Do LEDs fail open or short?

[hamiltont]

I'm playing with red LEDs (https://www.sparkfun.com/products/9590) to learn more about failure modes. Using a very simple arrangement - connecting one LED directly to a current-limiting power supply and catastrophically destroying it by raising the max allowed current. Oddly the I_F is 20mA (I_PF is 30mA) but I can supply these continuously with much more (typically about 120mA before they start to "dim" and 330 mA before they are failed short and acting as a resistor)

Some of the LEDs fail open dramatically e.g. bright flash then black burned spot on the LED. However, most of them fail short much less dramatically - slowly get dimmer and dimmer as the current increases and then eventually offer only a minor resistance e.g. 300  .

Is there a clear reason why some LEDs fail open and others fail short? In a "good" design, would you need to plan for both types of failure?

[chemelec]

If an LED is supplied with the Correct Current, It should Last a LONG TIME.
Can't remember when I have Ever had an LED Fail.

[Tom45]

I don't know about the open vs. short failure modes.

But a "good design" will limit the current through the LED anyway, so as chemelec points out, the LED is unlikely to fail at all. And if it does, the current limiting of the circuit will protect against bad things happening with a failing short LED.

Failing open obviously isn't going to be a problem either. Although if the compliance voltage of the current source is really high, I suppose failing open could expose other circuit elements to damage.

[MrAl]

Hi,

I think it depends highly on the method of destruction.  You could buy a pack of cheap LEDs and do some destructive tests.
Supply one with larger current, supply another with high reverse voltage, etc.  Test the outcomes.

I had one 'blow' but it didnt really blow it just got very dark purple.  It was a white LED that got too much current.  It still produced light but ti was very dim and purple in color.

[David Hess]

At lower currents, diodes fail with a short due to overheating of the semiconductor.  At higher currents, especially with LEDs which use a bond wire for the top connection, they fail open.

[Zero999]

At lower currents, diodes fail with a short due to overheating of the semiconductor.  At higher currents, especially with LEDs which use a bond wire for the top connection, they fail open.

That's true and in some cases, the bond wire can fail leaving the semiconductor intact.

I don't think I've ever seen an LED fail a very low resistance though. Normally they still have a significant resistance, compared to the PCB track and will limit the current enough to stop it or the wiring from melting.

[msuffidy]

Well a few days ago I wound up a wound up led light too fast and all the lights went out at the same time. I took it apart and it had a wind up motor, 4 rectification diodes, one current limiting resistor, and maybe a supercap or nicad across the bridge output. The diodes tested ok with the meter, the resistor I tried shorting, and the nicad was outputting. So I looked for the resistance of the diode and they were near zero, and I argued they were shorting the whole thing out. I cut them off and tried a led in its place and it lit up, but did not with them attached. At least in this case, they definitely failed closed.