Phase-Segregation and Photothermal Remixing of Mixed-Halide Lead Perovskites

Abstract

Mixed-halide lead perovskites (MHPs) are promising materials for photovoltaics and optoelectronics due to their highly tunable bandgaps. However, they phase-segregate under continuous illumination and/or electric field, whose mechanism is still under debate. Herein, we systematically measure the phase-segregation behavior of CH₃NH₃Pb(Br_xI_1-x)₃ MHPs as a function of excitation intensity and nominal halide ratio by in situ photoluminescence micro-spectroscopy. We encapsulate the MHPs with a layer of polystyrene polymer film to isolate them from the effects of the immediate atmosphere. Under this passivated condition, the phase segregation of the MHPs is very different from those without polymer passivation reported in the literature. The initial phase segregation to I-rich and Br-rich phase is observed followed by the formation of a new mixed-halide phase within several seconds that has not been reported before. We observe that the photothermal effect is amplified at the small-size I-rich domains which significantly changes the local phase segregation in the otherwise uniform film as early as milliseconds after illumination.

Graphical Abstract

Hybrid organic-inorganic mixed tri-halide perovskites (MHPs) are materials with a chemical formula of ABC₃ where A is a filling cation (or mixed cations), B is the central cation (e.g. Pb), and C is the structural anion C = X_xY_yZ_z. where X, Y, Z are either Cl⁻, Br⁻, and I⁻, and the stoichiometry number x+y+z = 1. The perovskite structure is maintained by the average sizes of A, B, and C that follow the Goldschmidt tolerance factor.¹ Their advantages over pure halide perovskites are tunable bandgap² with simple doping procedures^2–5 and still maintain excellent carrier mobility,^6–9 and high solar-cell efficiency¹⁰ or photoluminescence (PL) efficiency^4,11 Thus, it is promising in applications such as efficient and color-tunable LEDs,¹² tandem solar cells,¹⁰ and hetero-structured microelectronic devices.¹³

Phase segregation has been a major observation in MHPs, which can either hurt the device’s performance¹⁴ or enhance efficiency.^15,16 Unusual blue-shift of the emission pick was also observed¹⁷ and was confirmed by Hoke and coauthors to be the phase segregation due to the halide migration.^18–26 The Hoke effect (or light-induced phase segregation) is reversible.²⁷ The bandgap of the materials ranked wide to narrow from Cl-rich, Br-rich, to I-rich domains. As such, the carriers can be intrinsically split among the domains which either extend the lifetime of the carriers or hindered them from being collected depending on the locations of the domains relative to the electrodes.^14–16 The phase segregation has been suspected to be initialized by either single-exciton²⁸ or the accumulation of one kind of carrier on the surface of a grain or a nanocrystal. This internal electric field breaks Pb-halide bonds and moves halides around,^29,30 which often generates neutral X₂ species.^29,31,32 When the neutral X₂ (e.g. I₂) is lost into the environment or reacts with oxygen, irreversible degradation is observed,^29,31,32 which should be prevented by passivation.³⁰ Under the dark, the halides migrate back in the order of several minutes to hours at room temperature.^{19,26,33–35}

While we have the big picture of the phase segregation, the detailed mechanism is far from being settled. This open question requires a lot of more experimental data and theoretical simulations to answer. Using UV-Vis extinction and PL spectra to study MHPs’ phase separation has been well established in the literature,^17,19,20,35 The vertical ion migration on the film and the lateral ion migration in nanowires have been reported by measuring PL with a confocal microscope.^13,33,36 In this report, we track the phase segregation of methylammonium lead halide MHP films (CH₃NH₃Pb(Br_xI_1-x)₃) over space and time using a home-built PL emission micro-spectroscope.³⁷ Briefly, the films are illuminated under a wide-field fluorescence microscope (supporting information, SI Figure S1). A slit and a transmission grating are inserted before the EMCCD camera. A narrow slice of the spatially-resolved 0^th order image of the illuminated film (through the grating) is selected by the slit and its associated 1^st order diffraction image is collected by the same camera. The first order image provides the spatially-resolved spectra of the film whose emission wavelengths are calibrated using the standard emission peaks from a few noble gas lamps.

To begin with, CH₃NH₃Pb(Br_xI_1-x)₃ thin films with a range of Br:I ratio from 0.5:1 to 9:1 are prepared by the solvent-washing method,³⁸ which is then passivated by a ~1 μm thick layer of polystyrene film (MW = 280 kDa). The experimental details are described in the SI. The initial extinction edge and PL peak of the as-prepared MHP thin films blue-shift as a function of the increasing Br:I ratio (Figure 1A, 1B), which is consistent with the literature.^2,19 The spectra of both pure-Br and pure-I perovskites are also included in Figure 1 as references. The film is then placed under continuous illumination for 200 seconds (473 nm CW laser). The initial PL spectra of the pure-halides stay and those of the mixed-halides red-shift over time and the degree of the shift is dependent on the illumination power density (Figure 1C). The lower the illumination intensity, the more red-shift. At a constant power density of 16 W/cm², the extent of the red-shift appears to be strongly dependent on the initial Br:I ratio (Figure 1D).

Figure 1.

(A) Extinction and (B) PL spectra of as-prepared MHP films with different Br:I ratios measured with 473 nm laser at 16 W/cm² excitation intensity and 0.1 s exposure time. Inset showing a scheme of the samples. (C) The PL of the MHP (Br:I = 9:1) after 200 seconds illumination with varying excitation intensity (labeled, ‘0’ is the initial PL). (D) The PL of MHPs with varying Br:I ratios after 200 seconds illumination at constant power density (473 nm, 16 W/cm²).

To gain insights into these observations, we obtain the time-dependent PL spectra of each MHP film. Figure 2 shows one set of data under 16 W/cm² illumination. At time zero (t = 0), the initial MHP PL peak (a) appears instantly upon illumination (the same peaks are shown in Figure 1B). A PL peak (b) around 750 nm quickly shows up within 3 seconds. The degree and rate of the shift of peak (b) are consistent with the typical segregation of an I-rich phase described in the literature.^19,31 Surprisingly, a new peak (c) between peak (a) and (b) emerges at t ~ 6 s. Overtime, this peak keeps on growing and slowly red-shifting. At the same time, peak (b) gradually disappears. Interestingly, for MHPs with lower Br:I ratio (<5:1), the peak (b) re-appeared as the formation of the peak (c) levels-off.

Figure 2.

(A-F) Time-dependent PL spectra of MHPs with various Br:I ratio under 16 W/cm² illumination. The bottom half of each figure is the zoom-in of the first 10 s data. The arrows indicate the initial PL peak (a) at t = 0 s, the new peaks are labeled as (b) and (c) in (A). The intensity represented by the color bar is normalized to the maximum intensity of the series with the lower half of (A-C) boosted two times for better contrast.

Figure 3 shows the full sets of data in addition to the data shown in Figure 2. While the three-peak shift is similar among all samples with different Br:I ratios at high power densities (4-16 W/cm²), different dynamics is observed at lower laser power densities. Briefly, the three-peak dynamics is not apparent for samples with larger Br:I ratios at lower laser power (left-bottom corner of Figure 3A, 3B).

Figure 3.

The evolution of the PL spectra of MHP films with various Br:I ratio (column) under various excitation intensity (row) over time (color bars) during the first (A) 10 seconds and (B) 200 seconds illumination. The wavelength axis is from 450-900 nm for each spectrum. The height axis of the spectra is normalized to the maximum intensity of the spectra series.

We fit peak (b) and peak (c) in Figure 2 and Figure 3 (see SI Figure S3 for fitting examples) and assign the peaks to species with molar fraction of iodine X_I according to the calibration curve of peak (a) over the initial composition (Figure 4 inset). The linear relationship between the X_I and the peak center suggests a uniform mixing of the as-prepared film after thermal annealing. Then we plot X_I for the peaks of all MHP films as a function of illumination intensity in Figure 4. The MHP with Br:I = 0.5:1 that has a less significant peak shift is not included in the figure. The initial composition of the film is represented by peak (a) shown in Figure 4 at I_exc = 0. The arrows point to the transition from the peak (a) to peak (b) and (c). With increasing excitation intensity, peak (b) and (c) split further on this diagram.

Figure 4.

The iodine molar fraction X_I of peaks (a-c) in Figure 2 at various excitation intensities I_exc The X_I is calculated based on the calibration line shown in the inset. The arrows indicate the peak shifting direction from the initial composition to the new components. Circle: represents the composition of peak (a); diamond: peak (b); square: peak (c).

These results (Figure 1–4) clearly indicate phase segregation. The initial appearance of peak (b) at t < 10 s represents the I-rich phase, associated with the typical phase-segregation observed on MHPs.^{2,19,21,28,33,39} Majority of the photo-generated carriers are funneled to the I-rich phase that has the lowest conduction band-edge.^31,40,41 Thus, the Br-rich phase left behind is under a dark state (i.e. no emission), which if emitted would have shown a peak blue-shifted from the initial peak (a). In the literature, the I-rich phase disappears along with an onset of the Br-rich phase due to the photodegradation of the I-rich phase in the air or in vaccum.^29,31,42 Surprisingly, we don’t observe the Br-rich phase, rather a new I-rich phase, peak (c), emerges that is reversible under dark or under lower excitation intensity (SI Figure S4–S5).

The formation of the new I-rich phase (peak c) under light is because of the polymer passivation on our samples, which blocks the evaporation of I₂. This passivation shifts the equilibrium to a new I-rich phase under light, which is suspected to be driven by the photothermal heating.²⁶ The Helmholtz free energy of the film is a function of internal energy, temperature, and entropy, F = U − TS.^43,44 Increasing temperature enhances the effect of entropy that drives the remixing of the I-rich and Br-rich phases.^26,43 According to the mass balance, the dark Br-rich should shift back to merge the new I-rich phase. The actual temperature of the film is still challenging for us to directly measure. However, we can estimate the temperature indirectly using the glass transition temperature of the polystyrene thin film, which is T_g = 370 K.⁴⁰ We titrate the laser power density at room temperature T_R = 297 K and find that the polystyrene film on top of the perovskite layer changes morphology above 20±5 W/cm² (SI Figure S6). Assuming a linear correlation between ΔT and I_exc,⁴⁵ the temperature of the perovskite film under illumination is estimated to be ΔT = I_excΔT_g/20, where ΔT_g = T_g − T_R. Thus, the equilibrium temperature of the film under 16 W/cm² is estimated to be 330 K, around the peak of the spinodal temperature of the phase diagram of MHPs calculated in the literature.⁴⁴ However, since the energy is mostly funneled to the I-rich domains, their temperature can be significantly higher than this value. Thus, they may remix under even lower laser intensities due to this local heating effect. With this proposed mechanism, we hypothesize that there will be formation and dissolution of I-rich domains over time under illumination. Thus the spatial behavior of the film is explored using the same PL micro-spectroscope (SI).

The MHP film with Br:I = 9:1 is used in Figure 5 as an example to discuss the spatial behavior. The associated video and the complete data are shown in the SI (SI video and Figure S7, S8). The 0^th order diffraction image of the film (through the grating) shows the spatially-resolved grains in a thin slice of the film (~2.7 μm wide and 40 μm long). The resolution is diffraction limited ~300 nm. At the same time, the spectra of the grains can be observed in the 1^st order diffraction image. At the initial stage (the first few seconds), all the grains appear to have very similar PL peak (a) suggesting relatively uniform initial composition throughout the film. In a few seconds, randomly distributed I-rich domains show up over space with emission peak shifted to peak (b). Overtime, the PL spectra of the grains coalesce back into the same PL spectra (Figure 5, 200 s), suggesting re-homogenization of the phase-segregated materials.²⁶ This behavior is consistent with the proposed photo-thermal mechanism.

Figure 5.

Time-dependent PL micro-spectroscopy of the MHP film with Br:I = 9:1 under the illumination power density 16 W/cm² with the 473 nm laser. (A) The 0^th order PL image of a vertical slice of the MHP film. (B) The corresponding PL spectra of the locations in (A) indicated by the dashed lines, respectively. The hot-spots have I-rich emission spectra. The full video is shown in the SI movie S1.

The degree of the local heating is anti-correlated to the size of the I-rich domains. For the MHPs with a large Br:I ratio (> 5:1), the I-rich phase quickly dissolves under light (Figure 2, Figure 3, Figure 5). For these films, there is less amount of iodine that nucleates into small domains, which gain a lot of energy from their Br-rich surroundings through energy transfer. Once dissolved, the local temperature drops and another round of phase segregation can happen at this location again (SI movie S1). For the MHPs with small Br:I ratio (< 3:1), the I-rich domains can grow big and do not over heat which could explain why they never completely dissolve even under strong light intensity (Figure 2).

Our results on the phase segregation under passivation and relative strong laser excitation are relevant to many applications of MHPs in photovoltaic and optoelectronic devices. A few methods have been developed to prevent phase segregation from affecting the device efficiency. Enhancing crystal thermodynamic stability of MHPs using FA as the A-site cation and carefully adjust Cs doping level obtained mixed halide perovskite that is more stable than those use MA and Cs,^46,47 which either raise the kinetic barrier for ion migration or thermodynamically lower the spinodal decomposition temperature.^28,44,48 Film thickness, precursor composition and ratio, synthetic process and steps, have effects on the phase stability.^33,39,47,49 Under passivation, we have demonstrated that the phase segregation of the MHP films can be different from bare films. Under illumination, segregated domains with different compositions form and homogenize dynamically, which we hypothesize to associate with the local heating effect that should be investigated in the future.