Surface Passivation with Selected Phosphine Oxide Molecules for Efficient Pure‐Blue Mixed‐Halide Perovskite Quantum Dot Light‐Emitting Diodes

Passivation of defects in halide perovskite using phosphine oxide or alkyl‐phosphonate has recently obtained a few remarkable achievements. However, effective application of phosphine oxide or alky‐phosphonate in passivating perovskite quantum dots (QDs) are seldom reported due to solubility issue or difficulty of amount control. In this work, two bifunctional organic molecules containing phosphine oxide groups, 2,4,6‐Tris[3‐(diphenylphosphinyl)phenyl]‐1,3,5‐triazine (PO‐T2T) and 2,7‐bis(diphenylphosphoryl)‐9,9′‐spirobifluorene (SPPO13), are deposited on QDs films by thermal evaporation. The molecules, both as passivation agents as well as electron transporting materials, exhibit stark contrast in passivating QDs and in light‐emitting diodes (LEDs) performance. A competition between charge transfer and defect passivation between the QDs and the molecules is proposed. In film, electron transfer from the QDs to PO‐T2T dominates and quench the QDs, while the passivation effect of PO‐T2T on the QDs dominates in driving device and enhances luminance of the LEDs. In contrast, passivation effect of SPPO13 on the QDs dominates both in films and in LEDs. A maximum EQE of 2.67% is obtained for the pure‐blue LED based on SPPO13‐passivated QDs films. This work provides a guide on the selection of passivation agents based on phosphine oxide and a promising passivation method for high‐efficient perovskite QD LEDs.


Introduction
Light-emitting diodes (LEDs) with halide perovskite as the emissive materials make big strides in recent years. [1] The highest external quantum efficiency (EQE) of perovskite LEDs with nearinfrared, red, green, and pure-blue emission have reached, respectively, 21.6%, 21.3%, 28.9%, and 10.3%. [2][3][4][5] Clearly, the efficiency for pure-blue perovskite LEDs relatively lags behind. Apart from the factor of LED structure and consequent issue of carrier injection efficiency, defects have a significant impact on the device performance. [6][7][8][9][10][11][12][13] Through passivation of defects in perovskite, photoluminescence quantum yield (PL QY) of the perovskite and efficiency of the perovskitebased devices could be improved. Phosphine oxide and Softer Lewis bases like alkyl-phosphonate have received increasing attention as passivation agents. [14][15][16][17][18][19][20][21][22][23][24][25] In the case thin film perovskite, defects mainly exist at the grain boundaries and in the grains. [26,27] Addition of molecules with phosphine oxide groups in the anti-solvents or in the precursor have been used to improve crystallinity and passivate defects of the perovskite. [14][15][16][17][18] Sargent group reported that triphenylphosphine oxide (TPPO) could effectively passivate the edge of green-emitting reduceddimensional perovskites and improve the PL QY from 40% to 98%, delivering a great increase of the maximum EQE from 4.5% to 14% with a prolonged operation lifetime. [14] Further, through fluorinated TPPO, tris(4-fluorophenyl)phosphine oxide, they obtained quantum wells with monodispersed thickness and pushed the PL QY from 80% to 100%, achieving a maximum EQE of 25.6% for the green-emitting LED. [15] Later, the group applied bis(4-fluorophenyl)phenylphosphine oxide in protecting Sr 2+ -doped perovskite from H 2 O and obtained sky-blue LEDs with a maximum EQE of 13.8%. [16] Recently, 2,7-bis(diphenylphosphoryl)-9,9′-spirobifluorene (SPPO13) was used to replace TPPO. The maximum EQE was improved from 14% to 22.3% and the maximum luminance at a wavelength of 530 nm was pushed from 45 230 to 127 841 cd m −2 . [17] Additionally, Lee group applied benzylphosphonic acid solution onto perovskite film and obtained perovskite nanocrystals shelled by the molecules, greatly reducing the defects and delivering a maximum EQE of 28.9% with a maximum luminance of 470 000 cd m −2 and a remarkable operation lifetime. [4] Octylphosphonic acid (OPA) was also used to passivate surface defects in preparing Ruddlesden-Popper perovskite sky-blue LEDs, improving the PL QY from 7.3% to 53.2% and the maximum EQE from 2.6% to 3.7%. [19] Zhao et al. synthesized a phosphonate/phosphine oxide dyad molecule and proposed that the phosphine oxide group plays the defect passivation effect to enhance PL QY and the phosphonate group could increase local carrier concentration due to strong electron affinity to accelerate radiative recombination in devices. [23] The maximum EQE of the obtained LEDs improves from 15.9% to 25.1% after introduction of the dyad molecule.
In the case of perovskite QDs, the high ratio of surface to volume makes QDs more sensitive to surface defects. In particular, chlorine vacancy defects on perovskite QDs are much easier to form than bromine and iodine vacancy defects, delivering a lower PL QY for chlorine-based blue-emissive perovskite. [20,28,29] OPA was used to replace oleic acid/oleylamine during hot-injection synthesis of CsPbBr 3 QDs to improve optical stability of the QDs during purification process, achieving a maximum EQE of 6.5%, which increases by seven times compared with the control sample. [21] Alky phosphonic acids with different alkyl chain lengths were later explored in preparing CsPbBr 3 QDs with nearunity PL QY. [30] The QDs were found to be in the shape of truncated octahedron with Pb-terminated facets due to strong binding between the phosphonate group and the lead ions. In addition, hexylphosphonate was used to passivate the surface bromine vacancy of perovskite QDs in the post-treatment process and achieved near-unity PL QY. [20] However, the actual amount of the hexylphosphonate in the colloidal QDs is strict to control, making passivation using phosphine oxide or alkyl-phosphonate in the purification stage of QDs difficult. [20] To avoid this issue, direct deposition of the molecules onto perovskite QDs films using thermal evaporation could be an alternative. Song group ever reported bi-lateral passivation of perovskite QDs films using thermal-evaporated diphenyl [4-(triphenylsilyl)phenyl]phosphine oxide for highly efficient green perovskite LEDs. [24] In this work, we selected two bifunctional organic molecules containing phosphine oxide groups, 2,4,6-Tris[3-(diphenylphosphinyl)phenyl]-1,3,5-triazine (PO-T2T) and SPPO13, as the passivation agents as well as the electron transporting materials for pure-blue mixed-halide perovskite QD LEDs. The theoretical and experimental passivation effects of the two molecules on the QDs films and performance in LEDs are compared and analyzed. PO-T2T could quench the QDs in film or in solution, while it could enhance the luminance of the QDs in LEDs. SPPO13 always presents defect passivation effect on the QDs. As for these observations, we propose that there is a competition between electron transfer and defect passivation between the QDs and the molecules. Based on SPPO13 modified QDs films, we obtained pure-blue perovskite QD LEDs with a maximum EQE of 2.67% at a peak of 469 nm. Our work provides a guide on the selection of passivation agents based on phosphine oxide and a promising passivation method for high-efficient perovskite QD LEDs.

Results and Discussion
To passivate defects of mixed-halide perovskite QDs without causing damage to the QDs and ensure charge transport at interface, we introduced organic molecules PO-T2T and SPPO13 ( Figure S1, Supporting Information) between the QDs layer and the electron transporting layer. The two types of molecules both have electron transporting ability and contain phosphine oxide groups for potential passivation of under-coordinated Pb atoms on the surface of the perovskite QDs. [17,31] First-principle calculation based on DFT was carried out using VASP program to obtain the adsorption energies of the molecules on the perovskite QDs. A 2 × 2 × 1 cell for CsPbBrCl was used. According to our X-ray diffraction (XRD) result of the QDs, in which (011) plane presents the strongest diffraction intensity ( Figure S2, Supporting Information), we chose the (011) lattice plane as the surface in the calculation. The calculation models (Figure 1a,b) present that there is a stronger interaction between the oxygen atoms in the molecules and the Pb atoms on the QD surface than the other elements in the QDs, which is in line with previous reports. [14,23,30] In addition, only one of the phosphine oxide groups in a molecule could adsorb effectively onto the QD surface at a time due to steric hindrance effect and complex field among the atoms. The calculation results show that PO-T2T has a higher adsorption energy of −1.54 than −0.75 eV for SPPO13 (Figure 1c), suggesting a better passivation effect of PO-T2T than SPPO13.
However, we obtained the opposite results in experiments. QD/SPPO13 films exhibit enhanced PL intensity without changing the PL emission peak, while QD/PO-T2T films show PL intensity even lower than that of the pristine QDs films (Figure 1d). The average PL QY of the QD/SPPO13 films increases by 87% compared with the QDs films, while the PL QY of the QD/PO-T2T films collected in glove box cannot be identified (Figure 1e). The PL decay plots present a much longer average lifetime of 64.5 ns for the QD/SPPO13 film than 9.72 ns for the QD/PO-T2T film, which is much lower than the average lifetime of 45.72ns for the pristine QDs film ( Figure 1f and Table S1, Supporting Information). These results indicate that SPPO13 has a positive passivation effect on the QDs film and PO-T2T could quench the emission of the QDs. To elucidate the possible reasons for the distinct passivation effect, we characterized the crystal structures of the QDs and intermolecular interactions in the films. X-ray diffraction (XRD) patterns (Figure 2a) show almost no difference after treating the QDs with PO-T2T, while the QD/SPPO13 film shows an obvious increase of the ratio of (002) to (011) lattice planes. FTIR patterns (Figure 2b) of the films also exhibit no shift of the transmittance peaks after treating the QDs with PO-T2T, while there is a slight blue shift from 1195 to 1197 cm −1 after treating the QDs with SPPO13. X-ray photoelectron spectroscopy (XPS) spectra of Pb 4f 5/2 and 4f 7/2 present a bit peak shift toward higher binding energies of the QD/SPPO13 film relative to the QD film ( Figure S3, Supporting Information). These characterizations suggest that SPPO13 molecules may have restructured the surface of the QDs and effectively passivate the surface defects on the QDs, while the PO-T2T molecules have weaker effect on the surface of the QDs.
In addition, considering that the PO-T2T molecule occupies larger space than the SPPO13 molecule ( Figure S4, Supporting Information), it is expected that steric hindrance, which may limit the numbers of the interaction sites between PO-T2T and the QDs, may exert effect. Thus, we calculated the intensity distribution of steric hindrance between the original ligands, didodecyldimethylammonium chloride (DDAC), on the surface of the QDs and the PO-T2T or SPPO13 molecules. The independent gradient model based on Hirshfeld partition of molecular density (IGMH maps) (Figure 2c,d) visualize the intensity of the steric hindrance by colors. Herein, the blue color represents lower steric hindrance and the red color represents higher steric hindrance. There are more red signals between DDAC and PO-T2T molecules (Figure 2c) than those between DDAC and SPPO13 molecules (Figure 2d), verifying stronger steric hindrance caused by PO-T2T molecules and weaker surface interaction between the QDs and PO-T2T molecules. However, the above results still cannot explain why the PL seriously quenched in the QD/PO-T2T film. According to literature, this kind of PL quenching is usually attributed to energy transfer or charge transfer, such as triplet-triplet energy transfer (TET), Dexter-type energy transfer, Förster resonant energy transfer (FRET), hole transfer, electron transfer and charge trapping. [32][33][34][35][36] However, there are three main differences of the current case from the previous cases. First, the current case happens in solid film rather than in solution. Second, the QDs used in this work has an average size of 12nm ( Figure S2c, Supporting Information), which is beyond quantum confinement required for efficient TET. [35][36][37] Third, the phosphine oxide molecules used in this work are macromolecule rather than small molecule, which may make a difference in interaction between the QDs and the molecules.
To initially analyze the kinetic mechanism, we performed transient absorption (TA) measurement on the films using 375nm laser to make sure that only QDs were pumped ( Figure S5, Supporting Information). The TA difference spectra show clear exciton bleaching (XB) signals ≈462nm in the three films and there is no extra XB signals after introducing PO-T2T or SPPO13 onto the QDs film (Figure 3a--c). The normalized TA signals at 462nm were plotted (Figure 3d) and the fitting results of the time constants using the bi-exponential decay function show a decrease of the average lifetime from 635 to 529 and 543ps ( Figure S6 and Table S2, Supporting Information), respectively, after treating the QDs with PO-T2T and SPPO13. According to the energy levels (Figure 3f), [38][39][40][41][42] both PO-T2T and SPPO13 have LUMO levels lower than the conduction band (CB) of the QDs and deeper HOMO levels than the valence band (VB) of the QDs ( Figure S7, Supporting Information), which indicate that ultrafast electron transfer from the QDs to the phosphine oxide molecules is energetically favorable and hole transfer from the QDs to the phosphine oxide molecules is energetically unfavorable. Considering that the driving force from the QDs to PO-T2T and SPPO13 are, respectively, 0.72 and 0.13 eV, [34] the electron transfer from the QD to PO-T2T should be faster, in line with the shorter lifetime of the QD/PO-T2T film. In addition, it is reported that the dark state of perovskite QDs may lie above the bright states with a small energy spacing, [43] which may make TET from the QDs to PO-T2T, which has a high triplet energy of 2.99eV, [38] possible. However, we did not detect photoinduced absorption signals corresponding to triplets of PO-T2T (Figure 3e; Figure S8, Supporting Information), which should otherwise present increasing normalized TA signals as a function of lifetime in a 50 μs time scale. [36] Dexter-type energy transfer, which is a short-range process, [35,36] could be excluded taking the steric hindrance of PO-T2T into account. FRET could also be excluded because there is no overlap between the absorption spectrum of PO-T2T with a rather wide energy gap and the PL spectrum of the QDs. [36] Thus, the PL quenching of the QD/PO-T2T film could be attributed to the ultrafast electron transfer from the QDs to PO-T2T.
Another difference of the current case from the previous reports is that there is a saturation value for the PL quenching after treating the QDs with PO-T2T solution (Figure 4a), while previous researches basically report that the PL quenching of perovskite QDs becomes increasingly serious with increas-ing amount of molecules. [34][35][36][37][38][39][40][41][42][43][44] The saturation value for the PL quenching in the case of PO-T2T is ≈75% of the original PL intensity, no matter solutions with 0.5, 1, or 5 mg mL −1 PO-T2T in chlorobenzene (CB) is added into the QDs solution (Figure 4c). In contrast, the PL intensities of the QDs solutions increase after introduction of SPPO13/CB solutions, and with more SPPO13/CB solution is used, the PL intensity of the QDs solution is higher (Figure 4b,d). Considering that CB, a polar solvent, could itself quench the PL intensity of the QD solution to a degree, we attribute the increase of the PL intensity in the case of SPPO13 to its effective passivation effect on the QDs. Thus, passivation effect dominates over electron transfer when the QDs is modified with SPPO13 molecules.    to be able to suppress EL spectra shift in our recent report. [45] Performance of LEDs based on QD/PO-T2T and QD/SPPO13 show stark contrast. With similar current density (Figure 5e), a typical QD/SPPO13 LED exhibits an obviously higher maximum EQE than that of a PO-T2T modified device (Figure 5f and Table S3, Supporting Information). But the maximum luminance of the two modified devices are similar. This may be due to suppress of charge transfer from the QDs to PO-T2T and enhancement of interaction between the QDs and PO-T2T in passivation effect under driving bias. Both LEDs present stable EL spectra with increasing bias (Figure 5g,i). The statistics of the performance of 32 QD/PO-T2T LEDs and 36 QD/SPPO13 LEDs show that the highest maximum EQE for QD/SPPO13 LEDs is 2.67%, four times >0.62% for the QD/PO-T2T LEDs (Table S4, Supporting Information). The average maximum EQE and the average maximum luminance for QD/SPPO13 and QD/PO-T2T LEDs are 1.82% and 0.43%, as well as 323 and 277cd m −2 (Figure 5h,i), respectively. The driving time of the LEDs were collected. The half-lifetime (L 50 ) of the QD/PO-T2T LED and QD/SPPO13 LED are, respectively, 19.5 and 13.4 s, both of which are longer than the L 50 of the QD LED ( Figure S11, Supporting Information). This together with the improved luminance of the LEDs suggests that both PO-T2T and SPPO13 have passivation effect on the QDs in running devices. Combining the previous discuss on films, we propose that electron transfer from the QDs to PO-T2T dominates over passivation effect of PO-T2T on the QDs, leading to PL quenching of the QDs in films; while electron transfer from the QDs to PO-T2T is suppressed by the external bias in device, and thus passivation effect of PO-T2T on the QDs dominates, enhancing the luminance and driving lifetime of the LEDs.
Herein, the low EQEs of QD/PO-T2T LEDs is attributed to the poor band alignment between the PO-T2T and the mixedhalide perovskite QDs (Figure 5d). First, the LUMO for PO-T2T is 0.72eV lower than the CB of the QDs, while the LUMO for SPPO13 is merely 0.13eV lower than the CB of the QDs. [17,31,46] This big energy spacing makes the electron transfer from the PO-T2T to the QDs difficult, while the well-matched band alignment between the SPPO13 and the QDs is conducive to the higher device efficiency. Second, some electrons directly transfer from the LUMO of PO-T2T to the VB of the QDs ( Figure S12, Supporting Information), leading to a weak emis-www.advancedsciencenews.com www.advmatinterfaces.de sion near 620 nm ( Figure S13, Supporting Information). Due to the low LUMO/HOMO and high triplet energy, PO-T2T was ever designed to form exciplex in organic LED. [38,39] Interestingly, we found PO-T2T could also form exciplex exciton with perovskite QDs even in a planar structure.

Conclusion
In summary, we selected two bifunctional organic molecules containing phosphine oxide groups, PO-T2T and SPPO13, as passivation agents for the mixed-halide perovskite QDs and electron transporting materials for the pure-blue LEDs. In films, PO-T2T could quench PL of the QDs due to dominating electron transfer from the QDs to PO-T2T over the passivation effect of PO-T2T on the QDs. In contrast, SPPO13 could effectively passivate defects in the QDs because the passivation effect of SPPO13 dominates over electron transfer from the QDs to SPPO13. In LEDs, although PO-T2T could not effectively transfer electrons to the QDs under external bias and form a parasitic exciplex with the QDs, which leads to low EQE, PO-T2T shows passivation effect on the QDs and enhances the luminance and lifetime of the LEDs. SPPO13-passivated QDs delivers pure-blue perovskite QD LEDs with a maximum EQE of 2.67% at a peak of 469 nm. Our work provides a guide on the selection of passivation agents based on phosphine oxide and a promising passivation method for high-efficient perovskite QD LEDs.
The cesium, potassium, and formamidine precursors together with OA dispersed in toluene were injected into lead bromide precursor at room temperature under magnetic stir in the open air. Two minutes later, the CuCl 2 precursor was injected into the reaction system. After another two minutes, MA was poured into the solution with a volume ratio of 2:1. After centrifugation for 1min, the precipitates were collected and dispersed in toluene. The second purification followed the first one, except a certain amount of OA and DDAC were supplemented before adding MA. The collected precipitates after the second centrifugation were dispersed in octane to obtain Cs x FA 1−x PbBr 1.75 Cl 1.25 :Cu:K QDs.
Theoretical Calculations: First-principle calculation was performed in the framework of density functional theory as implemented in the VASP program. [49,50] The generalized gradient approximation of Perdew-Burke-Ernzerh was employed for the electronic exchange and correlation. [51] The plane wave pseudopotential with a kinetic cutoff energy of 450eV within the projector augmented wave (PAW) method was used. [52] The van der Waals dispersion-corrected DFT (DFT-D3) was also carried out, as pro-posed by Grimme et al. [53] The cell dimensions (a×b×c) were as follows: 2 × 2 × 1 for CsPbBrCl (011). All surface models were taken in a vacuum condition of ≈20 Å along c-axis to avoid periodic effects. The convergence criteria of the total energy was <10-5eV and the geometry optimization was terminated when the forces on all atoms were <0.01eV Å −1 . The Monkhorst-Pack method was applied to perform the Brillouin zone integrations, in which the k-point values of 2×2×1 was used for surface structure optimization. [54] Calculations for steric hindrance were performed using Gaussian 16 quantum chemistry program package. The initial starting point geometries were built from the Materials Studio and optimized to a minimum, followed by analytical frequency calculations (Hessian) to confirm that no imaginary frequencies were present. The geometry optimizations were performed using the Baker's three-parameter hybrid (B3LYP) correlation functional with DFT-D3 dispersion correction and the 6-311(d, p) basis set. Using the optimized geometries, single point calculations were performed at the M06-2X-D3/def-TZVP. IGM based Hirshfeld partition of molecular density which was obtained by Multiwfn was used to analyze the weak interaction. [55] Device Fabrication: Cleaned indium tin oxide (ITO)-coated glass substrates were treated with oxygen plasma for 10min. Then, poly (3,4ethylenedioxythiophene):polystyrene sulfonate (PEDOT:PSS) solutions (Clevious4083, filtered by a 0.45 um filter) and Poly[bis(4-phenyl)(2,4,6trimethylphenyl)amine] (PTAA) (5 mg mL −1 in chlorobenzene, filtered by 0.22 μm filter) solution were spin-coated on ITO substrates at 3000rpm for 50 s and annealed, respectively, at 150°C for 15min and 120°C for 25min. Subsequently, the QDs were spin-coated at 4000rpm for 45 s. Later, under a vacuum of ≈4×10 −4 Pa, 2,4,6-Tris[3-(diphenylphosphinyl)phenyl]-1,3,5triazine (PO-T2T) or 2,7-bis(diphenylphosphoryl)-9,9′-spirobifluorene (SPPO13), 1,3,5-tri[(3-pyridyl)-phen-3-yl]benzene (TmPyPB), and LiF/Al electrodes (1nm/100nm) were deposited sequentially in a thermal evaporator with shadow masks. The active area of the light-emitting diodes (LEDs) was 4 mm 2 as defined by the overlapping area of the ITO and Al electrodes.
Characterizations: XRD was performed on a Rigaku Ultima IV setup. UPS was collected by a Escalab Xi+ setup. LED cross-sectional samples were prepared using Focused Ion beam on a HELIOS 5 UX setup, and TEM images were collected on a Talos F200E setup. PL spectra were collected by OmniFluo900. TA spectra was collected on a Helios Femtosecond Transient Absorption Spectrometer. PL QY was measured using an integrating sphere and a QE65000 spectrometer (Ocean Optics) with a pump intensity of ≈0.3 mW cm −2 . The current density-voltage-luminance characteristics of the LEDs were measured by a Keithley 2400 source meter coupled with a QE-Pro spectrometer (Ocean Optics) in an N 2 -filled glove box.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.