Re-cyclic photophysics in perovskite Cs₄PbBr₆

Perovskites are one of the most widely used materials in the solar cell industry,^1–4 LED research,^5–8 and other promising future energy-related research.^9–12 In the past, perovskite films have drawbacks of low photoluminescence quantum yields (PLQYs) due to the fast dissociation of the excitons.¹³ This problem was solved through quantum dots' (PQDs) systems, an confinement of the electron–hole pair, but the conversion of PQDs into solid forms significantly reduces their PLQY because removing the stabilizer-ligand promotes particle aggregation.^13,14 Recently, apart from these issues, solar energy harvesting research are suffered from re-absorption of materials.^15,16 Various methods have been devised to collect sunlight over a large spectral area.^15,17,18 However, the loss due to the re-absorption process of the material is inevitable. Because of overlapping of absorption and emission, the perovskite re-absorb the photons emitted.^19–21 To overcome the wasting situation, the efforts of adjusting the spectral broadness and the maximum distance of absorption/emission for high efficient luminescent solar concentrators are ongoing.¹⁶ 

Confinement of the electron–hole pair by the “octahedral isolation” of perovskites was realized by a low dimensional crystal structure.²² Cs₄PbBr₆ is a perovskite-related semiconductor with an unconnected octahedral crystal structure (0-D), and it has the general formula A_nBX_2+n.^13,23,24 Cs₄PbBr₆ is composed of isolated octahedral PbBr₆⁴⁻ interspersed with Cs⁺ cations in the crystal structure. The isolated octahedron of the 0D perovskite-related semiconductor imparts unique quantum behavior, particularly, narrow photoluminescence (PL) emission at ∼520 nm (∼2.384 eV) and high PLQY (∼40%). Although the PL origin of Cs₄PbBr₆ remains controversial, many studies have suggested that the PL occurs because the Br^- ion vacancy state in Cs₄PbBr₆ yields a green emission (∼520 nm).^25–27 In this article, based on a careful characterization of Cs₄PbBr₆, we focused on re-absorption and re-emission properties of Cs₄PbBr₆.

Synthetic procedures and details of characterizations are described in the supplementary material. Pure Cs₄PbBr₆ and CsPbBr₃ perovskite used in this study are synthesized and purified following the published process with a minor modification.^13,24 For optical measurements, we defined the sampling procedure as shown in Fig. S6 of the supplementary material. To make the surface of the sample as uniform as possible, the powder was sandwiched between two glass slides. Afterward, the two glass slides were pressed together and wrapped with a tape. The illustration of a process for generating “surface-emission” (I_S) and “re-emission” (I_R) is shown in Fig. 1(a). In Fig. 1(a), (1) under laser (375/405/488 nm) irradiation, emission is generated at the surface (Surface-emission, I_S), (2) emitted photons are re-absorbed by the crystal (Self-absorption or Re-absorption), and (3) repeated Re-absorption/Re-emission cycles are followed by Re-emission (I_R). In Fig. 1(b), the two emission bands (I_S and I_R), and the absorption band (red) display significant spectral overlaps. The different shapes of the emission bands were obtained from the sample surface (I_S) and a rear surface (I_R) of Cs₄PbBr₆. Under the incident 375 nm laser, a bluish-green PL (I_S) was generated with an E_max (the energy at the highest intensity) of 2.38 eV (≈ 520 nm). Meanwhile, the yellowish-green PL is observed from the back (I_R) with an E_max of 2.30 eV (≈ 539 nm). The shape of the asymmetric emission spectrum of I_R is a characteristic of I_R that results from various forms of an interaction (i.e., scattering or reflection during repeated re-absorption/re-emission cycles) inside the crystal. The absorption spectrum of the Cs₄PbBr₆ was obtained by transforming the diffuse reflectance spectrum [Fig. 1(b)]. As mentioned, Fig. 1(b) shows that a large portion of the I_S spectrum overlaps with the absorption spectrum, which results in active re-absorption and the generation of red-shifted re-emission (I_R).

To understand the pump-power dependence of Cs₄PbBr₆, the power-dependent PL intensity was plotted on a double-logarithmic scale (I_PL ∝ I_Excitation^β). First, β < 1 means free-to-bound recombination, and second, 1 < β < 2 represents recombination similar to that in excitons.²⁸ In Fig. 2, the power-dependent PL intensities fitted with a linear line are shown [Figs. 2(a) and 2(b): irradiation of 375, Figs. 2(c) and 2(d): irradiation of 405, and Figs. 2(e) and 2(f): irradiation of 488 nm]. The β values (β = 1) show no significant variation with 375, 405, and 488 nm laser irradiation. When the pump-power was increased, the β values slightly deviated from 0.9 to 1.1 under an incident laser irradiation within the pump-power range of 30–50 mW (375 and 405 nm laser in each case). Under 488 nm laser irradiation, there was no change in the trend of β value with a pump-power range of 2.6–38.7 mW. For further investigation of the exciton relaxation dynamics of I_S and I_R, time-correlated single-photon counting (TCSPC) measurements were conducted. The PL spectra of I_S and I_R under the 405 nm TCSPC laser are shown in Fig. 3(a). The TCSPC lifetime profiles were measured at the maximum point of the PL intensity (I_S: 2.374 eV, I_R: 2.30 eV). The time-resolved PL (TRPL) decay profiles fitted with the tri-exponential function: I(t) = A₁ exp (−t/τ₁) + A₂ exp (−t/τ₂) + A₃ exp (−t/τ₃). The parameters of lifetimes (τ) and the fractional intensities (A) are summarized in Table I. According to the generally accepted assignment of decay, two fast decay (A₁, τ₁, A₂, τ₂) components represent radiative recombination of excitons, and the third (slow decay) components (A₃, τ₃) relate to a thermally induced decay.²⁹ TRPL decay profiles are plotted in Fig. 3(b). As shown in Fig. 3(c) [the magnification region of −1.5–5 ns in Fig. 3(b)], the rising delay of I_R compared with I_S (∼ 0.5 ns) implies that I_R occurs after several more processes than I_S. The fast-decay (A₁) value in Table I can be used to evaluate the defect density.³⁰ The I_S showed a slightly higher fraction (34%) of defect state originated emission than the fraction for I_R (23%). In the slow decay, the lifetime (τ₃) and the fraction (A₃) of I_R (A₃: 30% and τ₃: 37.64 ns) were higher than that of I_S (A₃: 0.21 and τ₃: 33.75 ns). Moreover, the intensity weighted average lifetime (τ_avg) was obtained using the function: τ_avg = [(A₁ × τ₁²) +(A₂ × τ₂²) + (A₃ × τ₃²)]/[(A₁ × τ₁) + (A₂ × τ₂) + (A₃ × τ₃)]. τ_avg values were 24.06 and 30.22 ns for I_S and I_R, respectively. These results imply that the bright color-emitting electronic states of I_S with I_R are different, and the radiative recombination of I_S is faster than that of I_R.

The optical setup was developed to observe the first re-emission (I_FR) that was generated after the first re-absorption (Fig. 4). Briefly, the goal of a setup is to identify “I_S1-excited surface-emission” (I_S2) and “I_S1-excited re-emission” (I_R2) from sample 2 (S2). Under laser irradiation, I_S1 from sample 1 (S1) is generated, after which S2 absorbs I_S1 and generates I_S2 and I_R2. All PL measurements were conducted at a position 1 cm away from the PL generating position. Also, we specified the distance between S1 and S2 as 5 cm. Between S1 and S2, longpass (LP, 400/500 nm longpass) filters are located. In the supplementary material, the color difference between PL without a LP filter and PL with a LP filter is shown (Fig. S7). Whether the stray photons of the laser beam (scattered/reflected by S1) were completely removed by the LP filter was confirmed with a power meter. Due to the opaque nature of the polycrystalline powder material, the PL generated slightly behind the surface would not be transmitted forward. Therefore, only an I_S was obtained. Because of the specific separation of the I_S with I_R, isolated I_S1 irradiation on S2 was possible. (In the case of a single crystal, the re-emission is transmitted forward and observed from the front.) Interestingly, as shown in Fig. 5(a), I_S2 under 375 nm laser irradiation (I_S2,375) was generated with a slight red shift and one-fourth of the emission intensity of I_S1,375. Moreover, I_R2,375 exhibited a very weak intensity with its E_max identical to that of I_R1,375. The emission from S1 and S2 is represented separately on the right-hand side of Fig. 5(a). The two red-dashed lines are the E_max positions of the surface-emission (I_S1,375 or I_S2,375), and the re-emissions (I_R1,375, I_R2,375) from the two samples (S1, S2) have the same energy (blue dashed line). As previously mentioned, the E_max of I_S1,375 is 2.38 eV, and the E_max of I_R1,375 is 2.30 eV. However, the E_max of I_S2,375 is 2.353 eV, which is different from the E_max of I_S1,375. Also, the E_max of I_R2,375 is 2.30 eV, which is identical to I_R1,375. As stated earlier, the different colors (energy) obtained from the rear sides of the sample are due to the photon re-absorption/re-emission process. Therefore, the I_S2,375 (2.353 eV) is I_FR in Cs₄PbBr₆. In short, the E_max of I_S2,375 (2.353 eV) obtained by irradiating I_S1 on S2 is the I_FR results of the first re-absorption in Cs₄PbBr₆ under 375 nm laser irradiation (I_S2 = I_FR). The same experiments were repeated using 405 and 488 nm laser irradiation, as shown in Figs. 5(b) and 5(c). In Fig. 5(b), the E_max of I_S1,405 and I_S1,488 are 2.374 and 2.358 eV, respectively, which are different from the E_max of I_S1,375. The E_max of I_S1 depends on the wavelength of the laser irradiation. The E_max of I_S2,405 and I_S2,488 are 2.348 and 2.337 eV, respectively, which are red-shifted from the E_max of I_S2,375. Interestingly, the E_max of re-emissions (I_R1, I_R2) under 405 and 488 nm were the same at 2.30 eV. It is noteworthy that six re-emissions (I_R1,375, I_R2,375, I_R1,405, I_R2,405, I_R1,488, and I_R2,488) with the same E_max of 2.30 eV imply that regardless of the sample thickness, E_max of re-emission is always 2.30 eV. The E_max of 2.30 eV means the emission generated from the lowest edge of the conduction band corresponds to the position (2.30 eV) at which absorption begins in the absorption spectrum [Fig. 2(b)]. All the E_max values for the various experimental conditions are summarized in Table II. Figure 5(d) shows the relative intensities with E_max positions of I_S1, I_S2, I_R1, and I_R2. A small difference of three I_S1 E_max depending on the laser wavelength (375 nm: 2.38 eV, 405 nm: 2.374 eV, and 488 nm: 2.358 eV) is shown in Fig. 5(d). In addition, noticeable changes were also observed in the E_max of I_S2 (375 nm: 2.353 eV, 405 nm: 2.348 eV, and 488 nm: 2.337 eV). As previously mentioned, all I_R showed the same E_max of 2.30 eV. The I_S2 is I_FR after the first re-absorption in Cs₄PbBr₆.

Figure 6(a) shows the illustration of the re-absorption/re-emission cycles in Cs₄PbBr₆. Based on PL from the 375 nm laser (a significant energy difference with ∼2.30 eV, which is the absorption edge of Cs₄PbBr₆), the 405 nm laser (closer energy to the absorption edge of Cs₄PbBr₆ than the 375 nm laser), and the 488 nm laser (energy near to the absorption edge of Cs₄PbBr₆) irradiation, we found that least three times of re-emission/re-absorption cycles are required for I_R generation (2.30 eV). Figure 6(b) illustrates the various PL and real absorption spectra. The PL with various colors (energy) from the Cs₄PbBr₆ material has several E_max values that are listed in Table II. We experimentally found that E_max of I_S was 2.38 eV and that of I_FR was 2.353 eV. The I_FR is the observation of the re-emission after the first re-absorption. Based on these results, the number of re-absorption/re-emission cycles for generating final re-emission (2.30 eV, I_R) is estimated. In Fig. 6(b), under 375 nm laser irradiation, the energy difference between I_S and I_FR was 27 meV, and the energy difference between I_FR and I_R was 53 meV. These energy gap differences between 27 and 53 meV represent that the energy intervals between I_S and I_FR and between I_FR and I_R (final re-emission) are close to integer multiples. Therefore, we claim that the 2.30 eV energy value of I_R is a result of approximately three re-absorption/re-emission cycles. Under the irradiation of the 488 nm laser, the difference between I_S and I_FR was 21 meV, and the energy difference between I_FR and I_R was 37 meV. This result shows that the I_R energy is generated in approximately three steps (2.358 → 2.337 → ∼2.316 → 2.30 eV) under the irradiation of the 488 nm laser. The 2.316 eV was determined by considering a difference of 21 meV at the step and not the experimental value. However, a gap of 37 meV (not 42 meV as expected) indicates that the minimum edge of the conduction band was reached. Similarly, the emission approaches to conduction band minimum emission under 405 nm laser irradiation. Despite the difference between the I_S1 and I_S2 gap and the I_S2 and I_R gap is not an integer multiple: 26 vs 48 meV, three times of re-absorption would occur.

In this research, photphysical properties of re-absorption/re-emission cycles in Cs₄PbBr₆ are demonstrated. The time-resolved PL (TRPL) profile suggests that I_S and I_R have originated from different states and have different lifetimes (24 and 30 ns for I_S and I_R, respectively). The Cs₄PbBr₆ showed I_S (2.38, 2.374, and 2.358 eV) and I_R (2.30 eV) under three different irradiation wavelengths (375/405/488 nm). I_S was irradiated on sample 2 (S2) to confirm the I_FR. The surface-emission of the S2 (I_FR) showed E_max values of 2.353, 2.348, and 2.337 eV. The analysis of the I_FR revealed that the energy intervals between I_S and I_FR and between I_FR and I_R are close to integer multiples. Therefore, we suggest that I_R is the result of least three spatially moving re-absorption/re-emission cycles.

See the supplementary material for experimental details of materials synthesis, characterization tools, spectroscopic tools, and characterization details. In Fig. S1, illustration of the synthesis of Cs₄PbBr₆ is shown. XRD patterns and TEM images for as-prepared sample/sample after UV exposure are given in Figs. S2 and S4, respectively, and SEM images and EDS spectra are shown in Fig. S3. XPS spectra of Cs₄PbBr₆ are shown in Fig. S5. The sampling procedure of Cs₄PbBr₆ is shown in Fig. S6. In Fig. S7, picture of color difference between PL without LP filter and PL with LP filter is shown.

K.T.L. was supported by the grant awarded from GIST in 2020 through the Research Institute (GRI) program and by the National Research Foundation (NRF) of South Korea (Grant Nos. 2020R1F1A1073442 and 2021R1A2C2010557). H.L. was supported by the grant awarded from GIST in 2020 through the Research Institute program and by the National Research Foundation of South Korea (Grant No. 2019R1A2C1089731).