Self-defect-passivation by Br-enrichment in FA-doped Cs1−xFAxPbBr3 quantum dots: towards high-performance quantum dot light-emitting diodes

Halide vacancy defect is one of the major origins of non-radiative recombination in the lead halide perovskite light emitting devices (LEDs). Hence the defect passivation is highly demanded for the high-performance perovskite LEDs. Here, we demonstrated that FA doping led to the enrichment of Br in Cs1−xFAxPbBr3 QDs. Due to the defect passivation by the enriched Br, the trap density in Cs1−xFAxPbBr3 significantly decreased after FA doping, and which improved the optical properties of Cs1−xFAxPbBr3 QDs and their QD-LEDs. PLQY of Cs1–xFAxPbBr3 QDs increased from 76.8% (x = 0) to 85.1% (x = 0.04), and Lmax and CEmax of Cs1–xFAxPbBr3 QD-LEDs were improved from Lmax = 2880 cd m−2 and CEmax = 1.98 cd A−1 (x = 0) to Lmax = 5200 cd m−2 and CEmax = 3.87 cd A−1 (x = 0.04). Cs1–xFAxPbBr3 QD-LED device structure was optimized by using PVK as a HTL and ZnO modified with b-PEI as an ETL. The energy band diagram of Cs1–xFAxPbBr3 QD-LEDs deduced by UPS analyses.

www.nature.com/scientificreports/ than organic cations, so that they much improved the defect passivation efficiency 15 . Similarly, the internal hydrogen bonding between halides and ammonium cations (MA or FA) within the perovskite framework affects the geometry of perovskites 11 . Herein, we report the effects of the FA doping on CsPbBr 3 QD-LEDs. We found that partial substitution of Cs with FA led to the significant increase of Br concentration in Cs 1−x FA x PbBr 3 framework by hydrogen bonding and ionic interaction between FA and Br -, and which dramatically decreased the defects in Cs 1−x FA x PbBr 3 . Accordingly, inverted-type perovskite LEDs prepared with Cs 1−x FA x PbBr 3 QD in ambient condition exhibited much better performance than those prepared with undoped CsPbBr 3 QDs. The performance of QD-LEDs with Cs 1−x FA x PbBr 3 at the optimized composition (x = 0.04) exhibited the maximum luminance (L max ) of 5200 cd m −2 at 5.3 V and the maximum current efficiency (CE max ) of 3.87 cd A −1 at 5.0 V. These are much better than those values for undoped (x = 0) and over-doped (x = 0.055) ones: L max = 2880 cd m −2 at 6.2 V and CE max = 1.98 cd A −1 at 5.9 V for CsPbBr 3 , and L max = 2250 cd m -2 at 5.6 V and CE max = 2.73 cd A −1 at 5.0 V Cs 0.945 FA 0.055 PbBr 3 , respectively. This FA doping strategy enables us not only to suppress the non-radiative recombination in luminance layer to improve the performance of QD-LEDs but also to realize the high efficiency in optoelectronic devices. Figure 1a-b show the stack configuration of a solution-processed inverted-type QD-LED with multilayer heterojunctions of ZnO NCs modified with b-PEI (ZnO/b-PEI), Cs 1−x FA x PbBr 3 perovskite QDs, PVK, and V 2 O 5−x , which were sequentially spin-coated on ITO-coated glass substrates under the ambient condition, for electron transport layer (ETL), luminescence layer, hole transport layer (HTL), and hole injection layer (HIL), respectively. The thicknesses of ZnO/b-PEI, Cs 1−x FA x PbBr 3 , PVK, and V 2 O 5−x layers was approximately 32, 18, 20, and 10 nm, respectively, which were determined by the cross-sectional SEM image shown in Fig. 1a. Figure 1c presents a schematic illustration of Cs 1−x FA x PbBr 3 with a cubic structure consisting of the Pb cation in sixfold coordination surrounded by an octahedron of Br anions and the Cs cation (and partial substitution of FA) in 12-fold cuboctahedral coordination. Figure 1d shows the photograph of the Cs 1−x FA x PbBr 3 QD-LED device fabricated on a 2 × 2 mm 2 active area exhibiting uniform emission. Figure 1e presents the normalized electroluminescence (EL) spectra of devices prepared using Cs 1−x FA x PbBr 3 QDs. The EL peaks located at 508, 512, and 513 nm for Type A (x = 0), Type B (x = 0.04) and Type c (x = 0.055), respectively. All EL spectra show narrow emission (FWHM = 19 nm) and high color purity, which is solely attributed to the band-edge emission of Cs 1−x FA x PbBr 3 QDs with a slightly redshifted emission from the PL spectrum taken in the QD colloidal solution 32 . Additionally, no notable parasitic emission originated from the PVK layer was observed (Fig. 1e) 31 . Figure 1f shows the electronic energy level diagram (with respect to the vacuum level) of the layers applied for the QD-LEDs. The www.nature.com/scientificreports/ ionization energy (IE) and the electron affinity (EA) of these layers were estimated from the UV-visible absorption spectra (not shown here) and UPS analysis, which were in good agreement with the previous studies 3,31,33-35 . The electrical and electroluminescent performances of Cs 1−x FA x PbBr 3 QD-LEDs were examined by measuring the J-V-L characteristic curves. Figure 2a shows the J-V characteristic curves of Cs 1−x FA x PbBr 3 QD-LEDs. All QD-LEDs exhibited high electrical rectification behavior with an inflection point around at 0.54 V and steeper increase of the current density above the inflection point (inset of Fig. 2a). In Fig. 2b, J-V characteristic curves plotted on double-logarithmic axes. In the Ohmic conduction region (J ∝ V 1.2 ), the leakage current of devices doped with FA (Type B and Type C) was slightly lower than that of Type A. In the trap-limited conduction region, the J-V curve slops for all QD-LEDs (J ∝ V 8.4 ) were slightly deviated from the typical power law (J ∝ V 7 ), which was attributed to similar charge injection/transport energy band diagram of the QD-LEDs structure. Interestingly, the Type B and C QD-LEDs exhibited higher luminance, higher current efficiency (CE), higher external quantum efficiency (EQE), and lower turn-on voltage than those of Type A, as shown in Fig. 2c-e respectively. The turn-on voltage (calculated with a luminance of 1 cd m −2 ) was 4.1, 3.2, and 3.5 V for Type A, B and C, respectively. Originally, the maximum luminance (L max ) of the Type A was 2880 cd m −2 at 6.2 V. It significantly increased to 5200 cd m −2 (at 5.3 V) when CsPbBr 3 was slightly doped with FA (x = 0.04, Type B). We attribute this enhancement to the decreased trap density by doping with FA. However, L max decreased to 2250 cd m −2 at 5.6 V when CSPbBr 3 was over-doped with FA (x = 0.055, Type C). This is because the valence band maximum (VBM) of the CsPbBr 3 is slightly upshifted by FA doping, which makes difficult inject electrons (this will be discussed in Fig. 6). The EQE and CE were also maximized for Type B devices (Fig. 2d,e). The highest EQE of Type B device was EQE max = 1.36%, which was much higher than that of Type A (EQE max = 0.72%) and Type C (EQE max = 0.96%), respectively (Fig. 2d). At the same time, Type B device showed much better CE max = 3.87 cd A −1 than that of Type A (CE max = 1.98 cd A −1 ) and Type C (CE max = 2.73 cd A −1 ) (Fig. 2e). All devices exhibited a bright and uniform pure green color from the entire pixel area under bias voltage of 4.0-6.0 V, and their corresponding CIE (Commission Internationale de l'Eclairage 1931) chromaticity coordinates were Type A (0.048, 0.711), Type B (0.057, 0.733), and Type C (0.064, 0.742), respectively (Fig. 2f,g).

Results and discussion
Above all data clearly suggest that FA doping significantly improves the device performances including luminescence, EQE and CE of Cs 1-x FA x PbBr 3 QD solutions. To understand the effects of FA doping on CsPbBr 3 QDs, we first investigated them with TEM. Figure 3a,c, and e show representative TEM micrographs of Cs 1-x FA x PbBr 3 QDs. They formed the well-defined cubic particles with an average diameter of d avg = 6.0 ± 0.16 nm, 6.5 ± 0.19 nm, and 6.5 ± 0.21 nm for Type A, B and C, respectively. HR-TEM images clearly show the cubic crystal structure of Type A QDs (Fig. 3b). The measured d-spacing for (100) and (110) plane was 0.58 nm and 0.41 nm, respectively. With FA doping, the d-spacing for (100) and (110) plane slightly increased to 0.59 nm and 0.42 nm for both Type B and Type C (Fig. 3d,f). The enlarged HR-TEM images in the top-insets of Fig. 3b,d, and f well match with the atomic arrangement of cubic Cs 1-x FA x PbBr 3 QDs crystal. The bottom-insets of Fig. 3b,d, and f show  The elemental composition of Cs 1-x FA x PbBr 3 QDs was investigated by using XPS. As shown in Supplementary Figure S3a, six major XPS peaks were assigned as Br 3d, Pb 4f, C 1s, N 1s, O 1s, and Cs 3d, respectively. The chemical state of nitrogen was carefully characterized by multiple-peak fitting of the N 1s peak using symmetric Voigt functions. Figure 4a shows the characteristic N 1s spectra of Cs 1-x FA x PbBr 3 QDs (Type A-C). Only ammonium (-NH 3+ ) peak was observed at 401.6 eV for Type A, and which is originated from DDAB added as a ligand for synthesis 4,36 . It is notable that the TOAB does not act as a ligand owing to the large steric effect 6 . For Type B and C, however, the primary amine group (-NH 2 ) as well as ammonium group (-NH 3+ ) was observed at 399.0 eV. The presence of the primary amine group suggests the formation of FA doped Cs 1-x FA x PbBr 3 QDs. Based on the XPS analysis, the relative concentration of FA to Cs was determined as 0.04 and 0.055 for Type B and C, respectively. Similarly, XPS data for C 1s also well support the formation of FA doped Cs 1-x FA x PbBr 3 QDs ( Figure S3b, Supplementary Information). The characteristic peak of FA, C=N (287.6 eV) was observed for Type B and C along with C-C (284.8 eV) and C-N (285.6 eV) peaks, while only two peaks of C-C and C-N bonding were observed for Type A.
The elemental composition including Cs, Pb and Br in Cs 1-x FA x PbBr 3 was further investigated (Fig. 4 and Supplementary Figures S3-S5). As summarized in Fig. 4c, it is notable that the concentration of Br was significantly increased by FA doping. Originally, undoped CsPbBr 3 QDs (Type A) have a composition ratio of Cs:Pb:Br = 1.00:0.79:3.11. The composition ratio for CsPbBr 3 NC with large size (d avg = 8.0 ± 0.16 nm) was Cs:Pb:Br = 1.00:0.73:2.90. In this case, the terminology of CsPbBr 3 NC was used to clarify that its average size (d avg = 8.0 ± 0.16 nm) is larger than the Bohr diameter (d Bohr = 7 nm) 7,8,37,38 , while the size of CsPbBr 3 QD is smaller than the Bohr diameter. More careful analysis of XPS peaks for Br 3d revealed that Brions are concentrated at the surface of QD. The concentration of Brions located at the surface (highlighted in red shadow, 3d 5/2 = 67.90 eV and 3d 3/2 = 68.95 eV) was higher than that of the Brions located inside (highlighted in cyan www.nature.com/scientificreports/ shadow, 3d 5/2 = 67.40 eV and 3d 3/2 = 68.40 eV) (Fig. 4b) 39 . The Br-rich surface of CsPbBr 3 QD was attributed to the ionic interaction of Brions with ammonium groups in DDAB 6 . The composition ratio of (Cs + FA):Pb:Br was changed to 1.00:0.84:3.35 and 1.00:0.91:3.54 as increasing FA doping concentration to x = 0.04 (Type B) and x = 0.055 (Type C), respectively. In this case, the ratio of anion to cation [Br/(Cs + FA + Pb)] increased almost linearly with the FA content in Cs 1-x FA x PbBr 3 (Fig. 4d, the left axis). At the same time, FA doping increased the inner Br content as well as the surface Br content (Fig. 4d, the right axis). The inner/surface Br ratio was maximized for Type B. These results strongly suggest that hydrogen bonding and ionic interaction of FA with Brions led to the increase of Br content. Since FA molecules have both a primary amine group and an iminium group, the hydrogen bonding as well as ionic interaction in Cs 1-x FA x PbBr 3 QDs is highly effective 40 . The effects of Br-enrichment on the optical properties of Cs 1-x FA x PbBr 3 QDs have been investigated (Figures S6-S8 and Table S1, Supplementary Information). Supplementary Figure S6 shows the absorbance and PL of Cs 1-x FA x PbBr 3 QD solutions. As increasing FA contents, both absorbance and PL were slightly red-shifted. PLQY of Cs 1-x FA x PbBr 3 QD solutions varied from 76.8% (Type A) to 85.1% (Type B), and 82.6% (Type C) ( Figure S7, Supplementary Information). CsPbBr 3 NC solution with large size showed a lower PLQY of 67% ( Figure S7, Supplementary Information). PLQY data well agreed with TR-PL measurements ( Figure S8 and Table S1, Supplementary Information). The PL decay curves were well-fitted with the biexponential decay function consisting of a fast-decay lifetime (τ 1 ) and a slow-decay lifetime (τ 2 ) (Table S1, Supplementary Information) 10 . In this case, τ 1 is originated from the initially populated core-state recombination, and τ 2 is related with the surface emission 41 . As summarized in Table S1, τ 2 was dramatically increased by FA doping while τ 1 showed relatively small changes. Hence, the average PL decay times <τ> increased from 16.5 ns (Type A) to 30.51 ns (Type B) and 28.51 ns (Type C). Furthermore, the contribution ratio of the surface-related emission (W 2 ) to the intrinsic core-state recombination emission (W 1 ) was maximized for Type B. These results strongly suggest that FA doping reduced the trap-states in the Cs 1-x FA x PbBr 3 QD, and which was apparently originated from Br-enrichment by FA doping.
The effects of FA doping on the electronic structure of QD-LEDs also have been investigated. UPS measurements were carried out in the secondary electron cutoff (Fig. 5a) and/or VBM (Fig. 5b)   www.nature.com/scientificreports/ A, B and C, respectively. As the FA content increased from 0 to 0.055, the Φ value slightly increased from 5.06 eV to 5.35 eV. Also, we observed a slight shift of the VB level towards E Fermi with FA doping (Fig. 5b). For effective hole injection, PVK was used as a hole transfer layer (HTL). PVK layers exhibited the similar electronic energy level as Cs 1-x FA x PbBr 3 . From the onset values shown in Fig. 5c, the Φ of PVK was calculated to be 4.60, 4.62, and 4.62 eV, for Type A, B and C, respectively. The Φ values of other layers were also obtained ( Figure S9a, Supplementary Information). The IE values of PVK for all Type A, B and C were estimated as same (IE ~ 5.73 eV). The electronic energy level of PVK was almost unaffected by FA doped Cs 1-x FA x PbBr 3 QDs sublayer (Fig. 5c,d). The hole barrier height (Δh) was estimated from the difference of HOMO (and VBM) level between overlayer and underlayer ( Figures S9b,c). Due to the significantly lowered energy level which was attributed to the laying-down assembly of PVK chains on the Cs 1-x FA x PbBr 3 layer ( Figure S10, Supplementary  Information) 34 , PVK layer formed the quantum-well-like energy alignment rather than the hole-barrier. The Δh 2 formed at the PVK/QD heterointerface was determined to be − 0.43 eV, and which helps to inject holes to the Cs 1-x FA x PbBr 3 layer without large barrier. This was well supported by the J-V characteristics of the hole-onlydevice (HOD) ( Figure S11a, Supplementary Information). The J-V curve of the HOD (Al/V 2 O 5-x /PVK/CsPbBr 3 / ITO) was quite similar as that of HOD (Al/V 2 O 5-x /PVK/ITO) while the leakage current was much lower in the low bias voltage. This imparts that there is no significant hole-barrier between CsPbBr 3 layer and PVK layer.
b-PEI was used to block the undesirable parasitical electron injection (denoted yellow arrow in Fig. 6) due to the moderately high interface dipole at the interface of CsPbBr 3 /b-PEI and the decreased Φ of ZnO from 4.71 eV to 3.65 eV (1.41 eV, see Supplementary Figure S9). As shown in Supplementary Figure S11b, the excessive electron injection was well suppressed in the electron-only-device (EOD) (Al/CsPbBr 3 /b-PEI/ZnO/ITO), and which improves the charge carrier balance of Cs 1-x FA x PbBr 3 QD-LEDs.
Finally, the space-charge-limited current (SCLC) analysis was performed for EODs made of Type A, B and C to prove the reduction of trap-states by FA doping (Figure S12, Supplementary Information). The trap density (n trap ) was calculated by the equation n trap = 2εε 0 V TFL /(ed 2 ), where ε is a relative dielectric constant (ε = 16.46) 42 , ε 0 is the permittivity constant in free space, e is the elementary electronic charge, and d is the thickness. As shown in Supplementary Figure S12, FA doping significantly decreased the onset voltage of the trap-filled limit regime (V TFL ) from 1.8 V (Type A) to 1.3 V and 1.6 V (Type B and Type C). The calculated n trap values were 1.5 × 10 −18 cm −3 , 9.4 × 10 −17 cm −3 , and 1.2 × 10 −18 cm −3 for the Type A, Type B, and Type C, respectively. This result clearly shows that the trap states were significantly reduced by FA doping. For the synthesis of CsPbBr 3 QDs, Cs precursor solution (1 mL) was quickly injected into PbBr 2 precursor solution (9 mL), and then the solution was stirred for 5 min at room temperature. Afterward, didodecyldimethylammonium bromide (DDAB; Aldrich, CAS No. 3282-73-3, 3 mL) dissolved in toluene (10 mg mL −1 ) was added dropwise to the reaction solution, and which was stirred more for 5 min. Lastly, the reaction solution was rapidly quenched by cooling in ice bath. CsPbBr 3 QDs were purified by centrifugation (g-force = 2800 RCF) to remove large particles and aggregates. After centrifugation, the colloidal CsPbBr 3 solution (with a bright green color and a green emission) was obtained. The CsPbBr 3 solution was further purified based on the previous reports 6 . Ethyl acetate was added to the CsPbBr 3 solution with a 2:1 volume ratio. After centrifugation, the precipitate was collected and dispersed in toluene. The precipitate solution (in toluene) was mixed again with ethyl acetate with a 2:1 volume ratio, and which was centrifugated. Finally, the collected precipitate was re-dispersed in n-octane with a concentration of 10 mg/mL for further use. diluted in isopropyl alcohol) on ITO-coated glass substrates, and which were subsequently annealed at 120, 100, 70, 160, and 120 °C for 1 min, 20 min, 5 min, 10 min, and 1 min, respectively instantly after spin-coating each layer (Fig. 1a). In this case, FA concentration in Cs 1-x FA x PbBr 3 QDs was controlled to x = 0 (denoted as Type A), x = 0.04 (denoted as Type B) and x = 0.055 (denoted as Type C). Aluminum (Al; 120 nm thick) cathode was deposited onto the V 2 O 5-x layer by thermal evaporation in a vacuum chamber through a patterned shadow mask, for all the QD-LEDs, electron-only devices (EODs), and hole-only devices (HODs). It is notable that all devices were processed using the spin-coating method under ambient conditions (20-24 °C and 10-30% humidity), except for the deposition of electrodes. The fabricated devices were capped with a glass lid and ultraviolet curable epoxy resin. No significant damage was observed after the coating of each layer.

characterizations.
A bright-field high-resolution transmission electron microscopy (HR-TEM), with an acceleration voltage of 200 keV (JEOL, JEM-2100F) was used to inspect the mean diameter and crystallinity of Cs 1−x FA x PbBr 3 QDs. Energy-dispersive X-ray spectroscopy (EDX) equipped in a scanning transmission electron microscopy (STEM) (STEM-EDX) was used for elemental mapping of Cs 1−x FA x PbBr 3 QDs. X-ray diffraction (XRD) patterns were taken on an X-ray diffractometer (PANalytical, X'Pert PRO). Time-integrated photoluminescence (PL) spectra were obtained on a fluorescence spectrophotometer (PerkinElmer, LS55). The photoluminescence quantum yield (PLQY) was obtained on an absolute photoluminescence quantum yield measurement system (Jasco FP-8500) with an integrating sphere at room temperature. Optical absorbance was characterized by using an UV-Vis-NIR spectrometer (Agilent Technologies, Cary 5000). The time-resolved photoluminescence (TR-PL) were measured using a time correlated single photon counting system (Horiba Jobin Yvon iHR320). A pulsed InGaN multiple quantum-well LED (λ = 405 nm, repetition rate 1 MHz and optical pulse duration 200 ps) was used as an excitation source for the TR-PL measurements. Surface morphologies of the QDs layers were characterized by field emission scanning electron microscopy (FE-SEM, Hitachi SU-8010), and atomic force microscopy (AFM, Park Systems XE-100) with a silicon probe (Nanoworld 910 M-NCHR) under non-contact mode. The root-mean-square (rms) surface roughness was averaged from at least five different areas