Two recent blogs about radiation and analog integration have me wondering about the practicality of large-scale integration for space applications. The first blog was written by Paolo Scalisi. He discussed the mechanism by which radiation can cause ICs to temporarily or permanently malfunction. This blog was in three parts. The third part is here; you can get at all three parts from that link.
The second blog was written by Steve Taranovich. He discusses the need to squeeze lots of functionality into a small space for applications related to space travel. Obviously, you want to keep the weight of your launch vehicle as low as possible, so small ICs and small PC boards are needed. Power management ICs and analog front ends (AFEs) are a couple areas where Steve suggests we should concentrate our efforts:
First of all, it takes larger rocket engines and more fuel as onboard weight increases on a mission, so you want to keep weight and size down to keep costs down. For an electronics payload, guidance, communications, and control systems that means high levels of integration… AFEs, Power Management Integrated Circuits (PMICs), and microcontrollers with lots of analog integrated or lots of analog with a micro integrated are recommended. You also want efficient power — avoid heavy heatsinks and transformers if possible. This means switching regulators, low resistance MOSFETs…
My concern with regard to putting all this functionality on a small piece of silicon is that you are obliged to use some very fine line processes and make devices that have very low breakdown voltage ratings. This means the problems to which Paolo referred in his blog will be exacerbated:
To understand the effect on an IC let’s examine the effect of the radiation on a power MOSFET switch, which can be considered an elementary component of an IC. Due to ionizing radiations, e− /p+ pairs are created inside the gate oxide (SiO2 ) and the electrical field moves the more mobile charges to the gate terminal. Meanwhile, the holes (shown as h+ in the Figure) remain inside the oxide. Some of the electrons that flow from the source to the drain will recombine with the holes inside the gate oxide, resulting in some holes disappearing from the oxide.
Paolo goes on to explain how this can cause a shift in the gate threshold voltage of the MOSFET that had been irradiated. At least the FET would still function (though perhaps not quite properly). Beyond merely a threshold shift, you could expect the ultra-thin oxide layers would be damaged.
My conclusion here is that we may be stuck with the older processes for fabricating devices intended for high-radiation environments. What do you think? Can we do large-scale analog integration of rad-hard parts?