The schematic below is of a CD4007 MOS transistor chip.
From the schematic it can be seen that on either side, PMOS or NMOS, the arrows indicating what type of MOS transistor it is are all connected to a single pin - pin 7 for the NMOS side, and pin 14 on the PMOS side. In this case, pin 7 is a source and pin 14 is a drain.
Why is that? Does anything need to be connected to those pins to use the other transistor in the chip?
That is the body connection. All MOSFETs have a fourth connection to the substrate, but for discrete MOSFETs this is normally connected internally to the source which results in a parasitic body diode anti-parallel to the source-drain. So in a typical discrete MOSFET you see the body connection tied to the source terminal and you get the MOSFET symbol which you are used to see which looks like this:
simulate this circuit – Schematic created using CircuitLab
But if you are designing on silicon wafer you get to do whatever you want with the body connection. On a very few discrete transistors, the body connection is run out so you can use it however you want. One such application is a CMOS transmission gate.
Taken from: https://learnabout-electronics.org/Digital/dig55.php
The NMOS body is connected to the most negative rail in the system and the PMOS connects to the most positive rail in the system (much like in your schematic). Doing this allow the MOSFET to block current in both directions which is something it cannot do when a body diode is present. As a result, you can't make a CMOS transmission gate if the transistor body connection is fixed to the source terminal.
I wish more MOSFETs did run the body connection out. There might be other ways or reasons to connect the body, but I don't know them.
The voltage between body and source modulates the drain current in similar fashion to the voltage between gate and source. The body behaves similarly to a JFET gate. For switching, it's generally fine to connect body connections to the power rails. But for linear circuits, you may want something different. For example, if the source voltage varies and the body is fixed, the gate-source voltage for a given current will vary, which may be undesirable. It may also be desirable to bootstrap the body to minimize source capacitance. I believe these considerations drove the design of some of these circuits, where the P-channel transistors in the differential amplifiers have body connected to source, even when source is not Vdd.
In common bulk CMOS processes, the body of the N-channel transistors is the P-doped chip substrate, so all N-channel transistors share the same body connection. You see this in the 4007. The body for a P-channel is an N-doped "well", and there can be separate wells for different parts of the circuit. But I believe the 4007 has a single well for its three P-channel transistors, thus a single body connection for them, too.
I remember years ago seeing a frequency mixer using the body as a gate. I can't point you at a reference. There are better ways to do this.
