Layered CMOS structure question I am trying to understand the workings of a CMOS image sensor.  I understand that increasing wavelength results in an increased penetration depth in the silicon often used in CMOS image sensors.  What I am in need of assistance in understanding is how the wavelengths (corresponding to the primary colours, red green and blue) relate to the vertical structure of the CMOS sensor.
After reading several articles and performing some experiments, I am aware that CMOS sensors are indeed sensitive to UV wavelengths, but I am trying to understand how I was able to get a reddish-pink response using an incident UV source.
 A: A follow up from links extending from the links that @krs013 kindly supplied in his answer.
One of the key factors that is mentioned on page 19 of this Hamamatsu* article, is the intensity of incident light on the sensor (higher intensity = greater response). Another article is this older article from UDT sensors* (from 1982 - dated, but a good basis) that suggests that the different UV vs visible light absorbance properties of silicon and silicon dioxide (used in CMOS sensors) as another factor (something I am now going to research).
Now, in the experiments mentioned in my opening post (validated and published recently**), most of the responses at near-UV were reddish-pink when viewing the sun (2nd paper), as I was taking measurements at different incident light intensities, this may account for the seemingly counterintuitive reddish response to near-UVA light.
*Note, I have no affiliation with either Hamamatsu and UDT, butare giving credit for the work they have done in these references.
**I am including this link to one of my published papers only as it demonstrates the increasing camera response to increasing UV radiation.
A: After reading the wikipedia page and this article, I understand your question a little better. A CMOS sensor detects colors by letting light pass through three layers of detector, each with a certain spectral response and absorption. I don't think that each layer is specifically red, blue, and green, rather that their responses overlap and that the processor decodes this output and maps it to a color gamut, much like our eyes do.
However, the design is for visible light, and not for UV radiation. The UV probably isn't detected and filtered in a way that would allow it to be characterized; it would just look like some random other-colored light. So yes, CMOS sensors detect UV light, but probably not in a useful way. 
I think that the discrepancy could come in two ways. Either the photodiodes that constitute the detector part of the sensor respond to UV radiation in an unintended way (I wouldn't know if this is likely), or the filtering of light as it passes through the sensor doesn't block UV as well in the blue and green layers (note the order in the second link), so it ends up all in red. The second's more likely, but they both could be causes. I hope that helps!
