Understanding the Ligand Effects on Photophysical, Optical, and Electroluminescent Characteristics of Hybrid Lead Halide Perovskite Nanocrystal Solids.

There has been a tremendous amount of interest in developing high-efficiency light-emitting diodes (LEDs) based on colloidal nanocrystals (NCs) of hybrid lead halide perovskites. Here, we systematically investigate the ligand effects on EL characteristics by tuning the hydrophobicity of primary alkylamine ligands used in NC synthesis. By increasing the ligand hydrophobicity, we find (i) a reduced NC size that induces a higher degree of quantum confinement, (ii) a shortened exciton lifetime that increases the photoluminescence quantum yield, (iii) a lowering of refractive index that increases the light outcoupling efficiency, and (iv) an increased thin-film resistivity. Accordingly, ligand engineering allows us to demonstrate high-performance green LEDs exhibiting a maximum external quantum efficiency up to 16.2%. The device operational lifetime, defined by the time lasted when the device luminance reduces to 85% of its initial value, LT85, reaches 243 min at an initial luminance of 516 cd m-2.

The emergence of colloidal NCs of hybrid lead halide perovskites has generated considerable research effort aimed at demonstrating their optoelectronic devices, including the light-emitting diodes (LEDs).[1−4] Compared to the bulk counterparts,[5−10] the perovskite NCs have three important merits in LEDs. (i) The versatile colloidal chemistry enables the stabilization of emissive cubic or orthorhombic phases at room temperature across a wider range of anion and cation mixing.[11] (ii) The ligand-induced dielectric confinement increases the thin-film photoluminescence (PL) quantum yield, ηPL, and lowers the refractive index, n, together boosting the theoretical efficiency upper limit in device. (iii) The surface ligand layer surrounding individual NCs may hinder ionic diffusion of ionic species, thereby enhancing the device stability upon electrical stress. Early investigations had been focused on the synthetic approaches giving the desirable properties, including the bandgap tunability and narrowband emission. Later on, further research has highlighted the importance of surface ligands in determining the perovskite NC characteristics. Indeed, the nature of high surface-to-volume ratio induces surface traps and dangling bonds that often promote nonradiative recombination[12] and deteriorate the colloidal and chemical stabilities.[13−15] Accordingly, ligand passivation becomes increasingly attractive to enhance their characteristics.[16,17] As the ionic perovskite NCs are usually stabilized in nonpolar solvents, amphiphilic ligands, including the primary and secondary aliphatic amines,[18] aliphatic carboxylic acids,[19−22] bidentate ligands,[20,23] aromatic amines/acids,[24,25] zwitterionic surfactants,[26] and quaternary ammonium surfactants,[27] had been explored. However, beyond the enhancement of chemical stability and ηPL, although a number of reports have discussed about the ligand effects on the NC morphologies,[28,29] it remains unclear how they influence the photophysical, optical, and transport characteristics of the resulting NC solids, which act as the emission layer in device. A more complete fundamental picture and analysis are required to elucidate the ligand effects on the perovskite LED characteristics. In the present work, using primary alkylamines as the model ligand system, we aim to systematically investigate the ligand effects on the electroluminescent (EL) characteristics in LEDs.

Effects on Morphologies. The binary cations hybrid perovskite NCs with chemical formula FA0.5MA0.5PbBr3, where FA = formamidinium, CH3(NH2)2+ and MA = methylammonium, CH3NH3+, were synthesized at room temperature using the ligand-assisted reprecipitation (LARP) technique (see Methods).[11] Alkylamine amphiphiles with different hydrophobic tail groups, including hexyl- (C6), octyl- (C8), decyl- (C10), dodecyl- (C12), tetradecyl- (C14) and hexadecyl- (C16), were used to control the ligand hydrophobicity.

First we discuss the ligand effects on the NC morphologies. Figure presents the cryogenic transmission electron micrographs of synthesized NCs. Upon increasing the alkyl group length from C6 to C16, we observed a gradually morphological transition from nanoplatelets (NPLs) to nanocubes by reducing the lateral dimension. The statistical analysis shows (Figure b) that the average NC lateral size changes from 15.4 ± 3.9 to 6.4 ± 0.9 nm. The NPLs synthesized with the short ligands have an average thickness of approximately 5.5 nm (Figure S1). The observed ligand effects on size and shape basically follow the same trend for the CsPbBr3 NC system synthesized using the hot-injection method,[28] but we did not find NPL thinning upon decreasing the ligand length. Clearly, by increasing the length of hydrophobic tail, the ligand molecules become more hydrophobic, thereby lowering the interfacial tension between the ligand-passivated surface and nonpolar solvent, which facilitates nucleation at an early stage that limits the NC size. Upon assembly to form NC solids, the interparticle distance also increases from 1.1 ± 0.1 nm (C6) to 2.1 ± 0.1 nm (C16) (Figure inset). We recognize the grafting density and molecular structure of surface ligands also influence the size of nanocrystals.[28,30]

Figure 1

Ligand effects on the NC morphologies. (a) TEM images of synthesized perovskite NCs with different alkyl chain lengths from C6 to C16 (scale bar: 20 nm). (b) Histograms of lateral size distribution by analyzing the TEM images with respect to alkyl chain length. Inset: Interparticle distance as a function of alkyl chain length.

Effects on Photophysical Properties.Figure presents the ligand effects on the photophysical properties of perovskite NCs. The absorption and PL spectra (Figure and Figure S3) reveal an increase of the optical bandgap for the two long-ligand NC systems (C14 and C16); the solution and film PL maximum blue shifts from 531 to 510 nm and from 526 to 513 nm, respectively. The small degree of quantum confinement cannot be completely explained by the reduction of lateral size (Figure ), since the smallest dimension for these NCs remains larger than the Bohr radius. We further examined the X-ray diffractograms for the NC solids (Figure S2) and found that although all NC samples are of the cubic phase, the lattice constant exhibits a small degree of contraction, 5.86 and 5.82 Å for C14 and C16 NCs, respectively, as compared to that for the rest of the NCs (5.90 Å).[31−35] Some degree of peak splitting was also observed. Since the surface-to-volume ratio reaches maximum for C14 and C16 NCs, we hypothesized that the observed lattice contraction results from a degree of stoichiometric imbalance that leads to some phase separation within individual NCs, presumably owing to inhomogeneous adsorption of ligands on the NC surface.

Figure 2

Ligand effects on the photophysical characteristics of NC solids. (a) PL emission spectra of NC solid samples with different alkyl chain lengths from C6 to C16. (b) Solution and thin-film ηPL values as a function of alkyl chain length. (c) Thin-film ηPL versus time for the 6 NC solid samples considered here. (d) Time-resolved PL response for the 6 NC solid samples at room temperature. (e) Extracted average excited state lifetime, τavg, in NC solution and solids as a function of alkyl chain length. (f) τavg of NC solid samples as a function of temperature.

Under the assumption that radiative recombination within individual NCs is fully contributed by excitons, we estimated the exciton binding energy, EB, of the NC solids by characterizing the PL intensity at different temperatures, T, followed by fitting with the Arrhenius relation (Figure S4), revealing a monotonic increase from 65 meV (C6) to 131 meV (C16). These values are all higher than that reported in the bulk.[36] The observed trend is not surprising, because the NC matrix is composed of the high-dielectric-constant (high-ε) perovskite lattice and low-ε ligands, surrounding individual NCs. Accordingly, a larger interparticle distance due to a longer ligand would effectively lower the surrounding dielectric constant giving a stronger dielectric confinement effect. Figure characterizes the solution and thin-film ηPL as a function of alkyl chain length. The solution ηPL values are within the range of 85 ± 2% and nearly independent of ligand length. Nevertheless, interestingly, the thin-film ηPL increases with the ligand length, going from 89% (C6) to 100% (C16). We attributed the anomalous increase of ηPL to the “aggregation-induced emission (AIE)” phenomenon reported by our group,[37] hypothetically because of more restricted motion of surface cations induced by the compression of neighboring ligands. We further examined the shelf stability in ambient (at room temperature with the relative humidity of ∼55%) by monitoring ηPL with time (Figures and S5). The NC thin films with C8, C10, and C12 exhibit excellent stability, while a small degree of decrease is observed in C14 and C16 samples, presumably due to the more contracted lattices. Some samples experienced a small degree of ηPL increase during early times, which may result from the surface trap healing through physisorption of oxygen and moisture molecules from air.[38]

The ligand-induced dielectric quantum confinement effect is also reflected by the exciton lifetime.[39]Figure compares the room-temperature time-resolved PL (TRPL) responses considering different alkyl chain lengths. The average exciton lifetimes, τavg, were calculated by fitting with the biexponential functions, decreasing from 144.6 (C6) to 8.2 (C16) ns in both solution and thin-film samples (Figure e). In addition to the dielectric quantum confinement effect, we recognize that the free carriers also considerably contribute to the radiative recombination,[11] so the long-ligand-induced reduction of NC lateral size (see Figure ) results in confined diffusion that can also shorten the lifetime. Interestingly, exciton lifetimes in NC solids remain nearly unchanged as compared to those in solutions, which suggest that the inter-NC exciton diffusion does not quench PL. In all solid samples considered, τavg decreases with lowering temperature (Figure and Figure S6). As the NC size is larger than the Bohr radius, upon lowering temperature, the probability of exciton dissociation to free carriers becomes smaller due to lowering of thermal energy. As a result, the measured exciton lifetime becomes smaller.

The PL emission bandwidth, corresponding to the full-width at half-maxima, Γ, decreases with lowering temperature (Figure S7) for all ligands. We numerically fit the observed temperature-dependent inhomogeneous Γ broadening, Γ(T), considering the exciton–phonon coupling,[40] which nicely describes all experimental data (Figure S7). The interesting ligand effects on the transition of interactions between excitons and acoustic/LO phonons may be of importance for further characterization.

Effects on Optical Properties. Following the path of organic LEDs,[41−45] the optical properties of NC solids determine the light outcoupling efficiency, ηout, which corresponds to the fraction of light escaping from the LED dielectric stack, deducting from the interfacial total-internal-reflection (TIR) and Purcell-effect losses.[46] We notice that, however, the ligand effects on the thin-film optical properties are largely ignored in literature. First we characterized the refractive indices of the NC solid samples using ellipsometry, which all nicely follow the Sellmeier dispersion equation.[47] As expected, since the refractive index of the NC solid, nNC, is equivalent to within the visible frequency region, a longer ligand leads to a lower nNC (Figure c). Note that if exciton transition dipole moments (TDMs) are randomly oriented, the outcoupling efficiency is approximately given by .[48]

Figure 3

Ligand effects on the optical characteristics of perovskite NC solids. (a) Experimental (exp) and optical simulation (sim) radiation patterns characterizing the PL intensity as a function wave vectors k and k. (b) Comparison of experimental and simulated s-pol (k/k0 = 0) and p-pol (k/k0 = 0) cuts of the radiation patterns. (c) Ellipsometry-characterized refractive indices of the NC solids as a function of alkyl chain length. (d) Extracted probability of horizontal dipole, ΘH, and calculated light outcoupling efficiency, ηout, as a function of alkyl chain length. (e) Calculated ηout as a function of ETL thickness using the extracted ΘH values with different alkyl chain lengths. (f) Calculated power dissipation to outcoupled, substrate, absorption, waveguide, and evanescent fractions as a function of ΘH for the C10 NC solids in device.

Each NC solid sample (air/NC solid film/glass) was attached to a hemispherical glass prism, followed by carrying out the polarization- and angle-dependent PL spectroscopy[49] that differentiates between the p-polarized (p-pol) emission from the transverse-magnetic (TM) dipoles and the s-polarized (s-pol) emission from the transverse-electric (TE) dipoles (Supporting Information section 1.8). The generated radiation pattern (e.g., Figure and Figure S8) resolves the PL intensity I on the substrate plane (x–y) projection of emission wave vector k, k and k, which informs the TDM orientation in the NC solids within the k space domain, k/k0 < nglass, where k0 is the wave vector in air and nglass = 1.52 is the refractive index of substrate. It follows that k/k0 = 1 corresponds to the critical angle of TIR at the glass/air interface. Optical simulations were then performed to fit the p-pol profile (e.g, Figure b), using the probability of horizontally oriented TDMs, ΘH, as the only fitting parameter. Note that isotropic TDM orientation corresponds to ΘH = 0.67;[50] a higher ΘH value implies that the TDM orientation is preferably horizontal, so that more radiation can be coupled into air (k/k0 < 1).[50]

Figure d compares that extracted ΘH values for the NC solids considered here. We find the samples can be categorized into two groups: (i) the NC solids with short alkyl chains (C6, C8, and C10) giving ΘH = 60 ± 1.8%, a slight degree of vertical orientation, and (ii) the NC solids with long alkyl chains (C12, C14, and C16), giving ΘH = 67 ± 1.3%, isotropic orientation. Interestingly, the former is composed of NPLs with the surface normal vector predominantly perpendicular to the substrate plane, as revealed by AFM (Figure S1). The geometrical confinement appears to not promote the horizontal alignment. Clearly, the NPL thickness is larger than the exciton Bohr radius, so the degree of quantum confinement is not sufficient to restrict the TDM orientation. The vertical orientation may result from a relatively large difference between nNC (1.73–1.76) and nglass (1.52) in the NPL NC solids (Figure S9), inducing a vertical electric field that polarizes excitons. This effect, on the other hand, becomes less profound in the cubic NC solid samples due to their relatively low nNC values (1.59–1.64).

We further carried out optical simulations to estimate ηout as a function of the electron transport layer (ETL) thickness, dETL, in device (Figure ; Supporting Information section 1.9). The ηout-dETL curves are oscillatory due to a Purcell effect of the resonant cavity formed in the dielectric stacks. At a practically relevant ETL thickness, dETL = 45 nm, the calculated ηout increases from 20 ± 0.7 (C6) to 25 ± 1% (C16) (Figure e). More simulations were performed to analyze ηout, together with other losses, as a function of ΘH.[51,52] Taking the C10 NC solid as an example (Figure f), if the exciton dipoles can be perfectly horizontal (ΘH = 1), ηout reaches its maximum at 44%. We anticipate that engineering the optical properties of the perovskite NC films would be essential to take the device performance to the next level.

Effects on Electroluminescent Properties. The NC solid thin films were employed in LED devices to examine their EL characteristics. After extensive process optimization, we applied the following device architecture: indium tin oxide (ITO; 120 nm)/PEDOT:PSS (32 nm)/perovskite NC film (15–30 nm)/3TPYMB (45 nm)/LiF (1 nm)/Al (70 nm) (Figure S11). Parts a and b of Figure compare the device current density, J, and luminance, L, as a function of voltage, V, respectively. The characterized turn-on voltage, Von, and the maximum efficiencies ηCE, ηext, and ηPE, corresponding to the current efficiency, external quantum efficiency, and power efficiencies, respectively, are shown in Table S1. A clear trend is that both J and L decrease with the alkyl chain length, and Von increases with the alky chain length. It reflects the impeded inter-NC charge transfer due to the insulating nature of the alkyl chains.[53]

Figure 4

Ligand effects on the EL characteristics of perovskite NC solids. (a) Current density as a function of voltage. (b) Luminance as a function of voltage. (c) Maximum current efficiency (ηCE) as a function of alkyl chain length. (d) EL spectra with different alkyl chain lengths from C6 to C16.

Figure presents ηCE as a function of the alkyl chain length. Except for the two long alkyl chain length samples, with increasing length, the device performance is decent with ηCE (ηext) spanning from 18.3 (3.8%) to 38.1 (9.9%) cd A–1, reaching the maximum at C10 but dropping again in the C12 samples, exhibiting 27.1 cd A–1 (6.8%) (Figure S13). The observed trend for the ηext change may be qualitatively explained by the trade-off between the transport and the ηout/ηPL characteristics, in which the former decreases while the latter increases with the alkyl chain length (see Figures and Figure c). The LEDs made by C14 and C16 NC solids exhibit relatively poor performance, despite the NC solids have near-unity ηPL. We notice that, however, in dichalcogenide (e.g., CdSe) NC-based LEDs, capping with long-chain organic ligands and high-bandgap shell still gives exceptional efficiencies, and the thin-film ηPL seems to be a dominant factor in these devices.[54−56] More research efforts are required to elucidate fundamental principles governing EL performance of semiconductor NCs. Figure shows the normalized EL spectra for the NC solid samples with different alkyl chain length, which are nearly identical to the PL spectra (see Figure a).

By further optimizing the NC layer thickness, Figure presents the J–L–V characteristics of our champion perovskite NC device using the C10 alkyl chain, which was reached with the NC solid film thickness of 32 ± 2 nm (Figure a). Moreover, the decylamine capped NC thin-film deposited on PEDOT:PSS is smooth and nearly pinhole free (Figure S14). The maximum ηext reaches 11.7% (Figure b), which is among one of very few reports demonstrating an ηext of >10% in green perovskite NC LEDs.[57−62] Combining with the simulated ηout (see Figure d), the maximum internal quantum efficiency, ηint, is therefore estimated to be ηext/ηout ∼ 60%. We further utilized the light outcoupling technique giving a maximum ηext of 16.2%.

Figure 5

EL performance of our optimized device. (a) Current density and luminance as a function of voltage. (b) External quantum efficiency (ηext) as a function of current density. (c) Relative luminance as a function of time under continuous electrical stress at constant current density of 6 mA cm–2 corresponding to the initial luminance L0 of 516 cd m–2.

The operational lifetime of an encapsulated device was tested at a constant driving current density of 6 mA cm–2, corresponding to an initial luminance of 516 cd m–2. The device exhibits an LT85, defined by the time lasted when the device luminance reduces to 85% of its initial value, of 243 min (Figure c). The EL spectrum after the electrical stress remains consistent as compared to the fresh device (Figure S16).

In summary, we have systematically investigate the ligand effects on the morphological, optical, photophysical, and electroluminescent properties of perovskite NC solids. Our results uncover a more complex role of surface ligands in determining the LED device performance beyond the solution ηPL. We anticipate the fundamental principles presented here will facilitate the development of perovskite NC LEDs exceeding the light outcoupling efficiency limit by ligand chemical engineering.