Stable and Efficient Red Perovskite Light-Emitting Diodes Based on Ca2+-Doped CsPbI3 Nanocrystals

α-CsPbI3 nanocrystals (NCs) with poor stability prevent their wide applications in optoelectronic fields. Ca2+ (1.00 Å) as a new B-site doping ion can successfully boost CsPbI3 NC performance with both improved phase stability and optoelectronic properties. With a Ca2+/Pb2+ ratio of 0.40%, both phase and photoluminescence (PL) stability could be greatly enhanced. Facilitated by increased tolerance factor, the cubic phase of its solid film could be maintained after 58 days in ambient condition or 4 h accelerated aging process at 120°C. The PL stability of its solution could be preserved to 83% after 147 days in ambient condition. Even using UV light to accelerate aging, the T50 of PL could boost 1.8-folds as compared to CsPbI3 NCs. Because Ca2+ doping can dramatically decrease defect densities of films and reduce hole injection barriers, the red light-emitting diodes (LEDs) exhibited about triple enhancement for maximum the external quantum efficiency (EQE) up to 7.8% and 2.2 times enhancement for half-lifetime of LED up to 85 min. We believe it is promising to further explore high-quality CsPbI3 NC LEDs via a Ca2+-doping strategy.

According to the Goldschmidt tolerance factor (τ) function for perovskite (ABX 3 ), doping ions with suitable size can realize precise tuning of τ to stabilize cubic phase [23,24]. Based on this principle, A-, B-, or X-site doping can maintain α-CsPbI 3 . It should be noted that common methods to synthesize CsPbI 3 NCs need high temperature (>170°C) [6]. As a result, organic A-site ions (methylamine or formamidine) will be easily decomposed at high temperature [24,25]. Currently, B-or X-site doping is the hotspot for stabilization of α-CsPbI 3 . Due to the ionic nature of

Result and Discussion
Ca 2+ -doped CsPbI 3 NCs were synthesized by using different feed ratios of Ca(Ac) 2 /PbI 2 (0%, 15%, 25%, and 35%) via hot injection method (Experimental Section). As shown in Table S1, the real Ca 2+ /Pb 2+ ratios for Ca 2+ -doped CsPbI 3 NCs were determined by ICP-MS. According to ICP-MS results, the real Ca 2+ /Pb 2+ ratios were 0%, 0.35%, 0.40%, and 1.20% for the condition of 0%, 15%, 25%, and 35% feed ratios of Ca(Ac) 2 /PbI 2 . Such results implied that Ca 2+ was successfully doped into CsPbI 3 NCs. The Pb 2+ is located in the center of the octahedron (PbI 6 4-). The exchange and migration of the B-site ion is more difficult as compared to that of A-site and X-site ions. Therefore, B-site ions do not easily incorporate into the CsPbI 3 lattice, and the actual doping ratio of the B-site ion is much lower than the feed ratio [18,34,38,40,43]. The size of Ca 2+ (1.00 Å) is smaller than Pb 2+ (1.19 Å), and Ca 2+ doping must induce the lattice contraction, which can be confirmed by XRD. Figure 1(a) shows XRD patterns for Ca 2+ -doped CsPbI 3 NC films with different real Ca 2+ /Pb 2+ ratios. The patterns for all films match well with α-CsPbI 3 (PDF#97-018-1288), and the diffraction peaks at 14.09°a nd 28.58°correspond to the (100) and (200) planes, respectively [34]. It should be emphasized that the splitting peaks at around 14°may be related to γ-CsPbI 3 [44,45], while the characterization patterns for γ-CsPbI 3 from 25-30°cannot be observed. It is possible that a slight lattice distortion occurred due to the influence of water and oxygen during the test [20,46]. With increasing Ca 2+ /Pb 2+ ratio, no additional diffraction peaks were observed, which indicated that doping Ca 2+ did not significantly change the CsPbI 3 crystal phase. Their fine XRD patterns are shown in Figure 1(b). There is a slight shift toward a higher diffraction angle for the α-CsPbI 3 (200) plane (28.62°for 0%, 28.77°for 0.35%, 28.78°for 0.40%, and 28.82°for 1.20%), which verifies the lattice contraction resulted from the partial replacement of Pb 2+ (1.19 Å). An enhanced peak at 20°is attributed to the (110) crystal plane for CsPbI 3 . Based on the HRTEM images, the cubic NCs become truncated cubic NCs (Figure S1a-d). Therefore, (110) crystal planes are more exposed and enhance the peak at 20° [47,48]. Furthermore, the morphology and size characterization for Ca 2+ -doped CsPbI 3 NCs were studied by TEM. As shown in Figures 1(c)-1(f), with increasing Ca 2+ /Pb 2+ ratio, Ca 2+ -doped CsPbI 3 NCs maintain a cubic shape with uniform size distribution. The average size decreased from 11:11 ± 0:90 nm, 10:62 ± 0:98 nm, 10:40 ± 1:00 nm, and 9.89 ± 1.07 nm ( Figure S2), which is mainly because Ca 2+ doping influences lattice contraction or nucleation and growth processes. HRTEM images (Figures 1(g)-1(j)) reveal that Ca 2+ -doped CsPbI 3 NCs are highly crystalline with displayed lattice fringes. The lattice distances of all samples are 0.31 nm corresponding to the (200) plane of α-CsPbI 3 [49]. This is due to the limited TEM accuracy, which makes it difficult to distinguish the difference of 0.001 nm. According to the XRD test results and Scherer's formula, the lattice distances of (200) are 0.311 nm (0%), 0.310 nm (0.8%), 0.310 nm (3.1%), and 2 Research 0.309 nm (5.0%). Therefore, only XRD results can confirm that Ca 2+ is successfully doped into CsPbI 3 NCs. Furthermore, the elemental mapping images were measured by EDS ( Figure S3a-d). We chose the elemental mapping area of Ca 2+ -doped CsPbI 3 NCs in a high-angle annular dark field scanning transmission electron microscopy image (HAADF-STEM) (Figure 2(a)). Elemental mapping images of Cs, Pb, Ca, and I (Figures 2(b)-2(e)) can be observed clearly. We merged these elemental mapping images to obtain the overlapped image ( Figure 2(f)), which shows that the positions of Cs, Pb, Ca, and I are uniformly distributed in NCs. Therefore, these results can directly demonstrate that Ca 2+ is doped in CsPbI 3 NCs.
The optical properties of Ca 2+ -doped CsPbI 3 NCs are shown in Figure 3. The UV-vis absorbance and PL spectra of Ca 2+ -doped CsPbI 3 NCs are compared in Figure 3(a) and Figure S4a-b. With the ratio of Ca/Pb increasing, both of their UV-vis absorption and PL peaks appear as blue shifts due to their lattice contraction, size decrease, and Ca 2+ orbitals to influence the electronic structure of NCs [34,40,50,51] (Figure S4a-b and Figure S2). Figure 3(b) summarizes the UV-vis absorption peaks, PL peaks, and PLQY evolution with different ratios of Ca/Pb. The UVvis absorption peaks exhibit a blue shift from 666 nm to 665, 664, and 661 nm with the ratio of Ca/Pb increasing from 0% to 0.35%, 0.40%, and 1.20%, respectively. Simultaneously, their PL peaks also exhibit a blue shift from 682 nm to 676 nm with increase of the ratio of Ca/Pb. Furthermore, benefitting from Ca 2+ doping, their PLQYs can be improved. With increasing Ca/Pb ratios from 0% to 0.40%, their PLQYs increased from 89% (±1.5%) to 93% (±1.3%), whereas further increasing the Ca/Pb ratio to 1.20% led PLQY lower to 91% (±1.4%). This is due to that the smaller NCs (Ca/Pb = 1:20%) with larger surface-tovolume ratio must have more surface defects. In addition, Ca 2+ cannot be as emission centers. Therefore, excess Ca 2+ doping may decrease PLQY. Figure 2(c) shows the decay curves of Ca 2+ -doped CsPbI 3 NCs. All of the decay curves can be fitted by a double-exponential (Table S2). The   In addition to boosting their PLQYs, the increase in τ for Ca 2+ doping in CsPbI 3 can much improve their stability. We measured the stability of their solutions and solid films. Firstly, all samples were stored in the conditions of 20 ± 5°C and 40−50% humidity. Figure 4(a) shows PL intensity evolution with storage time. After a 147-day storage, PL intensity of CsPbI 3 NCs decreased to 36% of the initial one, whereas PL stabilities of Ca 2+ -doped CsPbI 3 NCs exhibited much improvement. After 147-day storage, an obvious color bleaching was observed for the CsPbI 3 NC solution, while the red color of Ca 2+ -doped CsPbI 3 NC solutions can be maintained (Figure 4(b)). PL intensities of Ca 2+doped CsPbI 3 NC solutions can be preserved to 80% (0.35%), 83% (0.40%), and 76% (1.20%) of the initial intensities after 147-day storage (Figure 4(a) and Figure S5). We further studied their morphology changes by TEM. The size of CsPbI 3 NCs became nonuniform and noncubic shape and tended to be aggregated, which must lead to their PL decrease (Figure 4(c) and Figure S5). In contrast, cubic shape and uniform size distribution of Ca 2+ -doped CsPbI 3 NCs can be kept with relatively high PL performance (Figure 4(c) and Figure S5). Secondly, Ca 2+doped CsPbI 3 NCs exhibited improved stability against to UV. All of the samples were placed under 365 nm UV light (8 W), and their PL intensities decayed with UV irradiation time increasing. We periodically measured their PL spectra, which were used to calculate their half-lifetimes (T 50 ) under UV light ( Figure S6 and Figure S7a). On the basis of these data, T 50 were 52 min (0%), 74 min (0.35%), 92 min (0.40%), and 85 min (1.20%), separately. About 1.8 times enhancement of UV stability was observed for the 0.40% Ca/Pb ratio. Further increasing UV irradiation time to 100 min, it was hard to observe the red color for CsPbI 3 NCs, while Ca 2+ -doped CsPbI 3 NCs could still emit red PL ( Figure S7b). As a result, Ca 2+ -doped CsPbI 3 NCs exhibit enhanced UV stability as compared to CsPbI 3 NCs.
According to the above results, Ca 2+ -doped CsPbI 3 NC solutions exhibited improved stability in ambient condition and UV light, whereas the application of these NCs must form solid films. Therefore, we further studied their stability in solid films, such as in ambient condition (40−50% RH at 20 ± 5°C) and heating condition (120°C). As shown in Figure 5(a), α-CsPbI 3 (cubic phase) film gradually converted to δ-CsPbI 3 (orthorhombic phase) film after 28  . Such results demonstrate that the phase stability of CsPbI 3 can be enhanced in ambient condition facilitated by Ca 2+ doping. Additionally, we verified their thermal stability in the solid state. All the films were placed on a hot plate at 120°C, and we periodically measured their XRD patterns. The CsPbI 3 film exhibited the poorest phase stability, and its phase gradually converted to the orthorhombic phase after 120°C heating for 2 h ( Figure S8). On the basis of previous reports, the small size effect can reduce CsPbI 3 lattice distortion to partly prevent phase transition [39,52]. As an ionic nature of CsPbI 3 , all ion migrations can be accelerated at high temperature. As a result, ion migration and crystal fusion easily occur in the solid state, which leads to increasing the size of CsPbI 3 NCs. The enlarged size of CsPbI 3 must induce phase transition [14,53,54]. Benefitting from small-sized Ca 2+ doping, the increase in τ should improve the phase stability of CsPbI 3 . With increase of the Ca/Pb ratio from 0% to 1.20%, their cubic phase could be maintained at least for 4 h at 120°C ( Figure S8). Therefore, the thermal stability of CsPbI 3 films can be enhanced via Ca 2+ doping.
To identify the positive effect of Ca 2+ -doped CsPbI 3 NCs for LEDs, we used Ca 2+ -doped CsPbI 3 NC films as emitting layers to fabricate red LEDs. Figure 6

Research
and LiF are the electron-transport and electron-injection layer, respectively; and Al is the cathode. Figure 6(b) presents the current density as functions of voltage. The LED with Ca 2+ -doped CsPbI 3 NCs (Ca/Pb = 0:40%) exhibited obviously enhanced current density, which implied the decreasing of defect density as well as the improvement for carrier transport efficiency.
Firstly, hole-only devices with a structure of ITO/PED-OT:PSS/CsPbI 3 NCs or Ca 2+ -doped CsPbI 3 NCs/MoO 3 /Al ( Figure S10) were used to quantitatively measure the defect density of NCs films. The defect density (N t ) is calculated according to the following equation: where ε 0 and ε are the vacuum dielectric constant and the relative dielectric constant, respectively; V TFL is the trapfilled limit voltage; L is the thickness of the NC film; and e is the elementary electronic charge. By assuming that ε = where ε 0 is the vacuum permittivity; ε r is the average relative dielectric constant of CsPbI 3 (ε r ≈ 6:32); L is the thickness of the perovskite film; and J, μ, and V are the measured current density, carrier migration rates, and applied voltage, respectively [34]. The hole mobilities of CsPbI 3 and Ca 2+ -doped CsPbI 3 NC films were 2:8 × 10 −8 and 6:0 × 10 −8 cm 2 V −1 s −1 , respectively. With the Ca 2+ doping, the hole mobility increased, which enhanced hole transport efficiency.
Then, the improvement for carrier transport efficiency can be confirmed by ultraviolet photoelectron spectra (UPS). Combining with their optical bandgaps (E g ) for Ca 2+ -doped CsPbI 3 NC film ( Figure S11a) and their UPS spectra of CsPbI 3 NCs (Figure S11b), their conduction band minimum (CBM), valence band maximum (VBM), and Fermi level can be obtained. For CsPbI 3 NCs, its CBM, VBM, and Fermi level were -3.79 eV, -5.60 eV, and -3.93 eV, respectively. Benefitting from Ca 2+ doping (0.40%), its CBM, VBM, and Fermi level shifted to -3.73 eV, -5.55 eV, and -4.05 eV, respectively. The higher VBM reduces the energy barrier between PEDOT:PSS and the emission layer to boost hole transport efficiency. Additionally, the lower Fermi level reveals the transformation of CsPbI 3 NCs from n-type to a more nearly ambipolar nature facilitated by Ca 2+ doping, which also enhances hole transport efficiency [34,37].
Benefitting from the enhanced current density, the Ca 2+doped CsPbI 3 NC LEDs exhibited stronger brightness than CsPbI 3 NC LEDs in the whole driving voltage range (Figure 6(c) and Figure S12 7 Research (7.2 V), which was doubled as compared to the nondoping one. In addition, the maximum and average EQEs for Ca 2+ -doped CsPbI 3 NC (0.40%) LEDs were 7.8% and 6.3%, both of which enhanced to about 3 times as compared to the nondoping one, respectively (Figure 6(d) and Figure S13). Therefore, our work using Ca 2+ as a new B-site doping ion to boost CsPbI 3 NC LEDs is quite promising (Table S3). Their electroluminescent (EL) spectra exhibit stable and sharp peaks at 683 nm with a full width at half-maximum (FWHM) of 35 nm on various driving voltages (Figure 6(e) and Figure S14). A bright red emission could be observed at 5.0 V (inset in Figure 6(e)), and its Commission Internationale del'Eclairage (CIE) color coordinate was (0.72, 0.27) ( Figure S15). Furthermore, the operational stability of the LED was evaluated at a constant current density of 5.0 mA/cm 2 . The luminance of CsPbI 3 NC LEDs decreases to half at 39 min, while the half-lifetime of Ca 2+ -doped CsPbI 3 NC (0.40%) LEDs could significantly increase to 85 min. A 2.2-fold improvement for the halflifetime of LED confirmed that Ca 2+ doping is a powerful strategy to promote both stability of CsPbI 3 NCs and their LEDs. In a word, the boosting of both the efficiency and the stability of LEDs is mainly attributed to a decrease in defect density, improvement on hole injection efficiency, and reduction in phase transition.

Conclusion
In summary, we explored stable and high-performance CsPbI 3 NCs based on Ca 2+ (1.00 Å) doping. With a Ca 2+ / Pb 2+ ratio of 0.40%, the phase stability could be greatly enhanced. Ca 2+ -doped CsPbI 3 NC solid films could maintain the cubic phase after 58-day storage in ambient condition or 4 h accelerated aging process at 120°C. Moreover, the PL stability could be also improved. The PL intensity of Ca 2+ -doped CsPbI 3 NC solutions could be preserved to 83% after 147-day storage in ambient condition. Even using UV light to accelerate aging, the T 50 of PL could be boosted 1.8-folds as compared to that of CsPbI 3 NCs. Red LED based on Ca 2+ -doped CsPbI 3 NCs exhibited about triple enhancement for maximum EQE up to 7.8% and 2.2 times enhancement for half-lifetime of LED up to 85 min. These were mainly attributed to the decreased defect densities of films and reduced hole injection barrier facilitated by Ca 2+ doping. Ca 2+ as a new B-site doping ion can efficiently boost both stability and performance for CsPbI 3 NC LEDs, which has the potential to promote the development of CsPbI 3 NC LEDs.

4.2.
Synthesis of Cs-OA. 0.39 g Cs 2 CO 3 , 18 mL ODE, and 2.0 mL OA were mixed into a 100 mL three-neck flask and dried in vacuum at 120°C for 1 h. Then, the mixture was heated to 150°C under N 2 until the Cs 2 CO 3 powders were completely dissolved to form a transparent solution. The solution was cooled down to room temperature via icewater bath and preheated to 110°C before use.

4.3.
Synthesis of OLA-HI. 20 mL OLA and 2 mL HI were mixed into a 100 mL three-neck flask. Then, the solution was heated to 120°C for 2 h under vacuum to remove the water. Then, the solution was cooled down to 60°С to obtain the OLA-HI solution and preheated to 90°C before use.

Synthesis of Ca 2+ -Doped CsPbI 3 NCs.
In a typical synthesis, PbI 2 (0.4 mmol), Ca(Ac) 2 (0, 0.06, 0.10, and 0.14 mmol), and 10 mL ODE were mixed into a 100 mL three-neck flask. The mixture was degassed and dried in vacuum for 1 h at 120°C. Then, 1.0 mL OLA, OA, and preheated OLA-HI were injected into the reaction flask, separately. The mixed solution became clear and was kept under vacuum for 30 min at 120°C. Finally, the temperature was increased to 260°C, and 1.0 mL Cs-OA solution was injected immediately. After 1 min, the reaction flask was placed into an ice-water bath to stop the reaction.

4.5.
Purification of NCs. The as-prepared NC solution was mixed with the same volume of MeOAc and was centrifuged at 15000 rpm for 5 min at 19°C to remove the supernatant. The precipitate was redispersed into toluene. Then, the NC toluene solution was mixed with the same volume of MeOAc and was centrifuged at 15000 rpm for 5 min at 19°C. After removing the supernatant, the precipitate was dispersed into toluene. Finally, the redispersed NC solution was centrifuged at 15000 rpm for 5 min at 19°C to discard the precipitate, and the supernatant was preserved for characterization and device fabrication. 4.6. Fabrication of LED Devices. Indium tin oxide-(ITO-) coated glass substrates were cleaned by ultrasonic cleaning for 15 min in acetone, ethanol, and deionized water, separately, and dried with nitrogen flow. The clean ITO glasses were treated by UV-ozone for 15 min. Then, PEDOT:PSS solutions were spin-coated onto the ITO substrates at 3000 rpm for 30 s and annealed at 120°C for 15 min in air. NCs were spin-coated onto the PEDOT:PSS layer at 1500 rpm for 30 s. This process was repeated five times. Finally, 40 nm of the TPBi layer, 1 nm of the LiF layer, and 100 nm of Al electrode were deposited in sequence by a thermal evaporation system under a high vacuum (∼2 × 10 −4 Pa). The active area of the devices was 10 mm 2 as defined by the overlapping area of the ITO and Al electrodes. Research spectra were collected by a RF6000 spectrofluorometer with an excitation wavelength of 500 nm. The PL lifetimes of NCs were measured by a FLS920 fluorescence spectrometer with a pulse laser at 375 nm. The chemical compositions were measured by a PerkinElmer NexION 2000 inductively coupled plasma mass spectrometry (ICP-MS). X-ray diffraction (XRD) data were collected by a Bruker D8 Advance X-ray powder diffractometer with Cu Kα radiation (λ = 0:154 nm). Photoluminescence fluorescence quantum yield (PLQY), which is defined as the ratio of emitted photons to absorbed ones, was determined by a FLS920 fluorescence spectrometer equipped with an integrating sphere. The morphology and size of NCs were confirmed by transmission electron microscope (TEM) (Hitachi, HT7700), high-resolution TEM (HRTEM) (Talos, F200X), and energy-dispersive X-ray spectroscopy (EDS). The electroluminescent (EL) spectra and luminance-(L-) current density-(J-) voltage (V) characteristics were collected by using a Keithley 2400 source and PR-655 spectra scan spectrophotometer (Photo Research). The characterization for LED devices was measured at room temperature in air.

Data Availability
All data needed to evaluate the conclusions in the paper are presented in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Conflicts of Interest
The authors declare no conflict of interest.

Authors' Contributions
W. Shen and J. Zhang contributed equally to this work and are listed as co-first authors. S. Chen and W. Huang conceived and designed the research and revised the manuscript. W. Shen and J. Zhang performed the synthesis, data analysis, and article writing. R. Dong, Y. Chen, and S. Chen prepared and characterized the LED. L. Yang, Z. Su, and Y. Dai acquired the stability data. L. Liu and K. Cao assisted in all data analysis and interpretation. All authors participated in drafting the manuscript and approved the final version.