I see. Space agencies are running out of 238Pu.
www.techno-science.net/?onglet=news&news=5341

OK, I had a look at my Handbook of Chemistry and Physics. 90Sr (supposedly produced massively in uranium reactors) is among the least bad candidates, according to

90Sr -> 90Y -> 90Zr where each transition is 100% beta-.

My edition of the Hdbk gives no indication about the energy of the gamma emitted together with the beta, but I've never heard it could be absent. Has someone data?

The 0.5MeV beta is absorbed very quickly by any solid, but this process radiates photons. Use light elements to radiate softer photons which are then less difficult to shield.

Li melts readily, so it's unsuitable as a beta absorber for a spacecraft. Be is said to be toxic - though this is debated. Rather B or C or B4C3 or BN as a beta absorber. A thin coating within the radioactive pellet is enough.

However, Sr and Y and Zr will stop some beta rays, and these elements are heavy and produce harder photons. Could be nice to dilute them (as atoms or thin strips) in a lot of C to make the pellets.

However, I've read somewhere that a mission to Mars planned to use an RTG, and this really looks like a bad idea. We can still use Sun energy on Mars, especially as a thermal powerplant. No need to risk a radioactive accident at launch for that.

Ahum, yes, RTG based on 90Sr were built and used in the Soviet Union, and are a major contamination issue:

http://en.wikipedia.org/wiki/Radioisotope_...ctric_generator
http://en.wikipedia.org/wiki/Strontium-90 -> gamma is weak
http://en.wikipedia.org/wiki/Isotopes_of_yttrium#90 -> nothing about gamma

More about the Soviet design:
www.bellona.no/bellona.org/english_import_area/international/russia/navy/northern_fleet/incidents/31772

A proper design can be safer than what Soviet engineers did (together with subsequent bad administration and rocky history), but shielding 90Sr is definitely more difficult than 238Pu.

Apparently, 90-yttrium doesn't radiate much gamma neither - to be checked more carefully. So one would just need to concentrate on braking radiation (Bremsstrahlung), to be minimized by absorbing the betas with light atoms.

One design would alternate very thin layers of 90Sr with thin layers of B or Be etc, possibly by CVD or similar.

Direct conversion methods seem to have all big drawbacks, especially for the safety of a generator falling on Earth after a launch accident. My (quick) impression is that only thermal conversion methods can be somewhat safer.

But then, thermocouples are by no means the only solution. It may be the most reliable for being static, but a true thermal engine boasts a better efficiency. And I wouldn't choose spontaneously a Stirling engine as Nasa did, but rather a turbine, which runs quieter and longer. The whole turbine+generator can be enclosed in the hermetic vessel. A few heat exchangers bring a turbine the same efficiency as a reciprocating engine has. Gas bearings look nice.

As a working fluid, I would avoid any phase change and associated issues. Simple gases like Kr, (N2), SF6 must be stable under irradiation, leak little and bring moderate speeds that simplify the turbine. A double-hull pressure vessel might allow to pump any leak from the intermediate volume back to the working chamber.

The safety vessel that resists a launch impact can't be the whole gas circuit. But it can include heat pipes between the pellets and the vessel.

In case the RTG falls on Earth at launch, pellets made of very thin layers of Sr0 alternating with thin layers of BN should be less bad than Zr alternating with B or Be. At least, they won't catch fire and spread fumes.

Make by CVD as well.

First, Nist publishes for free some very useful diagrams, for instance the absorption of electrons and photons through various elements:
http://physics.nist.gov/PhysRefData/Star/Text/ESTAR.html
http://physics.nist.gov/PhysRefData/XrayMassCoef/tab3.html
really nice of them.

Then, I've read the confirmation that the 2MeV electrons radiated by 90Y create important amounts of photons when braking, and shielding these requires about 10cm lead. Not welcome at all onboard a satellite.

Little can be done against this braking radiation (Bremsstrahlung). Braking by lighter elements would reduce the amount of hard photons, but as the volume increases since 90Sr is diluted in C or BN for instance, the shielding hull has also more surface and I guess more volume.

So I would say: on a satellite, stick to alpha emitters like 238Pu, or maybe use tritium (only 18keV betas). Replace with solar panels where possible. Improve efficiency elsewhere.

On Earth...
- There are so many better ways to produce energy!
- At Sellafield and La Hague, one may want to extract some energy from the waste. No additional dissemination, OK.
- Don't make many small generators. Your enemies will use them as dirty bombs.
- As an electricity and heat source for a base in Antarctica, maybe.
- As a town generator in Northwest Territory, it's already too dangerous.

Tritium : I've found no way to pack enough of it in a mass reasonable for a satellite. Just as anybody. The container is too heavy, not tritium itself.

Some propose tu use a tube of carbon fibres. Too dangerous to my taste in a crash or a reentry. I'd only trust a resilient alloy.

Soaking palladium doesn't improve the weight so much, and apparently, palladium won't absorb the helium produced by radioactivity. One might try to find a vessel material that leaks helium and is securely tritium-tight, but I doubt such a thing exists.

So: tritium doesn't seem to be usable for a satellite. Pity, as all its radiations are easily stopped.