Controlled phase distribution of quasi-2D perovskite enables improved electroluminescence

Quasi-two-dimensional (quasi-2D) perovskites are increasingly explored for integration into light-emitting diodes (LEDs) as light-emissive layers. However, the quasi-2D perovskite films likely exhibit non-uniform dimensional phase distribution and irregular internal crystal structures. These characteristics are known to contribute to undesirable effects, including non-radiative recombination losses and radiative recombination in perovskites of various dimensions, impeding the realization of efficient electroluminescence and high color purity in LEDs. In this study, we present an investigation on the correlation between the dimensional distribution of quasi-2D perovskites and charge carrier behavior by modulating anti-solvent dripping during the film fabrication processes. We provide a comprehensive analysis of the impact of controlled dimensional distribution on charge injection and recombination processes associated with the performance of quasi-2D perovskite LEDs. Our work emphasizes the crucial role played by controlled dimensionality in quasi-2D perovskites in realizing efficient and stable perovskite-based LEDs.


Introduction
Metal halide perovskites have attracted significant attention in the field of light-emitting diodes (LEDs) [1,2] due to their advantageous properties.In recent studies, bulky organic cations separating metal halide octahedra to form a 2D quantum well structure have been introduced to improve exciton binding energy in the materials and operational stability of the devices, which have been a main hindrance of prototypical three-dimensional (3D) ABX 3 -perovskite based LEDs [3,4].Since the 2D perovskite-forming bulky organic cation (L) participates in crystal formation, the structure of perovskite is expressed to L 2 A n-1 B n X 3n+1 (so called quasi-2D perovskite), where n is the number of metal halide layers between organic molecules.(e.g.n = 1: pure 2D perovskite, n = ∞: pure 3D perovskite).The presence of multiple 'n' phases within the perovskite layer plays a crucial role in the quantum confinement effect and carrier funneling processes [5,6].Thanks to the high exciton binding energy and carrier funneling processes, the external quantum efficiency (EQE) of perovskite LEDs (PeLEDs) has surpassed 20% [7][8][9].However, in solution processes, controlling the proportion and distribution of quasi-2D perovskites with different 'n-value' poses challenges, even when starting with precursor materials featuring a dimensional homogeneity [10,11].The quasi-2D perovskite films likely exhibit non-uniform dimensional phase distribution and irregular internal crystal structures.These characteristics are known to contribute to undesirable effects, including non-radiative recombination losses and radiative recombination in perovskites of various dimensions, impeding the realization of efficient electroluminescence and high color purity in LEDs.To address this issue, various strategies have been employed in perovskite solutions to regulate the phase distribution.For instance, Li et al achieved a high efficiency in quasi-2D perovskite LEDs by forming a homogeneous dimensional-phase distribution [12].They employed the molecule with carbon chains to reinforce their interaction with perovskite structure, leading the dominant n = 3 phase.This single n-phase dominated distribution reduces energy loss and enhances the overall emission performance.This result shows that the realization of high performance PeLEDs requires the precise control of the n-value and their distribution [13][14][15][16].
The fabrication process, including stages such as crystallization and post-treatment, significantly affects the formation and distribution of quasi-2D phases, which are crucial for carrier dynamics [17].Recent studies have attempted to manipulate phase distribution and orientation within the spin-coating system by adjusting precursor solutions [18], composition [19,20], additives [21,22], and antisolvents [23].For the fabrication of perovskite film, regardless of dimension, antisolvent engineering has been a common method to produce highly uniform and dense film with minimized non-radiative recombination [24,25].The antisolvent injection process impacts the crystallization dynamics and consequent phase distribution in the film, potentially influencing charge carrier behavior and device performance.However, the understanding and prediction of changes in phase distribution due to antisolvent addition and annealing processes is not yet clear.
This study investigates the mechanisms behind phase distribution influenced by antisolvent addition and establishes a link between phase distribution and carrier dynamics in LEDs.Phase distribution and charge carrier dynamics were investigated using transient absorption spectroscopy and time-of-flight secondary ion mass spectroscopy (ToF-SIMS), which correlated with device performance.Critical impacts of anti-solvent dripping were unraveled in terms of phase distribution and device performance.We believe that understanding this relationship is crucial for optimizing the fabrication process and improving the performance and stability of perovskite-based LEDs.

Perovskite precursor preparation
The perovskite precursor solution was prepared by dissolving the corresponding number of precursor chemicals (MABr, PEABr, FPEABr and PbBr 2 ) to achieve 0.16 M of (FPEA 0.5 PEA 0.5 ) 2 MA 2 Pb 3 Br 10 in the DMSO.The solution was stirred for 4 h.The solution was filtered by a 0.2 µm hydrophilic PTFE filter before use.

Perovskite film fabrication
The bare glass substrates were cleaned by ultrasonication in detergent, acetone, and ethanol baths for 10 min, respectively.Subsequently, these substrates were dried with N 2 blowing and treated with O 2 plasma for 10 min.The hole transporting layer was spin-coated on the cleaned glass substrates at 500 rpm 7 s and then at 4000 rpm 90 s.The coated substrates were annealed at 150 • C for 20 min.To coat the perovskite layer, the substrates were moved into the N 2 glovebox.The perovskite precursor was spin-coated on the GraHIL at 1000 rpm for 10 s and then at 5000 rpm for 60 s.During the spin coating of the perovskite precursor solution, 0.1 ml of chlorobenzene was dropped onto the spinning substrate.The as-spun films were annealed at 100 • C for 5 min.

Device preparation
Patterned indium-doped tin oxide (ITO) glass substrates were cleaned with detergent and with ultrasonication in deionized water, acetone, and ethanol for 10 min, respectively.These substrates were dried with N 2 blowing and treated with O 2 plasma for 10 min.The hole transporting layer and perovskite film were spin-coated on the cleaned ITO substrates as described above.Finally, 50 nm of 2,2 ′ ,2 ′′ -(1,3,5-Benzinetriyl)tris(1-phenyl-1-H-benzimidazole) (TPBi), 1 nm of lithium fluoride (LiF), and 100 nm of Al were sequentially evaporated in a vacuum chamber.All the devices were encapsulated with CaO getter, encapsulation glass, and epoxy resin under a nitrogen atmosphere before the measurement.

Steady-state photoluminescence measurement
Steady-state photoluminescence (PL) spectra were measured using an FLS1000 (Edinburgh Inst).The monochromatic excitation light with wavelength of 372 nm was generated from Xenon lamp and the 395 nm long-pass filter was placed at the detector.The scan range was 380-600 nm with 1 nm scan step and a dwell time of 1 s.

Absorbance
The absorbance of the perovskite films was measured using a UV-vis spectrometer (Lambda 45, PerkinElmer).

Time-of-fight secondary ion mass spectroscopy
Time-of-fight secondary ion mass spectroscopy (TOF-SIMS) was used for depth profiling of the perovskite.Depth profiling was completed with a 25 keV Bi1 (1 pA pulsed current) rostered over an area of 30 µm × 30 µm and a 1 keV cesium ion sputter beam (50 nA sputter current) rostered over an area of 100 µm × 100 µm.

Atomic force microscope
An atomic force microscope (AFM) was employed to reveal the surface morphology of perovskite films.The Park Systems NX10 with OMCL-AC160TS tip was used for the AFM measurements.AFM images were analyzed using the open-source software Gwyddion©.

Device performances
The current-voltage-luminance characteristics were obtained using a source-measurement unit (Keithley 2400), spectroradiometer (CS-2000, Konica Minolta).In addition, the stability of the PeLEDs was assessed by a CS-2000 spectroradiometer (Konica Minolta) with a Keithley 2450 sourcemeter in N 2 filled glovebox without encapsulation.

Femtosecond transient absorption spectroscopy
Femtosecond transient absorption spectroscopy (fs-TA), a femtosecond Ti: Sapphire amplifier (Legend Elite, Coherent, USA) was operated with a repetition rate of 1 kHz and a pulse duration of 30 fs.The fundamental laser pulses, centered at a wavelength of 790 nm, were divided into two parts.A stronger beam was guided into an optical parametric amplifier (TOPAS Prime, Light Conversion, Lithuania) and adjusted to 350 nm (±10 nm) to serve as the pump pulse, each delivering an energy of 10 nJ per pulse.The second beam was focused onto a nonlinear crystal to generate a supercontinuum laser in the UV-visible spectral range to utilize as the probe pulse.The probe beam was responsible for detecting the pump-induced differential absorption changes (∆A) as a function of the time delay.The time delays between the pump and probe pulses were controlled by a motorized linear stage in the probe beam path.Both the pump and probe beams were focused onto the same position in the sample and the TA data were collected by a TA spectrometer equipped with a high-resolution 1024 pixel CMOS sensor (HELIOS Fire, Ultrafast Systems, USA).The relative presence of each n-value (figure 3) was quantified by the amplitude of the transient absorption signal at t = 0.5 ps as below:

Results and discussion
To explore the mechanism behind the distribution of quasi-2D phases in antisolvent-assisted spin coating processes, we fabricated the quasi-2D perovskite films with different anti-solvent dripping conditions (figure S1(a)).Without antisolvent (no dripping), crystallization would primarily occur via the vaporization of the DMSO solvent during prolonged spin-coating.On the contrary, the introduction of antisolvent would lead to the formation of an intermediate phase during spinning, where the grains continue to grow during annealing as residual DMSO evaporates [26].We presumed the formation and distribution of the quasi-2D perovskite should be strongly dependent on the anti-solvent dripping condition during the spinning process.S1.Under the 'No dripping' condition, the EQE and current efficiency nearly approach zero, as shown in figures 1(b) and S2.However, upon introducing the antisolvent, there is a notable surge in these values.The EQE steadily ascends up to the 25 s device but shows a decline beyond that point.This pattern remains consistent across all parameters.Closer inspection of the current density-voltage curve revealed a difference in the leakage current between the devices (figure 1(a)).In the 'No dripping' device, a substantial leakage current arises from carriers seeping into the injection layer without recombining, whereas the leakage current is relatively low with the anti-solvent dripping.In addition, the 'No dripping' device exhibit turns on voltage at 3.2 V, whereas the other devices turn on at 3.0 V.The difference in leakage current and turn-on voltage implies a change in the energy landscape with the dripping conditions [27][28][29].Despite the reduced turn-on voltage of antisolvent-treated devices, the maximum current density and luminance strongly increase with lower operating voltage than devices without antisolvent treatment.This implies that electrons and holes are able to recombine effectively in antisolvent-treated devices owing to their well-aligned energy levels.The improved alignment of energy levels might foster effective transport of charge carriers and increased exciton recombination, which enables a larger current flow as well.As a result, the operational voltage has been decreased for a certain current or luminance, which results in the increase in device efficiency of the antisolvent-treated devices.The dramatic decrease in EQE at higher operating voltage is probably due to unbalanced carrier injection (mainly for 'no dripping' condition) and/or excessive radiative recombination (mainly for anti-solvent treated devices), which is able to degrade metal halide perovskite crystals.The voltage dependent electroluminescence study confirms that the PeLEDs operate with 1-2 nm shift of the electroluminescent peak position for all conditions (figure S3).The minor shift at higher operating voltages might be from carrier/phonon interaction due to Joule-heating [30].The 'No dripping' device exhibits relatively poor performance upon increasing voltage.While the highest electroluminescent intensity in this device reaches 4.8 V, it rapidly decreases.In contrast, the other devices achieve their peak intensity above 5.0 V.In addition, the 'No dripping' and antisolvent-treated condition ('25 s') show large differences in operation stability test (figure S10).The quasi-2d perovskite devices are operated at a constant voltage to compare the stability performance from matched initial luminance [31,32].Whereas the luminance of 'No dripping' device drops under 50% of initial luminance after 44 s, that of '25 s' keeps over 50% of initial luminance over 400 s.This improvement in the stability might be explained by the optimized of charge balance and reduced injection energy loss [33,34].
To elucidate the difference in current-voltage characteristics and PeLED performance, we first delved into evaluating the morphology of the perovskite film by AFM measurements in figure S4.While the roughness is slightly higher for the films with antisolvent dripping (except for '40 s' condition), the grain size obviously decreases when the antisolvent is dropped during spin-coating, consistent with previous studies [35][36][37].To investigate the photo-physical properties of the quasi-2D perovskite, we first measured the steady-state PL spectra of the films (figure 1(c) and table S2) (sample images in figure S5).In figure 1(e), the PL intensity corresponds closely to the current-voltage-luminance characteristics.The maximum PL intensity substantially increases with anti-solvent dripping compared to no dripping condition (0.93 × 10 6 ), reaching the highest value of 2.45 × 10 6 with 25 s condition while it decreases with earlier or later dripping timing.The PL peak position is blue-shifted with delayed anti-solvent dripping time and 'No dripping' conditions (figure 1(d)), which implies the presence of quasi-2D perovskite phases with low-n values in these films.In figure 1(c), a noticeable PL peak around 440 nm and 465 nm were visible for the 'No dripping' film, indicating the presence of low-n phases (n = 2, 3 phases).The presence of low-n phases was also confirmed from characteristic excitonic absorption features in UV-vis absorption spectra (figure 1(f)) [38,39].The 'No dripping' perovskite exhibits strong n = 1, 2, and 3 phase peaks at 405 nm, 433 nm, and 460 nm, respectively.This implies that the 'No dripping' perovskite comprises multiple n phases, ranging from the pure 2D phase to the high-n phase.This distinct quasi-2D multiphase structure features multiple energy barriers between different phases, resulting in energy losses during the energy transfer process, which might be the possible origin of the poor PeLED performance [40][41][42].Such an absorption peaks seem to disappear with anti-solvent dripping in earlier timing ('15 s' and '25 s' conditions) and re-emerging with increasing spinning time before the dripping ('30 s' , '35 s' and '40 s' conditions).In figure S9, a trend, in line with the context, appears on the XRD patterns of perovskite films with different anti-solvent dripping conditions.All of the conditions show a 3D perovskite peak at 15.14 • .Interestingly, the significant peaks of low-n phases (n = 1 phases at 5 • , 11.08 • , 15.78 • and n = 2 phases at 11.84 • ) appear in 'No dripping' in XRD data and these peaks disappear in films with antisolvent dripping at earlier timing.For the'40 s' condition, the n = 2 peak emerged again [43,44].Based on these observations, we are further motivated to perform an in-depth investigation of the phase distribution under different dripping conditions.
The femtosecond transient absorption spectroscopy (fs-TA) spectra can offer valuable insights into the distribution of charge carriers between different phases in quasi-2D perovskite films (figures 2 and S6).All the films demonstrated characteristic bleaching peaks that can be assigned to quasi-2D perovskites with different n values [45,46].Right after excitation, bleaching peaks for lower n-values were observed at around 430 nm (n = 2) and 455 nm (n = 3) were observed, which was followed by the appearance and intensification of the peak for higher n-values at 500 nm (n ⩾ 5).This indicates that charge carrier funneling progressively occurs from the low-n phases to low n-phases.Between different dripping conditions, certain systematic tendency was observed; The bleaching peaks for lower n phases at 430 nm and 455 nm were most This observation suggests a smooth and efficient progression of carrier transfer within the system.Whereas the bleaching peaks for the lower n-phases were still observable even after 100 ps for 'no dripping' and'40 s' conditions, indicating inefficient charge carrier funneling in this system (figure S7).In figures 3(a) and (b), we utilized the normalized TA spectra to tentatively quantify distribution of quasi-2D phases.The ratio between the characteristic bleaching peaks was calculated from the normalized TA spectra measured at 0.5 ps.The 'No dripping' condition exhibited a unique presence of a pure 2D phase (n = 1), alongside n = 2 and n ⩾ 3 phases.On the contrary, the '25 s' perovskite showcased negligible signals for the lower n-phases The distribution of the phases with different n-values can significantly impact the behavior of the injected carriers and the resultant electroluminescence efficiency.For an in-depth exploration of the n-value distribution in the device, we employed time-of-flight secondary ion mass spectrometry (ToF-SIMS).We carefully selected three conditions (no dripping, 15 s, 25 s and 40 s) that exhibit comprehensive changes in the phase distribution.We tried to identify the depth dependent distribution of the bulky 'A' site cation, distribution of F-PEA that might provide insight into the distribution of phases with different n-values.As the F − ion reached saturation, we utilized the F 2 − ion to provide a clearer depth profile within the perovskite layer [18,47,48].Interestingly, the F 2 − ion intensity significantly differs at the surface under different dripping conditions.Without dripping (figure 4(a)), the concentration of F 2 − ions is notably high at the surface, indicating the accumulation of low-n phases in the film surface region [49].With the 15 s and 25 s dripping conditions, the F 2 − ions at the surface decrease, and the intensity of the F 2 − ions at the film bottom surface increases.With the longer spinning time (40 s), the intensity of the F 2 − ions at the film surface increases again.From an energy formation standpoint, the phases with the lower n-value have relatively lowformation energy, implying that low-n phases are more readily formed [40,41,50,51].Hence, the concentrated low-n phases at the surface of the 'No dripping' film indicates crystallization primarily initiated from the film surface during the extended spin time caused by DMSO solvent evaporation from the surface, which was also reported in previous studies [30,52].With anti-solvent dripping, on the contrary, the formed intermediate phase might be retained during the spinning, which is then crystallized during the annealing process on the hot plate.In this case, crystallization might be partially initiated from the film's bottom surface, and thus low n-phases with relatively low formation energies might be formed at the bottom surface.The longer the spinning time before the anti-solvent dripping (40 s), the vaporization of the DMSO from the film surface becomes pronounced, and thus low n-phases start to form at the surface region.The formation of low n-phases at the surface region without dripping (no dripping) or longer spinning time (40 s) is supported by higher intensity of the detected F 2 − ions at the film surface (figures 4(a) and (d)).Steady-state PL measurements with different directions of the incident excitation laser confirmed the corresponding tendency.In the 'No dripping' case, the PL peak position was blue-shifted when the excitation beam was incident from film's top surface whereas the blue-shift was observed for 25 s sample when the excitation laser was incident from film's bottom surface (figure S8).
Figure 5 illustrates the quasi-2D phase distribution and energy band alignment for both 'No dripping' and optimal ('25 s') conditions, building upon the observations derived from the previous analyses.In figure 5(a), the 'No dripping' perovskite layer displays a surface primarily constituted by low-n 2D phases, leading to a distinctive charge imbalance and hole leakage due to a substantial energy barrier present (figures 5(a) and (c)) [11,[53][54][55][56][57].Moreover, variations in multi-n phases result in the accumulation of carriers and subsequent recombination, evident from the PL intensity graph (figure 1(c)).On the other hand, in figure 5(b), the '25 s' perovskite shows a more uniform and even distribution of quasi-2D phases, allowing for a more seamless and efficient funneling of carriers toward a homogeneous perovskite layer.

Figure 1 .
Figure 1.(a) Current density-voltage and luminance and (b) EQE-current density curve of PeLEDs.(c) Steady-state photoluminescence (PL) spectra and (d) normalized PL spectra of perovskite films coated on glass/GraHIL.(e) Maximum PL intensities and peak positions as a function of antisolvent dripping condition.(f) Normalized absorbance of the perovskite films coated on glass/GraHIL.

Figure 2 .
Figure 2. TA spectra at different time delays and TA data map of selected quasi-2D perovskite films, (a) No dripping, (b) 25 s, and (c) 40 s.(excitation wavelength at 350 nm).

Figure 3 .
Figure 3. (a) The normalized amplitude of ground-state bleaches (GSBs) in TA spectra at 0.5 ps delays under 350 nm excitation.(b) Phase distributions for quasi-2D perovskite were calculated by GSBs from TA signal.

Figure 5 .
Figure 5. Schematic illustration of phase distribution in (a) No dripping and (b) 25 s perovskite films.And the band structure in (c) No dripping and (d) 25 s perovskite films.