Close-Packed Ultrasmooth Self-assembled Monolayer of CsPbBr3 Perovskite Nanocubes.

The use of colloidal self-assembly to form the complex multiscale patterns in many optoelectronic devices has been a long-standing dream of the nanoscience community. While great progress has been made using charged colloids in polar solvents, controlled assembly from nonpolar solvents is much more challenging. The major challenge is colloidal clustering caused by strong van der Waals (vdW) attraction between long-chain surface capping ligands passivating the surface of nanocrystals. Such clustering degrades ordering in packing during the self-assembly process. While ligand exchange to provide colloidal stability in polar phases is often an option, this is not the case for the exciting new class of halide perovskites due to the material's solubility in essentially all polar solvents. Here, we report surface-functionalized self-assembly of luminescent CsPbBr3 perovskite nanocubes by partially replacing long-chain oleyl groups (18 carbon chain) with short-chain thiocyanate (SCN-). This enables the fabrication of ultrasmooth monolayer thin films of nanocubes with a root-mean-square (RMS) roughness of around 4 Å. This ultrasmooth large area self-assembled layer could act as high-efficiency optoelectronic devices like solar cells, light-emitting diodes (LEDs), transistors, etc. We correlate our experimental results with simulations, providing detailed predictions for lattice constants with chain conformations showing reduced free energy for cubes grafted with short-chain thiocyanate compared to long-chain oleyl groups, thus facilitating better self-assembly.

Introduction

Self-assembly of nanomaterials to form complex two-dimensional (2D) and three-dimensional (3D) patterns with control from nanometer to centimeter length scale has been a long-standing challenge for the nanoscience community.[1−7] The prospect of engineering complex functional devices by simple wet chemistry and self-assembly has largely been inspired by nature, where hierarchical ordering and reproducible complex pattern formation are achieved at or near room temperature in an aqueous environment.[8,9] Although there has been tremendous progress in bioinspired and colloidal self-assembly, most work has been conducted either in polar solvents with charged colloids[3] or at the interface between polar and nonpolar solvents.[10] In polar solvents, surface charges on nanoparticles provide excellent colloidal stability ensuring facile monolayer 2D[11,12] and 3D[3] self-assembly and patterning via capillary assembly[13] or electrophoretic deposition.[14] In nonpolar solvents, they can be stabilized using soluble organic surface capping ligands[15] and self-assembly proceeds, for example, via solvent evaporation[4,16−21] and nanoimprinting.[22] However, the colloidal aggregation due to strong van der Waals (vdW) interactions[23,24] between organic chains prevents controlled assembly into single-crystalline monolayers. The role of capping ligands therefore is not only to ensure stability but also to guide self-assembly into ordered or disordered structures.

There is an emerging class of materials, namely the halide perovskite nanocrystals,[25−27] which has outstanding optoelectronic properties. The most stable inorganic perovskite CsPbBr3 nanostructures[25,28−32] are stabilized by long-chain oleyl groups as surface capping ligands. Nanocubes of CsPbBr3 have a tendency to fuse into single crystals,[33−36] making them ideally suited for a self-assembled thin film and luminescent nanopatterned devices. However, the major problem behind their self-assembly is that due to their ionic nature they readily dissolve in a polar environment, while in a nonpolar solvent, they have a tendency to cluster[37] due to strong vdW attractions between the long capping ligands. Therefore, preventing colloidal aggregation to enable close-packed monolayer thin films remains an open challenge. Here, we demonstrate that besides improving the optical quality,[38] short-chain thiocyanate ligands also lead to a substantial reduction in cube–cube interactions, greatly improving self-assembly. This enables the formation of monolayer thin films with root-mean-square (RMS) roughness of 4 Å.

Results and Discussion

We perform colloidal synthesis of monodispersed CsPbBr3 nanocubes using a modified literature report,[28] and carry out surface modifications to achieve better control in self-assembly (details in the Experimental Section). In short, lead bromide (PbBr2), octadecene (ODE), oleic acid (OAc), and oleylamine (OAm) are heated at 200 °C in a three-necked round bottom flask. Cesium oleate is injected, and 5 s after the injection, the temperature is decreased to 160 °C.

The reaction mixture is then annealed at 160 °C for 10 min. Annealing at this stage is essential to obtain monodispersed cubes with sharp facets, which is crucial for a long-range, defect-free, close-packed ordering of cubes. After that, long-chain oleyl groups on the nanocube surfaces are partially exchanged with shorter-chain thiocyanate (details in the Experimental Section). The thiocyanate-capped cubes dispersed in toluene are dropcast on a silicon (Si) substrate, with slow evaporation of toluene (Figure ), leading to a beautiful monolayer assembly of cubes with uniform packing, as shown in Figure a.

Figure 1

Schematic presentation of droplet evaporation technique inside a closed chamber to get a close-packed monolayer self-assembly.

Figure 2

(a) High-resolution scanning electron microscopy (SEM) image of the monolayer; inset of (a) shows FFT pattern of the monolayer film. (b) Low-resolution SEM image of a self-assembled CsPbBr3 nanocube monolayer. Light green false color is used in the SEM image to indicate the nanocube monolayer on Si (black area). (c) Atomic force microscopy (AFM) image of a monolayer film. (d) Three-dimensional view of the area in (c) and measured RMS roughness value. (e) X-ray diffraction (XRD) pattern of monolayer (NH4SCN-treated film) and multilayer (untreated with NH4SCN) nanocube assembly.

A low-resolution SEM image of the film is shown in Figure b (also in Figure S1), showing large area continuous assembly on a Si substrate. The assembled layer is truly a monolayer, shown in Figure S2, observed from the fact that the cleaning procedure (rinse one time with methyl acetate solution very gently) used to avoid charging effects during SEM led to partial delamination in a few places. The AFM image in Figure c,d reveals that the surface of the assembled monolayer is ultrasmooth, with a measured RMS roughness of 4 Å (Figure d). The surface of the nanocubes is atomically flat; the 4 Å roughness can be attributed to residual ligands crystallized on the surface after solvent evaporation and size dispersion of the nanocubes. RMS value of an entirely different part of the assembly shows similar roughness (Figure S3), indicating the uniform nature of surface roughness throughout the whole sample. Assemblies of the as-synthesized (without thiocyanate exchange) cubes result in short-ranged domains, mostly with multilayer stacking, as shown in Figure S4. Quantitative measurement of the height profile from AFM of different layers stacking is shown in Figure S5. We have analyzed the crystallographic orientation of close-packed monolayer films and multilayer stacked films using XRD, presented in Figure e. In both cases, the peak corresponding to the [100] direction is the strongest, but thiocyanate-capped nanocube films only have [100] and their equivalent peak contributions, which can be attributed to the fact that these cubes are aligned along the same [100] crystallographic orientations to form a close-packed film. This is also clear from the SEM images where even before ligand exchange (Figure S4) there is reasonably good ordering, though often multilayers rather than monolayers prevent perfect ordering. This imperfect ordering leads to the substantial 110 reflection shown in the untreated sample (Figure e), which is absent after thiocyanate exchange, demonstrating the improved assembly and perfect orientation.

Although exchanging the long-chain ligands with thiocyanate improves the assembly in the large area monolayer case, one concern with such postsynthetic ligand exchange is always the accompanying changes in material properties.[39−41] The absorption and photoluminescence (PL) spectra of CsPbBr3 are identical before and after treatment (Figure S6). However, a significant increase in the PL quantum yield (PLQY) from 72.9 to 90.9% is observed after thiocyanate treatment, which is consistent with the literature.[38] Details of the PLQY measurement are explained in the Experimental Section, and full schematic of the setup is explained in the “PLQY Measurement” section in the Supporting Information (Figure S7 and Table S1). Thiocyanate is not completely soluble in toluene but partially interacts with the surface of the cubes forming Cs/Pb–SCN bonds,[38] which is confirmed by Fourier transform infrared (FTIR) spectroscopy, as shown in Figure . The inset of Figure shows a broad peak around 2060 cm–1 that represents C≡N vibrational mode of SCN. This clearly indicates that thiocyanate binds at the surface of nanocubes and replaces some of the long hydrocarbon chains from the surface of the cubes, which is essentially the key factor for the controlled self-assembly.

Figure 3

FTIR spectra of bare silicon (red) without thiocyanate treatment (blue) and thiocyanate-treated CsPbBr3 nanocubes (black).

We perform molecular dynamics simulations using the HOOMD-Blue software[42] with the OPLS united force field[43] and some adapted CHARMM parameters[44] to describe CsPbBr3 nanocubes with either oleyls or thiocyanate at the surface (Figure S8).

We compute the free energy of assembly for a 4 × 4 two-dimensional square lattice with variable lattice constant (ann) and cube size ranging from 13 unit cells (N13—7.63 nm cube edge length) to 2 unit cells (N2—1.17 nm cube edge length) following previous methods (details in Table S2 and Figure S9).[45] Thiocyanate shows a slightly weaker attraction compared to oleyl chains for N13, as shown in Figure . The corresponding internal energy (U), entropy (S) and pressure (P), of the system are shown in Figure S10, displaying the usual energy entropy cancellation described in previous studies.[45] The same trends are also observed for N11, N9, N7, and N2, as shown in Figures S11–S14, respectively. In equilibrium, the system is dry (free of any toluene), and the minimum bonding free energy results from a competition between the attractive enthalpic vdW attraction and the repulsive entropic chain packing (see Figures and S10). The bonding free energy is 40% stronger with oleyl chains than with thiocyanate (−1514 kBT vs −920 kBT for 7.63 nm cube). In both cases, the strong attractive forces are holding the cubes together. This points to a finer assembly control as the thiocyanate nanocubes only get pulled together strongly when they are already close to the proper orientation for ideal packing, whereas the former starts to have strong long-range attractive interactions for less ideal binding configurations; the longer the ligands, the more difficult it is to equilibrate the system. This is related to the strength of the vdW forces that increase very significantly with larger chains, leading to a “cascade effect” where longer ligands may wrap around the nanocrystals. This suggests that the shorter chains facilitate better monolayer assembly. Analysis of the structure of the capping ligands reveals that oleyl chains are completely stretched along the cube faces and curve along the corners (see Figure S15). A detailed formula for the equilibrium lattice constant, derived by adapting the Orbifold Topological Model[46−48] to the case of a cube geometry, is provided in Supporting Information and Figure S16.

Figure 4

Free energy plots of N13 perovskite nanocubes grafted with long oleyl chains and thiocyanate. For oleyl chains, equilibrium distance (r0) = 11.79 nm, minimum bonding free energy (F0) = −1514 kBT, while for thiocyanate r0 = 8.66 nm and F0 = −920 kBT.

Conclusions

By modifying the surface of CsPbBr3 perovskite nanocubes, we have described a novel route to control the interactions for self-assembly, where the long-chain hydrocarbon ligands are replaced by a shorter thiocyanate moiety. This enables high-quality nanocube self-assembly into close-packed monolayers. The monolayer films could also act as seed layers to create halide perovskite single-crystalline thin films for better performances in optoelectronic devices.

Experimental Section

Materials

Cs2CO3 (99.9%), lead bromide (PbBr2, 99.9%), octadecene (ODE, 90%), oleylamine (OAm, 90%), oleic acid (OA, 90%), and ammonium thiocyanate (NH4SCN, 90%) are purchased from Sigma-Aldrich. All of the chemicals are used without further purifications.

Methods

Preparation of Cesium Oleate

Typically, 0.814 g of Cs2CO3 is loaded into 100 mL three-neck flask along with 40 mL of octadecene and 2.5 mL of OA, dried for 1 h at 120 °C, and then heated under N2 to 150 °C until all Cs2CO3 reacts with OA. Preheating of cesium oleate to 100 °C before injection is necessary as it precipitates out from ODE at room temperature.

Synthesis of CsPbBr3 Nanocubes

In a typical synthesis, 0.188 mmol of PbBr2 with 5 mL of ODE, 0.5 mL of OAm, and 0.5 mL of OA is loaded into a three-neck round bottom flask and dried under vacuum at 120 °C for an hour. Then, the reaction atmosphere is made inert by passing N2. After complete solubilization of PbBr2, the temperature is raised to 200 °C and 0.4 mL of the preheated cesium oleate is injected into the three-neck flask. After the injection, the color of the solution turns from colorless to greenish yellow, indicating the formation of perovskite cubes. Then, we lower the temperature to 160 °C and anneal the solution at that temperature for 10 min to get uniform size dispersion of the cubes. After that, we cool the solution using ice water bath for further use.

Isolation and Purification of CsPbBr3 Cubes

After the synthesis, we use two-step centrifugations to collect the cubes for our self-assembly study. First, we take 1 mL from the stock solution just after the synthesis and centrifuge at 8000 rpm for 20 min to collect all CsPbBr3 particles from the solution. We discard the supernatant, wash gently the inner wall of the tube using a tissue paper, and add 2 mL of toluene to disperse the whole CsPbBr3 solid. Then, we perform the second step of centrifugation at 2000 rpm for 5 min to get rid of all the large particles. Now, from the supernatant, we have 2 mL of toluene containing CsPbBr3 nanocubes with a size distribution around 10–15 nm.

Surface Modification of CsPbBr3 Nanocubes

To get well-controlled self-assembly, we have performed surface modification of the cubes from long hydrocarbon oleyl chain to short thiocyanate ion. We use NH4SCN as a source of thiocyanate. About 50 mg of the NH4SCN powder is added in excess into 2 mL perovskite cube solution in toluene in a 4 mL vial containing a magnetic bar. The solution is stirred for 30 min inside a glovebox. To get rid of the insoluble NH4SCN, we centrifuge the solution at 2000 rpm for 2 min and use the supernatant for self-assembly study. NH4SCN is insoluble in toluene but replaces some of the oleate ligands from the surfaces of the cubes, as confirmed from FTIR spectroscopy (Figure ).

PL Quantum Yield Measurement

We measure the PLQY using a custom-modified GPS-033-SL integrating sphere built by LabSphere.[49] A laser diode (Thorlabs, L405P20, 405 nm) is used as an excitation source, passing through an optical chopper (Thorlabs, MC2000B-EC), hitting into the integrating sphere. The incident intensity is controlled with neutral density filters (Thorlabs). The beam hits the sample within a cylindrical cuvette. Light leaving the exit port of the sphere, fitted with a baffle to prevent direct reflections, hits onto a low-noise Newport 818-SL calibrated photodetector, which is connected to a Stanford Research Systems SR830 lock-in amplifier. We measure the excitation and emission separately using a short-pass filter (Thorlabs FESH0450) and long-pass filter (Thorlabs FELH0450) in front of the photodetector. The comparison of the emission and excitation results in the quantum yield. The sensitivity as a function of wavelength is calibrated with the spectral responsivity of the photodetector. A more detailed description of this calculation can be found in the Supporting Information.

Self-Assembly Process and Theoretical Calculations

Details of the monolayer assembly including details of the setup used are described in the Supporting Information. Theoretical calculations with modeling of nanocubes to calculate the bonding free energy during the self-assembly process is also explained in Supporting Information.

Instrumentations

UV–Vis Spectroscopy

Absorption of CsPbBr3 nanocubes in toluene is measured using a Perkin Elmer UV/VIS/NIR spectrometer.

PL Measurement

A 405 nm laser diode (Thorlabs S1FC405) is used as an excitation source, where the PL intensity of the CsPbBr3 perovskites is collected in reflection mode through a NA 0.9 objective using a spectrometer (UHTC 300 VIS, WITec).

XRD

We use Bruker Cu Kα XRD to get the crystallographic orientation of our sample.

FTIR

We perform FTIR to get the surface composition of the nanocubes using VERTEX 80V Bruker instrument.

AFM

We use Veeco Dimension 3100 AFM to measure different thicknesses and to measure RMS roughness of the nanocube film.

SEM

We use the FEI Verios 460 to take all of the scanning electron microscope images shown in the manuscript.