Why anyone connect two MOSFETs in series?
I recently worked on a battery powered project with a buck-boost converter. When the battery is new, the converter steps down the voltage. As the battery ages, the converter boosts the voltage. An added wrinkle is a microprocessor varies the output voltage to levels optimized for specific operations.
When the battery voltage is near the desired output voltage, we can save power by connecting the output to the battery and disabling the power supply. Some power supply controller chips have a built in bypass feature. The part on this board does not, so we need a separate bypass circuit.
A P-type transistor connects the battery to the load when its gate is low. (See diagram on the right.) An N-type transistor is connected to this gate to ensure that the gate voltage can rise high enough to turn off the P transistor reliably. A diode is needed to prevent current from flowing backwards from the output, through the P-type FET’s body-drain diode, and back into the battery. 
The trouble with this approach is some voltage is dropped across the R[DS,on] of the transistor, and more is dropped across the Schottky. The power lost can be as much as the power saved from not running the switching power supply.
There are transistors with amazingly low V[th] and R[DS,on], but nothing can be done about the Schottky’s drop. It will drop at least 0.2V at a few milliamps of current.
A colleague suggested using two FETs in series. If the output never exceeds two diode drops above the battery voltage, however, the Schottky isn’t needed. I found with 1V across the two transistors, just under 1uA flowed. If the supply output is 2.5V and input is two 1.5V cells, no current will flow through the body-drain diodes until the cells are exhausted. The extra R[DS,on], assuming the voltage at the source is large enough for the transistor to be fully tuned on, results in much less voltage drop than a Schottky. This is important because it wastes less battery power and allows the device to run without the power supply at lower battery voltages.
I’d be interested to hear other ideas about other high efficiency approaches to bypassing a supply, such as transistors that bring out the body connection or mechanical relays.