How to choose a MOSFET with as low Vgs as possible?

Question

I'd like to build a schematic which would work in a wide span of output voltages. The low boundary of this span will be defined by the Vgs voltage of chosen FET. The test schematic is the following:

The current is pretty low (up to 500 uAmps) so this mode usually don't described in devices datasheets. If I'd use any of part searcher like digikey - all the things I can filter of is Vgsth at some given current and drain-source voltage. Usually they are pretty high for me, moreover many of devices with Vgsth specified digikey as 400mV @ 100uA by digikey have no confirmation:

But NO confirmation in the datasheet, all graphs start with much higher Vgs:

Earlier I tested a MOSFET and I find out that in practice Vgs can be much lower at small currents (I've got the schematic worked at 0.6 V when anything predicted 1,67 V). I found an article "The MOS Transistor in Weak Inversion" which was pretty helpful for understanding of what happens inside the FET below specified "ON" state.

However, I can not be satisfied by only knowing the theory - I need the MOSFET able to work with as low voltage as possible. But: how I can pick one?

I tried to pick the devices by opting Vgsth with as low Vgsth as possible (digikey for instance starting with 400mV) but I need to read out all datasheets and compare devices in the region which is usually not specified. I tried to pick the devices with Rds specified at 2.5...2.7 Volts (assuming that those FETs can probably work at low Vgs)...

Now I stuck with BSS138 and IRMLM6244 which are chip and available at local dealer. So I can put them to the test to compare datasheets with my measurements and test in the low current region. But I'd like to have more devices to choose from.

The main question: is there any parameters of MOSFET which directly or indirectly can affect on a possibility to work under Vgsth and I can use to filter devices out? For instance: gate capacitance? Maximum Vds voltage?..

Answer 1

The number is correct, but the designation (at whatever distributor site- looks like Digikey) is wrong- it is 400mV minimum. You can't really trust such info (even) as much as datasheets (which also contain errors from time-to-time).

Vgs(th) is best used as an indication of where the MOSFET will be mostly 'off' so the minimum is usually of more interest than the maximum.

As far as finding a part that meets your specifications, you can use the imperfect sites and try to verify the data by looking at the datasheets one by one. I don't know of any better way.

By the way, if you're interested in particular in extremely low Vgs(th), you can use depletion mode transistors, where the voltage is actually negative, or you could consider boutique parts from ALD which are tweaked to have 0V +/-10mV Vgs(th).

Answer 2

Just turning this on its head to offer other options.

Why not use an energy harvesting chip to produce a substantially higher voltage such as 3 or 5 volts. The "real" input (say) 0.5 volts can be boosted to several volts and this is used to drive a more commonly available MOSFET.

TI's BQ25504 will operate (from cold) as low as 0.33 volts and I'm sure LTI have offerings in this area.

The downside is that the MOSFET switching cannot be near-instantaneous because energy has to be taken from the input signal and this might take a few tens of milli seconds.