# How does a MOS capacitor work?

#### HunterDX77M

Joined Sep 28, 2011
104
I'm starting on a design where capacitors and diodes must be made from transistors. I know diodes can be made from connecting the gate to the drain (like in a current mirror), but what about a MOS capacitor?

From searching around, I've found that shorting the drain and source together and using them as one terminal, along with the gate as the other makes a capacitor. Am I correct in assuming the oxide capacitance/area (C_ox) multiplied by the area (width and length of the device) determines its capacitance? What other factors are involved?

Also nagging me: how does current flow through the device if the gate oxide is an insulator?

#### crutschow

Joined Mar 14, 2008
25,431
Also nagging me: how does current flow through the device if the gate oxide is an insulator?
All capacitors have an insulator between their plates. That's how capacitors are made.
The MOS capacitor is no different.
That's why capacitors block DC and only pass AC.

#### dl324

Joined Mar 30, 2015
11,537
I'm starting on a design where capacitors and diodes must be made from transistors. I know diodes can be made from connecting the gate to the drain (like in a current mirror),
If you're talking about MOSFETs, you don't get a diode by connecting drain to gate. There's a parasitic diode between the source and drain.

Shorting the BE or CB junctions on a BJT would give you a diode, but the diodes are there whether or not one junction is shorted.
From searching around, I've found that shorting the drain and source together and using them as one terminal, along with the gate as the other makes a capacitor. Am I correct in assuming the oxide capacitance/area (C_ox) multiplied by the area (width and length of the device) determines its capacitance? What other factors are involved?
For a MOSFET, gate capacitance will be given in the datasheet. Generally, it's
$$\small C = \frac{\epsilon A}{d}$$
But most datasheets won't give you the gate area, dielectric constant, or gate oxide thickness.

Also nagging me: how does current flow through the device if the gate oxide is an insulator?
This might help: http://www.electronics-tutorials.ws/capacitor/cap_1.html

Last edited:

#### shortbus

Joined Sep 30, 2009
7,942
Won't the "gate capacitor" also be very limited in both farads and voltage? A very ineffective capacitor if it would work at all.

#### takao21203

Joined Apr 28, 2012
3,695
Won't the "gate capacitor" also be very limited in both farads and voltage? A very ineffective capacitor if it would work at all.
Of course it works at certain frequency. A capacitor is a capacitor.

#### WBahn

Joined Mar 31, 2012
26,155
If you're talking about MOSFETs, you don't get a diode by connecting drain to gate. There's a parasitic diode between the source and drain.
I sure used a LOT of diode-connected MOSFETs in the ASICs I've designed.

https://en.wikipedia.org/wiki/Diode-connected_transistor

http://web.eecs.utk.edu/~jhollema/classes/ece336/lectures/slides06.pdf

http://web.mit.edu/6.012/www/SP07-L25.pdf

It's diode characteristic is most definitely NOT due to the parasitic diodes -- you almost never want to forward bias those.

For a MOSFET, gate capacitance will be given in the datasheet. Generally, it's
$$\small C = \frac{\epsilon A}{d}$$
But most datasheets won't give you the gate area, dielectric constant, or gate oxide thickness.
And a big reason for that is that gate capacitance is HIGHLY non-linear with voltage -- so that equation does you almost no good for a FET capacitor.

#### WBahn

Joined Mar 31, 2012
26,155
Won't the "gate capacitor" also be very limited in both farads and voltage? A very ineffective capacitor if it would work at all.
For a discrete transistor this is probably true since you are comparing it to discrete capacitors. But on an IC the FET capacitors almost always provide WAY more capacitance per unit area than any other structure can provide precisely because they use thin oxide instead of field oxide.