Thermoisotopic generators generally rely on plutonium, which is in very low supply, and generate only about 100-200We output AFAIU.
That's enough for electronics and comms on a deep-space mission, but doesn't offer a whole lot of surplus power for planetary activity. Among the reasons that lander craft are so slow-moving (they basically manage a slow crawl) is power demands, though other factors, including navigation and stress, are considerations.
https://en.m.wikipedia.org/wiki/Radioisotope_thermoelectric_...
> plutonium, which is in very low supply.
That's just because we haven't been making more plutonium recently, because we haven't needed it. We could simply make more.
NASA has also been buying Pu-238 from the Russians at around $1.5 million/kg[3], which would be $7.5 million for Curiosity, or 0.3% of the cost of the program.
I'm interested in deep-space exploration (but am no expert), and as far as I can tell the reason the US is running out of Pu-238 has absolutely nothing to do with cost.
It's been happening because only NASA has really wanted this, but producing it has been the purview of the military or the DOE.
So the issue has been stuck in some bureaucratic nightmare for decades, the only people allowed to produce Pu-238 didn't need it, and NASA couldn't simply spend a crapload of money to pay another agency or branch of the government to restart production due to the way money politics works, even though it was an overall good investment.
So rather than simply restart production at a trivial cost compared to what NASA otherwise spends, it's been easier to buy it from Russia.
As [1] shows there's now a real possibility that Pu-238 will run out, so the US has restarted production.
1. https://www.nasa.gov/pdf/636900main_Howe_Presentation.pdf
2. https://en.wikipedia.org/wiki/Mars_Science_Laboratory
3. http://dailycaller.com/2017/03/28/nasa-wants-to-stop-buying-...
I'm sure the demand for a colony would scale pretty fast to 100kW or more. Now we are talking about Billions.
So I think that your original point that cost for Pu-238 is not an issue for a colony is wrong.
Small reactors are a much better option as a single ton of fuel could supply a colony with a few MW for decades. You just have to make sure that the largest part is less than 37 tons so it can be transported with a Long March 9.
I said they were a "bad idea for general power generation", so I'm not suggesting that they be used to power a Mars colony, but rather that the reasons not to do so don't include Pu-238 being in low supply.
We also have a low supply of rockets & other infrastructure to colonize Mars, but we can simply decide to make them.
Still, I think any extrapolation of current numbers to say that Pu-238 would be categorically unsuitable for such a purpose is probably premature. Nobody's tried to produce it on a truly industrial scale, which would bring costs down. It has a half-life of around 90 years degrading at 1%/year, so once you produce it it'll power the colony for a long time.
It's also around 6% efficient at generating electricity[1], but 100% efficient at generating heat, which is a huge part of energy requirements on Mars when it comes to human habitation. The Curiosity rover generates 120W of electricity but 2000W of heat.
1. https://physics.stackexchange.com/questions/34203/mars-curio...
Pu-238 isn't found in nature, but has to be synthesized. The precursor material, U-238 is the most abundant form of Uranium, which is convenient. It's also produce via neptunium-237.
The real complication is that the nuclear synthesis processes are also those used for nuclear weapons production, which makes for some serious complications.
You have to be pretty insane to launch large amounts of nuclear material from earth in a rocket that can randomly explode in our atmosphere.
That's why you build them on earth and transport them to mars.
There are some small designs such as the 4MW(el) u-battery that are small enough for a transport.