Introduction

Tin-based perovskites, with their small hole effective masses, efficient charge transport, high charge carrier diffusion lengths, weak ion migration, and solution-processability, are promising candidates for low-cost, scalable production of high-performance electronic devices1,2,3,4,5. However, the lower activation energy and faster crystallization rate of the three-dimensional (3D) structure in tin-based halide perovskites often lead to reduced film quality and increased defect density6,7. These defects, such as grain boundaries and point defects, can severely degrade the material’s performance by acting as recombination centers for charge carriers, reducing the efficiency of devices like solar cells and field-effect transistors (FETs). Additionally, the oxidation of Sn2+ to Sn4+ generates tin vacancies within the film, contributing to a high level of p-type doping8,9,10,11. In contrast, quasi-two-dimensional (quasi-2D) perovskites—formed by introducing bulky organic spacer cations into the 3D lattice—exhibit enhanced structural and electronic properties12,13,14. The hydrophobic nature of these spacer cations mitigates ion migration and moisture/oxygen ingress while simultaneously regulating crystal growth and passivating interfacial defects, thereby reducing both morphological and electronic disorder15,16.

The general formula of these quasi-2D systems, (A′)2(A)n-1BnX3n+1, reflects the number of perovskite layers (n) separated by organic spacers (A′). Early FET studies focused primarily on n = 1 pure 2D perovskites17,18,19,20,21, which, despite offering excellent environmental stability and interfacial passivation, exhibit severely constrained out-of-plane charge transport. This is largely due to the complete disruption of corner-sharing connectivity between adjacent [SnI6]4− octahedra by the insulating organic layers, resulting in deep quantum wells that localize charge carriers3. In contrast, n = 2 phases offer a suitable balance between structural dimensionality and electronic performance22. They retain the defect tolerance and moisture resistance of lower-dimensional structures while exhibiting narrower bandgaps and enhanced inter-octahedral coupling, thereby enabling more efficient charge transport. In addition, owing to their intrinsically anisotropic layered structure, n = 2 phases preferentially crystallize with strong out-of-plane orientation, which, when acting as a structural template, directs the subsequent vertical epitaxial growth of high-n or 3D components and thereby enhances the overall crystallinity and orientation uniformity of the perovskite film. However, quasi-2D perovskite films processed with a specific stoichiometric ratio still generate a series of phases with significantly varying n values, each exhibiting distinct orientations18,23,24. Typically, the growth rate of high-n-value phases is faster than that of low-n-value ones25. The competitive growth among these components results in small grain sizes, random orientation, and structural disorder, leading to increased film defects and reduced charge carrier mobility26. Therefore, rational design and precise control of the crystallization of quasi-2D perovskites are required to achieve highly vertically oriented growth of high-n-value structures.

In the study, we propose an effective strategy for sequentially growing various layer (n) phases in quasi-2D perovskites, producing high-quality films with low defect density and high orientation. The incorporation of PEASCN induces well-formed PEA2FAn-1SnnI3n-1SCN2 (n = 2) templates at room temperature. Simultaneously, replacing FAI with FAHCOO and NH4I prevents FAI-SnI2 binding, thus inhibiting uncontrolled formation of the 3D FASnI3 phase. Based on the optimized perovskite films, the fabricated FETs exhibit mobility of 43 cm2 V−1 s−1 and an on/off ratio exceeding 108, with minor changes in electrical performance after 30 days of storage in a nitrogen environment.

Results

Bilayer template-guided crystallization and phase evolution

We fabricated tin-based perovskite thin films using a single-step spin-coating method with anti-solvent dripping, followed by annealing at 100 °C for 10 min. The reference quasi-2D perovskite films, denoted as PEAI-based films, were synthesized from a precursor solution, which contained phenethylammonium iodide (PEAI), formamidinium iodide (FAI), SnI2, and SnF2 in a molar ratio of 0.34:0.83:1:0.1 and was dissolved in a solvent mixture of dimethyl sulfoxide (DMSO) and N,N-dimethylformamide (DMF). In contrast, the target samples, referred to as PEASCN–FAHCOO-based films, were synthesized from a precursor solution comprising phenethylammonium thiocyanate (PEASCN), FAHCOO, NH4I, SnI2, and SnF2 in a molar ratio of 0.34:0.83:0.83:1:0.1. More details of the molar ratio optimization and the spin-coating optimization are shown in Supplementary Figs. 1–5.

The effects of introducing PEASCN and replacing FAI with FAHCOO and NH4I on the crystallization process of perovskite thin films were investigated by analyzing X-ray diffraction (XRD), glancing wide-angle X-ray scattering (GIWAXS), and photoluminescence (PL) spectra at different annealing times. The XRD patterns (Fig. 1a) show that PEASCN-FAHCOO-based films exhibit pronounced diffraction peaks of the bilayer quasi-2D phase PEA2FAn−1SnnI3n-1SCN2 (n = 2) prior to annealing, accompanied by weak signals from the 3D FASnI3 phase. To verify the identity of this n = 2 phase, we compared the experimental XRD data with simulated patterns derived from the crystallographic structure of PEA2FASn2I5SCN2 (Supplementary Fig. 6). The simulated peaks show excellent concordance with the low-angle diffraction features observed experimentally, validating the structural consistency of the n = 2 phase. This close match between simulation and experiment provides supporting evidence for the existence and oriented growth of the bilayer phase in our films. With annealing, the bilayer peaks gradually weaken while the 3D-related peaks intensify, indicating that the low-n bilayer structure acts as a template to direct the growth of the 3D domains and concurrently transforms into higher-n structures. This phase evolution mitigates dielectric and quantum confinement effects, promoting vertical grain orientation and efficient charge transport. The structural transition is schematically illustrated in Fig. 1b, visually depicting the gradual transformation from the bilayer quasi-2D phase to higher-n quasi-2D or 3D phases, emphasizing the template-guided crystallization mechanism.

Fig. 1: Crystallization dynamics of quasi-2D thin films.
Fig. 1: Crystallization dynamics of quasi-2D thin films.
Full size image

a XRD patterns of the PEASCN–FAHCOO-based films at different annealing times. 2 L denotes the bilayer (n = 2) Ruddlesden–Popper perovskite phase corresponding to PEA2FASn2I5SCN2, while 3D denotes the three-dimensional FASnI3 phase. b Schematic diagrams of the structural evolution of PEASCN–FAHCOO-based films during the annealing process. c GIWAXS patterns of the unannealed PEASCN–FAHCOO-based films. d GIWAXS patterns of the PEASCN–FAHCOO-based films after annealing for 1 min. e PL spectra of the PEASCN–FAHCOO-based films at different annealing times. f XRD patterns of the PEAI-based films at different annealing times. g XRD patterns of the annealed PEASCN–FAHCOO-based and PEAI-based perovskite films, with a partially enlarged view of the (100) diffraction peak. h Azimuth curves of the integrated ring at qr = 1.0 Å−1.

The thermodynamic and kinetic mechanisms underlying this process were elucidated by density functional theory (DFT) calculations and nuclear magnetic resonance (NMR) analysis. The DFT results indicate that the SCN−-doped bilayer phase PEA2FASn2I5SCN2 possesses a reduced formation energy compared to the iodide-only analog PEA2FASn2I7, suggesting that SCN− incorporation thermodynamically favors bilayer phase formation (Supplementary Fig. 7). Meanwhile, the 119Sn NMR spectra show a pronounced upfield chemical shift in FAHCOO-containing samples relative to FAI-based ones (Supplementary Fig. 8), indicating the formation of stable Sn2+–FAHCOO complexes. This interaction slows down the nucleation of the 3D phase, stabilizes the bilayer intermediate, and promotes a gradual and controlled phase transition. During spin-coating and subsequent annealing, FAHCOO and NH4I gradually react to generate FAI, while the byproduct NH4HCOO volatilizes upon annealing. The delayed release of FAI provides a sufficient time window for the formation of low-dimensional (e.g., n = 2) intermediate phases, which act as structural templates guiding the ordered growth of the 3D FASnI3 phase.

To verify whether FAHCOO is permanently incorporated into the perovskite lattice, the TOF-SIMS depth profiling was conducted (Supplementary Fig. 9). The results show negligible presence of HCOO⁻ in the final bulk phase, indicating that FAHCOO mainly functions as a transient crystallization modulator rather than a permanent lattice dopant. Further evidence of crystallization kinetics regulation was obtained via chlorobenzene anti-solvent extraction tests. When 1 mL of chlorobenzene was added to 50 μL of FAI + SnI2 solution, immediate precipitation of yellow solids was observed, reflecting rapid nucleation of perovskite microcrystals (Supplementary Fig. 10). In contrast, the FAHCOO + NH4I + SnI2 solution remained clear, indicating significantly suppressed crystallization. The quaternary PEASCN + FAHCOO + NH4I + SnI2 formulation exhibited delayed precipitation, with yellow solids emerging only after ~5 s, consistent with the controlled formation of low-dimensional (n = 2) perovskite phases. These results suggest that replacing FAI with FAHCOO and NH4I effectively slows down nucleation and crystallization, providing an extended time window for the self-assembly of bilayer templates.

GIWAXS and PL measurements further substantiate the above conclusions. Before annealing, PEASCN–FAHCOO-based films exhibit distinct Bragg spots characteristic of the bilayer phase, corresponding to the (002), (004), and (006) planes, together with faint Debye–Scherrer rings from the FASnI3 (100) plane (Fig. 1c), indicating the coexistence of the bilayer phase and minor 3D components. After annealing, the Bragg spots disappear, and the (100) Debye-Scherrer ring becomes sharper and more defined (Fig. 1d), signifying the dominance of higher-n quasi-2D or 3D phases. Correspondingly, PL spectra collected at different annealing stages (Fig. 1e) show a redshift in the emission peak, reflecting increased structural dimensionality. In addition, perovskite thin films containing PEASCN but without FAHCOO and NH4I substitution (Supplementary Fig. 11) also exhibit the initial formation of the bilayer phase. However, substantial 3D phase signatures are already present before annealing, suggesting a limited templating effect and insufficient guidance for ordered phase evolution. During annealing, although the film undergoes phase transformation, the resulting structure shows slightly inferior out-of-plane orientation, as evidenced by GIWAXS intensity integration. This moderate loss in structural coherence correlates with a subtle decline in device performance, highlighting the key role of FAHCOO and NH4I in stabilizing the intermediate bilayer phase, promoting gradual dimensional growth, and improving crystallographic orientation (Supplementary Figs. 12 and 13). Additionally, the PEAI-based films exhibit fewer bilayer structures before annealing (Fig. 1f), suggesting that SCN− plays a crucial role in bilayer formation. The PEAI-based films also show rapid 3D crystallization, evident from the strong diffraction peaks at room temperature.

Enhanced crystallinity and morphology of perovskite films

XRD analysis of post-annealing films (Fig. 1g) shows that the PEASCN–FAHCOO-based films display narrower, more intense diffraction peaks at the (100) plane compared to PEAI-based films, indicating higher crystallinity. Numerical integration of the diffraction intensities (Fig. 1c, h, and Supplementary Fig. 12) reveals that the PEASCN–FAHCOO-based films exhibit enhanced out-of-plane orientation, as reflected by the diffraction signal concentration around 90°, further confirming their higher crystallinity.

Moreover, observations from scanning electron microscopy (SEM) indicate that introducing PEASCN and replacing FAI with FAHCOO and NH4I enhances the morphology of tin-based perovskite films (Fig. 2a, b). The PEAI-based film shows abundant small grains and a relatively rough surface, indicating the presence of disordered 3D components and a random distribution of bulky PEA+ cations within the bulk, which hinders uniform crystal growth. In contrast, the PEASCN–FAHCOO-based film exhibits a compact and smooth surface with substantially fewer pinholes. These minimal pinholes are likely attributed to the incomplete coverage of the substrate by the initial low-n quasi-2D components during the early crystallization stage. Atomic force microscopy (AFM) measurements provide precise information on the surface roughness of the two films (Supplementary Fig. 14), with Rq values of 7.4 and 5.5 nm for the PEAI-based and PEASCN–FAHCOO-based films, respectively. The reduced roughness ensures tighter contact between the perovskite and the interface, facilitating smooth carrier transfer.

Fig. 2: Characterizations of perovskite films.
Fig. 2: Characterizations of perovskite films.
Full size image

Typical SEM images of PEASCN–FAHCOO-based (a) and PEAI-based (b) films. Scale bar, 200 nm. c High-resolution XPS spectra of the Sn 3d regions in PEAI-based and PEASCN–FAHCOO-based films. d Contact angles of PEAI-based and PEASCN–FAHCOO-based films. e PL spectra of PEAI, PEASCN–FAHCOO, and FASnI3 films. f TRPL spectra of PEAI-based and PEASCN–FAHCOO-based films.

The suppression of Sn2+ oxidation was further elucidated through a combination of theoretical calculations, precursor-level evaluation, and surface-sensitive characterizations. Bader charge analysis revealed that the amount of electron loss from Sn atoms decreased from 1.11 e in the undoped control sample to 0.91 e upon SCN− incorporation. This reduced electron loss indicates a higher electron density around Sn, which can be attributed to the coordination interaction between the nitrogen lone pair of the SCN− anion and Sn2+, thereby enhancing the local electron shielding. The increased electron density around Sn2+ effectively mitigates its oxidation to Sn4+, contributing to chemical stability of the perovskite under ambient conditions. Furthermore, theoretical calculations show that SCN− incorporation leads to a more negative formation energy compared to I− (Supplementary Fig. 15), suggesting a thermodynamically favorable doping process. This enhanced thermodynamic stability is attributed to the Sn–SCN⁻ coordination, which reinforces the local lattice structure and contributes to the long-term durability of the perovskite film. This atomic-level stabilization is further supported by X-ray photoelectron spectroscopy (XPS) analysis (Fig. 2c), which shows a lower surface Sn4+ content in the PEASCN–FAHCOO-based film compared to the PEAI-based counterpart. Furthermore, the ToF-SIMS analysis (Supplementary Fig. 16) reveals that, after annealing, PEA⁺ cations predominantly accumulate near the film surface, forming a PEA⁺-rich interfacial layer. This surface-enriched layer not only facilitates defect passivation by reducing undercoordinated Sn-related trap states but also acts as a hydrophobic barrier that effectively suppresses the ingress of moisture and oxygen, thereby mitigating the oxidation of Sn2+. Water contact angle measurements (Fig. 2d) support this interpretation, with the PEASCN–FAHCOO-based film exhibiting a much larger contact angle than the PEAI-based film, indicative of improved surface hydrophobicity. In addition, the antioxidation capability of the precursor solutions was qualitatively evaluated by monitoring color changes under ambient conditions. As shown in Supplementary Fig. 17, colorimetric monitoring under ambient conditions (25 °C, 40% RH) reveals that the PEAI-based precursor rapidly turns dark red within 30 min, indicating severe oxidation of Sn2+ to Sn4+. In contrast, the PEASCN-FAHCOO-based precursor remains yellow with only slight surface darkening, indicating improved resistance to oxidation. To identify the key contributor to this antioxidation behavior, we conducted comparative tests using binary FAI + SnI2 and FAHCOO + SnI2 solutions under identical conditions (Supplementary Fig. 18). The FAI + SnI2 solution showed rapid oxidation, whereas the FAHCOO + SnI2 solution maintained its original state over time. Such stabilization can be ascribed to two factors: (i) the strong coordination between the carboxylate group of FAHCOO and Sn2+, which protects Sn2+ from oxidative species, and (ii) the weak acidity of FAHCOO, which creates a less oxidative chemical environment in the precursor solution. Collectively, these effects help preserve Sn2+ in its reduced state and suppress premature degradation during film formation.

Subsequently, PL and time-resolved photoluminescence (TRPL) spectroscopy were utilized to analyze the prepared perovskite thin films’ steady-state optical properties and fluorescence carrier dynamics. In Fig. 2e, the fluorescence peak of PEASCN–FAHCOO-based film is closer to that of the 3D structure, suggesting a higher prevalence of high-n-value structures in the film, thereby supporting three-dimensional carrier transport. Additionally, the fluorescence intensity of the PEASCN–FAHCOO-based film is greater than that of the PEAI-based film, suggesting reduced defects and inhibited non-radiative carrier recombination within the film. In the TRPL spectrum (Fig. 2f), the fitting results reveal that the fluorescence lifetime of the PEASCN–FAHCOO-based film is 3.6 ns, markedly more prolonged than that of the PEAI-based film at 0.9 ns, aligning with the observed trend in the fluorescence spectra. The longer carrier lifetime in the PEASCN-FAHCOO-based film can be attributed to its enhanced orientation and reduced oxidation, associated with lower defect density27. These effects result from multiple synergistic contributions: PEASCN induces bilayer template formation for ordered crystallization; FAHCOO slows down 3D nucleation through Sn2+ coordination; and the PEA+-enriched surface layer passivates traps and blocks moisture/oxygen, contributing to perovskite films with improved crystallinity, orientation, and stability.

High-performance transistors via optimized perovskite films

We constructed FET devices with a bottom-gate/top-contact configuration, as depicted in Fig. 3a, b, illustrating carrier transport with the utilization of the two perovskite thin films as semiconductor active layers. The transfer curves in the saturation region (Fig. 3c and Supplementary Fig. 19) and the output curves (Fig. 3d) demonstrate that the PEASCN–FAHCOO-based FETs deliver higher drain currents than the PEAI-based counterparts. The key parameters, extracted from the transfer curves in Fig. 3c, show that the PEASCN–FAHCOO-based devices achieve a mobility of up to 43 cm2 V−1 s−1 (with statistical distribution summarized in Fig. 3e), along with a threshold voltage of 0.2 V, an on/off ratio of 1.62 × 108, and a subthreshold swing of 0.66 V dec−1. In comparison, the PEAI-based FETs exhibit a lower mobility of 8 cm2 V−1 s−1 and an on/off ratio of 6.93 × 107. The observed improvements in device performance can be attributed to the optimized crystallization of the high-n quasi-2D perovskite structure and its enhanced vertical alignment, which is associated with a reduced defect density and supports efficient charge carrier transport. To clarify the impact of crystallization kinetics on device performance, we optimized the heat treatment temperature (Supplementary Fig. 20) and systematically investigated the properties of PEASCN–FAHCOO-based FETs under different heat treatment durations (0−10 min at 100 °C) at the optimal heat treatment temperature. FETs fabricated from unannealed films (predominantly low-n 2D phases) and those annealed for 30 s (with coexisting low-n and high-n phases) both exhibit significant hysteresis and operational instability (Supplementary Fig. 21). This is attributed to the presence of substantial low-n phases, which lead to severe interfacial trapping—manifested in scan-rate-dependent transfer characteristics and notable threshold voltage shifts over 100 consecutive operating cycles. In contrast, films annealed for at least 1 min, which achieve stable crystallization, result in devices with negligible hysteresis and excellent operational stability. This improvement arises from the completion of phase transition to high-n phases, reducing interfacial traps and enhancing overall film quality, underscoring the critical role of sufficient annealing time in ensuring reliable device performance.

Fig. 3: Device performance characterization.
Fig. 3: Device performance characterization.
Full size image

Main carrier transport pathways in (a) PEAI-based and (b) PEASCN–FAHCOO-based bottom-gate top-contact FETs, including carrier injection, vertical transport, and interface transport. c Transfer curves under VDS = −30 V. d Output curves of PEAI-based and PEASCN–FAHCOO-based FETs. e Forward-scan saturation mobilities of PEAI-based and PEASCN–FAHCOO-based FETs. f Contact resistance extracted using the TLM method. Drain current noise power spectral density of (g) PEAI-based and (h) PEASCN–FAHCOO-based FETs at different VGS. i Mobility statistics of 200 PEAI-based and PEASCN–FAHCOO-based FETs.

To ensure the reliability of the transistor performance metrics, we analyzed the transfer curves in the linear region (VDS = −3 V), given that mobility in the generally saturated region typically surpasses that in the linear region (Supplementary Fig. 22). The findings, with mobilities of 41 cm2 V−1 s−1 for the PEASCN–FAHCOO-based FET and 6.9 cm2 V−1 s−1 for the PEAI-based FET, closely mirroring those in the saturation region, validate the reliability of mobility extraction. This suggests that even at lower VDS, the carrier mobility of the PEASCN–FAHCOO-based FET can be accurately reflected, further demonstrating that the mobility extracted under different operating conditions is consistent and reliable. Subsequently, the contact resistance was evaluated using the Transmission Line Method (TLM), which involves measuring the total resistance (\({R}_{{{{\rm{T}}}}}\)) of transistors with varying channel length L in the linear regime. The obtained data were fitted to a linear RT – L plot, where the total resistance comprises both the channel resistance and the contact resistance. As shown in Fig. 3f and Supplementary Fig. 23, the slope of the fitting line yields the sheet resistance \({R}_{{{{\rm{s}}}}}\), and the y-axis intercept corresponds to 2Rc. From this, the normalized contact resistance Rc × W (where W is the channel width) is extracted, showing a decrease from 114 Ω cm for the PEAI-based FET to 76 Ω cm for the PEASCN–FAHCOO-based FET. This reduction in contact resistance indicates an improved interfacial contact between the perovskite layer and the gold electrode, facilitating more efficient carrier injection and reducing energy losses at the interface. We employed low-frequency noise (LFN) analysis to assess the trap density of PEAI-based and PEASCN–FAHCOO-based FETs28. Figure 3g, h illustrates the correlation between drain current fluctuation power spectral density (PSD) and frequency ( f ) across different gate voltages for both types of transistors. Each line, corresponding to a different gate voltage, exhibits a clear 1/f-type dependence on f. Notably, the PEASCN–FAHCOO-based FETs consistently exhibit lower drain current fluctuation (SI) than PEAI-based devices across all tested gate voltages, indicating reduced trap-assisted carrier scattering and improved interfacial quality. To quantitatively evaluate the trap density, we extracted the interfacial trap density (Ntrap) based on the subthreshold swing (SS) using the equation:

$$\begin{array}{c}{N}_{{{{\rm{trap}}}}}=\left[\frac{{{{\rm{SS}}}}\times {{{\rm{loge}}}}}{\frac{{kT}}{q}}-1\right]\frac{{C}_{{{{\rm{i}}}}}}{q}\end{array}$$
(1)

where k, T, and q are Boltzmann’s constant, absolute temperature, and elementary charge, respectively. The PEASCN-FAHCOO-based FET exhibits an Ntrap of 1.57 × 1012 cm−2 eV−1, more than two-fold lower than that of the PEAI-based FET (3.70 × 1012 cm−2 eV−1, corresponding to an SS of 1.1 V dec−1). These results collectively suggest that the PEASCN–FAHCOO strategy effectively reduces interfacial trap density and enhances device performance. To further elucidate the intrinsic mechanisms behind the observed device performance enhancement, we investigated the energy band structure and Fermi level position of the perovskite films, which offer direct insight into their doping characteristics and charge transport behavior. Based on ultraviolet photoelectron spectroscopy (UPS) and ultraviolet-visible absorption spectroscopy (UV–Vis) measurements (Figs. S24–S26), the optimized PEASCN–FAHCOO-based perovskite film exhibits a conduction band minimum at −3.47 eV, a valence band maximum at −4.87 eV, and a Fermi level at −4.49 eV (relative to vacuum). These values indicate that its Fermi level lies 0.38 eV above the valence band and 1.02 eV below the conduction band, consistent with stable p-type conductivity. In comparison, the reference PEAI-based film has a Fermi level 0.43 eV above the valence band and 0.98 eV below the conduction band. The further downward shift of the Fermi level in the optimized film suggests enhanced p-type doping, likely associated with the incorporation of SCN− into the perovskite lattice. Density of states (DOS) calculations based on first-principles simulations confirm that the introduction of SCN− anions into the perovskite thin film shifts the DOS distribution closer to the valence band, thereby resulting in stronger p-type electronic characteristics (Supplementary Fig. 27). Additionally, the structural optimization influences the interfacial energy alignment between the perovskite and the Au contact. The work function of the optimized film increases from 4.36 to 4.49 eV, reducing the hole injection barrier at the Au/perovskite interface from 0.74 to 0.61 eV. This increased work function and reduced injection barrier facilitate more efficient hole injection and transport across the interface, contributing to the lower VTH, higher on-state current, and improved subthreshold swing observed in our FETs.

To assess the repeatability of different devices, we fabricated 200 PEAI-based and PEASCN–FAHCOO-based FETs, respectively, and recorded their performance metrics (Supplementary Fig. 28). The results are summarized in Fig. 3i, indicating consistent repeatability for both materials. Notably, PEASCN–FAHCOO-based FETs exhibit an average mobility of 38 cm2 V−1 s−1, suggesting that the fabrication process outlined in this work can consistently produce high-quality tin-based perovskites. PEASCN–FAHCOO-based FETs generally demonstrate observable improvements across all performance metrics relative to PEAI-based FETs. To provide a comprehensive comparison, we compiled the electrical properties of representative tin-based perovskite transistors in Supplementary Table 1. Notably, in addition to the high on/off ratio, the reported mobility in this work ranks among the higher values reported for 2D and FA-based quasi-2D perovskite FETs.

Operational stability in high-performance transistors

We assessed the devices’ positional stability and operational reliability, which are critical factors for their practical utility. The unencapsulated test devices were stored in a nitrogen glovebox at around 20 °C. As depicted in Fig. 4a–c and Supplementary Fig. 29, the electrical performance of the PEAI-based FET significantly deteriorates with prolonged storage time. After 7 days of storage, the PEAI-based FET loses its gate control capability due to oxidation. In contrast, after 30 days of storage, the mobility of the PEASCN–FAHCOO-based FET only decreased from 42 to 34 cm2 V−1 s−1, retaining 82% of its initial value, indicating sustained performance. The operational stability exhibits a similar trend. After conducting a hundred consecutive transfer curve tests (Fig. 4d, e and Supplementary Fig. 30), the threshold voltage of the PEASCN–FAHCOO-based FET only shifted by 0.2 V, much less than the 3.9 V shift observed in the PEAI-based FETs. Furthermore, all PEASCN–FAHCOO-based FETs with different channel lengths (65–350 μm) show minimal performance degradation over 100 cycles and constant bias, with ΔVTH < 1 V and stable on-current (Supplementary Fig. 31), indicating consistent operational behavior regardless of channel length. To systematically evaluate the hysteresis behavior and its underlying mechanism, we performed both scan rate-dependent and temperature-dependent measurements. As shown in Fig. 4f and Supplementary Fig. 32, the PEAI-based FETs exhibit pronounced hysteresis in their transfer characteristics, with the hysteresis window significantly widening at lower scan rates. This behavior indicates the presence of a high density of interfacial trap states, where slower gate voltage sweeps allow more sufficient charge trapping and detrapping processes, thereby enhancing hysteresis. In contrast, the PEASCN–FAHCOO-based FETs show nearly invariant hysteresis windows across all scan rates, suggesting a substantially reduced density of interfacial traps and thus more stable and reproducible charge transport behavior29,30,31. Temperature-dependent measurements further elucidate the carrier transport mechanism. Extracting μh and hysteresis from Supplementary Fig. 33a, b, both PEAI-based and PEASCN–FAHCOO-based FETs exhibit a positive correlation between μh and temperature (Supplementary Fig. 33c), indicating that charge transport follows a thermally activated mechanism adhering to the Arrhenius relation, which can be expressed as:

$$\begin{array}{c}\mu={\mu }_{0}\exp \left(\frac{-{E}_{{{{\rm{A}}}}}}{{k}_{{{{\rm{B}}}}}T}\right)\end{array}$$
(2)

where EA is the activation energy for charge transport and kB is Boltzmann’s constant, μ0 is the pre-exponential factor6. Moreover, PEASCN–FAHCOO-based FETs show negligible hysteresis (ΔV < 0.5 V) across the temperature range, with hysteresis further decreasing at lower temperatures (Supplementary Fig. 33d). This contrasts with PEAI-based FETs, which display significant temperature-dependent hysteresis (ΔV = 4.1 V at room temperature) due to thermally activated ion migration (e.g., from VI defects). The suppressed hysteresis in PEASCN–FAHCOO-based FETs arises from enhanced crystallinity and effective defect passivation, which collectively block ion migration pathways. To further assess their operational robustness, we next examined the electrical reliability of the devices under repeated switching and bias stress conditions. After 5000 consecutive switching cycles, the on/off ratio of PEAI-based FETs degraded to 85.7% of its initial value, whereas the PEASCN–FAHCOO-based FETs retained 98.3% (Fig. 4g, h). Furthermore, under more rigorous constant bias stress, the on-state current of PEASCN-FAHCOO-based remained nearly unchanged (Fig. 4i), indicating the improved stability of PEASCN–FAHCOO-based FETs. These results indicate that PEASCN–FAHCOO-based films exhibit high material quality, where controlled crystallization and the use of hydrophobic spacers reduce defects, leading to suppressed charge trapping and improved repeatability.

Fig. 4: Stability characterization of PEAI-based and PEASCN–FAHCOO-based FETs.
Fig. 4: Stability characterization of PEAI-based and PEASCN–FAHCOO-based FETs.
Full size image

Storage stability of (a) PEASCN–FAHCOO-based and (b) PEAI-based FETs without encapsulation. c Changes in on-state current and saturation mobility of PEASCN–FAHCOO-based FETs over storage time. d Consecutive transfer curve measurements for 100 cycles (VDS = −30 V) of the PEASCN–FAHCOO-based FET. e Threshold voltage (Vth) variation in the two types of FETs during cyclic measurements. f Transfer curves measured at different scan steps. g Continuous on/off switching test with 5000 cycles (VGS = ± 30 V) of PEAI-based and PEASCN–FAHCOO-based FETs. h Partial magnification of (g). i Negative bias stress stability of the two types of FETs (VGS = VDS = −30 V).

Discussion

In summary, the introduction of PEASCN promotes the formation of high-quality PEA2FAn-1SnnI3n-1SCN2 (n = 2) templates, while substituting FAI slows down the uncontrolled growth of 3D FASnI3 structures. These effects result in perovskite films with reduced defects and improved vertical orientation, facilitating efficient charge transport. The resulting FETs exhibit high mobility (43 cm2 V−1 s−1), a high on/off ratio (>108), and negligible hysteresis. Furthermore, the devices maintain their electrical performance after 30 days of storage in a nitrogen environment, demonstrating good operational and environmental stability. These findings highlight the effectiveness of our crystallization control strategy in enabling the fabrication of high-performance and stable quasi-2D tin-based perovskite FETs, providing a reliable foundation for their future applications in electronic devices.

Methods

Materials

N, N-dimethylformamide (DMF, anhydrous, 99.8%), dimethyl sulfoxide (DMSO, anhydrous, 99.9%), chlorobenzene (CB, anhydrous, 99.8%), tin fluoride (SnF2, 99%) were obtained from Aladdin and used as received with no further purification. 2-phenylethylamine thiocyanate (PEASCN, 99.9%) was purchased from Greatcell.

Preparation of the SnI2 solution

To prepare a 0.2 M SnI2 solution, 2.5 mmol I2 was dissolved in 2.5 ml DMSO and 10 ml DMF successively, and then sufficient Sn powder was added to the I2 solution after stirring at room temperature overnight.

Preparation of the perovskite precursor solution

For the PEAI-based precursor, a 0.2 M solution was prepared by dissolving PEAI, FAI, SnI2, and SnF2 in a molar ratio of 0.34:0.83:1:0.1. This mixture was stirred at room temperature overnight. Similarly, for the PEASCN-FAHCOO-based precursor, a 0.2 M solution was prepared with PEASCN, FAHCOO, NH4I, SnI2, and SnF2 in the molar ratio of 0.34:0.83:0.83:1:0.1 and stirred overnight at room temperature. The SnI2 used herein was in situ synthesized in a glovebox via the reaction between I2 and Sn powder. This in-situ synthesis method not only features lower cost but also avoids the risk of oxidation during transportation. A performance comparison with commercial SnI2 is presented in Supplementary Fig. 34.

Device fabrication

The substrates used were highly doped P-type silicon with a 100 nm SiO2 layer. To enhance surface wettability, the Si/SiO2 substrates were treated with argon plasma for 20 s. The perovskite precursor solution was spin-coated onto the substrate at 1000 rpm for 5 s and 5000 rpm for 60 s. Following this, 100 μL of chlorobenzene was dropped onto the surface as an anti-solvent during the second spin-coating step, after which the films were thermally annealed at 100 °C for 10 min. Antisolvent dripping time during spin-coating was optimized by testing chlorobenzene addition at 5, 15, 25, and 35 s in the second step. Dripping at 15 s achieved optimal performance (high mobility, minimal hysteresis, good stability) due to balanced supersaturation promoting uniform nucleation. Early (5 s) or late (25 s, 35 s) dripping degraded device behavior, with issues like poor switching, hysteresis, or threshold voltage shifts indicating compromised crystallization. Gold electrodes (50 nm) were thermally evaporated to form the source and drain contacts. The channel length and width of the FETs were set to 150 μm and 1200 μm, respectively. All device fabrication steps were carried out in a nitrogen-filled glove box. The optimized perovskite film thickness, measured by a surface step profiler (Zeptools JS100A), was approximately 40 nm (Supplementary Fig. 35).

Perovskite film characterization and device measurement

GIWAXS measurement was performed by employing a beam energy of 10 keV and a PILATUS detector at the BL02U1 beamline of Shanghai Synchrotron Radiation Facility (SDRS), Shanghai, China. The SEM images were acquired with a JEOL JSM-7800 instrument. XRD patterns were collected using a Bruker D8 Advance diffractometer with Cu Kα radiation (λ = 1.5406 Å). AFM measurements were obtained with a Bruker Dimension Edge03040155. Time-resolved photoluminescence (TRPL) spectra were measured using a streak camera (Hamamatsu, C6860). PL spectra were recorded with a HORIBA Fluorolog-3 spectrometer equipped with a CCD detector. The electrical and noise measurements of the perovskite FET devices were carried out using the FS-Pro semiconductor measurement system inside a nitrogen-filled glove box.