Ambient-Processed, Additive-Assisted CsPbBr 3 Perovskite Light-Emitting Diodes with Colloidal NiO x Nanoparticles for E ﬃ cient Hole Transporting

: In this study, the electrically driven perovskite light-emitting diodes (PeLEDs) were investigated by hybridizing the organic polyethylene oxide, 1,3,5-tris (N-phenylbenzimiazole-2-yl) benzene (TPBi), and bis(3,5-diﬂuoro-2-(2-pyridyl)phenyl-(2-carboxypyridyl) iridium III (FIrpic) with CsPbBr 3 in the emission layer and adopting the colloidal NiO x nanoparticle (NP) hole transport layer. The synthesized NiO x NPs, having an average size of ~5 nm, can be spin-coated to become a smooth and close-packed ﬁlm on the indium–tin–oxide anode. The NiO x NP layer possesses an overall transmittance of ~80% at 520 nm, which is about the peak position of electroluminescence (EL) spectra of CsPbBr 3 emission layer. The coating procedures of NiO x NP and CsPbBr 3 layers were carried out in ambient air. The novel PeLED turned on at 2.4 V and emitted bright EL of 4456 cd / m 2 at 7 V, indicating the remarkable nonradiative-related defect elimination by organic additive addition and signiﬁcant charge balance achieved by the NiO x NP layer.

Despite the ease of ion exchanges for adjusting the optical and electrical properties, all inorganic halide perovskites have been firmly demonstrated to be superior to their organic-inorganic hybrid counterparts in environmental and thermal stabilities [9]. As a result, the main stream of related research has been focused on the improvement of inorganic lead-free perovskites in recent few years [10].
As for the application of LEDs, there are many reports about the electrically driven CsPbBr 3 perovskite LEDs (PeLEDs), which merely shows moderate performance, despite the outstanding stability of CsPbBr 3 , as compared to that of the hybrid CH 3 NH 3 PbBr 3 [11][12][13]. One of the key issues for

Synthesis
Prior to the deposition of CsPbBr 3 -related EML, the CsPbBr 3 powder was firstly synthesized as previously mentioned [8]. Briefly, 1.9 mmol PbBr 2 and CsBr were added in 3 mL DMSO and the solution was vigorously stirred for 30 min at room temperature. By adding 3 mL of the hydrobromic acid (HBr) into the transparent PbBr 2 /CsBr solution, the orange CsPbBr 3 powder was obtained after subjecting the turbid solution to centrifugation and vacuum drying the precipitates overnight. Then, the fluorescent powder ( Figure S1, see the Supplementary Materials) was preserved in a sealed vial in the ambient environment for CsPbBr 3 film deposition. The preparation of the CsPbBr 3 precursor solution was done through the mixture of 262.5 mg of CsPbBr 3 powder and 28.5 mg of CsBr in 3 mL of DMSO. Here, the transparent precursor solution was referred to as Solution A for individually blending with the organic additives in the following device fabrication.
For the preparation of NiO x NPs, the synthesis was based on the protocol revealed in a previous report [22]. In a typical synthesis, 0.05 mmol of nickel nitrate hexahydrate (Ni(NO 3 ) 2 ·6H 2 O) was dissolved in 20 mL of de-ionized (DI) water to produce dark-green solution through vigorous stirring. Six milliliters of NaOH-water solution (10 mol/L) was then added dropwise into the dark-green solution while continuously stirring for another 5 min to generate plenty of light-green precipitation of Ni(OH) 2 . After subjecting to centrifugation and vacuum drying, the precipitation was calcined at 270 • C for 2 h. Finally, the derived black powder of NiO x NPs with an average particle size of~5 nm was dispersed in DI water with 30 mg/mL concentration as the NiO x NP ink.

Device Fabrication
In PeLED fabrication, as shown in Figure 1a, firstly the patterned indium tin oxide (ITO)-glass substrates were sequentially cleaned by detergent, acetone, isopropyl alcohol (IPA), and DI water, respectively. The substrates were then UV-ozone treated for 25 min to remove the cleaning agent residues and to modify the work function of indium tin oxide (ITO). As the HTL, PEDOT:PSS or colloidal NiO x NPs was immediately one-step spin-coated onto the substrates (inset of Figure 2). Secondly, Solution A was doped with and without 19 mg of PEO, 6 mg of TPBi, and 0.6 mg of FIrpic addition, for the deposition of EML. The devices with different compositions for EML were named and  Table 1. It should be noted that the processes were all carried out in ambient air, based on our previous report that the spin-coated CsPbBr 3 films were highly air-stable [8]. Finally, to complete the fabrication, TPBi acting as the electron transport layer (ETL), and Ca/Al as the electrode were deposited in the vacuum chamber at 5 × 10 6 torr. In the corresponding energy band diagram shown in Figure 1b, the VBM of NiO x NP HTL, CsPbBr 3 EML, and TPBi ETL was measured from the ultraviolet photoelectron spectroscopy (UPS) spectra ( Figure S2, see the Supplementary Materials) and referenced from a previous report [23], respectively. The measured VBM (−5.47 eV) value of the NiO x NP HTL is very close to that (−5.4 eV) of the NiO x NP film derived in a similar method [24].
Coatings 2020, 10, x FOR PEER REVIEW 3 of 8 colloidal NiOx NPs was immediately one-step spin-coated onto the substrates (inset of Figure 2). Secondly, Solution A was doped with and without 19 mg of PEO, 6 mg of TPBi, and 0.6 mg of FIrpic addition, for the deposition of EML. The devices with different compositions for EML were named and listed in Table 1. It should be noted that the processes were all carried out in ambient air, based on our previous report that the spin-coated CsPbBr3 films were highly air-stable [8]. Finally, to complete the fabrication, TPBi acting as the electron transport layer (ETL), and Ca/Al as the electrode were deposited in the vacuum chamber at 5 × 10 6 torr. In the corresponding energy band diagram shown in Figure 1b, the VBM of NiOx NP HTL, CsPbBr3 EML, and TPBi ETL was measured from the ultraviolet photoelectron spectroscopy (UPS) spectra ( Figure S2, see the Supplementary Materials) and referenced from a previous report [23], respectively. The measured VBM (−5.47 eV) value of the NiOx NP HTL is very close to that (−5.4 eV) of the NiOx NP film derived in a similar method [24].    Coatings 2020, 10, x FOR PEER REVIEW 3 of 8 colloidal NiOx NPs was immediately one-step spin-coated onto the substrates (inset of Figure 2). Secondly, Solution A was doped with and without 19 mg of PEO, 6 mg of TPBi, and 0.6 mg of FIrpic addition, for the deposition of EML. The devices with different compositions for EML were named and listed in Table 1. It should be noted that the processes were all carried out in ambient air, based on our previous report that the spin-coated CsPbBr3 films were highly air-stable [8]. Finally, to complete the fabrication, TPBi acting as the electron transport layer (ETL), and Ca/Al as the electrode were deposited in the vacuum chamber at 5 × 10 6 torr. In the corresponding energy band diagram shown in Figure 1b, the VBM of NiOx NP HTL, CsPbBr3 EML, and TPBi ETL was measured from the ultraviolet photoelectron spectroscopy (UPS) spectra ( Figure S2, see the Supplementary Materials) and referenced from a previous report [23], respectively. The measured VBM (−5.47 eV) value of the NiOx NP HTL is very close to that (−5.4 eV) of the NiOx NP film derived in a similar method [24].

Characterizations
As for the thin film and device characterization, the X-ray diffraction (XRD) pattern was obtained using a Rigaku MiniFlex II X-ray diffractometer. The absorption and electroluminescence (EL) spectra were measured by the UV-VIS spectrometer (LINKO, LKU-5200) and a fiber-optic spectrometer (Ocean Optics, USB 4000), respectively. The size of nanoparticles, surface morphology of films, and UPS spectra were obtained using a transmission electron microscope (TEM) (JEOL, JEM-2100F), an atomic force microscope (AFM) (Bruker, Innova), and an electron spectrometer (Chigasaki, ULVAC-PHI), respectively. Current-voltage (I-V) curves of the PeLEDs were measured using a Keithley 2400 source meter. Moreover, the luminance properties under various voltages were obtained by calibrating the photocurrent of a silicon detector using a luminance meter (Minolta, LS-110).

Results and Discussion
In Figure 2, the transmittance of NiO x NP layer is slightly lower than that of PEDOT:PSS in the visible light region, respectively. The overall transmittance at 520 nm was larger than 80% for both of the films coated on ITO-glass. The XRD patterns of the spin-coated EML from the Solution-A with and without the organic additives are shown in Figure 3. Basically, all the patterns are dominated by the peaks corresponding to orthorhombic CsPbBr 3 polycrystals. In the EML coated with only Solution A, the pattern was dominant by the (220) diffraction peak, the same as that of the CsPbBr 3 powder. However, by comparing the variation of the main peaks of the doped films, it is evident that the addition of PEO and FIrpic was especially favorable for the grain growth along the [110] direction. The decrease of intensity related to the (220) peak in the EML additionally doped with FIrpic can be possibly attributed to the variation of lateral grain size (see Figure 4h). Moreover, it was also revealed that FIrpic could definitely suppress the grain growth of (020) phase. By adding the TPBi, a new diffraction peak was shown at~31.6 • . This peak should not be related to the formation of 2-D CsPb 2 Br 5 or 0-D Cs 4 PbBr 6 crystals [25]. Nonetheless, the real attribution is still under investigation. No diffraction signals associated with PEO, TPBi, and FIrpic were detected.

Characterizations
As for the thin film and device characterization, the X-ray diffraction (XRD) pattern was obtained using a Rigaku MiniFlex II X-ray diffractometer. The absorption and electroluminescence (EL) spectra were measured by the UV-VIS spectrometer (LINKO, LKU-5200) and a fiber-optic spectrometer (Ocean Optics, USB 4000), respectively. The size of nanoparticles, surface morphology of films, and UPS spectra were obtained using a transmission electron microscope (TEM) (JEOL, JEM-2100F), an atomic force microscope (AFM) (Bruker, Innova), and an electron spectrometer (Chigasaki, ULVAC-PHI), respectively. Current-voltage (I-V) curves of the PeLEDs were measured using a Keithley 2400 source meter. Moreover, the luminance properties under various voltages were obtained by calibrating the photocurrent of a silicon detector using a luminance meter (Minolta, LS-110).

Results and Discussion
In Figure 2, the transmittance of NiOx NP layer is slightly lower than that of PEDOT:PSS in the visible light region, respectively. The overall transmittance at 520 nm was larger than 80% for both of the films coated on ITO-glass. The XRD patterns of the spin-coated EML from the Solution-A with and without the organic additives are shown in Figure 3. Basically, all the patterns are dominated by the peaks corresponding to orthorhombic CsPbBr3 polycrystals. In the EML coated with only Solution A, the pattern was dominant by the (220) diffraction peak, the same as that of the CsPbBr3 powder. However, by comparing the variation of the main peaks of the doped films, it is evident that the addition of PEO and FIrpic was especially favorable for the grain growth along the [110] direction. The decrease of intensity related to the (220) peak in the EML additionally doped with FIrpic can be possibly attributed to the variation of lateral grain size (see Figure 4h). Moreover, it was also revealed that FIrpic could definitely suppress the grain growth of (020) phase. By adding the TPBi, a new diffraction peak was shown at ~31.6°. This peak should not be related to the formation of 2-D CsPb2Br5 or 0-D Cs4PbBr6 crystals [25]. Nonetheless, the real attribution is still under investigation. No diffraction signals associated with PEO, TPBi, and FIrpic were detected.  Coatings 2020, 10, 336 5 of 8 more PEO. Too much of the electrically resistive PEO, however, was not beneficial for device performance [27]. In Figure 4e,f, the influence of TPBi and FIrpic on CsPbBr3 films was very different to that of PEO. When TPBi is added, the original small grains in Figure 4d tend to interconnect with one another, becoming a 2-D crystal network, which was enhanced by adding FIrpic. In Figure 4h the additive-assisted CsPbBr3 film on NiOx NPs, the CsPbBr3 crystals coalesced and became large and flat grains with few voids. In PeLED characterization, the luminance-current density-voltage (L-J-V) and the current efficiency-current density curves are shown in Figure 5a. For Device-PA and Device-NA with pure CsPbBr3 EML, the leakage current at 0~2.4 V was remarkably high because of the rough EML surface with plenty of pinholes, which acted as the shunt pathway. Furthermore, the large difference between the turn-on voltages for current injection (~2.5 V) and light threshold (VT~3.5 V) in Device-PA indicates that the injection charges were extremely unbalanced, while the defective CsPbBr3 crystals led to very low luminance. The clear trend of current efficiency (ηCE) increasing with current density shown in Figure 5b for Device-PA is strong evidence of the competition between leakage and injection current.
For Device-PB, Device-PC, and Device-PD, the crystal quality of EML was improved by reducing the roughness and eliminating the defects, which caused lower VT and higher luminance. In circumstances of neglectable leakage current, the higher current efficiency in Device-PD exhibited higher radiative recombination rate of excitons in the EML, probably owing to the phosphorescent capability of FIrpic for efficient Förster energy transfer [28].
Furthermore, Device-ND achieved the low light threshold voltage VT of 2.4 V, high luminance maximum (Lmax) of 4456 cd/m 2 , and high current efficiency of 2.9 cd/A by adopting the NiOx NPs and the additive-assisted EML that gave a better hole transporting and high radiative recombination rate, respectively. It can be clearly observed that the threshold of the minority carrier (holes) injection at 2.4 V and the onset of majority carrier (electrons) injection at ~2.3 V matched pretty well, which suggests that those negative effects such as Auger recombination (trions) and improper shift of exciton recombination zone can therefore be prevented [29,30]. Overall, compared with Device-PA, Device-ND had roughly 17-and 94-fold increased Lmax and ηCE, respectively. The comparison of surface morphology of each spin-coated layer is shown in Figure 4. Through calculation, the value of rms surface roughness (R q ) was 2.0, 5.7, 10.5, 8.4, 5.1, 4.8, 15.3 and 5.5 nm, respectively. It can be seen that the PEDOT:PSS (R q = 2.0 nm) was homogeneous and flat whereas the NiO x NP layer was slightly rough (R q = 5.7 nm) owing to the aggregation of NPs (inset of Figure 2). The surface of pure CsPbBr 3 film on PEDOT:PSS or NiO x NP layer was quite rough (R q > 10 nm), with distinct and diverse-sized 3-D CsPbBr 3 crystal grains separated by clear boundaries. As shown in Figure 4d, the addition of PEO reduced the film roughness by reducing the grain size and shrinking the size distribution [25,26]. It was found that R q could be further decreased by adding more PEO. Too much of the electrically resistive PEO, however, was not beneficial for device performance [27]. In Figure 4e,f, the influence of TPBi and FIrpic on CsPbBr 3 films was very different to that of PEO. When TPBi is added, the original small grains in Figure 4d tend to interconnect with one another, becoming a 2-D crystal network, which was enhanced by adding FIrpic. In Figure 4h the additive-assisted CsPbBr 3 film on NiO x NPs, the CsPbBr 3 crystals coalesced and became large and flat grains with few voids.
In PeLED characterization, the luminance-current density-voltage (L-J-V) and the current efficiency-current density curves are shown in Figure 5a. For Device-PA and Device-NA with pure CsPbBr 3 EML, the leakage current at 0~2.4 V was remarkably high because of the rough EML surface with plenty of pinholes, which acted as the shunt pathway. Furthermore, the large difference between the turn-on voltages for current injection (~2.5 V) and light threshold (V T~3 .5 V) in Device-PA indicates that the injection charges were extremely unbalanced, while the defective CsPbBr 3 crystals led to very low luminance. The clear trend of current efficiency (η CE ) increasing with current density shown in Figure 5b for Device-PA is strong evidence of the competition between leakage and injection current.
For Device-PB, Device-PC, and Device-PD, the crystal quality of EML was improved by reducing the roughness and eliminating the defects, which caused lower V T and higher luminance. In circumstances of neglectable leakage current, the higher current efficiency in Device-PD exhibited higher radiative recombination rate of excitons in the EML, probably owing to the phosphorescent capability of FIrpic for efficient Förster energy transfer [28].
Furthermore, Device-ND achieved the low light threshold voltage V T of 2.4 V, high luminance maximum (L max ) of 4456 cd/m 2 , and high current efficiency of 2.9 cd/A by adopting the NiO x NPs and the additive-assisted EML that gave a better hole transporting and high radiative recombination rate, respectively. It can be clearly observed that the threshold of the minority carrier (holes) injection Coatings 2020, 10, 336 6 of 8 at 2.4 V and the onset of majority carrier (electrons) injection at~2.3 V matched pretty well, which suggests that those negative effects such as Auger recombination (trions) and improper shift of exciton recombination zone can therefore be prevented [29,30]. Overall, compared with Device-PA, Device-ND had roughly 17-and 94-fold increased L max and η CE , respectively.
Finally, the normalized EL spectra of Device-PB, Device-NA, and Device-ND are shown in Figure 6 for observing the effect of organic additives on the emission of perovskite crystals. The spectrum of Device-PA is not shown because of the low EL intensity, while the spectra of Device-PC and Device-PD are very similar to that of Device-ND. No emission from the TPBi or FIrpic was detected, possibly due to their low doping concentration, low emission efficiency, and exciton energy transfer just mentioned [28]. However, the spectra are slightly different in peak position. By adding the additives, the peak position of the green light was shifted from 521.7 (Device-NA) to 519.5 nm (Device-ND), with no obvious change in spectral width (~17 nm). Also, the spectrum of Device-PB was broadened to 18.3 nm. Finally, the normalized EL spectra of Device-PB, Device-NA, and Device-ND are shown in Figure 6 for observing the effect of organic additives on the emission of perovskite crystals. The spectrum of Device-PA is not shown because of the low EL intensity, while the spectra of Device-PC and Device-PD are very similar to that of Device-ND. No emission from the TPBi or FIrpic was detected, possibly due to their low doping concentration, low emission efficiency, and exciton energy transfer just mentioned [28]. However, the spectra are slightly different in peak position. By adding the additives, the peak position of the green light was shifted from 521.7 (Device-NA) to 519.5 nm (Device-ND), with no obvious change in spectral width (~17 nm). Also, the spectrum of Device-PB was broadened to 18.3 nm.

Conclusion
In conclusion, the fabrication and characterization of an ambient-processed, electrically driven PeLED with a multiple-additive-assisted CsPbBr3 EML and a colloidal NiOx NP HTL were demonstrated. From the experimental results, it can be understood that the deeper VBM (-5.47 eV)  Finally, the normalized EL spectra of Device-PB, Device-NA, and Device-ND are shown in Figure 6 for observing the effect of organic additives on the emission of perovskite crystals. The spectrum of Device-PA is not shown because of the low EL intensity, while the spectra of Device-PC and Device-PD are very similar to that of Device-ND. No emission from the TPBi or FIrpic was detected, possibly due to their low doping concentration, low emission efficiency, and exciton energy transfer just mentioned [28]. However, the spectra are slightly different in peak position. By adding the additives, the peak position of the green light was shifted from 521.7 (Device-NA) to 519.5 nm (Device-ND), with no obvious change in spectral width (~17 nm). Also, the spectrum of Device-PB was broadened to 18.3 nm.

Conclusion
In conclusion, the fabrication and characterization of an ambient-processed, electrically driven PeLED with a multiple-additive-assisted CsPbBr3 EML and a colloidal NiOx NP HTL were demonstrated. From the experimental results, it can be understood that the deeper VBM (-5.47 eV) of 5 nm-sized NiOx NPs facilitated the injection of holes, the minority carrier, into EML by decreasing the energy barrier without sacrificing the transmittance of the HTL. The doping of PEO was significantly beneficial to reduce the roughness of EML and thus reduce the device leakage current, whereas TPBi and FIrpic helped to improve the crystal quality by interconnecting the adjacent crystal grains to reduce the PL-inefficient boundaries. In terms of the crystal orientation, additives tended to favor the grain growth along the [110] direction. Because of the improved crystal quality of the spin-coated perovskite film and fairly balanced charge injection, the novel PeLED turned on at a very low voltage of 2.4 V, and exhibited a high luminance maximum of 4456 cd/m 2 and an enhanced current efficiency of 2.9 cd/A, which is about 17-and 94-fold of that of the PeLED

Conclusions
In conclusion, the fabrication and characterization of an ambient-processed, electrically driven PeLED with a multiple-additive-assisted CsPbBr 3 EML and a colloidal NiO x NP HTL were demonstrated. From the experimental results, it can be understood that the deeper VBM (−5.47 eV) of 5 nm-sized NiO x NPs facilitated the injection of holes, the minority carrier, into EML by decreasing the energy barrier without sacrificing the transmittance of the HTL. The doping of PEO was significantly beneficial to reduce the roughness of EML and thus reduce the device leakage current, whereas TPBi and FIrpic helped to improve the crystal quality by interconnecting the adjacent crystal grains to reduce the PL-inefficient boundaries. In terms of the crystal orientation, additives tended to favor the grain growth along the [110] direction. Because of the improved crystal quality of the spin-coated perovskite film and fairly balanced charge injection, the novel PeLED turned on at a very low voltage of 2.4 V, and exhibited a high luminance maximum of 4456 cd/m 2 and an enhanced current efficiency of 2.9 cd/A, which is about 17-and 94-fold of that of the PeLED with a pure/untreated CsPbBr 3 EML and a conventional PEDOT:PSS HTL.
Supplementary Materials: The following are available online at http://www.mdpi.com/2079-6412/10/4/336/s1, Figure S1, Absorption and photoluminescence (PL) spectra of the synthesized CsPbBr 3 powder. Insets are the photographs of CsPbBr 3 powder under white and UV light irradiation, respectively; Figure S2, UPS spectra of the NiO x NP HTL. According to the cutoff values, the valence band maximum (E v ) of NiO x NP HTL can be determined as 5.47 eV.

Conflicts of Interest:
The authors declare no conflict of interest.