The metal halide perovskites have a general chemical formula of ABX₃: A is a monovalent organic or inorganic cation, e.g., CH₃NH₃⁺(MA⁺), CH(NH₂)₂⁺(FA⁺), Cs⁺, Rb⁺; B is a divalent metal cation, e.g., Pb²⁺, Sn²⁺; X is a halide anion, e.g., Cl^–, Br^–, I^–. The perovskites have attracted tremendous interests due to their excellent properties such as easy solution-based fabrication process, high absorption coefficient, tunable energy bandgaps, narrow emission linewidths, and long exciton diffusion length^[1-3]. These outstanding properties lead perovskites to be used in a wide range of optoelectronic applications, including solar cells, light-emitting diodes (LEDs), and photodetectors. In this minireview, we will focus on perovskite light-emitting diodes (Pero-LEDs), in which perovskite is used as an emitting layer. In the past few years, the Pero-LEDs have shown great potential in applications like lighting, displays, and light communication.

In 2014, the first bright and room-temperature operative Pero-LEDs were fabricated using MAPbI_3–xCl_x (near-infrared) and MAPbBr₃ (green) as emitting layers, which obtained an external quantum efficiencies (EQEs) of 0.76% and 0.1%^[1], respectively. Then, several strategies were demonstrated to enhance the efficiency of Pero-LEDs, including compositional engineering, morphology controlling, and device structure engineering^[2-7]. Recently, the EQEs of green, visible red, and near-infrared emitting Pero-LEDs have all exceeded 20%, reaching 20.3%^[7], 21.3%^[8], and 21.6%^[6], respectively. However, the efficiency growth of blue Pero-LEDs lags behind their counterparts, as shown in Table 1, the EQE of deep-blue (440–470 nm), pure blue (470–480 nm), and blue-green (480–510 nm) emitting Pero-LEDs is lower than 3%, 6%, and 11% respectively^[9]. Blue is one of the three primary colors, and the blue LEDs play an essential role in displays and white lighting^[10]. In other words, developing high-performance blue Pero-LEDs takes a lot of effort.

In this review, the recent progress of blue Pero-LEDs, especially for the strategies of preparing the blue perovskite emitting layer, are summarized. We categorize the strategies into two: compositional engineering and size controlling of the emitting units. Compositional engineering, including A-, B-, and X-site doping and lead-free perovskites, is the simplest and most effective strategy to tune the bandgap (emitting color) of perovskites^{[11, 12]}. Size control of the emitting units could be realized directly by synthesizing nanoplates (NPs) and quantum dots (QDs)^{[13, 14]}. It could also be realized by forming low-dimensional (2D and quasi-2D) perovskites (forming small emitting perovskite units by cutting the 3D emitting center with the long-chain organic ligands)^{[15, 16]} or reducing the thickness of perovskite films to the nanoscale^[17]. Once the size of perovskite emitting crystals were reduced to the nanoscale, the bandgap of perovskites will be enlarged owing to the quantum-confined effect, and the emitting color will be blue-shift^[13]. Here the advantages and disadvantages of both methods to prepare blue-emitting perovskites will be discussed, and we will also give a perspective of preparing high-performance perovskites emitting layers and the corresponding Pero-LEDs.

To prepare the blue-emitting perovskites, researchers have proposed a mass of methods. Fundamentally, these methods could be categorized into two basic aspects. One is enlarging the energy bandgap by replacing the elements of perovskite crystals. Due to the excellent tolerance of perovskite crystals, lots of elements could be used to form the stable perovskite crystals^{[18, 19]}. Recently, Bartel et al.^[20] reported a new tolerance factor to predict the stability of perovskite oxides and halides. They not only correctly predicted 92% of compounds as perovskite or non-perovskite for an experimental dataset of 576 ABX₃ materials, but also they generalized outside the training set for 1034 experimentally realized single and double perovskites (A₂BB’X₆) to identify the stability of 23 314 new double perovskites. According to their research, abundant stable perovskites could emit blue have not been studied by researchers. However, one of the fatal drawbacks is that non-homogeneous phase distribution and phases segregation would be likely inevitable if the components of perovskites are partly replaced^[11]. The other strategy is to reduce the size of perovskite crystals to create the quantum confined effect. When the size was reduced to the nanoscale, notably less than the exciton Bohr radius, the energy bandgap will increase and induce emitting peak shifting to the blue region, and this theory is widely applied in fabricating of the traditional blue-emitting quantum dot (QD)-LEDs^[21]. Nevertheless, the disadvantages of this strategy are also apparent. Specifically, the electrical conductivity will decrease dramatically after adding the insulating organic ligands^{[13, 22]}, and the size of crystals is hard to control and tend to be inhomogeneous when the crystals are synthesized to the nanoscale^{[13, 23]}.

Generally, according to the tight-binding approximation, the bandgap of crystals will get widened if the elements are replaced by the ones in the smaller period of the same family^[24]. Hence, the mixing of halide anions was firstly applied to prepare blue-emitting perovskites. In 2015, Kumawat et al.^[11] firstly reported the blue emissive Pero-LEDs based on MAPbBr_1.08Cl_1.92. The bandgap changes when varying the ratio of Br^– and Cl^–. Fig. 1(a) shows that the absorption moves to higher energy with the increasing of chlorine concentration, indicating that the bandgap increases with the growth of chlorine fraction. Furthermore, the photoluminescence (PL) peaks of the corresponding perovskites cover the range from 408 to 535 nm, as shown in Fig. 1(b). Sadhanala et al. conducted similar research^[12], and the bandgap tuning range of the MAPb(Br_xCl_1–x)₃ (0 ≤ x ≤ 1) perovskites is ~3.1–2.3 eV. Fig. 1(c) shows that the electroluminescence (EL) peaks of Pero-LEDs based on such perovskites could be tuned from 427 to 570 nm by controlling the chloride content. At present, halogen doping is widely used as the most effective and simplest method for preparing blue-emitting perovskites.

However, using the simple halogen doping method to prepare the Pero-LEDs, the problem of phase segregation will be inevitable when the devices operated at high voltage. Hence, researchers developed alternative methods by A- or B- site doping to control the bandgap of perovskites and stabilize the crystalline phases. Todorovic´ et al.^[25] reported a tunable and stable electroluminescence perovskite enabled by Rb doping of CsPbBr₃ nanocrystals (NCs), which could obtain tunable emission from 460 to 500 nm. Impressively, unlike the halide mixing, the Pero-LEDs based on Rb-doped perovskites achieved stable emission peaks at 464 and 490 nm with EQE_max of 0.11% and 0.87% respectively. The Rb-doped quasi-2D perovskites were also demonstrated by Jiang et al.^[26]. As shown in Fig 1(d), the PL peaks of PEA₂(Rb_xCs_1–x)₂Pb₃Br₁₀ could be tuned from 450 to 510 nm by controlling the ratio of Rb⁺ and Cs⁺. Compared to the halogen-doped perovskites, the Rb-doped perovskites exhibit excellent spectral stability, in which both the PL peak position and the full width of half maximum (FWHM) changed negligibly after 4 h of annealing (Fig. 1(e)). And the corresponding Pero-LEDs obtained spectrally stable emission at 475 nm (Fig. 1(f)) and 490 nm based on PEA₂(Rb_0.6Cs_0.4)₂Pb₃Br₁₀ and PEA₂(Rb_0.4Cs_0.6)₂Pb₃Br₁₀, respectively.

Compared to the regulation of A- and X-site of perovskite crystals, B-site doping, and lead-free perovskites could own a stronger ability to adjust the bandgap and achieve deep-blue emission. Leng et al.^[27] reported a lead-free perovskite of Cs₃Bi₂Br₉ QDs, showing a deep-blue emission at 410 nm with a photoluminescence quantum yield (PLQY) up to 19.4%. Moreover, the emission range could be adjusted from 393 to 545 nm by incorporating with the halogen-doping method. Furthermore, they applied Cl-passivation to boost the PLQY of MA₃Bi₂Br₉ QDs to 54.1% at the wavelength of 422 nm in their later work. Tan et al.^[28] reported a bismuth (Bi)-doped lead-free perovskite of Cs₂SnCl₆ crystal with the outstanding anti-water stability, which could preserve 97.1% of the initial PL intensity after 120 min soaking. Furthermore, the optimized perovskite exhibits a deep-blue emission at 455 nm and a high PLQY close to 80%. Although such high PLQYs of lead-free deep blue-emitting perovskites have reported, there are still lacking studies of their application on LED devices, due to the poor-quality morphology of the perovskite film.

According to the summaries in Table 1, it is obvious that few papers have reported high EQE by using the strategy of compositional engineering, even though the PL emission of perovskites could easily be adjusted by this method. In other words, it is hard to fabricate the spectrally stable and high efficiency deep blue-emitting Pero-LEDs only by compositional engineering.

As the particle size decreases to the nanoscale, the continuous energy levels near the Fermi energy level will be divided^[24]. Hence, reducing the size of perovskite crystals in some dimensionality should be an efficient approach to enlarge the bandgap of perovskites. Recently, 2D/quasi-2D perovskites and the pre-synthesis NPs, NCs, and QDs are widely used to prepare blue-emitting perovskites.

The 2D/quasi-2D perovskites could have a general formula of L₂(ABX₃)_n–1BX₄, where L is a monovalent long alkyl chain, and n is the number of stacking perovskite units. The bandgap of perovskites could be tuned by controlling the n values. Liang et al.^[15] reported the first Pero-LED based on 2D perovskite, using (PEA)₂PbBr₄ (PEA⁺ = C₆H₅CH₂CH₂NH₃⁺) as emitting layer (Fig 2(a)), which obtained an emission at 410 nm (Fig. 2(b)) and an EQE_max of 0.04%. By controlling the n values, Wang et al.^[37] reported a dual emission (as shown in Fig. 2(c), 473 and 485 nm) Pero-LEDs based on (EA)₂MA_n–1Pb_nBr_3n+1, which achieved an EQE_max of 2.6%. However, it is hard to control the n values precisely, and multiple n values always exist in the quasi-2D perovskite crystals. Chen et al.^[29] studied the charge-transfer and energy-transfer (Fig. 2(d)) among crystals with different n values in quasi-2D Pero-LEDs. According to their research, four distinct emissions at 425, 452, 452, and 505 nm corresponding to n = 1, 2, 3, and 4 of the quasi-2D perovskite of PA₂(CsPbBr₃)_n–1PbBr₄ (here, PA is propylammonium) were existing, and the main emission peak of 505 nm was due to the charge-transfer. After optimization, a cyan Pero-LED emitting at 505 nm (Fig. 2(e)) and with an EQE_max of 3.6% was obtained. To control the formation of quasi-2D perovskites with a desired n values, Xing et al.^[30] demonstrated a strategy replacing the long ligands (PEA) with shorter ones (iso-propylammonium, IPA) to reduce van der Waals interactions, which successfully slowed the formation of n = 1 phase and improved the monodispersity of n = 2, 3, 4 phases. With this strategy, the as-synthesized quasi-2D perovskite films exhibit a single emission peak and color-stable blue emission at 477 nm. Moreover, the as-fabricated sky-blue Pero-LEDs (emission at 490 nm, Fig. 2(f)) obtained a maximum luminance (Lv_max) of 2480 cd/m² and EQE_max of 1.5%. To further stabilize the quasi-2D perovskite crystals and block the carrier diffusion in a bias condition, Shang et al.^[31] introduced the inorganic crystalline Cs₄PbBr₆ to homogeneously surround the quasi-2D perovskite (Fig. 2(g)). The Pero-LEDs based on such a structure achieved a Lv_max of 3259 cd/m² and an EQE_max of 4.51% with emission at 500 nm. Besides, the performance of the operational stability of devices was enhanced dramatically. As shown in Fig. 2(h), the lifetime was increased to more than 1 h, and the EL spectra curves of 0 and 12 h are almost completely unchanged.

Although some progress has been made in phases stability and performance enhancement of Pero-LEDs based on quasi-2D perovskites, the efficient pure and deep blue-emitting devices are still difficult to fabricate owning to the poor electrical conductivity caused by the excess organic ligands adding. To overcome this obstacle, the as-synthesized QDs, NPs and ultra-thin quasi-2D perovskites are reported. With these strategies, the size of the perovskite emitting units could also be reduced to the nanoscale, and the bandgap will be enlarged due to the quantum confined effect. In 2015, Song et al.^[11] firstly reported the Pero-LEDs based on all-inorganic QDs of CsPbBr₃, and the emission peaks were successfully shifted from 470 to 515 nm by controlling the size of QDs (Fig. 3(a)). The particle size control around from 2 to 8 nm was realized by changing the reaction temperature, the reaction temperature was 140, 155, 170, and 185 °C, respectively. Besides, the NPs of CsPbBr₃ was reported by Yang et al.^[32] and Bohn et al.^[14], and the emission covered the range of 442–459 nm (Fig. 3(b)) and 432–520 nm (Fig. 3(c)) respectively. Moreover, deep blue-emitting Pero-LEDs based on CsPbBr₃ NPs were fabricated and obtained an EQE_max of 0.057% with emission at 464 nm. Recently, an ultra-thin quasi-2D perovskite was reported by Liu et al.^[17], the thickness was reduced to ~ 9 nm (as shown in the cross-sectional STEM-HAADF image, Fig. 3(d)). The corresponding Pero-LEDs achieved an EQE_max of 9.5% (Fig. 3(e)) at a luminance of 54 cd/m² with emission at 483 nm (Fig 3(f)). This impressive growth of EQE was achieved mainly by the reduction of the insulating organic ligands, and the blue-shift emission was due to the effective quantum-confined effects caused by the ultra-thin thickness and quasi-2D structure of perovskite.

However, there are still many problems for QDs, NPs, and ultra-thin quasi-2D perovskites, such as current leakage caused by the low coverage of the emitting layer. To fabricate efficient and color stable blue-emitting Pero-LEDs, we should combine the several methods of preparing blue-emitting perovskites. Firstly, the halogen doped method could be used in the synthesis of QDs and NPs to improve the ability of bandgap adjustment. Song et al.^[13] demonstrated that the CsPbBr₃ QDs could obtain an emission from 470 to 515 nm by controlling the size of QDs, while the CsPb(Br_xCl_1–x)₃ QDs could emission from 420 to 515 nm (Fig. 4(a)). And A-site doping in the synthesis of QDs was reported by Meng et al.^[33], the Pero-LEDs based on (Rb_0.33Cs_0.67)_0.42FA_0.58PbBr₃ achieved a Lv_max of ~ 1000 cd/m² and an EQE_max of 3.6% with emission at 502 nm (Fig. 4(b)). And a deep blue-emitting Pero-LED (emission at 466 nm, Fig. 4(c)) based on (Rb_0.33Cs_0.67)_0.42FA_0.58PbBr_1.75Cl_1.25 was also fabricated, and an EQE_max of 0.61% was obtained. Recently, Wang et al.^[9] reported a blue-emitting Pero-LED (emission at 485 nm) with a recorded Lv_max of 9040 cd/m² (Fig. 4(d)) and a recorded EQE_max of 11% (Fig. 4(e)), which based on the perovskite of CsPbBr₃:PEACl:2%YCl₃ prepared by halide mixing and B-site doping in the quasi-2D perovskite. Besides, an impressive excellent EL spectrum stability was also obtained. As shown in Fig. 4(f), there was almost no shift of EL peaks with a continuous bias of 3.2 V for 120 min.

The summaries in Table 1 show that researchers have started to combine the strategies of compositional engineering and size controlling of emitting units. Pero-LEDs with high EQEs were fabricated based on the perovskites prepared by the comprehensive strategies. However, we believe that a more effective combination of the strategies could be carried out to fabricate the blue-emitting Pero-LEDs with spectrally stable emission and high EQE.

Although great progress has been achieved in the blue-emitting Pero-LEDs fabrication in the past few years, the high-efficiency deep-blue-emitting (440–470 nm) devices are still lacking in demonstrations. We list and draw an EQE evolution curve in Fig. 5 using the reported EQEs of blue-emitting Pero-LEDs, most EQEs of Pero-LEDs with the emission peak at less than 480 nm are no more than 3%, and that ones with the emission peak at less than 440 nm are no more than 1%. Besides, improving the EL spectra stability and operational stability of the blue-emitting Pero-LEDs are still big challenges. To overcome these obstacles, we need to utilize the two strategies (compositional engineering and size controlling of the emitting units) comprehensively and optimize the device structure of Pero-LEDs.

Considering the side effects of using a single method to tune the emission of Pero-LEDs to deep-blue gamut, the strategies using several methods simultaneously were studied. Shang et al.^[31] and Zou et al.^[34] reported a similar strategy using the inorganic large bandgap Cs₄PbBr₆ to surrounding the quasi-2D perovskites. With this strategy, they successfully obtained the spectrally stable emission of Pero-LEDs. However, the emission colors are not blue enough, which are 500 and 484 nm respectively. If more methods such as halogen-doped and A-site doped are applied, the spectra stable emission may obtain a blue-shift. Tan et al.^[35] demonstrated a stable ultra-pure blue (468 nm) emitting of Pero-LEDs based on the perovskite of 2-phenoxyethylamine-passivated CsPbBr_xCl_3−x, and a square-wave alternating voltage was applied at the same time. And some ions doping such as potassium ion (K⁺), Rb⁺, manganese ion (Mn²⁺), and nickel ion (Ni²⁺) could effectively passivate the perovskite crystals and stable the emission spectra of Pero-LEDs.

To fabricate the high-efficiency and operational stable blue-emitting Pero-LEDs, only improving the performances of perovskites is not enough. It is also critical to improve the charge injection ability and control the injection balance of electrons and holes by optimizing the structure of Pero-LEDs. Gangishetty et al.^[22] demonstrated a transport layer structure using poly[(9,9-dioctylfluorenyl-2,7diyl)-co-(4,4’-(N-(4-sec-butylphenyl) diphenylamine)] (TFB) and Nafion perfluorinated ionomer (PFI) as hole transport layer instead of traditional NiO_x to maintain robust nanocrystal emission, and an enhanced EQE_max of 0.5% with emission at 469 nm was achieved. Ren et al.^[36] demonstrated an efficient hole transport bi-layer structure composed of poly(sodium-4-styrene sulfonate) (PSSNa) and NiO_x, which could simultaneously inhibit the nonradiative decays between NiO_x and perovskite films by reducing NiO_x surface defects and improve quasi-2D perovskite thin film quality by minimizing its pin-holes and reducing the film roughness. With this architecture, an EQE_max of 1.45% and a remarkable luminance of 4359 cd/m² were obtained.

We have summarized the strategies of preparing blue-emitting perovskites and discussed their advantages and disadvantages. The compositional engineering strategies could precisely adjust the bandgap of perovskites. However, the drawbacks (e.g., phase segregation and poor quality of film morphology) could not be ignored. More environmentally stable blue-emitting perovskites could be obtained by size controlling of perovskite crystals. However, the size is hard to be controlled precisely and tending to be inhomogeneous. Hence, the strategies should be utilized comprehensively to improve the performances of blue-emitting perovskites. Moreover, we also discussed the challenges of fabricating high-efficiency stable blue-emitting Pero-LEDs. The most urgent challenge is increasing the EQE and keeping the deep blue emission at the same time. Researchers should then pay more attention to the spectral stability and the long operational lifetime of blue-emitting Pero-LEDs. We believe that the high-efficiency stable blue-emitting Pero-LEDs will be fabricated by improving the performances of perovskites and optimizing the architecture of Pero-LEDs.

This work was supported by the National Natural Science Foundation of China (Nos. 51802102, 21805101 and 51902110).