Friday, November 10, 2023

EETimes article about imec's new thin film pinned photodiode

Full article:

Imec Taps Pinned Photodiode to Build a Better SWIR Sensor

‘Monolithic hybrid’ prototype integrates PPD into the TFT structure to lower the cost of light detection in the nonvisible range, with improved noise performance. 

Silicon-based image sensors can detect light within a limited range of wavelengths and thus have limitations in applications like automotive and medical imaging. Sensors that can capture light beyond the visible range, such as short-wave infrared (SWIR), can be built using III-V materials, which combine such elements as gallium, indium, aluminum and phosphorous. But while those sensors perform well, their manufacture requires a high degree of precision and control, increasing their cost.

Research into less expensive alternatives has yielded thin-film absorbers such as quantum-dot (QD) and other organic photodiode (OPD) materials that are compatible with the CMOS readout circuits found in electronic devices, an advantage that has boosted their adoption for IR detection. But thin-film absorbers exhibit higher levels of noise when capturing IR light, resulting in lower image quality. They are also known to have lower sensitivity to IR.

The challenge, then, is to design a cost-effective image sensor that uses thin-film absorbers but offers better noise performance. Imec has taken aim at the problem by revisiting a technology that was first used in the 1980s to improve noise in early CMOS image sensors: the pinned photodiode (PPD).
The PPD structure’s ability to completely remove electrical charges before starting a new capture cycle makes it an efficient approach, as the sensor can reset without unwanted background noise (kTC noise) or any lingering influence from the previous image frame. PPDs quickly became the go-to choice for consumer-grade silicon-based image sensors. Their low noise and high power efficiency made them a favorite among camera manufacturers.

Researchers at imec integrated a PPD structure into thin-film–transistor (TFT) image sensors to yield a hybrid prototype. The sensor structure also uses imec’s proprietary indium gallium zinc oxide (IGZO) technology for electron transport.

“You can call such systems ‘monolithic hybrid’ sensors, where the photodiode is not a part of the CMOS circuit [as in CMOS image sensors, in which silicon is used for light absorption], but is formed with another material as the photoactive layer,” Pawel Malinowski, Pixel Innovations program manager at imec, told EE Times Europe. “The spectrum this photodiode captures is something separate … By introducing an additional thin-film transistor in between, it enables separation of the storage and readout nodes, making it possible to fully deplete the photodiode and transfer all charges to the readout, [thereby] preventing the generation of kTC noise and reducing image lag.”

Unlike the conventional thin-film-based pixel architecture, imec’s TFT hybrid PPD structure introduces a separate thin-film transistor (TFT) to the design, which acts as a transfer gate and a photogate—in other words, it functions as a middleman. Here, imec’s IGZO technology serves as an effective electron transport layer, as it has higher electron mobility. Also acting as the gate dielectric, it contributes to the performance of the sensor by controlling the flow of charges and enhancing absorption characteristics.
With the new elements strategically placed within the traditional PDD structure, the prototype 4T image sensor showed a low readout noise of 6.1e-, compared to >100e- for the conventional 3T sensor, demonstrating its superior noise performance, imec stated. Because of IGZO’s large bandgap, the TFT hybrid PPD structure also entails lower dark current than traditional CMOS image sensors. This means the image sensor can capture infrared images with less noise, less distortion or interference and more accuracy and detail, according to imec

Figure 1: Top (a) and cross-sectional (b) view of structure of TF-PPD pixels

By using thin-film absorbers, imec’s prototype image sensor can detect at SWIR wavelengths and beyond, imec said. Image sensors operating in the near-infrared range are already used in automotive applications and consumer apps like iPhone Face ID. Going to longer wavelengths, such as SWIR, enables better transmission through OLED displays, which leads to better “hiding” of the components behind the screen and reduction of the “notch.”

Malinowski said, “In automotive, going to longer wavelengths can enable better visibility in adverse weather conditions, such as visibility through fog, smoke or clouds, [and achieve] increased contrast of some materials that are hard to distinguish against a dark background—for example, high contrast of textiles against poorly illuminated, shaded places.” Using the thin-film image sensor could make intruder detection and monitoring in dark conditions more effective and cost-efficient. It could also aid in medical imaging, which uses SWIR to study veins, blood flow and tissue properties.

Looking ahead, imec plans to diversify the thin-film photodiodes that can be used in the proposed architecture. The current research has tested for two types of photodiodes: a photodiode sensitive to near-infrared and a QD photodiode sensitive to SWIR.

“Current developments were focused on realizing a proof-of-concept device, with many design and process variations to arrive at a generic module,” Malinowski said. “Further steps include testing the PPD structure with different photodiodes—for example, other OPD and QDPD versions. Furthermore, next-generation devices are planned to focus on a more specific use case, with a custom readout suitable for a particular application.

“SWIR imaging with quantum dots is one of the avenues for further developments and is also a topic with high interest from the imaging community,” Malinowski added. “We are open to collaborations with industrial players to explore and mature this exciting sensor technology.”

No comments:

Post a Comment

All comments are moderated to avoid spam and personal attacks.