We note that the mechanism for the formation of ChPbI 3 upon the addition of CA into the PbI 2 solution is expected to be slightly different from the formation of the powder as described above since the former does not include HI.To explore this mechanism, PbI 2 films with and without CA were analyzed by X-ray photoemission spectroscopy (XPS).The O 1s spectrum of the PbI 2 -CA-20 sample reveals the presence of several oxygen species, including C─O (532.1 eV) and C═O (533.7 eV) species that originate from the carboxylate ions (COO − ) of the acetate (Figure S5a, Supporting Information). [40]We note that these two peaks can also include contributions from the hydroxyl moiety in the choline cation due to their similar binding energies.Another peak at lower binding energy (≈530 eV) can be assigned to the Pb─O species suggesting the formation of Pb(OAc) 2 in the PbI 2 -CA-20 sample. [40]In contrast, the intensity of all peaks in the control PbI 2 film is significantly lower.To exclude possible contributions from surface adsorbed water and oxygen, the samples were etched for 1 min using argon gas clusters and the corresponding O 1s spectra are shown in Figure S5b (Supporting Information).While no oxygen peaks are visible in the PbI 2 film, in the PbI 2 -CA-20 sample the same series of peaks as at the surface is observed, albeit at a lower intensity.This indicates the presence of CA and Pb(OAc) 2 -also in the bulk of the PbI 2 -CA-20.The formation of Pb(OAc) 2 is further consistent with a slight blue-shift of the absorption spectra observed for CA-containing PbI 2 films (Figure S3b, Supporting Information) since Pb(OAc) 2 is a wide bandgap material.The presence of Pb(OAc) 2 and the evidence for the presence of the 1D ChPbI 3 by XRD allows us to propose that the following reaction takes place (eq.( 1)) upon the addition of CA into PbI 2 : The conversion of the CA-PbI 2 films into perovskite layers was achieved during the second spin-coating step, in which a solution of FAI and MAI is cast on top of the PbI 2 films, and subsequently annealed (Figure 1a).For both the reference and the CA-20 samples, the O 1s spectra of the perovskite films (Figure S5c, Supporting Information) contain only one peak at ≈532 eV, which is associated with a hydroxyl group.This peak can originate from both absorbed water and, in the case of CA-20, from the hydroxyl group in the choline cation.Importantly, the absence of the species assigned to lead acetate suggests that upon the deposition of the organic halides, it is converted to a perovskite via the following proposed reaction: Based on these reaction schemes, we propose that upon the addition of the CA, the 1D and 3D perovskites are formed sequentially during the first and second steps of the spin-coating, respectively.The XRD measurements on both reference and CA-FA x MA 1-x PbI 3 layers confirmed the formation of a high-quality 3D crystalline perovskite (Figure S6, Supporting Information).Similar to the results of CA-PbI 2 , the corresponding perovskite layers also exhibit a clear signature of a 1D phase.The intensity of the 10.1°peak is also enhanced with the increasing CA amount.To investigate in detail the presence of the 1D phase in the perovskite layers, we characterized the microstructure and crystal structure at both their surface and buried interface.The latter was exposed by a mechanical peeling method, which was reported to completely preserve the morphology of the buried surface of the perovskite film. [24,25]In this method-illustrated in Figure 2a-the perovskite layer is coated with an epoxy glue and capped with a glass substrate.Next, the layer is peeled off the original substrate to reveal the buried interface.The morphology of the top and buried surfaces of the films were characterized by SEM (Figure 2b-e).While the microstructure of the top surface remained largely unaffected, the buried surface exhibited significant differences.In Figure 2c, we observe many line-shaped structures at the buried surface of the reference sample, which are accompanied by many noticeable voids.To identify the origin of these structures, we performed Energy Dispersive X-ray (EDX) measurements (Figure S7, Supporting Information) that revealed that they primarily consist of Pb and I elements, but do not contain N, so we speculate that these structures are unconverted PbI 2 .On the other hand, samples fabricated with CA resulted in a much more homogeneous microstructure without voids and the aforementioned line-shaped PbI 2 .To investigate the long-range order at the top and bottom surfaces of the CAmodified films, we performed grazing incidence XRD (GIXRD) experiments by varying the incidence angle Ω from 0.5°to 2.0°( Figure 2f,g), thus enabling to vary the probing depth of the XRD characterization.Measurements performed on the top surface of the perovskite layers (Figure 2f) revealed essentially no reflection at 2 ≈10.1°regardless of Ω, which suggests that nearly no 1D phase is present at the top of the perovskite layer.On the other hand, a clear reflection associated with the 1D phase can be observed when the buried surface is characterized.This reflection shifts to a slightly larger 2 at low Ω angles, suggesting a slight lattice compression of the 1D phase formed directly near the substrate.
To further explore the composition of the perovskite layers at both the top and bottom interfaces, we performed XPS measurements.The C 1s spectra of the reference and CA-20 samples are shown in Figure S8a,c (Supporting Information), respectively.In all cases, the peak at 288.5 eV is assigned to C═N double bonds, whilst the peak at 287 eV originates from the C─N single bonds (highlighted in orange for clarity), while the peak at 285 eV is assigned to C─C bonds.We observe no notable difference between the spectra collected at the top surface of the reference and the CA-20 samples, suggesting they are compositionally similar.On the other hand, at the buried interface a significant difference is evident in the intensity of the C─N peak at 287 eV: the contribution of this peak in the CA-20 sample is stronger than that of the reference sample, while the intensities of the other two peaks (C═N and C─C) are similar.N 1s spectra of the reference and CA-20 samples are also shown in Figure S8b,d (Supporting Information).In these spectra, the peak centered at 400.5 eV is assigned to C═N double bonds, whilst the peak at 402.5 eV is attributable to C─N single bonds.Similar to the C 1s spec-tra, the signal from C─N bonds at the buried interface of the CA-20 sample is far clearer than that at the reference sample.The increase in the C─N signal is a consequence of the presence of the choline cations at the buried interface of the CA-20 samples.Taken together with the GIXRD and SEM results, these measurements confirm that the formed 1D ChPbI 3 phase tends to gather at the bottom surface of the 3D perovskite active layer.
To explore the factors that influence the accumulation of the 1D ChPbI 3 perovskite at the buried interface, we first examined the potential role of the choice of HTL.We utilized NiO x , PTAA, and MeO-2PACz to represent various material classes typically used as HTLs, i.e., metal oxides, polymers, and self-assembled molecules, respectively.In all three cases, we exposed the buried interfaces using the method illustrated in Figure 2a and performed GIXRD measurements with Ω = 0.5°, 1.0°, and 2.0°to probe for the presence of the 1D phase (Figure S9, Supporting Information).We observe that the reflection at 2 ≈10.1°originating from the 1D phase is present for each of the HTLs with similar intensities.This suggests that the choice of HTL does not play a role in the accumulation of the 1D phase at the buried interface, implying the absence of any specific interactions between the choline cations and the HTLs.
Next, we examined the impact of reversing the sequence in which the choline cations are introduced during the fabrication of the perovskite layers.Specifically, we added choline iodide into the organic salt solution used in the sequential deposition in order to introduce the choline cations to the second step of fabrication, instead of to the first step as was described until now.We found that in such a case, the ChPbI 3 characteristic reflection appears in the GIXRD patterns probed at the top surface (Figure S10a, Supporting Information), and the intensity decreases at the buried surface (Figure S10b, Supporting Information).These results suggest that when introduced in the second fabrication step, the 1D phase is present throughout the bulk of the layer.Moreover, it demonstrates that to promote the formation of the 1D phase at the buried interface, it is crucial to introduce the CA with PbI 2 prior to the deposition of other organic cations.Finally, we examined the role of tuning the crystallization dynamics during film formation.In this case, we compared the original process of pre-annealing at 70 °C for 1 min followed by bottom annealing at 150 °C for 15 min (two-stage annealing) with a rapid annealing process in which the film was annealed at 150 °C simultaneously at both the front and back sides of the layer.In the latter process, the annealing from the top surface was achieved by using a dry air heating gun.The rapid annealing procedure resulted in the presence of a small amount of 1D phase at the top surface of the layers (Figure S10c, Supporting Information) and a small reduction in the 1D signal at the buried interface (Figure S10d, Supporting Information).These results indicate that the crystallization rate can also influence the vertical distribution of the 1D phase within the samples, suggesting that slower crystallization is beneficial for the accumulation of the 1D phase at the buried interface, in agreement with previous observations for Cs and FA cations reported in literature. [41]To summarize, the formation of the 1D phase at the buried interface is a consequence of the sequence of the introduction of the choline cations during fabrication and the low-temperature/high-temperature step-by-step annealing process but is unrelated to the choice of HTL.

Phase Purity of the 3D and 1D Phases
While our results indicate that a ChPbI 3 perovskite with 1D phase forms at the buried interface of the perovskite layer, they do not exclude the possibility that in addition to that some CA is also incorporated into the 3D perovskite, thus acting as an additive that may alter its properties.To explore the local interactions between the choline cation and the FA x MA 1−x PbI 3 3D perovskite at the sub-nanometer to the nanometer distances, we employed solid-state nuclear magnetic resonance spectroscopy (NMR), [42,43] which can offer structural insights into the possible integration of the choline cation into the perovskite lattice.
Figure 3 presents the 1D 1 H and 207 Pb NMR spectra of the CA-20 sample, the reference FA x MA 1−x PbI 3 3D perovskite, and the synthesized ChPbI 3 .For the reference compound, the 1 H signals associated with MA + (CH 3 = 3.4 ppm and NH 3 = 6.3 ppm) and FA + cations (NH 2 = 7.5 ppm and CH = 8.2 ppm) cations are well-resolved (Figure 3a).In the case of the CA-20, additional 1 H signals in the 2.8-4.4 ppm range are observed, which are similar to those measured on the 1D perovskite and the CA itself (Figure S11, Supporting Information), thus indicating the presence of the choline cation in the CA-20 sample.In agreement with the XPS and XRD results presented above, no 1 H signals associated with the acetate group (─CH 3 = 2.05 ppm) [44] are observed for the CA-20 sample, indicating that no acetate is present in the final perovskite film.[47][48][49] While the reference FA x MA 1−x PbI 3 3D perovskite exhibited a broad 207 Pb peak centered at 1520 ppm (green band), indicating an iodine-rich 3D phase, [45,47] the CA-20 samples show an additional broad feature at ≈900 ppm (blue band).Considering that 207 Pb NMR spectra measured on the synthesized ChPbI 3 powder exhibit a clear peak at ≈900 ppm, this peak in the CA-20 can be assigned to the formation of the 1D ChPbI 3 phase modulated by the lead octahedral distortions.
Insights into the through-space spatial proximities between choline cations can be obtained by 2D 1 H-1 H correlation NMR spectroscopy.In 2D 1 H-1 H spin-diffusion (SD) NMR experiments, magnetization exchange occurs between spatially proximate and dipolar coupled spins.The resulting 2D spectrum contains on-and off-diagonal peaks, the intensities of which depend on the spin-diffusion time (also referred to as mixing time).In particular, the off-diagonal peaks are characteristic of the magnetization exchange between distinct chemical inter-and intramolecular 1 H-1 H sites in the framework and extra framework cations in interface-modified perovskites. [50]igure 3c, d compares the 2D 1 H-1 H SD NMR spectra of reference FA x MA 1−x PbI 3 3D perovskite and the CA-20 sample.For the reference compound (Figure 3c), the off-diagonal peaks are a consequence of the close proximities between the CH 3 and NH 3 sites of MA (gray box) and between the CH and NH 2 sites of FA (green box) cations.The off-diagonal peaks between the MA (CH 3 ) and FA (NH 2 ) (highlighted by the arrows) indicate the proximity (inter-and intramolecular) between the FA + and MA + cations at sub-nano to nanometer distances.The presence of these features suggests that the MA + and FA + cations are randomly distributed in the reference 3D sample. [51]In the case of the CA-20 sample (Figure 3d), the off-diagonal peaks corresponding to MA-FA proximities are observed, and a wellresolved peak associated with the 1D perovskite (vertical red dashed line, 4.4 ppm) is also detected.This spectrum does not display strong off-diagonal peaks corresponding to the throughspace interactions between choline and the 3D perovskite peaks, which indicates that the 3D and 1D perovskites co-exist, yet are phase-separated.The very weak intensity of the off-diagonal peak at 4.4 ppm (CA) and 7.7 ppm (FA) is likely to arise only from the interface between the phase-separated 3D and 1D regions.
Overall, the 1 H and 207 Pb NMR and 2D 1 H-1 H SD correlation experiments indicate that the 1D and 3D components are phase-separated in the CA-20 sample.Taken together with the observations by SEM and XRD, these experiments suggest that the choline cation is not incorporated in the 3D perovskite, but rather leads to the formation of a well-separated phase at the buried interfaces between the 3D perovskite and the HTL.

Photovoltaic Performance
To investigate the impact of the 1D perovskite formed at the buried interface between the 3D and the HTL on the photovoltaic performance, we fabricated devices with the structure indium tin oxide (ITO)/(2-(3,6-dimethoxy-9H-carbazol-9yl)ethyl)phosphonic acid:(4-(3,6-dimethyl-9H-carbazol-9-yl)butyl)phosphonic acid (MeO-2PACz:Me-4PACz)/perovskite/[6,6]phenyl-C61-butyric acid methyl ester (PCBM)/2,9-dimethyl-4,7-diphenyl-1,10-phenanthroline (BCP)/Ag (Figure 1c).The dependence of the device performance on the CA concentration (Figure 4a-d, Table 1) reveals a gradual increase of the opencircuit voltage (V OC ) and fill factor (FF) until 20 mg mL −1 of CA in the precursor solution, followed by a decrease in both the FF and the short-circuit current density (J SC ) for higher CA  concentrations (30 mg mL −1 ).As a consequence, the power conversion efficiencies (PCE) increase from an average of 20.5% to 22.5% upon the introduction of the 1D perovskite at the buried interface using the optimal CA concentration.With a further increase in the concentration of CA to 30 mg mL −1 , the PCE decreases to 20.4% on average.We attribute the performance decline to an unbalanced ratio between the 1D and 3D phases.An excess in the 1D phase may adversely affect processes such as photon absorption, charge transport and extraction.The photovoltaic parameters of the champion devices of each concentration of CA are summarized in Table 1, while the average values are listed in Table S4 (Supporting Information).The J-V curves of the champion reference and CA-20 devices are shown in Figure 4e.In both cases, the J SC remains largely unchanged, which is consistent with the corresponding EQE measurements (Figure S12, Supporting Information).However, the V OC and FF substantially increased in the CA-20 sample, leading to a maximum PCE of 24.03%.The steady-state output for the CA-20 device is 23.7%, while for the reference 3D perovskite, the steady output is 21.8% (Figure 4f).
To investigate whether the improved performance might be related to any residual lead acetate left in the perovskite layers, we performed a control experiment in which we partially replaced PbI 2 with Pb(OAc) 2 in the same molar ratio as CA and fabricated solar cell devices.Their photovoltaic performance is shown in Figure S13 (Supporting Information).The introduction of Pb(OAc) 2 indeed improves the device V OC from 1.08 to 1.12 V on average.The improvement can be explained by the anion exchange reaction between OAc − and I − upon coating with iodide salt solution, which improves the quality of the resulting perovskite films. [52,53]However, the overall performance of the device does not improve, due to the almost unchanged FF and decreased J SC .These results confirm that the performance improvement of the CA-containing devices is not a consequence of residual lead acetate in the films.These results are in agreement with the XRD, XPS, and NMR results, which show no evidence of Pb(OAc) 2 in the CA-perovskite layers.
To preliminarily evaluate the impact on device stability, we monitored the device performance under continuous illumination, thermal stress, and exposure to humidity (Figure 4g; Figure S14, Supporting Information).We observe that under continuous illumination the CA-20 devices maintained ≈80% of their performance after 400 h, while reference devices degraded faster and retained only ≈30% of their performance.Upon continuous heating at 60 °C or exposure to 40% relative humidity for 288 h, the CA-20 devices retained ≈10% higher performance in comparison to the reference devices.These results suggest that the modification of the buried interface by the 1D perovskite increases not only the efficiency but also the stability of the devices upon exposure to light, thermal stress, and humidity.
To explore the applicability of this approach to other perovskite compositions, we fabricated single-cation (FAPbI 3 ) and triplecation (FA 1−x−y MA x Cs y PbI 3−z Br z ) films and PSCs.Similar to the observations for the dual cation system (Figure 2), no changes in the microstructure of the top surface of the 3D perovskites can be observed (Figure S15a-d and S16a-d, Supporting Information), but the presence of the 2 ≈10.1°reflex in the XRD diffractograms of FAPbI 3 (Figure S15e, Supporting Information) and FA 1−x−y MA x Cs y PbI 3−z Br z (Figure S16e, Supporting Information) confirms the formation of ChPbI 3 .The performance of the FAPbI 3 and FA 1−x−y MA x Cs y PbI 3−z Br z solar cells significantly improved upon the incorporation of the optimal amount of CA, which varies depending on the specific composition (Figure S15f-j and S16f-j, Tables S1 and S2, Supporting Information).For example, the performance of FAPbI 3 devices increases from a maximum of 21.45% for reference devices to 23.67% for the champion device with a CA concentration of 10 mg mL −1 .For the triple cation perovskite composition, the PCE increases from 20.88% for reference devices to 23.01% for a CA concentration of 30 mg mL −1 .These results suggest that the spontaneous formation of 1D perovskite at the buried interface can be applied to enhance the efficiency of devices with different perovskite compositions.

Origin of the Improved Photovoltaic Performance
The increase in the V OC and FF of the devices, with minimal change in the J SC may arise from different factors such as a suppressed non-radiative recombination at the buried interface and/or an increased built-in potential in the 1D/3D device.To probe the former, we characterized the optical properties of the reference and CA-20 samples by UV-vis absorption, steady-state photoluminescence (PL), and time-resolved PL (TRPL).The absorption measurements (Figure 5a) show the same bandgap of 1.55 eV for both reference and CA-20 samples.The PL spectrum of the reference sample shows emission at ≈825 nm, as is expected based on the bandgap of the 3D perovskite (Figure 5a).The PL signal of the 1D/3D (CA-20) film is not only enhanced but is also slightly blue-shifted to ≈820 nm.This result is consistent with a reduced density of trap states at the buried interface once modified with the 1D perovskite. [13]Figure S17 (Supporting Information) depicts wide-field PL microscopy images acquired at the top and bottom surfaces of the reference and the CA-20 samples, while Figure S18 (Supporting Information) displays the wide-field PL images with normalized intensity and background removal, as well as their spectra.The images reveal that the buried interface of the reference sample exhibits more inhomogeneities and areas with weaker PL, while the PL intensity distribution of the CA-20 sample at the buried interface is far more uniform and the PL intensity is significantly higher.This suggests that the formation of the 1D phase at the buried interface not only reduces nonradiative recombination but also enhances the buried interface uniformity.The PL spectra shown in Figure S18e (Supporting Information) do not show any contributions of LD perovskite phases, which is associated with the fact that 3D perovskites have a lower bandgap, so any excitations are likely to lead to emission by the 3D phase by either energy or charge transfer. [30]A reduced density of defects is further suggested by the increase in the lifetime of the PL, which rises from 206.5 ns for the reference sample to 365.5 ns for the 1D/3D case (Figure 5e).
To investigate whether the introduction of the 1D perovskite at the buried interface impacts its energy level alignment, we carried out ultraviolet photoemission spectroscopy (UPS) measurements.As shown in Figure S19 (Supporting Information), the photoemission onset of the glass/ITO/HTL sample suggests its work function (WF) is 4.85 eV.In the case of the reference sample (3D perovskite) deposited on the HTL, the WF is reduced by 0.45 eV, corresponding to a downward change in the vacuum level at the HTL/3D perovskite interface (Figure 5b).The valance band spectrum of the reference samples shows that the Fermi level is located very close to the conduction band of the 3D perovskite layer, which in turn would lead to a reduced built-in potential once the device is completed.On the other hand, the WF of the CA-20 sample is increased by 0.30 eV as compared to the HTL (Figure 5f), corresponding to an upward shift in the vacuum level at the 1D/3D interface (Figure 5h).These measurements suggest that the introduction of the 1D perovskite at the buried interface leads to a change in the energetic landscape that would enhance the built-in potential of the device.We speculate that the change in the energetic landscape at the buried interface is related to the directional distribution of anions and cations in the 1D perovskite, but considering the interface is buried, a direct investigation of such a process is complex and is beyond the scope of this work.
Taken together the PL and UPS results suggest that the 1D perovskite has a dual beneficial effect on the performance of the solar cells: on the one hand, it leads to the passivation of defects at the buried interface (thus reducing the degree of non-radiative recombination), while on the other hand, it leads to an increase in the built-in potential of the device.The combination of these effects would lead to an increase in the V OC and FF of the solar cells, which is consistent with the photovoltaic performance results displayed in Figure 4a,c.
The consequence of these effects is also evident when the solar cells are measured as light-emitting diodes, with the corresponding electroluminescence (EL) spectra and the EL quantum efficiency (ELQY) shown in Figure 6a.Devices with a 1D/3D structure show an enhanced EL intensity with approximately one order of magnitude higher ELQY than the reference 3D solar cell.These results agree with the PL and TRPL measurements and indicate that the extent of the non-radiative recombination in the perovskite layer has been largely suppressed.This conclusion is further supported by the light intensity-dependent V OC measurements (Figure 6b), which reveal an overall lower degree of trapassisted recombination in the 1D/3D device, evidenced by a reduction in the ideality factor (n) from n = 1.701 for the reference devices to n = 1.308 for the CA-20 device. [54]ransient photovoltage and photo-current (TPV and TPC, respectively) decays measured at open-and short-circuit, respectively, allow us to further explore the carrier recombination and transport in the devices. [55]The charge recombination time constant ( r ) of the 1D/3D device is 12.4 μs-more than double the  r of the reference device at 5.7 μs (Figure 6c)-suggesting a slower surface charge recombination at the buried interface upon the introduction of the 1D perovskite.The transport time constant ( t ) decreases from 1.6 to 1.1 μs after forming the 1D/3D structure (Figure 6d), suggesting an improved hole extraction at the 1D/3D interface as a consequence of the favorable energy level alignment.The fitting details are presented in Figure S20 and Table S3 (Supporting Information).

Conclusion
In summary, we introduce a method to generate a 1D/3D heterojunction at the buried interface of the perovskite active layer, which enables a substantial improvement in both the V OC and FF of the photovoltaic devices.This increase is a result of a suppression of non-radiative recombination processes and a beneficial energy level alignment at the buried interface upon the introduction of the 1D perovskite.Consequently, a maximum power conversion efficiency of 24.03% can be achieved based on a 1D/3D heterojunction device, surpassing the performance of reference 3D perovskite devices (22.30%).The bulk and local structure of the buried interface is characterized by a combination of X-ray diffraction/scattering and NMR spectroscopy.In addition, we demonstrate that the approach can be applied to a range of perovskite compositions, leading in addition to an increase in the devices' light stability and is promising for large-area applications.Our work highlights the importance of modifying the buried interface of the perovskite layer and offers a versatile approach to improve the performance of inverted architecture PSCs.
Photovoltaic Device Fabrication and Characterization: Pre-patterned ITO substrates were ultrasonically cleaned with 2% Hellmanex detergent, deionized water, acetone, and isopropanol, followed by oxygen plasma treatment at 100 mW for 10 min.MeO-2PACz and Me-4PACz are dissolved with 10 mmol L −1 in anhydrous ethanol, separately.The solution was ultrasonicated for 15 min at 30-40 °C to obtain a stock solution.Each stock solution is mixed 1:9 with anhydrous ethanol to obtain 1 mmol L −1 solution, then mixed with the ratio of 3:2 (MeO-2PACz:Me-4PACz) for spincoating.30 μL of mixed solution were spin-coated at 4000 rpm for 30 s, followed by annealing at 100 °C for 10 min in an N 2 -filled glovebox.
Solution-Processed FAPbI 3 : The PbI 2 solution was prepared by dissolving 816 mg PbI 2 in 1 mL anhydrous DMF:DMSO (9:1, v:v) solvent mixture.After the PbI 2 solution was fully dissolved, it was spin-coated on ITO/HTL at 2800 rpm for 30 s in a dry air-filled glovebox (relative humidity < 1.0%).Then, the PbI 2 film was transferred into a vacuum chamber for 5 min to remove extra solvent.After the evacuation, the films were transferred back to the dry air-filled glovebox, and a mixed organic cation solution (FAI:MACl = 90 mg:9 mg in 950 μL anydrous isopropanol) was dynamically spin-coated at 2300 rpm for 30 s to form a wet the precursor film.The as-coated precursor film was placed onto a 70 °C hot plate to anneal for 1 min then transferred onto a 150 °C hot plate to anneal for 15 min in dry air condition.
Solution-Processed FA x MA 1−x PbI 3 : The mixed organic cation solution was prepared by mixing FAI, MAI, and MACl in the amounts of 90, 6.39, and 9 mg, respectively, in 950 μL anhydrous isopropanol.Other details were the same as the fabrication method of FAPbI 3 perovskite.
Solution-Processed FA 1−x−y MA x Cs y PbI 3−z Br z : PbI 2 solution was mixed with 5% CsI of the molar ratio.The mixed organic cation solution was prepared by mixing FAI, MABr, and MACl in the amounts of 90, 3.08, and 5 mg, respectively in 1 mL anhydrous isopropanol.The mixed organic cation solution was dynamically spin-coated at 2300 rpm for 30 s.Other details were the same as the fabrication method of FAPbI 3 perovskite.
For CA-modified devices, 10-30 mg mL −1 of CA was dissolved in the PbI 2 solution and stirred at room temperature for 12 h before use.
After the deposition of perovskite layers, the samples were transferred into a nitrogen-filled glovebox, where PC 61 BM (20 mg mL −1 dissolved in anhydrous chlorobenzene, 99.8%) was dynamically spin-coated at 2000 rpm for 30 s followed by a 10 min annealing at 100 °C.At last, a bathocuproine (BCP) (0.5 mg mL −1 dissolved in anhydrous isopropanol, 99.5%) hole-blocking layer was spin-coated at 4000 rpm for 30 s, followed by an 80 nm thermally evaporated Ag cathode (Mantis evaporator, base pressure of 10 −7 mbar).
For exposing the buried interface of the layers, a mechanical peeling-off procedure was performed following a previously published method. [24,25]hotovoltaic Device Characterization: EQE spectra of the devices were recorded using a monochromatized light of a halogen lamp from 400 to 800 nm, and the reference spectra were calibrated using a National Institute of Standards and Technology (NIST)-traceable Si diode (Thorlabs).J-V characteristics of solar cells under a solar simulator (Abet Sun 3000 Class AAA solar simulator, AM 1.5 conditions) were recorded at room temperature in ambient conditions using a computer-controlled Keithley 2450 source meter unit.The incident light intensity was calibrated via a Si reference cell (NIST traceable, VLSI Standards Inc.) and tuned by measuring the spectral mismatch factor between the real solar spectrum, the spectral response of the reference cell, and the perovskite devices.All devices were scanned from short circuit to forward bias (1.2 V) and reversed with a

Figure 1 .
Figure 1.a) Schematic presentation of the fabrication procedure of the perovskite active layer.b) Chemical structure of choline acetate.c) Solar cell device architecture used in this work.

Figure 2 .
Figure 2. a) Schematic presentation of the flipping process to characterize the buried interface.SEM of the b) top surface, and c) buried surface of reference perovskite film.SEM of the d) top surface, and e) buried surface of CA-20 perovskite film.GIXRD of f) the top surface, and g. the buried surface of the CA-20 with different incident angles.

Figure 3 .
Figure 3. Solid-state 1D NMR a) 1 H and b) 207 Pb spectra of CA-20, ref, and 1D component.In b., the green-colored vertical bands indicate the 207 Pb peaks associated with 3D perovskites, and the blue bar indicates the 1D component, respectively.Solid-state 2D 1 H-1 H spin diffusion spectrum of c) ref and d) CA-20 films were acquired using 500 ms of spin-diffusion mixing time.The line-cut spectra (row-and column) at 4.4 ppm and 7.5 ppm are shown in red d.All spectra are acquired at 18.8 T ( 1 H = 800.1 MHz and 207 Pb = 167.6MHz with 50 kHz magic-angle spinning) and room temperature.

Figure 4 .
Figure 4. PV performance of reference and CA-based solar cell devices a) V OC , b) J SC c) FF d) PCE e) J-V curves of the champion cells f) maximum power point (MPP) tracking.g) PCE evolution upon continuous illumination of encapsulated devices in air.The box plots in panels (a-d) display the mean, median line, and 25-75% box limits with 1.5× interquartile range whiskers.The number of devices used for statistical analysis is 75, 99, 49, and 58 for ref, CA-10, CA-20, and CA-30, respectively.

Figure 5 .
Figure 5. a) UV-vis and steady-state PL spectra of reference 3D perovskite and CA-20.b, c) UPS spectra of reference 3D perovskite on an ITO/HTL substrate.d) Energy-level diagram of reference 3D perovskite.e) TRPL of reference 3D and CA-20 perovskite on glass.f, g) UPS spectra of CA-20 on an ITO/HTL substrate.h) Energy-level diagram of CA-20.

Figure 6 .
Figure 6.a) ELQE and EL spectra (inset) and b) Light intensity-dependent V OC c) TPV d) TPC of reference and CA-20 devices.

Table 1 .
PV parameters of the champion devices produced with different amounts of CA.