Yes, you could, and you could do better with more moving parts and more stages, but they're optimizing for reliability not for power.

In theory, the best you could do is (T_hot - T_cold) / T_hot, or (430 - 210) / 430 ~= 50% , using some approximate values for fin root temperature and Martian climate. To exceed 10% of the theoretical optimum with something that contains no moving parts is pretty impressive, IMO.

Couldn't they use the heat to to drive the wheels directly, maybe they could go to more of a hybrid drive? Just seems like such a waste to produce that much heat for 15 years.

I'm not sure how heat would turn wheels directly. You need to convert the heat into either mechanical or electrical energy somehow. Any such system involves tradeoffs in efficiency (in terms of useful energy captured) against weight and reliability.

Specifically you wind up needing grease/oil more moving parts. The failure rates of reaction wheels on orbital craft speak well to the relative difficulty of keeping moving parts moving in space-environments without ongoing maintenance.

AFAIK NASA is already experimenting with Stirling engines + Plutonium for future missions. It’s called ASRG https://en.m.wikipedia.org/wiki/Advanced_Stirling_radioisoto...