image: Single junction type (a) and the proposed double junction type (b) solar cells are compared with the conventional TANDEM type (c) and the proposed face-to-face type (d) multi-junction connections. A cross-sectional view of a new AI Robot Vision chip in 3DIC multichip architecture.(e) shows one-pixel unit of a unique high-performance image sensor unit, with the solar cell capability, which is composed of a pinned-surface buried-channel P+PNPP+ double junction photodiode with an in-pixel source-follower current-amplifier, and also with a depletion MOS type charge transfer gate (CTG) and another depletion MOS type pre-charging gate (PCG) with a switching outlet diffusion drain (ODD) region, for draining out the photo signal electrons.
Credit: Yoshiaki Daimon Hagiwara, President Office, Sojo University, Kumamoto-city, Japan
Floating surface single-junction type photodiodes are mostly used in solar cell applications for simplicity and cost. On the other hand, pinned-surface and double-junction type photodiodes are used now in super high-performance image sensor applications. This paper first reviews the difference between the conventional floating-surface single-junction type photodiode and the pinned-surface double-junction type photodiode. The pinned-surface buried-channel P+PNPP+ double junction type photodiodes are very high-performance image sensors with no image lag and very high light sensitivity compared to conventional ones. The diode can be applied not only to image sensors but also to solar cells. In addition, this paper proposes a new AI robot vision chip in the modern 3DIC CMOS image sensor technology using this double junction type diode. So, the diode will be widely interested in process, device, and application researchers and engineers for image sensors and solar cells. A real-time AI smart robot vision chip is described as an example of application, which is composed of an array of N × N pinned-surface buried-channel P+PNPP+ double junction type photo diodes, N × N analog-data stream mask-and-match comparators, digital processing and SRAM cache buffer memory units, integrated in a 3-D multichip architecture. In the external power-off mode, the image sensor array of N × N pinned-surface buried-channel P+PNPP+ double junction type photo diodes also function as a solar cell unit for the AI self-energy robot vision chip.
Charge couple device (CCD) type charge transfer device (CTD), originally invented in 1970, is composed of a series of MOS large-capacitor gates, and consumes a large power for charging and discharging the large CCD/MOS capacitors in order to perform the complete charge transfer operations. Thanks to the advancements in the long history of modern low-power CMOS process scaling technology, CCD type CTD is now completely replaced by the low-power digital CMOS type CTD with the in-pixel source-follower current-amplifier circuit originally invented by Peter Noble in 1969. Since the size of the source-follower current amplifier circuit was too large to be included in each pixel, the buried-channel charge coupled device (CCD) type charge transfer device (CTD) was used till late 1990s.
Thanks to the modern CMOS scaled advanced process technology, in our high-definition digital TV era, MOS transistors and S/D contacts have now become small enough to be included in each pixel of image sensors. Charge Coupled Device (CCD) type was not the only charge transfer device (CTD) that transfers the information of one single photo electron for a long distance in a silicon chip.
Image sensor in general is composed of three basic parts, (1) a light-sensing photodiode, (2) a signal charge transfer device (CTD) and (3) the image information processing unit. Historically, the first original PNP double junction buried-channel type light-sensing photodiode was invented by Philips in Netherland on June 9, 1975, which showed the empty potential well in the buried channel for a single electron to move along for a long distance in the silicon chip. However, the surface P region was connected only to the high-resistivity substrate, with some undesired RC delay time to the substrate pinned ground voltage level. This device works only for low-frequency operations because of the RC delay time constant to the substrate.
This paper explains the pinned-surface completely-majority-carrier-depleted charge-collecting buried-channel/storage region with the complete charge transfer capability for high-frequency operations, realizing the completely-mechanical-free film-less Global and Electrical Shuter functions.
At low frequency, the surface P-region can remain grounded and the buried N-region can be kept completely depleted of the majority carrier electrons. And with no minority carrier hole present for recombination in the buried N-channel region, this PNP double junction photodiode can transfer one single photo electron for a distance inside of the silicon chip without recombination at low frequency.
For high frequency operations, however, the surface hole accumulation region must be pinned in order to function as a pinned virtual gate for realizing the no-image-lag feature with the high-frequency global and electronic shutter function capability, in order to realize high-performance video camera operations, completely free from any mechanical parts and any film media. The difference between the pinned-surface and the floating-surface photodiodes makes a big difference.
High performance pinned-surface double and triple junction photodiodes invented by Hagiwara at Sony in 1975 were hinted by Sony original double and triple junction type bipolar dynamic transistor technology. They had a unique in-pixel vertical overflow drain (VOD) built-in structure with the global and electronic shutter capability, suitable for consumer video camera applications.
Sony filed in 1975 a series of Japanese patent applications on the in-pixel pinned-surface and buried-channel double and triple junction photodiodes, but never disclosed the details until recently. Sony pinned-photodiodes have the unique global and global shutter capability for capturing quick action pictures, free from any mechanical parts and replacing completely the film media by electronic media.
A high-energy ion implantation technology with a unique lamp anneal method was used to fabricate the pinned-surface P+PNPP+ double junction photodiode. Sony reported the complete charge transfer capability and the high quantum efficiency in the SSDM1977 and the SSDM1978 international conferences, and also in another domestic conference in Tokyo. Then, Sony was invited to talk at the CCD1979 conference in Edinburgh, Scotland UK and also at the ECS1980 conference in St. Louis, USA. In 1979,
Hynecek developed a virtual-phase CCD delay line with the complete charge transfer capability with the pinned-surface buried-channel PNP junction photodiode. The pinned-surface of the PNP junction photodiode functions as a virtual gate for the complete charge transfer action.
In 1982 NEC developed the buried-channel photodiode and used it in an ILT CCD type CTD image sensor with detailed measurement data of the image lag. The NEC photodiode, used in the ILT CCD, reported in IEDM1982 was identical with the Philips 1975 invention of the floating-surface PNP double junction buried photodiode. The surface hole accumulation was connected to the high resistivity substrate.
In IEDM1984, Kodak emphasized the importance of the pinning the surface hole accumulation region to achieve the completely-no-image-lag feature. Kodak reported that there was no image lag and named the device as Pinned Photodiode, which is now widely known as the pinned-surface photodiode.
The SSDM1978 conference paper by Sony and the IEDM1984 conference paper by KODAK both reported the excellent short-wave blue-light sensitivity with a very high quantum-efficiency of 60% to 80% on image sensor chips.
The CCD type charge transfer device (CTD) is now completely replaced by the in-pixel source-follower current-amplifier circuit, originally invented in 1969 by Peter Noble, owing to the advancements of the scaled modern digital CMOS process technology.
However, the pinned-surface buried-channel P+PNPP+ double junction photodiode, invented in 1975 by Sony, has been used in the CCD video cameras in the analog TV era and also now widely in CMOS video cameras and smart phones in our modern digital TV era.
The diode can be applied not only to image sensors but also to solar cells. In addition, this paper proposes a new AI robot vision chip in the modern 3DIC CMOS image sensor technology using this double junction type diode. So, the diode will be widely interested in process, device, and application researchers and engineers for image sensors and solar cells.
A real-time AI smart robot vision chip is described as an example of application, which is composed of an array of N × N pinned-surface buried-channel P+PNPP+ double junction type photo diodes, N × N analog-data stream mask-and-match comparators, digital processing and SRAM cache buffer memory units, integrated in a 3-D multichip architecture.
In the external power-off mode, the image sensor array of N × N pinned-surface buried-channel P+PNPP+ double junction type photo diodes also function as a solar cell unit for the AI self-energy robot vision chip.
Hagiwara YD. Pinned-surface and double-junction photodiode type super high-performance image sensor with built-in solar cell structure. Electron. Signal Process. 2025(1):0003, https://doi.org/10.55092/20250003.
Journal
Electronics and Signal Processing
Method of Research
Experimental study
Subject of Research
Not applicable
Article Title
Pinned-surface and double-junction photodiode type super high-performance image sensor with built-in solar cell structure
Article Publication Date
12-May-2025