Contributed by: St.J. Dixon-Warren & Ray Fontaine
Chipworks recently opened up the new Nokia N95 camera cell phone – or, as Nokia puts it, "an all-in-one multimedia computer with a unique two-way slide design, integrated GPS functionality, a five-megapixel camera and support for high-speed mobile networks." With a spec. like that, one wonders if this is a camera with a phone, or a phone with a camera – and, as you can see, from the camera side it actually does look more like a camera.
Nokia N95 camera cell phone
However, when you flip it over, it becomes a two-way slide phone with a 2.6” screen and the usual keypad plus dedicated music playback keys at the other end.
This thing is so packed with features that we don’t have space here to go into them, nor should we, since we’re not a phone review blog – there are plenty of them, try PhoneScoop or Yourmobilephonereviews, for example.
Inside the Nokia N95 camera cell phone
Inside the camera module we found the Micron MT9P001 5 megapixel CMOS image sensor (CIS), shown in the following selected teardown photographs. To our knowledge, this is the first design win for Micron with Nokia, certainly for their high-end phones. Our teardown of the N90 showed that Nokia used a Toshiba ET8EA3 2-Mpixel sensor in that phone. And of course we’ve looked at both sensors, but by coincidence, we obtained the Micron chip and analysed it several months ago.
MT9P001 Camera Module from the Nokia N95
MT9P001 Die Photograph
MT9P001 Die Markings
The part is fabricated with a 0.13-µm three metal aluminum process, and features a 2.5-transistor (2.5T) pixel. This sounds a bit counter-intuitive, since how can a pixel have half a transistor? It’s actually image sensor jargon for a layout where two adjacent pixels share the row select, source follower, and reset transistors (see below).
Schematic of Shared Pixel Layout
Micron has actually taken this sharing further with a 1.75T pixel in their MT9T012 sensor, in which four pixels share the same three transistors. The pixel layout of the ‘001 at the polysilicon level is shown below, and we can see how the outputs of the T1 and T2 transistors are strapped to T3 and T4. The irregular-shaped regions in each pixel are silicon nitride anti-reflective (AR) layers. These AR layers are shaped fairly tightly to the photocathode, as shown by the scanning capacitance (SCM) image of the doped regions in the pixel array.
Pixel Array at Polysilicon
SCM Image of Pixel Array
In the cross-section below we see only two metal layers, since the top metal is not used in the array at all – unusually, the pixel aperture is defined by metal 1 and metal 2, instead of the top metal. This section is actually through the column-out and VDD contacts in the array, and the light areas on the substrate in between the contacts are the nitride AR layer.
.jpg "SEM Cross-Section of the MT9P001 CIS")
SEM Cross-Section of the MT9P001 CIS
Unfortunately, that's about as much as I have space for in the blog this week. Of course, our published reports have much more detailed cross-sectional and plan-view analyses of the sensors. We have a comparative analysis of the MT9P001 and the MT9T012, and a specific report on the MT9T012, not to mention the Toshiba ET8EA3 from the N90. And plenty of others too numerous to list!