Solution-processed metal halide perovskites have attracted considerable research interest as a new type of semiconductor materials, which has been utilized in solar cells, light-emitting diodes and other optoelectronic devices [1-8]. In particular, Perovskite light-emitting diodes (PeLEDs) have been conferred as the cutting-edging research due to their high performance [9-16], red emission is very important in the field of lighting and display. To date, the red PeLEDs have achieved significant external quantum efficiencies (EQE) that over 20%, most of them are based on the three-dimensional (3D) and quantum-dot (QD) perovskite material [17-22].
Compared with 3D and QD perovskite, quasi-two-dimensional (quasi-2D) perovskite could be more efficient emission material due to its higher exciton binding energy, multi-quantum-well structure and good charge transfer properties [23-28]. Ruddlesden-Popper perovskite (RPP) is generally described by the formula of L2An-1BnX3n+1 as typical quasi-2D perovskites, which consist of large organic cation L, a small organic or cesium cation A, divalent metal cation B (lead or tin), and halide anion X (Cl-, Br- or I-). Although the overall development of red LED is remarkable, the quasi-2D red LEDs are still in arrears of efficiency. The reason may be that fine control of quasi-2D perovskite phase (represents the thickness of the [PbI6]4- octahedra layers capped by the L cations [28-31]) distribution is a great challenge, which make the inferior charge injection and serious non-radiative recombination.
In this work, we report a general strategy of cospacer cations to control the phase distribution in RPP films, by mixing a representative of phenethylammonium iodide (PEAI)/ 3-fluorophenylethylammonium iodide (m-F-PEA) and 1-naphthylmethylammonium iodide (NMAI) to alloy the RPP films. These RPP films show significant suppression of small index phase (n=2), and the carrier transport has been improved effectively. At the same time, the mixing of NMA can reduce the metallic Pb states and halide defects in RPP films, moreover, a smooth and pinhole-free perovskite film has been obtained. As a result, larger than 90% photoluminescence quantum yield (PLQY) is achieved through the mixing of m-F-PEA and NMA with a mixing ratio of 6:4, the LED prepared by this method achieves the efficiency as high as 25.8% and the maximum luminance of exceeding 1300 cd∙cm-2 at 680 nm.
We prepared RPP films of (PEAxNMA1-x)2CsPb2I7 and (m-F-PEAxNMA1-x)2CsPb2I7 with different composition (x = 0, 0.2, 0.4, 0.6, 0.8 and 1) to study the effect of cospacer cations on perovskite properties. The molecular structures of PEA, m-F-PEA and NMA cations are shown in Fig. 1a, m-F-PEA represents that a halide F atom is contained on the para-position on the conjugated phenyl ring in PEA. Among them, we found that the mixed cation perovskite with x=0.6 is the optimized condition, which will be shown in the following. PEA cation only and cation mixture perovskites are expressed as P10N0, P6N4 and mFP6N4 in this study, here P means PEA, mFP means m-F-PEA, and N refers to NMA. Fig. 1b represents ultraviolet-visible (UV-Vis) absorption spectra and steady photoluminescence (PL) spectra of the three films. The multiple absorption peaks suggest that thin perovskite films likely contain mixed quasi-2D structures with different n values [32-33]. The film P10N0 shows several strong exciton absorption peaks at ~560 nm and ~610 nm, corresponding to the phase of n = 2 and 3. While introducing NMA, the films of P6N4 and mFP6N4 absorption peaks corresponding to n = 2 and 3 are decreased significantly, and the peak of n = 2 is almost negligible in the mFP6N4 film. Consisting with the absorption, the film P10N0 shows obvious emission peaks of n = 2 and 3. And the films P6N4 and mFP6N4 mainly contain large n phase, which lead to the red-shifted PL spectra and no obvious emission of n = 2. The comprehensive UV–vis absorption spectra and PL spectra from quasi-2D perovskite films with different compositions (Supplementary Fig. 1, 2) confirm this trend. It must be noted, along with the red-shift of PL for cospacer cations, the full width at half maximum (FWHM) decreased from 43 nm to 34 nm, indicating that the phase distribution in quasi-2D perovskite become narrow. The weak PL intensity of P10N0 and P0N10 samples could due to the trap-assisted nonradiative recombination and strong electron−phonon coupling [34-36]. We carried out PLQY measurements for these three films, as seen from Fig. 1c, the PLQY of perovskite sample increases from 53% (P10N0) to 78% (P6N4) and 91% (mFP6N4) while using cospacer cations, and the targeted films exhibit strong emission under ultraviolet illumination (Supplementary Fig. 3). Fig. 1d shows the time-resolved photoluminescence (TRPL) decay profiles of different samples, with the mixing of NMA, the average PL lifetime of the quasi-2D film gradually increases from 5.321 ns (P10N0) to 8.308 ns (P6N4) and 8.523 ns (mFP6N4), suggesting the suppression of nonradiative recombination while engineering cations.
Fig. 2a represents the X-ray diffraction (XRD) patterns of different quasi-2D perovskite films, the strong diffraction peak of n = 2 (2θ = 3.8°) can be observed in P10N0 film, indicating a dominant of small index of n = 2 phase. As NMA mixed with PEA/ m-F-PEA, the peaks of n = 2 phase reduce significantly, and the large-n phases become the main part in the P6N4 and mFP6N4 films. Compared with the (212) and (111) peaks in P10N0 film, the strong (110) and (220) peaks are observed from P6N4 and mFP6N4 films, implying that the preferred orientation has been formed in the cospacer cation perovskite. In addition, grazing-incidence wide-angle X-ray scattering (GIWAXS) measurements were performed to further investigate the crystal phase and crystallinity of the quasi-2D films, the Debye-Scherrer rings showed in P10N0 films indicate a random crystalline orientation, while stronger scattering spots along the qz direction are exhibited in P6N4 and mFP6N4 indicating a vertical alignment of (110) with respect to the substrate (Fig. 2b-d), which facilitates the charge vertical transport [37-39]. The crystallinity of perovskite with different crystal planes collected by high-resolution transmission electron microscopy (HRTEM) is shown in Fig. 2e-f. For the P10N0 film, the n = 2 phase crystal planes in P10N0 can be clearly observed, the crystal plane of (212) and (111) from large n phase (close to 3D perovskite) are also found in Supplementary Fig. 4. While for the cospacer cation films, the dominant (110) crystal plane with good crystallization were observed, consistent with our finding in the XRD results.
Time-resolved transient absorption (TA) measurements were carried out to derive the phase distribution and energy funnel process in quasi-2D perovskite films. For the P10N0 film, as seen from Fig. 3a, d, there are several ground-state bleaching (GSB) peaks around 560, 610, 635, 655 and 670 nm, which correspond to the phase of n = 2, 3, 4, 5, and n ≥ 6 in RPP film and represent the evolution of excited-states carriers, respectively.
And it can also be found that the photogenerated carriers are initially formed in the wide-bandgap quantum well of n = 2 phase, while the charge transfer process from n = 2 phase to large-n phase is more than 100 ps based on the absorption bleaching results, which could retard carrier concentration and reduce radiative recombination in the emitting center (large-n phases) [29]. For the cospacer cation films, as shown in Fig. 3b, e and c, f, the peaks of different phases still exist, the most significant difference is that the peak of n = 2 phase is almost negligible after attenuation of about 1 ps, indicating much faster charge transfer happened in the quantum well structure for the cospacer cation system.
To further study the energy transfer process, the kinetics of each GSB are fitted by a function that includes a fast decay component τ1 related to the carrier transferring from small-n phase to large-n phases in perovskite film, slow decay components τ2 and τ3 related to the energy decay and recombination, and the rapid increase component τet related to the establishment of GSB peak [40-41], Fig. 3g-i show the decay kinetics about different phases of three perovskite films, the results of fitted lifetime components are given in Supplementary Table 1. The fast decay time τ1 of n = 2 and n = 3 phases in P10N0 film are 0.44 and 0.19 ps, respectively, then the excitons continuously transferred to n ≥ 6 phases for the radiative recombination. Encouragingly, the fast decay time τ1 of n = 2 phase in P6N4 (0.05 ps) and mFP6N4 (0.05 ps) are 9 times shorter than that in the pristine one, with τ1 of n = 3 phase in P6N4 (0.10 ps) and mFP6N4 (0.10 ps), which are 2 times shorter than that in P10N0. The shorter τ1 means that the carriers are transferred out of the quantum well at a faster rate [25]. Meanwhile, the establishment time τet of the emission phase for P10N0, P6N4, and mFP6N4 is 1.21 ps, 1.10 ps, and 1.02 ps, respectively. The above results show that the cospacer cation strategy can effectively improve the energy transfer process in the perovskite films.
X-ray photoelectron spectroscopy (XPS) measurements were performed to check the effect of the cospacer cations on the chemical states of the elements in the perovskite. The core levels of Pb 4f7/2 from the three different compositions of perovskite are shown in Fig. 3a-c. The dominant peak around 138 eV is corresponding to the Pb-I bond in [PbI6]4-, and the shoulder peak around 136 eV corresponds to metallic Pb states. For the control film, an obvious Pb metallic state can be easily found, which is detrimental to the PeLED performance due to the serious non-radiative recombination [3]. After mixing the PEA with NMA, the metallic Pb peak becomes weak but still exist, and it is very encouraging that the metallic Pb peak is fully suppressed in mFP6N4 system. The reduction of metallic Pb states could be one of main reasons for the high PLQY for the cospacer cation system.
In addition to the suppression of metallic Pb states, we also found that there are some chemical shifts for the I core level. Compared with P10N0 and P6N4, the core level of I in mFP6N4 exhibits a blue-shift of 0.2 eV (Supplementary Fig. 5), means the stronger interaction between I and Pb and a more stable crystal structure [42-43]. The main effect could be due to the substitution of fluorine for hydrogen on the benzene ring of PEA, which will change the electronic structure of the molecule, then resulting slight differences in octahedral tilting, shorter average phenyl ring centroid-centroid distances and more tightly stacked perovskite sheets [42-45]. Meanwhile, the electron withdrawing of F atom could suppress the formation of halide dangling bonds [21].
Scanning electron microscopy (SEM) was used to investigate the surface morphology of thin quasi-2D perovskite films, as shown in Fig.4d-f. It is noticeable that the P10N0 perovskite film shows obvious pinholes and isolated large grains, which could be detrimental to achieving high device performance due to the current leakage. We found that the cospacer cations could effectively reduce the pinhole and suppress the formation of large isolated particles, and it is very encouraging that the perovskite film becomes very dense and continuous while mixing m-F-PEA and NMA. In previous studies, it has been reported that low-dimensional perovskite tends to aggravate the formation of larger grains and rougher perovskite films, due to the existence of layer perovskite phases [40-41]. This study has confirmed that the strategy of cospacer cations can reduce the small index of quasi-2D perovskite, which guarantees the formation of smaller grains and reduces the surface holes and defects of perovskite. Atomic force microscopy (AFM) results also confirmed the better morphology while mixing m-F-PEA with NMA (Supplementary Fig. 6), the root mean square roughness of three perovskite films decreased from 2.89 nm (P10N0) to 2.12 nm (P6N4) and 1.74 nm (mFP6N4).
We fabricate the perovskite films into configure LEDs based on device structure of ITO/ the modified PEDOT: PSS (m-PEDOT: PSS) (40 nm)/ poly-TPD (Poly(N, N′-bis(4-butylphenyl)-N,N′-bis(phenyl)-benzidine)) (10 nm)/ PVK (poly(9-vinlycarbazole)) (5 nm) / quasi-2D perovskite (30 nm)/ TPBi (2,2′,2′′-(1,3,5-benzinetriyl)-tris(1-phenyl-1H-benzi-midazole)) (30 nm)/ LiF (1 nm) Al (100 nm). Fig. 5a shows a schematic of the flat-band energy-level diagram of the multiple layers. The cross-sectional SEM image of devices is shown in Fig. 5b, and a clear sandwich structure is clearly observed. The typical electroluminescence (EL) spectra of P10N0, P6N4 and mFP6N4 are shown in Fig. 5c. Although the quasi-2D perovskite films exhibit a mixture of many phases with different bandgaps, only a single red EL peak is observed. Compared with the peak of P10N0 (~660 nm), the peaks of P6N4 and mFP6N4 (~680 nm) have a slight red-shift, which is consistent with the PL results (Fig. 1a).
According to the current density−voltage−luminance (J-V-L) plot (Fig. 5d), the leakage current of the P10N0 film is obviously higher than that of films with cospacer cations, but the subsequent injection current is lower, indicating that there is a higher defect density in P10N0 films and the redundant n = 2 phase will hinder the injection of current [46]. Furthermore, it could be found that the leakage current of P6N4 is slightly higher than that of mFP6N4, which consists with that mFP6N4 owns better morphology. In Fig. 5e, it is apparent that the performance of the devices with cospacer cations is better than that of the device with single organic cation, a very high EQE value of 24.2% is achieved with a luminance of 1300 cd∙m−2 in mFP6N4 film, while just 20.6% with 850 cd∙m−2 in P6N4 and 12.7% with 500 cd∙m−2 in P10N0, and the maximum EQE of champion PeLED based on mFP6N4 could reach 25.8% (Fig. 5f). The EL spectrum of the device under different voltages is given in Supplementary Fig. 7. As the bias voltage increases, the device shows good spectral stability and also with nice EL performance at Commission Internationale de l’Eclairage (CIE) coordinates of (0.722, 0.277) (Supplementary Fig. 8). We also collected the EQE data for 27 devices, the device shows only 5.1% relative deviation (Supplementary Fig. 9), demonstrating the good reproducibility of the cospacer cation strategy. In addition to increase the device efficiency, we also found that the cospacer cation method improves the operational stability, the T50 operational time was increased to over 5 times with an initial brightness of 100 cd∙m−2 (Supplementary Fig. 10), which could be due to the passivation of the defects in the perovskite layer (Fig. 3a-c) and also the increase of the chemical bonding (Supplementary Fig. 5).
In conclusion, cospacer cation quasi-2D (PEA/m-F-PEA)xNMA1-x)2CsPb2I7 RPP films were developed for efficient red PeLED. Low-dimensional with small index (n = 2) perovskite phase is suppressed by cospacer cation modulation, which is helpful for charge transfer. Meanwhile, the metallic Pb states and halide defects in perovskite films are passivated. Accordingly, efficient red PeLED with a maximum EQE of 25.8% at 680 nm is achieved. Our method provides an efficient strategy to control phase distribution and passivate defects simultaneously, which is beneficial to realize more efficient PeLEDs and could also be good for solar cells.