Linear LED? Non-linear LED? Strange LED Phenomenon.

[BlownUpCapacitor]

So I recently got a large batch of old LEDs free with an oscilloscope I bought from someone. I always test second-hand components and throw out any bad components.

So I was testing these LEDs when I came across this LED that was linear at first, indicating a short of some kind and thus turning the LED into a resistor, but as I cranked up the emitter/collector voltage on my curve tracer, the LED suddenly started glowing and the LED curve looked fairly normal for a diode. I found this quite funny and tried it again only to have the same result.

So curious me decided to try reverse biasing the LED to check for any resistive components across the LED, as the LED would block reverse current, allowing me to isolate the resistive component. But the strange bit was how the linear line kept shifting and decreasing the "slope ratio" as you would find in simple linear mathematical functions such as y = mx + b. I tried it again a couple times only to get the same result. A "stepping" linear graph. The LED changes its resistance according to voltage but keeps that resistance until it is forward-biased again.

Here is a video that will probably explain the phenomenon far better than my words can describe: https://youtu.be/h_juq_nrUU4?si=vR2cqZOlBnNJthyy

Any thoughts on what could be happening? Maybe there is an insulating barrier that breaks down with more voltage, something like the Coherer effect? But that doesn't really explain why it resets when applying a reverse voltage.

[Kim Christensen]

How does this LED behave on pure DC with a current limiting resistor?
Could it be a flashing LED that keeps getting reset by the sweep of the curve tracer? Or a faulty flashing LED and you're seeing the active circuity behavior on the curve tracer?

[moffy]

It reminds me of automatic scale changing, maybe as Kim Christensen suggests, if you plot it by hand and remove any complexities introduced by the curve tracer.

[BlownUpCapacitor]

How does this LED behave on pure DC with a current limiting resistor?
Could it be a flashing LED that keeps getting reset by the sweep of the curve tracer? Or a faulty flashing LED and you're seeing the active circuity behavior on the curve tracer?

It acts just like a normal blue LED. Nothing special, no blinking. Just a blue LED.

if you plot it by hand and remove any complexities introduced by the curve tracer.

The curve tracer doesn't quite seem to be the problem. I used my 2465 and a simple resistor & LED set up to measure the current flowing through the circuit and the voltage across the diode to plot them on my scope, much like a curve tracer, except the thing doesn't output a triangle wave to sweep the voltage 50 times per second.

I have attached a schematic and a video. I apologize for my poor camera handling ahead of time, but I think it gets the information and data across.

https://youtu.be/qY6TyUI7Nuw?si=mqX9xCRzQ_ffZHP4

[mikerj]

Looks like it has some damage causing a high leakage current.  The reverse voltage characteristic in particular is not right, with current quickly increasing immediately below 0v.  Typically you'd expect very low reverse currents until you start approaching the breakdown voltage.

[moffy]

Fascinating, thanks for the video. It looks like a resistor in the forward direction until a threshold is reached, and something with negative resistance, like SCR, UJT etc. fires bypassing the resistor and it becomes just an LED/diode. It doesn't reset until 0 current is flowing through it again suggesting something like an SCR, maybe parasitic SCRs and the substrate? doesn't quite sound right but a guess.

[BlownUpCapacitor]

Looks like it has some damage causing a high leakage current.  The reverse voltage characteristic in particular is not right, with current quickly increasing immediately below 0v.  Typically you'd expect very low reverse currents until you start approaching the breakdown voltage.

This is a very interesting phenomenon. The reverse trace is very linear, which boggles my mind. As voltage increases, I guess parts in the LED breakdown in successive order, decreasing the resistance and thus creating that stepping effect.

Fascinating, thanks for the video. It looks like a resistor in the forward direction until a threshold is reached, and something with negative resistance, like SCR, UJT etc. fires bypassing the resistor and it becomes just an LED/diode. It doesn't reset until 0 current is flowing through it again suggesting something like an SCR, maybe parasitic SCRs and the substrate? doesn't quite sound right but a guess.

Negative resistance stuff is exactly what I thought at first. But it doesn't quite follow.

The reset stuff doesn't seem to happen immediately at 0V. It is immediate in reverse. Maybe there is something capative at play and the LED didn't really reach 0v and I need to short it for that to happen. I'll try that tomorrow.

It's also remarkable how consistent this effect is with this LED. I would think that it would beball over the place for a "defect", but no I guess.

Perhaps I've discovered a new semiconductor phenomenon. Maybe someone will find a way to make it into a useful effect  Then I'll be accepted into the bestest of best colleges. (wishful thinking likely not to come true :'( ).

[Kleinstein]

It looks like that a "high" reverse voltage charges some parts that seem to keep the "charge" / memory for quite some time. So I don't think it is thermal, but more like a charges surface or island that switches the LED to a different state, more like a resistor. The effect seem to happen even in multiple stages, likely multiple such elements in parallel.

Chances are that just waiting with zero volatage after it reached the linear state would also reset the LED to normal, just slower.

[CatalinaWOW]

If nothing else this is an excellent example of how the simple models we use for components don't always apply.  Much of this behavior would have escaped notice in a typical application.  The LED would have been forward biased, and glowed blue just as expected.

I don't have a good explanation, though Kleinstein's theory has merit, and the thought that it is really one of those pulsing diodes could fit in with this.  If such a pulsing diode was damaged by overvoltage or overheating the failed control IC would offer many more opportunities for the kind of parasitics mentioned by Kleinstein.

[Phil1977]

I´m no semiconductor expert, but remember some stuff:

A diode in reverse polarity creates a so called depletion zone. This is free of charge carriers, until the breakdown voltage is reached.

In reverse breakdown, charge carriers pass through this zone with quite high energy. This seems to induce some kind of F-centers, in other words atoms that are ionized in the otherwise neutral volume.

So far I think it´s solid physics. Now it gets speculation: The F-centers may act as crystallographic defects around the junction. Crystallographic defects can transform a semiconductor to a resistor.

What´s really astonishing is that this effect seems to be reversible, but maybe the flow of current through the diode in forward direction neutralizes the defects again.

It´s just a theory - maybe it´s also just some artifact from the curve tracer.

Did you try this with other components? With normal diodes? Zener diodes? Other color LEDs?

In case it really is a resetable material property it would be an exotic form of 1-bit-memory 

[Phil1977]

It looks like that a "high" reverse voltage charges some parts that seem to keep the "charge" / memory for quite some time. So I don't think it is thermal, but more like a charges surface or island that switches the LED to a different state, more like a resistor. The effect seem to happen even in multiple stages, likely multiple such elements in parallel.

Chances are that just waiting with zero volatage after it reached the linear state would also reset the LED to normal, just slower.

I didn't see you reply in time, but this observation seems to match.

I also have one strange white LED the famous transistor tester always reports as "unknown device". It has a very low breakdown voltage of only around 5V, and I can also confirm it has some memory effect if it was in reverse breakdown. In my case the "annealing" was never suddenly as soon as it lights up, but only with time and higher currents. In my case the annealing could also be just thermal.

[Positron Enthusiast]

Sounds like the diode may have been damaged. Stuff like this usually proceeds device failure. I had an APD fail a couple months ago by exhibiting a parasitic negative resistance with a weird step response. Wasted a couple pages in my notebook only for it to fail completely the following week. I think the cause was static damage from stray fingers on the device but hard to say