I'm just a plain old (non-physicist, non-degreed) person, but i have to say this:
aaron: (probably been answered adequately but...)
what applications are willing to risk having radioactive materials on board just to avoid changing the battery more often? yeah. pacemakers currently. just about every military vehicle in existence as backup power for communications and environmental equipment if they are light, rugged, and cheap. solar photovoltaics just aren't always there for you.
low energy beta radiation, btw, is not the same as the myriad other forms and energy levels of radiation. I humbly suggest that the chances of breaking open a betavoltaic battery and sniffing some tritium (as the low energy beta from tritium cannot penetrate even your outer skin) are still preferable to the certain huffing of diesel generator fumes as exist for remote terrestrial power applications today. the risk compared to other battlefield hazards make it trivial. (oh, say like getting shot, friendly fire, chemical weapons, vaporised DU, etc.)
I'm certainly no expert, but if you're going to bother with posting to a physics site, you might as well alleviate some of that FUD regarding the spooky term "radiation" by visiting wikipedia or a high school chemistry class more often.
even Greenpeace has FINALLY come around to see the light (ie. long term impact) on radioactive power generation, including responsibly designed, integrally reprocessed, traditional uranium fission. it took long enough -- so let's not retreat back into a knee-jerk shell of ignorant fear now. I mean, really, the stuff is here, and is going to decay anyway, we might as well get something from it.
Jill England: The cost of producing tritium in large quantities is the other radioactive isotopes generated... ...could be used to create other dangerous radioactive materialsI look at it the other way -- aren't there neutron producing processes that are valuable in other ways who's neutrons are just lonely for some nitrogen? In other words can't tritium generation be used to increase end to end efficiency of other already existing processes? Maybe not, but I'm certain the price would be vastly reduced in volume. Ice was once a fantastically expensive luxury in some climates (if it could be had at all). No longer. Besides isn't the end product of Helium-3 quite valuable right now?
As for the other... well I hope I don't sound like a jerk but dangerous things can always be used in dangerous ways. But that is a social-economic cause, and should not be the sole limit on technology. The total number of people killed by building bricks to the head must be a tremendous number, but restricting the use of building bricks or iron pipe is obviously overkill and in itself destructive to society. The phenomenon of music downloads, DRM, etc. is another display of this, but in the more comfortable area of theft (vs weapons). There is no technological solution, only a societal one. Terrorists with high level encryption are vastly harder to observe, but restricting my use of encrypted email to a business partner (or quicken file) will not help with this problem, as terrorist will probably not balk at a software export law (to say the least)!
There will always be clever people like the Radioactive Boy Scout (don't need to be "smart", just clever), and it is the job of society to make sure the the boy scouts of the future are otherwise decent people like David Hahn, rather than evil sociopathic beastards like Ali Muhammed. But that is not the job of the physicist, IMHO.
Besides tritium is not cesium or strontium-90 or apple juice or a mouse. It is tritium, with it's own unique characteristics and should be used accordingly. The commonly used "lay-logic" or "fox-news-logic" that using tritium in a different method, different machine than cesium or strontium somehow means that it tritium is equally dangerous is infuriating to me.

I could care less if tritium is heavier or lighter than a duck. (obligatory car analogy: a Ford Pinto does not equal a 2007 Mercedes in terms of safety. yes they are both cars with the potential to maim and kill, but their potential to injure the driver in similar circumstances is enough different that many billions of industry dollars were spent on developing that difference in the potential to injure. The differences in radioactive substances and processes may seem small to us lay-people but they are NOT insignificant.)
Guest_David2.5 *10^4 Ci of Tritium are needed per Watt of beta power. Yes, but how many grams is that. No really, I have no idea. But isn't the power to weight ratio rather good?
my questions for all you smart ones:Speaking of which, couldn't a sealed chunk of silicon aerogel-like material be used as a medium to further lighten the load? Or is that actually what they are using? Or does it go too far and is not dense enough for effective capture? Several comments here concern "low power" of such a battery, but would not total power be a function of the mass of tritium? Surely, then, practical applications for a given power requirement are limited instead by VOLUME?
BTW, a use of unmanned air vehicles (other than the violent ones mentioned) that I'm interested in is geosynchronous (or "city-synchronous") high altitude aircraft as communication relays instead of the hassles of orbital satellites. The planes already exist, but solar panels on the wings just don't quite cut it. A very lightweight power source, even just to trickle charge capacitors for night time would make the difference!
I'd much rather have some of these guys in the air than a bazillion cell towers crowding cities and every telephone pole spewing microwave radiation for city wide 802.xx networks-- such a waste.
Anyway, best of luck to the researchers, but I'm filing this under "killed by irrational fear / corporate politics" along with a SABRE/aerospike powered Delta Clipper, IFR/LFR fission reactors replacing mercury barfing coal plants, and profitable inertial confinement fusion by 2006 (the Z-machine's power generating successor, whatever they were going to call it before de-funding it to pay with some admittedly NIFty lasers.)
Update: I apologize for my cynicism. Though the descriptive web pages I bookmarked are long gone, somebody put the basic concept on wikipedia, so maybe it WILL get re-funded someday. After as many billions are sunk into NIF as were into superconducting toroids, of course. After all, why fund a concept that already works?

just joking, guys! Oh noooo! ((runs away from horde of enraged optical physicists))