Aluminium Oxide as a dielectric

**tommyers** · Jun23-06, 03:48 PM

Hi,

I have been doing some experimentation with parallel plate capacitors, see http://www.physicsforums.com/showthread.php?t=124150

For one arrangement of aluminium plates (air as the dielectric) I was expecting the measured capacitance to be approx. 7.9pF and what I measured was 65pF. Do you think that aluminium oxide (dielectric 9.1) was adding to the air dielectric of 1.

So instead of my calculations using 1 I should have used 9.1 which would have brought my theory nearer my practise?

Regards

Tom

**lalbatros** · Jun23-06, 04:15 PM

Hello Tom,

Assume that there are three dielectric layers {1,2,3} in between two conducting plates of area A.
Then the capacity of this system is given by:

$LaTeX Code: C = A \\frac{1}{\\frac{d_1}{\\epsilon_1}+\\frac{d_2}{\\epsilo n_2}+\\frac{d_3}{\\epsilon_3}}$

From your explanation, I don't see clearly the geometry you consider, but I guess it is a series of layers, isn't it? If this is correct, then this formula applies. It shows clearly that the capacity can indeed vary between that for air and that for Al2O3 depending simply on the thickness of the layers. I don't know the thickness of air and Al2O3 you consider, but if air is much thicker than the Al2O3 layers, the capacity would be dominated by the air.

Hope it helps,

Michel

**tommyers** · Jun24-06, 12:50 PM

Hi Michel,

The geometry is basically two parallel aluminium plates separated by air. I was then wondering if the surface of the plates had an oxide (aluminium oxide, 9.1) on them, this could be thought of as the dielectric.

Regards

Tom

**Bystander** · Jun24-06, 04:18 PM

How are you doing the measurement?

**lalbatros** · Jun24-06, 04:43 PM

Hi tom,

Clearly the thin oxide layer on your plate has a negligible effect.
You should investigate parisitic capacities in your setup.

Remember that coaxial cable connecting your meter to your device could bring a large additional capacity (say more than 60pF/m). This geometry is the easiest to calculate. But if you try parallel wires you will find it is not much better. Keeping the wires far apart can only bring a limited improvement (look at the log in the formula). Keeping them short is certainly a good idea (10 cm could still bring 6pF !!!).

So, you are left to answer some essential questions:

how can you reduce the parasitic capacities
how could you eventually compensate them by a special arrangement
(zeroing the parasitic capacity somehow)
how you could reliably correct for them by some calibration (small corrections ...)
could you increase the tested capacity to reduce the parasitic effects
how did the fathers in the 19th century perform these measurements

I have no answer right now for you. But I think this is basic electrical metrology and you could find the answers is appropriate handbooks. Let us know about your findings. I will do it too if I find something.

Michel

**tommyers** · Jun25-06, 07:50 AM

Hi,

For measuring the capacitance of the arrangment I am using a commercially avaliable capacitance meter, see http://www.farnell.com/datasheets/24173.pdf

This has a 'zero-adjust' which allows my to 'zero' out leads between connecting them from the meter to the plates. On the 200pF range setting which I have the meter set for these experiments, an 820Hz excitation voltage is produce by the meter, but this should be of no concern to me right? because the meter takes all this into account when it produces the reading?

This should answer 'bystanders' question:
How are you doing the measurement?

and hopefully the first two of michels

how can you reduce the parasitic capacities
how could you eventually compensate them by a special arrangement
(zeroing the parasitic capacity somehow)

I will continue to look at various measurement techniques.
Any further help in resolving the matter will be greatly appreciated.

Regards

Tom

**lalbatros** · Jun25-06, 10:09 AM

Tom,

You should be absolutely sure that your meter and your method are reliable. Make some tests with a known capacity in situations (cables, connectors, ...) as close as possible to your setup. If such tests are not reliable, try other devices and methods.

Additionaly, I also express some doubt on the possiblity to cancel properly parasitic capacities that are much larger than the capacity you would try to measure. Therefore, you should certainly minimize these unwanted capacities as much as possible. For those parasitic capacities that you cannot remove, you should investigate for the best possible cancellation technique: I doubt that the built-in zeroing mechanism of your meter could eliminate accurately 50 pF when you try to measure 1 pF. I am confident that this is possible.

Michel

**tommyers** · Jun25-06, 01:06 PM

The zeroing mechanism is in the form of an adjustment knob which is manipulated by the user.

Tom

**lalbatros** · Jul24-06, 03:52 PM

Tom,

The following link explains some practical problems when measuring capacities as you were:

http://www.gyogyitokezek.hu/fe/capresist.htm

Maybe this will help you improve your experiment,

Michel

**TGarzarella** · Jul26-06, 12:43 PM

You're probably measuring parasitic capacitance from your cables/leads. Testing a known, commercial capacitance (e.g., Radio Shack) that has a similar C is a good idea.

I've measured parasitic C as high as 50pF in my high speed coaxs. It will help if you make the leads as short as possible, and remove as much stray metal/wires/etc. as possible.