Non-competitive additives regulate molecular assembly for compact monolayers in perovskite/silicon tandems

Abstract

Achieving densely packed self-assembled monolayers (SAM) is essential for constructing well-defined functional interfaces, yet monolayer formation is often hindered by molecular aggregation and additive-induced competition for surface anchoring sites. Here we report a non-competitive additive strategy that regulates molecular assembly during SAM deposition. Using 1,3,5-tris(trifluoromethyl)benzene (TTMB) as a non-anchoring additive, aggregation of the SAM molecule is suppressed while all substrate binding sites remain available for molecular anchoring. Because TTMB interacts only weakly with the surface, it does not participate in surface binding but increases the population of anchoring-capable molecules and strengthens SAM-substrate interactions, enabling the formation of densely packed monolayers. The resulting interfaces exhibit improved energetic alignment, reduced non-radiative recombination, and suppressed shunt pathways. When implemented in monolithic perovskite/silicon tandem solar cells, the compact SAM enables a certified stabilized power conversion efficiency of 33.66%. This non-competitive additive concept provides a general strategy for directing molecular assembly in monolayer materials.

Dense self-assembled monolayers are vital for functional interfaces, but molecular aggregation and additives can block surface anchoring. Zeng et al. used a weakly binding additive to guide assembly, forming compact layers that improved tandem solar-cell efficiency.

Introduction

Self-assembled monolayers (SAMs) offer a powerful platform for tailoring interfacial properties through the spontaneous organization of molecular building blocks^1–3. However, achieving compact and uniformly anchored monolayers remains challenging because intermolecular interactions during deposition often disrupt compact assembly^4,5. In particular, amphiphilic SAM molecules tend to form aggregates in solution and during adsorption, which reduces the population of anchoring-active species and leads to incomplete surface coverage and heterogeneous interfacial structures^6,7.

In perovskite/silicon tandem solar cells, SAM is widely used as hole-transport layers at buried interfaces^8–10. However, aggregation of commonly used SAM molecules limits the achievable surface density and compromises charge extraction across the interface. Considerable efforts have therefore been devoted to increasing the effective surface coverage of SAM^10–14. One strategy focuses on substrate engineering, such as increasing surface hydroxylation to create additional anchoring sites^15–17. A second strategy modulates intermolecular interactions through molecular design, including the introduction of steric hindrance to suppress aggregation^10–12, tailoring carbazole units to promote ordered packing^18–20, or employing highly polar solvents to adjust intermolecular affinity^21,22. A third and widely adopted approach is co-adsorption, in which SAM molecules are deposited together with additives to alleviate aggregation and cover exposed substrate region by additives^23–28. However, existing co-adsorption strategies rely on competitive adsorption, in which additive molecules occupy surface anchoring sites that would otherwise be available for SAM molecules attachment. Given that these co-adsorbed additives typically possess inferior charge-transport characteristics compared with SAM molecules, this competition potentially limits the effective SAM coverage and interfacial uniformity.

In this study, we introduce a non-competitive adsorption strategy that decouples molecular packing from surface anchoring during SAM formation. By incorporating the non-anchoring additive 1,3,5-tris(trifluoromethyl)benzene (TTMB), intermolecular aggregation of the SAM molecule is suppressed while preserving all available anchoring sites on the substrate. Owing to its weak interaction with the surface, TTMB is removed during adsorption and does not participate in surface binding. At the same time, TTMB strengthens SAM-substrate interactions and promotes a tilted adsorption geometry that favors compact molecular packing. These combined effects substantially increase the effective surface coverage of the monolayer, leading to improved interfacial energetics, reduced recombination losses, and mitigated shunt pathways. When implemented in monolithic perovskite/silicon tandem solar cells, the resulting compact SAM enable devices with enhanced open-circuit voltage, fill factor, efficiency, and stability. Although the reported co-deposition strategy and our non-competitive adsorption strategy both employ additives during SAM formation, they differ fundamentally in interfacial chemistry and assembly pathway. In our approach, the additive does not bind to the substrate, thereby leaving all anchoring sites available for SAM molecules adsorption.

Results

Non-competitive adsorption strategy for compact SAM

We introduce a non-competitive adsorption strategy that suppresses intermolecular aggregation while preserving all available surface anchoring sites, enabling the formation of high-coverage SAM. Guided by this principle, we selected TTMB (Fig. 1a), which contains three electron-withdrawing –CF₃ groups that effectively suppress (2-(9H-carbazol-9-yl)ethyl)phosphonic acid (2PACz) aggregation while containing no functional groups capable of bonding with the indium tin oxide (ITO) substrate. For comparison, two carboxyl-containing additives were investigated: trimesitic acid (TMA, three –COOH groups) and 3,5-bis(trifluoromethyl)benzoic acid (BTBA, one –COOH and two –CF₃ groups) (Fig. 1a). The -COOH groups in TMA and BTBA can bind to ITO, leading to competitive adsorption with 2PACz.

Fig. 1. Additive-assisted enhancement of 2PACz surface coverage.

a Chemical structures of 2PACz and the investigated additives. b P/In XPS atomic ratios of SAM/ITO/glass samples prepared from pristine 2PACz and 2PACz with different additives. c C 1s XPS spectra of SAM/ITO/glass samples deposited from pristine 2PACz and additive-containing 2PACz solutions. d DLS measurements of 2PACz solutions without and with different additives. e P/In XPS atomic ratios of SAM/ITO/glass samples prepared from unfiltered and filtered 2PACz solutions. f P/In XPS atomic ratios of SAM/ITO/glass samples prepared from filtered 2PACz with TTMB added after filtration (denoted as f-2PACz + TTMB) and from multiple deposition cycles using filtered 2PACz solution.

The influence of additives on monolayer formation and surface coverage was quantified by X-ray photoelectron spectroscopy (XPS), contact-angle measurements, Kelvin probe force microscopy (KPFM) and energy-dispersive X-ray spectroscopy (EDS) mapping in scanning transmission electron microscopy (STEM-EDS) mode. SAMs were deposited by spin-coating 0.4 mg/mL 2PACz with or without additives (additive:2PACz = 1:4, optimized via concentration-dependent experiments, Supplementary Fig. 1), followed by annealing, washing, and re-annealing. The P/In atomic ratio from XPS served as an indicator of effective SAM surface density Supplementary Fig. 2)^17,19,22,29. Compared to P/In atomic ratio of 0.130 for pristine 2PACz, the P/In ratio increased to 0.145 for 2PACz/TMA, 0.170 for 2PACz/BTBA, and 0.232 for 2PACz/TTMB (Fig. 1b), corresponding to a 78% enhancement in effective surface density. Consistently, contact-angle measurements on both planar ITO and textured silicon substrates showed increased values after TTMB addition (Supplementary Figs. 3 and 4), indicating greater exposure of hydrophobic carbazole moieties and hence higher SAM coverage. KPFM revealed a systematic shift in surface potential from −190 mV (pristine 2PACz) to −245 mV (2PACz/TMA), −324 mV (2PACz/BTBA), and −385 mV (2PACz/TTMB) (Supplementary Fig. 5). A larger surface potential shift reflects a higher effective dipole density directly correlated with SAM coverage, thus the largest shift for 2PACz/TTMB sample confirms the highest SAM packing density. STEM-EDS elemental mapping on ITO-coated TEM grids further showed stronger P signal for TTMB-treated samples (Supplementary Fig. 6), corroborating the increased coverage.

To distinguish non-competitive adsorption from conventional competitive co-adsorption during SAM formation, the interfacial binding behavior of the additives was investigated by XPS. No –CF₃ signal was detected for 2PACz/TTMB (Fig. 1c), confirming that TTMB does not bind to ITO. In contrast, 2PACz/TMA produced a clear –COOH signal, and 2PACz/BTBA gave both –COOH and –CF₃ signals (Fig. 1c), indicating that TMA and BTBA co-adsorb and occupy anchoring sites. For BTBA, anchoring occurs via –COOH, while the –CF₃ signal arises from BTBA molecules co-assembled within the SAM, not from substrate binding. Importantly, even when the TTMB concentration was increased by two orders of magnitude, no TTMB-related signal was observed immediately after spin coating (Supplementary Fig. 7), indicating spontaneous desorption of TTMB prior to annealing and washing. This non-adsorbing behavior was also verified on pyramidal textured silicon bottom cells under various conditions (unannealed/unwashed, annealed/unwashed, and annealed/washed), where no –CF₃ signal was detected even at elevated TTMB concentrations (Supplementary Fig. 8). Time-of-flight secondary ion mass spectrometry (TOF-SIMS) after deposition further confirmed the absence of TTMB-related –CF₃ on textured substrates (Supplementary Fig. 9). These results demonstrate that TTMB does not adsorb on either flat or textured interfaces prior to annealing and washing, confirming that the non-competitive assembly mechanism is valid for textured substrates as well.

Molecular assembly behavior during SAM deposition

To elucidate how TTMB regulates molecular assembly, we investigated the aggregation behavior of 2PACz in solution using dynamic light scattering (DLS). As shown in Fig. 1d, pristine 2PACz solutions (0.4 mg/mL) form large micellar aggregates exceeding 100 nm in size. After addition of TTMB, TMA, or BTBA, the aggregate size decreased to below 5 nm, with the 2PACz/TTMB system exhibiting the smallest aggregates (~ 1 nm) (Fig. 1d).

Scanning electron microscopy (SEM), KPFM and XPS measurements further reveal that aggregates in pristine films are only weakly adsorbed on the surface and can be readily removed by washing (Supplementary Figs. 10–12). Consistently, filtering the precursor solution to remove aggregates does not change the P/In ratio of the resulting SAMs (Fig. 1e and Supplementary Figs. 13–15). These results indicate that large micellar aggregates do not participate in SAM anchoring on the ITO surface.

Taken together, these observations suggest that solution aggregation effectively reduces the population of anchoring-capable molecular species, thereby limiting the achievable surface coverage during SAM formation. By suppressing intermolecular clustering in solution, TTMB increases the fraction of individual molecules available for surface anchoring, enabling more efficient monolayer assembly.

Beyond suppressing aggregation, we examined whether TTMB also facilitates anchoring of individual 2PACz molecules. TTMB addition to filtered, aggregate-free solutions could increase the P/In ratio from 0.131 to 0.146 (Fig. 1f and Supplementary Fig. 16). By contrast, repeated deposition of filtered 2PACz without TTMB produced only marginal increases (0.133 after two cycles and 0.141 after ten cycles). These results demonstrate that TTMB directly promotes the anchoring of individual 2PACz molecules on the ITO substrate.

Molecular mechanism of non-competitive adsorption

To elucidate the origin of the non-competitive adsorption behavior enabled by TTMB, we combined molecular dynamics (MD) simulations and first-principles density functional theory (DFT) calculations to probe the adsorption statistics, packing configuration, and underlying energetic interactions among 2PACz, TTMB, and the ITO substrate.

MD simulations show that the simulated monolayer coverage increases markedly in the presence of TTMB. MD simulations were initiated with 300 uniformly distributed 2PACz molecules, either without additives or in the presence of 75 additive molecules, above a 12 × 12 nm ITO substrate without initial bonding. After reaching equilibration (Fig. 2a), unanchored molecules were removed to emulate the experimental washing process (cross-sectional view in Fig. 2b and top-view in Supplementary Fig. 17). The surface coverage of anchored molecules increased from 22.8% for pristine 2PACz to 43.2% for 2PACz/TTMB (Fig. 2c and Supplementary Fig. 18), with the number of anchored molecules rising from 39 to 86 (Fig. 2d). TMA and BTBA also enhanced coverage but to a lesser extent (Fig. 2c, d). These trends are insensitive to the In-O cutoff distance used to define anchoring (2.46 Å here; see Supplementary Fig. 19).

Fig. 2. Influence of additives on the adsorption behavior of 2PACz.

a Classical MD simulations showing the equilibrium configurations of 2PACz on ITO, starting from 300 randomly distributed, unbonded 2PACz molecules, in the absence or presence of 75 additive molecules. b Cross-sectional view of anchored 2PACz and additives after removing unbonded species from (a). c Surface coverage of anchored molecules on ITO. d Number of anchored 2PACz and additive molecules. e Number of 2PACz monomers and clusters at equilibrium from (a). f Anchoring numbers and anchoring probabilities of different 2PACz species on ITO for pristine 2PACz and 2PACz/TTMB at equilibrium. g Distribution of anchoring angles for bonded 2PACz in pristine 2PACz and 2PACz/TTMB. h DFT-calculated binding energy of 2PACz on ITO for pristine 2PACz and 2PACz/TTMB. i DFT-calculated binding energies of representative functional groups in 2PACz, TMA, BTBA, and TTMB on ITO, respectively. j Number fraction of total bonded species contributed by additives and 2PACz in each system from (b) and Supplementary Fig. 17.

Cluster analysis reveals that TTMB suppresses higher-order 2PACz aggregates on ITO substrate while increasing anchoring-capable species. The number of 2PACz monomers rises from 90 to 172, accompanied by a decrease in trimers and tetramers (Fig. 2e). Trimers and tetramers exhibit much lower anchoring probabilities than monomers and dimers (Fig. 2f). Solution-phase MD simulations further confirm that TTMB suppresses 2PACz aggregation in solution (Supplementary Figs. 20 and 21). This redistribution increases the population of anchoring-active species, enhancing monolayer formation. Moreover, TTMB shifts 2PACz from a nearly flat-lying configuration (~ 10°) to a tilted orientation (~ 50°) (Fig. 2g), facilitating denser packing and minimizing steric blockage of anchoring sites.

Fourier-transform infrared spectroscopy (FTIR) and DFT calculations show that TTMB preferentially interacts with 2PACz. FTIR reveals a redshift of the carbazole C–H vibration upon TTMB addition, indicating intermolecular interaction between TTMB and 2PACz (Supplementary Fig. 22). DFT shows that the 2PACz-TTMB interaction (−0.445 eV) is stronger than 2PACz-2PACz (−0.365 eV), while TTMB-TTMB is much weaker (−0.199 eV) (Supplementary Figs. 23 and 24). Thus, TTMB disrupts 2PACz self-association, increasing intermolecular spacing and reducing binding energy (Supplementary Figs. 25–27). In the presence of TTMB, the binding energy of 2PACz on ITO substrate strengthens from −4.53 to −5.05 eV (Fig. 2h and Supplementary Fig. 28). Adsorption-energy calculations show that the –CF₃ groups in TTMB interact only weakly with ITO (around −0.1 eV), whereas –COOH groups in TMA and BTBA bind strongly (around −2.4 eV) (Fig. 2i and Supplementary Fig. 29). Consequently, TTMB does not occupy anchoring sites, while TMA and BTBA compete directly with 2PACz for anchoring—accounting for ≈21.1% and 10.6% of anchored species, respectively (Fig. 2j and Supplementary Fig. 17). Transient and equilibrium MD simulations confirm that TTMB rapidly migrates away from the interface and does not adopt a stable adsorption configuration (Supplementary Figs. 30 and 31), demonstrating that TTMB is intrinsically non-adsorbing and leaves all anchoring sites available for SAM molecules.

Together, these MD and DFT results demonstrate that the non-competitive TTMB additive avoids competitive site occupation, suppresses SAM aggregation, strengthens SAM anchoring, and promotes a denser packing geometry, thereby enabling substantially enhanced SAM coverage on ITO substrate.

Interfacial electronic properties of high-coverage SAM

To evaluate the interfacial electronic properties enabled by the high-coverage SAM formed via non-competitive adsorption, TTMB-assisted monolayers were further investigated in perovskite films.

Ultraviolet photoelectron spectroscopy (UPS) was employed to evaluate the energy-level alignment at the SAM/perovskite interface. As shown in Fig. 3a and Supplementary Fig. 32, the valence band maximum (VBM) of 2PACz/TTMB is located at −5.78 eV, slightly deeper than that of pristine 2PACz (−5.74 eV) on ITO. This downward shift results in improved energetic alignment with the perovskite VBM (−6.04 eV), thereby facilitating more efficient hole extraction. The VBM downshift arises from the enhanced interfacial dipole due to increased SAM coverage, not from a direct electronic contribution of TTMB. To verify this, we tuned 2PACz coverage by varying deposition concentration without TTMB. Higher coverage led to a progressive VBM downshift, accompanied by improved open-circuit voltage (V_OC) and power conversion efficiency (PCE) (Supplementary Figs. 33 and 34), confirming that the energetic shift originates from coverage-induced dipole enhancement.

Fig. 3. Effect of TTMB additive on optoelectronic properties of perovskite films.

a Energy-level alignment between SAM (pristine 2PACz and 2PACz/TTMB) and the perovskite film. E_F, E_VAC and E_VBM denote the Fermi level, vacuum level, and valence band maximum, respectively. b Steady-state PL spectra of perovskite films on SAM/ITO/glass under excitation from perovskite side. c Reconstructed pseudo-J-V curves and corresponding photovoltaic parameters derived from Suns-QFLS plots under excitation from perovskite side. d Suns-QFLS plots with fitted diode ideality factors under excitation from perovskite side. e Steady-state PL spectra of perovskite films on SAM/ITO/glass under excitation from ITO-side. f SCLC characteristics of ITO/SAM/perovskite/Spiro-OMeTAD/Cu devices. g Bulk carrier lifetime and SRV extracted from thickness-dependent TRPL decay fitting. h FF loss analysis relative to the SQ limit based on pseudo-J-V curves and measured J-V curves. i Dark J-V characteristics of perovskite solar cells.

We performed photoluminescence measurements under both perovskite-side and ITO-side excitation. Under perovskite-side excitation, where the signal reflects bulk perovskite properties, the 2PACz/TTMB sample exhibits higher photoluminescence (PL) intensity (Fig. 3b), higher implied open-circuit voltage (iV_OC) (Fig. 3c), higher quasi-Fermi level splitting (QFLS) with a smaller diode ideality factor (Fig. 3d), and longer carrier lifetime (Supplementary Fig. 35a), consistently indicating improved perovskite film quality and reduced non-radiative recombination. Under ITO-side excitation, photogenerated carriers are created near the HTL/perovskite interface, making the PL signal sensitive to interfacial charge transfer. More efficient hole extraction enabled by TTMB reduces the carrier population available for radiative recombination, leading to weaker PL (Fig. 3e) and faster time-resolved photoluminescence (TRPL) decay (Supplementary Fig. 35b) under ITO-side excitation, even though the perovskite quality is improved (as confirmed by a smaller ideality factor from Suns-QFLS analysis in Supplementary Fig. 36)^21,28,30. Thus, TTMB-assisted SAM both improves perovskite film quality and enhances hole extraction.

Space-charge-limited current (SCLC) measurements (Fig. 3f) revealed a decrease in trap density from 1.68 × 10¹⁵ cm⁻³ for pristine 2PACz to 1.21 × 10¹⁵ cm⁻³ for the TTMB-assisted device, reflecting the improved perovskite film quality. Thickness-dependent TRPL analysis using a double-heterostructure model^31–33 showed the surface recombination velocity (SRV) is significantly lower for perovskite films on 2PACz/TTMB (Fig. 3g and Supplementary Fig. 37). DFT calculations suggest that carbazole groups in 2PACz passivate deep-level Pb-vacancy defects (V_Pb²⁺) (Supplementary Fig. 38), and thus the higher SAM coverage afforded by TTMB further reduces SRV by strengthening interfacial passivation.

We performed FF-loss analysis relative to the Shockley-Queisser (SQ) limit (Fig. 3h). The TTMB-assisted device shows reduced non-radiative recombination losses and a smaller gap between the theoretical maximum (FF_max) and the measured FF. The reduced transport loss is primarily attributed to the suppression of leakage pathways enabled by the higher surface coverage of the 2PACz monolayer. The denser and more uniform SAM effectively minimizes exposed ITO regions, thereby increasing the shunt resistance and reducing carrier losses through local shunting channels. In addition, the improved interfacial uniformity creates a more homogeneous energetic landscape and reduces local extraction barriers. These effects, together with a possible reduction in interfacial contact resistance, facilitate more efficient hole extraction and contribute to the smaller gap between FF_max and the measured FF. Consistently, the lower dark saturation current density in Fig. 3i confirms suppressed leakage and recombination.

Overall, the non-competitive adsorption enabled by TTMB optimizes energy-level alignment, accelerates hole extraction, improves perovskite film quality, and mitigates shunting pathways, leading to superior optoelectronic performance of perovskite films.

Solar cell performance

To demonstrate the practical implications of the high-coverage SAM enabled by the non-competitive adsorption strategy, we implemented this approach in perovskite solar cells and perovskite/silicon tandem solar cells.

Single-junction perovskite solar cells were fabricated with the architecture depicted in Fig. 4a. Devices incorporating 2PACz/TTMB achieved a PCE of 23.17%, V_OC of 1.270 V, and FF of 85.49%, outperforming control devices based on pristine 2PACz (PCE of 22.05%, V_OC of 1.250 V, FF of 82.78%; Fig. 4b). The stabilized efficiency reached 23.10% with negligible hysteresis, and the measured short-circuit current density (J_SC) matched the external quantum efficiency (EQE)-integrated current density (Supplementary Fig. 39). The superior performance arises from simultaneous increases in V_OC and FF (Fig. 4c, d), underscoring the effectiveness of the non-competitive adsorption strategy in mitigating interfacial losses. By comparison, devices incorporating TMA or BTBA additives exhibited only moderate performance gains (Supplementary Fig. 40), underscoring the advantage of the non-competitive adsorption mechanism enabled by TTMB. Consistent with the formation of a more compact monolayer, TTMB-treated devices exhibited lower 1/R_sh (Fig. 4e), indicating suppressed leakage pathways. Statistical analyses of R_sh indued FF and PCE losses further revealed reduced average losses and narrower variation (Supplementary Fig. 41), highlighting the improved reproducibility.

Fig. 4. Performance of single-junction cells and perovskite/silicon tandem solar cells.

a Device architecture of single-junction perovskite solar cells. b J-V curves, c PCE statistics, d FF statistics, and e 1/R_sh statistics for perovskite solar cells employing SAM of pristine 2PACz and 2PACz/TTMB. Box plot represents the standard deviation, the central line shows the median, and the white square represents the mean value. f Device architecture of perovskite/silicon tandem solar cells. g J-V curves of tandem devices based on pristine 2PACz and 2PACz/TTMB. h EQE and total absorptance (1-R, where R is the reflectance) of the 2PACz/TTMB-based tandem device. i MPP tracking of unencapsulated tandem devices under 1000 W/m² illumination from LED-based solar simulator at room temperature in N₂ glovebox.

To evaluate the generality of this strategy, we extended the approach to multiple SAMs, including (3-((9-ethyl-9H-carbazol-3-yl)oxy)propyl)phosphonic acid (HTL201), 1-cyano-2-(4-(3,6-dimethoxy-9H-carbazol-9-yl)phenyl)vinyl)phosphonic acid (MCZ-CPA)), and (4-(3,6-Dimethyl-9H-carbazol-9-yl)butyl)phosphonic acid (Me-4PACz), as well as alternative non-anchoring CF₃-containing additives such as (1-(trifluoromethyl)naphthalene and bis[3-(trifluoromethyl)phenyl]methanone. In all cases, the additives increased the P/In ratio and improved device performance (Supplementary Figs. 42–44), demonstrating the broad applicability of the non-competitive adsorption concept across different SAM/additive combinations.

We further extended this strategy to perovskite/silicon tandem solar cells (Fig. 4f). The control tandem device with pristine 2PACz achieved a PCE of 32.52%, V_OC of 1.972 V, J_SC of 20.40 mA/cm², and FF of 80.84%. In contrast, the champion device incorporating TTMB reached a PCE of 34.13%, V_OC of 1.992 V, J_SC of 20.43 mA/cm², and FF of 83.86% (Fig. 4g), delivering a stabilized power output of 34.03% with negligible hysteresis (Supplementary Fig. 45). The measured J_SC was consistent with EQE integration (Fig. 4h). An independently certified stabilized PCE of 33.66% was confirmed by the Fujian Institute of Metrology (Supplementary Fig. 46). Statistical analysis (Supplementary Fig. 47) demonstrates that TTMB incorporation consistently enhances device performance, with notable gains in V_OC, FF, and PCE.

Operational stability assessed via maximum power point (MPP) tracking showed that the 2PACz/TTMB-based tandem device retained 90% of its initial efficiency after 980 h of continuous 1000 W/m² illumination from a full-spectrum LED solar simulator at room temperature, whereas control devices exhibited a T₉₀ of 222 h (Fig. 4i). The corresponding illumination spectrum is provided in Supplementary Fig. 48, demonstrating coverage across the UV, visible, and NIR regions. The substantially enhanced operational stability is attributed to the formation of a denser 2PACz monolayer enabled by TTMB. By minimizing uncovered substrate regions and reducing interfacial defect density, the compact SAM establishes a more robust buried interface that is less susceptible to light-induced interfacial degradation.

Discussion

We demonstrate a non-competitive adsorption strategy that employs the non-anchoring molecule TTMB to regulate the formation of SAMs. By suppressing the aggregation of 2PACz molecules in solution while preserving all available surface anchoring sites, TTMB increases the population of anchoring-capable molecules and promotes a tilted, densely packed adsorption geometry on the ITO substrate. Molecular simulations further reveal strengthened SAM-substrate interactions in the presence of TTMB, which together enable the formation of high-coverage monolayers. The resulting compact SAM interfaces improve interfacial energy-level alignment, suppress non-radiative recombination, and mitigate shunting pathways in perovskite devices. When implemented in monolithic perovskite/silicon tandem solar cells, this strategy enables a certified stabilized PCE of 33.66% together with improved operational stability.

Beyond device performance, this work highlights the importance of controlling molecular assembly during monolayer deposition. By decoupling intermolecular packing from surface anchoring competition, the non-competitive adsorption concept provides a general strategy for directing molecular organization at functional interfaces, offering new opportunities for the rational construction of high-quality SAM in optoelectronic and energy-conversion systems.

Methods

Materials

Formamidinium iodide (FAI), methylammonium chloride (MABr), and propane-1,3-diammonium iodide (PDAI₂) were purchased from Greatcell Solar. Lead bromide (PbBr₂), cesium iodide (CsI), N,N-dimethylformamide (DMF), dimethyl sulfoxide (DMSO), isopropanol (IPA), acetone, and anisole were purchased from Sigma Aldrich. Lead iodide (PbI₂) and 2PACz were purchased from TCI. TTMB was purchased from Macklin. Trimesitinic acid (TMA) and BTBA were purchased from Aladdin Holdings Group Co., Ltd. C₆₀ and bathocuproine (BCP) were purchased from Xi’an Yuri Solar Co., Ltd. Magnesium fluoride (MgF₂) was purchased from Yingkou Libra New Energy Technology Co., Ltd. Polytetrafluoroethylene (PTFE) filter was purchased from Jiangsu Green Union Science Instrument Co., Ltd. All materials were used as received without further purification.

Fabrication of the single-junction perovskite solar cells

ITO substrates were sequentially cleaned by ultrasonication in IPA, acetone, and deionized water for 20 min each. Prior to SAM deposition, the ITO surfaces were treated with UV-ozone for 25 min. 2PACz was dissolved in IPA at a concentration of 0.4 mg/mL and spin-coated at 3000 rpm for 30 s, followed by annealing at 100 °C for 10 min, IPA washing at 3000 rpm, and a second annealing at 100 °C for 5 min. For additive-modified 2PACz solutions, TTMB, TMA, or BTBA was added to the 2PACz solution with additive:2PACz molar ratio of 1:4. A 1.7 M perovskite precursor solution with the composition Cs_0.05MA_0.15FA_0.80Pb(I_0.72Br_0.28)₃ was prepared by dissolving CsI, FAI, MABr, PbBr₂, and PbI₂ in a mixed solvent of DMF and DMSO (4:1 v/v). The solution was stirred for 8 h and filtered through a 0.22-μm PTFE filter before use. The precursor was spin-coated at 500 rpm for 5 s and 4000 rpm for 45 s, 300 μL of anisole was dropped as an antisolvent at the 12th second before the end of the second spin step. The films were then annealed sequentially at 60 °C for 1 min and at 100 °C for 30 min. For surface passivation, a 0.5 mg/mL PDAI₂ in IPA was spin-coated onto the annealed perovskite films at 4000 rpm for 30 s, followed by annealing at 100 °C for 10 min. Finally, C₆₀ (25 nm) was thermally evaporated at a rate of 0.1 Å/s, followed by 4 nm of BCP at the same rate, and a 100 nm Cu electrode thermally evaporated at 1.0 Å/s (ZHDS500, Beijing Technol Science Co., Ltd.).

Fabrication of textured silicon bottom cells

Silicon heterojunction bottom cells were prepared using n-type Czochralski silicon wafers with double-side submicrometer pyramidal texturing (210 × 210 mm, resistivity of 1–5 Ω·cm, thickness of 150 μm). The intrinsic, n-type and p-type hydrogenated amorphous silicon layers were deposited by plasma-enhanced chemical vapor deposition. Subsequently, ITO electrodes with thicknesses of approximately 80 nm on the rear side and 20 nm on the front side were deposited by sputtering through a 10.5 × 10.5 mm shadow mask. Silver rear electrodes were then deposited via screen printing. Laser dicing separated the wafers into 25 × 25 mm cells, centering the 10.5 × 10.5 mm ITO squares by aligning to the wafer corners.

Fabrication of perovskite/silicon tandem solar cells

Prior to tandem integration, silicon bottom cells were cleaned by spin-rinsing with IPA at 3000 rpm for 30 s to remove surface contaminants. After UV-ozone treatment, the SAM was deposited by 0.4 mg/mL 2PACz with TTMB (TTMB:2PACz = 1:4), after which the perovskite film and surface passivation layer were prepared according to the procedures described above. An electron-transporting C₆₀ layer with a thickness of 15 nm was subsequently thermally evaporated at a rate of 0.1 Å/s. A conformal SnO₂ buffer layer (~ 10 nm) was then deposited by atomic layer deposition (ALD) at 100 °C for 80 cycles (TALD-150D, Jiaxing Kemin Electronic Equipment Technology Co., Ltd.). Next, a 70 nm indium zinc oxide transparent electrode was deposited onto the SnO₂ layer by sputtering (JCPS400, Beijing Technol Science Co., Ltd.). Metallic grid contacts were formed by thermally evaporating 150 nm of Ag through a patterned shadow mask. To complete the tandem architecture, a 100 nm magnesium fluoride (MgF₂) layer was thermally evaporated at 1 Å/s as an antireflection coating.

Film characterizations

XPS and UPS measurements were conducted using an ESCALAB XI electron analyzer (Thermo Fisher Scientific). The contact angle was measured with a KRUSS DSA100 system. KPFM measurements were performed in tapping mode using SPM-9700HT atomic force microscope (Jiesen Industrial Co., Ltd.) and Oxford Instrument MFP-3D. STEM-EDS elemental mapping was performed using a probe-corrected JEM-ARM200F transmission electron microscope (JEOL, Japan) operated at an accelerating voltage of 200 kV. TOF-SIMS measurements were performed using a TOF.SIMS 5 instrument (IONTOF GmbH, Germany). DLS measurements were carried out using a Malvern Zetasizer Nano ZS90. A field-emission scanning electron microscope (FEI VERIOS 460) was used to analyze the surface morphologies. PL and TRPL spectra were measured using a Horiba DeltaFlex fluorescence lifetime spectrometer equipped with a 405 nm pulsed laser as the excitation source. QLFS was determined using a LuQY Pro LP20-32 radiative efficiency meter (QYB) under a range of illumination intensities. Suns-QFLS analysis was performed according to established methods^31,34.

Device characterization

J-V characteristics of single-junction perovskite solar cells and perovskite/silicon tandem devices were measured at room temperature under ambient conditions using a Keithley 2400 source meter coupled with a dual-beam AM 1.5 G solar simulator (RHS-50SS, Ravi). The illumination intensity was calibrated to 1 sun (1000 W/m²) using two certified silicon reference cells from the Fujian Metrology Institute: one with a KG5 filter and one with a quartz filter (both calibrated on July 14, 2025). Device aperture areas were defined by photomasks: 0.0544 cm² for single-junction cells and 1.0 cm² for tandem cells. J-V scans for single-junction devices were recorded from −0.1 to 1.3 V with 10 mV steps and 500 ms dwell per step, while tandem devices were measured from −0.2 to 2.1 V with 20 mV steps and the same dwell time. TPV measurements were conducted using a transient photocurrent/photovoltage system (LST-TPCV, Shanghai Jinzhu Technology Co., Ltd.). EQE spectra were obtained using an Enlitech QE-R system. For sub-cell characterization of tandem devices, selective light biasing was applied using an 850 nm long-pass filter for top-cell measurements and a 550 nm short-pass filter for bottom-cell measurements. During the measurements, bias voltages of 0.5 and 1.0 V were applied to the tandem devices for the top- and bottom-cell EQE measurements, respectively. Device operational stability was evaluated at the maximum power point under 1000 W/m² illumination of an LED-based solar simulator (LST-LED20, Shanghai Jinzhu Technology Co., Ltd.) at room temperature in N₂-filled gloveboxes, using an MPP tracking algorithm (Shenzhen Purui Material Technology Co., Ltd.).

Simulations

Classical MD simulations were performed using the LAMMPS package³⁵ to investigate the adsorption behavior of 2PACz and its mixtures with three additives (TMA, BTBA, or TTMB) on an ITO surface. The ITO (001) substrate (120 × 120 Å) was described using the Buckingham potential with parameters adopted from previous studies³⁶. All organic molecules were modeled using the OPLS-AA force-field, with initial parameters generated by LigParGen^37,38, and a modified force-field scheme was applied to accurately describe the phosphonic acid anchoring group⁶. Interactions between organic molecules and the ITO substrate were modeled using Lennard-Jones and Coulombic potentials, with cross-interaction parameters derived from geometric mixing rules based on universal force-field parameters³⁹. Initial configurations were generated using PACKMOL⁴⁰ by randomly placing 300 2PACz molecules, with or without 75 additive molecules, 10 Å above the ITO surface, while enforcing a minimum intermolecular separation of 2 Å. All systems were equilibrated in the NVT ensemble at 300 K for 5.0 ns with the ITO slab held rigid. Periodic boundary conditions were applied in all directions, and the equations of motion were integrated using the velocity-Verlet algorithm with a time step of 1 fs⁴¹. All subsequent analyses were performed using Python code.

First-principles DFT calculations were performed by using planewave pseudopotential methods as implemented in the Vienna Ab initio Simulation Package^42,43. For the exchange-correlation functional, the Perdew-Burke-Ernzerhof functional was adopted^44,45. The 5s²5p² (Sn), 5s²5p¹ (In), 2s²2p⁴ (O), 3s²3p³ (P) were treated as valence electrons. A kinetic cutoff energy of 520 eV was used for electronic wavefunction expansion and k-point grid spacing of 0.25 Å⁻¹ was used for Brillouin zone integration. Structure optimization was performed using the conjugate gradient technique, which ensures residual forces smaller than 0.05 eV/Å. In addition, to properly take into account the long-range van der Waals (vdW) interaction that is non-negligible for perovskites involving organic molecules, DFT-D3 method was adopted.

To demonstrate the effect of the additives on the intermolecular interaction of 2PACz molecules, the binding energy (E_b) of 2PACz/2PACz/additive was defined as E_b = E_AB–E_A–E_B, where E_AB, E_A, and E_B are the total energies of the optimized system, 2PACz, and 2PACz/additive, respectively. The surface adsorption was carried out on the ITO (001) slab with a size of approximately 20 × 10 Å in the x and y directions, together with a vacuum layer of 15 Å in the out-of-plane direction. The ITO slab was constructed based on the method of Dong et al.⁴⁶. We considered as many adsorption configurations as possible to explore the most energetically stable adsorption configuration between additives and the ITO (001) facets⁴⁷. During the structure optimization, the bottom atomic layers were fixed, and atoms of top layers and the additives were allowed to relax. The binding energy of 2PACz (or additive) on the ITO substrate was defined as the energy difference between the total energy of the 2PACz-ITO (or additive-ITO) system (E₃) and the sum of the energies of the isolated ITO surface (E₁) and the free 2PACz (or additive) molecule (E₂): E_b = E₃–(E₁ + E₂).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Supplementary information

Supplementary information ^{(28MB, pdf)}

Reporting Summary ^{(1.3MB, pdf)}

Transparent Peer Review file ^{(16.7MB, pdf)}

Source data

Source Data ^{(248.6KB, xlsx)}

Author contributions

B.C. and L. Zeng conceived the idea. B.C. and G.Y. designed and supervised the research. L. Zeng carried out the DLS, XPS, and SCLC measurement. Y.W., Z.L. and T.L. performed KPFM measurement. X.Z. carried out MD simulations. B.Q., X.W., and L. Zhang carried out the DFT calculations. Y.G. performed the PLQY, QFLS, and pseudo-J-V characterization. Y.Z. conducted PL and TRPL measurement. L. Zeng and Y.L. fabricated perovskite solar cells. N.L. optimized ALD and sputtering. K.Q. fabricated silicon bottom cells. L.Y., M.W., C.Z., and Q.C. fabricated tandem solar cells. B.C., L. Zhang, and L. Zeng wrote the paper, and all authors reviewed the paper.

Peer review

Peer review information

Nature Communications thanks Zhong’an Li, Zhiqin Ying, Jiang Liu and Zonglong Zhu for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by National Natural Science Foundation of China [Grant No. 52273273 (B.C.), 62321166653 (L.Z.), 62125402 (L.Z.), and T2521005 (L.Z.)], Shaanxi Qinchuangyuan Project [Grant No. 2025QCY-KXJ-186 (G.Y.) and QCYRCXM-2023-081 (B.C.)], and Jilin Province Science and Technology Development Program [Grant No. 20260602032RC (X.W.)].

Data availability

All data generated in this study are provided in the main text, Supplementary Information, and Source Data file. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76266-0.