It is called a tee because the 3 ports are often arranged in the shape of a T.[1] Conceptually, the bias tee can be viewed as an ideal capacitor that allows AC through but blocks the DC bias and an ideal inductor that blocks AC but allows DC.
At a high enough frequency, the stray capacitance presents a low-impedance shunt path for the RF signal, and the bias tee becomes ineffective.
[a][2] For example, a Picosecond Pulse Labs model 5580 bias tee works from 10 kHz to 15 GHz.
To show the advantage of additional components, Johnson provided a simulation of a bias tee that used just inductors and capacitors without Q suppression.
Bias tees are used in a variety of applications, but are generally used to provide an RF signal and (DC) power to a remote device where running two separate cables would not be advantageous.
[7] Biasing is often used with photodiodes (vacuum and solid state), Microchannel plate detectors, transistors, and triodes, so that high frequencies from the signal do not leak into a common power supply rail.
[10] The telephone line for plain old telephone service and some early microphones use a bias tee circuit—often with a gyrator replacing the inductor—this enables a thin cable with only 2 conductors to send power from the system to the device, and send audio from the device back to the system.
This front-end of a bias "T" typically consists of a bandpass filter, a low noise amplifier, and a mixer coupled to a local oscillator.
[11] At one point a small slice is cut out of the center conductor, therefore a capacitor is formed and low frequencies are blocked.
Frequencies between 1 MHz and 1 GHz do leak into this port, so there is a second coil with a cone shaped core outside of the outer conductor, but inside of a housing to avoid interference with other components.
It starts with a high impedance, so a lot of power will be reflected, but the rest will travel down the coil and there is some leakage into the low frequency port.