1
\$\begingroup\$

(The question is at the end of the text.) In have a recurring application, where I want to clamp ESD strikes into a sensitive high-impedance into a guard node using antiparallel diodes.

schematic

simulate this circuit – Schematic created using CircuitLab

The two basic requirements for D1 and D2 are:

  • low ideality factor (i.e. \$n\approx1\$) to have low leakage at very low bias [as Sphero notes, it's actually the product of \$n\$ and \$I_S\$ that's important with \$I_S\$ being dominant.]
  • good peak current handling

There are a few devices which do well to mediocre at this task, e.g. diode-connected transistors, JFET gate junctions or TVS diodes in forward-mode.

In contrast, as a matter of fact, nearly everything marketed as a "regular diode" completely fails at this task, as do Schottky diodes. Those devices (with a few exceptions) seem to be intentionally made with rather high ideality factors giving them abysmal leakage at low bias.

Question:

What is the advantage of these high ideality factors and how are they realized in the actual devices from a device-making point of view?

\$\endgroup\$

1 Answer 1

1
\$\begingroup\$

According to Wikipedia, n (emission coefficient, ideality factor):

mainly accounts for carrier recombination as the charge carriers cross the depletion region.

I think you're a bit off base with this though, the main variable is not n, which is typically between 1 and 2 for normal diodes, rather it's Is (saturation current).

For example, a 1N5817 model has Is = 31.7uA, n = 1.373 1N4148 has Is of 2.52nA, n = 1.752

Obviously, the current through a 1N5817 at (say) 0.2V forward bias will be much higher than through a 1N4148.

Aside from the relatively small effect of n (you can put two diodes in series and you'll have an effective n of < 1 for most normal diodes) there is a straightforward tradeoff between reverse leakage and forward voltage drop at normal operating current. Schottky diodes have relatively low forward voltage drop but can have enormous reverse leakage at high temperatures (sometimes amperes for a large rectifier).

If you want low leakage with relatively small forward (and reverse) biases, one kind of diode with very low Is is the LED (however you have to keep them in the dark). For example, a Nichia NSSW008CT-P1 has an Is of 0.23fA and an n of 3.43

Compare the currents in this circuit: enter image description here

From left to right, 5.07616e-009, 1.0048e-013, 9.78401e-005, 2.03773e-008, 0.000496834.

So the best is the LED at < 0.1pA and the worst is the single Schottky at 500uA.

\$\endgroup\$
3
  • \$\begingroup\$ Thanks for the extensive answer, one thing that is maybe missing before I can accept is: "What is the advantage of higher \$n\$ ? And yeah, you are certainly right about \$I_S\$ being dominant. Which leads me to the question (which is probably off-topic here) why conventional diodes seem to have such high \$I_S\$.. LEDs are a bit too fragile for direct ESD protection I guess (more of a gut-feeling than a fact). \$\endgroup\$
    – tobalt
    Feb 1, 2022 at 9:00
  • 1
    \$\begingroup\$ You want Is to be high for low Vf (which translates into low losses in a rectifier) and you want Is to be lower if reverse leakage is very important. I don't think there is any particular advantage to n being high, but if it comes along with other useful characteristics it is fine. The only case I can think of where it really matters is circuits that depend on n being 1 for log/antilog, temperature sensing/compensation and such like. A diode-connected BJT has n very close to 1, but very low breakdown voltage (<10V). \$\endgroup\$ Feb 1, 2022 at 9:19
  • \$\begingroup\$ Aha. Good point about the reverse breakdown voltage. All the examples mentioned by me in the question are indeed cases where a weak reverse blocking behavior doesn't matter. So it could be that there is some tradeoff in doping profile between high reverse blocking voltage and low \$n\$. \$\endgroup\$
    – tobalt
    Feb 1, 2022 at 9:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.