Twin domains modulate light-matter interactions in metal halide perovskites

Metal halide perovskites (MHPs) are a promising class of materials for optoelectronics.^1–6 To date, the recorded solar to electricity power conversion efficiency (PCE) of MHP solar cells has achieved 25.2%.⁷ The PCE of a solar cell is closely related to light absorption, charge carrier separation and transport, and recombination processes. High photoluminescence (PL) intensity and slow PL decay dynamics are generally taken as indicators of better performance.^8–11 Nanostructures—such as grains and grain boundaries, twins and twin walls—that are associated with local defect sites have critical effects on the performance of MHPs.^12–14 

Numerous studies have shown the important role of grains and grain boundaries in MHP photophysics.^13,15–19 Correlating scanning electron microscopy (SEM) and confocal fluorescence microscopy revealed that the PL intensity and lifetime varied between different grains and that the PL decayed faster at grain boundaries.¹³ Furthermore, grain boundaries are associated with charge separation and collection,^17,18 and effective passivation of grain boundaries results in the increase in the PL intensity and lifetime.¹⁷ It was also suggested that grain boundaries acting as potential barriers do not dominate recombination, as evidenced by similar PL lifetime at grain boundaries and grain interiors.²⁰ These insights have greatly improved the performance of MHP solar cells,^{14–16,21–25} suggesting that nanostructures are crucial for MHP optoelectronic devices. Although a few studies showed the difference in photocurrent of twin domains in MHPs,^26,27 the research interest of MHP twin domains has mostly focused on their ferroic nature and their effect on optoelectronic properties is barely understood.^26–35 Theoretical simulations have predicted that twin domains and domain walls in MHPs can significantly affect the electronic bandgap,³⁶ charge separation,^36,37 charge transport,³⁸ and photovoltaic performance.³⁹ Therefore, it is critical to understand how twin domains affect the photophysics of MHPs.

In this work, we synthesized CH₃NH₃PbI₃ microcrystals with twin domains of several micrometers wide. We investigate these samples by scanning probe microscopy, scanning electron microscopy, and confocal PL microscopy. We found that twin domains have minor effects on PL, evidenced by the uniform PL lifetime and the same wavelength of emitted light in all regions. Furthermore, according to the PL intensity variation between neighboring domains we propose that the twin domains affect the light-matter interaction but have a negligible impact on the local electronic properties of CH₃NH₃PbI₃.

The CH₃NH₃PbI₃ was synthesized by spin-casting a CH₃NH₃PbI₃ precursor—1M CH₃NH₃I and 1M PbI₂ in 0.75 ml DMSO and 0.25 ml DMF—directly on the indium tin oxide (ITO)/glass substrates. The as-casted film was annealed at 100–110 °C until it fully converted to a dark perovskite. SEM was used to check the morphology of the films. The SEM micrograph [Fig. 1(a)] shows the hexagonal CH₃NH₃PbI₃ crystals. Twin domains around several micrometers wide are observed in the sample [Fig. 1(a)]. We performed band excitation piezoresponse force microscopy (BE-PFM) to study the twin domains. The BE-PFM was performed in vertical on a commercial atomic force microscopy (AFM) system [Cypher (Asylum Research and Oxford Instruments Co.)] using a Pt/Ir coated tip with a nominal stiffness of 3.0 N m⁻¹. It is worth noting that the twin domains in MHPs were also studied using single frequency PFM near resonance frequency.²⁹ However, resonance frequency could couple with changes in local mechanical property and results in amplitude and phase contrast in PFM measurement, which can be misinterpreted as piezoelectric contrast.^34,40 The BE-PFM tracks resonance frequency and hence provides information on the local mechanical properties.⁴¹ The BE-PFM results [Figs. 1(b)–1(e)] show the same twin domain structures in amplitude [Fig. 1(c)], phase [Fig. 1(d)], and resonance frequency shift [Fig. 1(e)] maps, indicating that the amplitude and phase channels are coupled to the resonance frequency channel. This suggests that the PFM contrast is affected by the mechanical variation between the neighboring domains.^29,34

We also investigated how twin domains impact local PL behavior by correlating SEM and PL spectroscopy. After finding domains with SEM [Fig. 2(a)], the sample was transferred into a confocal Raman microscope instrument with a CCD detector, which allowed us to detect the PL spectrum on a microscale region. We note that no polarization filter was used in this measurement. Some twin domains are also visible in the optical image [Fig. 2(b)], enabling precise locating the region of interest. We monitored the PL under continuous-wave (CW) 532 nm laser illumination at an intensity of 0.1 mW and a spot size of ∼1 µm and obtained the PL spectra of four different spots locating in the neighboring domains, as marked in Fig. 2(b). The PL spectra in neighboring domains [Fig. 2(c)] are very similar, which shows a <1 nm variation in peak position. PL at certain energy (or wavelength) is generally ascribed to the electronic bandgap of the material; thus, these results indicate that the neighboring domains have a similar bandgap.

Confocal PL intensity and PL lifetime measurements were also conducted to study the twin domains. The confocal PL measurement was performed on an Olympus IX81 inverted microscope (0.50 NA 20× dry objective) using a 405 nm pulsed diode-pumped solid state laser at 5 MHz repetition rate and 10 nW average power at the sample. PL in the range from 532 nm to 800 nm was collected in the epi-illumination scheme, spectrally separated from the excitation laser light by a dichroic (Semrock, DiO-442), and spatially filtered by a 100 µm pinhole and then imaged onto a single photon counting avalanche photodiode [Micro-Photo-Device (MPD) Picoquant] coupled to a time-analyzer (PicoHarp 300, PicoQuant). No polarization filter was used in this measurement, either. The time resolution of confocal PL measurement is 60 ps. Data acquisition and data analysis were performed with the Symphotime 64 analysis software (Picoquant). The acquired data were analyzed and plotted using Python 3.6. Figure 2(e) shows the PL intensity map, where a small intensity variation between neighboring domains is observed. These are consistent with the twin domains shown in SEM [Fig. 2(d)]. Additionally, a spatially resolved PL lifetime measurement was performed in situ. The PL lifetime map [Fig. 2(f)] shows the local PL decay dynamics. The uniformity of Fig. 2(f) suggests that the variation in local PL decay dynamics is negligible. A PL intensity vs time diagram measured by using a Horiba Fluorolog III spectrometer with 401 nm NanoLED as the excitation source is shown in Fig. 2(g), detailed lifetime information on this sample is labeled in Fig. 2(g).

In another region, the twin domains in the SEM micrograph [Fig. 3(a)] are not consistent with the striped variation in PL intensity [Fig. 3(b)]. We speculate that two different kinds of twin domains—one kind of domain shows up in SEM [Fig. 3(a), this kind of domain does not affect PL intensity] and the other kind of domain results in the striped PL intensity variation [Fig. 3(b)]—may coexist in this region. The coexistence of different twin domains in the same region has been reported previously.⁴² Therefore, we propose that only a portion of twin domains can affect PL behavior, while other twin domains may not impact the PL behavior of CH₃NH₃PbI₃. Exactly what kind of domains can affect PL intensity may depend on the domain configuration (e.g., crystallographic orientation and chemical composition) and needs further exploration. Nonetheless, there is still no variation in PL lifetime [Fig. 3(c)] in this region.

The PL intensity and lifetime generally correlate with the quality of the crystal, and strong PL response and low PL decay dynamics are widely regarded as an indication of a high-quality sample with a low density of nonradiative recombination centers, e.g., defect sites. Therefore, in our sample, the variation in PL intensity is unlikely to originate from the difference in local defect density (as evidenced by the uniform PL lifetime). This conclusion is also supported by the same PL spectra between neighboring domains because the difference in defect density may also induce a shift in PL peak position. Therefore, we propose that the PL intensity variation between neighboring domains is induced by the difference in light-matter interaction. In principle, ferroelastic twin domains have different crystallographic orientations (as previously indicated),⁴³ which can result in anisotropic optical properties, and as a consequence, change the light-matter interaction between domains. This light-matter interaction difference could be due to light-absorption, light-transport, and light reemission (due to light-trapping). Finally, the difference in light-matter interaction leads to the variation in the PL intensity. However, PL lifetime and PL peak position are not affected as this anisotropic optical behavior is likely not associated with defect sites and electronic properties. Altogether, these results suggest that the impact of twin domains on the PL behavior is minor, implying that the twin domains are likely not a determining factor for photovoltaic action in MHPs.

The proposed mechanism of light-matter interaction based on our PL studies may also be a reason of the previous discoveries of the difference in photocurrent between neighboring domains.^26,27 For example, if different light absorption between neighboring domains is an origin of the PL intensity difference, such a different light absorption can also result in a difference in photocurrent between neighboring domains. We would like to note that both our PL study here and previous photocurrent studies focus on the properties of the center of twin domains.^26,27 In contrast, theoretical studies demonstrated that it is the domain wall rather than the center of domain that affects the optoelectronic properties of CH₃NH₃PbI₃.³⁷ In addition, it has been found that the effect of twin domains in CH₃NH₃PbI₃ also depends on domain size.^44,45 These imply that the density of the domain wall (or, in other word, domain size) is important to the effect of these twin domains. However, limited by the spatial resolution of the PL microscope, we only studied the domains that are submicrometer to several micrometers wide. According to theoretical studies, the effects of domain walls, as well as domains with different sizes, may be different. In particular, considering the ferroelasticity of these domains, the strain condition near domain walls is expected to be different, which may have a significant impact on optoelectronic properties.³⁷ Therefore, further exploration that sheds light on the nanoscale crystal structure will be useful for further clarifying relevant questions.

In conclusion, we studied the PL behavior of the twin domains in the CH₃NH₃PbI₃. Our results indicate that the twin domains have a minor impact on the PL behavior of CH₃NH₃PbI₃, which only alter the local PL intensity but do not affect the PL decay lifetime and the wavelength of emitted light. According to the PL results, we suggested that the twin domains could alter local light-matter interactions in CH₃NH₃PbI₃. We expect that this work will motivate more exploration on the photophysics of twin domains in MHPs.

This research was conducted at the Center for Nanophase Materials Sciences, which is a DOE Office of Science User Facility. The research was supported by the Center for Nanophase Materials Sciences at Oak Ridge National Laboratory (Y.L., L.C., A.V.I., S.J., K.X., A.B., and O.S.O.). The research was partially sponsored by the Air Force Office of Scientific Research (AFOSR) under Grant No. FA 9550-15-1-0064, AOARD (Grant No. FA2386-15-1-4104), and the National Science Foundation (Grant No. CBET-1438181) (B.H.). This research used resources of the Center for Functional Nanomaterials, which is a U.S. DOE Office of Science Facility at Brookhaven National Laboratory under Contract No. DE-SC0012704.

The authors declare no competing financial interests.

This manuscript has been authored by UT-Battelle, LLC, under Contract No. DE-AC0500OR22725 with the U.S. Department of Energy. The publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. The Department of Energy will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan).