Wednesday, July 06, 2005
Zero-threshold MOSFETs are not a new concept. Previously, however, the threshold was determined by dopants (trace impurities) in the transistor channel, and there wasn't enough accuracy to guarantee a threshold close to zero. ALD appears to be getting around that by using floating gates to compensate for channel variability. Rather than perfectly dope the channel, the factory simply has to measure the imperfect channel and program the floating gate to compensate. A side benefit is that all the transistors are well-calibrated to the same settings; engineers normally have to design with sloppy (as in +/-50% on a good day) transistors in mind.
Downsides: The transistors can only handle low voltages. High voltages will cause electric charge to quantum-mechanically tunnel onto the floating gate, which would wreck the delicate calibration. (I bet they handle static electricity poorly.) Ionizing radiation is also a concern, as it can toss charge onto the floating gate, erasing the calibration. (What about airport and postal x-ray scanners?) Ambient radiation will certainly cause slow drift over time.
As their data sheet doesn't talk about the problems, I'd be leery of using these parts in an aircraft, military, or long-life industrial application, and certainly not in spacecraft. They talk about the low threshold being useful for medical apps, but say nothing about what would happen during a long session with a fluoroscope (the kind where the patient comes out with x-ray burns). Until I see some more numbers, I'll have to limit these parts to less-critical applications.