A CMOS Indirect Time-of-Flight Sensor with Pull-and-Split Charge Transfer Pixel Structure for High Depth Resolution
Cited 0 times inCited 0 times in
- A CMOS Indirect Time-of-Flight Sensor with Pull-and-Split Charge Transfer Pixel Structure for High Depth Resolution
- Lee, Seunghyun
- Kim, Seong-Jin
- CMOS Image Sensor, Time-of-Flight
- Issue Date
- Graduate School of UNIST
- In recent years, as the demand of various vision applications including robot vision and VR/AR systems increases, many Time-of-Flight (TOF) sensors are developed and commercialized which can measure depth not only with cm to mm-scale depth resolution in a few meters range, but also maximized target range over 100m. As the history of color image sensor has shown the direction, TOF sensors’ development would be continued towards pixel scaling to have spatial resolution while keeping the depth resolution and the cost.
Then, what kind of applications can be possible through 3D imaging? First, an automotive application can be an example. Imagine the situation of driving a car in a dark environment, and we almost can’t see the people around. With 3D imaging technology, the depth perception is possible in the dark, as well as the detection of an object. The next one is the user interface. There are many SF movies with scenes that the actor controls the device or computer with his or her gestures, not using a conventional keyboard, or mouse. This can be possible with the person’s gesture recognition through 3D imaging. Last one is the robot vision. With the sensors on top of the robot, the depth recognition can help the machines to interact with environments around it. For example, the robot can perceive the obstacles around it, and they reach the destination without colliding other objects, such as bookshelf, tables, or people moving around.
To realize these applications, there are many 3D imaging techniques. Triangulation, time of flight, and interferometry can be an example. Triangulation is roughly composed of two categories; one is structured light and the other is stereo vision. Structured light shoots the patterned light, and the reflected light is received onto two or more sensors, and the pattern’s difference among them gives the depth output. The stereo vision works as same as human eyes. This can also detect depth through the difference among sensors’ output through the calculation of trigonometric functions, except this one does not use the active light. There are two big disadvantages of the triangulation technique; one is the number of sensors, which needs at least 2 sensors, and the other one is the computational power, which is higher than other methods.
The next one is the time-of-flight technique. The time-of-flight basically measures the round-trip time of the illuminated light until it comes into the sensor. As the name states, the DTOF measures the time directly, and ITOF measures the phase delay. DTOF uses a single photon avalanche diode (SPAD) for its photonic device, and ITOF can use a photogate , pinned photodiode, and current-assisted photonic demodulator (CAPD). The TOF technique’s strength is that it needs only 1 sensor for depth detection. Due to the strength, this research is focused on ITOF method with pinning photodiode structure.
The interferometry detects distance using optical coherence of the illuminated light and reflected light, thus micrometer to nanometer scaled depth detection can be done, but the maximum distance is too short, and the system is too bulky, so it is used for scientific use.
Various factors affect Indirect Time-of-Flight (I-TOF) sensor’s depth resolution performance. From the pixel’s perspective, Quantum Efficiency (QE), and modulation frequency are the most critical factors among them. Since most of the TOF system uses near-infrared (NIR) region as its light source for end users’ eye convenience, silicon’s low QE in this wavelength can cause a low photo-generated signal comparing to emitted light power , resulting low depth resolution. Thick epi-layer can solve this problem, however, there is a trade-off between the epi-layer thickness and the maximum modulation frequency .
It is known that the depth resolution becomes favorable as the modulation frequency increases since the I-TOF sensor detects the depth result from the phase difference between the emitted light and the received light, and speed of light is always fixed. However, epi-thickness limits maximum modulation frequency since the electrons must be transferred through longer distance to reach the designated FD within the limited time interval to detect the phase difference caused by the target object’s actual distance from the sensor.
In this research, the proposed pixel has grab-and-split charge transfer using different n-doping profile inside pinned photodiode (PPD) which generates lateral electric field to push the photo-generated electrons towards designated FD. The unit PPD is composed of two regions, one is lightly doped with Arsenic which has low pinning potential, and the other is highly doped near the FD to have higher pinning potential, therefore, lateral electric field can be generated. The lightly doped area is designed to have wide area to get a high fill factor, and the highly doped area is designed to have minimal distance between two TX gates. The two TX gates’ distance is carefully designed to have proper lateral e-field through TCAD simulation. Through this structure, the electrons are transferred towards the highly doped area via pinning potential gradient, and the TX gates transfer electrons inside highly doped area. Unit pixel pitch is 14.4um with the fill factor of about 48% without a microlens. It is composed of 8 PPDs to reduce the electrons’ lateral travel distance. This chip is fabricated with a 0.11um CIS process with the minimal change from conventional CMOS Image Sensor technology. In addition, the implementation and measurement of I-ToF sensor are reported with different parameters such as epi-layer thickness, different PPD structures, and PPD’s n-dose for performance comparison.
- Department of Electrical Engineering
- Go to Link;
- Appears in Collections:
- Files in This Item:
A CMOS Indirect Time-of-Flight Sensor with Pull-and-Split Charge Transfer Pixel Structure for High Depth Resolution.pdf
can give you direct access to the published full text of this article. (UNISTARs only)
Show full item record
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.