LED with IV curve showing current driven negative resistance ... Any clue ???

[Enea]

Hi Folks,
I was testing some 650nm GaAs/AlGaAs LEDs when I found a batch showing this very strange IV curve (attached) with current driven negative resistance. Any idea what is going on ?

The affected LEDs have all an unusual higher forward voltage. Even up to 4V for 1mA  instead of a regular 1.8/2V for my processed LEDs

The negative resistance curve has a threshold-like behavior. Increasing just few mA brings down the Voltage of roughly 1V (from 5V where the threshold seem to be) and increase current from7/8 mA to 20/25mA

Increasing the current above 40mA rectifies the IV curve, brings back the curve to a textbook graph, decrease the initial higher forward voltage of a couple of Volts, but also decrease of 50% light output at 1mA (I believe simply because I am pumping too much current in my very thin hereto junction)

Any helps or advice would be useful. The structure of the LEDs and metalization has nothing fancy at all, just regular stuff for this kind of red LEDs

Please Ignore the lines at 2V in the picture. Is just an artefact created by the machine!!

[Zero999]

I've not seen that before. Are you sure you're not reverse biasing the offending LEDs?

[Enea]

Supersure. I have hundreds of LEDs with IV curve like that!! An entire batch

[T3sl4co1l]

 

Cool!

They still light up?

Maybe it's a processing error, like oxide over the metallization, and you're getting tunneling through it or something.  This would have an equivalent circuit of LED in series with TD.

Does the negative resistance occur in reverse?  (You may need quite a large voltage to avalanche the LED junction, 30 to 200V; if the negative resistance is symmetrical, it will appear as a ~3V drop in breakdown voltage, on top of whatever the nominal breakdown voltage is.)  After reverse breakdown, does it still work in forward?

Is the negative resistance modulated by light, at all?  (You may need a quite intense source to test this, like a 3W white LED and collimating lens, or a xenon flash with a synchronized electronic trigger on the sweep.)

Tim

[floobydust]

I've seen this with white 5mm LED's and thought I was going crazy.  Also on failed automotive LED's that go intermittent.

It seems to be some strange "SCR"-like action occurring at the bonding wire, as if an extra semiconductor at the contact is there. Heating up the LED makes it go zany, intermittent or fixes it. Taking it up to rated current and back down seems to make it latch.

I run 5mm white LED's off +350VDC with 220k series resistor. Several would not work consistently, but did on multimeter diode test. Added a reverse-diode, no change.
Drove me nuts, I kept changing LED's until I found one that consistently works.

[T3sl4co1l]

OH-- does it depend on temperature!?

I've seen that before, LEDs can start flashing if the bond wire becomes broken.  Perhaps a marginal break can be a (vacuum) tunnel junction?  Perhaps the hysteresis is actually a small arc*, or thermally driven?  If the latter case, then the curve should be very different depending on how fast the curve is traced.

*Unlikely below 20V (ionization of most gasses), actually.

Tim

[Enea]

The characteristic appears at room temperature and only in forward voltage, nothing on reverse voltage.

Seems a sort of tunneling somewhere, but I cannot understand where!!

Moreover and more interesting, light output BEFORE the NDR is bigger than after the strange effect takes place, suggesting a tunneling effect somewhere but unclear where and who are the actors in this tunneling!!

[Jay_Diddy_B]

Enea,

Can you do a few simple tests on your curve tracer and post the results here?

1) replace the LED with a 200 or 220 Ohm resistor.

2) 3 or 4 small signal diodes (1N4148 or similar) connected in series.

Keep the settings of the curve trace the same, 1V and 5mA.

I just want to see what your curve trace displays for some known components.

Regards,

Jay_Diddy_B
 

[Enea]

Hi Jay,
the test equipment has been tested and checked already and works just fine!! Is not an equipment related problem unfortunately!

[Chalcogenide]

Does the effect become stronger / weaker when the LED itself is strongly illuminated by an other light source?

[jolshefsky]

Does the effect become stronger / weaker when the LED itself is strongly illuminated by an other light source?

Indeed: GaAs green LEDs can source a tiny amount of current when exposed to light. There's a trick circuit somewhere out there that uses a powered green LED in a tube coupled to an un-powered one to provide a small DC voltage source for rail-to-rail op-amp operation I think. I would be that a tiny DC current source would look a lot like negative resistance.