Optoelectronic devices like photovoltaic (PV) cells, light-emitting diodes (LEDs), and radiation monitors are very interested in halide perovskite semiconductors because they can be used in a lot of different ways. When these materials are added as active layers to these devices, they work very well. In single junction configurations, they perform better than 26%, and in tandem configurations, they perform 33.7% better. Perovskite PV goods are almost ready to be sold commercially. Several companies are already using well-established production lines to make them.
These materials don’t follow the rules in most semiconductor textbooks because they are disorganised, poorly treated, and have a lot of defects, but they still work very well in devices. The disorder shows up in things with chemical, molecular, optical, and physical qualities on a range of length scales, from the biggest to the smallest. But there are still some problems that make it hard to use halide perovskites in everyday life.
Ion movement and other material changes that happen when external stressesors like light and/or external bias are applied shorten the life of the gadget in the end. This problem happens a lot in X-ray direct detectors and LED device designs where strong electric fields are put across the active layer. These problems may keep making these gadgets less useful.
Also, not all flaws are so harmless; there are energetically deep carrier traps that work as nonradiative recombination centres, which means they lose a lot of power. These traps lower the amount of luminescence that can be produced, and device contacts are likely to affect or even cause them. When electrical circuits are put under stress from the outside, even “standard” bandgap materials that are close to their radiation efficiency limits when they are first made become unstable.
To make technology better, we need to fully understand the chemical, structural, and morphological properties of materials on different length scales, as well as how these properties affect carrier recombination (both radioactive and nonradiative) and transport, as well as how these relationships change over time while the device is in use.
Researchers are trying to find these kinds of relationships between properties and figure out what causes the disorder in halide perovskites by using powerful microscopic methods to collect data from different measurement methods, also known as multimodal measurements. These can be done at the same time or one after the other on the same scan area.
The grain-to-grain, sub-grain, and atomic length scales are the most important but also the least known. By overcoming these problems and pushing the limits of both testing and theory even further, we can learn more and come up with smart ways to make these materials work better and be more stable as they become commercially available.
Multimodal methods are strong ways to figure out how key features of materials and gadgets relate to their structure, function, and performance. Parallel measurements on nominally similar samples can, in theory, lead to different conclusions about their properties. However, multimodal approaches make it possible to draw direct relationships with little uncertainty, even when the materials are used in device-like environments. This view will be mainly about using tiny methods to look at the qualities of films and gadgets as they are made and how they change as they are used.
Early Multimodal Characterization Work
Nonradiative recombination in semiconductor materials has a direct effect on device performance by lowering effective charge-carrier densities and quasi-Fermi level splittings, which are important measures of PV performance. Early computer studies of theory showed that flaws on the atomic level in metal-halide perovskites make shallow traps, which makes them harmless. This idea caused people to think of perovskites as “defect-tolerant” materials.
The performance of perovskite materials can be on par with III–V semiconductors like GaAs, even when there are a lot of defects (1015–1017 cm−3 for polycrystalline perovskite films). This is because III–V semiconductors can handle defects much better and can get defect levels as low as 107 cm−3. But, in experiments, a lot of nonradiative recombination is still seen. This keeps perovskite devices from working as well as they should (30% and 35% for single-junction and tandem PV devices, respectively).
Photoluminescence (PL) maps that show changes in space make nonradiative recombination easier to see. In 2015, de Quilettes et al. used scanning electron microscopy (SEM) and PL to look at MAPbI3 thin films that had been formed with Cl-based predecessors in the same area. This showed that the luminescence of perovskites varied in different places. Nonradiative recombination, which was especially bad at grain boundaries, slowed down the performance in these dark areas.
It was odd for other thin-film semiconductors for PV, like GaAs, to have different light qualities in a material that was already getting over 20% efficiency in PV devices. It looked like the variety was connected to the underlying shape. This meant that the grains in the perovskite were either different in makeup or structure, which affected how well they worked.
In conclusion, getting rid of nonradiative recombination and making material features more uniform could lead to big changes in performance and show how useful correlated microscopy is for studying new materials.
The study looks into how time-resolved ion migration (PL) can be used to understand how the features of perovskite change quickly when they are exposed to outside forces. Halide perovskite is a mixed electronic-ionic conductor. To fix stability problems that happen when there is voltage bias, light, and high temperatures, it is important to understand how ion migration, carrier recombination, and performance work together. Mobile ions’ ability to react has been directly linked to how perovskite materials break down for use in PV uses.
As an example, think about how the sample’s qualities, like PL, change over time when it is lit up. The environment, photon energy, and photon dose all have a big impact on the sign of this change. When very small amounts of oxygen are present, they can stop defect-induced electronic trap states from working and make the PL effects brighter.
De Quilettes et al. lit up a small area of a MAPbI3 sample and used TRPL to see how the sample got brighter and the lives of charge carriers got longer. By comparing the same part of the sample that was mapped with TRPL to the local chemical distribution using time-of-flight secondary ion mass spectrometry (ToF-SIMS), they saw that the amount of iodide dropped in the area that was lit up and went up in the area next to it. This clear picture of light-driven ion movement in halide perovskite films shows how important these processes are for the short-term changes in PL.
The work of Choi et al. using correlative nano-SIMS showed how these impacts show up on a grain level. It’s these moving ions that make redox chemistry possible, which, based on the sample conditions, can either passivate or make traps.
Researchers can look into the features of halide perovskite in more depth and with better precision thanks to scanning probe readings. For example, Xiao et al. made MAPbI3 perovskite on lateral device designs and used PL and Kelvin probe force microscopy (KPFM) to see how the material properties changed in real time as a field was applied across the perovskite layer all the time. It was shown by Gratia et al. that nanoscale chemical difference in the form of halide segregation could be linked to nanoscale changes in surface potential seen with KPFM and helium ion microscopy with secondary ion mass spectroscopy (HIM-SIMS).
Birkhold et al. used both in-situ PL and scanning Kelvin probe microscopy over time in a lateral electrode shape to look into how charged particles and moving ions interact with each other. Comparing trap creation with charge injection and blocking contacts revealed that new nonradiative defects (Pb0) were only formed in MAPbI3 when electrons were introduced. This shows how important redox processes are in these short-lived processes.
Garrett et al. used heterodyne KPFM to figure out the open-circuit voltage of MAPbI3 solar cells in both space and time in more scanning probe work. They found nanoscale sites with voltage changes greater than 0.3V when exposed to sunlight. They also showed that these sites changed voltage quickly both when they were lit up and when they were left in the dark for minutes after the light was turned off.
Leblebici et al. used photoconductive atomic force microscopy on MAPbI3 films to show that the short-circuit current and open-circuit voltage could change in the same areas up to 0.6V. They believe this is because the levels of trap states change depending on the facet. In places where grains had certain features, having both high Voc and high Isc could help the local solar performance get close to theoretical limits. But in many other places, the performance was much lower.
More research has shown that how well the AFM tip removes charge carriers can also change the differences in current from c-AFM readings. Voc and Isc are still not all the same, though. This could be because of certain crystalline or facet traits or structure phase defects. The overall performance of halide perovskite devices might be improved by getting rid of local differences in Voc. This could be done by engineering materials to grow in certain ways, such as with favoured grain patterns or preferred orientations. These results showed how important the nanoscale is for performance and gave us a hint that learning more about the nanostructure of perovskites could lead to better device stability and efficiency.
Recent Multimodal Characterization
Multimodal studies look at the connections between photophysics, morphology, and makeup. One important question that comes up from de Quilettes et al.’s work is where the differences in space come from in local non-radiative recombination events. Diffraction methods are the best way to look at structure, especially when making techniques based on X-ray diffraction or electron imaging. Electrons are sped up to almost the speed of light by synchrontron light sources, which shine light from electrons that have been pushed aside by a magnetic field. This very strong energy, usually X-rays, can be controlled and pointed at a sample of interest to find out about its structure, makeup, and photophysics. Synchrotron sources are many times brighter than regular sources of light. They offer better signal-to-noise ratios and information that is spatially resolved on nanometre or smaller length scales. Electrons are sent out through thermionic or field emission in an electron microscope. They are then focused and sent to interact with an object of interest. Signals like cathodoluminescence, diffraction, incoherent scattering, and elemental X-rays are made by this interaction. These can show how the structure, photophysics, and makeup of a sample change over time and space.
Subgrain Features
Recent research in the field of perovskite has shown that morphological grains in polycrystalline perovskite films have complex subgrain structural traits. As Li et al. showed, these hidden features can have big effects on performance. For example, they saw subgrain boundaries in FAPbI3 films that weren’t visible from simple visual or morphological maps of the same area when they used PL and EBSD mapping.
Twinned crystals are groups of two crystals that are tied together by a certain symmetry operation and share some of the same lattice points (the twin border). It is possible for twins to form during growth, during a phase change, or when the structure of something changes. We used density functional theory to look into how twin borders change the electrical features of two types of perovskite: pure FAPbI3 and FA0.875Cs0.125Pb(I0.666Br0.333).3. At room temperature, the formation energy of the {111} twin border in pure FAPbI3 is very low. This means that these flaws should be easy to make in real materials when the temperatures are higher. However, these twins’ electronic qualities aren’t too bad for single-junction solar uses; they only make it slightly harder for electrons and holes to move between the boundaries.
In the mixed-cation, mixed-halide perovskite FA0.875Cs0.125Pb(I0.666Br0.333)3, twin borders cause I− and Cs+ to separate, which creates a local valence band maximum that is 0.2 eV more energetic than the mass. Different valence band energies could cause holes to gather near the twin edges, which would increase the amount of electron–hole recombination events and, in the end, change how well the device works.
Recent testing results mostly agreed with these DFT predictions. Li et al. were able to figure out how intragrain planar defects affect the performance of MA1-−xFAxPbI3-based perovskite PV devices by changing the composition of the twin boundaries and using selected area electron diffraction (SAED) to see the boundaries. They also kept an eye on parameters like Voc, Isc, and charge-carrier lifetimes. They came to the conclusion that there was a direct link between the density of twin boundaries and how well the device worked. Perovskite compositions with lower densities of twin boundaries worked better than those with higher densities. But it might be hard to tell the effect of the number of twin boundaries apart from other factors. This is probably why SAED will need to be correlated with local Voc and Isc maps.
Ferroelasticity in perovskites is a property that can be linked to differences in the structure and make-up at the nanoscale level. This happens because of an elastic hysteresis that happens when external force causes the crystals to mechanically switch between two different positions. Using correlated force imaging and compositional mapping methods, it was found that ferroelasticity and local chemical segregation are naturally linked in MAPbI3 grains. The results from piezoresponse force microscopy show striped areas that are ferroelastic twin regions and show local ion segregation. The sample’s own strain energy is released, which causes the chemicals to separate. The chemicals may also have different crystalline directions.
This difference in chemicals at the subgrain level could have big effects on the electronic properties of halide perovskites, either for the better or for the worse. For instance, the variability could make it easier for charge carriers to separate near the domain walls, which would improve solar features, or it could slow down the device because of the presence of low-conductivity domains.
A new study by Xiao et al. has added to what we know about how ferroelastic twin limits in MAPBI3 affect performance. It was discovered that ferroelastic twin limits don’t normally affect electronics and don’t act as nonradiative recombination centres or get in the way of carrier movement. But the twin borders seen in tetragonal MAPbI3 are not the same as those seen in FA-rich pseudodocubic perovskites. The twin borders in MAPbI3 are not as solid as those in FA-rich perovskites. They disappear quickly when heated or exposed to an electron beam. We need to do more research to fully understand how these small traits affect the performance and stability of halide perovskites, especially in wide-bandgap compositions that can be used in tandem photovoltaic (PV) systems.
Strain and Photophysics
Strain, which is how a material reacts to stress, is a key factor in changing the local photophysics of a semiconductor. To look at how pressure changes the space between atoms in a material, methods like EBSD, microbeam X-ray Diffraction (μXRD), and nanobeam X-ray diffraction (nXRD) can be used. The PL lifetime tracks from areas that don’t emit light at all and areas that do match to areas with higher and lower compression strain in a MAPbI3 film are different.
Lower strained areas last a lot longer than highly strained ones. This shows that local strain is linked to nonradiative recombination events that are tied to traps. More nXRD tests on these samples showed “super-grain”-like structures with horizontal areas as large as 25 μm2. These are long-range, uniform regions that share some kind of common direction and are at least a factor of ten bigger than the average grain sizes found by SEM.
Changes in strain may be linked to crystal orientation or relative misorientation. Jariwala et al. used EBSD, SEM, and PL to look at how crystal direction, subgrain structure, and strain affect nonradiative recombination and efficiency losses. Inverse pole figure maps showed local crystal orientations and limits that could not be seen with SEM alone. If you look at the misorientation between grains, you can see that grain boundaries with more intragrain misorientation and, as a result, more local strain are connected to higher trap densities as found in PL measures.
In the perovskite industry, strain is still not well known. There are many things that can cause it, including a difference in the coefficient of temperature expansion of the perovskite and the substrate it is grown on, a difference in the structure of the perovskite as it is made and the substrate, mechanical bending, and atmospheric pressure. At the local level, strain has also been connected to things like atomic flaws, phase changes, light/bias stimulation, and the edges of grains.
In addition to changing how well something works, strain can also make it less stable by controlling the thermodynamics of defect formation or lowering the energy barrier for ionic migration. This can cause degradation through atomic rearrangement and redox chemistry between migrated ions and other species.
A lot of research has been done on how strain affects the features of perovskite materials, but many of them have problems, like not having a “unstrained” reference state or knowing how Poisson’s effect works when a film is under uniaxial strain. There have been some general claims made about how strain affects perovskites. For example, strain has been linked to the breakdown of perovskites while also keeping the desired photoactive phase of FAPbI3 stable. There are times when changes in makeup may be important, because perovskites with different compositions may react to pressure in different ways.
When studying strain, or any structure trait for that matter, it is important to separate its impact from changes in makeup. For instance, changes in the halide (or cation) content of a perovskite film over time and space can cause the unit cell size to change in mixed halide and cation perovskite mixtures. In the case of halides, the ionic radii of I− and Br− are very different from those of Cs+, FA, and MA, where the cations are concerned. In experiments, any change or shift in the local unit cell size caused by makeup will show up as changes in the interplanar spacing seen in structure measures, which could be mistaken for mechanical strain.
One way to get around this problem and separate the effect of strain from that of composition is to use multimodal studies to get compositional and structure information at the same time on the same part of a sample. Recently, it has become possible to look for connections between structure, chemistry, and photophysics on the same areas of halide perovskite materials with very high spatial precision. This is especially true at synchrotron light sources. Correa-Baena and others studied the location and effects on performance of alkali metals (like Rb and Cs) and halides (like I and Br) in mixed cation and mixed halide “alloyed” perovskite materials using nXRF, SXDM, EBIC, and TAM.
They used photoemission electron microscopy (PEEM) to look at where charge-carrier traps are on the surface of thin films of (Cs0.05FA0.78MA0.17)Pb(I0.83Br0.17)3 and (Cs0.05FA0.78MA0.17)Pb(I)3. They saw separate nanoscale groups of photo-excited hole traps that were strongly linked to differences in makeup and grain borders. These trap groups were mostly found at the edges of grains that were made up of pure perovskite and “inhomogeneous” grains that were low in Br and high in I.
We used scanning transmission electron microscopy energy dispersive X-ray spectroscopy (STEM-EDX) along with PEEM and STEM-EDX data to show that the unevenly shaped grains connected to deep trap clusters were physically different from the pure perovskite. The diffraction images show that most of the trap groups are connected to grains that are low in Br. This study helped show the physically and chemically disorganised environment that directly led to the differences in space-time non-radiative recombination that were first seen by de Quilettes et al.
Kosar et al. explained what these physically different grains are and how they help make trap groups. Researchers found three types of nanoscale structure entities (defects and phase impurities) connected to trap groups. These were PbI2, hexagonal perovskite polytypes, and grain boundary type defects. Time-resolved PEEM (TR-PEEM) showed that PbI2 was not too harmful to electronics, while hexagonal perovskite polytypes helped trap charge carriers. The flaws along the grain boundaries were the worst for performance. They had the biggest drop in TR-PEEM strength after photoexcitation, which shows that they have a very large cross section for hole capture.
These defects at the grain boundaries are too small for the nXRD methods used in this work to resolve. They are most likely very small phase impurities or atomic-scale defects like empty spaces and jumbled interstitials. This will lead to more studies using atomic resolution methods, like high-resolution STEM, to figure out exactly what these grain boundary flaws are.
“Alloyed” perovskite mixtures that are a mix of mixed bromide and mixed cation have achieved high power conversion efficiencies, topping the most recent PV efficiency charts. Recent research that ties together local structure, chemical, and photophysical data has helped explain why there is high performance when there is a high number of deep charge-carrier traps that should limit performance. Feldmann, Macpherson, et al. found that changes in the amount of halides in a material cause changes in the bandgap that act as channels for charges on low-bandgap sites. This creates “hot spots” of high radiative recombination at these sites.
Frohna et al. built on this work by connecting local luminescence measurements on thin films with nanobeam X-ray fluorescence and nXRD measurements to show where the chemicals of (Cs0.05FA0.78MA0.17)Pb(I0.83Br0.17) are distributed in space.We found that areas with high absolute PL quantum efficiency (PLQE) and low Urbach energy were strongly linked to areas that were high in bromide. However, the local PL spectrum in these areas with a lot of emission has a slightly reddened shoulder compared to other areas. This is because the emission is coming from places with more iodide. The results give a complicated picture of these “hot spots”: they seem to have too much bromide when looking through the bulk region, but they may also have small pockets of lower-bandgap material with a little more iodide. This could be a surface layer or tiny inclusions of iodide-rich material where strong radiative recombination happens. It’s amazing that this local substance is a clean semiconductor with a very low Urbach energy.
This kind of chemical disorder controls the optical reaction in these alloyed perovskite systems, even when there are nanoscale strain changes (∼1%) in other places. These small energy differences direct carriers and keep them from recombining with other carriers that are in trap groups that are linked to electronic disorder. This effect could be seen as a type of better tolerance for traps that happen by chance in these mixed-composition materials that have been made better through empirically efficient device developments.
FAPbI3 has a near-ideal bandgap and is more thermally stable, which makes it a good choice for industrial solar photovoltaic (PV) uses. However, it is hard to make and keep stable because the desired photoactive average cubic phase, which is made up of corner-sharing PbI6 octahedra, can only stay stable at temperatures above 150°C. At room temperature, the material quickly changes into face-sharing hexagonal polytypes with a wide bandgap. These polytypes are not good for collecting sun light. Combining FA on the A-site with Cs+, MA, or combinations is thought to be the most effective way to stabilise the system.
Doherty, Nagane, et al. demonstrated that these stable, alloyed, FA-rich perovskites have a small amount of octahedral bending at room temperature, which means they are not normally cubic. In the perovskite world, people thought that these steady, alloyed, FA-rich mixtures had the same cubic crystal structure as FAPbI3. This is a big finding. It was also shown that the octahedral tilting prevented changes from corner-sharing to face-sharing structures, which makes the desired photoactive corner-sharing structure stable at room temperature.
The authors also used AFM-IR readings to show that differences in the A-site cation in FA-rich perovskites created areas that did not have octahedral tilting and are therefore able to change to hexagonal phase flaws. Finally, the writers demonstrated that octahedral tilting can be created in FAPbI3 films without any cationic ingredients by using template-setting chemicals like ethylenediaminetetraacetic acid.
This method gives us more chances to fine-tune the nanoscale structure and phase stability by looking into different tilt-inducing growth techniques. This will lead to better performance with fewer unwanted trap groups in the end.
Atomic Scale Multimodal Studies
Halide perovskite research is a complicated area that needs atomic-scale multimodal studies to figure out the specific flaws that can trap charge carriers and cause degradation. It’s hard to get information on these length scales in organic-inorganic (thin film) halide perovskites because of problems like getting the samples ready, beam damage, and figuring out what the data means. Recent studies, on the other hand, have shown that the community can get very useful information on these length scales.
Rothmann et al. used a scanning transmission electron microscope (TEM) with low-angle annular dark field (LAADF) imaging to get atomic-level pictures of FAPbI3 and MAPbI3 thin films made of polycrystalline material. The LAADF pictures showed that there were smooth, defect-free, low-strain surfaces between the leftover precursor PbI2 and FAPbI3 grains. This explains why a small amount of extra PbI2 might not hurt the performance of the PV. Aligned point defects in the form of gaps on the Pb–I sublattice in FAPbI3 were also seen. This was the first time that defects that had been predicted by theory could be seen directly in an experiment.
Researchers led by Cai and others used a focused ion beam (FIB) to make thin films of FA1–xCsxPbI3 perovskite and measured them using high-angle annular dark field (HAADF) STEM. The films were made up of cross-sectional lamellas. This work was the first to look at mixed-cation thin film perovskites at the atomic level. Cross-sectional samples can tell us a lot about how the perovskite active layer interacts with the contact layers and interfaces in full device stacks. This is a very important area of study because many of the leftover device losses happen at the contacts.
More study needs to be done to find out if the changes seen in FA happen even if the sample hasn’t been introduced to the FIB first. This will help us understand any changes caused by the beam that might happen in these kinds of studies.
Pressing Challenges and Opportunities
To look at qualities like structural, chemical, and optoelectronics on different length scales, you need high-tech tools for characterisation. This needs a directed tool to shine light on a sample, which is hard to do technically and logistically.
Challenges: Beam Damage
There has been a lot of interest in studying halide perovskites because electrons can hurt them. These are beam sensitive semiconductors. Xiao et al. used cathodoluminescence (CL) spectroscopy to look at how electrons damaged MAPbI3, FAPbI3, and CsPbI3 perovskite thin films in a range of testing settings. After only 30 seconds, they saw the excitonic peak widening and the CL signal going down when the probe currents were smaller and the voltages were rising faster. Along with the peak widening, new photon peaks with higher energies showed up in the CL spectrum as the electron energies increased. This was explained by the electron beam creating flaws. The high energy peaks were seen because the electron beam caused damage through heat, which created intermediate phases with bigger bandgaps than MAPbI3.
Kleine-Kedem et al. used EBIC to look into what happened to MAPbI3, MAPbBr3, and CsPbBr3 when they were exposed to electron beams. They saw that fully artificial perovskites, like CsPbBr3, could handle electron doses that were two orders of magnitude higher than those that could handle organic perovskites. Milosavljevic et al. looked into what low-energy electrons could do to change MAPbI3 by using energy levels ranging from 4.5 to 60 eV. They showed that even these electrons with energies as low as 10 eV can change the shape of the MAPbI3 crystal in a big way. As the radiation damage got worse, big changes were seen in the shape of the sample. It became rougher overall as holes and cracks started to show up.
As the field has grown, it has become possible to look at the effect of electron radiation on perovskite structure in a more measured way. Rothmann et al. used selected area electron diffraction to look into how electron and gallium ion beams changed the structure and chemistry of MAPbI3 at different dose rates and amounts. They showed that even at low dose rates and total collected dose, small changes can happen to the structure of a material that is otherwise perfect.
It is proven by the work by Xiao et al. that MAPbI3 forms intermediate stages because electrons hurt it. As electron exposure goes up, the (002) and (110) planes’ lattice parameters go down. This is because organic compounds are lost. Once the supercell is formed, the MAPbI3 film breaks down into a different structure. The I:Pb ratio drops at the same rate no matter what the dose rate is. This shows that the total collected dose, not the dose rate, is what matters with perovskites (at 200 kV).
It was also discovered by the writers that changes in structure can happen at room temperature with amounts as low as 100 e− ×−2. But, Kosasih et al. recently showed that samples made with FIB milling still showed emission that was consistent with perovskite, though with a slightly blueshifted luminescence that shows some of the perovskite structure is still there after FIB milling.
The findings of Rothmann et al. and Chen et al. have been supported by other researchers. Rothmann et al. and Chen et al. also used SAED to look into the structural instability and breakdown paths of MAPbI3. Over time, as the electron dose increased, they saw a number of morphological stages appear, which they linked to a middle MAPbI2.5 supercell. After this unstable phase formed, the material broke down even more into PbI2, with a critical amount of 50 to 100 e− Å−2. Alberti et al. also saw the appearance of new stages in MAPbI3 because of damage caused by the beam.
Low-temperature beam damage studies have also been done to make MAPbI3 and MAPbBr3 more resistant to damage from beams. Li et al. used cold electron microscopy to make them more resistant to damage from beams, but Rothmann et al. and Chen et al. found that MAPbI3’s beam damage problems got worse at low temperatures. VandenBussche et al. looked into whether the way the dose is delivered could change how quickly MAPbI3 is damaged when the total dose rate stays the same. It was found that the rapid electron beam did 17% less damage than the constant thermionic source.
The authors saw that dose thresholds at 300 kV were two to three times higher than those at 80 kV. They also found that MAPbI3 had facet-dependent electron beam sensitivity, with dose thresholds 10 times higher for the (100) plane than for the (001) plane. This sample, Br, has dose limits that are twice as high as the sample I only.
Recent research has focused on formulas that are more useful for technology. For example, Rothmann et al. used atomic resolution scanning-TEM (STEM) LAADF) images on samples of FAPbI3 to show how the material breaks down when electrons are exposed to it over time. Critical dose rates were about the same as for MAPbI3 (∼100 e− Å−2), and breakdown stages were the same (different structure phases like PbI2 appeared after a lot of damage).
An experiment by Ferrer-Orri et al. used SED and nXRD to look at how electrons and X-rays affected the structure of (Cs0.05FA0.78MA0.17)Pb(I0.83Br0.17)3. After accumulating doses of about 200 e− Å−2, they saw PbBr2 (within grains) and PbI2 (at high-angle grain boundaries) form. This was followed by the formation of pinholes and a change in the perovskite structure from a pure average tetragonal phase to an average cubic phase.
In conclusion, studies of beam damage have found a number of themes that are common to halide perovskites. Some of these are that compositions with an MA cation have lower critical doses, compositions with Br are more resistant to damage from beams, the critical dose changes depending on the imaging method, and total dose is more important than dose rate when tracking damage processes. When you use X-rays or electrons to study different types of perovskite, the damage usually follows the same pattern: the pure perovskite structure loses its crystallinity, and other crystalline and sometimes disordered materials or phases appear.
It is possible to get accurate structure readings of halide perovskites if the experimental settings are carefully picked. It can also be helpful to compare damage before and after beam exposure using alternative methods, such as PL. In some cases, like in STEM-EDX studies, doses higher than the critical dose limits may be needed to get a good signal to noise ratio. This means that damage may be unavoidable. But, if the measurements are done correctly, this kind of data might still be useful, especially when looking at how the makeup changes across a picture.
Challenges: Resolution Limits
It can be hard to get to the right length scales and features for multimodal microscope research of perovskite. Halide perovskites have different qualities that affect how well and how stable devices work at large scales, down to the atomic and nanometre levels. PL mapping is a good way to test how well optoelectronics work, but its spatial precision is very low, only about 300 nm at normal excitation/emission wavelengths. This is a bit bigger than the grain sizes of many perovskite materials that are useful in technology. Other luminescence-based methods probably won’t work on halide perovskites because they need stronger lasers.
Any changes made to the perovskite material that can be seen can cause quick changes in the perovskite’s qualities in that area. Another option is electron microscopy (CL), which works like electron microscopy and can get very clear images, but beam damage is a problem. To keep artefacts to a minimum when using CL, testing conditions must be carefully managed.
Tracking carriers locally with high temporal and spatial precision is possible with Tam beyond the diffraction limit, but carrier counts are usually high. Photo-emission studies can look at the local valence band edge and subgap trap states at the surface. If you connect them to the right electron photo-emission microscope (PEEM), you can get very detailed information (<20 nm). Time-resolved PEEM data give extra details about how carriers recombine in halide perovskites in real time and space, but they need to be used with great care to avoid creating surface artefacts. Scanning probe measures, like KPFM or photoconductive AFM, give information about electrical qualities and shape at the nanoscale, but they are mostly subject to the surface. With cross-section AFM measures, you can check out layered device stacks, but you need to be careful not to get errors from the tips, especially when working with soft materials.
Challenges: Image Registration and Sample Fabrication
For imaging data to be accurate, sample drift and picture errors between modalities need to be carefully thought through. It is very important to have accurate picture registration, and you can use fiducial marks if they don’t change the way the samples behave. For comparison tests, some groups have used geometrical gold nanoparticles that were cast on a film surface, while others have burned in fiducials by sacrificing parts of a sample using methods such as the FIB. Careful cuts on the sample surface can be useful in some situations, but they are less useful for films that are very regular because you need to measure away from scratch-related waste and still need a nearby reference point.
Image association is also hard because images can be distorted during measuring, such as by translations, transformations, and rotations that don’t follow a straight line. A lot of tests need to be done in places or with samples that are just right for the measurement method, not in standard device settings. For instance, TEM readings need very thin samples, which limits the thickness of device layers and makes it impossible to use any contact layers, especially solid ones.
It’s not easy to use lighting levels and biases that look like real devices. Because many scanning probe measurements are taken along the side, they can only show structures that are measured along the side. This means that the electric field is stronger in one direction than the other, and different features can be measured both vertically and laterally. Tomographic methods, which can recover 3D data by looking at the width of a film, are usually limited by the risk of beam damage. To get a full picture of the features being studied, you need sets of tools that work together.
Challenges: Facility Access and Data Acquisition
Access to big facilities like synchrotrons is needed for multimodal methods that probe structure and chemical properties with high spatial resolutions. This can be hard because they need to be customised over and over again and only have so much beamtime to give out. In the halide perovskite field, differences can happen between sample batches, between samples in the same batch, and between pixels on the same device base. This is because of how the materials are processed in different places. To draw results from multidimensional data, this means that more than one test is needed. The COVID-19 outbreak made these problems even worse by limiting access and stopping the creation of customised innovations. Characterisation facilities that are very advanced create huge amounts of data—up to 200 TB of data per hour—that can’t be analysed by hand. Researchers who are working on multimodal characterisation of halide perovskite need to know a lot about data science as well as how to do experiments and make materials.
Outlook and Opportunities
To fully understand how the different qualities of halide perovskites affect each other, we need to do multimodal characterisation studies. These studies should look at things like nonradiative recombination and local structure, ion movement and performance, and instability and material changes. To keep an eye on, understand, and solve the big problem of halide perovskites’ long-term operating stability, it is necessary to measure multiple features directly and at the same time while the device is running continuously. To get these results, standard but flexible operando rigs must be put together. These rigs should include controls for light, bias, and atmosphere on multimodal sets in research labs and facilities.
Around the world, updates to beamlines like ESRF, Diamond II, SPRING-8, and Advanced Photon Source will make detectors brighter and more sensitive. This will speed up progress by letting measurements happen more quickly and adding new features while keeping dose low. With the help of beamline scientists, COVID-19 has shown that many advanced tests can be run successfully from afar, making these sites easier to get to.
In the field of perovskite, it is very important to be able to run and characterise multiple full solar cell devices at the same time or very quickly one after the other. This way, you can get accurate information and strong data on degradation metrics. This makes high-throughput tests and automating hardware more difficult, but it will be an important part of the future of self-driving labs, where automated manufacture and characterisation will let researchers study a huge range of device manufacturing parameters. To explore the data space in a reasonable amount of time, we will need data analysis methods that are based on artificial intelligence, like picture recognition with neural networks. These new developments will also help with real-time data processing to guide studies in real time and find the best ways to collect data, such as figuring out the safest amounts for each experiment.
In the perovskite field, halide perovskite materials are sensitive to different chemicals, which have been seen to improve device performance and longevity in lab-scale tests that are doable. Some chemicals, like those found in FA-rich perovskites, cause a small amount of octahedral bending, which keeps the photoactive structure stable and stops the formation of performance- and stability-limiting phases in certain areas. But there are still impurities in the nanoscale phase, like hexagonal polytypes and lead iodide, that will eventually cause failure because they help redox processes happen in solar devices, even if they are only present on the nanoscale.
A big question is how to get rid of these stages for good, or is there a certain length or size where they are no longer a problem? At some point, these impurities will be smaller than the nanoscale precision of most useful laboratory sets. To get rid of them, we will have to move towards atomic scale methods. High-resolution TEM and STEM pictures of bulk perovskite absorption materials show that their structures are not all the same on an atomic level. It would be a good idea and, in the end, important to include device-like studies with atomic detail. For instance, studies that use both atomic resolution images and DFT calculations, along with photophysical data and different types of perovskite, will help everyone fully understand how traits at the atomic level affect the performance and stability of perovskites.
Quantitative nanoscale atomic resolution chemistry mapping in full device structures at low enough amounts to not damage the sample or change the way the results are interpreted in a big way would be a big step forward. This level of chemistry detail at the atomic level will be very important for viewing nanostructured alternatives to find flaws and losses in materials that are better for quantum or light-emission uses. For long-term stable testing (including mechanical effects like delamination) and basic understanding of these materials, it will also be important to find mapping methods that can consistently measure strain and stress on these length scales.
New research suggests that halide perovskites might not be as resistant to chemicals and changes in local stoichiometry as was previously thought, especially when looking at very long-term practical performance measures. As we move towards high-throughput production, it will be very important to understand these local defects and make sure that key qualities are the same across a big area. By adding the multimodal methods we talked about in this chapter to production lines, we can get real-time feedback about stoichiometry, structure, and photonic changes that aren’t wanted. Making sure that these multimodal platforms are tested in a way that is consistent with industry standards will make sure that these methods directly test parameters in situations that are similar to field testing, which will give manufacturers useful information.
The halide perovskite field is almost ready to go industrial, which is an exciting time in their history. But there are still some things that we don’t fully understand, like how power loss and failures happen. A lot of these things have to do with how traits on very small length scales interact in complicated ways that are hard to study.
Leave a Reply