Introduction

Direct emission of circularly polarized light (CPL) by circularly polarized light-emitting diodes (CP-LEDs) with chiral materials is highly desirable for three-dimensional displays, quantum information processing, and other applications1,2,3. However, CP-LEDs with both a high electroluminescence dissymmetric factor (gEL) and a high external quantum efficiency (EQE) are considerably challenging to realize (Supplementary Note 1)4,5,6, limiting their potential utilization. Chiral perovskites possessing efficient chiral-induced spin selectivity (CISS)7,8,9,10 effect have recently demonstrated spin-LEDs (a branch of CP-LEDs) with direct CPL emission (Supplementary Note 2)11,12,13,14,15,16,17,18,19,20. CISS effect of chiral perovskites efficiently generates a dissymmetric population of spin-polarized carriers, and this electrical spin-selective process can potentially amplify chiroptoelectronic properties of resulting devices despite the inferior chiroptic properties of chiral perovskites themselves, offering opportunities for high-performance LEDs with direct CPL emission.

Despite that, perovskite spin-LEDs heavily rely on two-dimensional (2D) and quasi-2D chiral perovskites, and are underexplored in terms of material properties and device mechanisms, leaving substantial room for improvement regarding their gEL and EQE (Supplementary Note 2). Pioneer perovskite spin-LEDs start with 2D chiral perovskites with efficient spin selectivity and fabrication simplicity8,21. However, steric hindrance of chiral molecules impedes the crystallization of 2D chiral perovskites, resulting in structural defects that accelerate non-radiative carrier losses22. Therefore, 2D chiral perovskites exhibit low photoluminescence quantum yield (PLQY) and are usually adopted as a spin-selective layer rather than an emission layer in spin-LEDs. In such a configuration, spin relaxation that impairs gEL can occur in adjacent achiral emission layer before the radiative recombination of spin-polarized carriers for CPL emission. Besides, unidirectional injection of spin-polarized carriers and non-radiative carrier losses at the 2D chiral perovskites are also unfavorable for realizing high gEL and EQE, respectively. Alternatively, quasi-2D chiral perovskites consisting of chiral lower-dimensional and achiral three-dimensional (3D) components possess intrinsic energy and spin funneling structures that promote the transfer of spin-polarized carriers to emissive 3D component, enabling their utilization as a CPL emission layer. However, relaxation of spin during their transfer from 2D/quasi-2D chiral components to 3D achiral components can be detrimental to gEL, as well as the inhomogeneous external injection of spin-unpolarized carriers from electrodes directly to 3D achiral components. Meanwhile, the presence of chiral lower-dimensional perovskites that are usually accompanied by relatively low crystalline quality can also result in non-radiative carrier losses and inferior EQE. To simultaneously realize high gEL and high EQE values in perovskite spin-LEDs, it is, therefore, a prerequisite to generate a highly dissymmetric population of spin-polarized carriers, suppress spin relaxation, and inhibit non-radiative losses of spin-polarized carriers.

Here, we show that high gEL and high EQE values can be simultaneously realized in spin-LEDs based on chiral perovskite quantum dots (PQDs). Specifically, chiral PQDs are shown to possess spin selectivity due to the CISS effect. Considering the large exciton binding energy resulting from the geometric confinement of PQDs, spin-polarized carriers in chiral PQDs are prone to form spin exciton and rapidly recombine for the emission of CPL. In this case, chiral PQDs functionalize as localized spin-selective units and localized radiative recombination centers at the same time (Fig. 1a), and spin relaxation during the transfer and diffusion of spin-polarized carriers in mainstream spin-LEDs based on 2D/quasi-2D chiral perovskites can be suppressed. Prior to that, it is requisite to ensure efficient spin selectivity and outstanding optoelectronic properties in chiral PQDs to maximize the population of spin-polarized carriers, as well as prevent undesirable non-radiative carrier losses. We show that the spin selectivity and optoelectronic properties of chiral PQDs are highly related to the coverage of chiral ligands that simultaneously imprint chirality/spin selectivity and passivate surface defects (Fig. 1a), and an enhanced ligand exchange strategy assisted by ultrasonic (US) treatment is proposed herein to facilitate the two critical factors. As a proof-of-concept, spin-LEDs based on the resulting chiral CsPbBr3 PQDs with R-/S-methylbenzylamine (R-/S-MBA) simultaneously demonstrate high gEL (0.285 for R-LEDs and 0.251 for S-LED) and high EQE (16.8% for R-LEDs and 16.0% for S-LED). The proposed ligand exchange strategy is also compatible with PQDs with different compositions, as well as various types of chiral ligands such as aromatics, alkanes, and amino acids. As a result, circularly polarized photoluminescence (CP-PL) ranging from 465 nm to 660 nm is demonstrated, which is beneficial for designing spin-LEDs with different wavelengths with a single chiral semiconductor platform. The above results reveal the potential of chiral PQDs for high-performance and multi-wavelength LEDs with direct emission of CPL.

Fig. 1: Optimization of chiroptic properties and spin selectivity in chiral PQDs for spin-LEDs.
Fig. 1: Optimization of chiroptic properties and spin selectivity in chiral PQDs for spin-LEDs.The alternative text for this image may have been generated using AI.
Full size image

a Schematic illustration of the mechanism of spin-LEDs based on chiral PQDs. ex: exciton, OAm: oleylamine, OAc: oleic acid. b CD and c CPL spectra of chiral CsPbBr3 PQDs with different ligand exchanged processes. US ultrasonic treatment. The corresponding d gabs and e glum values of chiral CsPbBr3 PQDs with different ligand exchange processes. f Schematic illustration of mCP-AFM measurements. Averaged I-V curves under different injected spin polarization of samples based on R-PQDs g with and h without US treatment. P: spin polarization efficiency. 30 I-V curves measured from different locations of a sample are averaged.

Results

Correlation between chirality and spin selectivity

Ligand exchange strategies have demonstrated their capabilities in realizing high-quality PQDs23,24 and high-performance optoelectronics25,26,27,28,29, providing feasibility to the synthesis of ready-to-use chiral PQDs that are compatible with the fabrication of spin-LEDs (Supplementary Note 3). Ligands directly determine the optoelectronic properties and functionalities of PQDs30,31,32,33, and they undergo highly dynamic desorption/adsorption processes due to the ionic nature of PQDs. Through the replacement of existing oleic acid (OAc)/oleylamine (OAm) ligands with chiral ligands, chirality can be effectively transferred to PQDs. However, solely relying on the natural competition between the original OAc/OAm ligands and the sequentially introduced chiral ligands (this strategy refers to ligand exchange strategy without US treatment herein) potentially yields limited ligand exchange efficiency. In this case, it is difficult to imprint sufficiently strong chirality that induces efficient spin selectivity into PQDs.

Here, an enhanced ligand exchange strategy assisted by US treatment is adopted, and the US treatment is anticipated to improve the exchange efficiency of chiral ligands by assisting the desorption of original OAc/OAm ligands. As an example, chiral CsPbBr3 PQDs with classic chiral molecules R-/S-MBA synthesized by ligand exchange strategies with and without US treatment are investigated. Specifically, achiral CsPbBr3 PQDs capped with OAc/OAm ligands are synthesized by conventional hot-injection methods (Supplementary Fig. 1), and chiral ligand exchange strategies with and without US treatment are conducted. Both types of chiral PQDs exhibit circular dichroism (CD) signals at around 515 nm (Fig. 1b). Besides, CPL signals at similar wavelengths arise from both types of chiral PQDs (Fig. 1c). These phenomena verify the replacement of OAc/OAm ligands by chiral ligands, as well as the chirality transfer from chiral ligands to PQDs8,34,35. More importantly, chiral PQDs with US treatment exhibit much-enhanced chiroptic properties when compared with those of chiral PQDs without US treatment (Fig. 1b, c) while their corresponding gabs and glum values are also significantly higher (Fig. 1d, e). We anticipate that the enhanced chiroptic properties are ascribed to the promoted exchange efficiency of chiral ligands during US treatment.

Given the difference in chiroptic properties between the two types of chiral PQDs, their corresponding spin selectivity is then characterized by the magnetic conductive probe atomic force microscopy (mCP-AFM) with a Co/Cr-coated tip that injects carriers with a single spin polarization state (denoted as spin-up or spin-down). R-PQDs are used for demonstration with a structure of ITO/poly-SAM/R-PQDs (Fig. 1f), which mimics the operational conditions of devices. Spin polarization efficiency (P) that quantifies the CISS effect can be calculated by the difference between spin-up current (Ispin-up) and spin-down current (Ispin-up) injection:

$$P=\frac{{I}_{{{\rm{spin}}}-{{\rm{up}}}}-{I}_{{{\rm{spin}}}-{{\rm{down}}}}}{{I}_{{{\rm{spin}}}-{{\rm{up}}}}+{I}_{{{\rm{spin}}}-{{\rm{down}}}}}\times 100\%$$
(1)

For thin-film R-PQDs with ultrasonic (US) treatment, P-values at 2 V and 3 V, which are close to the operational voltages of devices, are respectively calculated to be 86% and 89% (Fig. 1g). Such a result demonstrates the efficient spin selectivity of thin-film R-PQDs with US treatment despite their inferior chiroptic properties (Fig. 1d, e), which potentially results in amplified chiroptoelectronic properties in devices. Comparably, P-values at 2 and 3 V of the sample based on R-PQDs without US treatment are respectively determined to be 36% and 27% (Fig. 1h). The large difference of P between thin-film R-PQDs with and without US treatment also agrees with their difference of glum, indicating that the chirality of chiral PQDs directly determines the magnitude of spin selectivity.

Synergistic enhancement of optoelectronic properties

Besides imprinting chirality/spin selectivity, the binding of chiral ligands can simultaneously passivate surface defects on PQDs. Therefore, chiroptic and optoelectronic properties can be synergistically facilitated in chiral PQDs, which is highly desirable for high-performance spin-LEDs. Thin-film chiral PQDs with US treatment possess narrower diffraction peaks when compared with those of the thin-film chiral PQDs without US treatment during x-ray diffraction (XRD) characterizations (Fig. 2a). These results indicate that R/S-MBA can potentially passivate excess surface defects resulting from the weakly bonded OAc/OAm ligands on PQDs and, therefore, promote crystalline quality and suppress non-radiative carrier losses. Consequently, carrier dynamics that are strongly associated with non-radiative carrier losses are then studied by photoluminescence quantum yields (PLQY) and time-resolved photoluminescence (TRPL) measurements. Solution based on chiral PQDs with US treatment exhibit a high PLQY of 94% while that of solution based on chiral PQDs without US treatment is shown to be 86% (Fig. 2b), suggesting the suppression of non-radiative carrier losses in the former type of chiral PQDs. Carrier lifetimes revealed by TRPL measurements demonstrate a similar trend where thin-film chiral PQDs with US treatment show an enhanced carrier lifetime of 9.08 ns measurement when compared with that of thin-film chiral PQDs without US treatment (5.20 ns, Fig. 2c).

Fig. 2: Optoelectronic properties of chiral PQDs.
Fig. 2: Optoelectronic properties of chiral PQDs.The alternative text for this image may have been generated using AI.
Full size image

a XRD, b steady-state PL and PLQY, and c TPRLspectra of chiral CsPbBr3 PQDs with and without US treatment. I-V characteristic curves of devices based on chiral PQDs d without and e with US treatments. VTFL trap-filling limit voltage. f C-ω and g Nt-Eω spectra of devices based on chiral PQDs with and without US treatment. Chiral PQDs used: R-PQDs.

Space-charge limit current (SCLC) measurements are carried out to electrically quantify the trap density and resulting non-radiative carrier losses in both types of chiral PQDs. Specifically, the device based on chiral PQDs without US treatment exhibits a trap-filling limit voltage (VTFL) of 1.64 V, and the corresponding trap density (ntrap) was calculated to be 1.2 × 1018 cm−3 (Fig. 2d). As a direction comparison, the device based on chiral PQDs with US treatment possesses reduced VTFL (0.96 V) and ntrap (7.0 × 1017 cm−3) (Fig. 2e). Meanwhile, capacitance-frequency (C-ω) spectra of devices based on the two types of chiral PQDs show a distinct difference in magnitude (Fig. 2f). Device based on chiral PQDs without US treatment is shown to possess a higher capacitance, which is associated with the capture of carriers due to presence of higher density of traps36. Trap density of state (Nt) as a function of trap energy (Eω) can be further converted by using data of C-ω spectra. Considering that the binding of ligands is highly related to surface defects, we anticipate that these defects predominantly result in the formation of deep-level traps in the band37. Specifically, measured frequency ranging from 103 Hz to 105 Hz can be converted to Eω ranging from 0.3 eV to 0.43 eV, corresponding deep-level traps38. Device based on chiral PQDs with US treatment exhibits a distinct reduction of trap density by almost one order of magnitude when compared with that of the device based on chiral PQDs without US treatment (Fig. 2g), indicating the suppression of deep-level traps resulting from the passivation of surface defect. The above results show that optoelectronic properties of chiral PQDs can be synergistically facilitated with their chirality/spin selectivity, which is beneficial for enhancing the performance of spin-LEDs.

Mechanism of the enhanced ligand exchange strategy

To understand the underlying mechanism of the synergistically enhanced chiroptic and optoelectronic properties in chiral PQDs with US treatment, a transmission electron microscope (TEM) is carried out to characterize the morphological information of PQDs. Figure 3a shows the TEM image of the as-synthesized achiral PQDs where PQDs with clear boundaries and uniform size distribution (Supplementary Fig. 2a) could be identified. To solely study the desorption of OAc/OAm ligands, different ligand exchange processes without the addition of chiral ligands are performed on achiral PQDs. Specifically, achiral PQDs are either sonicated (denoted as achiral PQDs with US treatment) or stirred (denoted as achiral PQDs without US treatment) for an equal duration of 5 min. Compared with achiral PQDs without US treatment (Fig. 3b), a higher density of merged PQDs (marked by white dash boxes) can be observed in achiral PQDs with US treatment (Fig. 3c). Meanwhile, the size distribution profile of achiral PQDs with US treatment characterized by dynamic light scattering (DLS) is significantly broadened (Supplementary Fig. 2c) due to the presence of merged PQDs. This phenomenon can be attributed to the removal of OAc/OAm ligands, showing that US treatment promotes the exchange efficiency by assisting the removal of weakly bonded OAc/OAm ligands. Besides, the broadening of PL emission peak (Supplementary Fig. 3a) in achiral PQDs with US treatment further supports the above conclusion. By adding chiral ligands during the US treatment, chiral ligands effectively bond with the exposed surface of PQDs, and chiral PQDs with relatively uniform distribution can be obtained (Fig. 3d and Supplementary Fig. 2d), which can be further evident by the reduction of PL full-width at half-maximum (FWHM, Supplementary Fig. 3b).

Fig. 3: Mechanism of the enhanced ligand exchange strategy.
Fig. 3: Mechanism of the enhanced ligand exchange strategy.The alternative text for this image may have been generated using AI.
Full size image

TEM images of a as-synthesized achiral CsPbBr3 PQDs, b achiral CsPbBr3 PQDs without US treatment, i.e., achiral CsPbBr3 PQDs after continuous stirring, c achiral CsPbBr3 PQDs after US treatment without the addition of chiral ligands, and d chiral CsPbBr3 PQDs after ligand exchange with US treatment. Scale bar: 30 nm. Inset: corresponding HRTEM of different types of CsPbBr3 PQDs. Scale bar: 2 nm. e N 1s core level spectra and f FTIR spectra of different types of PQDs. Calculated g S(C=O)/S(-COO−) and h S(C=C)/S(C-H) values obtained from FTIR spectra. i 1H-NMR spectra of chiral ligands and chiral CsPbBr3 PQDs dispersed in toluene-d8. j 1H-NMR spectra of different types of PQDs dissolved in DMSO-d6. k Calculated density density and chiral ligand exchange efficiency of different types of PQDs dissolved in DMSO-d6. Chiral PQDs used: R-PQDs.

Meanwhile, crystallographic information of the above PQDs can be extracted from TEM images. In principle, the removal of surface ligands can release the surface distortion of PQDs and increase crystal plane spacing. For the as-synthesized achiral PQDs, crystal plane spacing was determined to be 4.11 Å (inset of Fig. 3a) while that of the achiral PQDs without US treatment slightly increased to 4.12 Å (inset of Fig. 3b). As a direction comparison, crystal plane spacing of achiral PQDs with US treatment significantly increases to 4.18 Å (inset of Fig. 3c), agreeing with the conclusion that US treatment effectively remove OAc/OAm ligands during the ligand exchange processes. By introducing chiral ligands during the US treatment, the crystal plane spacing of chiral PQDs is restored to 4.13 Å (inset of Fig. 3d), implying that the chiral ligands successfully bond with PQDs. Meanwhile, XRD spectra show that the (100) diffraction peak of achiral PQDs with US treatment shifts to a lower angle when compared with that of the as-synthesized achiral PQDs (Supplementary Fig. 4). This phenomenon refers to the expansion of perovskite lattices, which agrees with the increment of crystal plane spacing caused by the release of surface distortion due to the desorption of OAc/OAm ligands (Fig. 3a–c). Contrarily, the (100) diffraction peak of chiral PQDs retained its position, suggesting that the chiral ligands can bond with PQDs after the removal of OAc/OAm and protect the lattice structure of CsPbBr3 PQDs (Supplementary Fig. 4).

X-ray photoelectron spectroscopy (XPS) is performed to qualitatively study the ligand exchange efficiency and the surface chemistry of PQDs (Supplementary Fig. 5), and a weak shoulder peak corresponding to the benzene ring π-π conjugation in the chiral ligand R/S-MBA can be identified at around 288 eV in the C 1s core level peak of chiral PQDs (Supplementary Fig. 6), demonstrating the bonding of chiral ligands on PQDs. Meanwhile, N 1s core level peaks can be deconvoluted into two distinct peaks (Fig. 3e). Peaks at around 400 eV are attributed to ligands with -NH2 terminals while those located at around 402 eV are attributed to ligands with -NH3+ terminals. For achiral PQDs, N 1 s core level peaks solely originate from OAm ligands. OAm ligands undergo dynamic proton exchange with OAc ligands23,39. Therefore, both peaks corresponding to -NH2 and -NH3+ can be identified in the XPS spectrum of achiral PQDs (Fig. 3e), and part of OAm ligands with -NH3+ terminals can be considered as X-type ligands (one-electron donors) that weakly bind with PQDs31. Ligands with -NH2 terminals (two-electron donors) can be considered L-type ligands that bind stronger with PQDs. During the ligand exchange processes, OAc/OAm ligands are replaced by R-/S-MBA ligands with -NH2 terminals. In this case, the dynamic proton exchange processes between OAc/OAm ligands are suppressed, and the content of -NH2 in N 1s core level peaks increases. Correspondingly the -NH2/-NH3+ ratio increases from 2.3:1 in achiral PQDs to 4.3:1 in chiral PQDs without US treatment and 5.5:1 in chiral PQDs with US treatment (Fig. 3e), showing that the US treatment can improve the chiral ligand exchange efficiency. Correspondingly, Cs 3d (Supplementary Fig. 7a), Pb 4f (Supplementary Fig. 7b), and Br 3d (Supplementary Fig. 7c) core level peaks of chiral PQDs with US treatment also show an obvious shift, referring to the change of surface chemistry of PQDs after ligand exchange. Benefiting from the strengthened binding of ligands, chiral PQDs also possess enhanced stability by suppressing the natural desorption of ligands (Supplementary Fig. 8).

Furthermore, Fourier transform infrared (FTIR) spectra of achiral PQDs and chiral PQDs with different ligand exchange strategies (without purification) are studied (Supplementary Fig. 9). Peaks at 1710 cm−1 and 1530 cm−1 are respectively attributed to C=O in OAc ligands (including attached and detached OAc) and -COO− in attached OAc ligands (Fig. 3f), and, therefore, the ratio of the area of the C=O vibrational peak at 1710 cm−1 (S(C=O)) to that of the -COO− vibrational peak at 1530 cm−1 (S(-COO−)) reflects the amount of attached OAc ligands on PQDs. The S(C=O)/S(-COO−) value increases from 0.09 in achiral PQDs to 0.42 in chiral PQDs without US treatment, which further increases to 0.79 in chiral PQDs with US treatment (Fig. 3g). Such increment in S(C=O)/S(-COO−) generally indicates the desorption of OAc molecules after ligand exchange processes, as well as that the US treatment can effectively promote the detachment of OAc ligands. Meanwhile, Peaks at 1580 cm−1 and 1466 cm−1 are respectively attributed to C=C in chiral R-MBA ligands and C-H in OAc/OAm ligands (Fig. 3f), and the ratio of the area of the C=C vibrational peak at 1580 cm−1 in R-MBA (S(C=C)) to that of the C-H vibrational peak at 1466 cm−1 in OAc/OAm (S(C-H)) can be used to evaluate the ligand exchange between R-MBA and OAc/OAm. The S(C=C)/S(C-H) value increases from 0 in achiral PQDs to 0.25 in chiral PQDs without US treatment, which further increased to 0.47 in chiral PQDs with US treatment (Fig. 3h), indicating the increment of R-MBA on PQDs with the assist of US treatment. We also show that the C=C vibrational peak position of bonded chiral ligand R-MBA (1580 cm−1) is shifted when compared with that of pristine R-MBA (1599 cm−1, Supplementary Fig. 9b), which further proves the coordination between R-MBA molecules and CsPbBr3 PQDs. Meanwhile, the chemical shift (δ) of the chiral ligand R-MBA on chiral PQDs that are dispersed in toluene-d8 exhibited an obvious shift when compared with those of R-MBA in toluene-d8 (Fig. 3i and Supplementary Fig. 10) as characterized by proton nuclear magnetic resonance (1H-NMR) measurement, which can be ascribed to the interaction between chiral ligands and PQDs. Specifically, δ at around 1 ppm originates from the active hydrogens on -NH2 terminals of R-MBA, thereby exhibiting a pronounced shift due to the direct interaction between -NH2 terminals with PQDs (Fig. 3i).

Based on the above qualitative analysis, 1H-NMR measurements are then carried out on solutions of PQDs that dissolved in DMSO-d6 to quantify the ligand density and the chiral ligand exchange efficiency. δ at around 4.1 and 5.3 ppm can be respectively attributed to hydrogen atoms in R-MBA and OAc/OAm (Fig. 3j). Concentrations of R-MBA are then determined to be 0, 2.3, and 6.4 mM in achiral PQDs, chiral PQDs without US treatment, and chiral PQDs with US treatment, respectively. Meanwhile, Concentrations of OAc/OAm are then determined to be 12.9, 12.3, and 8.2 mM in achiral PQDs, chiral PQDs without US treatment, and chiral PQDs with US treatment, respectively. Combining the averaged size of PQDs (Supplementary Fig. 11), the density of different ligands, as well as the chiral ligand exchange efficiency, are calculated and summarized in Fig. 3k and Supplementary Table 1. Chiral ligand exchange efficiency can be effectively promoted from 15.7% to 43.9% with the help of US treatment that assists the desorption of OAc/OAm ligands. We also note that the correlations between chiroptic properties of chiral PQDs and experimental parameters during the enhanced ligand exchange processes are studied for optimization (Supplementary Figs. 12–14, as well as Supplementary Tables 2–4).

Tunable chiroptic properties

Besides chiral CsPbBr3 PQDs with R/S-MBA, the proposed strategy also shows compatibility with various PQDs and chiral ligands, which is beneficial for designing spin-LEDs with different emission wavelengths, as well as versatile passivators toward high-performance spintronics. Here, chiral CsPbCl1.5Br1.5 and CsPbI3 PQDs with R/S-MBA ligands are realized by the enhanced ligand exchange strategy for demonstration. The onsets of CD signals for chiral CsPbCl1.5Br1.5 and CsPbI3 PQDs are respectively determined to be 460 and 680 nm (Fig. 4a, b), and CPL can also be detected at corresponding wavelengths (Fig. 4c, d, chiral CsPbCl1.5Br1.5 ~ 465 nm, chiral CsPbI3 ~ 660 nm). Meanwhile, other types of chiral ligands besides aromatics (R/S-MBA), including alkanes (R/S-OcAm, Supplementary Fig. 15a) and amino acids (R/S-AOA, Supplementary Fig. 15b), are investigated based on achiral CsPbBr3 PQDs, and both CD (Fig. 4e, f) and CPL (Fig. 4g, h) signals can be detected from chiral CsPbBr3 PQDs with different types of chiral ligands.

Fig. 4: Tunable chiroptic properties of chiral PQDs.
Fig. 4: Tunable chiroptic properties of chiral PQDs.The alternative text for this image may have been generated using AI.
Full size image

CPL spectra of a chiral CsPbCl1.5Br1.5 PQDs and b chiral CsPbI3 PQDs with R-/S-MBA. CD spectra of c chiral CsPbCl1.5Br1.5 PQDs and d chiral CsPbI3 PQDs with R-/S-MBA. CPL spectra of chiral CsPbBr3 PQDs with e R-/S-OcAm and f R-/S-AOA. CD spectra of chiral CsPbBr3 PQDs with g R-/S-OcAm and h R-/S-AOA. The above results show that the enhanced ligand exchange method is compatible with PQDs with different compositions and different types of chiral ligands.

Spin light-emitting diode characterizations

As a proof-of-concept, spin-LEDs with an ITO/poly-SAM/chiral PQDs/PO-T2T/TPBi/LiF/Al structure (Supplementary Fig. 16) based on chiral CsPbBr3 PQDs with both R- and S-MBA (denoted as R-LED and S-LED) are demonstrated. Green-light emission with a center wavelength located at around 517 nm can be identified from the electroluminescence (EL) spectra from both R- and S-device (Fig. 5a). Besides, a turn-on voltage of around 2.1 V and maximum luminance over 28,000 cd m−2 can be extracted from the current density-luminance-voltage (J-L-V) characteristics curves (Fig. 5b) of the two types of devices (R-LED: 28,630 cd m−2, S-LED: 32,864 cd m−2), which is higher than the existing perovskite spin-LEDs in literature. Accordingly, the champion R- and S-device, respectively, show peak external quantum efficiency (EQE) values of 16.8% and 16.0% (Fig. 5c), which are both the highest values in the existing perovskite spin-LEDs. Peak current efficiencies of R- and S-LED are respectively determined to be 57.2 and 53.2 cd A−1 (Supplementary Fig. 17a) while peak power efficiencies of R- and S-LED are respectively determined to be 69.1 and 57.8 lm W−1 (Supplementary Fig. 17b).

Fig. 5: Spin-LEDs based on chiral CsPbBr3 PQDs with the enhanced ligand exchange strategy.
Fig. 5: Spin-LEDs based on chiral CsPbBr3 PQDs with the enhanced ligand exchange strategy.The alternative text for this image may have been generated using AI.
Full size image

a EL, b J-L-V, and c EQE spectra of spin-LEDs based on chiral CsPbBr3 PQDs with R- and S-MBA. Corresponding σ+ and σ− emission spectra of d R-LED and e S-LED. f Wavelength-dependent gEL of R-/S-LEDs. g Summary of gEL and EQE of perovskite spin-LEDs in the literature and this work. h Time-dependent EL intensity of R-LED at an initial luminance of 100 cd m−2. i T50 of R-LEDs at different initial brightness levels.

On the basis of the above optoelectronic performance, we then characterize the circularly polarized electroluminescence (CP-EL) performance of our spin-LEDs. Specifically, a home-customized measurement system is built to simultaneously measure the left-hand circularly polarized light (σ+) and right-hand circularly-polarized light (σ−) so that possible artifacts and device-to-device variation can be avoided (Supplementary Fig. 18). A distinct intensity difference between σ+ and σ− emission can be characterized from both R-LED (Fig. 5d) and S-LED (Fig. 5e), demonstrating the direct emission of CPL in our devices. gEL values of R- and S-LED are then respectively characterized to be 0.285 and 0.251 across 510 to 525 nm (Fig. 5f, the range of full-width at half-maximum of emission peaks), which also surpass those of state-of-the-art perovskite spin-LEDs. We also note that gEL values of our spin-LEDs are significantly larger than glum values of chiral PQDs, whether in solutions (Fig. 1e) or in the form of thin films (Supplementary Fig. 19 and Supplementary Table 5), while being in the same order of magnitude with P-values of thin-film chiral PQDs (Fig. 1g). Consequently, we anticipate that the efficient spin selectivity, rather than the optical stereogenic centers, of chiral PQDs contributes to the amplification of gEL in spin-LEDs despite the inferior glum of chiral PQDs.

More importantly, our devices are found to realize high EQE and high gEL values at the same time when compared with the existing perovskite spin-LEDs (Fig. 5g), showing the potentials of spin-LEDs based on chiral PQDs. We anticipate that chiral PQDs can serve as localized spin-selective units to generate spin-polarized carriers. Spin-polarized carriers can form spin excitons, which then rapidly recombine at the localized area due to the geometric confinement and large exciton binding energy of PQDs for CP-EL emission, suppressing undesirable spin relaxation during the transfer and diffusion of spin-polarized carriers as in spin-LEDs based on mainstream 2D/quasi-2D chiral perovskites. In addition, the enhanced chiral ligand exchange efficiency synergistically promotes the chirality/spin selectivity, as well as the optoelectronic performance, of chiral PQDs to generate highly dissymmetric population of spin-polarized carriers and suppress non-radiative carrier losses of carriers in devices. Together with the rapid localized radiative recombination of spin-polarized carriers, the resulting spin-LEDs can simultaneously exhibit high gEL and EQE.

Operational stability of devices that are equally important for their practical utilization is also characterized by using R-LED as an example, and no EL decay can be observed in an operational time of 220 min (Fig. 5h, at an initial luminance of 100 cd m−2). To estimate the half lifetime (T50, the time when luminance diminishes to 50% of its initial value) at an initial luminance of 100 cd m−2, we characterize T50 values at an initial luminance of 2000, 5000, 8000, 10,000, and 20,000 cd m−2 (Supplementary Fig. 20). T50 values under these conditions can be extracted to estimate T50 at 100 cd m−2 (Fig. 5i). Fitted T50 at 100 cd m−2 is then determined to be around 19.8 h.

As a direct comparison, spin-LEDs based on chiral PQDs without US treatment exhibit inferior chiroptoelectronic performance (Supplementary Fig. 21). Maximum luminance values for both R- and S-device are reduced to around 2000 cd m−2, as shown in J-L-V characteristics curves, and the peak EQE values are also reduced to 1.3% and 1.6%, respectively. Meanwhile, gEL values of R- and S-LEDs are respectively calculated to be 0.048 and 0.043, which are also evident by the negligible difference between σ+ and σ− emission from the corresponding device. The above EQE and gEL values are significantly lower than those of spin-LEDs based on chiral PQDs with US treatment, which is due to the insufficient chiral ligand coverage on PQDs. In this case, both surface defect passivation and chirality transfer through the bonding of chiral ligands are difficult to realize, leading to inferior chiroptoelectronic performance in the resulting spin-LEDs.

In conclusion, we have demonstrated the capability of chiral PQDs in constructing high-performance spin-LEDs that simultaneously exhibit high gEL and high EQE. Chiral PQDs are shown to possess spin selectivity that generates a dissymmetric population of spin-polarized carriers due to the CISS effect. Meanwhile, chiral PQDs can concurrently serve as localized radiative recombination centers of spin-polarized carriers for CP-EL emission due to the geometric confinement and large exciton binding energy of PQDs, thereby suppressing the relaxation of spin-polarized carriers. An enhanced ligand exchange strategy is proposed to enhance the chiral ligand exchange efficiency, which synergistically enhances the spin selectivity and optoelectronic properties of chiral PQDs. This strategy is also applicable to various types of PQDs and chiral ligands, which is beneficial for designing multi-wavelength spin-LEDs and versatile passivators for high-performance spintronics. Spin-LEDs based on chiral CsPbBr3 PQDs with R-/S-MBA are demonstrated as a proof-of-concept. For R-LEDs, EQE and gEL are respectively shown to be 16.8% and 0.285. For S-LEDs, EQE and gEL are respectively shown to be 16.0% and 0.251. Compared with the existing perovskite spin-LEDs, EQE and gEL in this work are both among the highest. We anticipate that the understanding of this work can potentially trigger the development of spin-LEDs based on chiral PQDs, providing insights into the construction of chiral light sources.

Methods

Materials

Lead bromide (PbBr2, 99.99%), lead chloride (PbCl2, 99.99%), lead iodide (PbI2, 99.99%), cesium carbonate (Cs2CO3, 99.99%), octadecene (1-ODE, ≥95%), Chlorobenzene (CB, 99%) methyl acetate (MeOAc, ≥99%) were purchased from Macklin Biochemical. Oleylamine (OAm, 80–90%), oleic acid (OAc, 85%), toluene-d8 (99.60%), (R)-2-Aminooctane (R-AmOc, 98%), (S)-2-Aminooctane (S-AmOc, 99+%) and Nickel Acetate (NiAc, 99.9% metals basis) were purchased from Aladdin. R-methylbenzylamine (R-MBA, 99%) and S-methylbenzylamine (S-MBA, 98%) were purchased from TCI. Ethanol (EtOH, 99.9%), (R)-2-Aminooctanoic Acid (R-AOA, 99%), and (S)-2-Aminooctanoic Acid (S-AOA, 97%) were purchased from Adamas. Toluene-d8 (Tol, ≥99.5%) and DMSO-d6 (Tol, ≥99.5%) were purchased from Sigma-Aldrich. 1,3,5-Tris(1-phenyl-1H-benzimidazol-2-yl)benzene (TPBi, 99%), 2,4,6-tris[3-(diphenylphosphinyl)phenyl]-1,3,5-triazine (PO-T2T, 99%), and poly-SAM (poly-4PACz, 99%) were purchased from Xi’an Yuri Solar Co., Ltd. Aluminum was purchased from ZhongChengXinCai Co., Ltd. All of the reagents were used as received without further purification.

Synthesis of achiral CsPbBr3\CsPbCl1.5Br1.5\CsPbI3 PQDs

Achiral CsPbBr3 PQDs were synthesized by mixing Cs2CO3 solution and PbBr2 solution. Specifically, Cs solution (0.4 g of Cs2CO3 and 2 mL of OAc in 15 mL of ODE) was firstly stirred in a three-necked round-bottom flask under vacuum at 90 °C for 1 h, which was then stirred under N2 atmosphere at 130 °C and was kept at this temperature. PbBr2 solution (160 mg of PbBr2 in 10 mL of 1-ODE) was stirred in a three-necked round-bottom flask under vacuum at 130 °C for 1 h. After that, 1 mL of OAc and 1 mL of OAm were injected into the flask. The PbBr2 solution was further stirred under N2 atmosphere at 130 °C until it became clear, which was then stirred under N2 atmosphere at 180 °C for another 1 h. After that, 0.8 mL of the above Cs2CO3 solution was quickly injected into the Pb precursor solution to obtain CsPbBr3 PQDs. At the 5th second after the injection of Cs2CO3 solution, the flask was immersed in an ice bath to quench the growth of CsPbBr3 PQDs. The as-synthesized CsPbBr3 PQDs crude solution was separated into two centrifuge tubes (each tube contained 6 mL of CsPbBr3 PQDs crude solution), and 18 mL of MeOAc was added to each tube for purification. The solutions were then centrifuged with a centrifuge force of 5630×g for 5 min, and the precipitates were collected and redispersed with 6 mL of toluene. The purification process was repeated for another than 1 time. After that, the CsPbBr3 PQDs solution was further centrifuged with a centrifuge force of 630×g for 5 min, and the supernatant was collected. To synthesize CsPbCl1.5Br1.5 PQDs and CsPbI3 PQDs, PbBr2 solution was replaced by PbClBr solution (80 mg of PbBr2 and 60 mg of PbCl2 in 10 mL of 1-ODE) and PbI2 solution (200 mg of PbI2 in 10 mL of 1-ODE), respectively.

Ligand exchange strategies for chiral PQDs

To conduct the ligand exchange with US treatment, 20 μL of R/S-MBA was added into 1 mL CsPbBr3 PQDs toluene solution. For ligand exchange with different chiral ligands, R/S-MPEA (10 μL), R/S-AmOc (10 μL), R/S-AOA (30 mg) and D/L-NE (30 mg) can be used. Subsequently, the vials containing the mixtures are placed in ice bath and are subjected to tip sonication (tip diameter: 2 mm, frequency: 25 kHZ, SCIENTZ-IID) at different power (100, 200, 400, 600, and 800 W, optimized: 400 W) for different duration (1, 2, 3, 5, 7, and 9 min, optimized: 5 min). Finally, the solution was centrifuged with a centrifuge force of 630×g for 5 min, and the supernatant was collected to get rid of large particles and by-products. To synthesize the chiral PQDs without US treatment, 10 μL of R/S-MBA and 1 mL CsPbBr3 PQDs toluene solution was added into a 5 mL bottle with a stirrer, and then, the mixed solution was placed and stirred on a magnetic stirrer apparatus at 1000 rpm for 5 min.

Material characterizations

CD measurements were carried out using a Jasco chiralscan and CPL measurements were conducted with a Jasco CPL-300. Thin-film chiral PQDs were deposited onto K9 glasses by spin coating while solutions of chiral PQDs were placed in quartz cuvettes. gabs was calculated by \({g}_{{{\rm{abs}}}}={CD}/32,980\times {{\rm{Abs}}}\) where CD (mdeg) represents the difference of absorbance between left-hand and right-hand CPL and Abs is the absorbance. glum was calculated by \({g}_{{{\rm{lum}}}}=2\times \left({I}_{{{\rm{PL}}}-{{\rm{L}}}}-{I}_{{{\rm{PL}}}-{{\rm{R}}}}\right)/({I}_{{{\rm{PL}}}-{{\rm{L}}}}+{I}_{{{\rm{PL}}}-{{\rm{R}}}})\) where IPL-L and IPL-R are the intensity of left- and right-handed CPL from PL spectra, respectively. gEL was calculated by \({g}_{{{\rm{EL}}}}=2\times \left({I}_{{{\rm{EL}}}-{{\rm{L}}}}-{I}_{{{\rm{EL}}}-{{\rm{R}}}}\right)/({I}_{{{\rm{EL}}}-{{\rm{L}}}}+{I}_{{{\rm{EL}}}-{{\rm{R}}}})\) where IEL-L and IEL-R are the intensity of left- and right-handed CPL from EL spectra, respectively. mCP-AFM measurements were carried out with a Bruker Dimension Icon. A Co/Cr-coated tip (MESP-V2) was used during the measurement. The tip was magnetized by a magnet for 1 h before each trail of measurement to ensure the injection of carriers with a single spin polarization state (denoted as spin-up or spin-down). 30 I-V curves measured from different locations of a sample are averaged for each trial of measurement. SCLC measurements were carried out with a semiconductor analyzer (Keithley 4200 A) based on devices with a structure of ITO/PEDOT: PSS/chiral PQDs/TPBi/LiF/Al. C-ω measurements were carried out with a semiconductor analyzer (Keithley 4200 A) based on devices with a structure of ITO/poly-SAM/chiral PQDs/PO-T2T/TPBi/LiF/Al. Nt-Eω spectra were converted from C-ω spectra. Specifically, \({E}_{{{\rm{\omega }}}}={kT} \, {\mathrm{ln}} \, ({\omega }_{0}/\omega )\) and \({N}_{{{\rm{t}}}}\left({E}_{{{\rm{\omega }}}}\right)=-{V}_{{{\rm{b}}}}{dC}(\omega )/{qkAtTd}{\mathrm{ln}}(\omega )\) where Vb is the build-in potential, q is the element charge, k is the Boltzmann constant, A is the device area, t is the thickness, T is the temperature, and ω0 is the attempt-to-escape frequency (~2 × 1010 s−1)36. TEM measurements were carried out with a FEI Tecnai G2 TEM. Specifically, CsPbBr3 PQDs solutions were dropped on the carbon-coated copper mesh grids. XRD measurements were conducted on an XtaLAB Synergy. XPS measurements were performed on a PHI VersaProbe 4 surface analysis system using Al as the excitation source. The samples for XPS measurements were prepared by dropping PQD solutions on the K9 glasses and then annealing at 60 °C for 5 min in an N2-filled glovebox. 1H-NMR measurements were taken on a Quantum-I 400M NMR using a standard proton pulse (zg). Each measurement was performed with 64 scans, 4.0 s collection times, and a 25.0 s delay between scans. PLQY measurement of CsPbBr3 PQDs solutions was carried out with an integrating sphere, a spectrometer, and a 405 nm laser (BQ405). FTIR measurements were carried out with an infrared Fourier transform spectrometer (ATR). For FTIR measurements, background was taken on blank K9 glass substrates, and subsequent sample measurements were taken with an average of 64 scans. For TRPL measurements, samples were excited at 400 nm with an amplified femtosecond laser (Spectra-Physics). After that, the emission spectra were collected by a commercial streak camera (XOPM streak camera 5200).

Spin-LED fabrication

100-nm patterned ITO substrates were sequentially cleaned by sonication in deionized water, acetone, and 2-propanol for 10 min each, followed by ultraviolet ozone treatment for 10 min. Subsequently, 50 μL of poly-SAM solution (1 mg mL−1 in a 1:1 mixed solution of chlorobenzene and ethanol) was spun on the substrates at 4000 rpm for 40 s and annealed at 90 °C for 10 min. 100 μL of PQD solutions were spun onto the substrates at 2000 rpm for 40 s. The samples were transferred to a thermal evaporator. 10 nm of PO-T2T, 5 nm of TPBi, 1.5 nm of LiF, and 100 nm of Al electrodes were sequentially evaporated.