Cheap solar energy is a critical component of a low-carbon energy economy to reduce or reverse the effects of anthropogenic climate change. However, the theoretical ~ 33% efficiency limit for single-junction solar cells featuring one absorber layer also imposes a limit to the specific power of such devices. One proven method to overcome this limit is by using multijunction solar cells featuring two or more absorber layers with different bandgaps that are active in complementary parts of the solar spectrum. Achieving that using cheap materials and low-cost coating processes further potentially reduces the levelized cost of energy for such technologies and enables faster and more widespread adoption in energy systems.

The field of solution-processed mixed-halide perovskite semiconductors has been developing for over a decade to achieve that. Here, materials with a nominal composition ABX₃ (A is a cation such as methylammonium (MA⁺), formamidinium (FA⁺), B is a metal cation such as lead (Pb²⁺) and X is an anion such as iodide (I^-) or bromide (Br^-)) are prepared using methods like spin-coating to fabricate high-efficiency solar cells that are competing with state-of-the-art silicon-based photovoltaics. The halide ratio (iodide-to-bromide) can be tuned to access bandgaps from 1.5 eV to 2.3 eV, and the bandgap range of 1.9 to 2.0 eV is ideal for the top/front sub-cell in multijunction devices with three absorber layers, known as triple-junctions.

However, as the perovskite composition becomes increasingly complex, it starts to affect film properties that make these devices difficult to fabricate and reproduce. In particular, and the primary focus of this study, the film surface wrinkles and forms large micrometer-sized peaks and valleys at high bromide contents necessary to achieve the 1.9 to 2.0 eV bandgap. This is an increasingly common observation and makes coating processes sensitive and device optimization difficult. Furthermore, other thin solution-processed layers, such as interfacial surfaces and charge-transport layers, cannot be conformally processed on such rough morphologies, reducing the overall performance and yield of devices.

We encountered these challenges in 2021 while developing compositions and recipes for wide-bandgap perovskites for triple-junctions and realized that the wrinkling behavior had a strong dependence on the perovskite composition. In particular, as the MA/FA ratio and/or the Br/I ratio increased, the wrinkling increased, and the perovskite film surface became rougher. The morphology became visually interesting, with long networks of ridges developing across the film surface (Fig. 1), but the impact on device performance was significant. We also realized that the film composition was the strongest determinant in this behavior since other processing variables (for example, solvent composition, antisolvent, annealing conditions, substrate) had a negligible effect on morphology.

In the summer of 2022, the Gordon Research Conference (GRC) on Unconventional Semiconductors and Their Applications (Ventura, USA) provided an opportunity to further study these materials. Here, a collaboration was struck up with researchers at the University of Washington (Seattle, USA) who specialized in using hyperspectral photoluminescence microscopy and time-of-flight secondary ion mass spectrometry mapping techniques to study local material composition and optical properties. Based on preliminary cross-sectional microscopy measurements, we had a hunch that the local compositions could be different in the peaks and valleys, and these measurements helped prove it systematically, showing a local decrease of bandgap that resembled the network of ridges (Fig. 1).

Later that year, the MRS Fall Meeting (Boston, USA) provided a second opportunity to involve other specialists in this study. We soon started collaborating with researchers at Georgia Tech (Atlanta, USA), Brookhaven National Lab (Upton, USA), and at Argonne National Lab (Lemont, USA) to study the local composition and structure of these films. Using X-ray fluorescence mapping at the Advanced Photon Source, we found that peak-like regions in wrinkled films corresponded to a local increase in iodide-content, which agreed with the bandgap distribution observed at the University of Washington. We also found that wrinkled films showed a stronger preferential orientation of crystallites which were later shown to have set in early in the crystallization process through in situ structural characterization conducted at the Advanced Light Source (Berkeley, USA). This could be compared against smoother films where the preferential orientation was absent. We therefore concluded that, in rough samples with wrinkled morphology, a bromide-rich perovskite would first precipitate owing to the poor solubility of bromide-based precursors, showing a characteristic preferential orientation. This was followed by the crystallization of the iodide-containing perovskite phase. The MA-rich environment further accelerated this process in some compositions.

Finally, using a combination of photoluminescence and mass spectrometry mapping, and sensitive photocurrent spectroscopy, we associated these compositional and morphological effects to properties such as defect density and ion migration. Taken together, the combination of advanced synchrotron-based structural and compositional characterization methods, combined with photoluminescence and photocurrent spectroscopy, helped us relate structural evolution with the local composition, morphology, optical properties and impact on device behavior. Tuning the crystallization of different halide-containing species is therefore critical to compositional homogeneity and the can help control wrinkling behavior to enable high-efficiency multijunction solar cells.

Beyond its scientific value, we also hope that this work highlights the merits of collaborative research to answer complex experimental questions and the extraordinary way that research gatherings provide spaces to develop personal networks that enable it. The GRC and MRS conferences were among the first large post-COVID conferences that allowed us to travel across the Atlantic. And the burst of fresh post-pandemic energy after the two-year lull certainly played its part in this story.

For more information, please read the full paper published in Nature Communications.