Elucidation of the suppression of photoinduced segregation in 2D mixed halide, A₂PbI₂Br₂: Critical role of A₂PbBr₄ photostability

Summary

2D lead halide perovskites are used to improve device operational stability due to increased environmental stability and reduced ion migration compared to 3D perovskites. However, the relationship between the 2D perovskite stability under illumination and spacer cation is still not well understood. Thus, we examine photoinduced halide segregation (PHS) in different 2D mixed-halide perovskites and show that PHS is suppressed in the materials which have photostable bromide halide phase. As the spacer cations provide the barrier to ion migration in 2D perovskites, PHS would be facilitated by the loss of spacer cations through their interactions with various mobile oxidized halide species, resulting in organic ammonium deprotonation and spacer cation vacancy formation. The existence of a photostable A₂PbBr₄ phase, which does not exhibit spacer cation loss, results in the suppression of PHS in 2D mixed halide perovskites due to reduced spacer vacancy formation and consequently reduced halide ion migration under illumination.

Graphical abstract

Highlights

• •Photoinduced halide segregation (PHS) in 2D perovskites is spacer cation dependent
• •Photooxidative degradation of 2D perovskites is also spacer cation dependent
• •PHS and photooxidative degradation occur due to photo/electrochemical redox reactions
• •PHS is suppressed by reduced spacer vacancy formation when photostable phase exists

Natural sciences; Applied sciences; Materials science

Introduction

2D lead halide perovskites have been commonly proposed to improve the stability of 3D lead halide perovskites APbX₃, where A is a small organic cation or Cs⁺ and X is halide anion.^1,2,3,4,5,6 These materials can be described with a formula C₂A_n-1Pb_nX_3n+1 for commonly used Ruddlesden-Popper perovskites, where C is the bulky organic spacer cation which separates the perovskite octahedral layers, while n is the number of layers.¹ They are commonly assumed to exhibit increased stability when exposed to humidity and/or elevated temperature, and to suppress ion migration,^1,2,5 and they have resulted in device stability improvements both for quasi-2D (n > 1) and 3D/2D (n = 1) perovskites.^4,5,6 The suppression of ion migration in 2D perovskites is particularly interesting for improving device operational stability, and it is commonly attributed to increased barrier to out-of-plane ion migration^1,2,5 caused by the presence of bulky spacer cations between the layers,^1,5 and reduced concentration of halide and small cation vacancies for in-plane ion migration.⁵ These factors also suppress photoinduced halide segregation (PHS),⁵ although in this case preferential distribution of Br and I sites within 2D perovskite could also play a role in the suppression^5,7 since I⁻ and Br⁻ exhibit different preferences for occupation of axial and equatorial sites.^7,8

To date, PHS and/or ion migration have been investigated in different 2D and/or quasi-2D perovskites both experimentally^{3,9,10,11,12,13,14,15,16,17,18,19,20,21} and theoretically.^22,23 In general, halide diffusion was more significant in quasi-2D (n > 1) than 2D (n = 1) materials,^{12,16,17,20,21} but there is very pronounced difference in behavior of 2D perovskites with different spacer cations. Significant PHS in 2D perovskites has been reported for some spacer cations,^9,10,11,19 while for others, such as phenethylammonium (PEA),^10,11,16,20 negligible photoinduced segregation is observed over similar timescales. For some spacer cations, such as commonly studied butylammonium (BA), not only there was a significant photoinduced segregation in BA₂PbI₂Br₂ but also expulsion of both I⁻ and Br⁻ into the solution was observed under illumination and bias.^9,24 The instability of some 2D materials under illumination,⁹ as well as the evidence of migration of both anions and cations (including bulky spacer cations) in this class of materials,^24,25,26,27 has resulted in questioning of the use of 3D/2D approach to stability improvements in the perovskite devices.⁹ It is therefore imperative to improve our understanding of ion migration and/or photoinduced segregation in 2D perovskites to facilitate rational selection of 2D capping layers for improved device stability.

Efforts have been made to elucidate which properties of the spacer cation are responsible for the suppression of halide ion migration.^{3,12,14,16,20,22,23,25} These studies generally agree that bulkier π-conjugated spacers can inhibit halide diffusion more compared to short chain aliphatic spacers.^3,12 Increased stability and reduced ion transport were linked to increased length of the spacer cation,²² increased interactions between spacer cations,²⁵ and increased lattice rigidity,^12,14 as bulkier and stiffer cations would result in higher energy for halide vacancy formation, which in turn would result in slower halide diffusion.¹⁴ In contrast, increased ligand polarization was proposed to increase ion transport.²² In addition, the suppression of halide diffusion was also attributed to non-binomial halide distribution²⁰ and to reduced charge separation and consequently lower halide mobility¹⁶ in 2D perovskites. Computational attempts to explain ion transport in 2D perovskites include diffusion coefficient calculations (calculated analytically and estimated from kinetic Monte Carlo simulations),³ thermodynamic calculations,²³ and molecular dynamic simulations.^12,22 However, these models do not take into account redox reactions which occur in halide perovskite materials (both 3D and 2D) under illumination. Furthermore, the number of spacer cations considered in the same study (to eliminate variations due to differences in methods among different studies) is much smaller for experimental (up to four)³ compared to computational (up to seven)²² studies, which hinders comparisons between experiments and theoretical predictions. Consequently, the mechanism of halide segregation as well as cation migration in 2D perovskites is still not well understood.²⁵

It should be noted that PHS phenomenon has been investigated more comprehensively in 3D perovskites^{28,29,30,31,32,33,34,35,36,37,38,39,40} compared to 2D perovskites. Various explanations have been proposed to clarify the PHS in 3D perovskites,^{28,29,30,31,32,33,34,35,36,37,38} including processes involving defects/charge trapping through various mechanisms,^28,29,30,31 polaron formation,^37,38 and photogenerated carrier funneling.^32,33,34,36 Consequently, different mathematical models of driving forces for halide segregation, namely miscibility gap model (internal lattice strain), polaron-assisted lattice strain, and photocarrier energy model were proposed.³⁵ In recent years, however, the role of photo/electrochemical redox reactions in PHS and ion migration has been gaining more recognition.^39,40,41 It should be noted that this explanation is fully consistent with the recognized roles of defects in the process,²⁸ as defects are generated in the electrochemical redox reactions, as well as I₂-induced photoinstability,^30,31 as I₂ is a product of I⁻ oxidation by photogenerated holes.^35,40 Due to differences in redox properties (iodide is more readily oxidized than bromide and the oxidation forms highly mobile oxidized species)^40,41 and bandgaps (charge carrier funneling to lower bandgap iodide-rich domains occurs),⁴⁰ the photo/electrochemical redox processes offer a possibility to provide a unified explanation of different experimental observations under illumination and/or bias. The important difference compared to other models is that photo/electrochemical redox reactions take into account continuous generation and annihilation of defects under illumination or bias, and the fact that oxidized halide species is not restricted to vacancy hopping mode of transport as oxidized halide species, such as interstitial iodide and I₂, are highly mobile.⁴¹ Consequently, these processes are expected to play an important role in all types of perovskite materials, and need to be considered when investigating PHS in 2D perovskites.

Results and discussion

Here, we examined PHS in 2D lead halide perovskites under simulated solar illumination using in-situ absorption measurements, as shown in Figure 1. In addition to commonly studied BA-based and PEA-based perovskites, we considered 6 additional spacer cations, namely benzylammonium (BZA), hexylammonium (HA), 2-thiophenmethylammonium (TMA), 2-thiophenethylammonium (TEA), 4-fluoro-phenethylammonium (4FPEA), and iso-butylammonium (i-BA). We can observe that for some spacer cations, 2D perovskite readily segregates, while others exhibit good photostability. The obtained results for the two commonly studied 2D perovskites, BA₂PbI₂Br₂ and PEA₂PbI₂Br₂, are in agreement with previous reports of significant segregation for BA^9,10,11 and negligible segregation for PEA,^10,11,20 as well as the report of one magnitude slower halide ion mixing for PEA compared to BA.¹⁴ It is also interesting to note that slower PEA cation migration compared to BA cation migration was also observed.²⁴ Perovskite with TEA spacer exhibited similar behavior to PEA, in agreement with similarity in ion transport for these two spacers, which exhibited much slower ion transport compared to BA-based perovskites in lateral diffusion experiments.¹² Thus, reported trends in the literature, with BA exhibiting significant PHS and PEA and TEA having improved photostability^{9,10,11,14,20} are in excellent agreement with our results. Nevertheless, it should be noted that direct comparisons between different studies of PHS could be affected by light source and light intensity used, as well as the film quality (grain size, crystallinity, and native defect concentrations). However, when materials studied exhibit such pronounced difference in behavior, as observed for BA and PEA, it is expected that the same trends are observed across different reports and this is indeed the case. One significant observation from these experiments is that while significant ion transport and consequently PHS is indeed observed for short aliphatic chain spacers in agreement with expectations,^3,15 we observe varied response in π-conjugated spacers, which are expected to exhibit suppression of ion transport.¹⁵ From a previous experimental observation of PHS in 2D perovskite with phenylpropylammonium (PPA) spacer,¹⁹ it is obvious that perovskites with π-conjugated spacers can indeed segregate, but the mechanisms responsible for fast segregation for BZA and suppressed segregation for PEA and 4FPEA are not fully clear.

Figure 1.

In-situ absorption spectra of 2D mixed halide perovskite under illumination

In-situ absorption spectra recorded at different times during the photoinduced segregation under 100 mW/cm² simulated solar illumination for mixed halide perovskite A₂PbBr₂I₂, where A denotes the following spacer cations (A) BZA (B) PEA (C) BA (D) 4FPEA (E) HA (F) TEA (G) i-BA (H) TMA.

Thus, we have selected BZA and PEA as model materials for a more detailed investigation, since the two spacers have very similar structure, differing only in the length of the alkyl chain on the phenyl ring (one and two carbon atoms for BZA and PEA, respectively). The X-ray diffraction (XRD) patterns of BZA-based and PEA-based perovskites are shown in Figure S1 and SEM cross-section of the films is shown in Figure S2, showing that in both cases thin films with similar thickness are obtained. To first examine the thermodynamic stability of these mixed perovskites, we calculated Helmholtz compositional energy²³ ΔF⁰ for PEA₂PbBr₂I₂ and BZA₂PbBr₂I₂, as shown in Figure 2A. Helmholtz compositional energy as a function of temperature T is defined as:²³ $\Delta F^0\left(T\right)=\Delta U-T\Delta S\left(T\right)$, where ΔU is the mixing enthalpy and ΔS is the mixing entropy (calculations are described in more detail in STAR Methods). We can observe that BZA₂PbBr₂I₂ is unstable (ΔF⁰ > 0) for temperatures up to ∼150°C, while PEA₂PbBr₂I₂ is stabilized by entropic effects already at low temperatures and is extremely stable at around room temperature. We have then examined PHS at low illumination power (6.5 mW/cm²) where Helmholtz compositional energy dominates over photocarrier free energy. We can indeed observe significant segregation for BZA and suppressed segregation for PEA over entire temperature range, as shown in Figures S3 and S4. From the experimental data on the change of absorption for Br-rich, I/Br mixed region, and I-rich region, as shown in Figures 2B and 2C, we can observe a significant difference in the behavior of Br-rich region for BZA and PEA, since no Br-rich region forms for PEA. This difference between BZA and PEA persists at different temperatures, as shown in Figure S5 illustrating the change in the absorption at 85°C.

Figure 2.

Helmholtz compositional energy and absorption change of BZA₂PbBr₂I₂ and PEA₂PbBr₂I₂ mixed halide perovskite

(A) Helmholtz compositional energy of BZA₂PbBr₂I₂ and PEA₂PbBr₂I₂. Dotted vertical lines serve to guide the eye.

(B and C) Changes in the absorption at peak wavelengths corresponding to Br, Br/I, and I peaks vs. illumination time at 25°C for (B) BZA₂PbBr₂I₂ and (C) PEA₂PbBr₂I₂.

As the film preparation for each spacer cation has been optimized to achieve well-mixed starting film and consequently optimal annealing temperature differ, we have also performed PHS experiments for BZA₂PbBr₂I₂ films annealed at different temperatures, since films without secondary phases could be obtained over a wider range of temperatures compared to PEA. From Figure S6, we can observe that segregation occurs for all annealing temperatures, indicating that the observed differences can be attributed to the spacer cation used and not film annealing temperature. In addition, photoluminescence (PL) measurements were performed to confirm observed behavior differences between BZA₂PbBr₂I₂ and PEA₂PbBr₂I₂. As shown in Figure S7, significantly larger red shift is obtained for BZA-based perovskite compared to PEA-based perovskite. However, it should be noted that PL measurements have a drawback in characterizing PHS phenomenon, as the emission typically occurs from the lowest bandgap I-rich phase,^9,34,40 which was reported to account only for 1–2% of the mixed halide perovskite film.³⁴ In contrast, XRD and absorption spectroscopy probe all compositions in the film,³⁴ which makes absorption spectroscopy more suitable for comprehensive investigations of PHS.

Reversibility of PHS was investigated by maintaining the previously illuminated samples in the dark for a set period of time and performing absorption measurements. To examine the possible effects of the loss of volatile degradation products, both samples illuminated in inert environment without encapsulation are examined. Obtained results are shown in Figures S8 and S9. We can observe that encapsulation is necessary to ensure the reversibility of the process, in agreement with a previous report.¹¹ However, remixing process at room temperature for both BZA₂PbBr₂I₂ and PEA₂PbBr₂I₂ is very slow, with segregation persisting up to 48 h, different from a previous report on BA₂PbBr₂I₂ which exhibited complete remixing^9,11 on the timescale around one order of magnitude longer compared to time needed for PHS.⁹ Possible reasons for the observed difference could be the use of different illumination source and different film microstructure (as the initial films exhibited very weak photoluminescence, as well as very fast PHS), as well as different spacer cation used. For PEA₂PbBr₂I₂, dark recovery was found to be incomplete¹⁶ or absent,²⁰ which was attributed to nonbinomial halide distribution in PEA₂PbBr₂I₂, which results in additional energy barrier for Br⁻/I⁻ to move into I/Br-rich octahedra.²⁰ Thus, very long time needed for remixing observed is in agreement with previous studies using the same spacer (PEA),^16,20 and likely due to the preferential halide distribution in axial and equatorial positions (see STAR Methods). When the samples are encapsulated and heated dark remixing occurs, as shown in Figure S9B and S9D, in agreement with observation of remixing at elevated temperature only for quasi-2D perovskite with PEA spacer cation.²⁰

To explain significant differences in the emergence of Br-rich region in BZA and PEA, let us consider how PHS occurs. Halide segregation in both 3D and 2D perovskites involves iodide oxidation by photogenerated holes as an initial step, with subsequent formation of interstitial iodine and iodide vacancies, followed by halide migration, which then results in halide segregation and/or loss of iodide.^25,39,41 The photo/electrochemical nature of the reactions involved in PHS was confirmed by iodide expulsion experiments into the solution under illumination and/or bias, with the detection of oxidized iodide species (I₂ and I₃⁻) in the solvent.^9,40 In addition, the critical role of organic cations in the degradation of 3D perovskites initiated by photogenerated or injected holes through photo/electrochemical processes has been elucidated recently.^39,42,43 It was found that perovskite decomposition occurs when both organic cations and lattice halides participate in electrochemical redox reactions,⁴² and that organic ammonium cations are readily deprotonated by I₃⁻,⁴³ which is a product of iodide oxidation. These processes include, but are not limited to, the following processes involving iodide (as there can be secondary reactions between different reaction products):^{39,40,41,42,43} I⁻+h⁺→I_i⁰+V_I, I_i⁰ + h→ I_i⁺, I₂+ I⁻↔ I₃⁻, I_i⁺ + I_i⁺ ↔ I₂, I⁻ + 2h⁺→I_i⁺+V_I⁺, I_i⁰+e⁻→ I_i⁻, I_i⁻+V_I⁺ → I⁻, I_i⁺+V_I⁺ + 2e⁻ → I⁻, as well as reactions between oxidized halide and organic cation, such as I₃⁻+MA⁺ →MA + HI₃, where I_i denotes iodine interstitial and V_I denotes iodine vacancy, MA⁺ denotes CH₃NH₃ cation and MA denotes corresponding amine CH₃NH₂.

In contrast with 3D Perovskites, the PHS and ion migration are not as well understood in 2D perovskites. While it is known that the degradation of BA₂PbBr₄ under illumination involves the loss of BA cations,⁴⁴ and similar observation is made for BA₂PbI₄ and PEA₂PbI₄,⁴⁵ the role of cation loss in the photoinduced halide segregation and/or ion migration in 2D perovskites has not been investigated. As both PHS and instability of both mixed and pure halides under illumination are related to the trapping of holes on the halide, followed by the halide expulsion,⁹ we investigated the photostability of pure bromide and iodide 2D perovskites. The stability tests were performed in dry air to accelerate the degradation due to the fact that oxygen serves as an electron scavenger, resulting in an excess of photogenerated holes⁴⁰ which then result in oxidation of iodide, triggering the chain reactions of perovskite decomposition.²⁵ From the absorption spectra of the 2D perovskite films illuminated in dry air shown in Figure 3, we can observe significant degradation for both BZA₂PbBr₄ and BZA₂PbI₄. Corresponding Fourier transform infrared spectroscopy (FTIR) spectra indicate clear loss of the organic cation, based on the weakening of various peaks corresponding to CH_x and NH₃ vibrations.⁴⁴ In contrast, while degradation is observed for PEA₂PbI₄ under illumination, in agreement with the literature,⁵ PEA₂PbBr₄ exhibits good photostability and there is no weakening of organic cation vibration peaks in the corresponding FTIR spectra. This finding is in agreement with the lack of emergence of Br-rich region for PEA₂PbBr₂I₂ observed in our investigation, as well as previous report that PEA₂PbBr₄/PEA₂PbI₄ lateral heterostructures remained stable under UV illumination, while in quasi-2D n = 2 and n = 3 heterostructures loss of I occurred, followed by migration of Br into I-rich region.²¹

Figure 3.

Absorption and FTIR spectra of 2D Br-based or I-based perovskite under illumination

Absorption spectra of perovskite films illuminated under simulated solar illumination (100 mW/cm²) for different times in dry air environment for (A) BZA₂PbBr₄, (B) BZA₂PbI₄, (C) PEA₂PbBr₄, and (D) PEA₂PbI₄; corresponding FTIR spectra of for (E) BZA₂PbBr₄, (F) BZA₂PbI₄, (G) PEA₂PbBr₄, and (H) PEA₂PbI₄.

For other spacer cations investigated, we can observe a similar pattern, namely illumination in dry air leads to degradation of both endpoint halides (Figures S10–S13) for those spacers which exhibit significant PHS for mixed halide A₂PbBr₂I₂ perovskite (as shown in Figure 1). From absorption spectra and XRD patterns, we can observe that bromides are generally more photostable than iodides, and that stable bromide perovskite effectively suppresses segregation. In the case of bromides, we observe that PEA-, 4FPEA-, and TEA-based materials exhibit only a very small decrease on XRD intensity, without appearance of any new phases while the decrease in intensity is quite pronounced in the case of BA-, BZA-, HA-, and i-BA-based bromide perovskites. Different from bromides, in the case of iodides we observe significant decrease in the absorption peaks after illumination in almost all spacer cations (except for TEA). In addition to a decrease in intensities, low intensity peaks belonging to PbI₂ appear in XRD patterns of iodides (as well as other phases in the case of PEA, i-BA, and TMA). The observed lack of photostability in BA₂PbBr₄, and BA₂PbI₄ is in agreement with previous work, which showed that BA₂PbI₂Br₂, BA₂PbBr₄, and BA₂PbI₄ are unstable under illumination and bias, with expulsion of both I⁻ and Br⁻ detected.⁹

PHS in both 3D and 2D perovskites has been commonly explained considering the oxidation of halides by photogenerated holes, predominantly iodide as it has lower oxidation potential,^25,42,46 as well as lower bandgap resulting in hole accumulation in iodide-rich phase.²⁵ While both Br and I can be transported through the lattice by vacancy hopping, the preferential oxidation of lattice iodide resulting in iodine interstitials creates an imbalance in iodide and bromide transport.⁴⁶ However, photoelectrochemical reactions involve both oxidation and reduction reactions, and the involvement of organic cations in these reactions is evident from the cation movement across 3D/2D interfaces under illumination.^25,26 Thus, we propose that PHS in 2D perovskites involves both oxidation of halide by photogenerated holes, and reduction/deprotonation of organic cation by photogenerated electrons or oxidized halide species. Oxidized halide species, such as iodine interstitials (I_i, I_i⁻, and I_i⁺), molecular iodine (I₂), and triiodide (I₃⁻), participate in various secondary reactions, which have been shown to include deprotonation of organic ammonium cations in 3D perovskites.⁴³ Deprotonation of organic cation would lead to the formation of cation vacancy, and spacer cation vacancies would lower the barrier for halide ion transport since spacer cations are responsible for suppression of ion migration in 2D perovskites.^3,5 If the spacer cations can react with oxidized halide species for both iodides and bromides, then a large number of spacer cation vacancies could readily form. On the other hand, the lack of reaction between spacer cation and oxidized bromide species, implied from photostability of A₂PbBr₄ perovskite, would lead to fewer vacancies of the spacer cations, which in turn inhibits the halide transport across the layers. The schematic diagram of the process for two types of 2D perovskites (with unstable and stable A₂PbBr₄ phases) is illustrated in Figure 4A.

Figure 4.

Schematic diagram and EDX measurent results

(A) Schematic diagram of photoelectrochemical reactions in the perovskites with both bromide and iodide phases susceptible to photooxidation (left) and only iodide phase susceptible to photooxidation (right); normalized halide:Pb ratios obtained from EDX measurements for different 2D perovskite films subjected to bias and illumination. The values have been normalized by halide content in the starting samples since we consider both A₂PbI₄ and A₂PbI₂Br₂.

(B) I:Pb.

(C) Br:Pb.

The proposed lack of interaction between oxidized bromide species and organic cation in photostable perovskites is also consistent with the fact that only iodide expulsion is observed from MAPbBr_xI_1-x, where MA denotes methylammonium,⁴⁰ where MAPbBr₃ has good photostability.^47,48 Due to differences in crystal structure (the lack of spacer cation barriers to ion transport^1,2,5 present in layered structures and the lack of different preferential occupation of axial and equatorial sites for Br⁻ and I⁻ which occurs in 2D perovskites^7,8), PHS still occurs despite stable bromide perovskite for 3D materials. In contrast, for BA 2D perovskites both I⁻ and Br⁻ expulsion was detected, although slower rate was observed for Br⁻,⁹ in agreement with the proposed mechanism and the instability under illumination for both BA₂PbI₄ and BA₂PbBr₄ (Figures S10–S13). The involvement of an organic cation in the photo/electrochemical reactions leading to PHS and ion migration is also confirmed by the fact that no iodide expulsion in the solution is observed for Cs-based 3D perovskite,⁴⁹ where Cs⁺ could be reduced by photogenerated electrons but cannot be deprotonated by oxidized iodide species. From the proposed mechanism, it follows that more significant segregation would be observed for higher iodine content for PEA₂PbBrI₃, as can indeed be observed in Figure S14, different from thermodynamically predicted higher stability for higher iodine content.²³ In all cases, BZA₂PbBr_xI_1-x shows significant segregation, also shown in Figure S14. Finally, to further test the hypothesis of photo/electrochemical reactions, we prepared devices containing BZA- and PEA-based perovskite (pure Br, pure I, and Br:I 1:1) and subjected them to electrical bias or illumination without encapsulation and compared X:Pb (X = I, Br) ratio in fresh and aged perovskite films after removal of electrode and electron transport layer. Starting stoichiometry of the samples obtained from energy dispersive X-ray spectroscopy (EDX) measurements is shown in Table S2. Obtained results, shown in Figures 4B and 4C, clearly show same trends across different materials, although amount of halide loss for the same material varies on the stress type (bias, illumination). In BZA-based materials, both loss of iodide and bromide are detected for all compositions. For PEA-based samples under illumination, we observe significant loss of iodide for pure PEA₂PbI₄ only, while the loss of both bromide and iodide is negligible for PEA₂PbBr₂I₂, and the loss of bromide is negligible for pure PEA₂PbBr₄.

We have also examined the expulsion of halide into the toluene solution under illumination, as shown in Figure 5. Both films exhibit good stability in toluene without illumination for 1 h. For BZA₂PbBr₂I₂, we observe similar behavior as previously reported for BA₂PbBr₂I₂ in dichloromethane, namely disappearance of the mixed halide and I-rich phase peaks, with pure bromide phase peak persisting for some time after disappearance of mixed and I-rich phase.⁹ In the case of PEA₂PbBr₂I₂, we can observe some loss of iodide phase, with illuminated films exhibiting clear mixed phase peak, as well as a smaller pure bromide peak, which can be attributed to the loss of iodide. From the absorption spectra of the toluene solutions after perovskite film immersion and illumination, Figure 5C, we can observe that expulsion of iodide is much smaller in the case of PEA-based perovskite, consisted with the changes in the absorption of films and overall observed behavior of PEA-based perovskite under illumination and bias.

Figure 5.

Absorption spectra of perovskite films and toluene solutions

(A and B) Absorption spectra of (A) BZA₂PbBr₂I₂ and (B) PEA₂PbBr₂I₂ films before and after immersion in toluene for 1 h in the dark and under illumination.

(C) Absorption spectrum of toluene solution after the immersion of 2D perovskite films under 1 Sun illumination for 1 h.

From the obtained results, we can conclude that perovskites alkyl chain spacer cations generally have lower photostability. The presence of a ring (phenyl or thiophene) is not sufficient ensure photostability. There is a trend with respect to the length of alkyl chain attached to the ring, with spacer cations containing two carbon atoms chain (PEA, FPEA, and TEA) exhibiting good photostability, while those with one carbon atom (BZA and TMA) are susceptible to PHS and photoinduced degradation. As the spacer cations with three carbon atoms chain are also susceptible to PHS,¹⁹ and the length of the alkyl chain affects the packing of the spacer cations and consequently their interactions, it is possible that ethyl chain attached to a phenyl or thiophene ring could lead to favorable conformal arrangements which would result in enhanced photostability. Alternatively, the length of the alkyl chain could affect the likelihood of deprotonation of the terminal ammonium group, which would also affect the photostability. In the absence of more comprehensive theoretical investigations looking into the effect of spacer cation structure and ordering on the 2D perovskite photostability, the photostability of pure bromide perovskite represents a good experimental predictor of the susceptibility of mixed halide perovskite to PHS and vice versa.

Conclusions

Thus, we show that PHS is suppressed in A₂PbBr₂I₂ 2D perovskites which exhibit photostable A₂PbBr₄ phase. From the known ability of triiodide I₃⁻ to deprotonate organic ammonium cations,³² leading to desorption of organic amine and creation of organic cation vacancy, and the finding that most of the 2D iodide perovskites degrade under illumination, we propose that the lack of reaction between the spacer cation and oxidized bromide species is responsible for the suppression of PHS in A₂PbBr₂I₂ 2D perovskites with photostable A₂PbBr₄ phase. This results in fewer spacer cation vacancies compared to 2D perovskites where organic cation can react with both oxidized bromide and iodide species, leading to the desorption of amine and facilitation of interlayer ion transport.

Limitations of the study

We have considered eight different organic ammonium spacer cations which form 2D Ruddlesden-Popper organic lead halide perovskites to investigate their photoinduced halide segregation and photostability. While this is more comprehensive compared to other investigations in the literature and selected cations have different structures to ensure that general conclusions could be reached, there is a huge number of possible organic ammonium cations so that an exhaustive investigation of all possibilities is not practically feasible. Nevertheless, we believe that selected cations are sufficiently representative to ensure valid conclusions for the photostability of 2D Ruddlesden-Popper halide perovskites. However, the conclusions apply for the case where 2D perovskite is formed across all compositions; the discussion of ion migration barriers would not necessarily apply to any non-perovskite structures formed (for some spacer cations, it is possible that perovskite structures would form only for certain values of x in A₂PbBr_xI_1-x, for example for tert-butylammonium).

Resource availability

Lead contact

Further information and requests for resources and materials should be directed to and will be fulfilled by the lead contact, Aleksandra B. Djurišić (dalek@hku.hk).

Materials availability

All materials used are commercially available and are used as received. This study did not generate new unique reagents.

Data and code availability

All data reported in this paper will be shared by the lead contact upon request. This paper does not report original code.

Acknowledgments

This work was supported by the Seed Funding for Basic Research, the Seed Funding for Strategic Interdisciplinary Research Scheme, and Research Output Prize of the University of Hong Kong, RGC CRF project 7018-20G and NSFC project 6207032617. A.W.H.-B. is supported by the Australian Research Council (ARC) Future Fellowship FT210100210. T. L. L. acknowledges support from the University of Sydney Faculty of Science Postgraduate Research Excellence Award (PREA).

Author contributions

Conceptualization, A.B.D., J.P., and T.L.L.; experimental measurements and figure plotting, Z.R., Z.Y., T.L.L., and Y.H.; calculations, J.O. and I.L.; supervision, A.B.D., A.W.Y.H.-B., and I.L.; writing the first draft, A.B.D.; writing-review and editing, A.B.D., J.P., .I.L, and A.W.Y.H.-B.

Declaration of interests

The authors declare no conflict of interest.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Chemicals, peptides, and recombinant proteins
N,N-dimethylformamide (DMF, anhydrous, 99.9%)	Thermo Scientific Chemicals	CAS: 68-12-2
Dimethyl Sulfoxide (DMSO, anhydrous, ≥99.9%)	Sigma-Aldrich	CAS: 67-68-5
Lead(II) bromide (PbBr2, ≥98%)	Tokyo Chemical Industry	CAS: 10031-22-8
Lead(II) iodide (PbI2, ≥98%)	Tokyo Chemical Industry	CAS: 10101-63-0
(2-(9H-carbazol-9-yl)ethyl)phosphonic acid (2PACz, >98.0%)	Tokyo Chemical Industry	CAS: 20999-38-6
γ-butyrolactone (GBL, >99.0%)	Tokyo Chemical Industry	CAS: 96-48-0
n-butylammonium bromide (n-BABr)	Greatcell Solar	CAS: 15567-09-6
n-butylammonium iodide (n-BAI)	Greatcell Solar	CAS: 36945-08-1
iso-butylammonium bromide (i-BABr)	Greatcell Solar	CAS: 74098-36-5
iso-butylammonium iodide (i-BAI)	Greatcell Solar	CAS: 205508-75-4
n-hexylammonium bromide (HABr)	Greatcell Solar	CAS: 7334-95-4
n-hexylammonium iodide (HAI)	Greatcell Solar	CAS: 54285-91-5
benzylammonium bromide (BZABr)	Greatcell Solar	CAS: 37488-40-7
benzylammonium iodide (BZAI)	Greatcell Solar	CAS: 45579-91-7
phenethylammonium bromide (PEABr)	Greatcell Solar	CAS: 53916-94-2
phenethylammonium iodide (PEAI)	Greatcell Solar	CAS: 151059-43-7
4-fluoro-phenethylammonium bromide (4FPEABr)	Greatcell Solar	CAS: 1807536-06-6
4-fluoro-phenethylammonium iodide (4FPEAI)	Greatcell Solar	CAS: 1413269-55-2
2-thiopheneethylammonium bromide (TEABr)	Greatcell Solar	CAS: 2490324-60-0
2-thiopheneethylammonium iodide (TEAI)	Greatcell Solar	CAS: 2414055-94-8
2-thiophenemethylammonium iodide (TMAI)	Greatcell Solar	CAS: 2247100-32-7
2-thiophenemethyl ammonium bromide (TMABr)	Xi'an Yuri	CAS: 59852-33-4
(6,6)-Phenyl C61 butyric acid methyl ester (PCBM, >99.8%)	Luminescence Technology	CAS: 160848-22-6
bathocuproine (BCP, >99.5%)	Luminescence Technology	CAS: 4733-39-5
Isopropyl alcohol (IPA, 99.5%)	Anaqua	CAS: 67-63-0
ethanol (99.9%)	Anaqua	CAS: 64-17-5
Other
Agilent Cary 60 UV-Vis spectrometer	Agilent	https://www.agilent.com/en/product/molecular-spectroscopy/uv-vis-uv-vis-nir-spectroscopy/uv-vis-uv-vis-nir-systems/cary-60-uv-vis-spectrophotometer
PDA-512 USB fiberoptic spectrometer	Control Development	http://www.controldevelopment.com/products/spectrometers/product-spectrometers-silicon-pda.php
Sunbrick™ Solar Simulator	G2V	https://g2voptics.com/products/led-solar-simulator-sunbrick/
Xenon lamp	Newport	https://www.newport.com/p/66906-150XF-R15
UV-curing Resin 3055B	ThreeBond	https://www.threebond.co.jp/en/product/threebond_3055b/
MiniFlex 600-C X-ray Diffractometer	Rigaku	https://rigaku.com/products/x-ray-diffraction-and-scattering/xrd/miniflex
Spectrum Two ATR-FTIR spectrometer	PerkinElmer	https://www.perkinelmer.com/Product/spectrum-two-ft-ir-sp10-software-l160000a
S4800 FEG SEM	Hitachi	https://www.hitachi-hightech.com/global/en/products/microscopes/sem-tem-stem/fe-sem/
Keithley’s Standard Series 2400 Source Measure Unit	Tektronix Technologies	https://www.tek.com/en/products/keithley/source-measure-units/2400-standard-series-sourcemeter
XploRA Plus Raman/PL spectrometer	Horiba	https://www.horiba.com/int/scientific/products/detail/action/show/Product/xploratm-plus-1528/
SU8230 FEG SEM	Hitachi	https://www.hitachi-hightech.com/global/en/products/microscopes/sem-tem-stem/fe-sem/
Ultim Max EDS detector	Oxford Instruments	https://nano.oxinst.com/products/ultim-max

Method details

Free energy calculations

Our aim is to estimate the free energy under illumination of mixed-halide Ruddlesden-Popper perovskites PEA₂PbBr₂I₂ and BZA₂PbBr₂I₂. The temperature-dependent free energy is:²³

$$\Delta F\left(T\right)=\Delta F^0\left(T\right)+\Delta F^{\ast }\left(T\right)$$ (Equation 1)

where $\Delta F^0$ is the Helmholtz compositional energy and $\Delta F^{\ast }$ is the photocarrier free energy. $\Delta F^0$ is defined as:

$$\Delta F^0\left(T\right)=\Delta U-T\Delta S\left(T\right)$$ (Equation 2)

where $\Delta U$ is the mixing enthalpy and $\Delta S$ is the mixing entropy. $\Delta U$ is given by:

$$\Delta U=E_M-0.5\left(E_{Br}+E_I\right)$$ (Equation 3)

where $E_M$, $E_{Br}$ and $E_I$ are the total energies per formula unit (f.u.) of the mixed I-Br and pure Br and I phases, respectively. Finally, for our use-case of identical bromide and iodine concentrations, the mixing entropy provides a constant contribution:

$$\Delta S=-3\mathit{ln}\left(2\right)k_B$$ (Equation 4)

where $k_B$ is the Boltzmann constant. At a given temperature, if ΔF (T ) > 0, the halide-segregated phase is (meta)stable under illumination, while the mixed phase is (meta)stable if ΔF(T) < 0. Similarly, ΔF⁰(T) > 0 indicates stability of the segregated phase and ΔF⁰(T) < 0 indicates stability of the mixed phase in the dark.

The crystal structure of PEA₂PbBr₂I₂ has been previously experimentally determined.⁵⁰ Similarly to (t-BA)₂PbBr₂I₂⁵¹ and BZA₂PbBr₂I₂ ,⁸ the halide distribution exhibits high structural ordering, with one species being almost exclusively found at equatorial positions of the perovskite layers, while the other is found at axial positions.

On the other hand, the crystal structure of BZA₂PbBr₂I₂ has not been experimentally determined. Our powder X-ray measurements show that the length of the c-axis of the unit cell amounts to 32.872(1) Å which is similar to the length of the c-axis of the pure bromide-based perovskite BZA₂PbBr₄ (33.35 Å). Therefore, we constructed guess- structures for BZA₂PbBr₂I₂ by substituting half of the bromide atoms in the experimentally determined structure of BZA₂PbBr₄ with iodines. As in our previous investigation,⁵¹ we performed the substitution in three ways, resulting in three guess-structures with the following halide distributions:

• (a)bromides in axial positions, iodines in equatorial positions;
• (b)iodines in axial positions, bromides in equatorial positions;
• (c)iodines and bromides alternately occupying halide lattice sites.

Starting from such guess structures, we optimized the atomic positions and unit cell parameters as described in the computational methods and compared the computed energies. We find that configuration (b) is the most stable, with configurations (a) and (c) being higher in energy by 0.12 eV/f.u. and 0.02 eV/f.u, respectively.

The halide distribution in the experimentally determined PEA₂PbBr₂I₂ crystal structure is the same as configuration (b) described above. We performed analogous computations on hypothetical PEA₂PbBr₂I₂ structures with halide configurations (a) and (c). We obtain that configuration (b) is indeed the most stable one, with configurations (a) and (c) being higher in energy by 0.14 eV/f.u. and 0.03 eV/f.u., closely resembling the trend observed for BZA₂PbI₄.

From the plot of the Helmholtz compositional energy ΔF⁰ for PEA₂PbBr₂I₂ and BZA₂PbBr₂I₂, shown in Figure 2A, of the main manuscript. We can see that BZA₂PbBr₂I₂ is unstable over a large range of temperatures, while PEA₂PbBr₂I₂ is stabilized by entropic effects already at low temperatures and is extremely stable at around room temperature.

It is worthwhile to note that we are using a theoretically obtained structure for BZA₂PbBr₂I₂, BZA₂PbBr₂I₂ which increases uncertainty in the corresponding calculated Helmholtz compositional energy. However, given the very large difference in the calculated enthalpies between the two systems (≈ 0.06 eV/f.u.), we may safely draw the main qualitative conclusion: the mixed-halide perovskite phase based on PEA is compositionally significantly more stable compared to the analogous phase based on BZA.

Under low illumination intensity, we expect that Helmholtz compositional energy will be the dominant factor contributing to free energy. To verify this, the photocarrier free energy ΔF^∗may be estimated as

$$\Delta F^{\ast }=n{E}_g$$ (Equation 5)

where n is the photocarrier density (number of photocarriers per f.u.) and E_g is the band gap of the mixed-halide phase.²³ We have calculated the band gaps of BZA₂PbBr₂I₂ and PEA₂PbBr₂I₂ and obtained the values 2.47 eV and 2.41 eV, respectively. The photocarrier density n may be approximated as:²³

$$n=G\tau$$ (Equation 6)

where G is the photocarrier generation rate and τ is the photocarrier lifetime. G is given by:

$$G=I\alpha V/h\nu$$ (Equation 7)

where I is the illumination intensity, α is the absorption coefficient, V is the volume per f.u. and hν is the photon energy. Taking the same values as were taken in the literature for PEA₂Pb(Br_xI_1-x)₄ (α = 10⁻⁵ cm⁻¹, V = 2.5 x 10⁻²² cm³, hν = 3 eV), the appropriate photocarrier lifetime τ = 1 ns¹ and the illumination intensity used in our experiments (I = 6.5 mW cm⁻²), we obtain n ≈ 3 x 10⁻²⁰ / f.u.

Given the similar values of the band gaps and the low estimated photocarrier density, at the illumination intensity used in our experiments and around room temperature, we may conclude that |ΔF^∗| ˂˂ | ΔF⁰(T)| for both BZA₂PbBr₂I₂ and PEA₂PbBr₂I₂, i.e., the tendency towards halide segregation is dominated by the compositional (in)stability of the considered systems, rather than a difference in the dynamical response of the systems to illumination.

Computational methods

We performed all calculations within density functional theory using the open-source Quantum ESPRESSO suite for quantum simulations of materials.^52,53 We employed the vdW-DF-cx density functional,^54,55,56,57 Garrity-Bennett-Rabe-Vanderbilt (GBRV) pseudopotentials,⁵⁸ a plane wave basis with a kinetic energy cutoff set to 60 Ry and a Monkhorst-Pack grid⁵⁹ with a 0.20 Å⁻¹ k-point distance. The atomic positions and cell parameters were relaxed until the change in the total energy between two steps of ionic relaxation was less than 10⁻⁴ a.u. and the forces on all atoms were less than 10⁻³ a.u. The target pressure was set to 0 bar.

Preparation of perovskite thin films

Precursor solutions of 2D bromide/iodide perovskites A₂PbBr₄/I₄ were prepared by dissolving monovalent spacer salt (ABr/I) and lead halide (PbBr₂/I₂) with molar ratios of 2:1 in DMF solvent (DMF:DMSO=9:1 for HA₂PbBr₄, and DMF:GBL=85:15 for HA₂PbI₄ as exception), where A denotes different spacer cations. Precursor solutions of 2D mixed halide perovskites were prepared by dissolving spacer bromide salt (ABr), spacer iodide salt (AI), lead bromide (PbBr₂), and lead iodide (PbI₂) with molar ratios of 4:0:1:1 (for A₂PbBr₃I), 2:2:1:1 (for A₂PbBr₂I₂), and 0:4:1:1 (for A₂PbBrI₃) in DMF/DMSO solvent (See Table S1 for details). Quartz substrates were cleaned by ultrasonic with detergent, DI water, ethanol and IPA in sequence. O₂ plasma treatment (60 seconds) was employed to improve wetting. All perovskite thin films were deposited by spin-coating the precursor solution on a substrate at 4000 rpm for 30 s in an argon filled glovebox, followed by annealing at 100°C for 10 mins (for A₂PbBr₄/I₄ films). Annealing condition for A₂PbBr₂I₂ can be found in Table S1. The deposition conditions (solvents used, annealing temperature and time) for each spacer cation were carefully optimized to ensure that starting films exhibit high quality mixed halide phase, as determined from absorption and XRD measurements. The optimization target was to obtain a sharp single mixed phase peak in the absorption spectra, with the verification of the presence of single mixed phase by XRD measurements. Spin-coating speeds were 3000 rpm for thicker films for FTIR measurement. All precursor solutions were stirred overnight before use, and the concentration of Pb²⁺ was 0.2 M, except for FTIR samples, the concentration of Pb²⁺ was 0.4 M to get thicker films. For all compositions, solvent selection and annealing conditions were optimized to obtain high quality films based on absorption spectra and XRD patterns.

Thin film characterizations

Absorption spectra were measured with Agilent Cary 60 UV-Vis spectrometer. In-situ absorption spectrums were converted through Beer–Lambert law from transmittance spectrums (A =log(I₀/I), where I₀ and I were the light intensity before and after film absorption), which were recorded by a PDA-512 USB fiberoptic spectrometer (Control Development Inc.) under 100 mW cm⁻² simulated solar illumination (Sunbrick™ Solar Simulator, G2V) or 6.5 mW cm⁻² white light source (150 W Xenon lamp) with an AM0 filter. Films for in-situ absorption measurements were encapsulated with a cover glass and ThreeBond TB3055B epoxy to avoid ambient exposure. XRD patterns of thin film were measured by Rigaku MiniFlex 600-C X-ray Diffractometer with Cu radiation source. FTIR spectra were obtained from Spectrum Two ATR-FTIR spectrometer (PerkinElmer). The structures shown in Figures S11 and S13 were confirmed by Rietveld refinement utilizing previously published structures. The periodicity (along the longest axis) was refined and amount to: 27.577(1) Å for BA₂BrBr₄ (according to Ref.⁶⁰), 33.303(4) Å for BZA₂PbBr₄ (according to Ref.⁶¹), 17.577(4) Å for PEA₂PbBr₄ (according to Ref.⁶¹), 33.497(1) Å for (4FPEA)₂PbBr₄ (according to Ref.⁶²), 35.618 Å for HA₂PbBr₄ (calculated based on Ref.⁶³), 31.356(2) Å for TEA₂PbBr₄ (according to Ref.⁶⁴), 28.908(3) Å for TMA₂PbBr₄ (calculated based on Ref.⁶⁵), 27.629(1)Å for BA₂BrI₄ (according to Ref.⁶⁶), 28.743(1) Å for BZA₂PbI₄ (according to Ref.⁶⁷), 32.871(1) Å for PEA₂PbI₄ (according to Ref.⁶¹), 16.731(7) Å for (4FPEA)₂PbI₄ (according to Ref.⁶⁸), 35.618 Å for HA₂PbI₄ (according to Ref.⁶³), 32.374(4) Å for TEA₂PbI₄ (calculated based on Ref.⁶⁴), 28.908(3) Å for TMA₂PbI₄ (according to Ref.⁶⁵), 27.498(2) for (i-BA) ₂PbBr₄ (according to Ref.⁶⁹) and 13.733(4) for (i-BA)₂PbI₄ (according to Ref.⁷⁰).

To characterize Br:Pb and I:Pb ratios after illumination or bias, ITO/2PACz/2D perovskite/PCBM/BCP/Ag devices were prepared, where ITO denotes indium tin oxide coated glass substrates. 2PACz was spin-coated on ITO substrates at 4000 rpm for 30 s, followed by annealing at 100°C for 10 min. 2D perovskite films were then deposited as previously described. After cooling to room temperature (RT), PCBM layer was deposited on top of 2D perovskite by spin-coating PCBM in CB (20 mg/ml) solution with 1200 rpm for 30 s. After annealing at 100°C for 10 mins, BCP in IPA (0.5 mg/ml) was spin-coated on the top with 4000 rpm for 30 s. Then, the samples were transferred into a thermal evaporator for depositing 80 nm Ag. To evaluate effects of electrical bias up to 8 V (0-8 V, 0.3 V/s, double scan, ITO positive, Keithley 2400 Source Measure Unit) or 4 hours illumination under 1 Sun (Sunbrick™ Solar Simulator, G2V), the Ag electrode was peeled off by a carbon tape. Then, 100 μl CB was dynamically spin-coated on the top of substrates at 5000 rpm for 30 s to remove PCBM layer. Br/Pb and I/Pb ratio were measured by energy dispersive X-ray spectroscopy (EDX) using Hitachi S4800 FEG SEM and Hitachi SU8230 FEG SEM with Oxford Ultim® Max detector (analysis area 170 mm²). Samples for EDX measurements were not encapsulated, which ensures that stoichiometry changes, if any, would not be reversible due to the loss of volatile degradation products. Perovskite cross-section SEM images were measured using Hitachi S4800 FEG SEM. Photoluminescence (PL) spectrum was measured using XploRA Plus Raman/PL spectrometer with 405 nm excitation.

Published: March 4, 2025