From Smallest to Quantum: Rounding Up Image Sensor 2021 World “Firsts”May 27, 2021 by Adrian Gibbons
It’s a busy quarter-year for image sensor technology. Sony with Microsoft, OmniVision, and Gigajot have all released new product classes claiming “world firsts.”
The year for CMOS image sensors might just end up being 2021. It started early this year with companies like Sony and OmniVision rolling out several new product classes.
Since then, several image sensor announcements merit an industry roundup to review what is going on with image sensors in 2021.
Sony's IMX661 CMOS, which was released earlier this year. Image used courtesy of Sony
This article will breakdown the following press releases, and discuss the technology and applications:
- OmniVision launches “world’s smallest” 1080p HD image sensor
- Sony & Microsoft team up to launch a smart camera with Azure AI and their IMX500 image sensor
- Gigajot launches the first commercially available Quanta image sensor
OmniVision: World’s Smallest, Again?
Claiming the "worlds smallest" is nothing new in today's industry when everyone strives to be the smallest. This concept applies specifically to OmniVision, which designs some of the industry's smallest image sensors, often claiming to be the smallest on record, and the new OVO2C is no exception.
The OVO2C is a 1/7 image format, 2 megapixels (MP) image sensor designed for high definition video embedded into tablets, notebooks, and IoT devices.
The device incorporates OmniVision's PureCel Plus-S stacked technology, which enables functionality in the smallest die size. According to OmniVision, PureCel technology was designed to improve low-light image quality, higher dynamic range, and lower power consumption.
An example of DTI for reducing pixel cross-talk. Video screenshot [1:21] used courtesy of OmniVision
The technology can have a higher full-well capacity and lower cross-talk with deep trench isolation between pixels, which helps them accomplish these metrics.
Though Qualcomm's newest sensor claims "worlds smallest" Sony is claiming a "worlds first."
Sony & Microsoft: Die-stacked Hardware with AI Analytics
As mentioned earlier, Sony already released new classes of image sensors. This new IMX500 sensor from Sony is a "one-chip" solution for AI processing of high-dynamic-range (HDR) images. It also claims to be the "worlds first intelligent Vision Sensor with edge processing."
As an overview, the IMX500 has an image format of 1/2.3 (or 7.857 mm diagonal) and can achieve 4k resolution at 60 fps or 30 fps with AI processing.
The die-stacked IMX500 from Sony. Image used courtesy of Sony
Using this sensor, Sony is teaming up with Microsoft to embed the Azure AI platform and create a cutting-edge smart camera as a way for businesses to open up new revenue streams.
Terushi Shimizu, Representative Director/President, Sony Semiconductor Solutions Corporation mentions that this team-up of Sony's imaging and sensing technology and Microsoft cloud AI services can deliver a convenient platform within the smart camera market.
Though this may be the world's first intelligent vision sensor, another first is the development of quanta image sensors.
Gigajot QIS Products: GJ01611 & GJ00422
Most things in the "quantum realm" can be difficult to develop, however, more and more research is pushing this type of technology into reality. Though Quanta Image Sensor (QIS) technology as a concept can be attributed to Eric Fossum, recent research of his and two former Ph.D. students from their startup, Gigajot, has led directly to the two new image sensors announced last week.
The GJ01611 has a resolution of 16 MP at 1.1 µm pixel size, the GJ00422 has a resolution size of 4 MP at 2.2 µm pixel size. Each sensor is housed in a 208-pin PGA package, which houses both the pixel array and the embedded electronics.
A high-level block diagram for the Gigajot stacked QIS. Image used courtesy of Gigajot
According to Gigajot, these two image sensors are the world's first implementation of room temperature photon counting CMOS image sensors. The sensors claim to have industry-leading specifications for dark current and read noise parameters, critical to room temperature operation.
Though this sensor sounds appealing, what exactly is the difference between CMOS and QIS?
What is a Quanta Image Sensor?
Traditional CMOS image sensors operate on pixel size and well capacity specifications, where numerous photons are collected inside a pixel well and converted to electrons.
QIS is different. QIS operates by detecting single photons hitting a pixel and has been purportedly limited to experimental lab settings until now.
Abstraction of the capture process of a Quanta image sensor. Image used courtesy of Gnanasambandam et al
So, what is the big-ticket advantage of QIS over traditional CMOS?
QIS is designed to register low-light environments with higher fidelity than other CMOS technologies in smaller pixel architectures.
A comparison of CMOS vs Quanta image sensors against photon count. Image used courtesy of Gnanasambandam et al
Additionally, using advanced backside illumination (BSI) technology, the Gigajot Quanta image sensors have greater sensitivity at smaller pixel sizes with better efficiency than front-side illumination fabrication.
Backside illumination places the circuitry underneath the photodetectors. Image used courtesy of Stemmer Imaging
Though these sensors seem groundbreaking in the world of quantum technology, there is still much progress to be made in this area. There are sure to be many more "firsts" as far as quantum technology is concerned.
Image Sensor Applications in 2021
Image sensors play essential roles in commercial, medical, scientific, and security applications. The advent of commercially available machine learning and artificial intelligence algorithms is a component in developing more powerful, environment-agnostic image sensors.
Using image recognition protocols to analyze events in real-time will become common in the next decade (based on industry trends). Pushing the limits of image sensor technology will allow for pattern recognition with AI, along with more efficient topologies like backside illumination with greater fidelity in all light levels.
Interested in other sensor innovations? Take a look at some recent advancements in the articles down below.