Specific link,
https://www.mouser.com/ProductDetail/Lumex/SML-LXT0805SIUGUBW?qs=sGAEpiMZZMseGfSY3csMkePlx4epE0bVaiw6Rw89oufT4s3ZlBHGtQ%3d%3dMany of these one-chip / three-diode packages seem to show having extra zeners between (but not all!), but what are the two zeners doing here? Why two that connect two specific diodes? I had a thought this may be because two of the LEDs have a different voltage than the third?
My plan was to use 5V common anode, give each LED it's own limiting resistor selected for it's forward voltage, and drive each cathode low with my micro. I don't immediately see how the Zener would be helpful for that, but only have a shadow of an idea what it's doing in other applications.
Simple answer?
Zeners possibly needed to protect from kickback voltage when long wires have high current turned off fast. Since there are 3 different dies, some of LEDs have zeners, and some do not.
Consider 1 meter of 1mm wire in free space carrying 80 mA. Inductance of this wire is 1500 nH, so energy of magnetic field is 5nJ. When shut with digital driver, the energy will turn into potential of capacitive load (closed FET or similar gate, looking like a capacitor). To not exceed 5V reverse, the capacitance of load must be higher than 400pF. If less than this, then you get overvoltage.
E=CU^2/2 = LI^2/2
One example of fault I seen is positional (pnp or npn) 24V sensors for pneumatics. They have some fast circuitry inside to drive output with 24V. The fault was that after weeks of being loaded to opto couplers, the optos died. The cause is long length of cable and optos having unprotected IR LED. And solution was to add zeners in parallel to LED.
I suspect they may be parasitic structures, like the body diode of a MOSFET; in the orientation shown they're not going to protect anything from overvoltage since they would just conduct, and in the other direction they would effectively disable the LED if their zenering voltage was less than Vf of the LED.
The higher voltage blue and green LEDs can be damaged by static electricity and the Zener diodes prevent it. The lower voltage red LED is not damaged by static electricity so does not need a Zener diode.
Three fairly different answers
Great, it's going to be one of those questions!
Audioguru is right: pure green and blue LEDs are easily damaged by reverse over-voltage, whilst red LEDs non-destructively conduct, like zener diodes, when the reverse breakdown voltage is exceeded.
I'm surprised they used zener diodes which will never conduct in reverse because the LED will conduct first. Ordinary silicon diodes would do.
Inductive kickback is unlikely to be a problem. Consider which end of the inductor is doing the kick.
ESD doesn't care about current flow, though: it just goes. GaInN (blue, high efficiency green) is indeed ESD sensitive, and dies suddenly at around 20V reverse.
Tim
I remember discovering this the hard way. I'd previously used a capacitive voltage dropper to power a red LED from the mains, with no anti-parallel diode and it worked perfectly, so decided to do the same with a nice blue LED and the LED died: no smoke, just a silent death. Blue LEDs were expensive at the time, this was around 1996, so I was a bit pissed off but I learned a valuable lesson. Now I always use a full-wave bridge rectifier or bipolar LED on AC, as I hate the flicker.