The image sensors found in smartphones and digital cameras distinguish colors in a similar way to the human eye. In our retinas, individual cone cells recognize red, green and blue (RGB). In image sensors, individual pixels absorb the corresponding wavelengths and convert them into electrical signals. The vast majority of image sensors are made of silicon. This semiconductor material normally absorbs light over the entire visible spectrum. In order to manufacture it into RGB image sensors, the incoming light must be filtered. Pixels for red contain filters that block (and waste) green and blue, and so on. Each pixel in a silicon image sensor therefore only receives around a third of the available light. Thin-film technology: One of the two perovskite-based sensor prototypes that the researchers have used to demonstrate that the technology can be successfully miniaturized. Courtesy of ETH Zurich. Researchers at ETH Zurich and Empa have proposed a solution to this problem, which allows them to utilize every photon of light for color recognition. The basis for their image sensor is lead halide perovskite, a semiconducting crystalline material. In contrast to silicon, however, it is particularly easy to process, and its physical properties vary with its exact chemical composition. If the perovskite contains slightly more iodine ions, it absorbs red light. For green, the researchers add more bromine, for blue, more chlorine, without any need for filters. The perovskite pixel layers remain transparent for the other wavelengths, allowing them to pass through. This means that the pixels for red, green, and blue can be stacked on top of each other in the image sensor, unlike with silicon image sensors, where the pixels are arranged side-by-side. Thanks to this arrangement, perovskite-based image sensors can, in theory, capture three times as much light as conventional image sensors of the same surface area while also providing three times higher spatial resolution. The research team was able to demonstrate this a few years ago, initially with individual oversized pixels made of millimeter-sized single crystals. Now, they have built two fully functional thin-film perovskite image sensors. “We are developing the technology further from a rough proof of principle to a dimension where it could be used,” said researcher Maksym Kovalenko. A normal course of development for electronic components, “The first transistor consisted of a large piece of germanium with a couple of connections. Today, 60 years later, transistors measure just a few nanometers.” Perovskite image sensors are still in the early stages of development. However, with the two prototypes, the researchers were able to demonstrate that the technology can be miniaturized. Manufactured using thin-film processes common in industry, the sensors have reached their target size in the vertical dimension at least. Through experiments, the researchers put the two prototypes, which differ in their readout technology, through their paces. Their results were that the sensors are more sensitive to light, more precise in color reproduction, and can offer a significantly higher resolution than conventional silicon technology. The fact that each pixel captures all the light also eliminates some of the artifacts of digital photography, such as the moiré effect, for example. Consumer digital cameras are not the only area of application for perovskite image sensors. Due to the material's properties, they can also be used in machine vision. The focus on red, green, and blue is dictated by the human eye; these image sensors work in RGB format because our eyes see in RGB mode. However, when solving specific tasks, it is advisable to specify another optimal wavelength range that the computer image sensor should read. Often there are more than three- so-called hyperspectral imaging. Perovskite sensors have a decisive advantage in hyperspectral imaging: Researchers can precisely control the wavelength range they absorb by each layer. “With perovskite, we can define a larger number of color channels that are clearly separated from each other,” said co-author Sergii Yakunin. Silicon, with its broad absorption spectrum, requires numerous filters and complex computer algorithms. Hyperspectral image sensors based on perovskite could be used in medical analysis or automated monitoring of agriculture and the environment, for example. In the next step, the researchers want to further reduce the size and increase the number of pixels in their perovskite image sensors. Their two prototypes have pixel sizes between 0.5 and 1 millimeters. “It should be possible to make even smaller pixels from perovskite than from silicon,” said Yakunin. The electronic connections and processing techniques need to be adapted for the new technology. “Today's readout electronics are optimized for silicon. But perovskite is a different semiconductor, with different material properties,” said Kovalenko. However, the researchers are convinced that these challenges can be overcome. The research was published in Nature (www.doi.org/10.1038/s41586-025-09062-3).