Researchers at ETH Zurich and Empa have developed a new image sensor made of lead halide perovskite, which promises better colour reproduction and fewer image artefacts with less light.
It's been a while since we've had a great leap forward in image sensing technology, with a huge amount of effort being put into working with what we can already capture via computational photography etc. But new research into lead halide perovskite image sensors at ETH Zurich and Empa looks like it could deliver some rather interesting results.
The tl;dr of this is that the vast majority of image sensors are made of silicon that absorbs light over the full visible spectrum but are 'downgraded' to process RGB information by filtering the incoming light. Pixels for blue contain filters that block green and red, and so on. The upshot is that each pixel in a silicon image sensor only receives around a third of the available light.
Maksym Kovalenko and his team have spent a decade researching perovskite-based image sensors and have just published a paper on the subject in Nature.
Lead halide perovskite turns out to be easy to process and its properties vary based on the details of its composition. As ETH Zurich explains, if the perovskite contains slightly more iodine ions, it absorbs red light. For green, the researchers add more bromine, for blue more chlorine — all without any need for filters. These pixel layers remain transparent to the other wavelengths, allowing all the available light to be 'used' and for the pixels to be stacked on top of each other rather than having to be arranged side-by-side.
That means that you end up with an image sensor that theoretically can not only capture three times as much light as sensors of the same size, but provide three times the resolution as well. And the fact that each pixel captures all the light also potentially eliminates some of the artefacts associated with digital photography, such as demosaicing and moiré.
Of course, at the moment this is all very much in the R&D stage. But things are progressing. The Kovalenko team has made functioning prototypes with super-sized crystals in the millimetre category in the past, and the recent breakthrough has seen it build two fully functional thin-film perovskite image sensors.
This is an important proof of concept. It shows that miniaturisation is possible, with the team reaching its target vertical dimension for the sensors already. "The first transistor consisted of a large piece of germanium with a couple of connections," points out Kovalenko. "Today, 60 years later, transistors measure just a few nanometers.”
There is a way to go yet. The two prototypes have pixel sizes between 0.5 and 1 millimetres, whereas pixels in current commercial image sensors fall in the micrometer range and are much smaller. The prototypes are 500–1000 µm — pixel pitches of 0.5–5 µm are the current ballpark.
The team reckons that it should be possible to make even smaller pixels from perovskite than from silicon. But there are other hurdles: perovskite requires different readout electronics; perovskite materials are typically less stable under humidity, heat, and long-term illumination; and the most common form of it contains lead which is, of course, poisonous and not the sort of thing you want to be reintroducing into the human environment when you've spent the past century getting rid of it.
But perovskite crystals are showing potentially game-changing applications in solar panel technology, where they are 25% efficient and improving against the silicon-based standard 20-23%, as well as other areas, so the amount of interest and focus on them is ratcheting up.
Stacked perovskite sensors could mark one of the biggest shifts in image sensor design since the advent of CMOS if they can be scaled down and stabilised. Potentially they could allow for not only smaller sensor sizes but also for higher quality images with less need for post-processing, especially in low light situations.