Thickness of oxide in MOS using capacitance-voltage?

[lcr2139]

Hello, I am looking to use capacitance and voltage calculations to calculate the thickness of a thin oxide of a MOS capacitor. Here is what I think I should do:
1. grow poly on oxide on silicon
2. hook up substrate to ground and poly to a potential using a voltage source.
3. Use an LCR meter to measure capacitance.
-Do this for different voltages from -5V to 5V (or voltage ranging from accumulation to depletion to inversion), with small steps in between.
4. Generate a C/V graph and calculate the thickness using small-signal capacitance and voltage equations pertinent to MOS capacitors.
Is this a correct way to measure a thin oxide thickness? What changes can I make to make this procedure better? How thick should the poly be?

 

[analogdesign]

What is your level of expertise? It sounds like you are planning to fabricate a MOS capacitor on a silicon wafer in a semiconductor processing lab, is that correct?

When you say thickness, do you mean physical thickness or electrical thickness? Because there is a limit to how much you can dope the poly, it will have a depletion region which will add to the effective (or "electrical") thickness. Also, there will likely be at least some depletion of the substrate that you will also have to take into account. Here is a free chapter on MOS capacitors from Prof. Chenming Hu at UC Berkeley. It discusses the difference between electrical and physical thickness and how to measure and interpret CV curves. https://people.eecs.berkeley.edu/~hu/Chenming-Hu_ch5.pdf

If what you want is electrical thickness, then what you describe is kind of workable but probably too crude for good results due to a couple of second-order effects. First off, I wouldn't use an LCR meter. A very sensitive ammeter is better because we're talking about really small capacitances here. Also, typically a semiconductor lab with have some Semiconductor Parameter Analyzers and those machines will make doing the CV measurement much easier (which is a very tricky thing to do yourself).

If what you want is physical thickness, depending on how thick the oxide is you're measuring this is usually done either by measuring the tunneling current through the oxide or if the oxide is thick enough it can be done optically using interferometry.

If you have a really thin oxide (and you probably don't because I suspect you're not a professional with access to a state-of-the-art facility) then you can't use CV measurements by you can use tunneling measurements. Here is an excellent paper that discusses the technique: https://web.stanford.edu/class/ee311/NOTES/GateOx_Gupta.pdf

If you have a relatively thick oxide you can use interferometry. The basic idea is because the oxides are transparent, they will scatter light differently at the poly-oxide and oxide-substrate interfaces and you can use this interference pattern to figure out the thickness of the oxide. There are machines that can do this for you. Where I work, a technician just puts a wafer into a machine and it measures the thin film thickness automatically.

Also, be careful about what voltages you use. It will take a bit of guess and check but be sure you know the substrate doping so you can calculate the breakdown voltage of your wafer.

Lastly, the poly thickness isn't super critical, but you certainly want it thick enough that a) it was a low resistance, and b) is stable enough not to get crushed when you attach a probe for measurement.