High efficiency green InP quantum dot light-emitting diodes by balancing electron and hole mobility

The industrialization of quantum dot light-emitting diodes (QLEDs) requires the use of less hazardous cadmium-free quantum dots, among which ZnSe-based blue and InP-based green and red quantum dots have received considerable attention. In comparison, the development of InP-based green QLEDs is lagging behind. Here, we prepare green InP/ZnSe/ZnS quantum dots with a diameter of 8.6 nm. We then modify the InP quantum dot emitting layer by passivation with various alkyl diamines and zinc halides, which decreases electron mobility and enhances hole transport. This, together with optimizing the electron transport layer, leads to green 545 nm InP QLEDs with a maximum quantum efficiency (EQE) of 16.3% and a current efficiency 57.5 cd/A. EQE approaches the theoretical limit of InP quantum dots, with an emission quantum yield of 86%. Developing green-emitting InP quantum dot light-emitting diodes (QLED) is lagging behind their red and green counterparts. Here, green InP quantum dots are prepared and incorporated into a QLED so to promote hole transport and reduce electron mobility, resulting in a maximum quantum efficiency of 16.3 %.

C olloidal quantum dots (QDs) are considered as the nextgeneration emissive materials for quantum light-emitting diodes (QLEDs) owing to their size controllable bandgaps, excellent photoelectric properties and high color purity [1][2][3][4] . The first QLED reported in 1994 5 suffered an extremely low external quantum efficiency (EQE) of less than 0.01%. It was not until 2011, a milestone for the breakthrough was made, where QLED was emerged with the introduction of n-type ZnO nanoparticles into the hybrid QLEDs as the electron transport layer (ETL), demonstrating then a high luminance (68,000 cd m −2 ) for green CdSe-ZnS QDs 1 . Ever since, progress has been continuously evolved for the CdSe relevant QLEDs, where EQE reaches the theoretical limit of 20% and luminance that attains 3.6 × 10 5 , 1.7 × 10 6 , and 6.3 × 10 4 cd m −2 for red, green, and blue devices, respectively 2,6-11 . Despite excellent performance for CdSe relevant QLEDs, however, the use of Cd has been banned in numerous consumer electronics because of its environment and health hazard 1,2 . In the quest of an alternative, the development of Cd-free QLEDs is thus demanding for their broad and practical implementation. In this regard, QLEDs incorporating a number of other QDs, such as InP, ZnSe, CuInS 2 , and lead halide perovskites, etc., have also been widely studied [12][13][14][15][16][17] , among which indium phosphide (InP)-based and zinc selenide (ZnSe)-based QDs are considered as promising benign alternatives, which are also capable of covering nearly the full range of natural colors 3,18-24 . Up to this stage, Cd-free QLEDs still show inferior performance to those with Cd-based QLEDs. Nevertheless, having spent intense efforts in recent years, blue ZnSeTe and red InP QLEDs have a leaping advance. The optimized blue ZnSeTe QLED with double emitting layers showed an EQE of 20.2% close to the theoretical limit with a brightness of 88,900 cd m −2 25 . In yet another approach, the red InP QLED has achieved an EQE of 21.4% with a high brightness of 1.0 × 10 5 cd m −2 23 . In comparison, the progress of InP-based green QLEDs still lags far behind. Under a conventional device structure that incorporates green InP/ZnSeS/ZnS QDs, a champion EQE of 13.6% was reported with a maximum luminance of 13,900 (cd m −2 ) 26 . As for a more practical inverted structure, the record high EQE for green InP QLEDs was even lower, which was reported to be 10.0% with a maximum luminance of 4000 (cd m −2 ) 27 . The inferior performance of green InP-based QLEDs mainly lies in the mismatched mobility between electron and hole. In the state-of-the-art organic/inorganic hybrid QLEDs, ZnO is commonly applied as an effective electron injection layer, rendering fast electron mobility 26 . Conversely, the hole injection is usually impeded due to the energy-level offset between emissive InP QDs and hole transporting layers. This leads to unbalanced electrons and holes in the recombining zone and hence low device efficiency 2,6,28 . Moreover, during the device operation, the excessive carriers accumulated at the barrier interface not only act as the nonradiative centers but also increase the driving voltage, limiting the device lifetime [29][30][31] . Theoretically, finding hole transport materials (HTMs) to match the energy levels of InP QDs is likely to improve hole injections, which unfortunately, is a great challenge because of the deep valance band of InP QDs [32][33][34] . Especially, theoretical approach on the energy-level alignments of a green InP/ZnSe/ZnS core/multi-shell structure indicates that the green InP QDs have a smaller e/h wave function overlap 35 , leading to further difficulty in adjusting the balance between electron and hole.
To circumvent the obstacle of electron/hole mismatch in green InP QLEDs, considerable efforts have been made to improve the energy-level alignment between the highest occupied molecular orbital (HOMO) of HTL and valence band of the InP QDs by tailoring the shell structure of QDs 7,36 . These methods are, however, nontrivial, which must undergo modification by the synthesis of QDs multilayer structure. Alternatively, it may be more facile and effective to post-modify the structures of QDs, which may simultaneously minimize several impeding factors such as energy transfer, Auger recombination and/or charge accumulation at the interfaces, promoting the charge balance. In light of this, ligand exchange plays one of the keys, which is able to influence the carrier mobility by altering the inter-QD dielectric environment as well as the tunneling distance. It has been reported that the mobility can increase exponentially upon decreasing ligand length in the absence of any other modification 37 . Moreover, appropriate ligands can passivate electronic trap sites on the QD surface arising from the structural aperiodicity and/or off-stoichiometry of the QD core. This result may increase carrier and exciton lifetimes as well as provide a degree of control over the doping level and type of the coupled QD film [38][39][40] . Similar to the modification of the shell structure, altering the identity of the chemical binding group and dipole moment of the ligand should change the strength of the QDligand surface dipole, shifting the vacuum energy hence the valence band maximum (VBM) and conduction band minimum (CBM) of QDs. The influence of surface chemistry on the QD energy levels has been reported in a number of studies [41][42][43][44] .
The above modification, in theory, depends on the properties of the as-prepared InP QDs in terms of core/shell structure, original ligand passivation and importantly the emission quantum yield. The latter should be sufficiently high for the QLEDs performance, which is commonly inferior for InP QDs (cf. CdSe QDs). In this work, based on the green QDs (GQDs) of InP, that consistently exhibit high emission yield of 86 ± 2% at 535 nm, we then performed post-modification of the QDs surface as well as interlayer modification of the energy levels. The key lies in the checks and balance of electron and hole mobility synergistically by post-passivation of the as-prepared InP GQDs. The systematic tuning brings closer the electron and hole mobility, resulting in a record high green InP QLED under an inverted device structure. Based on 30 devices being examined, an average EQE of 15.3% can be obtained, among which the champion EQE reaches 16.3% with the brightness of 1.26 × 10 4 cd m −2 at 545 nm. Detail of results and discussion is elaborated in the following sections.

Results and discussion
Ligand-dependent properties. For clarity, Fig. 1 depicts the progressive pathways to passivate the InP-based QDs. The asprepared green InP/ZnSe/ZnS core/shell/shell QDs containing oleic acid ligands were measured to have an average diameter of 8.6 ± 1.2 nm. These QDs, denoted as InP GQDs, consistently maintain their photoluminescence quantum yield (PLQY) of 86 ± 2% (see Supplementary Fig. 1 for the detail of characterization). The resulted InP GQDs were further modified by the passivation of various alkyl diamines consisting of two amino groups, respectively, at two terminals bridged by various lengths of the alkyl chain. Note that one of the alkyl diamines, 1,2ethylenediamine, used as a ligand, has been reported to boost the emission properties of CdTe/ZnS QDs by reducing the trapping site 45 . In addition to accommodate perhaps the trapping sites, the interaction of alkyl diamines with ZnS also has been reported to affect the ZnS structure, changing from cubic to hexagonal packing and hence altering the size of the particle to fine-tune the energy gap 46 . Similar alkyl diamines modification approach has not been explored in InP QDs. Due to the different lattice sizes and atom adduct properties, the exploitation of 1,2-ethylenediamine passivation in boosting CdTe/ZnS QDs may not work in InP QDs. In addition, as for various lengths of the alkyl chain, their insulation property may also largely affect the charge tunneling effect and hence the mobility. In this study, various alkyl diamines were selected and optimized via the performance of InP QLEDs where InP GQDs were modified by different alkyl diamines. As elaborated in the section of device fabrication it turned out that 1,4-butanediamine (BDA) rendered the top performance. Therefore, the results and discussion of the properties of post-modification are based on the BDA treated InP GQDs. The rest related data of the other alkyl diamines such as (1,2ethylenediamine, EDA) and (1,6-hexanediamine, HDA) are listed and elaborated in the Supplementary Fig. 2.
Followed by the alkyl diamine (BDA) modification, further modification of InP GQDs were carried out by incorporating the halogen anions ( Fig. 1). Addition of the halide salts to improve QDs optoelectronic performance has been reported in the case of CdSe/ZnS QDs 47 . Theoretical calculations have shown that the hole trapping sites originates from the coordination of chalcogenides on the surface of ZnS. Therefore, in theory, the small metal halide salts should be capable of filling the residual defects that BDA cannot accommodate 48 . Note that the optimization of zinc halides are mainly based on the performance of the corresponding QLED (vide infra). Among various zinc halide additives, it turns out that ZnI 2 added QLED exhibits the best device performance and will be the focus for the discussion. The rest related date of the other zinc halides such as ZnCl 2 and ZnBr 2 are listed and elaborated in the Supplementary Fig. 3.
Optical measurements. Figure 2a reveals the absorption spectra of the InP GQDs before and after passivation by BDA. Upon BDA passivation, non-negligibly, the absorption onset has been extended from 550 nm (2.25 eV) to 560 nm (2.21 eV). The decrease of bandgap after modification infers a number of possible combinations of the energy alignment between conduction band (CB) and valance band (VB). This is plausibly caused by different degrees of CB and VB shift induced by the surface dipoles 49 , which, in theory, is an opportunity to block the electron injection but facilitate the hole injection. Further evidence of the bandgap change is given by the emission red shift from 535 to 540 nm in toluene upon BDA modification (see Fig. 2b). This BDA-modification-dependent luminescent peak wavelength in solution seems to be a disadvantage for later QLEDs application. However, upon fabrication into the device, it is interesting to know that the electroluminescence spectrum has negligible shift (<1 nm) between those with and without BDA-modified QDs devices (see Fig. 2c). This perhaps is due to the dominant dielectric interlayer environment that offsets the spectral changes in the device (vide infra). Further addition of metal halogen ions such as ZnCl 2 , ZnBr 2 , and ZnI 2 do not change the absorption and emission properties QDs (see Fig. 2a and b for the addition of ZnI 2 ), implying that only rather small portion of trap sites remained after BDA modification. Nevertheless, due to the capping of the remaining trap sites at a large, the emission properties in terms of lifetime increase, as elaborated below.
The comparative studies were then made in the photoluminescence lifetime before and after the BDA modification; the corresponding data are listed in Table 1 (modification by other diamine and zinc halides, as shown in Supplementary Table 1). The time-resolved emission for the as-prepared and BDAmodified InP GQDs in toluene can be sufficiently fitted by two single exponential decay kinetics (τ 1, τ 2 ) of (19.5 ns, 68.6 ns) and (23.2 ns, 84.5 ns), respectively (see Fig. 2d and Table 1). Apparently, both τ 1 and τ 2 components increase with BDA post-modification. In general, the shorter τ 1 is related to a high rate of transition of interparticle charge and energy transfer, while the second exponent (τ 2 ) is attributed to charge carrier recombination [50][51][52][53] . The longer τ 1 and τ 2 observed in BDAmodified InP GQDs simply indicate less surface unfilled and oxidative defects, which is attributed to the uniform crystalline orientation on the QD surface 48,54,55 . The BDA modifying oxidation state of Zn in the InP GQDs is supported by XPS (see Supplementary Fig. 4), where the oxidation state of Zn is reduced, rendering the evidence of less surface unfilled and oxidative defects. Upon adding ZnI 2 to the BDA-modified InP GQDs, despite the unchanging of the absorption and emission spectral profile, τ 1 and τ 2 increase to 25.8, 85.5 ns, respectively, supporting the further suppression of the trapping sites.
Device performance. Prior to the device fabrication and performance, we firstly examined if the charge balancing has been improved by post-modification. This has been done by dividing the device structure into two parts, the electron-only and holeonly sections ( Fig. 3a and b), and measured the current density for each site to correlate with the electron and hole mobility, respectively. Each part is mimicking half of the real device and the integration of these two parts exactly reproduces the whole device architecture elaborated later.  Supplementary Fig. 7) suggests that the decrease of the electron mobility is due to the shift of vacuum level induced by the surface dipoles, resulting in the mismatch between the conduction band with the electron transport layer. Conversely, as mentioned early, the BDA post-modification decreases the hole trapping sites. As a result, the current density versus voltage of the hole-only device (HOD) reveals slight enhancement of the hole injection ability for the BDA-modified InP GQDs (see Fig. 3b). Evidently, after BDA modification, the decrease of the electron mobility as opposed to the increase of the hole mobility makes more balance between electron and hole injection. Further improvement can be made by the BDA modification followed by the addition of zinc halides using ZnI 2 as the additive. The results shown in Fig. 3b reveal further increase of the hole injection mobility whereas the change of electron mobility is not as significant. This may be rationalized, on the one hand, by the additional filling of vacancies by the zinc iodide salt, further reducing the hole traps 56 . On the other hand, the weak surface interaction as well as relatively small amount of the passivated iodide does not change the surface dipole and hence the band levels remain relatively unchanged (cf. the BDA modification only). We also noticed that the film morphology of InP GQDs has been improved accordingly via sequence of passivation. Figure 3c-e shows the atomic force microscope (AFM) height  can both anchor Zn on the InP GQD surface, it provides an opportunity to bridge the nearby QDs, stabilizing the film formation. Accordingly, we speculate that the decrease in RMS is related to some BDA bound to the proximal QDs on the surface. The decrease of RMS roughness along the sequence of BDA and then ZnI 2 modification manifests the improvement of the surface smoothness and hence suppression of the leakage current. Figure 4a shows the schematic diagram of the InP GQDs device structure applied in this study. Also depicted in Fig. 4a are the cross-sectional SEM images of the interlayer configuration. Note that the electron transport layer consists of ZnxMg 1-x O where Mg has been fined-tuned to 12.5% for the optimized performance. Table 2 systematically lists the device performance data for the as prepared, BDA and other alkyl diamines modified as well as the combination of BDA and zinc halides modified InP GQDs QLEDs. For each condition, data shown in Table 2 are the average value of a number of devices fabricated. Comprehensive data of all measured devices are listed in Supplementary Table 2 where several checking points can be pointed out here. First, among various alkyl diamines-modified InP GQDs devices, the ethylene 1,2-diamine (EDA)-modified InP GQDs devices has already shown better performance than that of the unmodified InP GQDs device. For example, typical device performance indices such as maximum external quantum efficiency (EQE max ) and maximum current efficiency (CE max ) for EDA modified InP GQDs, measured to be 7.9% and 35.9 cd A −1 , respectively, are higher than that (5.7% and 27.9 cd A −1 ) of the unmodified InP GQDs device. Upon increasing the alkyl lengths of diamines from none, EDA, BDA to HDA, both CE max and external EQE max exhibit a volcano shape-like plot, reaching the pinnacle at BDAmodified InP QLED with CE max and EQE max of 44.3 cd A −1 and 8.5%, respectively (see Table 2). Therefore, BDA with four carbon bridging lengths seems to be optimum for filling the trap sites as well as for tuning the electron/hole balance, consistent with the results of charge mobility measured in the electron and hole-only section of the device (vide supra). Upon further increasing the ligand length to six carbon atoms, i.e., the HDA, the long spatial distance between two terminal amines increases the structure variance, which is unfavorable for the diamino-chelation onto the surface of QDs. This, together with the increase of the hydrophobic/insulation to suppress the charge injection, leads to the inferiority in device performance, being dropping to only 14.4 cd A −1 and 1.8% for CE max and EQE max , respectively. Supplementary Table 2 lists other performance data such as CE, PE, and EQE at 1000 cd m −2 .
Based on the BDA-modified InP GQDs, we then move one step further by additionally passivating zinc halides to examine if the BDA/zinc halides modified InP GQDs can further improve the QLED properties. Note that more balanced electron/hole mobility has been proved by the current density versus voltage of the electron and hole-only device individually (see Fig. 3). Even though the devices show similar performance by passivation of different zinc halides (see Supplementary Table 2), ZnI 2 shown better solubility and hence better device efficiency upon combining BDA, which is supposed to be a synergistic effect. For a fair comparison, the best device performance for BDA/ZnI 2 passivated InP GQDs are depicted in Fig. 4b and c with corresponding data listed in Table 2. Apparently, the BDA/ZnI 2 passivated InP GQDs make leap advance in either the current density and luminescence versus the applied voltage or the We also examined the device performance of adding other zinc halides additives such as ZnCl 2 and ZnBr 2 to the BDA-modified InP GQDs. The resulted data listed in Table 2 and Supplementary  Table 2 all show better performance than that of the BDA modified only InP QLED. Nevertheless, champion performance still belongs to that of BDA/ZnI 2 passivated green InP QLED. It is also noteworthy that emission quantum yield of InP QLED is within 86%. Taking the internal quantum efficiency of 100% and 1/5 of all solid angles covered, a theoretically limited EQE of 17.2% can reach. The champion device of EQE 16.3% is thus close to the theoretical limit. The device also achieves a brightness of 1.26 × 10 4 cd m −2 under a bias of 8.8 V. Figure 5 highlights the systematic improvement in each stage, where the performance of devices treated by BDA/ZnI 2 modification process is clearly better than those of the BDA modified only and the as-prepared ones. As the modification process progresses, the EQE is increasing with the current density of HOD, and decreasing with current density of EOD. Other devices performances shown in Fig. 5, including L max , CE max , and PE max , also reveal superiority to those of the as-prepared one.
Comparison of green QD-LEDs. We then fairly compare the above results with those of the reported green InP QLEDs up to this stage. The corresponding data listed in Table 3 clearly manifest the breakthrough of the device performance in this work. Under the same inverted device structure, this work shows leaping enhancement, in which CE max PE max , and EQE are more than 12 folds, 13 folds, and 1.6 times better than those of the previous best data reported for each single parameter. Further extension to all InP QDs relevant devices, including those of the conventional structures, the current device renders the same level of PE max and luminance, but apparently is superior in terms of the lowest turn-on voltage of 2.2 V, the highest EQE of 16.3% and highest CE max of 57.5 cd A −1 .
Last but not the least, in this study we also carried out the device lifetime measurements to gain further insight into the relationship between electroluminescence efficiency and stability. To our knowledge, there is no practical and meaningful report in regards to the lifetime of the green-emitting InP-based QD-LEDs. Figure 6    shows the evolution of electroluminescence intensities and other parameters for the InP GQD devices operating over a period of 4 h at an initial luminance of 2000 cd m −2 . Also, the T 50 value at 100 cd m −2 was estimated by fitting the data with an empirical equation L n × T 50 = constant, where L is the initial luminance and n is the acceleration factor that was determined by fitting the data shown in Supplementary Fig. 9. As a result, T 50 at 100 cd m −2 increases from 373.9 to 545.8 h upon BDA modification. After BDA/ZnI 2 modification, T 50 at 100 cd m −2 lifts significantly to 1033.4 h. We must admit that the relatively crude packaging system currently existing in the lab cannot satisfy the rigorous lifetime test. Nevertheless, the trend of systematic increase of the device lifespan makes promising the superiority of BDA/ZnI 2 modification in addition to the record high performance in device efficiency.

Conclusion
In summary, we have carried out extensive post-modification of the colloidal green InP QDs by a sequence of ligand exchanges with various alkyl diamines and zinc halides. Our goal is to balance the electron and hole mobility and hence to promote the corresponding QLED performance. The alkyl diamines ligation provides bifunctionality, where upon surface passivation both energy level of CB and VB of InP GQDs lift due to the interface dipole interaction, facilitating the hole mobility, whereas the electron injection is retarded. In addition, the passivation fills the hole trapping sites as well as smoothens the surface of the film. Characterization. Absorption and PL spectra of QDs were recorded by Hitachi U-3310 spectrophotometer and Edinburgh FS980 fluorimeter, respectively. The emission quantum yield of InP QDs was measured by a diluted InP QDs in the toluene where the absorbance at the excitation wavelength (e.g., 450 nm) no more than 0.1 to avoid reabsorption. The integrated emission intensity was then compared with the standard Rhodamine 6 G in EtOH where the quantum yield of the 552 nm emission band was reported to be 95%. The PLQY is then calculated by following equation. Fig. 5 The comparison among as-prepared, BDA-modified, and BDA/ZnI 2 -modified InP GQD devices. a The EQE (red circle) increases in HOD (black square) and decreases in EOD (black triangle) upon increasing the current density. The error bars show standard deviation of as prepared for n = 5, BDA modified for n = 10, and BDA/ZnI 2 modified for n = 30, respectively. b The relative multiples with respect to the as-prepared device for L max , CE max , PE max , and EQE max .
Where Q S is the quantum yield of an unknown fluorescent sample and Q R is the quantum yield of the reference standard. Here, I S and I R are the integrated fluorescence intensities for the unknown and reference samples, respectively, while A S and A R are the respective absorbance values of the unknown and reference samples; n S and n R are the refractive indices of the unknown and reference samples. Cross-sectional scanning transmission electron microscopy (SEM) images were performed using a Hitachi S-4800 field emission scanning electron microscope. Atomic force microscopy (AFM) measurements were conducted on a Utek Material Nanoview 1000 with tapping mode at ambient temperature. The current density − voltage − luminance (J−V−L) and current efficiency-external quantum efficiency-luminance (CE-EQE-L) characteristics were characterized by a CR-280 spectrometer equipped with a source meter (Keithley 2400).

Data availability
Main data supporting the findings of this study are contained within the paper and its associated Supplementary Information. All relevant data of this study are available from the corresponding authors upon reasonable request.
Received: 19 May 2021; Accepted: 1 September 2021;  Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/ licenses/by/4.0/.