Nanocrystals of the inorganic perovskite, CsPbBr3, display outstanding photo-physical properties, making them ideal for next generation optical devices. However, the typical combination of oleic acid and oleylamine ligands employed in their synthesis is easily displaced, leading to poor stability that can hinder their applicability. In this work, we look toward the replacement of the oleic acid and amine with phosphorous-based ligands. We synthesize CsPbBr3 nanocrystals using an oleylamine/alkylphosphonic acid combination with near perfect monodispersity with the ability to tune the bandgap by varying the alkyl chain length. We further investigate the replacement of the oleylamine giving a ligand combination of alkylphosphonic acid/trioctylphosphine oxide for perovskite nanocrystal nucleation and growth. This combination is typical for the widely studied metal chalcogenide synthesis and in our study with CsPbBr3 yields a pure phase perovskite.
I. INTRODUCTION
Lead halide perovskites (CsPbX3, where X = I, Br, Cl) have been the focus of intense research over the past decade due to their potential in photovoltaics and light emitting devices.1,2 In particular, all inorganic cesium lead halide perovskite nanocrystals (NCs) display outstanding photophysical properties such as the high photoluminescence (PL) quantum yield and narrow full width half maximum (FWHM), which sets them above traditional semiconducting nanomaterials.3,4 Their encouraging properties originate from the defect-free electronic structure of the bulk perovskite, combined with the ability to tune their bandgap energy via compositional and nanoscale size variations, resulting in quantum confinement effects.5
Many procedures have been reported for the synthesis of different shapes of CsPbX3 nanoparticles with focused size distributions.6–12 A challenge is that the chemical and colloidal instability of CsPbX3 NCs when exposed to polar solvents limits their widespread utilization. The perovskite crystal structure exhibits a predominantly ionic behavior, in comparison to metal chalcogenide nanocrystals that have partial covalent character. This weakens their interactions with organic ligands increasing their vulnerability to ligand loss during washing, ultimately leading to chemical instability, ion migration, and worsening carrier transport.13,14
The ligand layer on the nanoparticle is the key parameter to controlling the particle’s solvent dispersibility, effective volume, adhesion to substrates, functionalization, and surface charge.15,16 CsPbX3 NCs are conventionally synthesized using non-coordinating solvents, employing a mixture of aliphatic primary amines and carboxylic acids to solvate PbX2, inhibit particle aggregation, and determine the extent of hydrophilicity on the surface of the NPs. The oleylamine (OAm), oleylammonium halides, and oleylammonium oleates are proposed as the primary surface-capping agents of CsPbX3 NCs with their unsaturated nature, providing a high dispersibility in organic solvents.16,17 Alkylammonium ligands are much more mobile than carboxylate ligands and are susceptible to detachment from the NC surface during centrifugation or contact with polar solvents, making it difficult to retain the PL properties of the crystals.18 Excess amines have been shown to drive the dissolution of perovskite NPs and their transformation into an insulating phase, Cs4PbBr6.19 Furthermore, the carboxylates are less effective than the alkylammonium ions in modulating the size of the NCs.18 Therefore, alternative surfactants are desired.
De Roo et al. showed that surface Cs and Pb ions bind to either oleate or Br−, while surface halides bind to oleylammonium ions via hydrogen binding or electrostatic interactions. Therefore, ligands that would remain attached during processing would form strong metal–ligand bonds or stable hydrogen bonds with halide ions. Phosphonic acids (PAs) are known to bind strongly to divalent cadmium and lead anions, and their strong hydrogen bonding to perovskites has been proven.20–22 The presence of tetradecylphosphonic acid (TDPA) produced colloidal perovskite NCs that withstood washing in H2O and has since been shown as suitable for use in room temperature synthesis.23,24 Wang et al.16 found that a phosphinic acid was capable of stabilizing the alpha phase of CsPbI3. They found through NMR and Diffusion Ordered Spectroscopy (DOSY) that the phosphinic acid exists in an ion pair with oleylammonium and does not play a surfactant role in stabilizing the α-CsPbI3, meaning that only the oleylammonium iodide and/or OAm are the predominant surface ligands. Separately, α-CsPbI3 was maintained through washing with trioctylphosphine (TOP), leading to the quantum yield efficiency stability of over a month.25 In related studies, octylphosphonic acid (OPA) dramatically enhanced the perovskite stability toward polar solvents, producing an LED with an external quantum efficiency (EQE) of 7.74%.26
The use of branched capping ligands has also been identified as a route for improving the NC stability as they can provide a strong steric effect.27 By incorporating TOPO into the reaction system, monodisperse CsPbX3 NCs were formed at higher temperatures than achievable using the traditional ligands with the further advantage of stability toward washing with ethanol. Almeida et al.28 performed a fully amine-free synthesis using TOPO and OA and investigated how the acidity of the system affects the reactivity of the PbBr2 and therefore the size of the NCs. Notably, through the NMR and FTIR study, they found that TOPO was virtually absent from the particle ligand shell. The addition of TOPO to the toluene antisolvent in a ligand assisted reprecipitation (LARP) synthesis significantly reduced the surface defects of the NPs and increased the photoluminescence lifetimes as well as the stability of LED devices.29,30 An alternative amine-free synthesis involving the use of quaternary alkylammonium halides in place of alkylamines worked well to produce NCs; however, the LED based on this work only yielded a peak EQE of 0.325% due to moderate electroluminescence intensity.31
The variety within these reports suggests that phosphorous-based ligand syntheses may be realistic routes toward a stable, reproducible, high yield production of perovskite NCs. Here, CsPbBr3 NCs were synthesized using a phosphine ligand-based protocol. For the first time, the effect of alkylphosphonic acids as a replacement for OA is systematically studied, and a clear influence over NC size is achieved. We use PAs of increasing chain length, in conjunction with OAm, to examine the feasibility of replacing OA in the precursor solution to find a more stable surfactant. Examination of the morphology, chemical structure, and optical properties presents small near monodisperse NCs with an ability to tune the size and therefore bandgap of the material. Following encouraging findings, we examined the use of TOPO as the second ligand in place of OAm. Similar characterizations demonstrated the feasibility of moving away from the traditional OA/OAm ligand combination. These results show that PAs offer numerous possibilities to broaden the ligand chemistry of perovskites, which extends to morphology control, stability, and expanded reaction regimes.
II. EXPERIMENTAL SECTION
A. Materials
Octylphosphonic acids (>99%, OPA), decylphosphonic acid (99%, DPA), dodecylphosphonic acid (>99%, DDPA), hexadecylphosphonic acid (>99%, HDPA), octadecylphosphonic acid (>99%, ODPA), and tetradecylphosphonic acid (>99%, TDPA) were purchased from PCI Synthesis. Acetone (99.5%) and toluene (99%) were purchased from Lennox. Cesium carbonate (Cs2CO3, 99%), oleylamine (70%, OAm), oleic acid (90%, OA), 1-octadecene (90%), PbBr2 (≥99%), and trioctylphosphine oxide (99%, TOPO) were purchased from Sigma-Aldrich. All chemicals were used without any further purification unless otherwise stated.
B. Preparation of Cs precursors
Cs-Oleate (0.15M) in 1-Octadecene. Cs2CO3 (0.407 g, 1.25 mmol) and OA (1.7 ml, 5.4 mmol) were degassed in 20.0 ml of octadecene in a three-neck round-bottomed flask under vacuum at 100 °C for 1 h, followed by the reaction under argon at 150 °C until all Cs2CO3 was dissolved. The Cs-oleate was stored in an argon atmosphere.
C. Synthesis of CsPbBr3 nanoparticles (NPs)
In a typical synthesis, PbBr2 (69 mg, 0.16 mmol) and PA (0.5 mmol):OPA (100 mg), DPA (110 mg), DDPA (125 mg), TDPA (140 mg), HDPA (150 mg), or ODPA (170 mg) were dried at 120 °C in ODE (6.0 ml) for 1 h. Under argon, OAm (0.25 ml) was injected, and the temperature was set to 210 °C to form a solution. Then, the system was set to the desired reaction temperature (180 °C), and Cs-oleate was injected (0.6 ml, preheated to 120 °C). The reaction using TOPO was identical, except for the replacement of OAm with TOPO (1.0 g, 0.88 g/ml, and 2.59 mmol), which was incorporated at the beginning.
After 5 s of growth, the vial was plunged into a water bath to quench the reaction. To wash the NCs, 5 ml of anhydrous toluene was added to the quenched solution, which was then centrifuged at 13 000 rpm. The washing step was repeated once more before re-dispersion in anhydrous hexane. All washing occurred in inert atmosphere.
D. Transmission electron microscopy (TEM)
NC dispersions were drop-cast on carbon-coated 200 mesh copper grids. We acquired bright field transmission electron microscopy (TEM) images on a JEOL TEM-2100 microscope (W filament) operating at an accelerating voltage of 200 kV.
E. X-ray diffraction (XRD)
Dried powder of NCs was analyzed on a zero diffraction silicon substrate. We conduct x-ray diffraction (XRD) measurements on a PANalytical Empyrean x-ray diffractometer equipped with a 1.8 kW Cu Kα ceramic x-ray tube and PIXcel3D 2 × 2 area detector, operating at 40 kV and 40 mA.
F. Steady-state UV–Vis extinction spectroscopy and steady-state photoluminescence spectroscopy
We recorded optical extinction and photoluminescence spectra of anhydrous toluene dispersions in quartz cuvettes with a 1 cm path length, employing a Agilent Cary 5000 UV–vis spectrophotometer and a Varian Cary Eclipse fluorescence spectrophotometer, respectively. Samples were stored in an argon atmosphere prior to measuring.
III. RESULTS AND DISCUSSION
Octylphosphonic acid (OPA) was the first and shortest PA investigated as an OA replacement. A TEM image of NCs produced by the hot injection of Cs-oleate into a solution of PbBr2 dissolved in OAm, OPA, and ODE is shown in Fig. 1(a). The particles were nearly monodisperse with an average NC diameter of 6.2 ± 0.3 nm. Selected Area Electron Diffraction (SEAD) of multiple NCs verifies the sample crystallinity with the three most intense diffraction rings of monoclinic CsPbBr3 highlighted in Fig. 1(b). These data agree with the High Resolution TEM (HRTEM) in Fig. 1(c), where a d-spacing of 0.58 nm corresponding to the (100) plane is shown. Figure 1(d) displays the Fast Fourier Transform (FFT) of one of the particles showing the (110) and (100) planes of monoclinic CsPbBr3 (Ref: 00-054-0751).
The corresponding X-Ray Diffraction (XRD) pattern of the OPA capped NCs [Fig. 1(e)] shows the expected peaks for monoclinic CsPbBr3 in agreement with the SAED, although additional impurities are present corresponding to lead depleted perovskite, Cs4PbBr6. PA ligands are known to bind with metal Pb ions and halide ions, while the oleylammonium ion, formed from OAm, binds to the halide.17 It is possible that the smaller amount of lead in the lead depleted sample correlates with a reduction of the OAm ligand coating on the particle surface and hence lowers dispersibility than that of the CsPbBr3 NCs that were collected for TEM.
For the synthetic control of perovskite materials with phosphorous based ligands, we can learn from knowledge developed in cadmium and copper chalcogenides.32–37 A number of reports have shown excellent control of size and morphology across the entire particle distribution.38–41 Through this type of work, there is an understanding of the importance of different ligand classifications on the growth of NCs.42–46 As an example, Wang et al.47 showed that varying PA length resulted in incremental size effects. In the interest of developing further shape and size control of CsPbBr3 NCs, PAs with chain lengths between 8 and 18 carbon atoms were employed in this synthesis alongside OAm (see Sec. II for details). The resulting TEM images are displayed in Figs. 2(a)–2(f). Figure 2(g) shows the size distribution of the samples as measured (N = 60 particles), which was determined from the profile analysis of the TEM images in Digital Micrograph. Short chain ligands OPA and DPA allowed the formation of nearly monodisperse NCs of about 6.3 nm and 7.5 nm, respectively. As the chain length increased to C12, the polydispersity increased, and in the DDPA/OAm sample, a bimodal distribution of crystal size was identified. An outlier to this study was the TDPA/OAm sample, which produced relatively small nanoparticles (7 nm). However, despite this, the general trend of the samples is an increase in size as the PA chain length increases.
As a comparison, the syntheses involving traditional ligands OA and OAm form NCs with a size of around 20 nm and are shown in Fig. S1. Compared with these, the PA capped NCs were considerably smaller. In the study of PbSe, phosphorous based ligands are shown to have a strong binding affinity toward lead; P–O− moieties passivate Pb on the surface of the final PbSe NC.20,48 The study by Brown et al.26 reported the improved LED performance with OPA incorporation, finding that OPA ligands attach strongly to CsPbBr3 and that octylphosphonates bind preferentially through P–O−, resulting in hydrogen bonds between neighboring octylphosphonate chains. This P–O− moiety is known to bond to lead from the PbSe study, and so the hydrogen network is likely formed around the PbBr2 precursor before the cesium precursor injection. Therefore, the formation of the perovskite lattice undergoes slower reaction kinetics. This decelerated growth mechanism would be responsible for the smaller nanocrystals in comparison to the carboxylic case.
Figure 3 shows the XRD analysis of the samples showing the CsPbBr3 perovskite but also the presence of Cs4PbBr6. This lead depleted Cs4PbBr6 perovskite is attracting attention in recent years due to its outstanding thermal, chemical, and photo-stability. The individual octahedra of [PbX6]4− are perfectly isolated by cesium cations in the crystal lattice.49,50 However, density functional theory (DFT) calculations show that it has a theoretical bandgap of 3.84 eV and so has been suggested as a beneficial impurity that increases CsPbBr3 stability, but not as a stand-alone luminescent material.50
Researchers have shown that excess OAm concentration results in the formation of lead depleted Cs4PbBr6.51 Here, in our studies, OAm is kept at a constant minimal concentration that is sufficient to dissolve PbBr2. This progression to a multiphase sample may be another impact of the lack of Pb-oleate in the precursor solution. Where the carboxylic acid has one bonding site, the phosphonic acid has two sites on which it can bond with the lead precursor. This may result in a more difficult reaction between the lead and other perovskite precursors, leading to a slower reaction mechanism.
Optical properties of the samples were characterized using UV–Vis absorption and Photoluminescence (PL) spectroscopy. The spectra are shown in Fig. 4 with Tauc plots and bandgap data in Fig. S2. The bandgap of the OPA/OAm sample is 2.44 eV, after which the lengthening of the PA generally shifts the absorbance onset toward lower energies to 2.37 eV for the ODPA/OAm sample. This correlates with the increase in the particle size shown in Fig. 1.
The slight red shift of the absorption onset as the chain length of the PA increases is in direct contrast with the findings of Pan et al., where a red shift occurs with shortening of the carboxylic acid chain length.18 However, in their case, as the acid length increased, the NC size decreased, whereas here, the NC size increases with the ligand length. Therefore, the red shift for both ligand types occurs as the NCs increase in size. Decreasing the particle size down to the exciton Bohr radius should lead to quantum size effects and a consequent blue shift of the absorption onset and fluorescence.52 This effect is seen here with the shortening of the PA chain length and explains the effect of the ligand length on the optical properties of CsPbBr3. The Photoluminescence (PL) properties of the samples show a less obvious progression. The peak shift is very slight but is still in agreement with the absorption data.
As shown by these results, PAs are a viable alternative to carboxylic acid ligands, allowing further methods for size control. The amine ligand, which deprotonates to take the form of oleylammonium halides, is proposed as the primary surface-capping agents of CsPbX3 NCs.16,17 Yang et al.53 showed that NCs with branched ligands are more dispersible than NCs with long straight chains such as OAm. The interaction of the amine with the precursors and surfactants in the synthesis is a major component of the PbBr2 precursor system, and so the effect of changing from a carboxylic acid to a PA is likely to have major consequences for all the precursors involved. Although the concentration of amine was consistently low in these reactions, it is possible that the removal of OA caused effects that replicated the increase of amine that causes the formation of the lead depleted Cs4PbBr6 phase. Liu et al. found that a pure OAm system with a very large increase (×1000) in amine relative to carboxylic acid leads to a transformation of CsPbBr3 to Cs4PbBr6, yet the addition of a thiol ligand markedly decreased the amount of amine required to drive the transformation. Interestingly, they found that the thiol ligand alone did not cause the transformation at all.51
Meanwhile, TOPO is a branched capping ligand that has potential to provide a strong steric effect to the NC. It has been widely used in the synthesis for group II–VI semiconductors and has many benefits.54,55 Its high boiling point allows reactions to proceed at high temperatures and is compatible with organic solvents, allowing a completely inert reaction environment and air sensitive precursors. It has the ability to dissolve PbBr2 and so can be used here to replace OAm.28
Figures 5(a)–5(f) show the TEM images of the six samples made with TOPO and PAs. There are a number of obvious differences immediately apparent. The particles are easier to wash than the OAm coated samples, leading to a more facile characterization procedure. The 3:1 mix of anhydrous toluene and acetone was sufficient for both sets of samples; however, in the OAm case, the samples remained as a paste and were more difficult to characterize by pXRD. When using TOPO, while the samples were cleaner, they aggregated in the solution much more readily. The NCs are of similar size and shape to the OAm samples, yet the polydispersity increased, especially in the case of TOPO/DPA and TOPO/ODPA combinations. It is interesting to note that it is again the DDPA and TDPA samples that are out of agreement, similar to the OAm combinations.
The true benefit of the TOPO/PA system is evident in the XRD patterns of Fig. 5(g). In all cases, the patterns are pure CsPbBr3 in the cubic phase without any evidence of Cs4PbBr6 impurities. It appears that the window for complete dissolution of the precursors without the formation of lead depleted perovskite is larger when using TOPO instead of OAm.
The UV–Vis absorption and PL data shown in Fig. 5(h) show a slight red shift with an increase in the PA chain length. The bandgap is tuned from 2.22 eV to 2.13 eV, which are considerably lower values than the OAm case. Figure SI3 displays the Tauc plots of the absorption data along with bandgap calculations. This movement corresponds well with the size distribution shown in Fig. 5(i) and can be taken together to show the size and quantum confinement effect of lengthening PA ligands. The size distribution, however, shows that the particles lose their monodispersity to some extent, especially in the DPA/TOPO, DDPA/TOPO, and ODPA/TOPO samples.
IV. CONCLUSION
In this work, we present the viability of alkyphosphonic acids as a replacement for oleic acid in the synthesis of CsPbBr3 NCs. When used as a ligand in combination with oleylamine, phosphonic acids are shown to provide very good control over the morphology of CsPbBr3 NCs, leading to monodisperse nanoparticles with the ability to increase their size by increasing the length of the phosphonic acid chain in the system. However, the system shows an increased propensity to form the lead depleted Cs4PbBr6 in conjunction with CsPbBr3, even with the limited use of oleylamine. The subsequent replacement of oleylamine with trioctylphosphine oxide allowed the formation of a pure monoclinic phase without the presence of any impurities. This purely phosphine based ligand system resulted in an overall lower bandgap of the perovskite material as well as possessing the ability to vary the bandgap through the phosphonic acid size. As such, this work presents a systematic way to vary the optical properties of CsPbBr3.
SUPPLEMENTARY MATERIAL
See the supplementary material for additional TEM and characterization data.
ACKNOWLEDGMENTS
This publication has emanated from research conducted with the financial support of Science Foundation Ireland (SFI), Grant Number 16/IA/4629, Irish Research Council (IRC) under Grant Number IRCLA/2017/285. K.R. further acknowledges SFI Research Centres MaREI, AMBER, and CONFIRM 12/RC/2278_P2, 12/RC/2302_P2, and 16/RC/3918. F.M. acknowledges the IRC Enterprise Award w/ Intel Ireland (Award/Contract No. EPSPG/2016/170).