Photoluminescence Mapping of Mobile and Fixed Defects in Halide Perovskite Films

TL;DR

Electrical methods often conflate ionic and interfacial effects in halide perovskites, hindering intrinsic ion-transport quantification. This work deploys localized IMPLS to map lateral ion diffusion optically, extracting $D_{\textrm{ion}}$ and introducing a defect contrast coefficient ($\kappa$) to separate mobile from immobile defects, with phase $\theta$ serving as a proxy for PLQY under suitable conditions. The results show mobile ionic defects diffuse laterally from the illuminated region, yielding $D_{\textrm{ion}}$ values in the $10^{-12}$–$10^{-10}$ cm$^2$/s range that align with literature, and beam-size–dependent maps reveal spatial heterogeneity in defect types. By combining frequency- and beam-size–dependent IMPLS with spatial mapping and Moran's I analysis, the study provides a robust, contact-free framework to identify dominant loss pathways and to spatially resolve defect types across perovskite films, with implications for device stability and performance.

Abstract

Metal halide perovskites exhibit coupled electronic and ionic properties that determine their photovoltaic performance and operational stability. Understanding and quantifying ionic transport are therefore essential for advancing perovskite optoelectronics. Conventional electrical methods such as impedance spectroscopy require fully integrated devices, and their interpretation is often complicated by interfacial and contact effects, limiting the ability to isolate intrinsic ionic behavior. Here, a localized adaptation of intensity-modulated photoluminescence spectroscopy (IMPLS) is utilized to optically probe lateral ionic transport in perovskite films. The frequency-dependent photoluminescence response is measured under controlled carrier injection levels and correlated with the photoluminescence quantum yield (PLQY). The proposed diffusion model indicates that mobile ionic defects laterally migrate from high light intensity regions, giving rise to characteristic photoluminescence modulations. Ionic diffusion coefficients extracted from IMPLS agree well with literature values obtained from electrical measurements. Importantly, IMPLS mapping separates mobile and immobile defect contributions through a defect contrast coefficient (DCC), which quantifies the normalized difference between the area-averaged photoluminescence intensity and phase data. This work ultimately demonstrates that localized IMPLS provides a contact-free means to extract lateral ion diffusion coefficients while spatially distinguishing defect types across the sample.

Figures (16)

• Figure 1: PLQY time series measured on an encapsulated triple-cation, mixed-halide perovskite film. The empty regions between the red data points are when the sample's absorptance was measured (see the SI for measurement details). The green curve represents the bi-exponential rise fit to the data, $\textrm{PLQY} = \textrm{PLQYsaturation} - A\exp(-t/\textrm{\texttau 1}) - B\exp(-t/\textrm{\texttau 2})$. The PLQY saturation value from the fit is 15.5% and the extracted time constants are 1 = 42.1 seconds and 2 = 11.08 minutes.
• Figure 2: (a) Schematic of the experimental setup. Laser excitation was applied to the top of the sample via a dichroic mirror and its reflection was blocked using a 420 nm long-pass filter. The LED provided an offset intensity of 38.1 mW/cm2, corresponding to a photon flux of DC,LED = 8.63$\times$1016 cm-2/s. The LED amplitude was set to 15.7 mW/cm2 (AC,LED = 3.56$\times$1016 cm-2/s). For experiment (i) -- the variable frequency experiment -- the laser intensity was fixed at 33.104 W/cm2 (laser = 6.75$\times$1019 cm-2/s). For experiment (ii) -- the variable intensity experiment -- the LED frequency was fixed at 50 mHz. (b) Schematic of experiment (i) for three representative modulation frequencies at a fixed offset intensity. As the modulation frequency decreases (dark blue LED curve), the PL AC amplitude and DC offset increase (red). The relative phase shift (exemplified by the shaded regions from the PL peak to the subsequent LED peak) also increases with decreasing frequency. (c) Schematic of experiment (ii) for three representative laser intensities at a fixed modulation frequency. Increasing laser intensity (light blue laser line) similarly increases the PL DC offset and phase shift. The PL AC amplitude decreases for increasing intensity. The amplitudes, offsets and phase shifts are exaggerated for clarity but follow the same trends observed experimentally in this work.
• Figure 3: Panels (a) - (c) show results from experiment (i), corresponding to Figure \ref{['fig2']}b. Panels (d) - (f) show results from experiment (ii), corresponding to Figure \ref{['fig2']}c. (a) Bode plot of the PL phase shift as a function of LED modulation frequency. (b) PL amplitude and (c) PL offset Bode plots over the same frequency range. The PL offset was normalized to the PL offset signal collected at $f =$ 1 Hz. (d) Bode plot of the PL phase shift as a function of the excess minority carrier density at a fixed modulation frequency of $f$ = 50 mHz. The carrier density was calculated from the combined photon flux of the laser and LED. (e) Corresponding PL amplitude and (f) PL offset over the same $\Delta n$ range, with the PL offset normalized to the signal collected at maximum $\Delta n$.
• Figure 4: (a) PL phase shift data (red markers) as a function of excess minority carrier density (left axis). Averaged PLQY values (25 measurements per point), with standard deviation error bars are shown with the blue markers (right axis). The blue curve represents the PLQY fit using Equation \ref{['eq:PLQY']} and red curve represents the phase fit using Equation \ref{['eq:logspaceG']}. (b) Re-visualizing the data from panel (a) directly as the normalized PLQY versus $\theta$. A linear fit (green line) was applied to the data to visualize the correlated trend. Outliers due to the difference in measurement setups at the high and low $\Delta n$ limits were omitted from the fit. (c) PL intensity phase shift (red, left axis) and corresponding relative PLQY (blue, right axis) as functions of modulation frequency. Both datasets were fitted using a log-linear model, $y(f) = A\log_{10}(f) +B$. (d) Re-visualizing the data from panel (b) directly as the normalized PLQY versus $\theta$. Both datasets in panels (b) and (d) are normalized to their common point (red square) at $f$ = 50 mHz, $\Delta n$ = 6.6$\times10^{16}$ cm-3.
• Figure 5: (a) Phase shift map and (b) corresponding peak PL intensity (defined as the maximum AC and DC PL intensity signal, and taken as the PLQY proxy) over a 25 µ m $\times$ 25 µ m region (1 µ m step size) at $f =$ 50 mHz, collected using the custom-built pulsed laser microscopy setup. High-intensity (DC) excitation was provided by a focused 405 nm laser with a 1.7 µ m beam radius, while modulating excitation was provided using the 450 nm LED. Maps are bicubic-smoothed for visualization; corresponding raw data maps are shown in Figure \ref{['figSIRawMaps']}. (c) Phase histograms from the fluence-matched datasets, measured with laser beam radii of 1.7 µ m (blue) and of 3.2 µ m (red). (d) Scatter plots of the peak PL intensity versus phase shift for both beam sizes, with corresponding linear fits.