The "nuclear" ones are ionization type, which have the useful quality of being cheap. Otherwise, they're almost, but not entirely, worthless as far as being a smoke detector goes.
The other type is the photoelectric type, which is generally more expensive, but rather radically more likely to alert for the sort of fires that kill people.
Ionization types are really good at detecting the emissions from flames. They're disturbingly bad at detecting actual smoke - so the name "smoke detector" for them is a bit misleading, IMO. If you've got a grease fire or something going with basically no smoke, an ionization type will detect this in a hurry, and they're also rather prone to false positives as noticed by anyone with one of them near a kitchen.
The photoelectric types aren't so good at detecting flames, but are quite good at detecting smoke in the air. In some tests, they alert 30-40 minutes before an ionization type notices a problem.
But the problem comes when you start looking at the type of fires that actually kill people. The bulk of fire deaths are from overnight fires, in which something is smoldering (and smoking...) for half an hour or longer before the heat gets to the point that the piece of (usually furniture) ignites. The photoelectric type will alert to this smoke. The ionization type waits until the [whatever] has actually caught fire to bother doing anything - which is far less useful, because by the time the couch has caught fire, the room and the rest of the house aren't far behind. Some realistic tests have shown that there's literally half an hour or more of alert from the photoelectric type, down to "A minute or two" for the ionization type.
Skip Walker has done a number of presentations, and some of his work can be found here:
http://www.propertyevaluation.net/Photoelectric%20vs%20Ioniz...
> In tests, ionization alarms will typically respond about 30 to 90 seconds faster to “fast-flame” fires than photoelectric smoke alarms. However, in smoldering fires ionization alarms respond an average of 15 to 50 minutes slower than photoelectric alarms. Several studies indicate that they will outright fail to activate up to 20-25% of the time. The vast majority of residential fire fatalities are due to smoke inhalation, not from the actual flames and almost two-thirds of fire fatalities occur at night while we sleep.
> In 2007, UL published the “Smoke Characterization Study”. This study tested both types of smoke alarms using current UL testing standards and materials; they also tested the alarms using UL test criteria integrating a variety of synthetic materials and current tests such as smoldering toast. The results are frightening. Ionization alarms failed the UL 217 test 20% of the time using the current standard test materials. This is the test that the alarms must pass 100% of the time to be offered for sale and installed in US homes. When tested using synthetic materials, ionization alarms DID NOT TRIGGER (DNT) in 7 out of 8 synthetic test scenarios. In the one test where the ionization alarm did trigger, it activated at a level exceeding maximum allowed under the UL standard and nearly 43 minutes after the photoelectric alarm in the same test.
He also has a presentation up that just drills down, over and over, into data, studies, etc. The results are quite clear: in the sort of fires that actually kill people, photoelectric smoke detectors radically outperform ionization types - and this is found in study after study that looks at "fires vs fire deaths vs smoke detector types" as well.
https://structuretech1.com/wp-content/uploads/2013/03/Ion-vs...
You can spider off those if you want, but you really should put at least a few photoelectric smoke detectors in your house.
Don't bother with the combo detectors. They're either an "or" gate, which gives you the worst of both worlds, or an "and" gate, at which point the photoelectric sensor can be screaming its head off about the smoke, but the alarm won't sound until the ionization type detects flame products.
> I'd rather have freaking IR cameras looking for too much heat (outside of kitchens).
They've been tested against smoke detectors and are worse than even ionization types for detecting early fires.