Tunable Angular Light Emission of Lead Halide Perovskite Nanocrystal Thin Films via Solution-Processed Substrate Treatment

Abstract

Lead halide perovskite (LHP) nanocrystals have demonstrated a significant electronic response to their local environment due to their ionic lattice nature. Here, we demonstrated their tunable dipole alignment via solution-processed methods. We synthesized LHP nanocubes and nanoplates in air and characterized them by UV–vis spectrophotometry and transmission electron microscopy. Using atomic force microscopy, UV–vis spectrophotometry, and back focal plane fluorescence microscopy, we characterized thin films of nanocubes on untreated glass, nanoroughened glass, and polymer film (poly­(methyl methacrylate), PMMA), as well as a perovskite nanocubes-nanoplate binary film on etched glass. Most notably, the dipole orientation factor can be modulated from 0.47 to 0.59 (effective transition dipole moment angle from 47° to 40°) by using glass or PMMA, respectively. Understanding the tunable anisotropic transitions in these materials at the nanoscale is required to control light emission into specific modes, which will maximize efficiency in devices such as light-emitting diodes, photovoltaics, and quantum information technology.

Introduction

Perovskite nanocrystals have emerged as one of the most promising semiconductor material classes. Their highly tunable light emission has already made them attractive for a wide range of applications, including lighting, encryption, solar, , laser, sensing, , scintillation, and now quantum information technologies. − Additionally, inorganic lead halide perovskite (CsPbX₃ (X = halide)) (LHP) nanocrystals have exhibited desirable properties such as remarkably high quantum yields, quantum coherence and quantum optical properties, − single photon emission, − and superfluorescence. − This young class of materials has the potential to address the grand materials challenges of next-generation devices.

However, what drives the unique photophysics of LHP nanocrystals is not fully parsed, especially in their alignment of electronic dipoles. The alignment and relative strength of emissive states determines the emission angle of light, characteristic energy transfer, and the ultimate efficiency of any optoelectronic device. − In most cases (i.e., dyes, chalcogenide nanoparticles), anisotropic dipole alignmentand therefore more exaggerated directional emissionis achieved through the alignment of confined or anisotropically shaped structures. , Either films of randomly oriented emitters or of unconfined, isotropic nanoparticles are generally expected to yield an isotropic dipole orientation. Recent work showed that the alignment of the electronic dipoles within films of unconfined perovskite nanocubes, which is expected to be isotropic, was tunable via changes to their local nanoscale environment through substrate modifications, , often showing a more exaggerated vertical dipole component and therefore more oblique light emission. This trend was also observed in confined perovskite nanoplates, which showed a higher vertical dipole than expected. This effect was exaggerated as surface coverage decreased: sparser films (particle assemblies <100 nm) showed even stronger vertical dipole components than their denser counterparts. Studies of this electronic surface effect on perovskite emitters are still limited. Further, films with nanostructuring or texturing of the emissive layer are of interest as this has been shown to improve light out-coupling efficiency. − Thus, understanding this interplay of dipole alignment and local environment will be key for maximizing optoelectronic device efficiency and accessing interesting photophysical behavior especially for future devices with ever-decreasing size footprints. −

Herein we quantify the unusual electronic anisotropy that nanostructured films of LHP nanocubes demonstrate in response to solution-processed substrate treatments. We investigated four substrates: untreated glass (untreated), nanoroughened glass (etched), insulating polymer film (polymer-treated), and perovskite nanoplate predeposition (perovskite-treated). After depositing films of LHP nanocubes on each film, we quantified their structural and optical properties by UV–vis spectrophotometry, transmission electron microscopy (TEM), atomic force microscopy (AFM), and back focal plane fluorescence microscopy (BFPFM). We demonstrated that nanoroughening, which may increase surface interactions, ultimately had the same enhanced vertical dipole response as untreated glass. However, the presence of bluer nanoplates, which introduced energy transfer effects, showed slightly more in-plane dipole alignment, and depositing LHP nanocubes on a polymer substrate yielded near-isotropic dipole orientation, even for these sparser films. Here we achieved a response by the effective transition dipole moment angle, which is determinant to the angular emission of light, thereby accessing advanced photophysical control via a facile benchtop process and expanding the library of materials to which perovskite nanoparticles uniquely respond. ,,

Synthesis and Characterization of Colloidal Perovskite Nanocrystals

LHP nanocubes were synthesized under ambient conditions via ligand-assisted reprecipitation adapted from Brown et al. In short, cesium carbonate was complexed with octanoic acid to form cesium octanoate and then injected into a lead bromide-octyl phosphonic acid precursor in air at room temperature under constant stirring. After 30 s, a solution of didecyldimethylammonium bromide in toluene was injected to quench the reaction. The colloidal product was cleaned and size-selected before further use (see Supporting Information). Confined LHP nanoplates were synthesized via a microemulsion approach adapted from Pan et al. In short, cesium bromide dissolved in hydrobromic acid was injected into a lead bromide solution in hexane containing N,N-dimethylformamide and an oleylamine-oleic acid ligand pair. Immediately upon injection, the reaction was quenched by a small addition of ethanol. The crude product was centrifuged, and plate-containing supernatant was retained as product without further treatment.

UV–vis spectrophotometry was used to characterize the absorbance and relative photoluminescence spectra of colloidal nanoparticles in solution (Figure (a) and (b)) and the relative photoluminescence of thin film samples (Figure (c)). The deposition process for thin films is described in the following section. For nanocubes in suspension, the absorbance spectra showed a characteristic exciton peak at 495 nm and the photoluminescence spectra showed a narrow emission peak at 503 nm with a full width-half-maximum (fwhm) of 20 nm. Nanoplates in suspension showed a blueshifted absorption exciton peak (450 nm) and emission peak (462 nm), indicating confinement. The nanoplate photoluminescence also had a larger red tail, indicating that some nanoplates fused into larger, thicker flakes. ,

1.

Absorbance and normalized relative photoluminescence spectra for colloidal suspension of (a) CsPbBr₃ nanocubes with emission at 503 nm and (b) CsPbBr₃ nanoplates with emission at 462 nm. (c) Relative photoluminescence spectra for thin films of CsPbBr₃ nanocubes (503 nm), nanoplates (462 and 489 nm due to plate fusing), and a bilayer of nanocubes and nanoplates (469 and 500 nm). The peak at 438 nm is a spectral artifact from the excitation light source. (d)-(e) Transmission electron micrographs of nanocubes (estimated length 7.4 nm) assembled in a monolayer. (f)-(g) Transmission electron micrographs of larger (100s nm long) fused populations and smaller (estimated length 9.5 nm, 3 monolayers thick) nanoplates used as a spectrally distinct emitter.

TEM was used to characterize the size and morphology of the nanocubes and nanoplates. Nanocubes were estimated to be around 7.4 nm in length. Nanoplates were estimated to be around 9.5 nm laterally, and the larger flakes of fused particles were 100s of nm laterally. The small nanoplates are estimated to be 3 monolayers thick based on the emission wavelength. Both populations of nanoplates had sufficiently blue-shifted emission that could be distinguished and optically filtered from the nanocubes being studied.

Fabrication of Perovskite Nanocrystal Thin Films

Thin films of LHP nanocubes were deposited by spin coating as-synthesized nanocube suspensions onto different substrates, creating four distinct environments: untreated borosilicate glass (untreated), nanoroughened borosilicate glass (etched), insulating polymer (polymer-treated), and a nanoplate predeposition (perovskite-treated). Untreated borosilicate glass was used to compare the behavior of CsPbBr₃ nanocubes to prior studies. , A caustic etch of potassium hydroxide in ethanol (pH 11–12) was used to treat the borosilicate glass and understand how roughening might also exaggerate the electronic substrate interaction by increasing surface area. We also investigated the use of an insulating polymer layer as the substrate. An approximately 40 nm-thick film of poly­(methyl methacrylate) (PMMA) was deposited via spin coating. PMMA has a similar refractive index to borosilicate glass (1.49 vs 1.51, respectively) and thus should only alter the local electronic environment and not the optical environment, allowing for a direct comparison between the glass and PMMA substrates. Finally, to elicit how Förster resonant energy transfer (FRET) or more complex exciton transport schemes affected dipole orientation, , perovskite nanoplates were deposited via spin coating before depositing nanocubes to create a binary film. Because these nanoplates had the same composition as the nanocubes (CsPbBr₃) and the emitting layer thickness is much smaller than the wavelength of light, any changes in the photoluminescence spectra and angular emission resulted from the interactions from high energy (nanoplates) to low energy (nanocubes) components and not from optical or chemical changes (e.g., halide diffusion of mixed samples).

There was no significant shift in the photoluminescence when the nanocrystals were deposited into the thin films. Thin films of nanocubes alone showed an emission peak at 503 nm, and thin films of the nanoplates alone showed an emission peak of 462 nm (Figure (c)). However, the nanoplate film also showed a significant secondary PL peak at 489 nm (Figure (c)), which was attributed to the fusion of plates. Finally, the perovskite-treated sample (binary population) showed two distinct emission peaks at 469 nm and at 500 nm, which aligned well with the peaks from the nanoplate and nanocube colloidal suspensions, respectively. The nanocube signal intensity was more than five times greater than the nanoplate signal intensity. From this we concluded that the less intense nanoplate signal could be sufficiently cut out using a long-pass filter, allowing us to isolate and study the angular signal of the nanocubes only.

We characterized the conformality and thickness of nanocube films on each substrate and substrate roughness by using atomic force microscopy (AFM). Figure shows the thickness variation over a representative area (2 × 2 μm) as well as a representative thickness profile of the nanocubes on each substrate. For all nanocube-only cases, the nanocube island thickness was estimated to be 4–8 nm (average 5.98 nm), which agreed reasonably well with the particle size of 7.4 nm estimated from TEM. The nanocube-only substrates showed similar total coverage of monolayer groupings or “islands” of nanocrystals, but islands of nanocubes were larger and more irregular across the untreated wafer (Figure (a)) as opposed to films on etched or polymer-treated wafers (Figure (b), (c)). The perovskite-treated sample formed slightly thicker islands on average (10.31 nm). Additionally, we confirmed that the etched substrate had a larger roughness than the untreated substrate (Root Mean Square roughness of 3.50 and 0.76 nm, respectively). The polymer-treated substrate had an RMS roughness of 1.56 nm. The nanocube-nanoplate binary film shows a higher total coverage of nanocrystals than the other three substrates, around 34% versus around 15% for cube-only samples. While the two species are indistinguishable using AFM, the presence of PL signal from both populations was observed in the binary film. See Supporting Information for further details.

2.

Atomic force micrographs of CsPbBr₃ nanocube thin films fabricated via spin coating onto a silicon wafer that was (a) untreated, (b) etched, (c) polymer-treated with 40 nm-thick PMMA, and (d) perovskite-treated with spectrally distinct CsPbBr₃ nanoplates. Islands of nanocubes assembled in a monolayer were present in each sample, and the untreated silicon had an irregular distribution compared to the other nanocube-only samples. The average height of these islands across nanocube-only samples was 5.98 nm (see Supporting Information and Figure S4). The average height of the nanocubes and nanoplates together was 10.31 nm. Illustrative extracted cross sections (background normalized to 0 nm) showing height profiles corresponding to nanocubes on (e) untreated, (f) etched, (g) polymer-treated, and (h) perovskite-treated wafers. The white line in each scan (a)-(d) shows where cross section was extracted.

Quantifying Angular Light Emission

We resolved the angular emission pattern of each sample using back focal plane fluorescence microscopy (BFPFM), also known as Fourier plane fluorescence microscopy. The focal plane in a traditional imaging microscope was transformed from the sample plane to the Back Focal Plane by inserting an additional lens (Bertrand lens) between the microscope objective and the tube lens. The Back Focal Plane shows a 2D projection of the angular emission pattern. , We performed BFPFM using a home-built inverted microscope with excitation by a 405 nm continuous wave laser and a 100× oil immersion objective. A linear polarizer allowed us to study the p-polarized angular emission. Longpass filters were used to isolate the nanocube photoluminescence, removing any background laser or nanoplate emission. Additional details are included in the Supporting Information.

The horizontal and vertical dipole contributions from a sample were determined by fitting the average collected emission pattern across 25 images to a simple three-layer analytical model which determines the angular emission pattern resulting from a thin emissive film on a substrate (here, glass) surrounded by air. ,,− Orientation factor (Θ) was defined as the relative strength of in-plane dipoles to all dipoles, according to the convention given by Kim et al.

$$\mathrm{\Theta }=\frac{p_{\parallel }}{p_{\parallel }+p_{\perp }}$$ 1

where p is relative dipole strength of the parallel and perpendicular dipoles. An effective transition dipole moment for the system as described in can be defined as

$$TDM=\arccos (\sqrt{\mathrm{\Theta }})$$ 2

for the convention where TDM is the angle measured between the horizontal and vertical dipole contributions. The following general equation describes the polarized light emission intensity projected onto a 2D collection plane (N _pol ) as a function of photon momentum vectors along the x- and y-directions (k _x , k _y ): ,

$$N^{pol}(k_x,k_y)=C[\frac{1}{2}({\rho }_{IP}^p+{\rho }_{IP}^s){\cos }^2(TDM)+{\rho }_{OP}^p{\sin }^2(TDM)]$$ 3

where C is a normalization constant arising from the integration time of the collection media and excitation intensity, ρ _ip and ρ _ip are the density of states of s- and p-polarized in-plane dipole contributions, and ρ _op is the out-of-plane dipole contribution. The density of states is a function of the photoluminescence wavelength, the refractive indices of the three layers, and the thickness of the emissive layer. When modeling density of states for our samples, we used a wavelength of 503 nm (peak photoluminescence of the nanocube films), a refractive index of 1.92 (estimated from TEM using an effective medium approximation , ), and a thickness of 0 nm, which is an accepted approximation for submonolayers of nanocrystals or when thickness is much smaller than the wavelength of light. , See Supporting Information for further details.

Because we investigated fluorescence from the same nanocrystals across substrates with nearly similar outcoupling behavior, the density of states (ρ _ip , ρ _ip , ρ _op ) remained the same across all substrates. Thus, any change in the angular emission pattern arose from changes only in the effective dipole orientation. Figure (a) models how the projected Back Focal Plane evolves as effective TDM, or dipole alignment, changes. This is shown clearly in the p-polarized cross section (k _y = 0), or the cross-section taken across the center of the image. When there are only horizontal dipoles present (TDM = 0°, Θ = 1), the p-polarized k-space axis (k _x ) decreases sharply to 0 at |k _x | = 1. As the vertical dipole strength increases (TDM > 0°, Θ < 1), the signal increases for |k _x | ≥ 1. When all dipoles are vertical (TDM = 90°, Θ = 0), the p-polarized axis shows instead an abrupt increase in photoluminescence intensity at |k _x | = 1. A TDM of 35° (Θ = 0.67) is expected for samples with isotropic dipole contributions (two horizontal dipoles and one vertical dipole), which can be seen in either unconfined, isotropically shaped nanostructures or randomly oriented assemblies of particles or dyes, such as CdSe quantum dots.

3.

(a) Modeled BFPFM signal for a thin film of known effective transition dipole moments of 0° (all dipoles in-plane), 35° (isotropic), 50°, and 90°. The extracted cross sections show how TDM changes the signal. (b) – (e) Experimental BFPFM signal, best fit BFPFM model, and effective TDM for each substrate: (b) untreated (TDM = 47 ± 1°), (c) etched (TDM = 46 ± 1°), (d) polymer-treated (TDM = 40 ± 1°), and (e) perovskite-treated (TDM = 44 ± 1°). (f) Extracted k _x cross sections for emission on each substrate. Shading indicates one standard deviation from the average k _x cross section. (g) Orientation factors for LHP nanocubes on a given substrate in the literature as compared to this work for high and low coverage films. Dashed line indicates isotropic dipole orientation.

Figure (b)-(e) (left-hand column) shows the experimental BFPFM images of perovskite nanocube thin films on substrates of untreated glass, etched glass, polymer-treated glass, and perovskite-treated glass. Additionally, the extracted k _x cross section is shown in Figure (f) to highlight the response to each substrate. The shading around each cross section shows one standard deviation from the average. The untreated and etched samples showed overlapping k _x cross sections, demonstrating the same dipole orientations. However, both the polymer-coated glass and the perovskite binary film showed a reduced signal at |k _x | ≥ 1 outside of the variance of the previous samples, demonstrating a more horizontal dipole alignment compared with the other treatments. The polymer-treated sample had a much stronger decrease, indicating the most horizontal alignment.

We then fit each BFPFM image to the theory in eq to quantify the orientation of the dipoles as the effective transition dipole moment angle (TDM) and the orientation factor (Θ) for these four substrates. The best fit to the theoretical angular emission pattern is shown in the right-hand column of Figure (b)-(e). An isotropic dipole alignment (TDM = 35°, Θ = 0.67) would be expected in the case of no structural or confinement anisotropy in the emitting film. However, as shown in Figure (g), perovskite nanocrystals have shown an enhanced vertical dipole component for most surface treatments studied in prior work ,, and for sparse films, in particular. This effect is likely due to charge redistribution in response to the substrate. , When two dissimilar materials are brought into contact, charge is generated due to a work function mismatch. The work function mismatch between glass (5.0 eV) and perovskite (4.0 eV) creates interfacial charge and, because of the ionic nature of the perovskite lattice, charges redistribute within the perovskite near the surface, generating a more exaggerated vertical electronic dipole that is not seen in traditional semiconductor nanocrystals with more covalent lattices. Additionally, it is theorized that this redistribution is even stronger for isolated groups of particles, and indeed more vertically oriented dipoles have been observed for sparser films.

The untreated glass substrate in our study fit to a TDM of 47 ± 1° and orientation factor Θ = 0.47, also showing an enhanced vertical dipole component. This corroborated prior work from Jurow et al., who showed that a similar density of nanocubes on a glass substrate had a TDM of around 46.8°. We hypothesized that the nanoroughening of glass might further influence the dipole alignment by increasing the surface area and thus the generation of interfacial charge. While the roughening did improve the film uniformity as shown by AFM, this treatment was shown to have no effect on the dipole alignment, yielding a TDM of 46 ± 1° and orientation factor Θ = 0.48, statistically the same as the untreated glass substrate.

As shown by the k _x cross sections, nanocubes on a PMMA substrate had a greater fraction of horizontal dipole contributions than those on the glass substrates. This led to a lower effective TDM angle (40 ± 1°, Θ = 0.59) that approached isotropic dipole behavior (TDM = 35°, Θ = 0.67); this change in orientation was statistically significant as compared to the etched treatment (p-value = 7.54e-07). Notably, this polymer-treated substrate achieved the same nearly isotropic behavior as more densely packed films from prior studies (Figure (g)), which is surprising given that sparse films tend toward more vertical dipole alignments. PMMA did not show a significantly increased contact angle with water over untreated glass (θ_water = 63 ± 7° versus θ_water = 59 ± 2°), indicating similar substrate polarity. Instead, our hypothesis for this TDM effect builds on intuition established by Jurow et al. The work function difference between PMMA (4.68 eV) and perovskite (4.0 eV) is smaller than glass (5.0 eV) and perovskite, and as a result, the polymer-treated substrate is expected to have fewer interfacial charges and therefore less charge redistribution. Our observations indicate that having a thin layer of PMMA is sufficient to reduce the vertical dipole strength by over 20%, bringing these nanostructured films much closer to their expected dipole alignment.

As a precursor to more extensive studies of the relationship between FRET and TDM, we studied nanocubes in the presence of nanoplates. Interestingly, as in other on-glass films, a vertically asymmetric TDM was again observed, though less than the untreated and etched substrates. After spectrally isolating the nanocube signal, the angular emission patterns fit to a TDM of 44 ± 1° and orientation factor Θ = 0.52. The extracted orientation for the binary film on an etched substrate was statistically different than the nanocubes alone on an etched substrate (p-value = 0.00015). While dipole-mediated energy transfer is a well-studied effect, ,, we note the effect of energy transfer on the TDM of perovskite nanocubes could not be decoupled from the effects of increasing film coverage (15% to 34%) in this study. , Further, charge generated at the nearby glass interface is highly relevant to exciton behavior in both the perovskite nanocubes and nanoplates. We note the work function of CsPbBr₃ increases with confinement, so work function mismatch of the different populations of nanoplates could contribute to this effect. Further studies should continue to investigate the mechanism of the unique optoelectronic response of CsPbBr₃ nanocubes in a binary film.

Conclusions

We showed the angular photoluminescence response of LHP nanocrystal thin films is due to changes in the local electronic environment through substrate modifications. LHP nanocubes were synthesized via in-air methods and deposited via spin coating onto four substrates that changed the electronic environment while maintaining a comparable optical environment. We compared untreated glass, nanoroughened glass, a thin insulating polymer film, and a distinctly emitting nanoplate treatment. We used back-of-flight fluorescence microscopy, which provided a 2D spatial projection of the 3D angular emission signal, to quantify the effects of the substrate on light emission. We fit the resulting emission patterns to a photonic density of states model and extracted the effective transition dipole moment (TDM) orientations. Nanostructured films of nanocubes on untreated borosilicate glass, nanoroughened glass, or in the presence of confined nanoplates exhibited an enhanced vertical dipole component (TDM = 44–47 ± 1°, Θ = 0.52–0.47), while nanocubes on a PMMA substrate showed nearly isotropic dipole alignment (TDM = 40 ± 1°, Θ = 0.59).

Generally, films of randomly oriented particles are expected to behave as isotropic emitters at an interface (TDM = 35°, Θ = 0.67). However, by leveraging this unique electronic response of perovskite nanocrystals to their local environment, we can tune the emission mode directly to the desired application, allowing improved external quantum efficiency and energy transfer in solar concentrators or LEDs. ,, Specificity of emission mode may also enable enhanced coherence for quantum information technologies, for which perovskite nanocrystals are already a promising candidate. ,, This work demonstrates that perovskite nanocrystals can access tunable photophysics via facile benchtop processes, opening the door to a greater suite of quantum and new-wave semiconductor research by reducing the processing expense of such materials systems.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsnanoscienceau.5c00054.

• Synthetic procedure, spin coating procedure, measurements and measurement analysis, derivation of PDOS model of light emission (PDF)

L.E.P. and C.N.E. conceived of the idea of the manuscript. L.E.P. carried out all experiments and analysis and wrote the manuscript with support from B.R. and C.N.E. B.R. wrote the code for fitting BFPFM signal to a model of dipole alignment and assisted on the fitting of experimental BFPFM images. C.N.E. advised the project. All authors have given approval to the final version of the manuscript.

L.E.P. is funded by the National Defense Science and Engineering Graduate (NDSEG) Fellowship. Material is based upon work supported by the National Science Foundation under Grant No. 2240140.

The authors declare no competing financial interest.