Nonradiative Energy Transfer between Thickness-Controlled Halide Perovskite Nanoplatelets

Abstract

Despite showing great promise for optoelectronics, the commercialization of halide perovskite nanostructure-based devices is hampered by inefficient electrical excitation and strong exciton binding energies. While transport of excitons in an energy-tailored system via Förster resonance energy transfer (FRET) could be an efficient alternative, halide ion migration makes the realization of cascaded structures difficult. Here, we show how these could be obtained by exploiting the pronounced quantum confinement effect in two-dimensional CsPbBr₃-based nanoplatelets (NPls). In thin films of NPls of two predetermined thicknesses, we observe an enhanced acceptor photoluminescence (PL) emission and a decreased donor PL lifetime. This indicates a FRET-mediated process, benefitted by the structural parameters of the NPls. We determine corresponding transfer rates up to k_FRET = 0.99 ns^–1 and efficiencies of nearly η_FRET = 70%. We also show FRET to occur between perovskite NPls of other thicknesses. Consequently, this strategy could lead to tailored energy cascade nanostructures for improved optoelectronic devices.

Halide perovskites are one of the hottest semiconducting materials for optoelectronic applications because of a range of fascinating properties.^1,2 With band gaps tunable throughout the visible range,³ high absorption cross sections,⁴ and photoluminescence (PL) quantum yields approaching 100%,⁵ potential uses range from solar cells^6,7 and photodetectors⁸ to light-emitting diodes (LEDs)^9,10 and lasers¹¹ and even more exotic applications such as gamma-ray detectors¹² and remote thermometers.¹³ With the initial focus on fabricating large-grain thin films for photovoltaics, fabrication has since spread to two-dimensional (2D) perovskite phases and nanocrystals (NCs).¹⁴⁻¹⁷ These possess an additional tuning mechanism, as shrinking any dimension below the excitonic Bohr radius induces quantum confinement, strongly affecting the optoelectronic properties.¹⁸ Examples of this have been demonstrated for 2D nanoplatelets (NPls) and nanosheets, 1D nanowires and nanorods, and 0D quantum dots (QDs).^2,19 Halide perovskites represent an interesting conundrum, as their marvelous properties are counterbalanced by a number of partially significant limitations. Instability to moisture, heat, and strong light exposure;^20,21 migration of ions during operation of devices;²² and the (current) necessity of containing lead²³ are some of the issues currently impeding commercialization. Ion migration, especially, has presented a problem, as it prevents the formation of heterostructures and thereupon reliant strategies like energy funnels.^24,25 A possible workaround could be the exploitation of quantum confinement in perovskite nanostructures, such as NPls, whose absorption and emission properties can be tuned over wide ranges by controlling their thickness.^26,27 Previous studies on NPls of different materials have already shown efficient energy-transfer mechanisms due to large spectral overlap and efficient lateral ordering in thin films.^28,29 This energy transfer was shown to occur very efficiently by the Förster resonance energy transfer (FRET) mechanism, even outpacing detrimental Auger recombination. The FRET process relies on a spectral overlap of the emission of a donor material and absorption of an acceptor material, the orientation of their dipole moments, and a small separation between them. Such an energy transfer was shown in perovskite thin films with varying grain thicknesses and in inhomogeneously broadened NC aggregates, but to date, not on separate monodisperse NCs.³⁰⁻³² Herein, we demonstrate the first such occurrence, using previously gained expertise on precisely controlling the thickness of CsPbBr₃ NPls.³³ Mixing colloidal dispersions of 2 and 3 monolayer (ML) thick NPls, we first show an enhanced PL emission of the thicker NPls, which act as acceptors. Proof of a nonradiative energy-transfer mechanism comes via time-resolved PL measurements in thin films comprising mixtures of 2 and 3 ML NPls. We show an increasing PL acceptor lifetime and a simultaneous decrease of the donor lifetime as the molar acceptor:donor ratio (A:D) of the NPls is varied. With only a narrow ligand spacer between adjacent NPls, the process is highly efficient (η_FRET ≈ 70%) and extremely fast (k_FRET ≈ 1 ns^–1). We also demonstrate that energy transfer is prevalent in NPl mixtures of other thicknesses. These results open the way for highly defined cascaded energy-transfer structures for realizing efficient energy harvesting, high optical gain, and even electrically pumped lasing.

Halide perovskite NPls were synthesized according to a modified version of our previously reported method (see Experimental Methods in the Supporting Information for details).³³ In this past work the main goal was to boost the radiative efficiency of the NPls, which we achieved by applying a passivating step after the synthesis. A drawback of this process is that the NPls tend to be unstable and, especially in films, often rapidly grow thicker, as evidenced by a progressive redshift of the PL maxima. For the energy-transfer measurements, however, it was critical that we produce thin films comprising exactly two distinct NPl thicknesses, otherwise, the lifetimes would be skewed and produce unreliable values.

Accordingly, we modified the synthesis to obtain less enhanced, but very stable NPl dispersions. The most essential parameter for this was the applied volume of enhancement solution, because it turns out that the balance between PbBr₂ and ligands in the dispersions is crucial for the long-term stability of the NPls. For this study, we focused on NPls of two thicknesses, 2 MLs and 3 MLs. To assess the quality of the two individual dispersions we acquired UV–vis and PL spectra (Figure 1a), which reveal sharp excitonic absorption peaks and single narrow PL peaks at 435 and 457 nm for the 2 and 3 ML NPls, respectively, matching our previous results.³³ Importantly, there is a strong spectral overlap between the emission spectrum of the 2 ML and the absorption spectrum of the 3 ML sample. This is an essential prerequisite for realizing a donor/acceptor energy-transfer system. According to this spectral alignment, the 2 ML NPls appear suitable as donors, while the 3 ML NPls can accept the energy from the donor system. Any subsequent reference to donor/acceptor in this Letter refers to 2 ML/3 ML NPls. The narrow PL peaks with full widths at half-maximum (fwhm) of 76 and 104 meV and the small Stokes shifts (60 and 61 meV) are indicative of the monodispersity of the dispersions. Further evidence for this is provided through transmission electron microscopy (TEM) images, as shown for the 2 ML NPls (Figure 1b,c). The 2 ML NPls possess a quadratic shape with a side length of 19 ± 1 nm and a thickness of 1.3 ± 0.2 nm, matching the reported lattice constant of 0.59 nm for CsPbBr₃.³⁴ The 3 ML NPls are also quadratic but slightly larger with a side length of 32 ± 1 nm and a thickness of 1.9 ± 0.2 nm (see the Supporting Information, Figure S1). Often self-assembling into long stacks during the drying process, the NPls passivated with long alkyl chain ligands are separated by 2.3 and 2.6 nm for the 2 and 3 ML samples, respectively. This arrangement and the small distance are important for enabling energy-transfer processes based on dipole–dipole interaction.²⁹

Figure 1.

Thickness-controlled halide perovskite nanoplatelets (NPls). (a) Absorption (dashed lines) and PL (solid lines) spectra of 2 ML (purple) and 3 ML (blue) NPls. (b and c) TEM images of 2 ML NPls from which a thickness of 1.3 ± 0.2 nm and side lengths of 19 ± 1 nm are determined.

To investigate possible energy-transfer processes, we mixed the two NPl dispersions in fixed molar ratios (A:D). To this end, we initially determined the absorption cross sections at 400 nm for the individual NPls and used this in combination with their optical density to calculate the concentration of the dispersions, which we found to be c_2 ML = 1.20 × 10¹⁴ cm^–3 and c_3 ML = 2.47 × 10¹⁴ cm^–3 for the 2 and 3 ML NPls, respectively (see the Supporting Information for details). UV–vis spectra of the mixtures were then acquired and compared to calculated spectra obtained from a weighted addition of the individual NPl spectra (see Figure S2). Clearly, these two match very closely, demonstrating that the absorption of the mixture resembles a simple superposition of the two individual absorption spectra. More importantly, the two NPl species are not affected by the mixing process. PL spectra of the mixtures, however, exhibit a different behavior, because for all mixtures containing a significant amount of both NPl species, the PL emission of the donor is decreased while that of the acceptor is increased (see Figure S3). This suggests that energy transfer already occurs in the mixtures; however, at these NPl concentrations, this is likely due to photons being emitted by the donor and reabsorbed by the acceptor NPls.³⁵ Thus, solid-state films were prepared by drop-casting these mixtures onto SiO₂-coated silicon substrates, and PL emission spectra of these films were acquired (Figure 2, colored curves). To investigate whether energy transfer occurs, we again calculated the expected PL spectra for the case that no energy transfer was present according to the molar ratios of the donor/acceptor NPls (black curves). Here, the curves are normalized to the donor emission peak for clarity. Clearly, in all cases the PL emission of the acceptor is enhanced, again pointing to energy transfer between the NPls. While the drop-casted films are considerably thinner than the cuvettes used for the dispersion, making radiative energy transfer less likely, these results do not exclude reabsorption as the transfer mechanism. For this, time-resolved PL measurements are necessary.

Figure 2.

Energy transfer between NPls in thin films. PL spectra (colored lines) of mixed 2 and 3 ML NPl thin films obtained via drop-casting and ranging from pure 2 ML (top) to pure 3 ML NPls (bottom). Shown in black are the calculated PL spectra obtained by adding the pure 2 and 3 ML spectra in the mixing ratio of the respective species. The increase in intensity of the 3 ML emission peak represents a clear sign of energy transfer between the two NPl types.

Accordingly, we employed a time-correlated single-photon counting (TCSPC) setup to investigate the PL decay and corresponding lifetimes of the donor and acceptor NPls in the solid-state heterostructures. While the donor and acceptor PL emission maxima are easily discernible from one another, as can be seen in Figure 1a, there is a slight overlap of the two spectra. However, for the lifetime measurements, it is crucial to ensure that only the PL emission from either the donor or the acceptor reaches the detector, which is difficult for the more unbalanced mixtures, because of this overlap. To achieve this, we installed two separate edge filters for the 2 and 3 ML samples (see Figure S4a). The PL of the 2 and 3 ML samples decays within several nanoseconds, exhibiting a multiexponential nature, possibly because of different dielectric surroundings or because of bright and dark subpopulations (see black curves in Figure 3a,b).³³ We thus extract the times at which the PL intensity has dropped to 1/e of its original value as a measure for the PL lifetimes, obtaining donor and acceptor lifetimes τ_D = 2.26 ns and τ_A = 1.45 ns, respectively. In the mixed films, this acceptor lifetime increases up to a maximum value of τ_AD = 5.36 ns as progressively more donor is added (Figure 3a). This is a clear indicator of an energy-transfer process, with a feeding of the acceptors leading to a slower PL decay. We were not able to observe a delayed onset of the acceptor emission, likely because we were exciting both samples simultaneously, with the excitation wavelength located above the continuum onset of the two NPl systems. As before, the change in the acceptor lifetime does not define the actual transfer mechanism. Therefore, we investigate the PL lifetime of the donor in the mixed samples (Figure 3b). Here, the opposite trend to the acceptor is visible, with the decay accelerating up to τ_DA = 0.7 ns as the molar acceptor concentration increases. The fact that the donor emission becomes faster means that an additional decay pathway must be present, as simple reabsorption of photons by the acceptor would not change the donor lifetime. Hence, we postulate that FRET occurs from the donor to acceptor NPls as previously observed for CdSe NPl films.^28,29

Figure 3.

FRET-modified PL lifetimes in mixed-NPl thin films. The PL decays of the pure 2 and 3 ML NPls are represented by the black curves. (a) PL decay curves of 3 ML NPls show increasing lifetimes as the molar donor concentration is increased (dark to light coloring). (b) The corresponding PL decays of 2 ML NPls show diminishing lifetimes for increasing acceptor molar ratios. These two trends show the nonradiative nature of the energy-transfer process in the films. The inset shows a schematic of the experimental setup with the filtered emission from the mixed NPl thin films.

To quantify the efficiency of the energy-transfer process, we calculate the FRET rates, k_FRET, and FRET efficiencies, η_FRET, from the donor PL lifetimes in the pure and mixed samples according to³⁶

These two physical quantities are plotted against the A:D ratio (Figure 4). For the smallest amount of acceptor present (A:D = 0.11), both the FRET rate and FRET efficiency assume low values of k_FRET = 0.12 ns^–1 and η_FRET = 22%. As the amount of acceptor is increased, both values rise rapidly. The FRET efficiency saturates at a ratio A:D = 20, reaching a maximum of 69%. The FRET rate increase slows down for the highest A:D ratios but still does not seem completely saturated with a maximally determined value of k_FRET = 0.99 ns^–1. Both behaviors make sense, as a higher amount of acceptor will increase the likelihood of a donor NPl having acceptor NPls in its proximity to which it can transfer its energy. However, at one point all donor NPls will obtain their maximal transfer probabilities. These are limited by the NPl spacing given by the organic ligand layers and by the overlap of emission and absorption spectra of the donor and acceptor, respectively. A further increase of both values might be possible by using shorter passivating ligands or by enhancing the order in the films, as stacks of NPls should be favorable for a more efficient dipole–dipole interaction. Nevertheless, the values are comparable to those reported recently for a similar CdSe-NPl system.³⁷

Figure 4.

Extracted donor PL lifetimes obtained from the mixed NPl thin films were used to calculate both the FRET efficiency (purple circles) and the FRET rates (gray circles). Both values grow with increasing A:D ratio, saturating for large acceptor concentrations.

To confirm the energy-transfer process, we also synthesized NPls of increased thickness, with 4 and 6 ML (see Figure S4b). The PL emission from these two samples peaks at 477 and 503 nm, respectively. The latter, as evidenced by a broader PL spectrum (124 meV fwhm), is clearly not monodisperse, and both samples suffered from stability issues upon mixing with other NPl dispersions. Consequently, we cannot use these to determine the FRET rates and efficiencies between NPls of specific thicknesses. However, we can use them to confirm that nonradiative energy transfer also occurs in these systems. Accordingly, we performed a similar set of measurements for 2 and 4 ML mixtures and for 3 and 6 ML mixtures (see Figures S5 and S6). In both sets, we were again able to observe a slowing of the acceptor decay accompanied by an accelerating donor decay, again indicative of a FRET-mediated process. The FRET rates in all cases tended to be similar while the efficiencies for these latter systems peaked at 49% and 52%, respectively (see Tables S1 and S2). These lower values could be due to the aforementioned issues or due to less efficient processes. More work will need to be done to optimize the syntheses and stabilize the NPls to elucidate this question.

In summary, we have presented the first demonstration of FRET between two defined halide perovskite nanocrystal populations. For this we used CsPbBr₃-based NPls with a precisely adjusted thickness, because these have strongly thickness-dependent energy levels, enabling nonradiative energy transfer. By fabricating thin films comprising different A:D ratios of NPls, we show a progressive lengthening of the acceptor PL lifetime and a shortening of the donor lifetime, indicative of the FRET mechanism. We use the reduced donor lifetimes in the mixtures to estimate the efficiencies and transfer rates of the process. Both increase strongly with increasing A:D ratio, reaching maximum values of η_FRET = 69% and k_FRET = 0.99 ns^–1. The FRET process is not limited to these thicknesses, as we show for two other thickness combinations. With energy transfer between perovskite NCs typically impossible because of halide ion exchange, size-controllable NCs present an interesting alternative for realizing cascaded energy-transfer structures. However, the efficiency of the NCs and most importantly their stability must be improved for this and for achieving optoelectronic integration.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsenergylett.0c00471.

• Experimental Section; materials and methods for the synthesis; concentration calculation of NPl dispersions; TEM characterization; and optical characterization, including UV–vis, PL spectroscopy, and TCSPC measurements (PDF)

The authors declare no competing financial interest.