I received this announcement from Albert Theuwissen:
Dear imaging friend,
I am happy to inform you about the fact that the organizing committee of the International Image Sensor Workshop 2009 is working on a special Symposium focusing on Back-Side Illumation of Solid-State Image Sensors. This symposium is scheduled on the first day of the workshop, being Thursday June 25th, 2009.
We invited 8 world experts in the field of BSI. At this moment 4 of them already confirmed that they will give a talk at the symposium. We are convinced that this new initiative will be a great success because the BSI is a very "hot" topic. For professional applications as well as consumer applications, for CCDs as well as CMOS, engineers are researching the possibility to produce back-side illuminated image sensors.
On the other hand, I would like to take this opportunity to [remind] you about the fact that the deadline for paper submissions is within 6 weeks from now, being Feb. 26th, 2009. If you have in mind to submit an abstract, make sure you do not miss this deadline !
Looking forward to a successful IISW 2009 !
Albert Theuwissen, on behalf of the organizing committee.
This sounds like a great idea to me. BSI is going to be the next great thing and such an open discussion can help the industry to overcome the initial problems.
Tuesday, January 20, 2009
Saturday, January 17, 2009
Weekly Patent Review
One funny quote on pixel operation from Samsung application US2009000868, paragraph [0005]:
"...if light is detected in the photodiodes, electron-hole pairs (EHP) are generated and accumulated. The accumulated electron-hole pairs are transferred to the floating diffusion region via the transfer transistor. As a result, the potential in the floating diffusion region is changed, and the active pixel sensor detects and outputs the change in potential."
The application mostly talks about process ideas. I have seen that process engineers sometimes do not understand the intimate details of pixel operation, so it's somewhat excusable. Still, it's quite rare to see such a glaring mistake in an official document from one of the leading image sensor companies.
Siliconfile application US20090008737 has a title "Image sensor having anti-reflection film and method of manufacturing the same". Contrary to the common meaning, the anti-reflection films (250) prevent the reflections from metal layers, as shown:
The anti-reflection films are said to include "at least one of polysilicon, amorphous silicon, tungsten, and TiN". On the negative side, if these films are conductive, they can significantly increase pixel parasitic capacitances.
"...if light is detected in the photodiodes, electron-hole pairs (EHP) are generated and accumulated. The accumulated electron-hole pairs are transferred to the floating diffusion region via the transfer transistor. As a result, the potential in the floating diffusion region is changed, and the active pixel sensor detects and outputs the change in potential."
The application mostly talks about process ideas. I have seen that process engineers sometimes do not understand the intimate details of pixel operation, so it's somewhat excusable. Still, it's quite rare to see such a glaring mistake in an official document from one of the leading image sensor companies.
Siliconfile application US20090008737 has a title "Image sensor having anti-reflection film and method of manufacturing the same". Contrary to the common meaning, the anti-reflection films (250) prevent the reflections from metal layers, as shown:
The anti-reflection films are said to include "at least one of polysilicon, amorphous silicon, tungsten, and TiN". On the negative side, if these films are conductive, they can significantly increase pixel parasitic capacitances.
Thursday, January 15, 2009
Pixim Posts YouTube Video
info4security shows promotional Pixim video comparing its HDR technology with regular CCD - a lot of side by side pictures.
Update: Here is the direct link to YouTube video.
Update: Here is the direct link to YouTube video.
Omnivision Rushes TSMC Orders
Digitimes reports that TSMC has recently received rush orders from Omnivision.
Wednesday, January 14, 2009
Curved Photodetectors Fresh from the Lab
University of Wisconsin-Madison, The Register and Science Daily report that Associate Professor Zhenqiang (Jack) Ma and colleagues have developed a flexible light-sensitive material. Their technology is featured on the cover of the January 5 issue of Applied Physics Letters.
Ma and his group can create curved photodetectors with specially fabricated nanomembranes—extremely thin, flexible sheets of germanium. Researchers then can apply the nanomembranes to any polymer substrate, such as a thin, flexible piece of plastic. Currently, the group has demonstrated photodetectors curved in one direction, but Ma hopes next to develop hemispherical sensors.
Ma's curved photodetector could eliminate that distortion. "If you can make a curved imaging plane, you just need one lens," says Ma. "That's why this development is extremely important."
Ma and his group can create curved photodetectors with specially fabricated nanomembranes—extremely thin, flexible sheets of germanium. Researchers then can apply the nanomembranes to any polymer substrate, such as a thin, flexible piece of plastic. Currently, the group has demonstrated photodetectors curved in one direction, but Ma hopes next to develop hemispherical sensors.
Ma's curved photodetector could eliminate that distortion. "If you can make a curved imaging plane, you just need one lens," says Ma. "That's why this development is extremely important."
Aptina Earned $3M Last Quarter
Micron has published its 10-Q report telling that Aptina's quarterly income was $3M on sales $180M versus $9M loss on $169M sales a year ago. Gross margin on imaging products improved to 29% from 28% a year ago.
Tuesday, January 13, 2009
Intertech-Pira Image Sensors Europe 2009 Finalizes Program
Image Sensors Europe turns into a big conference this year. Just a list of presenting companies impresses:
Advasense Technologies
Aptina
BBC Research and Innovation
CMOSIS NV
ClairPixel
Dalsa Professional Imaging
Dblur Technologies
Delft University of Technology
DXO Labs
Eastman Kodak
E2V
Harvest Imaging
Micro-imaging Solutions
Motorola
Opera Optics
Oxford University
Pira International
Robert Bosch GMBH
Samsung
ST Microelectronics
Strategy Analytics
Sypro Optics
Tessera
Toshiba
University of Sheffield
VRmagic
Yole Developpement
The final program is much bigger and even more interesting than its preliminary version. It seems to retain its educative and marketing tint and probably becomes the best image sensor conference of this sort.
Advasense Technologies
Aptina
BBC Research and Innovation
CMOSIS NV
ClairPixel
Dalsa Professional Imaging
Dblur Technologies
Delft University of Technology
DXO Labs
Eastman Kodak
E2V
Harvest Imaging
Micro-imaging Solutions
Motorola
Opera Optics
Oxford University
Pira International
Robert Bosch GMBH
Samsung
ST Microelectronics
Strategy Analytics
Sypro Optics
Tessera
Toshiba
University of Sheffield
VRmagic
Yole Developpement
The final program is much bigger and even more interesting than its preliminary version. It seems to retain its educative and marketing tint and probably becomes the best image sensor conference of this sort.
Omnivision Announces New Webcam Sensor
Yahoo: Omnivision introduced OV7740, the sensor for notebook webcams. The VGA sensor is based on 4.2um pixel having 6800mV/lux-sec sensitivity. The OV7740 is capable of operating at 60fps in VGA resolution and 120fps in QVGA.
The OV7740 is immediately available for sampling with volume production slated within the first quarter of 2009. Techno Systems Research projects the notebook webcam market to exceed $6 billion by 2012, as demand for portable PCs continues to grow.
The OV7740 is immediately available for sampling with volume production slated within the first quarter of 2009. Techno Systems Research projects the notebook webcam market to exceed $6 billion by 2012, as demand for portable PCs continues to grow.
Saturday, January 10, 2009
Weekly Patent Review
Omnivision patent application US20090002528 describes the split photodiode HDR sensor that Chipworks showed us a month ago. Indeed, as written in comments, the read curtains of the two photodiodes are aligned. The short-exposed photodiode also acts as anti-blooming drains for excess charge overflowing from the long exposure time photodiodes. The photodiodes are arranged in such a way that long exposure time photodiodes are neighbored along vertical and horizontal directions by short exposure time photodiodes, so that the blooming protection is more efficient. Looks like a very nice idea, but blooming during the short exposure portion might still be a problem.
While we are at HDR issues, TI proposes an improved way to mix two differently exposed images in its application US20090002530. To cope with mismatches and noise, TI employs linear regression and adaptive filter for noise gap reduction between high-gain and low-gain outputs of an HDR sensor.
Application US20090001257 from unknown Korean company (looks like Samsung to me) proposes to align the microlens with the active area, rather than the top metal. Normally in the process each subsequent layer is aligned with the previous one, so as third metal is aligned to second metal, which is aligned to the first one, which is aligned to poly or active. So, the misalignment is accumulated over many layers, increasing optical path variations. Being able to align microlens to active is an obvious idea, but to the best of my knowledge, this is impossible with any meaningful accuracy. Unfortunately, the application does not describe how exactly the stepper should be modified to achieve this alignment. I'm unable to find what company filed this application, but if somebody is able to align microlens straight with active layer, it would be a tangible advantage.
Dongbu application US20090004770 looks like a process engineer report on 400-450C final annealing influence on dark current non-uniformity and metal sheet resistance - a lot of low-temp annealing comparison data with some analysis. The experimental data is based on what looks like Foveon X3 sensor with 3.3um pixels made in 0.18um 1-poly 5-metal process. I'm not sure it's worth a patent, but it's certainly an interesting read for those who learn to deal with pixel optimizations.
Micron's application US20080266435 talks about the ways of dark current reduction by biasing 4T pixel transfer. The claim is that with proper transfer gate engineering both positive and negative bias improve dark current and hot pixel rate. With positive bias depletion region under transfer gate creates a path for thermally generated to the floating diffusion. The negative bias increases the thermal electrons recombination rate. All these claims are supported by the nice graphs for different pixel sizes, such as one below:
The positive gate bias has an advantage of better blooming suppression, while the negative bias does not cause full well degradation.
I doubt that this patent would be granted in its broad form. Many companies for years use negative transfer gate bias exactly for the purpose of dark current reduction. Some use positive bias for different purposes too. But the application has an extension - a circuit which changes the bias depending on light conditions - this portion has chances to become a patent.
While we are at HDR issues, TI proposes an improved way to mix two differently exposed images in its application US20090002530. To cope with mismatches and noise, TI employs linear regression and adaptive filter for noise gap reduction between high-gain and low-gain outputs of an HDR sensor.
Application US20090001257 from unknown Korean company (looks like Samsung to me) proposes to align the microlens with the active area, rather than the top metal. Normally in the process each subsequent layer is aligned with the previous one, so as third metal is aligned to second metal, which is aligned to the first one, which is aligned to poly or active. So, the misalignment is accumulated over many layers, increasing optical path variations. Being able to align microlens to active is an obvious idea, but to the best of my knowledge, this is impossible with any meaningful accuracy. Unfortunately, the application does not describe how exactly the stepper should be modified to achieve this alignment. I'm unable to find what company filed this application, but if somebody is able to align microlens straight with active layer, it would be a tangible advantage.
Dongbu application US20090004770 looks like a process engineer report on 400-450C final annealing influence on dark current non-uniformity and metal sheet resistance - a lot of low-temp annealing comparison data with some analysis. The experimental data is based on what looks like Foveon X3 sensor with 3.3um pixels made in 0.18um 1-poly 5-metal process. I'm not sure it's worth a patent, but it's certainly an interesting read for those who learn to deal with pixel optimizations.
Micron's application US20080266435 talks about the ways of dark current reduction by biasing 4T pixel transfer. The claim is that with proper transfer gate engineering both positive and negative bias improve dark current and hot pixel rate. With positive bias depletion region under transfer gate creates a path for thermally generated to the floating diffusion. The negative bias increases the thermal electrons recombination rate. All these claims are supported by the nice graphs for different pixel sizes, such as one below:
The positive gate bias has an advantage of better blooming suppression, while the negative bias does not cause full well degradation.
I doubt that this patent would be granted in its broad form. Many companies for years use negative transfer gate bias exactly for the purpose of dark current reduction. Some use positive bias for different purposes too. But the application has an extension - a circuit which changes the bias depending on light conditions - this portion has chances to become a patent.
Friday, January 09, 2009
Sony BSI Sensor in Products
It looks like Sony leads the race to get BSI sensors into production. Just announced Sony 2009 camcorder lineup features Back-illuminated Exmor-R sensor. Sonystyle also has a video on the new sensors.
Sony says that its BSI sensor provides approximately 2x increase in sensitivity compared with conventional models. However, InfoSync observed just 50% improvement in brightness and exposure comparing BSI-based HDR-XR520V with conventional HDR-SR12.
It's hard to say what is the pixel size in the new sensor. HDR-XR520V camcorder featuring 1/3.15" Exmor-R sensor has 12.1MP stills resolution, which corresponds to 1.1um pixels, if no tricks are used. Most probably, Sony interpolates the its 45-deg angled ClearVid sensor by a factor of 2, meaning that the physical pixel size is close to 1.6-1.7um or so.
The HDR-XR520V and HDR-XR500V camcorders with BSI sensor will begin shipping in March 2009.
Update: Camcorderinfo.com reports that the actual pixel count of HDR-XR520V sensor is 6.6MP, which means that my pixel size estimations of 1.6-1.7um are correct.
Sony says that its BSI sensor provides approximately 2x increase in sensitivity compared with conventional models. However, InfoSync observed just 50% improvement in brightness and exposure comparing BSI-based HDR-XR520V with conventional HDR-SR12.
It's hard to say what is the pixel size in the new sensor. HDR-XR520V camcorder featuring 1/3.15" Exmor-R sensor has 12.1MP stills resolution, which corresponds to 1.1um pixels, if no tricks are used. Most probably, Sony interpolates the its 45-deg angled ClearVid sensor by a factor of 2, meaning that the physical pixel size is close to 1.6-1.7um or so.
The HDR-XR520V and HDR-XR500V camcorders with BSI sensor will begin shipping in March 2009.
Update: Camcorderinfo.com reports that the actual pixel count of HDR-XR520V sensor is 6.6MP, which means that my pixel size estimations of 1.6-1.7um are correct.
Subscribe to:
Posts (Atom)