Wanted: Simulators that Predict Process Variation Across a Wafer

Excel has a correlation data analysis feature that allows you to make predictions, but it requires input of the feature you are trying to predict based on existing process data. A word of advice, it is best to have values below and above the range you are trying to predict, assuming that the relationship is somewhat linear.

If it is not, then you will need to perhaps consider average results, which are always normally distributed. You can then predict what percentage of your process is within points a and b on the distribution with simple mathematical calculations.

What about the etching are you measuring? It's depth, clarity, etc.?

Thanks very much JC and the other guest for your comments.

Well the etching I referred to was Dry etching or Reactive Ion etching (RIE) process.

The RIE process is done with the help of plasma generated in an etching chamber where we want to etch the surface of a wafer. The surface of a wafer can be a layer of deposited film where we want to etch parts of it off in order to create patterns.

In real life, the plasma generated in the chamber is not evenly distributed and so there is variation of the etch rate across the whole surface of the wafer.

This causes the dimension and thickness of the deposited film to be varied. Therefore in order to predict the etch rate, the input of the simulation would be the RF power to generate plasma, the pressure in the chamber and etc.

The simulation tool should be able to calculate the plasma distribution hence the etch rate (how fast the plasme etches the film, nm/second) across the surface of a wafer.

Most of the softwares in the market now are more to do with the prediction of the process results on a single device and it does not account for the process variation due to unevenly distributed plasma concentration in the chamber.

Ideally I want to have a tool that can tell me the etch rate distribution across wafer when I input power, pressure, gas as in the real tool for at least on a bare flat 6"-8" rounded wafer.

Thanks again for your valuable time and kind comments,

Appreciated,

Regards,

Wutthinan

Hi,

how do you think you can simulate a process where you do not know the conditions of the process.

If your pressure is high (mbar or higher) you should measure the local velocities as you operate in a viscous flow.

If the pressure is below 1 mbar (sputtering and similar) the power distribution across the target is much more important.

Get the different existing probes (Langmuir) and measure the the plasma density and energies, also electron energy and - if existing - play with the magnetic fields to achieve homogeneity.

If your dry etch and RIE etch process is HF excited it would be useful to feed the gas in various ways through the exciting region and make the exciter as good as possible.

This is process optimisation, takes some time, I can help you with in depth discussion,
experimental verification is to be done with your equipment and your process.

Maybe you need another exciter and another gas feed.

RHABE

Crystal Ball is a good simulation package that works in conjunction w/ Excel. You need process data (mean, std dev, etc..), though, to build the foundation of the Monte Carlo. Not sure if this software would work in your application, but having worked as an I.E. I can attest it works very well in business process / financial modeling.

Thanks very much, I put more explanation in the earlier reply.

Appreciated your kind comment,

Regards,

Wutthinan

I'm looking for a substitute for the old package ARENA Process Modeling.

Any sugestion will be usefull. Have you the web site or e-mail of Cristal Ball?

Thanks