Self-assembled monolayer–based blue perovskite LEDs

Abstract

Blue perovskite light-emitting diodes (LEDs) have shown external quantum efficiencies (EQEs) of more than 10%; however, devices that emit in the true blue—those that accord with the emission wavelength required for Rec. 2100 primary blue—have so far been limited to EQEs of ~6%. We focused here on true blue emitting CsPbBr₃ colloidal nanocrystals (c-NCs), finding in early studies that they suffer from a high charge injection barrier, a problem exacerbated in films containing multiple layers of nanocrystals. We introduce a self-assembled monolayer (SAM) active layer that improves charge injection. We identified a bifunctional capping ligand that simultaneously enables the self-assembly of CsPbBr₃ c-NCs while passivating surface traps. We report, as a result, SAM-based LEDs exhibit a champion EQE of ~12% [CIE of (0.132, 0.069) at 4.0 V with a luminance of 11 cd/m²], and 10-fold–enhanced operating stability relative to the best previously reported Rec. 2100-blue perovskite LEDs.

Self-assembled monolayer based blue perovskite LEDs that meet Rec.2100 and exhibit a champion EQE of 12% are reported.

INTRODUCTION

In blue light-emitting diodes (LEDs), it is important to target the Rec. 2100 blue [full width at half maximum (FWHM), <20 nm] to span the full color gamut (1). Chalcogenide quantum dot (QD)–based LEDs have achieved impressive performance along many axes, but in the blue, these suffer from a broad FWHM (38 to 40 nm) (2–4). Metal-halide perovskite colloidal nanocrystals (c-NCs), which show narrow FWHM (~20 nm) (5–6), has enabled external quantum efficiencies (EQEs) of over 20% in the green, red, and near-infrared wavelengths (7–9), and have the potential to meet the requirement of Rec. 2100 blue (CIEy < 0.06) (10). However, until now, the best reported EQE in Rec.2100 blue range is ~10% with CIE coordinates of (0.13, 0.06) (11).

Pure-Br perovskite c-NCs in Rec.2100 primary blue LEDs (11–12) demand strong quantum confinement to meet spectral requirements; however, increased confinement results in c-NCs with a larger surface area–to–volume ratio and higher ligand density. Films formed from multilayer-stacked c-NCs suffer from a charge injection barrier and lower conductivity, hindering LED performance (13). In principle, it is possible to improve charge transport by reducing the active layer thickness to the monolayer limit, an approach expected also to reduce interdot transfer–based quenching and thus increase EQE (14).

The first monolayer LEDs (15), reported two decades ago using CdSe/ZnS QDs, demonstrated the benefits of a monolayer active layer. This early work used an inorganic shell to passivate the CdSe QDs, which allows versatile engineering of solvent polarity between QDs and hole-transporting layers (HTLs) to form a phase-segregated interface. This is of particular importance in monolayer LEDs as the emission region must be finely tuned to be within the active layer and a rough HTL surface could result in unwanted emission. One strategy to avoid this emission is to increase QD size (>10 nm), which has proven effective in red monolayer LEDs (16) but does not apply to blue, because of the coexistence of emission from HTL and QD (17) resulting from the small size required to emit blue. In a recent example from perovskite LEDs, reducing the layer thickness of sky-blue emitting (~480 nm) c-NCs led to HTL emission when the emission layer thickness was below 15 nm (18). In sum, developing monolayer strategies for ultrathin active layers (~5 nm) is a pressing challenge toward the goal of monochromatic and efficient Rec.2100 primary blue perovskite LEDs.

We investigated herein the origins of unwanted emission in monolayer perovskite LEDs when the active layer thickness approaches ~5 nm and found that using available fabrication techniques results in a rough perovskite/HTL interface: This leads to punch-through and direct electrical interaction between HTL and ETL (electron-transporting layer), and consequently, to undesired exciplex emission in LEDs (Fig. 1). We sought to control monolayer interfaces in Rec.2100 primary blue perovskite LEDs and recognized that a well-defined, ordered, and compact monolayer film could suppress HTL/ETL interaction. We reasoned that this could be achieved if we could alter the polarity of the CsPbBr₃ c-NC surface and thereby induce perovskite self-assembly down to the monolayer limit [i.e., self-assembled monolayer (SAM)] through the use of an HTL-compatible ligand. Self-assembled films with ordered nanocrystal arrangement maximize the interactions between nanocrystals and provide homogeneity needed for monolayer films with ~5-nm thickness.

Fig. 1. Interface control to remove exciplex emission.

(A) Previously reported ligands classified by functional group. (B) Electroluminescence at different current densities (10 to 40 mA cm⁻²) from monolayer LEDs that rely on previously reported ligands [type II from (A)]. These exhibit the superposition of exciplex emission and blue emission. Inset: Current density–voltage curves demonstrating a reduction of the injection barrier in the case of monolayer LEDs. a.u., arbitrary units. (C) Schematic illustrating how HTL/perovskite intermiscibility results in HTL/ETL interactions and exciplex emission. (D) Schematic of the HTL-compatible ligand strategy that enables self-assembly of perovskite c-NCs for a clean interface between HTL and perovskite.

We report a ligand strategy that exploits a liquid, short, and nonpolar C─Si bond that is miscible with nonpolar solvents and is HTL compatible. Our strategy passivates QD surface defects while simultaneously providing the needed polarity difference between HTL and perovskite surfaces that, when applied to uniformly sized c-NCs, enables self-assembly (perovskite c-NCs typically fail to self-assemble when the polarities of the nanocrystal solution and the HTL are similar to one another) and avoids dissolution of the HTL substrate (18–20) (fig. S1). In this way, we form a compact uniform monolayer. This enables a record-high EQE (~12%) in 463-nm–centered perovskite LEDs (fig. S2), surpassing the best previously reported Rec.2100 primary blue LEDs.

RESULTS AND DISCUSSION

Design, preparation, and characterization of SAM

We explored reducing the CsPbBr₃ c-NC layer thickness in LEDs to a monolayer (~5 nm) using reported surface ligands (Fig. 1A) and did see evidence of a reduced charge injection barrier in current density–voltage curves (Fig. 1B, inset). However, an extra emission peak in the 500- to 600-nm range appeared along with the perovskite emission (Fig. 1B). Ultraviolet photoelectron spectroscopy (UPS) (fig. S3) showed that this spectral range coincides with the highest occupied molecular orbital–lowest unoccupied molecular orbital difference (~2.1 eV) between HTL and ETL (21), similar to exciplex emission observed in organic LEDS when the HTL and ETL are spatially close (~3 nm) (22–23) (Fig. 1C).

We posited that the HTL may partially dissolve during CsPbBr₃ c-NC processing, forming rough perovskite films that produce pinholes in the active layer, a problem specific to monolayer-QD devices.

To investigate this further, we considered three types of c-NC capping ligand based on functional group: type I: aromatic, type II: ketone/carbonyl-based, and type III: inorganic (Fig. 1A). For the type I and III ligands, residual polar molecules (be these from solvents or ligands) are expected to lead to pinholes and/or the removal of HTLs in monolayer LEDs (see following morphology study); hence, we prepared PVK (polyvinylcarbazole) films and investigated morphology variation after we deposited CsPbBr₃ c-NCs using type II ligands (24–26) (fig. S4). We found that c-NCs capped with carbonyl-based ligands form discontinuous films with pinholes (fig. S4, A and A′): The discontinuous c-NC films and ligand-mediated HTL dissolution (partially mixed into the PVK) account for HTL-ETL interaction and exciplex emission (Fig. 2A and fig. S4A).

Fig. 2. The effect of using the TMSBr-based, solution-phase ligand exchange on the perovskite/HTL interface.

(A to C) Interface mixing illustration between HTL and perovskite (A), cross-sectional SEM (B), and cross-sectional TEM (C) of a control CsPbBr₃ c-NC film with type II ligands (the arrow indicates potential HTL ETL interaction sites). (D to F) Clean interface illustration between HTL and perovskite (D), cross-sectional SEM (E), and cross-sectional TEM (F) of CsPbBr₃ c-NC with the TMSBr treatment. (G) GIWAXS of a control CsPbBr₃ c-NC film on PVK. (H) GIWAXS of TMSBr-treated CsPbBr₃ c-NCs (~15-nm) film on PVK. (I) XPS of control and TMSBr-treated CsPbBr₃ c-NC films.

We then developed a solution-phase ligand exchange that relies on liquid bromotrimethylsilane (TMSBr) (Methods), an approach that led to smoother [0.7 nm peak-to-peak in atomic force microscopy (AFM)] roughness (fig. S4, B and B′). We saw the same trends on other HTLs {Poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine] (PTAA) and poly(N,N'-bis-4-butylphenyl-N,N'-bisphenyl)benzidine (poly-TPD)}, while the TMSBr-treated CsPbBr₃ films remain smooth and homogeneous on these HTLs (fig. S4, C to F and C′ to F′). Large-area AFM (30 μm by 30 μm) and optical microscopy (2000 μm by 2000 μm) indicate a continuous monolayer of dots on HTL (fig. S5).

We evaluated, via scanning electron microscopy (SEM) and transmission electron microscopy (TEM), cross sections of LEDs that show exciplex emission (type II ligands, Fig. 2, A to C) and LEDs using TMSBr-treated cNCs (Fig. 2, D to F). Controls were indeed rough (Fig. 2B), with evidence of direct interaction between HTL and ETL (Fig. 2C). By contrast, cross-sectional SEM (Fig. 2E) and TEM (Fig. 2F) images of LEDs using TMSBr show a uniform active layer thickness of ~5 nm, indicating a monolayer.

We used grazing incidence wide-angle x-ray scattering (GIWAXS) to characterize the self-assembly and ordering of CsPbBr₃ c-NC on HTLs. Controls exhibit isotropic orientation of perovskite (Fig. 2G); by contrast, the TMSBr CsPbBr₃ c-NC films (~15 nm, to improve signal-to-noise ratio) show both vertical and horizontal orientations atop the HTL (Fig. 2H) (18–20), suggesting self-assembly of TMSBr-treated CsPbBr₃ c-NCs. Grazing incidence small-angle x-ray scattering (GISAXS) (fig. S6A) reveals orientational ordering of the monodispersed c-NCs: Azimuthal integration of the diffraction pattern shows a sharp peak with an internanocrystal distance of ~4.8 nm (average c-NC size is ~4.5 nm) (fig. S6C).

We next studied the surface ligands of the TMSBr CsPbBr₃ c-NCs. X-ray photoelectron spectroscopy (XPS) of TMSBr c-NC films shows a distinct Si peak at ~103 eV, while the control films are featureless in this range (Fig. 2I), demonstrating an effective ligand exchange. High-resolution TEM (HRTEM) images of TMSBr c-NCs show an inter-NC distance of ~4.8 nm (center-to-center distance) (fig. S6C): We clarify that these HRTEM images should not be used to compare morphology because of the different substrate (Cu grid) used compared to those in LED fabrication. This is in good agreement with the inter-NC spacing obtained from GISAXS, consistent with a close-packed film and short interparticle distance (distance between c-NCs) of <3 Å.

To understand the origin of pinholes in monolayer films, we used contact angle measurements to investigate the polarity effect of surface ligands on HTLs. The contact angle of CsPbBr₃ c-NC solutions on different surface ligands (types I to III and TMSBr in Fig. 1A) show similar wettability (~10^o) with little contact angle difference (<2^o) (fig. S7, A to D). Similarly, fig. S7 (E to G) shows that changing solvent also has little effect on contact angle (~10° to 12°). We studied further the impact of introducing type I versus III ligands on HTL morphology (fig. S7, H to J). These ligands have low solubility in octane (<2 mg/ml even when sonicated for 12 hours). Only with the addition of polar solvents do they dissolve; yet the addition of polar solvents leads to rough films on PVK, accompanied with pinholes in the monolayer films (fig. S7, I and J). When we use PTAA as the HTL (18), the addition of polar solvents removes the HTL. In contrast, liquid TMSBr is miscible with nonpolar solvents and compatible with HTLs (fig. S7J). These studies of contact angle and the morphology lead us to conclude that wettability is not a limiting factor. Instead, the miscibility of the HTL substrate (which is determined by the polarity of solvent in which the c-NCs are dissolved, the functional groups on the c-NC surface, and the properties/types of the HTL, taken together) determines the monolayer film quality. In the case of thick active layers, slight dissolution at the contact/active-layer interface does not limit LED performance, but in monolayer films, miscibility causes ETL/HTL interactions resulting in detrimental exciplex emission. Levering the managed HTL/perovskite interface, we fabricated monolayer perovskite LEDs, and achieved a twofold higher current density than seen in controls (Fig. 3A), accompanied by emission approaching Rec. 2100 primary blue.

Fig. 3. Steady-state and transient spectroscopy of TMSBr films compared to controls based on type II ligands.

(A) TMSBr monolayer LEDs exhibit Rec.2100 primary blue emission. Inset: Current density–voltage curves showing a twofold increase of current density and reduction of injection barrier. (B) TRPL of type II ligand, TMS, and TMSBr-treated CsPbBr₃ c-NC films. (C) PL stability of TMSBr-treated CsPbBr₃ c-NC films under ambient conditions or heating at 100°C. (D to F) fs-TA of [(D) to (F)] control-ligands–treated CsPbBr₃ c-NC films and (F) TMSBr-treated films after antisolvent washing.

Optical properties and material stability

We then used time-correlated single-photon counting (TCSPC) to investigate charge-carrier kinetics in CsPbBr₃ c-NC films and found that TMSBr-treated films have ~1.5× longer radiative lifetimes than controls (type II ligands) (Fig. 3B), which coincides with the photoluminance quantum yield (PLQY) increase from 59 to 81%. In perovskite c-NCs, most of the traps are due to uncoordinated Pb²⁺ (Lewis acid), which require electron-rich (Lewis base) agents such as SCN⁻, COO⁻, or Br⁻ for passivation (18, 27). In this work, the TMSBr agent is liquid and chemically active, which provides sufficient Br⁻ to coordinate with Pb²⁺ during the ligand exchange process. To understand the role of Br⁻ in passivation, we tried the same ligand exchange process using tetramethylsilane (TMS), the structure which contains TMS but without Br (Fig. 3B). TMSBr-treated films showed a longer radiative lifetime than TMS-treated c-NC films, suggesting that the uncoordinated Pb²⁺ (Lewis acid) is effectively passivated by the electron-rich Br⁻ (Lewis base) provided by TMSBr. The TMSBr-treated CsPbBr₃ c-NC films showed increased stability under heat: no emission shift after 3 hours at 100°C (Fig. 3C) compared to ~5 min for controls. Prior studies have suggested well-passivated, highly ordered films are more robust under accelerated aging conditions (28). Femtosecond transient absorption (fs-TA) shows that, after we wash with methyl acetate, controls exhibit a shifted bleach peak (Fig. 3, D and E) and unstable photoluminance (fig. S8). By contrast, under the same conditions, the TMSBr-treated CsPbB₃ c-NCs have a consistent bleach peak at ~450 nm with no observable shift (Fig. 3F).

Device structure and performance

Encouraged by the improved current density and promising optical properties of the SAM CsPbBr₃ c-NC films, we sought next to translate this into optimized LEDs. We used a configuration consisting of indium tin oxide (ITO as anode, ~32 nm)/ poly(3,4-ethylenedioxythiophene)polystyrene sulfonate (PEDOT:PSS) doped with Nafion perfluorinated (PFI) ionomer [PEDOT:PSS:PFI as hole injection layer (HIL), ~30 nm]/ PVK (~10 nm)/CsPbBr₃ c-NCs (active layer, ~5 nm)/ 3′,3′″,3′″″-(1,3,5-triazine-2,4,6-triyl)tris([1,1′-biphenyl]-3-carbonitrile) (CNT2T as electron-transport layer, ~40 nm)/lithium fluoride (LiF as electron-injection layer, ~1 nm)/aluminum (Al as cathode, ~150 nm) (Fig. 4A). We used PEDOT:PSS:PFI as the HIL in view of its exciton-buffering and hole-injecting capabilities (29–30). We use PVK on top of PEDOT:PSS:PFI (HIL) to decrease the charge injection barrier between the PEDOT:PSS:PFI layer and the CsPbBr₃ c-NC layer (31–32).

Fig. 4. LED performance and operating stability.

(A) Device structure and energy band diagram. The energy levels of the CsPbBr₃ c-NCs were calculated from optical data and UPS (fig. S11). (B to D) Current density–voltage–luminance (B), EQE versus current density (C), and electroluminescence curves (D) of control devices [inset is of a monolayer (~5 nm) control device]. (E to G) Current density and luminance curves (E), EQE versus current density (F), and electroluminescence curves (G) of TMSBr-treated devices at voltages from 3.4 to 5.0 V. (H) EQE histogram of 15 devices based on TMSBr-treated CsPbBr₃ c-NCs. (I) Current density versus voltage curves of hole- and electron-only devices based on TMSBr-treated CsPbBr₃ c-NCs. (J) Operating stability of LEDs based on CsPbBr₃ c-NCs.

Control LEDs have a higher turn-on voltage (>3.8 V turn-on in J-V compared to 3.4 V for TMSBr) and lower maximum luminance (<100 cd/m²) (Fig. 4B); and the electroluminescence of controls shows spectral broadening (FWHM of 29 nm) along with exciplex emission (Fig. 4D). To achieve emission approaching the Rec. 2100 standard blue, we further blue-shift the c-NC emission by applying size engineering (Methods), and achieve c-NCs with emission at 463 nm. Treated LEDs produce an EL spectrum with CIE color coordinates of (0.132, 0.057) at 3.6 V, and the CIE color coordinates shift to (0.132, 0.069) when the voltage reaches 4.0 V. The treated LEDs exhibit a twofold higher current density and a threefold increase in luminance (~300 cd/m²) related to the control (Fig. 4E). Although the electroluminescence peak is stable, the CIEy increases to >0.07 when the voltage is ≥4.0 V because of the long tail at an increased driving voltage (Fig. 4, G and I, and fig. S10). Multilayer and monolayer control CsPbBr₃ c-NC–based LEDs showed EQEs of 0.8 and 0.07%, respectively (Fig. 4C), while TMSBr CsPbBr₃ c-NC–based monolayer LEDs achieved an EQE of 11.9% with an average of (10.5 ± 1.2)% (Fig. 4, F to H, and fig. S9)—the highest among blue perovskite LEDs with a peak emission <470 nm (5, 10, 11, 27, 33, 34).

At a constant current density of 2 mA cm⁻², which provided an initial luminance (L₀) of 100 cd m⁻², the half-life (T₅₀) of luminance [the time by which L₀ decreases to half (L₀/2)] was 17 min (Fig. 4J and fig. S12). Although this stability represents a fivefold improvement on that of the mixed-halide Rec.2100 primary blue perovskite LEDs (10), more work is needed to resolve the fragile perovskite structure under operating conditions and the change from Pb²⁺ to Pb(0) in perovskite LEDs, thus extending the stability.

In summary, SAM CsPbBr₃ c-NC films were synthesized using a bifunctional molecular additive that alters the polarity of surface ligands relative to the HTL and also functions as a surface passivant. This allowed us to fabricate compact, smooth, and uniform monolayer films with high PLQY and well-tuned recombination within the perovskite monolayer. Through the use of a SAM active layer (SAM-AL), we achieve CsPbBr₃ c-NC–based LEDs with a record EQE of 12% approaching Rec. 2100 blue. The SAM-AL strategy contributes to much-needed further progress in the deep-blue region for solution-processed LEDs.

MATERIALS AND METHODS

Materials

Cesium carbonate (Cs₂CO₃, Puratronic, 99.994%, metals basis, Alfa Aesar), lead (II) bromide (PbBr₂, 99%, Sigma-Aldrich), oleylamine (OAm, technical grade 70%, Sigma-Aldrich), oleic acid (OAc, technical grade 90%, Sigma-Aldrich), 1-octadecene (ODE, technical grade 90%, Sigma-Aldrich), methyl acetate (ReagentPlus, 99%, Sigma-Aldrich), Nafion PFI resin solution (tetrafluoroethylene-perfluoro-3,6-dioxa-4- methyl-7-octenesulfonic acid copolymer, 5 wt % in a mixture of lower aliphatic alcohols and water, containing 45% water) and lithium fluoride (LiF, >99.99%) were purchased from Sigma-Aldrich. PEDOT:PSS (Clevios PVP Al 4083) was purchased from Heraeus. All the chemicals were used directly as received.

CsPbBr₃ c-NC synthesis and purification

CsPbBr₃ c-NCs were synthesized and purified on the basis of a previously reported method with some modifications. In brief, 310 mg of Cs₂CO₃, 1.2 ml ofOAc, and 5.0 ml of 1-octadecene (ODE) were dried in a three-necked round-bottom flask at room temperature for 15 min. In another flask, 320 mg of PbBr₂, 320 of mg of ZnBr₂, and 200 of mg SrBr2 were mixed with 4 ml of OAm/OAc mixture and 8 ml of ODE followed by vacuum drying at 120°C for 20 min. Under N₂ atmosphere, the lead halide precursors were kept at reaction temperature (86°C) until all solids were dissolved. Then, 1.8 to 2.0 ml of Cs precursor solution were swiftly injected into the flask containing the lead halide precursor. After a chosen reaction time (2 min), the solution was cooled using an ice water bath. The solution was centrifuged at 7800 rpm to remove unreacted precursors. The precipitation was discarded and the supernatant was collected. For the purification process, the supernatant was precipitated by adding 50 ml of methyl acetate. Then, the mixture was centrifuged to collect the precipitant and dissolved in octane. The addition of SrBr₂ and ZnBr₂ provides better size control to obtain strongly quantum-confined nanocrystals and to blue shift the emission. For the c-NCs in LED fabrication, we applied further a size-engineering process.

Size engineering

We applied a multistep washing process to obtain c-NCs with a uniform size distribution. We added the methyl acetate to the nanocrystal solution until the solution became slightly turbid, then centrifuged for 2 min, and collected the supernatant; we repeated this process >10 times to get c-NCs with a narrow size distribution. We found that after 8 washing steps, we obtained 467-nm emitting dots, and after 2 additional washing steps (10 in total), we obtained 463-nm emitting dots. Using the 463-nm c-NCs, we fabricated monolayer LEDs with emission centered at 463 nm.

Ligand exchange optimization and SAM c-NCs preparation

c-NCs suspended in octane were purified two times by precipitating using methyl acetate. Briefly, ~20 ml of methyl acetate was added into ~5 ml of nanocrystal solution in octane. The mixture was vortexed for 1 min and then centrifuged at 7800 rpm for 2 min. The precipitants were collected and resuspended in ~2 ml of octane. Subsequently, 7 ml of methyl acetate was added to the nanocrystal solution to precipitate the nanocrystals. The same procedure was repeated one time. The final c-NCs precipitants were suspended into ~0.5 ml of octane for ligand exchange.

For the standard ligand exchange process that resulted in 7% EQE LEDs, 800 mg of TMSBr was dissolved into 5 ml of octadecene. Then, ~30 μl of TMSBr solution was added to 0.5 ml of c-NC solution. After ~1-min vortexing, ~6 ml of methyl acetate was added to the mixture to precipitate the c-NCs. After centrifuging, the supernatant was discarded while the sediment was resuspended in ~0.5 ml of octane. Additional centrifugation was necessary in cases where turbidity persisted after resuspension. The exchange process was subsequently repeated one more time with ~30 μl of TMSBr solution. Additional methyl acetate was used to precipitate the c-NCs after each exchange cycle. The exchanged c-NCs were suspended in 1.0 ml of octane. The ink was centrifuged at 14,500 rpm to remove any nonsoluble impurities before storage or use.

Ligand exchange of control samples

Type I, II, and III ligand exchange process: Type I and III ligands are halide salts or inorganic ligands, which are not soluble in nonpolar solvents like octane and toluene. In this case, polar solvents [dimethylformamide (DMF) or dimethyl sulfoxide (DMSO)] are needed to dissolve these ligands. However, these polar solvents could destroy the perovskite c-NCs because of their ionic property. To proceed with the ligand exchange process, 200 mg of type I and III was dissolved in 0.5 ml of DMF with the addition of 0.5 ml of toluene. The mixed of DMF and toluene helped to maximize the passivating effect from the ligands and minimize the negative effects of the polar solvents. Then, ~30 μl of type I and III ligand solution was added to 0.5 ml of c-NC solution. After ~1-min vortexing, ~6 ml of methyl acetate was added to the mixture to precipitate the c-NCs. After centrifuging, the supernatant was discarded while the sediment was resuspended in ~0.5 ml of octane, repeating the above ligand exchange process one more time. The exchanged c-NCs were suspended in 1.0 ml of octane. The ink was centrifuged at 14,500 rpm to remove any nonsoluble impurities before storage or use.

The type II ligands are partially soluble in the toluene, and the exchange process is the same as above except that there is no addition of DMF or DMSO polar solvents.

Contact angle

Contact angle measurement was done with a captive-drop goniometer (CAM 101, KSV instrument Ltd., Finland). Briefly, 1.5-μl droplets of solutions were dropped onto a polymer surface that was aligned on the focal plane of the magnifying camera of the goniometer, using a calibrated P10 pipette. An image of the droplet is captured within 2 s after the droplet is formed to prevent loss by evaporation. The image recorded was then analyzed for the contact angle on both sides, using the internal setting of octane (γ = 21.62 mN/m) and toluene (γ = 28.50 mN/m).

PL characterization

A Horiba Fluorolog system was used for PL characterization. Steady-state PL was measured with a monochromatized Xe lamp as the excitation source. A TCSPC detector and a pulsed ultraviolet laser diode (λ = 374 nm) were used to acquire transient PL. An instrument response function of Δt = 0.13 ns provides a limit to the overall time resolution. Time-resolved emission spectra were recorded by measuring individual transient PL traces at increasing emission wavelengths. PL stability was measured in ambient conditions with a relative humidity of approximately 20%, using the laser diode (Delta Diode 375, peak wavelength 374 nm, 100 MHz) with a peak power of 300 mW and average power of 1.8 mW. The spot size was ~0.01 cm², so the calculated excitation density was 180 mW cm⁻². Absolute PLQY values were obtained by coupling a Quanta-Phi integrating sphere to the Fluorolog system through optical fiber bundles. All the PLQY measurements followed published methods. Both excitation and emission spectra were measured for three cases: the sample directly illuminated by the excitation beam path in the integrating sphere, the sample offset within the integrating sphere from the beam path, and the empty sphere itself. For PLQY measurements, the Fluorolog was set to an excitation wavelength of 380 nm and a 2-nm bandpass for both the excitation and emission slits. By using these settings, the resulting spectra had high signal-to-noise ratios and provided an excitation intensity in a range of 1 to 30 mW cm⁻². The detector was calibrated for spectral variance with a Newport white light source. The excitation intensity was varied for intensity-dependent PL spectra by changing the slit width on the Fluorolog monochromator. Excitation intensity was obtained by recording the power with an Ophir LaserStar Dual Channel Power and energy meter and by calculating the beam area through the known dispersion relations for the monochromator.

TA measurements

A regeneratively amplified Yb:KGW laser (PHAROS, Light Conversion) was used to generate femtosecond laser pulses at a wavelength of 1030 nm as the fundamental beam with a repetition rate of 5 kHz. The fundamental beam was passed through a beam splitter, where most of the beam was used to pump an optical parametric amplifier (ORPHEUS, Light Conversion) to serve as a narrow-band pump (pulse duration ~200 fs, FWHM ~10 nm). The remaining part of the beam was focused on a translating sapphire crystal to generate a white light probe ranging between 400 and 800 nm. The pump and probe pulses were directed into a commercial transient absorption spectrometer (Helios, Ultrafast). The probe pulse was sent to a retroreflector mounted on a delay stage where multiple reflections off the retroreflector allowed for a delay relative to the pump pulse of up to 8 ns. Sample measurement was obtained with pump powers at 70 μW and a spot size of 0.3 μm² (assumption-Gaussian beam profile).

AFM and optical microscope measurements

AFM measurement was performed with an Asylum Research Cypher AFM operated in AC mode in air. Imaging was done using ASYELEC-02 silicon probes with titanium-iridium coatings from Asylum Research. The probes had a typical spring constant of 42 N m⁻¹. Optical microscope images we obtained by using a Zeiss X10 optical microscope.

SEM and TEM measurements

Cross-sectional SEM images were collected in secondary electron mode by a Hitachi SU5000. Samples for top-view TEM imaging were prepared above onto a copper TEM grid attached on a glass substrate by using the same spin-coating and annealing method used in LEDs. Cross-sectional samples for TEM were prepared using a cryo–focused ion beam (Hitachi NB5000). The accelerating voltage of the ion beam was 40 kV, with a gradual decrease in the accelerating voltage to 2 kV as the sample thickness decreases. Cryo-TEM were performed using a Hitachi HF-3300 scanning/transmission electron microscope, operated at 300 kV.

GIWAXS and GISAXS measurement

GIWAXS and GISAXS were conducted at the Hard X-ray MicroAnalysis beamline of the Canadian Light Source. An energy of 17.998 keV (λ = 0.6888 Å) was selected using a Si(111) monochromator. Patterns were collected on a SX165 CCD camera (Rayonix) placed at a distance of 175 mm from the sample. A lead beamstop blocked the direct beam. Images were calibrated using LaB₆ and processed with the Nika Igor plug-in and the GIXSGUI MATLAB plug-in.

XRD measurements

XRD measurement was conducted using a Rigaku MiniFlex 600 diffractometer (Bragg-Brentano geometry) equipped with a NaI scintillation counter detector and a monochromatized Cu Kα radiation source (λ = 1.5406 Å) operating at a voltage of 40 kV and a current of 15 mA. The crystallite size was calculated using Scherrer Equation

$$\mathrm{\tau }=\frac{K\mathrm{\lambda }}{\mathrm{\beta }\cos \mathrm{\theta }}$$

where τ is the mean size of the crystalline domains, which may be smaller or equal to the grain size; K is a dimensionless shape factor, with a value close to unity; λ is the x-ray wavelength; β is the line broadening at half the maximum intensity, after subtracting the instrumental line broadening, in radians; and θ is the Bragg angle.

UPS measurements

UPS measurement was performed on ITO substrates with high conductivity. Photoelectron spectroscopy was performed in a PHI5500 Multi-Technique system using nonmonochromatized He-Iα radiation (hv = 21.22 eV). All the work function and valence band measurements were performed at a take-off angle of 88°, with the base chamber pressure of 10⁻⁷ Pa. A bias of −15 V was applied to measure the work function.

LED fabrication

LEDs were fabricated on patterned ITO-coated glass substrates purchased from Thin Film Devices Incorporated. The thickness of the glass substrate is 0.70 ± 0.002 mm. The thickness of the ITO layer is 320 ± 25 Å, the resistivity is 100 ± 25 ohms sq.⁻¹, and the transmittance is >80% at 475 ~ 800 nm. We cleaned the substrates with detergent, deionized water, acetone, and isopropanol in an ultrasonic washer and then treated them with oxygen plasma for 5 min. Then, a mixed solution of PEDOT:PSS (Clevios PVP AL 4083):PFI (at a mass ratio of 5:1) was spin coated at 3000 rpm for 20 s, followed by annealing on a hotplate at 100°C for 10 min. After the substrate was cooled to room temperature, PVK in chlorobenzene solution (4 mg/ml) was spin coated onto the PEDOT:PSS:PFI layer and baked for 30 min at 170°C. CsPbBr₃ c-NC was then spin coated onto the PVK at 2000 rpm for 30 s. Last, the substrates were transferred into a high-vacuum thermal evaporator, where CNT2T (~40 nm), LiF (~1 nm), and Al (~150 nm) were deposited thereon layer by layer through a shadow mask at a pressure below 10⁻⁸ torr.

LED evaluation

The device active area was 6.14 mm² as defined by the overlapping area of the ITO and Al electrodes. The devices were encapsulated before the current density and luminance measurements, using an ultraviolet curable resin (exposure under ultraviolet light for 20 s) and covered on the edges between the device and a transparent glass chip. All the devices were tested under ambient conditions. All devices were tested under ambient conditions. The luminance versus voltages and the current density versus voltage characteristics were collected using a HP4140B picoammeter. The absolute EL power spectra were collected using an integrating sphere and an Ocean Optics USB4000 spectrometer by mounting the devices on the wall of the integrating sphere. The EQEs were then calculated through the measured absolute EL power spectra and the current density. The devices were mounted on the open aperture of the integrating sphere to allow the light emitted from the glass surface to be collected, while the emission from the substrate edges was not collected. We have reproduced the measurements multiple times and also for several repetitions of the same experiment. The lifetime of encapsulated devices was measured in ambient conditions. LEDs were driven using a Keithley 2400 source meter at a constant current, and the luminance intensity was measured with a QEPro calibrated spectrometer (Ocean Optics). The current used to drive the LED for stability tests was first determined using an EQE measurement system. The radiance of our spectrometer (QEPro, Ocean Optics) used in our system was calibrated by a radiometric calibrated light source (HL-3P-INT-CAL, Ocean Optics), and to validate our test system, we cross-checked the whole system using a commercial planar organic light-emitting diodes (purchased from Lumtec. Corp., T70 > 10,000 hours) with a reference system that is separately calibrated in a different laboratory, the EQE values measured at two different systems gave similar results (average variations <2%) (fig. S13).

Supplementary Materials

This PDF file includes:

Figs. S1 to S13