Buried Interface Smoothing Boosts the Mechanical Durability and Efficiency of Flexible Perovskite Solar Cells

Abstract

Flexible perovskite solar cells (F-PSCs) have the advantages of high power-per-weight, solution processability, and bending durability and have emerged as a competitive photovoltaic technology for various applications. As the core electron transport layer (ETL) in n-i-p-type device configurations, the solution-processed SnO₂ generally suffers from serious defect stacking on films, compromising the charge transport properties and the performance of resulting devices. Herein, we proposed a media-filling strategy to optimize the contact quality at the buried interface by introducing Al₂O₃ nanoparticles on the SnO₂ surface. Rather than forming a compact insulating layer, the Al₂O₃ can fill the grain boundaries of SnO₂ and smooth the substrate surface. Optimized interfacial contact under careful concentration control can rationally minimize the contact area of the perovskite with the surface imperfections of SnO₂ to mitigate trap-assisted charge recombination. Furthermore, the reduced surface roughness of SnO₂ facilitates the uniform deposition and oriented growth of upper perovskite film. As a result, the target F-PSCs achieved an impressive efficiency of 23.83% and retained 80% of the initial performance after 5000 bending cycles at a radius of four mm.

1. Introduction

Flexible perovskite solar cells (F-PSCs) have demonstrated significant advantages, such as high power-per-weight ratios, flexibity [1,2], and compatibility for large-scale preparations by solution processing and have attracted tremendous academic and commercial attention [3,4,5]. These advantageous features position them as a promising candidate [6,7] to supply constant energy for wearable electronic devices within the evolving landscape of the Internet of Things (IoTs) [8,9,10]. Benefiting from the exceptional optoelectronic properties of metal halide perovskite, F-PSCs have delivered a rapid increase in power conversion efficiency (PCE) of over 25% to date [11]. However, the current record efficiency of F-PSCs remains inferior to that of their rigid counterparts. The practicality of this lightweight photovoltaic technology has also been restrained by their compromised mechanical stability over long-term bending cycles [12]. These limitations are primarily attributed to the intricate low-temperature procedure [13] and inhomogeneity of flexible substrates, which raises the difficulty for the high-quality deposition of the subsequent functional layers in devices [10,14]. The rough contact interface between the charge transport layer and perovskite film can deteriorate the weak interfacial adhesion, thus resulting in problems with mechanical delamination and the bending fragility of F-PSCs [15]. Moreover, the significant energy loss at the buried interface has emerged as the dominant hindrance that limits the further development of F-PSCs.

Due to its high electron mobility, good optical transmittance, and chemical stability, SnO₂ has emerged as a promising electron transport layer for F-PSCs. However, the solution-processed SnO₂ layer using colloidal dispersion generally suffers from numerous Sn dangling bonds and hydroxyl groups on the resulting film surface. These active sites can absorb O₂ and H₂O from the environment, facilitating electron capture and elevating the charge transport barrier [16]. Furthermore, the presence of voids and surface imperfections in SnO₂ film may compromise the crystal growth and morphology quality of upper perovskite films, consequently affecting the overall performance of devices. To optimize the buried interface of F-PSCs, previous researchers have predominantly concentrated on molecular modification to tailor interfacial properties [17]. The functional molecules can passivate the surface traps through chemical coordination or ionic bonding, facilitating improved interface contact and enhanced charge transport kinetics in F-PSCs. However, organic molecular modification is generally limited by the bonding effect, which is inadequate for the efficient suppression of interfacial imperfections. An increased incorporation concentration would reversely deteriorate the charge transport due to the increased series resistance. Therefore, molecular modification largely relies on a tedious trial-and-error process, which inevitably increases the preparation cost and deteriorates the device reproductivity. Additionally, the high roughness of flexible substrates further threatens the compactness and conformal alignment of SnO₂ colloids, particularly affecting high-throughput and scalable fabrications. As such, the surface imperfections of SnO₂ pose a decisive risk to the structural integrity and functionality of F-PSCs, highlighting the importance of rational interfacial modulation in the deposition process [18].

Alternatively, incorporating metal-oxide nanoparticles has been confirmed to have an effect on optimizing SnO₂ film. For example, Cheng et al. constructed a SnO₂/TiO₂ bilayer by spin-coating the TiO₂ nanoparticles on the SnO₂ ETL, resulting in the formation of cascade-aligned energy levels and reducing the energetic barrier for electron transfer [19]. In contrast to utilizing n-type semiconductors with a high electron conductivity in a bilayered structure, some insulating metal oxides such as Al₂O₃ and MgO were employed to modify the SnO₂ layer by smoothing substrates. For example, Brown et al. introduced the MgO on the top of SnO₂ ETL to boost the rectification ratio, reduce interfacial recombination, and improve the shelf-life stability of PSCs. Xu et al. deposited AlO_x via atomic layer deposition (ALD) as a buffer layer on an ITO substrate to promote the dense and conformal coating of SnO₂ nanoparticles and improve electron extraction at the cathode interface [20]. Despite these advances, the feasibility of using wide-bandgap metal oxidates as the buffer layer or interfacial filler for the F-PSCs has rarely been reported. The influences of media filing on the interfacial properties of SnO₂ and upper perovskite layers are still being explored. Thus, developing a controllable optimization strategy for the buried interface is imperative to optimize the charge transfer and boost the efficiency and mechanical robustness of F-PSCs.

In this study, we proposed an interfacial smoothing strategy for the buried contact of F-PSCs by introducing porous Al₂O₃ nanoparticles at the SnO₂-perovskite interface. In contrast to traditional molecular passivation, the Al₂O₃ nanoparticles can improve the carrier kinetics of F-PSCs by reducing the contact area of the perovskite layer with the defective substrates while maintaining the necessary electron extraction channels. Rather than forming a dense insulating layer on the SnO₂ surface, the Al₂O₃ nanoparticles were inserted into the grain boundaries of polycrystalline SnO₂ and created randomly distributed nanoscale openings. This optimized distribution of Al₂O₃ can largely avoid the thickness-dependent electron tunneling from compact stacking, mitigate the impacts of surface imperfections on the charge transport, and suppress trap-induced non-radiative loss at the cathode interface. Simultaneously, the surface-filling effect of Al₂O₃ nanoparticles can improve the uniformity and roughness of the SnO₂ layer, which subsequently facilitates high-quality perovskite deposition featuring enhanced crystallinity and well-connected grain stacking. The synergistic optimization of the buried interface by the addition of Al₂O₃ contributed to efficient charge transport with significantly suppressed energy loss in devices. Finally, the target F-PSCs achieved the optimal efficiency of 23.83%, outperforming the control device, which had an efficiency of 20.98%. The Al₂O₃ decoration also improved the humidity resistance and mechanical stability of F-PSCs, which retained over 80% of the initial performance after ambient aging for 30 days or 5000 bending cycles at a circus of 4 mm. This work provides inspiration for buried interface modulation beyond molecular modification for high-performance F-PSCs, which may increase the broad applicability of physical contact optimization for the large-scale production of F-PSCs.

2. Materials and Methods

2.1. Materials

Formamidinium iodide (FAI), lead iodide (PbI₂), Spiro-OMeTAD powder, and methylamine hydrochloride (MACl) were purchased from Advanced Election Technology Co., Ltd., Shenyang, Liaoning, China, Methylammonium iodide (MAI), cesium iodide (CsI), and 2-Phenylethanamine hydroiodide (PEAI) were purchased from Xi’an Yuri Solar Co., Ltd., Xi’an, China. Dimethylsulfoxide (DMSO) and N,N-dimethylformamide (DMF) were purchased from Aladdin. 4-tert-butylpyridine (TBP), Li-bis-(trifluoromethanesulfonyl) imide (Li-TFSI), tin(IV) oxide (SnO₂), colloidal dispersion liquid (15% in H₂O), and aluminum oxide ethanol dispersion (Al₂O₃, particle size of ca. 100 nm) were purchased from Alfa Aesar, Ward Hill, MA, USA. Isopropanol (IPA) was purchased from Sinopharm Chemical Reagent Co., Ltd., Beijing, China. Chlorobenzene (CB) and ethanol anhydrous (Et(OH)₂) were purchased from Sigma-Aldrich, Shanghai, China. All chemicals and solvents were used directly as received.

2.2. Preparation of Perovskite Precursor Solution

The perovskite solution was prepared by dissolving 645 mg PbI₂, 221 mg FAI, 15 mg CsI, 9 mg MAI, and 38 mg MACl in 1 mL mixed solvent of DMF and DMSO (volume ratio of 4:1). The solution was stirred for 12 h and then filtered by 0.22 μm syringe filter for later use.

2.3. Fabrication of F-PSCs

The flexible substrate was cleaned and adhered to the rigid soda-lime glass using PDMS, securing the ends with tape and leaving an area for the electrodes. The clean flexible substrate polyethylene terephthalate/indium tin oxide (PET/ITO) was treated with UV−ozone for 16 min. SnO₂ was dissolved in ultrapure water at a ratio of 1:6 and stirred for 2 h before its use. After the PET/ITO cooled down, 150 μL of SnO₂ colloidal solution was spin-coated on PET/ITO at 4000 rpm for 30 s and then annealed at 100 °C for 30 min after spin coating. Isopropanol was used to dilute the aluminum oxide ethanol dispersion to different concentrations (volume ratios of 1:2, 1:4 and 1:6), and the mixture was stirred for 2 h. The Al₂O₃ ethanol dispersion (150 μL) was spin-coated on the SnO₂ substrate at 3000 rpm for 30 s and then annealed in air at 100 °C for 10 min. Before spin coating, SnO₂ ETL was treated with UV–ozone for 12 min. The perovskite film was prepared by a one-step spin-coating procedure at 6000 rpm for 30 s. Twenty s before the end of the process, 130 μL antisolvent (chlorobenzene) was dropped onto the substrates, which were then annealed at 120 °C (under the following conditions: ~25 °C temperature; ~5% humidity) for 30 min. 2-Phenylethanamine hydroiodide (PEAI) was dissolved in isopropyl alcohol and stirred for 4 h. PEAI was coated with 3000 rpm for 30 s. Then the hole transport layer (HTL) was fabricated by spin coating 45 μL of Spiro-OMeTAD solution (90 mg Spiro-OMeTAD, 22 μL of Li-TFSI, and 36 μL TBP in 1 mL CB). The 80 nm Au electrode was deposited on the HTL.

2.4. Device Characterizations

Scanning electron microscope (SEM) images were acquired using field-emission SEM (SU8020, Hitachi, Tokyo, Japan). The AFM images were acquired using an atomic force microscope (Dimension Icon, BRUKER, Billerica, MA, USA). X-ray diffraction (XRD) patterns were recorded on a X-ray diffractometer (Smart Lab, Rigaku, Japan). UV-vis absorption spectra were obtained by a UV-vis spectrophotometer (PerkinElmer UV-Lambda 950, Waltham, MA, USA). Steady-state photoluminescence (PL) and time-resolved photoluminescence (TRPL) spectra were measured on a fluorescence spectrometer (PicoQuant FT-300, Berlin, Germany). The current density–voltage (J-V) curves were obtained in the air with a Keithley 2400 source meter under the AM 1.5 radiation condition (100 mW/cm²). The scan range was from 1.5 to 0 V with a 0.02 V bias step. The external quantum efficiency (EQE) measurements were conducted under a 300 W Xenon lamp light source with a QTest Station 2000ADI system (Enli Technology Co., Ltd., Taipei, Taiwan, China), with the monochromatic light intensity calibrated by a reference silicon photodiode. The EIS and C-V measurements were performed using a Zahner Zennium electrochemical workstation.

3. Results and Discussion

3.1. Interfacial Morphology and Electrical Properties of SnO₂

We first conducted atomic force microscopy (AFM) to evaluate the surface topography of the SnO₂ films. As shown in Figure 1a, the average roughness (Ra) of the pristine SnO₂ films was determined to be 19 nm, indicating the rough surface of the SnO₂ nanocrystals on the surface under solution processing. Then, we deposited the perovskite film onto the SnO₂ substrate, which was exfoliated from the substrate to expose the bottom interface [21]. The morphology of the buried interface was examined using scanning electron microscopy (SEM), illustrating a high density of pinholes on the bottom of the perovskite films, indicating the poor interface adhesion at the perovskite-SnO₂ interface (Figure 1b). The existence of holes could be attributed to the poor uniformity and flatness of the SnO₂ substrate, leading to the elevated difference in the surface energy across the different areas (Figure 1c). The subsequent perovskite deposition onto the irregular substrate can disturb the nucleation sites and intensify the randomness in the growth orientation of perovskite crystals. The lattice extrusion at the bottom surface causes the crystal deformation, thereby deteriorating the film quality.

After introducing Al₂O₃ nanoparticles on the SnO₂ surface, we noted a reduction in Ra from 19 to 16 nm, as demonstrated by the AFM results (Figure 1d). This indicates the Al₂O₃ nanoparticles could fill the grooves on the SnO₂ film, thereby reducing its roughness. The SEM images of the perovskite film on the SnO₂/Al₂O₃ substrate showed the densely packed grains at the buried interface with no obvious void residue (Figure 1e). Collectively, the enhancement in surface quality of SnO₂ could facilitate the high-quality deposition of perovskite films, potentially improving the interfacial contact and carrier transport kinetics (Figure 1f).

To explore the effect of Al₂O₃ modification on SnO₂ films, X-ray diffraction (XRD) was performed to assess the crystalline quality of SnO₂ ETL films before and after the addition of Al₂O₃. Figure 2a shows that introducing Al₂O₃ nanoparticles results in negligible variation in the crystal stacking structure of SnO₂ films. Similarly, the optical bandgap of different SnO₂ films retains the same value of 4.02 eV (Figure S1). Figure 2b shows the optical transmittance spectra of the SnO₂ substrates in the 300–900 nm region, illustrating the uncompromised transmittance of SnO₂ with the introduction of Al₂O₃ nanoparticles. The high light transmittance of bottom substrates ensures the adequate optical absorption of upper perovskite films. In addition, based on the space charge limited current (SCLC) method, the electron-only device structure of ITO/SnO₂/Ag was fabricated to assess the electrical properties of the ETL substrates (Figure S2). The electron mobility was calculated by the Mott−Gurney formula [22]:

J_D = 9μεε₀V²/8L³

(1)

where ε is the relative dielectric constant, ε₀ is the vacuum dielectric constant, and L is the film thickness. Compared to pure SnO₂, the electron mobility of SnO₂ films modified with Al₂O₃ was increased from 6.53 × 10⁻³ to 11.7 × 10⁻³ cm² V⁻¹ S⁻¹ (Figure 2c), indicating the enhanced carrier transport capacity of SnO₂/Al₂O₃ film. The conductivity of ETLs was further tested, which can be calculated by the following formula:

I = σAd⁻¹

(2)

where A is the device area and d represents the thickness of SnO₂ ETL [23]. According to the current–voltage (I-V) curves shown in Figure 2d, the Al₂O₃-modified SnO₂ showed increased conductivity from 2.27 × 10⁻³ to 4.21 × 10⁻³ cm² V⁻¹ S⁻¹ compared to the pristine SnO₂. These findings suggest that modification with Al₂O₃ nanoparticles is an efficient approach to improve the electron transport properties of ETLs on flexible substrates.

3.2. Morphology and Crystallinity of Flexible Perovskite Films

SEM and AFM measurements were employed to further evaluate the film formation quality of perovskite on SnO₂ and SnO₂/Al₂O₃ substrates, referred to as the control and target in the following discussion, respectively. As shown in Figure 3a,b, the target film exhibits a lower surface roughness than the control film (18 vs. 33 nm). This reduction in film roughness could be attributed to the flattening effect of the Al₂O₃ filling at the SnO₂ grooves. Moreover, the target film shows fewer grain boundaries and larger grain sizes, with the average grain size increasing from 0.5 to 1.1 μm (Figure 3c–f). These observations indicate that the Al₂O₃ nanoparticles can support a smooth interface condition for the uniform nucleation growth of perovskite crystals, thereby enhancing the crystalline quality of perovskite films. Figure 3g presents the XRD patterns of the perovskite films. Diffraction peaks at 13.94° and 28.08° were detected in the control and target films, corresponding to the (100) and (200) crystal planes, respectively. Notably, the films deposited on the Al₂O₃-modified substrate exhibit a stronger XRD diffraction intensity and a narrower half-peak width (FWHM) for the (100) crystal plane, as shown in Figure 3h. This indicates that the crystallinity of the films is significantly enhanced, which is mainly attributed to the improvement of film crystallization by the flattening filling strategy for Al₂O₃ nanoparticles. The above results indicate that the flattening of ETL layers can effectively promote high-quality crystallization and the growth of the perovskite films and can maintain similar absorption band edges (Figure 3i). Figure S3 shows that the bandgap of the control and target perovskite films is 1.55 eV.

3.3. Optical and Electrical Properties of Flexible Perovskite Film

We further explored the effect of Al₂O₃-modified SnO₂ on the optoelectrical properties of the films using steady-state photoluminescence (PL) and time-resolved photoluminescence (TRPL) characterizations. As shown in Figure 4a, the PL peaks for both the control and target films are around 780 nm, with no distinct shift observed. In contrast, the peak intensity of the target film is notably reduced, suggesting that the modified SnO₂ ETL has a higher charge extraction capacity than the control. The PL mapping further demonstrated faster charge extraction and a uniform film morphology in the target case, facilitating the suppressed non-radiative recombination loss at the buried interface (Figure 4b) [24]. TRPL spectra were further examined to determine the carrier kinetics, as illustrated in Figure 4c. The carrier lifetime was determined using the following double exponential function:

τ(t) = A₁exp(−t/τ₁) + A₂exp(−t/τ₂) + B

(3)

where A₁ and A₂ are the corresponding decay amplitudes, τ₁ and τ₂ are the slow and fast decay time constants, respectively, and B is a constant [25]. For the SnO₂/perovskite case, the fast decay process (τ₁) is driven by interfacial charge transfer, while the slow decay process (τ₂) is associated with the radiative recombination of free carriers within the bulk perovskite layers. For the fitting parameters listed in Table S1, the τ₁ was decreased from 47.4 to 9.5 ns and the τ₂ was reduced from 246.3 to 142.1 ns with the incorporation of the Al₂O₃ nanoparticles. The average carrier lifetimes of the control and target films were calculated to be 234.3 and 133.5 ns, respectively. The decreased carrier lifetime in the target film confirms that the Al₂O₃-modified buried interface delivers an improved charge extraction capability.

In order to further evaluate the defects density within perovskite films deposited on different substrates, we tested the dark current–voltage curves (I-V) of the devices with the structure of ITO/SnO₂/perovskite/PCBM/Ag via the SCLC method. The trap density was determined by the following formula [26]:

$$n_{trap}={\displaystyle \frac{2{\epsilon }_0{\epsilon }_r{V}_{TFL}}{\mathrm{e}{L}^2}}$$

(4)

where ${\epsilon }_r$ is the relative dielectric constant of perovskite (${\epsilon }_r$ = 62.3), V_TFL is the starting voltage of the limit region of defect filling, e is the amount of electron charge (e = 1.6 × 10⁻¹⁹ C), and L is the thickness of perovskite film. As shown in Figure 4d,e, the defect density of the device was reduced from 6.89 × 10¹⁵ to 3.86 × 10¹⁵ cm⁻³ after the modification with Al₂O₃ nanoparticles. Additionally, a Mott−Schottky analysis was applied to evaluate the built-in electric field (V_bi) of the devices [27]. As shown in Figure 4f, the Al₂O₃-modified device exhibits a higher V_bi compared to the control device, with values increasing from 1.08 to 1.16 V. This indicates that the modified device has a stronger driving force for charge transport, which could contribute to the improved carrier kinetics at the buried interface and the improvement in the V_OC of the device [28]. Figure 4g presents the electrochemical impedance spectra (EIS) of the devices. According to the fitting results in Table S2, the modified device shows a smaller transport resistance (R_tr) and a larger recombination resistance (R_rec) than the control after introducing Al₂O₃ nanoparticles, indicating suppressed trap-induced recombination [29]. The dark J-V curves further illustrate the lower leakage current of the target device concerning the control case, implying the blocking effect of Al₂O₃ nanoparticles on the carrier drift to the ETLs. Collecting the findings above, the Al₂O₃ modification can reduce the contact area of the perovskite under defect-enriched interface conditions and supplement a smoother film surface for carrier transport. These benefits effectively accelerate carrier extraction and alleviate the non-radiative recombination loss at the cathode layer, ensuring enhancements in device performance [30].

3.4. Flexible Solar Cells and Device Stability

Based on the SnO₂ smoothing strategy, we fabricated the n-i-p F-PSCs with the device configuration of ITO/SnO₂/Al₂O₃/perovskite/Spiro-OMeTAD/Au (Figure S4). We first evaluated the influence of the incorporation concentration of trace amounts of Al₂O₃ on the device performance. The corresponding J-V curves of the devices are illustrated in Figure 5a, and detailed parameters are summarized in Table S3. The control device showed a PCE of 20.98%, with a V_OC of 1.10 V, a J_SC of 24.58 mA cm⁻², and an FF of 77.58%. As the concentration of Al₂O₃ nanoparticles increased, the PCE of modified devices showed an increasing trend to reach a maximum at a concentration of 1:4 and declined at higher concentrations. The excessive incorporation concentration of Al₂O₃ nanoparticles can cause the formation of a compact insulating layer on the SnO₂ surface, blocking the charge transport and extraction. Under careful concentration control, the optimal device achieved a optimal PCE of 23.83%, with significant improvements in V_OC and FF. The external quantum efficiency (EQE) measurement in Figure 5b determined the integrated J_SC values of the control and target devices, which were 23.97 and 24.17 mA cm⁻², respectively, coinciding with the results from the J-V test, which verifies the accuracy of the device current [31]. Figure 5c–f gives the statistical diagrams of V_OC, FF, PCE, and J_SC, respectively. These results indicate that Al₂O₃-modified SnO₂ is capable of achieving an increased efficiency and good reproducibility.

Finally, the stability of unencapsulated F-PSCs based on different SnO₂ ETLs was evaluated under humidity and mechanical bending conditions, respectively. As shown in Figure 5g, the control device exhibited rapid degradation under ambient conditions with 30% relative humidity at room temperature, and the initial efficiency decreased to below 80% after 18 days. In contrast, the target device retained ca. 80% of the initial performance after 30 days. For the mechanical durability of F-PSCs, the control device began to degrade in the initial bending treatment at a bending circus of 4 mm and lost over 20% of its initial efficiency after 3000 bending cycles (Figure 5h). Encouragingly, the original efficiency of the target device was retained until 500 bending cycles, with the value remaining over 80% after 5000 cycles, illustrating the significantly improved mechanical robustness. With the Al₂O₃ decoration, the improved crystalline quality of perovskite films and the strengthened interfacial adhesion at the SnO₂-perovskite interface contribute to the collective optimization effect on the device longevity.

4. Conclusions

In this study, we proposed a novel media-filling strategy to smooth the buried interface of F-SPCs. The insulting Al₂O₃ nanoparticles were introduced at the perovskite-SnO₂ interface to fill the grooves on the SnO₂ surface and reduce its contact area with the absorber layer. The improved uniformity and roughness of the SnO₂ surface facilitate subsequent high-quality perovskite deposition with enlarged grain sizes and well-connected grain stacking. Our experimental results revealed that the addition of Al₂O₃ can effectively alleviate non-radiative recombination at the cathode interface through a synergistic effect on defect mitigation and charge extraction. As such, the optimized F-PSCs achieved a best-performing efficiency of 23.83%, which was higher than that of the control (20.98%). Moreover, the buried-interface smoothing can further enhance the mechanical durability of F-PSCs, which retained over 80% of the original efficiency after 5000 bending cycles. Our work illustrates the importance of buried interface management for the energy loss of F-PSCs, which may inspire further optimization research beyond molecular tailoring on the interfacial contact to realize the full device potential.