Quo vadis, perovskite emitters?

2019 marked the 10-year anniversary of the first application of halide perovskite in photovoltaics. From a humble 3.8% efficiency,¹ their performance has now surpassed 25% and is gaining parity with the ubiquitous Si solar cells. There has been much comparison between the two systems, and the accelerated developmental rate of perovskites is often touted as a phenomenal achievement over Si. However, one should also take into consideration the differences in the socioeconomic factors (e.g., increased investment by countries into research and development activities) and the greater urgency from worldwide phenomena of global warming and climate change in their respective eras that spur the development. Nonetheless, within this short decade of development, this large family of solution-processable halide perovskites has thrived across multiple optoelectronic domains. Their charm is primarily attributed to their low cost, facile processing, and outstanding optical and electronic properties (such as strong absorption, defect tolerance,² appreciable mobilities,³ and balanced ambipolar charge transport⁴), aptly referred to as a “poor man’s high efficiency semiconductor.”⁵ Specifically, perovskite’s advantages in the eyes of the light emission community lie in their high color purity (narrow emission linewidth), broad color tunability over the visible spectrum, and high luminescence quantum yields (QYs). Herein, this perspective examines the underlying photophysics in the various efficiency enhancement approaches undertaken and maps the possible new bearings for perovskite light emitting devices (PeLEDs) and perovskite lasers.

Electroluminescence (EL) from two-dimensional (2D) layered perovskites made their first appearance in a 1992 report by Hong et al. for (C₆H₅C₂H₄NH₃)₂PbI₄ and (C₆H₅C₂H₄NH₃)₂Pb₂I₇ at cryogenic temperatures.⁶ In spite of the efforts over the next decade, efficiencies of these 2D perovskite LEDs (PeLEDs) remained poor and most of these devices operated only at low temperatures.^7,8 Unfortunately, this brief chronicle overlapped with organic light emitting devices, which overshadowed these 2D perovskites that soon faded into obscurity.

The revival of PeLEDs is closely tied to the fortunes of perovskite photovoltaics beginning around 2013. The first 3D PeLED report⁹ by Tan et al. in 2014 for room temperature red and green devices with external quantum efficiencies (EQEs) of 0.76% and 0.1%, respectively, heralded a new development phase for PeLEDs. As seen from the chart progression in Fig. 1(a), a breakthrough in 3D PeLEDs efficiencies was achieved by Lee’s group in late 2015. This was attained by careful control of the MAPbBr₃ nanograins (a process coined “nanocrystal pinning”) to enhance the exciton confinement and reduce the exciton diffusion length. Concurrently, suppressing metallic Pb formation (with excess MABr) to minimize exciton quenching resulted in green-emitting devices with 8.53% EQE.¹⁰ 

From the photophysical perspective, the application of colloidal perovskite nanocrystals (PeNCs) in PeLEDs can be considered as a subset of this nanograin approach. Strong first-order excitonic emission can be achieved by leveraging the benefits of quantum confinement to enhance electron–hole spatial overlap. Zeng’s group¹¹ first reported PeLEDs fabricated with colloidal CsPbX₃ PeNCs in late 2015 with efficiencies of up to 0.12% for green devices. An improvement of 5.7% EQE for CsPbI₃ devices was achieved by Tan’s group,¹² who employed a trimethylaluminum (TMA) crosslinking method that not only helped to passivate the NC defects but also insolubilized them for subsequent solution processing. Of late, Yan et al. achieved record efficiencies of 12.9% EQE for colloidal MAPbBr₃ PeNCs through careful tuning of the charge balance at the recombination region.¹³ 

In a parallel development, the application of layered perovskites in PeLEDs was revived in the form of quasi-two-dimensional (2D) perovskites [also known as mixed Ruddlesden–Popper phase (RPP) perovskites or multi-quantum well (MQW) perovskites] that enable the excitation energy to funnel to the lowest emitting state. For instance, RPP perovskites have a general formula $(A′)mAn¯−1Bn¯X3n¯+1$ (⁠$n¯$ = 1, 2, 3, 4, …), where A′ is a large insulating organic spacer cation [e.g., 2-phenylethylammonium (PEA) and n-butylammonium (n-BA)—known as LC] and A represents a monovalent organic cation [e.g., CH₃NH₃⁺ (MA⁺) and HC(NH₂)₂⁺ (FA⁺)]. B is a divalent metal cation such as Pb and Sn, while X represents a halide anion, and $n¯$ is the number of [BX₆]⁴⁻ octahedral layers within each organic spacer. RPP is essentially an MQW system, comprising $n¯$ layers of [BX₆]⁴⁻ octahedral sheets sandwiched by LCs on either side. The LCs are held together by weak van der Waals forces between the repeating $n¯$ layers. Sargent’s group¹⁴ reported red devices with 8.8% EQEs using the (PEA)₂(CH₃NH₃)_n−1Pb_nI_3n+1; while Wei’s group¹⁵ demonstrated red devices with 11.7% EQEs with the NFPI₇ system [i.e., (NMA)₂(FA)Pb₂I₇]. The mosaic of MQW phases forms a “staircase of states” that are favorable for the excitons generated at the wider bandgap states to funnel to the lowest bandgap state for emission. These seminal works initiated a new branch in PeLEDs, and photophysical details will be discussed later. Readers interested in the broadband emission from low dimension perovskites are directed to a nice recent review by Cortecchia et al.¹⁶ 

Amplified spontaneous emission (ASE) is a distinct signature of optical gain and population inversion in a resonator-free gain medium. It refers to the amplification of spontaneously emitted photons by stimulated emission. Lasing on the other hand is strongly influenced by the cavity configuration. Unfortunately, there is some confusion in the literature with a misleading nomenclature such as “amplified stimulated emission.” Samuel et al. clearly laid down the following criteria¹⁷ for lasing: (i) narrow emission linewidth, (ii) clear emission intensity and linewidth thresholds, (iii) tunable light emission with laser cavity and gain medium, and (iv) beam-like output. It is important that “laser-like” phenomena^17,18 such as luminescent micro-cavities, waveguiding effects, and even grating resonant features are carefully checked and discounted prior to making claims about lasing.

An overview of the burgeoning perovskite optical gain field (179 papers in September 2019) is given in Fig. 1(b). Room temperature perovskite ASE, broadly tunable over the visible wavelengths (400–800 nm) from polycrystalline thin films as well as lasing from a small perovskite crystal, was first reported by Xing et al. in 2014.¹⁹ Extended wavelength tunability up to near IR (∼1 µm) was subsequently achieved with CsSnI₃.²⁰ At around the same period, Deschler et al. demonstrated optically-pumped lasing utilizing a mixed halide MAPbI_3−xCl_x perovskite layer sandwiched between a dielectric mirror and Au output coupler as a vertical cavity;²¹ while Sutherland et al. coated perovskites on silica microspheres to achieve whispering gallery mode lasing.²² Perovskite photonic crystal lasing with 2D resonators²³ as well as in 3D inverse opal template²⁴ was also demonstrated.

Given the molecular level mixing afforded by perovskite wet chemistry and vapor phase synthesis approaches, high phase purity perovskites of various morphologies and nanostructures could be fabricated with careful control of reaction conditions such as temperature, solvent, ligands, and substrates. Nanostructured perovskites (e.g., nanocrystals, nanowires/rods, nanoplates/disks, and RPP layered perovskites) are, thus, harnessed for both carrier and photon confinement in optical gain applications. Early works include CH₃NH₃PbI_3−yX_y platelets,²⁵ CH₃NH₃PbX₃ nanowires,²⁶ and CsPbX₃ nanocrystals²⁷. Polariton condensation is an exciting recent development in perovskite nanostructures,^28,29 but the detailed mechanism in these perovskite cavities is still a matter of debate. The interested reader is referred to an excellent review on this topic by Schlaus.³⁰ In the case of RPP perovskites, the first ASE work showcased the (NMA)₂(FA)Pb₂Br_yI_7−y system with tunability over 530 nm–810 nm.³¹ Leveraging the enhanced exciton and photon confinement afforded by the RPP system, Li et al. realized low threshold linearly polarized lasing from multilayered (OA)₂(MA)_n−1Pb_nBr_3n+1 RPP perovskite microplatelets.³² 

There are two electrical excitation approaches for perovskite lasing: indirect pumping with an external diode laser and direct electrical pumping. Given the low thermal conductivity of perovskites (i.e., CH₃NH₃PbI₃ ∼ 0.5 W m⁻¹ K⁻¹), thermal management is essential for both approaches. In the first approach, optical pumping of perovskite lasers with nanosecond diode lasers was demonstrated. Jia et al. demonstrated 20 nanoseconds distributed feedback perovskite lasing with 5 kW cm⁻² threshold optical pumping (2 MHz) at 160 K;³³ while Chen et al. showcased a perovskite vertical cavity surface-emitting laser (VCSEL) pumped at room temperature with continuous nanosecond pumping (5 ns, 10 Hz).³⁴ A breakthrough in continuous wave (CW) perovskite lasing was achieved by Giebink’s group in 2017, reporting a threshold excitation intensity of 17 kW cm⁻². Temperatures below CH₃NH₃PbI₃ tetragonal-to-orthorhombic phase transition of T ≈ 160 K were used to avoid lasing death in the pure tetragonal phase.³⁵ Most recently, phase stable triple cation mixed halide perovskite [Cs_0.1(MA_0.17FA_0.83)_0.9Pb_0.84(I_0.84Br_0.16)_2.68] was employed to circumvent lasing death attributed to the perovskite phase change phenomenon with the demonstration of CW ASE at low temperatures (for thermal management).³⁶ Progress in the later direct electrical excitation approach has been slow given the high current densities (kA cm⁻²) required. A step forward in this direction is the recent works on high current (pulsed) electrical excitation of MAPbI₃ PeLED³⁷ and NMA quasi-2D PeLED,¹⁸ examining the origins of efficiency roll-off (i.e., reduced EQE at higher current densities).

Despite critical advancements in this field, the stability of perovskite emitters is often treated like the elephant in the room. Currently, an in-depth understanding of this issue is still lacking, and simple devices that do not implement any stability control strategy can only last up to a few minutes under operational conditions. The PeLEDs instability goes beyond the well-known moisture-driven degradation of lead halide perovskite³⁸ because of the role played by the applied bias and the injected carriers in light-emitting devices. Recently, So et al. categorized three different sources of operational instability:³⁹ (i) ion migration, (ii) electrochemical reactions, and (iii) interfacial reactions. Although ion migration issues were discovered in the early days of this field, the presence of parasitic reactions at the electrodes poses a fundamental challenge to chemists and material scientists working in this field.⁴⁰ Among others, the possibility of reversing these reactions is a promising strategy to explore for extending the stability of these emitters.⁴¹ For a complete list of the advancements and strategies to improve the stability of PeLEDs, the reader is referred to the comprehensive account by So.³⁹ 

To illustrate the effect of the bimolecular recombination term, k₂, on η, Fig. 2(c) shows a complementary plot for fixed typical values of k₁ (=10⁷ s⁻¹) and k₃ (10⁻²⁸ cm⁶ s⁻¹), while k₂ is variable. It is clear that increasing k₂ will bring about a broadening of the PLQu peak that will extend into the lower carrier density regimes of n < 10¹⁴ cm⁻³. Figure 2(d) shows the relative peak, η, as a function of k₂ where a reduction of two orders of magnitude from the typical k₂ value (shaded vertical band) is needed to attain a near unity, η. It is, therefore, understandable why the early papers on PeLEDs based on the more bulk-like 3D MAPbI₃ yielded lower performances. From Figs. 2(a) and 2(c), reducing the trap densities (i.e., smaller k₁) and/or enhancing the bimolecular recombination rates (i.e., larger k₂) will help to increase the radiative efficiencies. Any improvement/change in k₁ has a much larger impact than on the second order process, k₂. Furthermore, it can be non-trivial to tune the perovskite stoichiometries and optimize them for desired k₂.

Within this context, the role and positioning of different 3D halide perovskites (i.e., chlorides and bromides) is similar to that of iodides. Namely, according to the Saha equation formalism for lead halide perovskites, in the relevant injection regime, all 3D perovskites behave as free carrier semiconductors.^45,46 As a general consideration, these materials are associated with higher exciton binding energies⁴⁷ and, thus, should have higher contributions from the monomolecular radiative recombination processes. However, quantum confinement plays a more decisive role in spatially confining the carriers, thereby promoting their excitonic recombination.

From a photophysical viewpoint, we can classify three broad approaches undertaken by the PeLED community to enhance the radiative efficiencies.

The earliest approach can be summarized as a “go excitonic” strategy using nanostructures. While small EBE and slow second-order electron–hole bimolecular recombination in 3D perovskites are favorable for photovoltaics, PeLEDs benefit from the opposite to ensure strong light emission. Shifting the radiative recombination from the second-to first-order removes the carrier dependence on the QY. This inherently increases the QY at low carrier densities where the dominant first-order trap-mediated recombination outcompetes the second-order radiative recombination. First-order excitonic recombination would level the playing field for efficient light emission. Lee’s group¹⁰ with their “nanocrystal pinning” is based on this approach. Essentially, some degree of quantum confinement in perovskites is also utilized to enhance their excitonic properties (i.e., increasing their EBE), thus promoting radiative excitonic recombination. Rand’s group further evolved the nano-morphology control by introducing surfactant agents that constrain and direct the growth of even smaller perovskite crystallites.⁴⁸ Further improvements require an optimal balance between the needs for increased first-order radiative recombination and preservation of the electronic conductivity. This challenge is even more evident in PeLEDs made from colloidal PeNCs. Indeed, PeNCs face challenges from insulating ligands necessary for their synthesis. Furthermore, ligands can also detach after film processing, exposing their vulnerable surfaces to charge trapping and degradation. Ligand exchange and/or some film processing (e.g., TMA) are necessary to improve the charge transport and passivate the surface defects.

Energy funneling in quasi-2D or RPP or MQW perovskites that is pioneered by Sargent’s group¹⁴ and Wei’s group¹⁵ is a promising approach to improving both the efficiency and stability of PeLEDs. It involves carrier excitation at the wider bandgap (lower n) phases. The ensuing excitons undergo rapid energy transfer (or funneling on a timescale of <0.5 ps)⁴³ to the higher n phases (narrower bandgaps) where recombination occurs. This ultrafast exciton transfer is effective in outpacing defect trapping, charge transfer at interfaces, and even Auger processes at higher carrier densities to channel the excitons to the lower bandgap phases for recombination. Indeed, due to a combination of excitonic recombination and energy funneling, QY measurements of MQW perovskite reveal high PLQY at lower carrier densities (between 10¹³–10¹⁵ cm⁻³) in contrast with 3D perovskite [Figs. 3(b)],⁴³ underpinning the basis of this approach. Using more sophisticated two-dimensional electronic spectroscopy with higher temporal resolution, Scholes’s group⁴⁹ narrowed down the energy funneling to occur within 150 fs. In addition, there is also evidence of a slower charge transfer process (tens of ns) occurring between the different phases.⁵⁰ A key challenge in the energy funneling approach is the control of the phases present in the film to limit the extent of funneling to ensure purity of the desired color.

With the advances in film synthesis and better control over their preparation conditions, the recent reports on PeLEDs with >20% EQEs leverage the trap reduction with passivating agents (e.g., TOPO,^51,52 amino acids,⁵³ and amines⁵⁴). The introduction of molecular or ionic species in perovskites is expected to provide trap passivation at the grain boundaries.^55–57 In the case of 3D perovskites, three recent papers published around the same period were based on this approach. Huang’s group⁵⁸ utilized amino acid additives to passivate their sub-micron FAPbI₃ structures (100–500 nm), which also has beneficial light out-coupling properties. Consequently, long PL lifetimes (6 µs) and 2 orders improvement in trap densities to 1.5 × 10¹³ cm⁻³ were achieved. A similar strategy was undertaken by Gao’s group who utilized amino-based molecular passivating agents to suppress the non-radiative recombination. Wei’s group⁵⁹ employed MABr as a passivation layer for CsPbBr₃ to form CsPbBr₃/MABr quasi-core/shell structures. For MQW perovskites, Di’s group pioneered the use of a perovskite-polymer bulk heterostructure configuration where poly(2-hydroxyethyl methacrylate) is mixed with the quasi-2D/3D perovskites [(NMA)₂(FA)Pb₂I₇ and (NMA)₂(FA)_m−1Pb_mI_3m+1, m ≫ 1]. Transient spectroscopy revealed PL lifetimes of 1.5 µs, and k₁ = 6.7 × 10⁵ s⁻¹ and k₂ = 3.2 × 10⁻¹¹ cm³ s⁻¹. From the photophysical perspective, quasi-2D MQW perovskites possess the unique advantage of high QYs at low carrier densities over 3D perovskites. Therefore, with effective trap suppression, Di’s group achieved the most efficient PeLEDs to date with 20.1% EQE at current densities of 0.1–1 mA cm⁻². Figure 4 shows comparisons of the calculated QYs for a 3D perovskite and a MQW perovskite with and without defect passivation using typical k values from the literature.

For optical gain applications, carrier densities >10¹⁸ cm⁻³ are needed. It has been routinely demonstrated in various perovskite structures that the buildup of the photon population by stimulated emission outcompetes the non-radiative Auger recombination. The majority of these reports [Fig. 1(b)] at both low and room temperatures are from 3D perovskites (including their nanostructures), while those from the quasi-2D systems are emerging. Nonetheless, it remains very challenging to realize room temperature lasing from the pure 2D phase (n = 1) perovskites (e.g., A₂PbX₄ systems),⁶¹ with only limited reports at low temperatures.^18,62

Despite the rosy picture of low optical gain thresholds, high Q factors, CW ASE, and lasing reported, the long-term stability and low thermal conductivity of perovskites remain major stumbling blocks in practical perovskite lasers. Saba et al. estimated that a CW pump intensity of approximately 15 kW cm⁻² (or an equivalent current density of 5.5 kA cm⁻² under electrical injection) is needed to sustain a carrier population of n ∼ 2 × 10¹⁸ cm⁻³.⁶³ With typical current densities of PeLEDs of <200 mA cm⁻², this huge current density is a major challenge for direct electrical pumping. Recent studies on intense pulsed electrical excitation (>200 A cm⁻²) on 3D and quasi-2D PeLEDs attributed Joule heating and charge imbalance as the dominant causes for the roll-off in the former 3D case,³⁷ while Auger recombination is believed to be the cause in the latter quasi-2D PeLED.¹⁸ Further photophysical studies to understand the roll-off mechanisms are urgently needed.

Apart from effective heat management in practical perovskite lasers, optimal resonator designs in high-quality single crystal thin films or nanostructures, and even photon management (e.g., high reflectivity cavity side walls) should also be explored. One could leverage perovskite’s excellent intracavity photon recycling and radiation trapping properties⁶⁴ to reabsorb the spontaneously emitted photons (i.e., increased carrier recombination lifetimes τ). For electrical pumping in a 4-level laser, the threshold power is inversely proportional to τ.⁶⁵ By ensuring that these photons remain in the laser cavity and contribute to carrier regeneration, threshold current densities can be decreased.^66–68 Further studies in this direction are warranted.

In retrospect, halide perovskites are the latest sensation for solution-processed solid-state light emitting applications. This exciting family of materials with their outstanding optoelectronic properties, rich chemistry, and vibrant photophysics provides new opportunities to break into the field long dominated by organics and semiconductor colloidal quantum dots. However, PeLEDs still face considerable hurdles that could derail their commercial aspirations (e.g., device lifetimes, I–V hysteresis, ion/halide migration, and color instability issues). Presently, the lifetimes and stability of these devices remain far from satisfactory—barely lasting 48 h of operation in air before a 50% reduction in performance in the best report.⁶⁰ 

From the photophysics, the QY of MQW perovskite maximizes over a wide range at low carrier densities, unlike 3D perovskite whose QY peaks over a narrower range at higher carrier densities, MQW perovskites, therefore, allows for more efficient PeLEDs. Furthermore, the layered quasi-2D nature of these systems is more resilient to oxygen and moisture penetration than 3D perovskites. Presently, the efficiencies of blue quasi-2D systems are found wanting, primarily facing challenges from defects and limiting the energy funneling to lower order phases. The vast library of large organic cations (LCs) for the intervening layers presents new prospects for finer control of the funneling, defect mitigation, and stability optimization. The ligands in colloidal PeNCs will continue to plague the charge transport in this class of emitters. The best possible approach is to emulate colloidal semiconductor quantum dots where colloidal PeNCs can be embedded in a protective matrix for LED-backlit displays, thus simultaneously addressing the charge injection and instability issues.

For perovskite lasing, the Holy Grail is a direct electrically driven laser. However, the jury on whether 3D or MQW perovskites are more suited for direct electrical driven lasing is still out. Nonetheless, it is clear from the recent studies that the adapted PeLED architecture must be optimized with compatible transport layers for balanced charge injections to mitigate the roll-off at high current densities. Light management strategies leveraging on perovskites’ excellent photon recycling properties to reduce the current densities is another interesting concept that can be explored for perovskite lasers.

Fundamental studies into the basic photophysics and the structure-function relations will continue to play a vital role in providing valuable foundational knowledge essential to guide material processing and device engineering efforts to optimize the material quality and device architectures. With the advances in materials synthesis and processing, the prospects for halide perovskites in solid-state light emitting applications remain optimistic. Their technological developments, particularly the stability and operational lifetime milestones over the next few years, will be extremely critical in determining whether they continue to shine in the foreseeable future. So, “Quo vadis, perovskite emitters?” (“Where are you going, perovskite emitters?”), our adventure continues…

The data that support the findings of this study are openly available in DR-NTU (Data) at https://doi.org/10.21979/N9/LCI3IF.

Financial support from the Nanyang Technological University start-up grant (No. M4080514); JSPS-NTU Joint Research Project (No. M4082176); the Ministry of Education Tier 2 grant (Nos. MOE2016-T2-1-034 and MOE2017-T2-2-002); the Agency for Science, Technology, and Research (A^*STAR) AME Individual Research Grant (No. A1883c0004); and the Singapore National Research Foundation (Program Nos. NRF-CRP14-2014-03 and NRF-NRFI-2018-04) is gratefully acknowledged.