Energy absorbed from ionizing radiation alpha, beta, gamma, cosmic rays frees electrons to move through the crystal lattice and some are trapped at imperfections in the lattice.
Subsequent heating of the crystal, or stimulation by absorption of light can release some of these trapped electrons with an associated emission of light - thermoluminescence TL or optically stimulated luminescence OSL respectively.
This is the technology used for dosimetry badges in areas where radiation safety is a concern. The time over which the badge has been exposed is well known, and the total radiation does controls the final luminescence.
The badges are heated TL , luminescence recorded, and total dose derived. Since we know the time period of exposure and total does, we know the average dose per unit time.
Now turn the process around; if you know the average dose per unit time, and the total dose from the luminescence, then you know the time period of exposure. This is the fundamental process behind luminescence dating TL and OSL , as well as electron spin resonance ESR dating, which uses a different technique to achieve the same result. But they can be used to measure significant periods nonetheless, especially in the range between about 40, to 50, years where radiocarbon dating cuts off, and the 1,, years or so required for most radiometric techniques to become reliable.
The idea here is that all materials carry extremely low concentrations of radiogenic isotopes, line Uranium, which in turn expose the material to extremely low doses of radiation over a long time. That radiation frees electrons that get trapped in crystal defects, just like dosimetry badges. The total population of trapped electrons in turn determines the total dose. If you know the average dose per unit time, by studying the geology of the site, you can then use the ratio of total dose over average dose, and get the time period.
Sunlight on a crystal will evict the trapped electrons much faster than background radiation puts them in. So once the crystal is buried, the "clock" starts. Dig up the crystal, measure its luminescence either optically or thermally stimulated , and you know the total dose. Compare with the average dose per unit time, and you know how long the crystal has been buried. This is a favorite means for dating buried sediments that are often rich in quartz and feldspar. For other materials, notably non translucent material, electrons become trapped in defects where the lattice potential is too deep and the electrons cannot be stimulated to come out.
In those cases, electron spin resonance ESR , which is much more complicated that luminescence techniques, can be used to count the number of trapped electrons by using a combination of microwaves and a variable magnetic field. The disadvantage of ESR is that it is much more complicated, and has larger uncertainties than luminescence techniques. The advantage of ESR is that, unlike luminescence, the electrons are not evicted from their traps, so the measurement can be repeated as desired on the same sample.
One of the key tests of reliability for any dating technique is the ability to intercompare with other techniques; they should all give the same age for the same sample, within the bounds of the usual experimental uncertainties. There is a lot of literature available that demonstrates intercomparison between these luminescence techniques and radiometric dating. But here is one recent, and very good example.
Australia's oldest human remains: All age results are considerably older than the previously assumed age of LM3 and demonstrate the necessity for directly dating hominid remains. We conclude that the Lake Mungo 3 burial documents the earliest known human presence on the Australian continent. The age implies that people who were skeletally within the range of the present Australian indigenous population colonized the continent during or before oxygen isotope stage 4 57,, years The authors performed the following dating measurements, with the indicated resulting ages expressed in thousands of years.