CMOS sensor process

[nbo10]

If I'm talking to someone about cmos sensors and they tell me one sensor is made in the .50um process (circa 2000) and the other sensor is made in the .35um process (circa 2004).

What are they tell me about the cmos sensor? Are there any good books on manufacturing cmos sensors?

thanks

[berkeman]

If I'm talking to someone about cmos sensors and they tell me one sensor is made in the .50um process (circa 2000) and the other sensor is made in the .35um process (circa 2004).

What are they tell me about the cmos sensor? Are there any good books on manufacturing cmos sensors?

thanks

Are these CMOS image sensors? Or some other kind of sensor?

Yeah, those are definitely pretty big geometries by today's standards. Current processes for mixed signal ICs are more like 0.18u, 0.13u and 90nm.

[nbo10]

Are these CMOS image sensors? Or some other kind of sensor?

Yeah, those are definitely pretty big geometries by today's standards. Current processes for mixed signal ICs are more like 0.18u, 0.13u and 90nm.

Yes they are image sensors.

[berkeman]

Ah, that's different. No wonder they are using larger geometries. Here's an interesting article that I found by googling image sensor IC design:

http://chipdesignmag.com/display.php?articleId=2175&issueId=27

That magazine also looks to be a good resource for IC design information.

And a search at Amazon.com on Image Sensor Design gave some very good hits:

http://www.amazon.com/s/ref=nb_ss?url=search-alias%3Daps&field-keywords=image+sensor+design

.

[davidrit]

If I'm talking to someone about cmos sensors and they tell me one sensor is made in the .50um process (circa 2000) and the other sensor is made in the .35um process (circa 2004).

What are they tell me about the cmos sensor? Are there any good books on manufacturing cmos sensors?

thanks

The process feature size doesn't tell you much. In general the per wafer cost for 0.5 um will be lower than 0.35 um. So for the same pixel pitch and number of pixels the 0.5 um chip cost might be lower. Then again if the 0.5 um is 6" wafer and 0.35 um is 8" wafer the latter may be cheaper.

The 0.35 um will probably have higher fill factor for a given size pixel, meaning the light collecting area will be a higher percentage of the pixel area. But that may not matter if the 0.5 um has microlenses.

The 0.35 um may have a longer stay in production but maybe not.

It doesn't hurt to know the process feature size but it really doesn't matter much as a user. You should focus on specs you care about: number of pixels, fill factor, noise floor(directly affects the minimum amount of light you can sense), full well capacity (the maximum amount of light generated signal the pixel can measure, linearity, pixel to pixel response uniformity, etc. Also one chip may have functions outside the pixels that you care about such as pixel output rate, digital or analog output, on-chip processing, power dissipation, cost etc.

The CMOS sensor manufacturing is basic CMOS process plus a photodiode "module" (some extra processing steps). So you can look at books that talk about CMOS processing to get a good idea. The photodiode is similar to other diodes but will have its parameters tweaked (depth, doping level) and possibly covered with an optical filter or lens. The substrate material (or probably the epitaxial material on top) will probably have a tighter specification as well to limit the detector dark current.

[nbo10]

What dimension is the the 0.5 or 0.35 um referring to? Thanks

[berkeman]

What dimension is the the 0.5 or 0.35 um referring to? Thanks

It generally refers to the smallest feature that can be drawn/fabricated using that process. Although there are variations on the number, depending on certain things. Here's an example explanation:

http://en.wikipedia.org/wiki/65_nanometer

.