A pass transistor is a MOSFET in which an input is applied not only to the gate, but also to the drain (see MOSFET if these terms are foreign to you). Pass transistor logic is an alternative to CMOS that is advantageous for some circuits, such as XOR's and decoders. As a simple example of pass transistor logic, consider the following AND gate.
B
|
----
----
| |
A-------- -----
_ |
B |
| |
---- |--A AND B
---- |
| | |
GND------ -----
Neglecting the fact that an inverted B is required (this might already be available), the pass transistor AND has only two transistors. As a comparison, a CMOS NAND gate requires four transistors, and a CMOS AND requires six. Thus pass transistor logic seems promising.
There's a catch. An NMOS transistor shuts off when its gate/source voltage falls below its threshold voltage. Consider the case when A is at the power supply voltage V and B switches from 0 to V. Since both A and B are binary 1, we expect the output to be V. However, as the output rises, the gate/source voltage of the upper NMOSFET drops. When the output reaches (V minus the threshold voltage), it quits rising because the NMOSFET is nonconducting and can no longer charge the output. For this reason it is said that NMOSFET's pass weak 1s. Similarly, PMOSFET's pass weak 0s.
One solution to the weak 1 and 0 problem is to put PMOSFET and NMOSFET pass transistors (with inverted inputs) in parallel with eachother. The parallel combination is called a transmission gate. Since one or the other passes a strong 1 and 0, the threshold voltage drop at the output does not occur. Of course, this requires more transistors.
Another solution is to use a level restorer to convert the weak signal to a strong one. The level restorer also adds transistors to the circuit.
Whether pass transistors or CMOS is a better logic style depends on the particular circuit. One example of a logic gate in which pass transistor logic is far more efficient than CMOS is the XOR.