Separating Bulk and Surface Contributions to ... - ACS Publications

Subscriber access provided by Olson Library | Northern Michigan University

Letter

Separating Bulk and Surface Contributions to Electronic Excited-State Processes in Hybrid Mixed Perovskite Thin Films via Multimodal All-Optical Imaging Mary Jane Simpson, Benjamin Doughty, Sanjib Das, Kai Xiao, and Ying-Zhong Ma J. Phys. Chem. Lett., Just Accepted Manuscript • DOI: 10.1021/acs.jpclett.7b01368 • Publication Date (Web): 04 Jul 2017 Downloaded from http://pubs.acs.org on July 4, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

The Journal of Physical Chemistry Letters is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry Letters

Separating Bulk and Surface Contributions to Electronic Excited-State Processes in Hybrid Mixed Perovskite Thin Films via Multimodal All-Optical Imaging

Mary Jane Simpson1, Benjamin Doughty1, Sanjib Das2, Kai Xiao3, Ying-Zhong Ma1* 1. Chemical Sciences Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831 2. Department of Electrical Engineering and Computer Science, University of Tennessee, Knoxville, Tennessee 37996 3. Center for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831 *Corresponding author: [email protected]

1 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Abstract A comprehensive understanding of electronic excited-state phenomena underlying the impressive performance of solution-processed hybrid halide perovskites solar cells requires access to both spatially resolved electronic processes and corresponding sample morphological characteristics. Here, we demonstrate an all-optical multimodal imaging approach that enables us to obtain both electronic excited-state and morphological information on a single optical microscope platform with simultaneous high temporal and spatial resolution. Specifically, images were acquired for the same region of interest on thin films of chloride containing mixed lead halide perovskites (CH3NH3PbI3-xClx) using femtosecond transient absorption, timeintegrated photoluminescence (PL), confocal reflectance, and transmission microscopies. Comprehensive image analysis revealed the presence of surface- and bulk-dominated contributions to the various images, which describe either spatially dependent electronic excitedstate properties or morphological variations across the probed region of the thin films. These results show that PL probes effectively the species near or at the film surface.

Table of Contents Graphic


Page 2 of 24

Page 3 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


The remarkable performance of organometallic halide perovskites in photovoltaic and lightemitting applications1-4 has stimulated increasingly extensive studies to understand the underlying fundamental photophysics.5-7 The spatial heterogeneity arising largely from the presence of crystalline grains with varying size, shape, and level of defects in these thin films,6, 812

however, imposes substantial challenges in deriving the underlying photophysics or for that

matter the origin of the observed signals in optical measurements. On one hand, to understand electronic excited-state processes and their spatial dependencies, one requires experimental approaches with simultaneous high spatial and temporal resolution. This need is clearly evidenced by both the observation of the substantially different transient absorption kinetics obtained by integrating either a finite sample area or a smaller region within it with a time window ranging from ∼1 ps time-resolved PL decays.15,

13

to hundreds of picoseconds,14 and reports of spatially dependent

16

This necessity has stimulated several time-resolved optical

imaging studies based on picosecond PL and femtosecond transient absorption microscopy (TAM), which have successfully identified the spatially dependent nature of elementary photoexcitations,14 mapped the spatial distributions of distinct photoexcited species17 and electronic trap states,18, relaxation,13,

15, 16

19

analyzed the spatial dependence of electronic excited-state

assessed thermal annealing induced changes of electronic transition dipole

moment orientations,20 and determined the characteristics of charge carrier diffusion.21, 22 On the other hand, to reveal the physical mechanisms underlying the observed spatial dependencies of various electronic excited-state phenomena, it is essential to learn about the corresponding morphology for the system under study. A combination of steady-state or time-resolved PL with structural characterization methods such as scanning electron microscopy (SEM) or/and atomicforce microscopy (AFM) has been successfully applied to gain new insights into solution-



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

processed perovskite films.15,

16

However, the methods to correlate morphology with spatially

resolved electronic excited-state spectra and dynamics are non-trivial, requiring precisely locating a purposely marked sample region of interest (ROI) in order to co-register images obtained using different microscope platforms. This technical difficulty becomes even more challenging for chemically unstable samples, like hybrid perovskites, that undergo rapid and irreversible degradation upon exposure to moisture and/or oxygen,23, 24 where the incorporation of protective layers on top of the samples to avoid this adverse effect necessarily makes them no longer suitable for SEM or AFM characterization.

Here, we report a multimodal all-optical imaging study of thin films of chloride containing mixed lead halide perovskite (CH3NH3PbI3-xClx), which includes femtosecond TAM, timeintegrated PL, transmission, and confocal reflectance imaging microscopies. Our results not only provide spatially and temporally resolved pictures of electronic excited-state dynamics in these complex films, but also yield morphological information that is intrinsically spatially registered on a single optical microscope platform. We observed remarkably rich features in both PL and confocal reflectance images along with clear pixel-to-pixel correlation between them, whereas strong pixel correlations are absent for other image combinations. These results reveal in an unambiguous way that the PL from the sample is dominated by emission from the surface/near surface species. This finding underscores a potentially inherent difference in the electronic excited-state properties probed with commonly used reflection-geometry-based PL and transmission-detected transient absorption measurements on condensed phase materials with strong optical absorption, a prerequisite for all light-harvesting materials. More importantly, this difference can impact both microscopic and spectroscopic results acquired using these


Page 4 of 24

Page 5 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


techniques as well as the physical mechanism and interpretations extracted from those measurements.

Our multimodal imaging apparatus is sketched in Fig. 1 and described in more detail in the Supplementary Information. Among the four imaging modalities, time-integrated PL and femtosecond TAM were applied to probe the spatial distributions of electronically excited species in a selected ROI, whereas transmission and confocal reflectance imaging were used to gain morphological information for the same selected sample region. Besides their distinctly different capabilities, these imaging modalities employ different laser beam geometries. Both PL and confocal reflectance images were acquired with reflection beam geometry, in which the same objective was used to focus the incoming laser beam and collect the PL and confocal reflectance signals. In contrast, the TAM and transmission images were collected using a transmission geometry, which involves the use of a second objective to collect the transmitted pump and probe beams after the sample. This all-optical, single microscope based imaging approach has several advantages over recently implemented schemes,15,

16

namely, it is non-

invasive, non-contact and capable of either sequential or simultaneous measurements on the same sample ROI with comparable spatial resolution, which eliminates the need for tedious sample marking and transporting, as well as image co-registering post data acquisition.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 1. Sketch of the multimodal microscope involving four modalities. To switch between modalities while preserving the ROI, a dichroic filter (DC) and a silver mirror are switched using a sliding cube mount. The inset shows a SEM image of the mixed lead halide perovskite thin film sample.

Fig. 2 shows the PL, TAM at ∼5 ps, transmission and confocal reflectance images acquired for the same ROI of a CH3NH3PbI3-xClx perovskite thin film. As detailed in the Experimental Methods and the Supplementary Information, the TAM measurements were performed using pump and probe pulses centered at 500 and 750 nm, respectively. Given that the probe pulse is resonant with the lowest electronic transition of the perovskite material under study, the transient absorption signal is dominated by a positive-signed photoinduced transmission signal owing to contribution from both photobleaching and stimulated emission. Acquisition of the other images was carried out using only the 500 nm pulses.


Page 6 of 24

Page 7 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 2: Femtosecond TAM at 5 ps delay time (a), time-integrated PL (b), confocal reflectance (c), and transmission (d) images acquired for a mixed perovskite thin film on a single microscope platform. The scale bar indicates 5 µm. Images are normalized to the maxima of the signal intensities or transient absorption amplitude, respectively. Image size is 20× 20 µm, corresponding to 128 × 128 pixels, and the black arrows indicate the corresponding spatial features.

Each image shown in Fig. 2 has been normalized to the maximum of the corresponding signal intensity / amplitude in order to account for the subtle differences in laser intensity, sample absorbance, optical alignment, and laser pulse widths used for acquiring these optical images that were performed one after another at the same sample position. This allows us to directly compare the various images acquired from the same sample ROI on a universal relative scale. From these images, one can immediately identify several striking features. First, distinct spatial features are observed in all images with the smallest ones less than 1 µm diameter, which are comparable to the small grains seen in the SEM image obtained for a thin film sample prepared under the same conditions but with no protective coating (see Supplementary Fig. S1 for a representative image). Second, corresponding spatial features can be readily recognized in at least three images as indicated by the black arrows. Although these corresponding features are sparse, their presence confirms that the same sample area was probed using these imaging 7 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

modalities, and negligible sample ROI displacement occurred during sequential acquisition of the images. Third, the substantial, and far more common, differences found in most portions of these acquired images point to different physical mechanisms that generate image contrast from one modality to another. While the image contrast for both confocal reflectance and transmission is obvious, those associated with TAM and PL are worth noting. The former arises from the product of a spatially and temporally dependent carrier density and the sum of photobleaching, stimulated emission and excited-state absorption cross sections, whereas the latter results from the product of the squared carrier density and spatially dependent PL quantum efficiency. Fourth, both PL and confocal reflectance images show significantly richer spatial features than the transmission and TAM images. To verify the spatial features observed in the confocal reflectance image indeed comes from the sample instead of the microscope coverslip, a control image from a coverslip was also taken. As shown in Supplementary Figure S3, the control image is featureless and therefore the features seen in Fig. 2c can be safely attributed to the perovskite surface.

As normalizing these images to their maximum intensity/amplitude can potentially wash out some spatial features with weak signal levels, we also performed normalization using the image averages and then set the color scales to two standard deviations below and above the average values (Supplementary Figure S4), where an enhanced contrast is obtained for low intensity spatial features while sacrificing contrast between some of the high intensity ones (i.e., saturation of the color scale). Upon careful inspection of these images one can find that all essential features can be readily identified in either set of these images.

To be more quantitative about the similarities and differences between the acquired images shown in Fig. 2, we present the normalized intensity/amplitude histograms shown in Fig. 3. 8 ACS Paragon Plus Environment

Page 8 of 24

Page 9 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Although all the distributions are asymmetrical, clear differences between them can be readily identified. First, the distributions obtained for the PL, transmission and confocal reflectance images are skewed left, whereas the one for the TAM image is skewed right. Second, the most striking distinction between these distributions lies in their spreads. Specifically, the histograms associated with both PL and confocal reflectance images have much larger spreads than those seen in the TAM and transmission histograms. Third, while all of these distributions are bimodal and require two Gaussian functions for satisfactory description, considerable differences in the center, full-width at half maximum (FWHM), amplitude, and area are found for each Gaussian component of all distributions. These differences can be seen qualitatively from the fits shown in Fig. 3 (dashed lines) and quantitatively from Supplementary Table S1, where all the fitting parameters are summarized.

Different spreads in histogram distributions have been observed recently in the TAM and PL images acquired from the thin films of pristine halide perovskites (CH3NH3PbI3) that were subjected to 1 and 2 hours of thermal annealing.19 For both samples, we found that the histogram distributions associated with the TAM signals are clearly narrower than the corresponding histogram distributions of the PL data. This previous observation is generally in line with the results shown in Fig. 3, except that the TAM histogram distributions observed in our previous study are symmetric instead of bimodal. In view of the fact that TAM is sensitive to both emissive and non-emissive photoexcited species, whereas PL microscopy probes only the emissive species, the observed differences were considered as strong evidence for the presence of spatially dependent PL quantum efficiency. It should be emphasized that the observation of remarkably different PL and TAM image histogram distributions in the current and previous



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

work even for different samples, pristine and mixed perovskites, suggests an intrinsic cause underlying the observed differences.


Page 10 of 24

Page 11 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 3: Histograms of the peak TA amplitudes (a) and the corresponding normalized intensities of the PL (b), confocal reflectance (c) and transmission (d) signals. The red solid lines are the fits to a sum of two Gaussian functions depicted by the dashed green and yellow lines.

To explore the potential correlations between these images, we generated pixel-to-pixel twodimensional (2D) histograms for every possible pair of images collected, where the normalized intensities (or amplitudes) of the individual signals are plotted. Among all six combinations (Fig. 4), obvious diagonal correlation is found only between the PL and confocal reflectance images, whereas no immediately obvious correlations are observed for the other pairs. A quantitative measure for the presence or absence of image correlation is to calculate the Pearson correlation coefficients for the 2D histograms in Fig. 4a-f, which are -0.123, -0.079, -0.053, 0.571, 0.433, 0.399, respectively. Based on these coefficients, one can see the obvious correlation between the PL and confocal reflectance images is characterized by the largest Pearson coefficient (0.571), whereas the absence of such correlation between the TAM and all other three images is reflected by small coefficients. Furthermore, both the coefficients for the transmission/PL (Fig. 4e) and transmission/confocal reflectance (Fig. 4f) correlations are comparably smaller than that obtained for the PL and confocal reflectance images. Furthermore, the presence and absence of these pixel-to-pixel correlations between the images are also verified using K-means clustering17, 25

analysis, which shows that the PL and confocal reflectance images have spatial regions that

strongly correlate as described in the Supplementary Fig. S5.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 24

Figure 4: 2D histograms for all six combinations of the normalized intensities of PL, transmission and confocal reflectance signals, and the corresponding peak TA amplitudes. (a) PL/TAM, (b) confocal reflectance/TAM, (c) transmission/TAM, (d) confocal reflectance/PL, (e) transmission/PL, and (f) transmission/confocal reflectance, respectively. The number of pixels at each position is represented by different colors; a color change from red to blue corresponds to a decreasing number of pixels. Each histogram has 128 bins in each dimension.

In order to understand the presence and absence of pixel-to-pixel correlations, we begin with identifying the dominant contributions to each of the modalities. To do so, we need to keep in mind that the laser beams were first incident on the glass coverslip, then propagated through the ∼275 nm thick perovskite film, and finally reached the conformably coated poly(methyl methacrylate) (PMMA) protecting layer. As light reflection occurs at any interface between


Page 13 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


materials with different refractive indices, it is clear that the primary contribution to the confocal reflectance image comes from the front surface of the thin film sample, which faces the incident laser beam (i.e., the glass coverslip/sample interface). Given the finite flatness of the sample surface with root mean square (RMS) roughness of 30–40 nm, one would expect that only those tall crystalline grains are in contact with the glass coverslip. For the short grains, there must be a gap between them and the coverslip, which presumably contains N2 as the sample was prepared in a N2-filled glovebox. This difference in the heights of the morphological features has been found to strongly influence the PL intensities of pristine lead halide perovskites measured using two-photon total internal reflection fluorescence microscopy.20 Based on the reported large refractive index for pristine perovskites (≈3.2 at 500 nm),26 and the comparably small indices of refraction for both N2 gas27 and glass,28 we can expect spatially variable reflection depending upon the grain height. This also explains why some small features observed in the confocal reflectance image are comparable to the grain size seen in the SEM image (Supplementary Fig. S1). If we further assume perfectly flat grains, a reflectance for the short and tall grains can be estimated to be 0.16 and 0.27, respectively, according to the Fresnel equation under normal incidence. This consideration is further supported by the negligible contribution from the back sample surface where a change in index of refraction occurs as well. Because of the high extinction coefficient of the thin film samples used, which is approximately 1.2×105 cm-1 at 500 nm for CH3NH3PbI3-xClx films,29 the incoming pump beam will be attenuated to

Separating Bulk and Surface Contributions to ... - ACS Publications

Recommend Documents