Enhancing the Interfacial Adhesion by a Novel Benzofuran‐Substituted Self‐Assembled Molecules for Thermal Cycle Stable Perovskite Solar Cells and Modules

ABSTRACT

Inverted perovskite solar cells (PSCs) employing self‐assembled molecules (SAMs) as hole transporting layer (HTL) have achieved markedly improved performance. However, emerging evidence demonstrates that the adhesion weakness between the HTL|perovskite interface hinders the charge transfer and limits the stability of the devices. Herein, a novel SAM featuring a benzofuran substitution, (2‐(5H‐benzofuro[3,2‐c]carbazol‐5‐yl)ethyl)phosphonic acid (BFC‐2PACz) has been designed and synthesized to address this issue. Molecular characterizations demonstrate that the BFC‐2PACz exhibits strong interactions with both the perovskite and NiO _x , thus not only improve the coverage of SAM, but also enhance the mechanical adhesion at the interface. By slot‐die coating the mixture of BFC‐2PACz and [4‐(3,6‐dimethyl‐9H‐carbazol‐9‐yl)butyl]phosphonic acid (Me‐4PACz) as HTLs, we have achieved efficiencies of 21.2% and 16.8% based on a wide bandgap (1.68 eV) perovskite for small‐area solar cells and larger‐area solar modules (10.0 cm²), respectively. More importantly, the devices maintain 87% of their initial efficiencies after 720 h of thermal cycling aging and 82% after 1000 h of maximum power point tracking.

Benzofuran‐functionalized BFC‐2PACz enables strong interfacial interactions with both NiOx and perovskite, improving the interfacial mechanical adhesion. A mixed SAM deposited by slot‐die coating delivers efficient and stable solar cells based on 1.68 eV bandgap perovskite, achieving 21.2% efficiency, while retaining 87% performance after thermal cycling and 82% under continuous operation.

1. Introduction

Over the past few years, perovskite solar cells (PSCs) have attracted extensive interest due to their high power conversion efficiencies (PCEs), compatibility with large‐scale fabrication, and low‐cost processing [1, 2, 3]. Wide bandgap (WBG) perovskites hold immense appealing properties, including tunable bandgaps, high absorption coefficient, and low exciton binding energy, making them excellent candidates for semitransparent and tandem cells that exceeding the theoretical efficiency limits [4, 5]. Recently, significant advancements in PSCs have been achieved with the application of self‐assembled molecules (SAMs) as hole transporting layers (HTLs) in inverted (p‐i‐n) structure [6]. Many reports demonstrate that SAM‐based HTLs can not only modulate the nucleation and crystallization of perovskite film, but also facilitates carrier transport, which has enabled p‐i‐n PSCs to reach a remarkable certified PCE of more than 27.0% [7, 8, 9].

However, despite the great promise, the stability concerns of PSCs remains a major obstacle to commercialization, with inadequate damp‐heat tolerance becoming particularly problematic when devices are fabricated using scalable processing methods [10, 11]. Increasing evidence shows that weak adhesion at the buried HTL|perovskite interface is a critical bottleneck, which can induce a non‐uniform crystallization of the perovskite, increase the interfacial recombination, and accelerate the interfacial degradation, ultimately leading to mechanical failure and performance degradation [12, 13, 14, 15]. Therefore, it is urgent to enhance the adhesion at the HTL|perovskite interface.

Recently, Zhang et al. demonstrated that the interface adhesion of a bilayer HTL structure is stronger than a single layer of NiO _x or SAMs [16]. Li et al. improved the interface properties by anchoring (4‐(7H‐dibenzo[c,g]carbazol‐7‐yl)butyl)phosphonic acid onto the NiO _x surface, resulting in improved interfacial robustness under thermal stress [17]. Wu et al. further improved the interface properties by a co‐assemble layer that contains two SAM molecules of 4‐(3,6‐bis(diphenylphosphoryl)‐9H‐carbazol‐9‐yl)butylphosphonic acid and [4‐(3,6‐dimethyl‐9H‐carbazol‐9‐yl)butyl]phosphonic acid (Me‐4PACz) [18]. Chen et al. improved the interface robustness by introducing a third layer of tris(2‐carboxyethyl)phosphine hydrochloride, which is due to the simultaneously formation of C═O···Ni coordinated bond and O─H···O─Ni hydrogen bond [19]. This enabled improvements of both efficiency and thermal stability of PSCs because of the enhanced the charge transfer properties between the HTL and perovskite. Although these modifications effectively strengthen the interface properties, the study of the adhesion at the SAM|perovskite interface remains limited. In addition, all HTLs in these strategies rely on spin‐coating method, which is difficult to translate to large‐area fabrication.

In response, this work designs and synthesizes a novel SAM, (2‐(5H‐benzofuro[3,2‐c]carbazol‐5‐yl) ethyl) phosphonic acid (BFC‐2PACz), aiming to improve the adhesion between the HTL and WBG perovskite layers. We systematically studiesd the interfacial adhesion between these layers by single‐lap shear tests. The mechanism of the mechanical adhesion variations has been studied by investigating the coverage of the SAMs, the molecular interactions between SAMs and perovskite as well as NiO _x . We have further studied the morphology of the HTL and the atop perovskite films by X‐ray diffraction (XRD) and scanning electron microscopy (SEM) imaging. Devices are fabricated to study the charger transfer dynamics and the photovoltaic performance. Moreover, solar modules with 6 series‐connected sub‐cells are fabricated to study the scalability of this strategy. The operational stability of the cells and modules is investigated in accordance with the International Summit on Organic Photovoltaic Stability (ISOS) protocol.

2. Results and Discussion

As shown in Figure 1a, we designed and synthesized a novel asymmetric SAM molecule by incorporating a benzofuran unit. The benzofuran‐substituent was selected for its favorable electronic properties, enhanced thermal stability, and cost‐effectiveness. Accordingly, the electronegative oxygen in the benzofuran unit strengthens interfacial interactions and facilitates hole transfer, while its aromatic framework interacts with undercoordinated Pb²⁺ to passivate defects [20, 21]. As shown in Figure S1, BFC‐2PACz also exhibits a larger dipole (1.78 D) than Me‐4PACz (1.64 D). To eliminate the impact of varying –OH orientations in the phosphonic acid moieties, the dipole moment of the molecules without the phosphonic acid anchoring group have been performed by density functional theory (DFT) calculations. The results reveal that the dipole of the benzofuro[3,2‐c]carbazole core in BFC‐2PACz (1.52 D) is higher than that of the 3,6‐dimethylcarbazole core in Me‐4PACz (1.37 D). The reduction in dipole moment upon removal of the phosphonic acid group arises from the loss of polar side‐chain contribution. The larger dipole can promote a downward shift of the substrate Fermi level and improving charge extraction [22, 23]. As shown in Figure S2, the decomposition temperature of Me‐4PACz is 375°C, while it is 445°C for BFC‐2PACz. The higher decomposition temperature suggests a better thermal stability of BFC‐2PACz [24]. In addition, the lab‐scale synthetic cost of BFC‐2PACz is lower than that of Me‐4PACz according to a previous established estimation method (Table S1) [25]. The molecular structure and purity of BFC‐2PACz have been confirmed by ¹H and ¹³C nuclear magnetic resonance (NMR) spectroscopy, and high‐resolution mass spectrometry (HRMS), while Me‐4PACz was obtained commercially (Figures S3–S5). Figure 1b illustrates the slot‐die coating used for the NiO _x and SAM bilayer HTLs.

FIGURE 1.

(a) Synthetic route of BFC‐2PACz. (b) Schematic illustration of slot‐die coating process for NiO _x and SAM. (c) Schematic diagram of sample structure for the single‐lap shear test. (d) Statistics of critical shear stress (σ_c) values from samples with different HTLs. The mole ratio of Me‐4PACz to BFC‐2PACz ranges from 1:0 to 0:1. (e) Statistics distribution of σ_c value for samples with different SAMs on ITO|NiO _x . (f) The shear stress‐displacement curves obtained from samples with different SAMs on ITO|NiO _x . (g) XPS of the buried Cs_0.23FA_0.77PbI_2.53Br_0.47 perovskite films peeled off from Me‐4PACz, BFC‐2PACz, and mixed SAM. The sample structure is ITO|NiO _x |SAM|perovskite. (h) The peak current vs. scanning rate of Me‐PACz, BFC‐2PACz, and mixed SAM on ITO|NiO _x substrates in cyclic voltammetry measurements.

To evaluate the interfacial adhesion strength between the SAM and the perovskite and NiO _x layers, we conducted single‐lap shear tests according to the ASTM D1002 standard with data collected from five independent samples for each condition (Figure 1c) [26]. The critical shear stress (σ_c) of mixing SAM with different Me‐4PACz/BFC‐2PACz ratios (from 1:0 to 0:1) deposited on ITO or ITO|NiO _x substrates has been studied. Figure 1d presents the statistics of σ_c values for samples with different HTLs, using a structure of ITO|HTL|perovskite. The black curve corresponds to ITO|NiO _x |SAM, while the red curve corresponds to ITO|SAM. The results on these two substrates exhibit similar trends, indicating that the 1:1 mixing ratio (hereafter, named mixed SAM) provides the strongest adhesion to perovskite among all HTLs. Additionally, the σ_c values for ITO|NiO _x |SAM are consistently higher than those for ITO|SAM, highlighting the beneficial effect of NiO _x on adhesion. Notably, in ITO|NiO _x |BFC‐2PACz or mixed SAM|perovskite samples, perovskite residues were observed on the HTL side after testing, indicating stronger adhesion of BFC‐2PACz to ITO|NiO _x than to perovskite (Figure S6). This enhanced adhesion likely arises from two factors: (1) strong interactions between the lone‐pair electrons of the oxygen atom in BFC‐2PACz and undercoordinated Pb²⁺ in the perovskite, and (2) the more compact and higher coverage of the mixed SAM on NiO _x , which improves interfacial contact with the perovskite [16, 17]. The average σ_c values for the Me‐4PACz, BFC‐2PACz, and mixed SAM‐based samples are 3.52 ± 0.2, 4.15 ± 0.26, and 4.92 ± 0.18 MPa, respectively (Figure 1e, f).

To investigate the reasons for the differences of σ_c, we probed the interaction between the SAM and the perovskite layers by performing X‐ray photoelectron spectroscopy (XPS) measurements on the bottom surface of perovskite layer which peeled off from the HTLs [27]. As shown in Figure 1g, the Pb 4f peaks of perovskite films from BFC‐2PACz and mixed SAM samples shift toward lower binding energies compared to that peeled‐off from Me‐4PACz. This shift indicates an increased electron cloud density around the Pb, possibly due to the electron donation from the SAMs [28]. In contrast, the O 1s peak shifts toward a higher binding energy from 531.0 eV (Me‐4PACz) to 531.1 eV (BFC‐2PACz) and 531.3 eV (mixed SAM), respectively, which is likely due to a decreased electron density around oxygen atoms (Figure S7). The complementary shifts in the Pb 4f and O 1s levels confirm the electron donation from the oxygen atoms of the benzofuran unit to undercoordinated Pb²⁺, thereby providing clear evidence of a Lewis acid‐base interaction between the benzofuran unit and the perovskite surface [29].

The XPS of Ni 2p _3/2 was conducted to evaluate the anchoring strength of SAMs on NiO _x (Figure S8) [30, 31]. The Ni 2p_3/2 spectrum of the Me‐4PACz sample displays peaks at 856.2 and 854.2 eV, corresponding to Ni³⁺ and Ni²⁺, respectively. In contrast, the Ni³⁺ and Ni²⁺ peaks of BFC‐2PACz and the mixed SAM shift to lower binding energies, appearing at 855.9 and 855.7 eV for Ni³⁺, and 853.8 eV for Ni²⁺, respectively. The systematic shift toward lower binding energies indicates increased electron density around Ni species, suggesting stronger interfacial interactions between the mixed SAM and the NiO _x surface [14, 18, 32].

In addition, Fourier transform infrared (FTIR) spectroscopy was employed to further investigate the interactions between perovskite and BFC‐2PACz (Figure S9). In the mixture of BFC‐2PACz/PbI₂, the C─O stretching vibration peak shifts from 1183 to 1195 cm⁻¹. This shift is primarily due to the electron delocalization of C─O bond, which is induced by the interaction between the PbI₂ and benzofuran unit as Lewis base [33]. Considering that BFC‐2PACz solely mixed with PbI₂, the observed shift of the C─O bond is most likely due to the interaction between benzofuran and Pb²⁺ [29]. Cyclic voltammetry (CV) measurements were employed to quantify the coverage of the slot‐die‐coated SAM films (Figure 1h; Figure S10). The surface density of mixed SAM is 1.85 × 10¹⁴ molecules cm⁻², surpassing the 1.46 × 10¹⁴ and 1.52 × 10¹⁴ molecules cm⁻² of Me‐4PACz and BFC‐2PACz, respectively. According to previous reports, the larger molecular dipole increases the intermolecular electrostatic interactions between adjacent SAM molecules, which promotes stronger lateral packing and facilitates the formation of a more compact and uniform monolayer on the NiO _x surface. This enhanced intermolecular interaction reduces the likelihood of molecular vacancies, leading to improved surface coverage [34]. Moreover, the larger dipole moment simultaneously improves the molecular packing of the SAM on NiO _x and strengthens the interfacial interaction with the perovskite, resulting in better surface coverage and stronger adhesion [35]. Additionally, as shown in Figure S11, mixed SAM exhibited the highest proportion of the P 2p_1/2 peak area, indicating the more effective bonding between the SAM molecules and the NiO _x surface through the formation of P─O─Ni bonds [36].

In combination of the above results, the enhancement of interfacial adhesion mainly arises from two factors: (1) Lewis acid‐base interactions boost the adhesion between mixed SAM and the perovskite; and (2) a stronger binding between mixed SAM and NiO _x .

Then, we characterized the electronic structure of these HTLs by employing ultraviolet photoelectron spectroscopy (UPS) and absorption spectra (Figures S12 and S13). The energy‐level diagram was constructed to further investigate the interface energy level matching in Figure S14. The energy level offset (ΔE) decreases from 1.04 (Me‐4PACz) and 0.81 eV (BFC‐2PACz) to 0.70 eV for the mixed SAM. This is possibly due to the mixed strategy suppressed the agglomeration of the single molecular, which homogenized the distribution and facilitated the formation of a more uniform HTL at the buried interface, thereby optimizing the interfacial energy‐level alignment [37, 38]. This result indicates the potential of facilitating the hole transfer and reducing the nonradiative recombination losses, thus finally increasing the open circuit voltage (V _OC) [39]. In addition, no discernible difference was observed among the three HTLs by testing the transmittance spectra (Figure S15) [40]. Thus, the mixed SAMs has a higher coverage, better interface energy level regulation, and potentially better the device performance.

The contact angle of perovskite precursor on different HTLs was measured (Figure S16). The contact angle of the Me‐4PACz (23.4°) is higher than those of BFC‐2PACz (20.1°) and mixed SAM (20.5°), indicating a better wettability for achieving high‐quality perovskite films on mixed SAM film. To evaluate the surface energy of different HTLs, we measured the contact angles of C₂H₆O₂ and H₂O at 12 points on a 10 cm × 10 cm substrate, respectively (Figure S17a). The distribution of the contact angle values was shown in Figure S17b, with the maximum fluctuation of 0.20°. Following the Owens‐Wendt model, the average values of contact angles were used to calculate the surface energy. the surface energies of Me‐4PACz, BFC‐2PACz, and mixed SAM are 33.30, 34.81, and 36.53 mN m^{− 1}, respectively (Table S2). A higher surface energy facilitates a reduction of the heterogeneous nucleation energy barrier, thereby enabling a more uniform nucleation and more compact perovskite grain growth [41, 42, 43, 44]. Moreover, we further investigate the perovskite formation process by tracking the absorption spectra of the precursor film at different temperatures to better understand the nucleation process [45]. As shown in Figure S18, Cs_0.23FA_0.77PbI_2.53Br_0.47 undergoes a classical nucleation‐growth process. Following previous established methods, the activation energy values of Me‐4PACz, BFC‐2PACz, and mixed SAM samples are 53.7, 52.3 and 51.7 KJ mol⁻ [1], respectively. The lower activation energy of mixed SAM sample indicates that more nuclei were produced during the deposition process [46, 47].

The bandgap of perovskites on all HTLs in Figure S19 remains at 1.68 eV, suggesting that the change of HTL does not alter the bandgap of the perovskite [48, 49]. The XRD patterns of the perovskite films on mixed SAM exhibit a sharper and more intense (100) diffraction peak, with a narrower full width at half maximum (FWHM) compared to those on Me‐4PACz and BFC‐2PACz (Figure S20).

The surface and cross‐sectional morphologies of the perovskite films have been examined using SEM images [50]. Figure 2a–c show the top surface morphology of perovskite films deposited on different HTLs. The average grain sizes extracted from grain size distribution histograms follow the trend of mixed SAM (480 ± 270 nm) > BFC‐2PACz (396 ± 276 nm) > Me‐4PACz (364 ± 290 nm). This difference in grain size is consistent with cross‐sectional SEM images. As shown in Figure 2d–f, perovskite on the mixed SAM shows intact grains in the vertical direction with the largest size and the fewest boundaries.

FIGURE 2.

Top‐surface SEM images of perovskite films on (a) Me‐4PACz, (b) BFC‐2PACz, and (c) mixed SAM, with corresponding grain size distribution histograms. The cross‐sectional SEM image of perovskite films on (d) Me‐4PACz, (e) BFC‐2PACz, and (f) mixed SAM. SEM images of the bottom‐surface of perovskite films peeled‐off from (g) Me‐4PACz, (h) BFC‐2PACz, and (i) mixed SAM. The sample structure is ITO|NiO _x |SAM|perovskite.

Next, we studied the morphology of the bottom surfaces of perovskite films by peeling off from the HTL using ultraviolet (UV) glue (Figure S21) [51, 52]. As demonstrated in Figure 2g–i, the bottom surface of the Me‐4PACz‐based perovskite exhibits few NiO _x particles. In contrast, the bottom surfaces of perovskite films on BFC‐2PACz and mixed SAM contains numerous NiO _x particles, confirming the enhanced adhesion between the HTL|perovskite interface [53]. These results are consistent with the above results that the HTL based on mixed SAM effectively enhanced the interfacial adhesion and the perovskite crystallization.

To further study the carrier transport dynamics and interfacial non‐radiative recombination, steady‐state photoluminescence (PL) and time‐resolved PL (TRPL) were carried out [47, 54]. To separate the charge transfer and recombination dynamics within perovskite films and at the interface, we first studied the PL and TRPL spectra of samples w/o HTL by peeling‐off the perovskite from the HTL. As shown in Figure 3a, when perovskite films were not contact with HTLs, the PL intensity shows a trend of mixed SAM > BFC‐2PACz > Me‐4PACz, indicating that the perovskite film grown on the mixed SAM exhibits the lowest defect density. As shown in Figure 3b and Table S3, the perovskite peeled‐off from mixed SAM shows the longest lifetime, with a τ ₁ of 193.5 ns and an τ _avg of 1209.7 ns, while they are 178.2 ns and 1046.4 ns, 184.2 ns and 1061.5 ns for Me‐4PACz and BFC‐2PACz, respectively.

FIGURE 3.

(a) PL and (b) TRPL spectra of perovskite films peeled‐off from different HTLs. The sample structure is glass|perovskite and excited from the perovskite side. The SCLC curves for the hole‐only devices based on (c) Me‐4PACz, (d) BFC‐2PACz, and (e) mixed SAM. The sample structure is ITO|NiO _x |SAMs|perovskite|spiro‐OMeTAD|Ag.

When perovskite films were contact with HTLs, as shown in Figure S22a, the PL intensities of the perovskite films follow a trend of mixed SAM < BFC‐2PACz < Me‐4PACz. Moreover, the mixed SAM sample shows the shortest τ ₁ of 163.1 ns (Figure S22b and Table S4), compared to 211.4 and 197.7 ns for Me‐4PACz and BFC‐2PACz samples, respectively. The mixed SAM exhibits the strongest charge extraction capability, which is attributed to optimized interfacial contact as well as less interfacial defects. These results indicate that mixed SAM not only promotes the transfer of charges from the perovskite to the HTL, but also reduces the non‐radiative recombination reactions at the bottom surface of the perovskite film [55, 56].

To delve deeper into the mechanism of enhanced interfacial adhesion, we further characterized the temperature‐dependent PL spectra. As shown in Figure S23, when the temperature decreased from 303 to 203 K, all samples exhibit a redshift in peak positions, a narrowing of full‐width at half maximums (FWHMs), and an increase in PL intensity. The increase in PL intensity at lower temperatures can be attributed to the reduced thermal energy of photocarriers, which confines them closer to the band edges [55]. Notably, compared to Me‐4PACz and BFC‐2PACz, the mixed SAM sample shows a smaller shift over the temperature range of 303‐223 K. These results suggest that a reduced lattice mismatch or stress release at the interface in the mixed SAM sample, indicating an improved crystallinity of perovskite and stronger adhesion [57, 58].

To quantify the trap density (n _t) of perovskite based on various HTLs, space charge limited current (SCLC) curves for HTL‐only devices were measured (Figure 3c–e). The defect density of n _t was calculated according to the following formula:

$$n_{\mathrm{t}}=2\epsilon {\epsilon }_0{V}_{\mathrm{TFL}}/e{L}^2$$ (1)

where the V _TFL, ε, ε ₀, e, and L are the trap‐filled limit voltage, the relative dielectric constant, vacuum permittivity, the electronic charge, and the thickness of the perovskite film, respectively [55]. Devices based on Me‐4PACz, BFC‐2PACz, and mixed SAM exhibit V _TFL of 0.26, 0.19, and 0.15 V, corresponding to the n _t values of 7.15 × 10¹⁵ cm⁻³, 5.23 × 10¹⁵ cm⁻³, and 4.13 × 10¹⁵ cm⁻³, respectively. This result suggests that the perovskite film on mixed SAM film has the lowest defect density [59, 60].

We then fabricated PSCs having a p‐i‐n structure of ITO|NiO _x |SAMs|Cs_0.23FA_0.77PbI_2.53Br_0.47|C60:BCP|Ag to study the photovoltaic performance (Figure 4a). Notably, the HTLs consisted of NiO _x and SAMs were deposited by scalable slot‐die‐coating method, and the perovskite active layer is Cs_0.23FA_0.77PbI_2.53Br_0.47 with a bandgap of 1.68 eV. As shown in Figure 4b, the solar cells based on Me‐4PACz exhibit a highest PCE of 19.0%, with a V _OC of 1.17 V, a short‐circuit current density (J _SC) of 20.3 mA cm⁻², and a fill factor (FF) of 0.78. When BFC‐2PACz is employed, the solar cells achieved a champion PCE of 20.2%, accompanied by a V _OC of 1.21 V, a J _SC of 20.8 mA cm⁻², and an FF of 0.79. In contrast, the devices based on mixed SAM show the highest PCE of 21.2%, with a V _OC of 1.23 V, a J _SC of 21.2 mA cm⁻², and an FF of 0.82, which are comparable with recent reported values [14, 61, 62, 63]. Compared with devices based on a single SAM, the mixed SAM exhibit significantly enhanced FF and V _OC, which will be examined in detail in the following device physics characterizations. The steady‐state power output (SPO) of Me‐4PACz, BFC‐2PACz, and mixed SAM devices are 18.3%, 18.9%, and 20.6%, respectively, which are consistent with the J‐V curves (Figure 4c; Figure S24). The integrated J _SC values obtained from the incident photon‐electron conversion efficiency (IPCE) spectra are 20.1, 20.4, and 21.0 mA cm⁻² for Me‐4PACz, BFC‐2PACz, and mixed SAM cells, respectively, which show good consistency with the J _SC values obtained from the J‐V measurements (Figure 4d). Moreover, all solar cells show a good reproducibility (Figure 4e; Figure S25). An additional benefit of the mixed SAM is the reduction in J‐V hysteresis (Table S5). We calculated the hysteresis index (HI) by the Equation (2):

$$\mathrm{Hysteresis}\mathrm{Index}=\frac{PC{E}_{\mathrm{F}}-PC{E}_{\mathrm{R}}}{PC{E}_{\mathrm{F}}}\times 100\%$$ (2)

where PCE _R is the PCE in the reverse scanning direction, PCE _F is the PCE in the forward scanning direction [64]. The hysteresis index values are 7.9% for Me‐4PACz, 7.4% for BFC‐2PACz, and 6.0% for mixed SAM based devices, respectively. This reduction of device hysteresis can be attributed to the faster charge transfer at the HTL|perovskite interface and reduced defects within the perovskite fabricated on the mixed SAM, both of which minimized the charge accumulation at interfaces [65, 66, 67].

FIGURE 4.

Photovoltaic performance of ITO|NiO _x |SAMs|Cs_0.23FA_0.77PbI_2.53Br_0.47|C60:BCP|Ag solar cells. (a) The cross‐section SEM image of a typical PSC based on mixed SAM. (b) J‐V curves of the champion devices (scanning from forward bias to short circuit with a scan speed of 100 mV s⁻¹). (c) The stabilized power output (SPO) and steady‐state photocurrent nearby the maximum power point (MPP) for champion device based on mixed SAM. (d) IPCE spectra (solid) and corresponding integrated J _SC (dash‐dotted). (e) The PCE distribution of PSCs based on different SAMs. (f) J‐V curves under dark condition; inset shows the current data as the dV/dJ versus J ⁻¹ plots, with solid lines representing the linear fits. (g) MPP tracking results of encapsulated cells. (h) Thermal cycling test of the cells between 8°C and 60°C. All photovoltaic parameters were obtained under AM 1.5 G 1‐sun irradiation with an aperture area of 0.16 cm².

To understand the device performance variation, we conducted a series of characterizations. As shown in Figure 4f, we measured dark J‐V curves and extracted the series resistance (Rs) and the ideality factor (m) by fitting the data with Equations (3) and (4).

$$J_{\mathrm{dark}}=J_0\left[exp\left(\frac{qV}{m{k}_{\mathrm{B}}T}\right)-1\right]$$ (3)

$$\frac{dV}{d{J}_{\mathrm{dark}}}=\frac{m{k}_{\mathrm{B}}T}{J_{\mathrm{dark}}q}+R_{\mathrm{s}}$$ (4)

where J ₀, q, k _B, and T are the reverse saturation current, the elementary charge, the Boltzmann constant, and the cell temperature [45, 68]. The R _S of mixed SAM device is 1.35 Ω cm⁻², which is lower than that of the devices based on Me‐4PACz (2.39 Ω cm⁻²) and BFC‐2PACz (1.97 Ω cm⁻²). Additionally, the m value of the mixed SAM device is 2.13, lower than the values of 2.46 and 2.35 measured for the Me‐4PACz and BFC‐2PACz devices. The smallest R _S value of the mixed SAM device is due to the better contact within the devices, while the smallest m value indicates the lowest non‐radiative recombination reactions [47, 56].

To further elucidate the charge extraction and recombination dynamics, we conducted transient photovoltage (TPV) and transient photocurrent (TPC) measurements. The TPC decay lifetime decreases from 1.43 µs (Me‐4PACz) to 1.32 µs (BFC‐2PACz) and further to 1.21 µs (mixed SAM), verifying that mixed SAM device has the fastest charge extraction process (Figure S26a) [69]. The TPV decay lifetimes are 12.42, 20.35, and 30.95 µs for the devices based on Me‐4PACz, BFC‐2PACz, and mixed SAM, respectively (Figure S26b). The longest TPV decay lifetime confirms that mixed SAM device has the lowest non‐radiative recombination [70].

Subsequently, the J _SC and V _OC values under different light intensities have been recorded. The J _SC exhibit a linear relationship with light intensity (I): J _SC ∝ I^α [68]. The extracted α values are 0.96, 0.98, and 0.99 for devices based on Me‐4PACz, BFC‐2PACz, and mixed SAM, respectively, demonstrating the most efficient charge collection and minimal bimolecular recombination losses in the mixed SAM devices (Figure S27a) [30, 71]. The mixed SAM devices exhibit the smallest slope (1.37 k _B T q ⁻¹) in the V _OC versus light intensity plot, compared to slopes of Me‐4PACz (1.73k _B T q ⁻¹) and BFC‐2PACz (1.51k _B T q ⁻¹) (Figure S27b). This result indicates mixed SAM device has the most efficient charge collection and minimal recombination losses [47].

The operational stability of the PSCs was investigated through performance tracking tests under continuous illumination at the maximum power point (MPP) following the protocol of ISOS‐L‐1 [72]. As shown in Figure 4g, the devices based on mixed SAM maintain approximately 82.0% of its initial PCE after 1000 h, whereas the Me‐4PACz based devices drop to 71.0%. Previous reports show that the mechanical adhesion between the HTL and perovskite layer is the most critical parameters that affects the thermal cycling stability of PSCs [73]. Therefore, we studied the device performance evolution following a thermal cycling protocol of ISOS‐T‐1, i.e. 12 h at 8°C followed by 12 h at 60°C. As demonstrated in Figure 4h, the mixed SAM based devices retain 87.0% of its initial PCE after 30 cycles, which is outperforming than the devices based on Me‐4PACz (72.5%) and BFC‐2PACz (81.2%). To reveal the mechanism of the improved stability, the cross‐sectional SEM images of PSCs after the thermal cycling have been studied. As shown in Figure S28, obvious interfacial delamination can be observed at the HTL|perovskite interface in the samples based on Me‐4PACz and BFC‐2PACz. In contrast, PSCs with the mixed SAM maintain a mechanically stable interface, showing no delamination. Moreover, to study the relationship between stability and mechanical adhesion, the σ_c values of the devices aged by different temperatures cycles (8°C–40°C, 8°C–60°C and 8°C–80°C) were measured (Figure S29). The σ_c values of the mixed SAM are higher than those of Me‐4PACz and BFC‐2PACz across all tested conditions, and exhibit the smallest decrease, confirming the robustness of this strategy under different thermal cycling.

Beyond the intrinsic better thermal stability of BFC‐2PACz and stronger interfacial adhesion, according to previous reports, the enhanced stability of mixed SAM based devices can be explained by two other observed reasons [74, 75, 76, 77]. First, the better perovskite quality with a lower defect density and larger grain sizes, which suggests a suppressed trap‐assisted non‐radiative recombination within the active layer. Second, the improved interfacial properties caused by the higher coverage and better contact between the HTL and perovskite film, which not only facilitated the charge transfer, but also suppressed the recombination and degradation at the interfaces [74, 78].

To better evaluate the intrinsic advantages of the BFC‐2PACz and mixed SAM, PSCs employing Me‐4PACz, BFC‐2PACz, and the mixed SAM were fabricated via spin coating under identical conditions. The champion PCE of the mixed SAM devices reaches 21.0%, compared to 18.6% and 20.8% for the devices based on Me‐4PACz and BFC‐2PACz, respectively. As shown in Figure S30, the higher efficiency of the mixed SAM devices mainly arises from the increased V _OC and FF, consistent with the observed trend in the slot‐die‐coated devices. These results indicate that the advantage of the mixed SAM primarily originates from its interfacial modulation effect, rather than the differences in the deposition technique.

To evaluate the compatibility of the mixed SAM strategy for large area deposition, a slot‐die‐coated HTL with a size of 10 cm × 10 cm has been cut into 40 pieces (1.25 cm × 2 cm each) to fabricate PSCs. As shown in Figure S31, the maximum efficiency fluctuation is 0.9% with an average PCE of 20.6 ± 0.6% for 40 devices. This good spatial uniformity of the PSC performance across the 100 cm² size demonstrates the superb scalability of the slot‐die coating process for the mixed SAM.

We subsequently used the mixed SAM to fabricate larger‐area perovskite modules. All modules were fabricated on 5 cm × 5 cm substrates with series‐interconnected 6 sub‐cells (Figure 5a). The modules based on mixed SAM achieved a champion PCE of 16.8%, with a V _OC of 6.41 V, a I _SC of 34.9 mA, and an FF of 0.76 (Figure 5b). This performance is superior to those modules based on Me‐4PACz and BFC‐2PACz. The average PCE extracted from 15 independent modules for each type are 15.9 ± 1.0%, 15.0 ± 1.3%, and 9.8 ± 2.1% for modules based on mixed SAM, BFC‐2PACz, and Me‐4PACz (Figure 5c). The decrease in PCE of modules compared to small‐area cells arises from two aspects (Table S6): (1) thermal damage at the scribed edges during the laser scribi results in directly contact between the perovskite layer and the Ag electrode, increasing the current leakage and inducing the FF loss [79]; (2) in the series‐connected module, electrode resistance, internal device resistance, and electrical contact resistance cause more electrical losses, which decrease the V _OC and FF [80].

FIGURE 5.

(a) Schematic diagram of the modules with P1‐P2‐P3 laser scribing. (b) J‐V curves of modules based on Me‐4PACz, BFC‐2PACz, and mixed SAM. (c) The PCE distribution of 15 independent modules, which were derived from the forward bias to short circuit by J‐V measurements. (d) Thermal cycling test of the modules between 8°C and 60°C for 20 cycles. The modules have a structure of ITO|NiO _x |SAMs|Cs_0.23FA_0.77PbI_2.53Br_0.47|C60|SnO₂|ITO. All photovoltaic parameters were measured under simulated AM1.5G 1‐sun illumination with a metal mask of 10.0 cm².

We further evaluated the thermal cycling stability of the modules by an identical protocol of the small‐area devices. As shown in Figure 5d, the mixed SAM modules retained 86.2% of their initial performance after 20 cycles, while it is 76.5% for the BFC‐2PACz devices. And the Me‐4PACz module drops to 70.0% of their initial efficiencies after only 12 cycles. These results demonstrate that the mixed SAM substantially enhanced the stability of both small‐area and large‐area devices. In addition, we investigated the reverse bias stability of larger area PSCs by thermal IR images under ambient conditions. A constant reverse bias of 10.0 V was applied to the 5 cm × 5 cm modules for 200 s. As shown in Figure S32, the devices based on Me‐4PACz was breakdown after the test, while the modules based on BFC‐2PACz and mixed SAM are robust and show a more uniform temperature distribution across the 25 cm² area throughout the testing period. These results are attributed to the more uniform and compact SAM film, which results in lower leakage current under reverse bias.

3. Conclusions

This work demonstrated the mechanical adhesion between the HTL and perovskite layer by designing new SAM molecules with a larger dipole moment. This not only improves the photovoltaic performance, but also enhances the thermal cycle stability of WBG perovskite cells and modules. As a promising example, we designed and synthesized a novel SAM molecule, i.e. BFC‐2PACz that featuring a benzofuran substation, which enabled a simultaneous interaction with both NiO _x and perovskite. By mixing BFC‐2PACz with Me‐4PACz, an increased film coverage, enhanced perovskite film quality, and most importantly, improved mechanical adhesion of the HTL|perovskite interface have been achieved. By using a scalable slot‐die coating method for this HTL, we achieved efficiencies of 21.2% and 16.8% for the small‐area cell and larger‐area modules based on a wide bandgap (1.68 eV), respectively. Notably, continuous operation and thermal cycle stability tests demonstrated that this approach also enhanced the stability of both small‐area solar cells and large‐area modules.

4. Experimental Section

4.1. Materials

Methylammonium bromide (MABr) and formamidinium iodide (FAI) were purchased from Greatcell Solar Materials Pty Ltd. Lead bromide (PbBr₂), BCP, Methylammonium chloride (MACl) and C60 were purchased from Xi'an Yuri Solar Technology. NiO _x was purchased from Advanced Election. PbI₂, Me‐4PACz and other materials were purchased from TCI. All materials were directly used without further purification. The synthesis details of BFC‐2PACz have been provided in the supporting information.

4.2. ITO Substrates Preparation

The pre‐patterned ITO glass substrates underwent a sequential cleaning process by ultra‐sonication in detergent, deionized water, and ethanol for 15 min each. After drying, ITO was treated by ultraviolet ozone (UVO) for 15 min before usage.

4.3. HTL Fabrication

The NiO _x layer was fabricated via slot‐die coating. 5 mg mL⁻¹ NiO _x precursor ink in isopropanol/deionized water (1:3, v/v) was filtered through a 0.22 µm polyether sulfone filter, then it was coated onto preheated ITO substrates (30°C) at a coating speed of 8 mm s⁻¹ and a solution feed rate of 8 µL s⁻¹. After coating, it was annealed at 140°C for 20 min. SAM precursor solutions were prepared by dissolving Me‐4PACz, BFC‐2PACz, or their 1:1 molar ratio mixture (mixed SAM) in isopropanol at a total concentration of 1.51 mM. The SAM solution was slot‐die coated onto the NiO _x film at 20 mm s⁻¹ with a feed rate of 20 µL s⁻¹, followed by thermal annealing at 100°C for 10 min. All slot‐die coating processes were performed at ambient conditions with a temperature of 25°C–30°C and a relative humidity of 20%–40%.

4.4. Perovskite Layer

1.34 M perovskite precursor of Cs_0.23FA_0.77PbI_2.53Br_0.47 was prepared by dissolving PbI₂, FAI, PbBr₂, PbCl₂, MACl, CsI, and 5 mol% MAPbCl₃ in DMF and DMSO with a volume ratio of 4:1. Then, the precursor was deposited onto the SAM substrates (area: 1.25 cm × 2 cm) via a one‐step spin coating method at 5000 rpm for 40 s, and 120 µL ethyl acetate was dropped on the film at the final 10 s before the finishing of spinning. The films were then annealed at 100°C for 30 min.

4.5. ETL and Counter Electrode

23 nm C60 and 8 nm BCP were thermally evaporated onto the perovskite film using a thermal evaporator. Finally, the device fabrication was finished by thermal evaporation of 90 nm Ag as a contact‐electrode.

4.6. Module Fabrication

A nanosecond laser was acquired from Suzhou Microtreat Intelligent Technology Co. Ltd to conducted the P1, P2, and P3 processes. P1 patterns were etched using an infrared laser with a wavelength of 1064 nm, while P2 and P3 patterns were etched with an ultraviolet laser with a wavelength of 355 nm. Initially, the ITO substrate was etched to create P1 lines and subsequently cleaned with detergent, deionized water, and ethanol to remove the residue. The P1 lines (≈ 40 µm wide) were produced using an average laser power of 10.8 W and a pulse repetition frequency of 20 kHz. The processes for preparing the NiO _x , SAM, perovskite, C60, and BCP layers were followed the same protocols as those for the small‐area device. The P2 lines (≈ 120 µm wide) were created with an average laser power of 4.2 W and a pulse repetition frequency of 50 kHz. Following the deposition of the Ag electrode, the P3 lines, approximately 300 µm wide, were patterned using an average laser power of 5.6 W and a pulse repetition frequency of 50 kHz. The module was tested under a 10.0 cm² shadow mask.

4.7. Characterizations

DFT calculations were performed using the Gaussian 16, Rev. A03 program package, with geometry optimization carried out at the B3LYP‐D3(BJ)/6‐31+G(d) level. XPS and UPS were characterized on a Thermo Fisher ESCALAB 250Xi system, with He I radiation for UPS measurements. PL and TRPL spectra were collected on a HORIBA Delta Flex Fluorescence Lifetime System with a 485 nm excitation source. UV–vis absorption spectra were recorded on a PerkinElmer Lambda 750S spectrophotometer. XRD patterns were recorded with a PANalytical B.V Empyrean diffractometer using a Cu K_α X‐ray tube operating at 40 kV, 60 mA. SEM imaging was performed on a Hitachi S4800 microscope. J‐V curves were measured using a Keithley 2400 source meter under AM 1.5G simulated solar irradiation (100 mW cm⁻²), calibrated with a standard silicon reference solar cell. IPCE was measured on a Newport PT‐QEM1000 system. Long‐term operational stability was evaluated via MPP tracking under continuous LED 1‐sun illumination. Thermal cycling stability was tested between 8°C and 60°C in the dark.

Conflicts of Interest

The authors declare no conflicts of interests.

Supporting information

Supporting File: smll74000‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.