LEDs typically specify Peak Current as something like 100mA @ duty cycle = 10%, and a "Test Condition" (I_F, i.e. Continuous Current) current of, for example, 20mA. That's a 2-dimensional constraint though. Is there some rule that dictates how much current an LED can handle up to some higher duty cycle given the extremes? I've never seen a 2-D graph relating maximum current to PWM duty-cycle in any LED datasheet... for example this one.
If you're talking about interpolating, you're probably reasonably safe. For example, if 100mA is okay at 10% and 30mA continuous is okay, then
Extrapolating is not safe. There are many different kinds of failure modes, and the one that is dominant at 10% duty cycle (perhaps thermal effects on the die) is not likely the same one that is dominant at 1% duty cycle (perhaps electromigration or bonding wire heating). You just can't tell. And it could manifest itself as a failure in a short time, or it might degrade the life so it's hundreds of hours.
Here are some examples from Cree of electromigration failures:
If you're looking for LEDs that will have a reasonable life with short duty cycles, you might consider the ones rated for flash use.
Cree has an excellent AN on pulsed over-current driving of their XLamp LEDs.
It says, for example:
A particular device subjected to Figure1 4 pulsed over-current driving repeated transients at an amplitude some percentage above the data-sheet limits but below the threshold required for single-pulse failure will still eventually fail. The failure mechanism will most likely be due to electromigration as enough metal ions are eventually shifted away from their original lattice positions. The other factor that can lead to a reduced lifetime is excessive heating of the p-n junction, which causes the LED’s output to degrade below 70 percent of its original luminous flux.
Their conclusion leads to the following guidelines:
Based on the 1-KHz pulse testing we have reviewed in this application note, Cree suggests the following guidelines for pulsed current operations:
For duty cycles between 51-100%, do not exceed 100% of the maximum rated current;
For duty cycles between 10-50%, do not exceed more than 200% of the maximum rated current;
For duty cycles less than 10%, do not exceed more than 300% of the maximum rated current
Those are guidelines, not specifications, and are not guarantees by Cree, nor would they necessarily apply to any other manufacturer's products.
I can say that I've done some testing of LEDs under pulsed conditions, which I cannot share due to NDA requirements, but I don't find the Cree guidelines especially surprising.
Is there some rule that dictates how much current an LED can handle up to some higher duty cycle given the extremes? I've never seen a 2-D graph relating maximum current to PWM duty-cycle in any LED datasheet...
There is no rule other than what the manufacturer shows in their datasheet. The Manufacturer, having the best access to the source material, a full understanding of their own production and test methods, will have tested the batches to provide the best (Read: Most Profitable) performance at constant or pulsed current/brightness/lifetime. No sense in testing say 500mA at 20% if it will die in a few hours.
That said, the 100mA @ 10% is for the stated Lifetime of the LED. You could try a higher current at a longer duty cycle.