High defect tolerance β-CsSnI₃ perovskite light-emitting diodes

Abstract

All-inorganic lead-free CsSnI₃ has shown promising potential in optoelectronic applications, particularly in near-infrared perovskite light-emitting diodes (Pero-LEDs). However, non-radiative recombination induced by defects hinders the optoelectronic properties of CsSnI₃-based Pero-LEDs, limiting their potential applications. Here, we uncovered that β-CsSnI₃ exhibits higher defect tolerance compared to orthorhombic γ-CsSnI₃, offering a potential for enhancing the emission efficiency. We further reported on the deposition and stabilization of highly crystalline β-CsSnI₃ films with the assistance of cesium formate to suppress electron–phonon scattering and reduce nonradiative recombination. This leads to an enhanced photoluminescence quantum yield up to ∼10%. As a result, near-infrared LEDs based on β-CsSnI₃ emitters are achieved with a peak external quantum efficiency of 1.81% and excellent stability under a high current injection of 1.0 A cm⁻².

---

New concepts

We report for the first time utilizing high defect tolerance tetragonal β-CsSnI₃ films for efficient and stable lead-free near-infrared (NIR) perovskite light-emitting diodes (Pero-LEDs). CsSnI₃ has emerged as a promising candidate for NIR Pero-LEDs due to its unique electronic configuration, optical property, and excellent intrinsic inorganic stability. However, the defect-induced nonradiative recombination hampers the optoelectronic properties of CsSnI₃-based Pero-LEDs. Herein, we propose a crystal phase engineering strategy to induce the formation of high defect-tolerance β-CsSnI₃ by introducing cesium formate into the precursor. The introduction of a small amount of CsFa plays a key role in suppressing the tilting of the [SnI₆]⁴⁻ octahedron, thus preventing the transformation of β-CsSnI₃ into γ-CsSnI₃ at room temperature. In the as-prepared β-CsSnI₃ films, electron–phonon scattering is effectively suppressed and non-radiative recombination is reduced, which increases the photoluminescence quantum yield. The highly defect-tolerant β-CsSnI₃ thin films introduced in this work offer new prospects for high-performance NIR LEDs and other optoelectronic devices.

Introduction

Near-infrared (NIR) light sources have a broad array of applications, notably in biomedical analysis where emission at 950 nm is preferred due to the penetration capability of infrared (IR) light through human tissues for imaging with high resolution.^1–3 However, there is a limited material selection for efficient NIR light emitting diodes (NIR-LEDs) with emission peaks longer than 800 nm. Organic semiconductors (emission wavelength ∼900 nm) and quantum dots (including heavy metal-based quantum dots and carbon dots, emission wavelength ∼850 nm) have been commonly used in NIR LEDs.^4–6 Recently, Pb-based perovskites have attracted huge attention in the LED community due to their unique optoelectronic properties, including ambipolar charge transport, long carrier diffusion length and tunable bandgap.^7,8 The state-of-art external quantum efficiency (EQE) of green and red perovskite light emitting diodes (Pero-LEDs) has rapidly surpassed 20%,^9–11 showing promising applications in lighting and colorful displays. In addition, white light-emitting diodes based on lead-free double perovskites have achieved a record luminance of 2793 cd m⁻² and an EQE of 0.76%.¹² Nevertheless, most studies on pure Pb-based Pero-LEDs have focused on emission wavelengths below 800 nm, due to their band gaps larger than 1.4 eV.¹³ In contrast, Sn-based perovskites exhibit an p-type direct semiconducting behavior with a narrow bandgap of 1.3 eV and strong near-infrared luminescence at 950 nm at room temperature causes the high energy level of the Sn 5s orbital to push up the valence band.^14–16

Recently, CsSnI₃ has shown promising potential in NIR Pero-LEDs due to its high photoluminescence quantum yield (PLQY), optical gain, and carrier mobility as high as ∼585 cm² V⁻¹ s⁻¹.^17–19 Despite its excellent optoelectronic properties, the applications of CsSnI₃ in Pero-LEDs still face great challenges, such as uncontrollable crystallization and high-dense defects. Various strategies have been pursued to modulate the crystallization process and suppress the defect-induced non-radiative recombination, including solvent engineering, reductant addition, and molecular passivation.^19–21 However, there is a need to explore approaches for further enhancing the light-emission properties of CsSnI₃.

In a typical CsSnI₃ perovskite structure, the off-center Cs atoms and the distortion of the [SnI₆]⁴⁻ octahedra cuase four different temperature-dependent polymorphs.¹⁵ Cubic α-CsSnI₃ is found to be stable above 500 K.²² As the temperature decreases to approximately 431.5 K, black tetragonal β-CsSnI₃ forms, which further transitions into the stable black orthorhombic γ-CsSnI₃ as the temperature reaches about 352 K.^22,23 Therefore, CsSnI₃-based optoelectronic devices are mostly based on γ-CsSnI₃ due to its preferred thermodynamic stability. However, the low crystal structure symmetry of γ-CsSnI₃ plays a negative effect on the light-emitting efficiency.

Herein, we propose a crystal-phase method for preparation of tetragonal β-CsSnI₃ perovskite films for high-efficiency and stable NIR PeLEDs. The theoretical calculations with density functional theory (DFT) indicate that β-CsSnI₃ has higher symmetry and defect tolerance than orthorhombic phase γ-CsSnI₃, which allows efficient light-emission. We fabricated β-CsSnI₃ perovskite films by incorporating a small amount of cesium formate (CsFa) in the precursor via a facile one-step deposition method. The β-CsSnI₃ exhibits an increased radiative recombination and a decreased electron–phonon scatterings than γ-CsSnI₃, leading to an enhanced PLQY from 4.0% to 10.0%. The corresponding NIR Pero-LEDs achieved an EQE of 1.81% and demonstrated outstanding operational stability with a T₅₀ (i.e., the time that a device lose 50% luminance) of 6 min under a high injection current density of 1000 mA cm⁻².

Results and discussion

To evaluate the potential of β-CsSnI₃ for light-emitting application, we first performed ab initio DFT calculations to investigate the crystal structures and band structures of CsSnI₃. Fig. 1a illustrates the crystal structures of all three CsSnI₃ perovskite phases and their relevance to defect tolerance. Compared to the ideal α-CsSnI₃ within a cubic Pm3̄m space group, the neighboring [SnI₆]⁴⁻ octahedra in the β-CsSnI₃ are tilted in the ab plane. However, for the case of γ-CsSnI₃, the tilting of the octahedra occurs both along the apical and equatorial directions. Consequently, the symmetry decreases successively from cubic to tetragonal and finally orthorhombic.^15,24 A high crystal symmetry is essential for maintaining the defect tolerance of the perovskite.^24–26 The slight octahedral tilting in the lattice of β-CsSnI₃ with a larger Sn–I–Sn bond angle of 167.9°compared to those in γ-CsSnI₃ (157.6° and 166.0°), leads to the strong antibonding state due to significant Sn–I overlap.²⁷ This attribute is crucial for maintaining defect tolerance. The latter is a key factor contributing to the performance of Pero-LEDs.^25,27 The PBE band structures and dipole transition matrix elements of three CsSnI₃ perovskite phases are shown in Fig. 1b–d. As shown in Fig. 1b, α-CsSnI₃ shows characteristics of a direct bandgap semiconductor, featuring the valence band maximum (VBM) and conduction band minimum (CBM) located at the high symmetry point R within the Brillouin zone. The crystal symmetry of β-CsSnI₃ is reduced, resulting in spin–orbit and crystal field splitting. Meanwhile, the R point in α-CsSnI₃ becomes the Z point in β-CsSnI₃ by rotation and folding (Fig. 1c). When the symmetry structure is further destroyed to form γ-CsSnI₃, the band can be folded along Z near the midpoint H (Fig. 1d). Despite of these alterations in the band structure, β-CsSnI₃ and γ -CsSnI₃ still maintain the direct bandgap and the narrow bandgap properties. The sum of the squares of the transition dipole moment elements (D²) at various k points reveals the high optical transition probabilities between the valence and conduction bands.²⁸ The D² values of different structures were compared in arbitrary units. The square magnitude of transition dipole moment elements indicates the coupling strength between Sn 5s and I 5p orbitals, which is influenced by the transient charge distribution (electrons and holes) within the perovskite structure. This serves as a figure-of-merit for the overlap of electron–hole wave functions.^28,29 As shown in Fig. 1b–d, all the perovskite phases exhibit no parity-forbidden transitions, in which α-CsSnI₃ and β-CsSnI₃ with higher symmetry has larger dipole transition matrix elements compared to γ-CsSnI₃ around their high symmetry points. This tunability of the electronic transition and the higher coupling strength make α-CsSnI₃ and β-CsSnI₃ more promising for efficient light emission than orthorhombic γ-CsSnI₃. Theoretically, α-CsSnI₃ and β-CsSnI₃ have higher symmetry and defect tolerance than orthorhombic γ-CsSnI₃, and thus have the potential for efficient light-emission. Unfortunately, α-CsSnI₃ is unstable below 500 K and exists only as an intermediate state.²² Therefore, it is a great chance and challenge to experimentally deposit β-CsSnI₃ films for efficient and stable near-infrared Pero-LEDs.

Fig. 1 DFT calculations. (a) Schematic illustration of defect tolerance in relation to the crystal structure for three crystal phases of the CsSnI₃ perovskite. PBE calculated band structures and corresponding transition matrix elements for (b) the cubic, (c) the tetragonal, and (d) the orthorhombic phase of CsSnI₃ perovskites.

A series of precursor solutions were prepared by dissolving equimolar amounts of CsI and SnI₂ in the precursor solution, with varying molar ratios of CsFa (0 to 8 mol%) added. The CsSnI₃ films were prepared using a single-step spin-coating method, followed by annealing at 110 °C for 10 min. X-ray diffraction (XRD) measurements were conducted on the scratched powder of as-deposited CsSnI₃ films to assess the impact of CsFa on their lattice structures. Strong XRD peaks at 14.45° and 29.11° are observed for the pristine CsSnI₃ (Fig. 2a), which are assigned to the (101) and (202) reflections of orthorhombic γ-CsSnI₃ (ICSD#262926). Strong XRD peaks at 14.25° and 28.94° are observed for the CsSnI₃ with the 4 mol% CsFa (Fig. 2b), which can be indexed to the (110) and (220) reflections of tetragonal β-CsSnI₃ (ICSD#262925). These XRD peaks are observed only above 230 °C.²³ High-resolution transmission electron microscopy (HR-TEM) characterization was further performed to clarify the formation of the β-CsSnI₃ phase. Fig. S1 (ESI†) shows the results. For the pristine CsSnI₃, the observed fringe spacing of 3.34 Å corresponds to the (202) plane of γ-CsSnI₃ (Fig. S1a and c, ESI†). For the CsSnI₃ treated with CsFa, 2-dimensional lattice fringes with inter-plane spacings of 3.42 Å and 6.12 Å indicated in the magnified HRTEM image (Fig. S1b and d, ESI†) are attributed to the (220) and the (002) planes of the β-CsSnI₃ phase, respectively. The zone axis is determined to be along the [101̄] direction and the angle between the two planes is 89.5° based on the indexed fast Fourier transform (FFT) pattern. The lattice spacing of these two planes and the angle between them are consistent with theoretical values (Table S1, ESI†), clearly indicating the existence of β-CsSnI₃ in the sample treated with CsFa.¹⁵ The film surface became rough when the concentration of CsFa exceeded 8 mol%. Therefore, we chose 4 mol% as the optimized concentration.

Fig. 2 Crystal structure evolution. XRD plots of the scratched powder from (a) the pristine CsSnI₃ and (b) the CsFa-treated CsSnI₃ films. The standard β and γ-CsSnI₃ patterns were calculated for Cu Kα1 radiation determined by Chung et al.¹⁵ The temperature-dependent XRD evolution for (c) the pristine CsSnI₃ and (d) the CsFa-treated CsSnI₃ films.

In situ temperature-dependent XRD measurements were performed to further explore the crystal structure evolutions of CsSnI₃. In Fig. 2c and d, both CsSnI₃ films exhibit similar XRD peaks at 29.15° at 300 °C, which are assigned to the (200) reflections of cubic α-CsSnI₃. As temperature decreases, additional peaks appear at positions between integral values of Miller indices of α-CsSnI₃, indicating a decrease in the crystal symmetry due to the tilting of the [SnI₆]⁴⁻ octahedra comprising the perovskite lattice.³⁰ In the pristine CsSnI₃ (Fig. 2c), as the temperature decreases, the peak of (200) reflection splits from 29.13° to 29.44°, indicating the phase transition from a cubic to an orthorhombic structure, while in the CsFa-treated CsSnI₃ (Fig. 2d), the peak of (200) only turns up from 29.15° to 29.28°, aligning with expectations for the tetragonal β-CsSnI₃ phase. The pattern distinctly contrasted with those observed in α- and γ-CsSnI₃ systems (Fig. S2, ESI†). These observations suggest that during the cooling process, the pristine CsSnI₃ undergoes a phase transition from the cubic phase to its tetragonal isomer. In contrast, introducing CsFa halts the phase transition, resulting in a final tetragonal phase. These results suggested that introducing CsFa can significantly affect the crystal phase of the final CsSnI₃ perovskite.

To probe the impact of CsFa on the formation of crystal structures, we focused on the precursor solutions due to their crucial role in crystallization. We utilized UV-vis spectroscopy (Fig. S3, ESI†) to observe each precursor state of CsSnI₃ in the solution phase. In the pure CsSnI₃ precursor solution, we noted an aggregated feature around 450 nm presumably due to the crystalline solvate interaction. This indicates that the CsSnI₃ precursor exhibits robust chemical interactions with the DMF/DMSO mixed solvent, leading to the formation of a solvate complex.³¹ After incorporating the CsFa, the absorption edge blue-shifted to a smaller edge wavelength of 400 nm, attributing to a lower degree of crystalline solvate interaction in the precursor solution. This may come from the strong coordination interactions between the Fa⁻ and the precursors as evidenced by a strong upward chemical shift (Δδ ≈ 2.3 ppm) of the –C=O functional group in the ¹³C NMR spectra (Fig. S4, ESI†). We suggest that this strong coordination interactions between the Fa⁻ and the precursors could transform the phase transition path of CsSnI₃, which can increase the formation energy of γ-CsSnI₃. This is evidenced by the slowdown of the black film forming process (Fig. S5, ESI†). With the increased formation energy of γ-CsSnI₃, it would tend to form β-CsSnI₃ during annealing. In addition, slower crystallization is beneficial to reduce the lattice distortion in the perovskite crystal, thus avoiding the rattling of Cs⁺ and the [SnI₆]⁴⁻ octahedral tilting during cooling.^32,33 This can effectively prevent the collapse of the highly symmetric β-CsSnI₃ into γ-CsSnI₃ at room temperature. Fourier transform infrared (FTIR) analysis was further applied to rule out the incorporation of any organic X-site anions (Fa⁻) (Fig. S6, ESI†). As there is no stretching vibration peak of the C=O bond in the final film even with a large incorporated ratio of 10 mol% CsFa, we could conclude that the Fa⁻ functional group acts as a volatile component for the controllable crystal phase of the CsSnI₃ perovskite.

Scanning electron microscopy (SEM) characterization shows that the CsFa-treated CsSnI₃ exhibits a textured and discontinuous morphology with distinct edges and corners (Fig. S7, ESI†). The average crystal size of the CsSnI₃ treated with CsFa is 170 nm, being smaller than that of the pristine CsSnI₃ (250 nm) (Fig. S8, ESI†). Such a discontinuous perovskite film with a smaller crystal size possesses a larger specific surface area, which is helpful for electron injection and reducing nondirective diffusion of excess holes.^1,7 In addition, the proportion of Sn⁴⁺ components in the films treated with CsFa (8.5%) is significantly lower than that of the pristine CsSnI₃ (18.7%), indicating the improved antioxidant properties (Fig. S9, ESI†). Meanwhile, β-CsSnI₃ maintains the unique narrow bandgap properties, with an absorbance edge at ca. 950 nm (Fig. S10a, ESI†). This is consistent with the small band gap difference (0.02 eV) between the β and γ phases as demonstrated by the theoretical calculations.^34,35 The photoluminescence (PL) spectra (Fig. S10b, ESI†) show that the perovskite film with CsFa exhibited considerably enhanced spectra intensity and narrowed full width at half maximum (FWHM) values (67.7 and 65.7 nm for the pristine and CsFa-treated CsSnI₃, respectively), indicative of β-CsSnI₃ with reduced non-radiative recombination sites in the film. In Fig. S11a (ESI†), time-resolved PL measurements show that the β-CsSnI₃ film has longer PL lifetimes (∼4.5 ns) than γ-CsSnI₃ films (∼1.2 ns). The excitation-intensity-dependent PLQY measurement shows a high PLQY of up to ∼10% for the β-CsSnI₃, which is about ∼3-fold higher than the γ-CsSnI₃ (Fig. S11b, ESI†), indicating that β-CsSnI₃ has more efficient radiative recombination process than γ-CsSnI₃. Despite being reported that Sn²⁺ oxidation¹⁷ and morphology³⁴ can affect the luminescence ability of Sn-based perovskites, we suggest that the enhanced defect tolerance of β-CsSnI₃ is the main factor for improving the PLQY.

The carrier recombination process can be extracted using the power dependence of the steady-state integrated PL intensity I_PL and the exciting laser radiation power density P_ex (excitation density) as I_PL ∝ P_ex^k. The normalized PL spectra at 300 K and 50 K under varying excitation densities of 532 nm laser are presented in Fig. 3a and b for γ-CsSnI₃ and Fig. 3d and e for β-CsSnI₃ films, respectively. The corresponding PL intensity graphs at 300 K and 50 K are given in Fig. 3c and f. In general, the PL intensity dependence with k≈1 was observed for both perovskite films, indicating the characteristics of the monomolecular recombination process. The recombination of photo-generated electrons with the background hole density resulting from unintentional doping in CsSnI₃ can explain the linear dependence on the excitation intensity of the PL. This mechanism is referred to as free-to-bound radiative recombination of degenerate holes from the valence band with non-equilibrium electrons in the conduction band tails.³⁶ For the pristine CsSnI₃ film, the PL intensity at room temperature exhibits a linear increase with excitation density up to approximately ∼10¹⁸ cm⁻³. This indicates that the quantum yield remains constant within this range. However, as the excitation density surpasses∼10¹⁸ cm⁻³, the power-dependent PL intensity exhibits a sublinear slope (saturation effect) instead of the expected superliner dependence on the injected carrier density, as typically observed in Pb-based perovskites. This phenomenon indicates the presence of significant auger recombination energy loss.³⁷ The CsFa-treated β-CsSnI₃ film, on the other hand, exhibits a PL intensity slope that is closer to 1 under both low fluency and high excitation density, indicating the suppressed non-radiative recombination in the films.³⁸

Fig. 3 Photoluminescence spectroscopy characterization. 2D color plot of power-dependent photoluminescence spectra of the γ-CsSnI₃ film at (a) 50 K, (b) 300 K, and the corresponding (c) logarithmic plot of the integrated PL intensity as a function of carrier density. The 2D color plot of power-dependent photoluminescence spectra of the β-CsSnI₃ film at (d) 50 K, (e) 300 K, and the corresponding (f) logarithmic plot of the integrated PL intensity as a function of carrier density.

The non-radiative decay rate of excited states is associated with the electron–phonon interaction in perovskites according to the Huang–Rhys parameter.^39,40 We further investigated the electron–phonon interaction via temperature-dependent PL to explain the improvement in light emission ability for the high-symmetry phase β-CsSnI₃. Fig. 4a and b depict the decrease in PL peak intensity and widening of the spectral linewidth as the temperature increases from 50 to 300 K. This phenomenon can be attributed to the fact that the hot carriers are not only thermalized to the band-edge state but also trapped in deep defect states. The pristine CsSnI₃ demonstrates significant thermal quenching even at lower temperatures, confirming that β-CsSnI₃ has less defect density than γ-CsSnI₃, therefore leading to a high PLQY in β-CsSnI₃. The temperature-dependent variation of the FWHM is depicted in Fig. 4c and is fitted using the independent Boson model. In this model, Γ₀ is the inhomogeneous broadening coefficient, Γ_LO represents the electron-LO phonon coupling coefficient, and E_LO denotes the LO phonon energy.⁴¹ It is extracted that Γ₀ = 80.9 ± 0.2 meV, Γ_LO= 380.7 ± 66.3 meV, and E_LO = 92.2 ± 5.5 meV for pristine CsSnI₃. Accordingly, Γ₀ = 79.5 ± 0.4 meV, Γ_LO = 167.1 ± 46.9 meV, and E_LO = 64.3 ± 7.6 meV were fitted for CsFa-treated CsSnI₃. Compared to pristine CsSnI₃, β-CsSnI₃ shows a notable reduction in electron-LO phonon coupling coefficient Γ_LO, indicating that the vibration of the [SnI₆]⁴⁻ octahedra cage involves less energies. The decreased LO phonon energy E_LO of β-CsSnI₃ suggests that the possibility of placing the non-luminescing energy state above the lowest radiative excitonic state is lower compared to γ-CsSnI₃. Consequently, the decreased LO phonon energy confirms that the high symmetry phase CsSnI₃ films formed by CsFa treatment can increase the number of excitation-generated electron–hole pairs in luminescing states and mitigate quenching through electron–phonon scatterings. We suggest that the higher symmetry and defect tolerance of the β-CsSnI₃ phase are the main reasons for the increased PLQY.

Fig. 4 Electron–phonon interaction. 2D color maps of temperature-dependent photoluminescence spectra of (a) the γ-CsSnI₃ and (b) the β-CsSnI₃ film. (c) The corresponding full width at half maximum (FWHM) evolution of PL spectra as a function of temperature.

The CsSnI₃ based Pero-LEDs were fabricated with a structure of ITO/poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate) (PEDOT:PSS)/CsSnI₃/4,6-bis(3,5-di(pyridin-3-yl)phenyl)-2-methylpyrimidine (B3PyMPM)/LiF/Al, where light emission occurs when injected electrons and holes encounter in the perovskite layer and recombine radiatively. The J–V curves and radiance–voltage (R–V) curves were measured and are depicted in Fig. 5a. The Pero-LEDs based on β-CsSnI₃ possessed higher radiance than the device based on γ-CsSnI₃ and reached a high radiance of 54 W sr⁻¹ m⁻² at 5.4 V. This indicates that the introduction of CsFa improves the radiative recombination process and crystallization quality of CsSnI₃. The peak positions of the PL spectra of both in Fig. S10b (ESI†) are both around 950 nm, while the corresponding peaks of the EL spectra are around 910 nm and 930 nm, respectively (Fig. S12, ESI†). Such a difference between the PL and EL spectra could be mainly ascribed to the tin-related defects involved in recombination processes. Usually, these defects can be found within the bandgap with a low energy level.²⁰ In the PL measurements which are carried out under illumination, some of the photogenerated carriers may not efficiently fill all the trap states, leading to defect-induced radiative recombination and consequently a red-shifting of the PL peak.⁴² In contrast, in the EL measurements which are carried out in the dark, the defect states may be readily filled by trapped carriers due to the high current injection, and thus a corresponding blue-shifted characteristic spectra peak.⁴³ In addition, current loading induces a change in the EL wavelength (called color-shift), which is mainly due to field-driven ionic migration, stemming from the inherent ionic nature of perovskite materials.²⁰

Fig. 5 Device performances. (a) J–V curves, radiance–V curves, and (b) EQE–J curves of PeLEDs based on γ-CsSnI₃ and β-CsSnI₃ and the device structure ITO/PEDOT:PSS/B3PyMPM/LiF/Al. Lifetime measurement result of the CsSnI₃ PeLEDs under constant injection of (c) 100 and (d) 1000 mA cm⁻².

Fig. 5b presents the current–EQE curves of the Pero-LEDs. The Pero-LEDs based on β-CsSnI₃ achieved the maximum EQE value of 1.81%, being twice higher than that of the device based on γ-CsSnI₃ (0.99%, Fig. S13, ESI†). We ascribed the EQE enhancement to β-CsSnI₃ for increasing radiative recombination and reducing electron–phonon scatterings. In addition, the β-CsSnI₃ based devices exhibited excellent EL stability even with increasing current injection, highlighting their immense potential for practical applications. We operated the Pero-LEDs at a high constant current density of 100 mA cm⁻² and noted the evolution of radiance (Fig. 5c). The operational lifetime (T₅₀) of the CsFa-treated device is 8.5 h, which is approximately ten times that of the pristine device. Surprisingly, the CsFa-treated device has excellent stability with a T₅₀ of 6 min under a very high injection rate of 1.0 A cm⁻² (Fig. 5d). To the best of our knowledge, this is the first report on β-CsSnI₃-based NIR PeLEDs (Table S2, ESI†). Our study provides a fundamental perspective to enhance the near-infrared emission of black-phase CsSnI₃ perovskites.

Conclusions

In summary, we propose crystalline phase engineering of the deposition and stabilization of tetragonal β-CsSnI₃ films for highly efficient and stable NIR Pero-LEDs. Introducing a small amount of CsFa could suppress the tilting of the [SnI₆]⁴⁻ octahedron during cooling, thus preventing the transformation of β-CsSnI₃ into γ-CsSnI₃ at room temperature. We revealed that β-CsSnI₃ has higher symmetry and Sn–I coupling strength than orthorhombic γ-CsSnI₃, which makes it have higher defect tolerance and potential for efficient light-emission. The β-CsSnI₃ exhibits increased radiative recombination and reduced electron–phonon scatterings compared to γ-CsSnI₃, leading to a notable enhancement in the PLQY of β-CsSnI₃ from 4% to 10%. Therefore, β-CsSnI₃-based NIR Pero-LEDs exhibit a high EQE of 1.81% and a T₅₀ lifetime of 8.5 h under the injection of 100 mA cm⁻². Our study highlights the impact of different crystal phases on the light emission properties of CsSnI₃ perovskites, offering a novel perspective for enhancing light emission. Additionally, the highly defect-tolerant β-CsSnI₃ thin films introduced in this work offer new prospects for high-performance photovoltaic and other optoelectronic devices.

Author contributions

Haixuan Yu: conceptualization and writing – original draft. Tao Zhang: investigation, methodology, formal analysis, and writing-original draft. Zhiguo Zhang: investigation, data curation, and writing – original draft. Zhirong Liu and Qiang Sun: investigation. Junyi Huang: investigation and writing – original draft. Letian Dai, Yan Shen, Xiongjie Li: writing – review & editing and funding acquisition. Mingkui Wang: supervision, writing – review & editing, and funding acquisition.

Data availability

The data supporting this article have been included as part of the ESI.†

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was financially supported by the National Key Research and Development Program of China (2022YFB4200305), the National Natural Science Foundation of China (22379049), the Department of Science and Technology of Hubei Province (2022EHB009), the China Postdoctoral Science Foundation (2022M711236), Innovation Project of Optics Valley Laboratory (OVL2023PY004), and Wuhan East Lak High-tech Development Zone (No. 2023KJB232). T. Zhang thanks the supports from the CSC scholarchip. The device fabrication and characterization were supported by Prof. Feng Gao and finished in his lab. at the Department of Physics, Chemistry and Biology (IFM), Linköping University, Linköping, Sweden.

References

1. F. Yuan, G. Folpini, T. Liu, U. Singh, A. Treglia, J. Lim, J. Klarbring, S. Simak, I. Abrikosov and T. Sum, Nat. Photonics, 2024, 18, 1–7.
2. D. Liu, P. Dang, G. Zhang, H. Lian, G. Li and J. Lin, InfoMat, 2024, e12542.
3. Y. Liu, F. Di Stasio, C. Bi, J. Zhang, Z. Xia, Z. Shi and L. Manna, Adv. Mater., 2024, 2312482.
4. J. Li, D. Cui, J. Huang, S. He, Z. Yang, Y. Zhang, Y. Luo and K. Pu, Angew. Chem., Int. Ed., 2019, 58, 12680–12687.
5. N. Davis, J. Allardice, J. Xiao, A. Karani, T. Jellicoe, A. Rao and N. Greenham, Mater. Horiz., 2019, 6, 137–143.
6. R. Chen, Z. Wang, T. Pang, Q. Teng, C. Li, N. Jiang, S. Zheng, R. Zhang, Y. Zheng, D. Chen and F. Yuan, Adv. Mater., 2023, 35, 2302275.
7. Q. Miao and K. Pu, Adv. Mater., 2018, 30, 1801778.
8. J. Luo, J. Li, L. Grater, R. Guo, A. Mohd Yusoff, E. Sargent and J. Tang, Nat. Rev. Mater., 2024, 1–13.
9. B. Zhao, M. Vasilopoulou, A. Fakharuddin, F. Gao, A. Mohd Yusoff, R. Friend and D. Di, Nat. Nanotechnol., 2023, 18, 981–992.
10. A. Vartanian, Nat. Rev. Mater., 2023, 8, 707.
11. Z. Yu, X. Shen, X. Fan, Y. Jung, W. Jeong, A. Dasgupta, M. Kober-Czerny, P. Caprioglio, S. Park and H. Choi, ACS Energy Lett., 2023, 8, 4296–4303.
12. S. Jin, H. Yuan, T. Pang, M. Zhang, J. Li, Y. Zheng, T. Wu, R. Zhang, Z. Wang and D. Chen, Adv. Mater., 2024, 36, 2308487.
13. X. Liu, W. Xu, S. Bai, Y. Jin, J. Wang, R. Friend and F. Gao, Nat. Mater., 2021, 20, 10–21.
14. G. Pacchioni, Nat. Rev. Mater., 2021, 6, 108.
15. I. Chung, J. Song, J. Im, J. Androulakis, C. Malliakas, H. Li, A. Freeman, J. Kenney and M. Kanatzidis, J. Am. Chem. Soc., 2012, 134, 8579–8587.
16. Z. Wang, S. Zheng, Q. Teng, C. Li, B. Zhuang, R. Zhang, F. Huang, D. Chen and F. Yuan, Innovations Mater., 2023, 1, 72–88.
17. X. Guan, J. Lu, Q. Wei, Y. Li, Y. Meng, K. Lin, Y. Zhao, W. Feng, K. Liu, G. Xing and Z. Wei, ACS Energy Lett., 2023, 8, 1597–1605.
18. Y. Sun, L. Ge, L. Dai, C. Cho, J. Ferrer Orri, K. Ji, S. Zelewski, Y. Liu, A. Mirabelli and Y. Zhang, Nature, 2023, 615, 830–835.
19. Y. Li, X. Guan, Y. Meng, J. Chen, J. Lin, X. Chen, C. Y. Liu, Y. Zhao, Q. Zhang and C. Tian, InfoMat, 2024, e12537.
20. H. Min, J. Chang, Y. Tong, J. Wang, F. Zhang, Z. Feng, X. Bi, N. Chen, Z. Kuang and S. Wang, Nat. Photonics, 2023, 17, 755–760.
21. A. Treglia, F. Ambrosio, S. Martani, G. Folpini, A. Barker, M. Albaqami, F. De Angelis, I. Poli and A. Petrozza, Mater. Horiz., 2022, 9, 1763–1773.
22. E. Da Silva, J. Skelton, S. Parker and A. Walsh, Phys. Rev. B: Condens. Matter Mater. Phys., 2015, 91, 144107.
23. Q. Sun, Z. Fang, Y. Zheng, Z. Yang, F. Hu, Y. Yang, W. Yang, X. Hou and M. Shang, J. Mater. Chem. A, 2022, 10, 3996–4005.
24. Y. Liu, W. Gao, C. Ran, H. Dong, N. Sun, X. Ran, Y. Xia, L. Song, Y. Chen and W. Huang, ChemSusChem, 2020, 13, 6477–6497.
25. Z. Xiao, Z. Song and Y. Yan, Adv. Mater., 2019, 31, 1803792.
26. M. de Jong, L. Seijo, A. Meijerink and F. Rabouw, Phys. Chem. Chem. Phys., 2015, 17, 16959–16969.
27. B. Zhao, S. Jin, S. Huang, N. Liu, J. Ma, D. Xue, Q. Han, J. Ding, Q. Ge and Y. Feng, J. Am. Chem. Soc., 2018, 140, 11716–11725.
28. W. Meng, X. Wang, Z. Xiao, J. Wang, D. Mitzi and Y. Yan, J. Phys. Chem. Lett., 2017, 8, 2999–3007.
29. H. Xie, S. Hao, J. Bao, T. Slade, G. Snyder, C. Wolverton and M. Kanatzidis, J. Am. Chem. Soc., 2020, 142, 9553–9563.
30. C. Yu, Y. Ren, Z. Chen and K. Shum, J. Appl. Phys., 2013, 114, 163505.
31. Y. Kim and D. Kang, J. Mater. Chem. A, 2022, 10, 4782–4790.
32. S. Xiang, Z. Fu, W. Li, Y. Wei, J. Liu, H. Liu, L. Zhu, R. Zhang and H. Chen, ACS Energy Lett., 2018, 3, 1824–1831.
33. Y. Wang, M. Dar, L. Ono, T. Zhang, M. Kan, Y. Li, L. Zhang, X. Wang, Y. Yang and X. Gao, Science, 2019, 365, 591–595.
34. J. Lu, X. Guan, Y. Li, K. Lin, W. Feng, Y. Zhao, C. Yan, M. Li, Y. Shen and X. Qin, Adv. Mater., 2021, 33, 2104414.
35. L. Huang and W. Lambrecht, Phys. Rev. B: Condens. Matter Mater. Phys., 2013, 88, 165203.
36. H. Shibata, M. Sakai, A. Yamada, K. Matsubara, K. Sakurai, H. Tampo, S. Ishizuka, K. Kim and S. Niki, Jpn. J. Appl. Phys., 2005, 44, 6113.
37. H. Fang, S. Adjokatse, S. Shao, J. Even and M. Loi, Nat. Commun., 2018, 9, 243.
38. Y. Wang, M. Zhi, Y. Chang, J. Zhang and Y. Chan, Nano Lett., 2018, 18, 4976–4984.
39. L. Zuo, S. B. Jo, Y. Li, Y. Meng, R. Stoddard, Y. Liu, F. Lin, X. Shi, F. Liu and H. Hillhouse, Nat. Nanotechnol., 2022, 17, 53–60.
40. W. Yin, T. Shi and Y. Yan, J. Phys. Chem. C, 2015, 119, 5253–5264.
41. T. Zhang, C. Zhou, X. Feng, N. Dong, H. Chen, X. Chen, L. Zhang, J. Lin and J. Wang, Nat. Commun., 2022, 13, 60.
42. D. Khadka, Y. Shirai, M. Yanagida, T. Tadano and K. Miyano, Chem. Mater., 2023, 35, 4250–4258.
43. M. Hajdel, M. Chlipała, M. Siekacz, H. Turski, P. Wolny, K. Nowakowski-Szkudlarek, A. Feduniewicz-Żmuda, C. Skierbiszewski and G. Muziol, Materials, 2021, 15, 237.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4mh00428k

‡ These authors contributed equally to this work.

---