3A diode with very small reverse current?

[dmendesf]

I´m looking for a diode that can handle 3A continuous in the forward direction with the lowest possible current (leakage current?) in the reverse polarization. Voltage will be low (10-15V maximum). I looked a lot of datasheets for different manufacturing technologies and the only conclusion was that Schottky diodes are bad. Anyone has any hint? I can live with 1A if it improves a lot the leakage. I´m targetting << 1uA. I´m even considering using a peltier to lower the temperature of the device but this is a last resort measure. Thanks for any help.

[Ice-Tea]

The search function of DigiKey is quite excellent.. I would suggest start with that ;-)

[mtwieg]

There are probably lots of options, even for schottky diodes. For example this one. Max If=6A, Max Ir=0.45uA at Tj=25C and Vr=100V.

But leakage is extremely dependent on temperature. At Tj=85C, leakage can increase by a factor of 250 (roughly doubles every 10C increase in temperature). So you would need to keep the diode from heating either due to ambient temperature or self-heating.

[Gyro]

I've found Vishay BYV27 (2A, round sintered glass package) to have usefully low reverse current at ambient (spec max 1uA at V_RRM) but way lower on test.

As pointed out above, you really need to clarify the conditions though, you're not going to get a diode to settle to low reverse leakage if it has just been heated by forward conduction. Also whether individual characterization is acceptable.

[David Hess]

Because of how they are constructed, bipolar transistor junctions are better than diode junctions.  So look for a bipolar transistor collector-base junction with the current and voltage requirements that you need.  Conveniently transistors also come in better power packages; you will not find many silicon diodes in TO-126 and TO-225 packages.

Nobody tests power diodes or power transistors for low reverse leakage current, not because they cannot have low leakage, but because testing costs money and their intended applications do not require it.  This means qualifying the parts you buy and confirming that they meet your specifications.  I used to do this with 2N3904s to get 1 picoamp diodes, and I would sample test incoming parts to make sure they met my specifications.  I currently have a batch of BC337-40s 45V 800mA (!) NPN transistors which have a leakage lower than 100 femtoamps; these are very handy for input protection diodes for multimeters.

So I just tested an MJE182 3A 80V NPN and its leakage at 20 volt was less than 1 nanoamp.  A 1N5400 3A 50V rectifier was 12 nanoamps under the same conditions.

[SiliconWizard]

There are some Schottky diodes which could fit your requirements (although read the fine print in datasheets), but keep in mind that with a few exceptions, "high-current" Schottky diodes tend to have a relatively high threshold voltage (higher than the typical 0.3 V of the low-current ones) and so will still dissipate significant power at 3 A.

If power loss is not a real concern, just pick a silicon diode instead, easiest. Calculate the max power loss and pick the diode accordingly (make also sure about heat dissipation).

If power loss is a concern, you can consider implementing some kind of "ideal diode" instead based off a MOSFET, but here again be careful with the reverse current. That may require using two transistors back-to-back.

[mtwieg]

Because of how they are constructed, bipolar transistor junctions are better than diode junctions.

Interesting, but shouldn't the junctions of a BJT be constrained by the same Shokley equation as normal PN diodes?

So look for a bipolar transistor collector-base junction with the current and voltage requirements that you need.

What specs would one look at for this? AFAIK most BJT datasheets don't say anything about forward conduction of the base-collector junction.

[TimFox]

As mentioned above, you won’t find what you want in the data sheet specified maximum reverse current, since current measurement at that level is too slow for automatic test equipment used to verify maximum values.
Yes, the Shockley equation holds for BJT junctions, but it has free parameters that vary dramatically from one construction to another.  I like the idea of using BJTs that are heat-sinkable to reduce junction heating by forward current.

[psydaddy]

What do you want to make? Reverse voltage protection?
If so:
- Can't you use a mosfet instead?
- Instead of using a series diode, use a parallel diode with a fuse.

[ejeffrey]

Interesting, but shouldn't the junctions of a BJT be constrained by the same Shokley equation as normal PN diodes?

Yes. But diode connected transistors are much closer to the ideal than silicon diodes.

[ejeffrey]

There are also more exotic options.I think silicon carbide shottky diodes have similar performance to silicon but less temperature dependence.  If your application requires working at high temperature that might be a better option.

[David Hess]

Because of how they are constructed, bipolar transistor junctions are better than diode junctions.

Interesting, but shouldn't the junctions of a BJT be constrained by the same Shockley equation as normal PN diodes?

It is, but the construction of a bipolar transistor protects most of the junction so there are fewer surface effects.  I suspect the diffusions are just better as well.  There is considerable variation between various diodes, as discussed below by Robert Pease.  The base-collector junction has worse conductivity and is slower than the base-emitter junction, which is almost perfect, but both have low leakage.

What specs would one look at for this? AFAIK most BJT datasheets don't say anything about forward conduction of the base-collector junction.

The base-collector maximum current is the specified maximum collector current, and the breakdown voltage is Vcbo or Vcb.

The base-emitter junction could be used, and it has advantages like better conduction and speed, but it has a lower maximum current and low breakdown voltage of about 5 volts.

Note that reverse recovery of the collector-base junction is relatively slow.

[PCB.Wiz]

I´m looking for a diode that can handle 3A continuous in the forward direction with the lowest possible current (leakage current?) in the reverse polarization. Voltage will be low (10-15V maximum). I looked a lot of datasheets for different manufacturing technologies and the only conclusion was that Schottky diodes are bad. Anyone has any hint? I can live with 1A if it improves a lot the leakage. I´m targetting << 1uA. I´m even considering using a peltier to lower the temperature of the device but this is a last resort measure. Thanks for any help.

Keeping the die cool is a good direction to go, but you can save a peltier by looking for very low thermal resistance and easily cooled packages.

Companies like ROHM have very well specified parts.
If we look at the RFN10RSM2STF for example.  10A 200V in a modern package.

That says MAX 1uA, but typical curves have it dropping under 1nA < 50V. Even at 75'C it is under 20nA below 40V 
The package has a low RθJC  3 ℃/W, so you can aggressively cool it with a big heatsink, or even a fan.

[dmendesf]

Thank you all for your valuable contributions. Searching for this parameter at Digikey is not an option because each manufacturer measures it in a different condition. I will try both the Rhom parts and some BJTs and see what works best.

[Ice-Tea]

It is a perfect starting point: it tells you what diodes might fit the bill. After that, it's plowing through datasheets and graphs..