Wednesday, July 24, 2024

Perovskite sensor with 3x more light throughput

Link: https://www.admin.ch/gov/en/start/documentation/media-releases.msg-id-101189.html


Dübendorf, St. Gallen und Thun, 28.05.2024 - Capturing three times more light: Empa and ETH researchers are developing an image sensor made of perovskite that could deliver true-color photos even in poor lighting conditions. Unlike conventional image sensors, where the pixels for red, green and blue lie next to each other in a grid, perovskite pixels can be stacked thus greatly increasing the amount of light each individual pixel can capture.

Family, friends, vacations, pets: Today, we take photos of everything that comes in front of our lens. Digital photography, whether with a cell phone or camera, is simple and hence widespread. Every year, the latest devices promise an even better image sensor with even more megapixels. The most common type of sensor is based on silicon, which is divided into individual pixels for red, green and blue (RGB) light using special filters. However, this is not the only way to make a digital image sensor – and possibly not even the best.

A consortium comprising Maksym Kovalenko from Empa's Thin Films and Photovoltaics laboratory, Ivan Shorubalko from Empa's Transport at Nanoscale Interfaces laboratory, as well as ETH Zurich researchers Taekwang Jang and Sergii Yakunin, is working on an image sensor made of perovskite capable of capturing considerably more light than its silicon counterpart. In a silicon image sensor, the RGB pixels are arranged next to each other in a grid. Each pixel only captures around one-third of the light that reaches it. The remaining two-thirds are blocked by the color filter.
Pixels made of lead halide perovskites do not need an additional filter: it is already "built into" the material, so to speak. Empa and ETH researchers have succeeded in producing lead halide perovskites in such a way that they only absorb the light of a certain wavelength – and therefore color – but are transparent to the other wavelengths. This means that the pixels for red, green and blue can be stacked on top of each other instead of being arranged next to each other. The resulting pixel can absorb the entire wavelength spectrum of visible light. "A perovskite sensor could therefore capture three times as much light per area as a conventional silicon sensor," explains Empa researcher Shorubalko. Moreover, perovskite converts a larger proportion of the absorbed light into an electrical signal, which makes the image sensor even more efficient.

Kovalenko's team was first able to fabricate individual functioning stacked perovskite pixels in 2017. To make the next step towards real image sensors, the ETH-Empa consortium led by Kovalenko had partnered with the electronics industry. "The challenges to address include finding new materials fabrication and patterning processes, as well as design and implementation of the perovskite-compatible read-out electronic architectures", emphasizes Kovalenko. The researchers are now working on miniaturizing the pixels, which were originally up to five millimeters in size, and assembling them into a functioning image sensor. "In the laboratory, we don't produce the large sensors with several megapixels that are used in cameras," explains Shorubalko, "but with a sensor size of around 100'000 pixels, we can already show that the technology works."

Good performance with less energy
Another advantage of perovskite-based image sensors is their manufacture. Unlike other semiconductors, perovskites are less sensitive to material defects and can therefore be fabricated relatively easily, for example by depositing them from a solution onto the carrier material. Conventional image sensors, on the other hand, require high-purity monocrystalline silicon, which is produced in a slow process at almost 1500 degrees Celsius.

The advantages of perovskite-based image sensors are apparent. It is therefore not surprising that the research project also includes a partnership with industry. The challenge lies in the stability of perovskite, which is more sensitive to environmental influences than silicon. "Standard processes would destroy the material," says Shorubalko. "So we are developing new processes in which the perovskite remains stable. And our partner groups at ETH Zurich are working on ensuring the stability of the image sensor during operation."

If the project, which will run until the end of 2025, is successful, the technology will be ready for transfer to industry. Shorubalko is confident that the promise of a better image sensor will attract cell phone manufacturers. "Many people today choose their smartphone based on the camera quality because they no longer have a stand-alone camera," says the researcher. A sensor delivering excellent images in much poorer lighting conditions could be a major advantage.

9 comments:

  1. It would be interesting to know what kind of performance is to be expected over different metrics and what kind of potential pitfalls exist. Anyone have any ideas?

    ReplyDelete
  2. Dark current? Effect of Humidity and Temperature? I wonder...

    ReplyDelete
  3. Lead is a no go for many integrators due to ROHS

    ReplyDelete
  4. FOVEON in perovskites. Good initiative. Let's see where it ends.

    ReplyDelete
  5. Sounds interesting for specialized applications

    ReplyDelete
  6. "The advantages of perovskite-based image sensors are apparent." - Well, not to me. Indeed there have been several Foveon-like proposals in the past. Also Sony stacked organic absorbers on top of CMOS to achieve a similar feat. Is this supposed to compete with CMOS? Because that's what the comparison to Bayer seems to suggest. If so, we'd need a lot more than the concept of stacking PDs which already exists in CMOS as well. (Amongst others) the reasons this feat didn't really succeed in CMOS are scalability (which this proposal doesn't seem to solve) and noise (hard to make all color channels 'pinned photodiodes' with low dark current, zero-noise/lag charge transfer, high conversion gain and FWC...). Just proposing "hey, we can also stack absorbers in Pervoskites" without addressing these issues feel naïve at best. Something is seriously missing here. This article is poorly worded advertisement at best - and if that's all there is, it's massively misleading or possibly deceiving. Please don't ignore history and try to learn from Foveon, Sony etc., identify why these approaches really failed and reevaluate how your new solution overcomes these issues. And then come back and write a proposal that reflects that.

    And regarding the light loss... Well, the CIS/CMOS world is now working on color routers. Let's see who beats whom to the punch. In the last 60 years CMOS beat every contender and I don't expect this to change (in visible).

    ReplyDelete
    Replies
    1. Indeed, this is basically a call for research money. Which is fine, research needs to be done, even one that fails. But as you said a lot seems to be missing here.

      Delete
  7. To me, there is few chance to see perovskite sensors for visible range. Foveon-like structures will lose interest because of color routers as said above. And in the visible range, perovskites cannot beat silicon because of reliability issue + process repetability and scalability for small pixels + high dark current due to defects in these materials (QE is not the sole metric, NEP is also important) + issues with cross talk. And no major companies will invest in long term development to change their whole sensors architectures for minor gains in the visible range. Differentiation from visible full-silicon CIS in term of value proposition is key for heterogeneous integration (UV, IR...) but perovskites cannot go too far in infrared so even for these applications it will be hard to compete.

    ReplyDelete
  8. Sensors employing a Foveon-like structure should take advantage of that, and sense more than three bands of light; rather than leaving anti-aliasing and possibly their 'fuzzy' seperation of those bands as their prime advantage over Bayer.

    The optical properties of silicon (as would perovskites, and other techniques), allow a much finer discrimination of color than a 4x4 Bayer filter would, yet no one takes advantage of that: https://www.pveducation.org/pvcdrom/materials/optical-properties-of-silicon versus https://en.wikipedia.org/wiki/Foveon_X3_sensor#Operation

    Instead using plasmonic arrays would double sensitivity over 3DCRs:
    https://pubs.acs.org/doi/10.1021/nl401641v allowing even polarimetric imaging, like MetaPolarIm.

    A plasmonic sensor would not only be able to sense light at various wavelengths but could also serve as a Surface-enhanced Raman spectroscope (detect chemical molecules, IE: it could "smell" the subject; of course analyzing and reproducing the smell would be a greater challenge over simply displaying the image.

    https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10046622/#:~:text=Plasmonics%20has%20made%20significant%20advances,near%20plasmonic%20surfaces%20%5B4%5D.

    With so many ways forward for image sensor development I feel like many of these "advancements" are more microimprovements that fail to fully exploit the advantages possible; not to mention the glacial progress compared to other technologies. [Kicks soapbox away]

    ReplyDelete

All comments are moderated to avoid spam and personal attacks.