The Seebeck effect results in the generation of electrical power across a Peltier semiconductor junction. Most Peltier junction arrays in use are operated as heat pumps. When DC power is applied, heat is transferred from the cold side to the hot side. Since they have no moving parts, unlike a mechanical refrigeration system, and can be operated directly off of low voltage DC power, they are almost perfect for a lot of different applications.
The Seebeck effect is the reverse of the Peltier effect. A temperature gradient between the two sides of the Peltier junction will generate an electrical current across the junction.
Peltier junctions used to be rather rare, used only to cool some exotic telecommunications and ultra-sensitive amplifier circuitry to reduce electrical noise. They are in much more widespread use now, in everything from CPU coolers to small DC-powered refrigerators. The hot and cold sides of a Peltier can be reversed merely by reversing the power supply polarity.
I had read about the Seebeck effect long ago, and was not under the impression that it provided much electrical energy. I guessed it would be somewhere in the range of a few millivolts across a fairly large Peltier array.
I have a DC-powered cooler on my desk, which is switchable between a cold mode and a hot mode. I had it set in hot mode for about an hour, which created about (this is a rough estimate) a 60 degree Farenheit temperature gradient between the aluminum cooler liner and the external heatsink. Closer inspection of the cooler revealed that a solid aluminum block couples the Peltier to the aluminum liner. The cooler's external heatsink is rather heavy as well; it's about a pound or so of aluminum. When I switched off the cooler and disconnected power, the red HOT LED indicator remained on for about two minutes! It started out at about half the brightness it works at when the cooler is operating on 12 volts DC, and slowly dimmed out. Since the cooler does not contain any filter capacitors, I knew that the power had to be coming off the Peltier array itself. This gave me an interesting idea...
Some new microcontrollers draw as little as a few microamps while in operation, and most are capable of being set to a 'sleep' state when no program instructions need to be run. The Peltier array in my cooler had to be generating at least 1.6 volts to light the LED, at about 10 milliamps. In conjunction with a storage battery or capacitor, a solar collector, and a heatsink, a Peltier array could be capable of providing a source of electrical power for a low-current device such as a weather station or data logger.
Hmm, this demands a bit of experimentation...