We've had this type of discussion before, and the conclusions we reached IIRC is that if an EMP is strong enough to take out other semiconductor components, then it will take out an LED.

An EMP generates a high amplitude burst of rf energy, amplitude in this case being voltage. Semiconductors junctions are susceptible to this voltage when the amplitude is reverse to the bias of the junction. If you can understand that semiconductors work like a one way valve. They can take a fair amount of forward flow, but don't like it much when the flow reverses and the pressure gets too high. This is called peak inverse voltage, or PIV. Typically most silicon junctions have a PIV limit of between 100 and 250 volts. The junction offers a relatively high impedence to current flow up to the PIV limit, then the "barrier" is compromised, usually resulting in a temporary short circuit at the junction with high instantaneous current flow until the junction overheats and then it blows open, meaning no more current flow period. LEDs are special diodes that generate photons as the electrons migrate across the junction in forward flow. As far as I can remember, LEDS have about the same PIV as normal silicon diodes, or somewhere in the same general range.

What this means is that when an EMP occurs, the pulse may not contain a lot of energy, but it will have a high voltage, and across a great range of frequencies (I believe there is a bell curve of frequency generation depending on the EMP event). As the wave encounters the junction of semiconductors, the undulating voltage creates a high potential, low current source of electricity. When the voltage is inverse to the polarity of the junction, and the amplitude exceeds the PIV, then the junction blows, and the component is rendered useless.

Since junction blows usually occur at the weakest point in the cross setional surface of the junction, it doesn't seem to matter too much how big the device is, although there is some heat sink protection in bigger components, but it is not significant enough to be much protection at all. Surface mount and complex Integrated circuits (ICs) being built today incorporate more exotic semiconductor materials that will be even more susceptible to EMP, so the cell phones, pdas, and laptops will almost certainly croak in most any EMP environment.

Faraday screens have limits as to how much voltage they can dampen. Basically they are sinking the signal to ground, and there are so many factors to consider that simply putting your device into a grounded metal box is far too simplistic. The magnitude of the EMP and the proximity to it are perhaps the two most significant factors that will determine the effectiveness.

I believe the conclusion was that unless the device was buried in the ground a thousand or so feet deep in a highly conductive medium, the chances it would be safe from EMP are pretty slim no matter what you do. It doesn't matter if the component is in a circuit or sitting in a little baggie, they are going to get popped just as bad either way. In the movie War of the Worlds with Tom Cruise, the mechanic replaced the solid state device(s) in the car that Tom then drove off. In reality, the replacement components would be just as useless as what was already in the vehicle.