Re: What makes incandesent light bulbs burn out?

Good question, but a pretty easy one to understand...

First, it would be good to talk a bit about how an incandescent bulb 
works.  All modern bulbs use thin metal filaments made of tungsten -- the 
Brits call it wolfram -- sometimes as coils and sometimes just as little 
wires.  Tungsten has some neat properties.  It's a pretty good electrical 
conductor, but not too good.  The "not too good" means that it has enough 
electrical resistance to cause it to heat up, which causes the resistance 
to go up some more, which causes it to heat up some more, and so on until 
it gets hot enough so that the amount of heat generated is equal to the 
amount of energy radiated by the yellow-hot wire.  The temperature at 
which the bulb stabilizes is high enough to radiate light -- which is 
good -- and tungsten has the property of not melting or vaporizing very 
quickly at the designed temperature (and these things are actually very 
carefully designed).  

After a bulb has operated for a really long time, however, a couple of bad 
things happen.  First, even though the tungsten wire doesn't vaporize very 
quickly, it does vaporize slowly -- reducing the size and, as a result, 
the strength of the wire.  Second, the very rapid heating of the wire 
causes it to grow in length quickly when the bulb is turned on -- and the 
cooling when it turns off causes it to shrink very quickly.  This "thermal 
shock" effect is analogous to what happens when you take a paper clip and 
subject it to periodic forces by bending it in your fingers.  After a 
while, the material becomes brittle (by a process involving buildup of 
microscopic flaws in a localized area that's too involved to talk about 
here).  Put it all together and you have a material that's being weakened 
by loss of material and embrittled by thermal cycling and you have a 
disaster in the making.  It can only take so many cycles and then, poof -- 
it dies.  

Interestingly, bulbs can fail due to the thermal shock either when being 
turned on or off -- but, of course, you'll only notice it when they turn 
on and fail to light.  My guess is that they actually fail more often when 
turning on, because it's a tougher (quicker) thermal transient, but I'm 
not certain of it.  In any case, after the wire breaks, the two loose ends 
sometimes remain near enough to each other so that they can touch briefly -
- if this happens when power is applied, you'll see a bright flash as the 
tungsten in the teeny contact point vaporizes and the vapor drives the 
wire ends apart, usually permanently.

Interestingly, tungsten halide bulbs operate have some similarities and 
some differences to regular tungsten incandescents.  They are designed for 
higher currents and hotter lamp temperatures, so the wires are "white hot" 
and give a lot of output for the electrical input.  This is very good, 
although the tungsten is boiling off from the filament at a ferocious 
rate.  By putting halogen vapor (I think they most often use iodine, but 
I'm not sure)in the bulb, the tungsten that boils off the filament and is 
deposited on the quartz bulb envelope reacts and goes back into the vapor 
form where -- if you get the chemistry just right -- it migrates back to 
the filament and decomposes into halogen gas and tungsten metal, which 
deposits back onto the filament.  Amazingly, the net result is a bulb 
which operates at high temperatures and brightness but which doesn't boil 
the filament away too rapidly... I believe that this process also helps to 
heal some, but not all, of the fatigue strain.  When enough tungsten goes 
to places on the envelope which are too cool for it to react chemically 
with the gas and/or when the residual fatigue is too great, this amazing 
bit of physical chemistry dies, too.  But it's wonderful science at work 
in a simple, useful way.

By the way, the physics of lighting -- particularly long-life, color-
corrected (i.e., so it looks like the sun) -- is NOT a dead science, and 
involves really interesting, ongoing work.  It's exceptionally difficult 
to find a way to make something last nearly forever, cost nothing, and 
produce lots of light at the right color(s), with little or no heat 
generated.

Thanks for the neat question!