Yole Développement, the market research firm, recently teamed up with its partner System Plus Consulting to disassemble the TrueDepth module inside the Apple iPhone X. "EE Times" had the pleasure of interviewing Yole and they deduced that in this module's near-infrared (NIR) An SOI wafer is used in the image sensor, and the SOI plays a key role in sensitivity for STMicroelectronics (ST) developed NIR sensors to meet Apple's stringent requirements .
Pierre Cambou, Yole Développement Imaging and Sensor Business Analyst, said that SOI-based NIR image sensors are a "very interesting milestone in SOI's evolution."
Many companies in the so-called 'Imaging Valley' near Grenoble, France, use SOI wafers from Soitec originally used for back-illuminated (BSI) image sensors and, according to Cambou, SOI research for NIR sensors dates back to 2005.
However, Cambou notes that Apple's ST NIR image sensor symbolizes SOI's inception for mass production of image sensors. "Because of the physical size of the light, image sensors feature a wide surface, so substrate suppliers such as Soitec This is a pretty good market. '
In the meantime, Jean-Christophe Eloy, president and chief executive of Yole, told "EE Times" that Apple adopted the best of both worlds when it came to designing TrueDepth - combining the benefits of both ST and ams 'products.' Apple adopted ST's advanced NIR Image sensors, and ams of dot-matrix sensors. "Eloy noted that ams' is very good at complex optical modules." Earlier this year, ams acquired Heptagon, which is known for stacking time-of-flight (ToF) technology.
Decrypt TrueDepth working principle
Apple integrated the TrueSense camera system called TrueDepth on the front of the iPhone X to recognize the user's face and unlock the phone.As Yole explained earlier, in order to achieve this, Apple combined a ToF distance sensor with an infrared structure Light 'camera, so you can use uniform' flood 'or' dot pattern 'illumination.
The operating principle of the 3D-sensing camera system is very different from that of a typical CMOS image sensor that takes a photo.First, the iPhone X combines an infrared camera with a pan-sensing element to project a uniform infrared light in front of the phone. Trigger face recognition algorithm.
However, this face recognition feature does not work continuously, and an infrared camera connected to the ToF range sensor signals that the camera take a photo when it detects a face, iPhone X then activates its dot matrix projector and then Generic images and bitmap images are sent to an application processing unit (APU) for neural network training to identify mobile phone users and unlock the phone.
Cambou points out that 3D imaging has not started at this time, and 3D information is contained in a bitmap image.'Another APU can use another algorithm for calculating depth maps to perform 3D applications, 'he added: The iPhone X leverages the power of the A11 chip due to its computationally intensive structured light approach, and using neural networks is the key to this design. "
Five sub-modules
According to a disassembly analysis by Yole and System Plus Consulting, a 'complex combination of five submodules' was found in Apple's TrueDepth Optical Hub: NIR cameras, ToF range sensors + IR flood sensors, RGB cameras, Dot matrix projectors and color / ambient light sensors.
As shown below, IR cameras, RGB cameras and dot matrix projectors are all aligned.
NIR image sensor
ST's NIR sensor can be found at the TrueDepth core of the Apple iPhone X. Yole and System Plus Consulting found within ST's NIR sensor that SOI was used at the top of deep trench isolation (DTI) .
The concept of DTI technology is well known.Generally, the problem with today's cameras that require high sense resolution is that the pixels are confined in the same space, causing noise, color change, or pixelation between adjacent sensors as the picture is taken DTI technology was used to avoid leakage between the photodiodes, and Apple is said to etch the actual trench between them and then fill the trench with an insulating material to block the current.
So why does Apple use NIR image sensors based on SOI wafers at the top of the DTI?
From an optical point of view, Cambou explains that SOI wafers are helpful because the insulating layer functions like a mirror, pointing out: 'Infrared light can penetrate deeper and reflect back into the active layer.'
Cambou points out that from the electrical point of view, SOI dramatically increases NIR sensitivity because it effectively reduces pixel-level leakage and improved sensitivity provides a good image contrast.
Cambou explains that this contrast is extremely important because "the operation of structured light is susceptible to sunshine."
Of course, the general goal of a CMOS or NIR sensor would be to see more light if it was to have a better image. "However, Cambou also pointed out that when a user tries to unlock the iPhone X in bright sunlight, Light will be a problem.
Cambou said: 'The problem is the contrast between the projection of the NIR light and the ambient light of the sun or any other light source, but the sun is usually the biggest problem.' So, Apple's use of SOI wafers to increase the NIR contrast is up to Important.
So ST NIR sensor whether to use FD-SOI or SOI wafer? Cambou said the company is still unable to judge.
As for the NIR sensor, is it now certain that Apple uses a NIR at 850nm or 940nm? Cambou states, 'We can not determine which one.' However, he speculates, 'Apple is most likely to use 850nm just like any other vendor - for example Intel's RealSense, Facebook, HTC, etc. However, ST is known for developing the 940nm single-photon avalanche diode (SPAD) proximity finder and may also intend to turn to this wavelength option in the future.
When asked about any surprises in dismantling, Cambou proposed the size of the ST NIR image sensor chip - about 25mm2 in size - but only 1.4 million pixels because of its large 2.8μm pixel, Cambou said: In spite of this, in this category, this pixel is considered "smaller" than competing products that typically use 3.0μm to 5μm.
The beginning of a new era
Yole positioned the iPhone X as the beginning of a new era in 3D imaging.
Cambou also believes that Apple is building the future for NIR sensors, noting that Apple recently acquired InVisage Technologies, and that "I think Apple intends to have InVisage provide NIR sensor capabilities, but there are probably several Way to explain the acquisition. '
Cambou believes that InVisage may not be able to match ST's performance in terms of performance but is able to provide solutions for miniaturization, he said: 'As a result, Face ID technology can be scaled down to other products such as augmented reality (AR) Type device
Business impact
On the one hand, the Apple iPhone X is creating tremendous business opportunities for SOI wafers such as Soitec, and just as importantly, it marks a meaningful comeback for ST as Cambou believes ST will be a part of the emerging ToF camera market.
Of course, the semiconductor business is often affected by the cyclical effects of brief periods of boom and bust, but according to Cambou, ST has "made a very subtle transition" despite a shrinking business after it lost Nokia in the handset market.
ST has created many different types of image sensor applications: moving from CMOS image sensors to future NIR and SPAD sensors while leveraging the company's assets and underlying technology developed in-house.