MESA+ University of Twente
Research Business & Innovation About MESA+ Education

Svetlana Bystrova (promotion date: 1 December 2004)

Diffusion Barriers for Copper Metallization

Promotion Date: 01-12-2004

Svetlana Bystrova

Our aim was to make a material, which can serve as a diffusion barrier against Cu. Such a material should be suitable for a fabrication of chips with Cu metallisation. The copper metallisation can be used to make chips faster. Also dimensions of interconnects can be reduced and this is important for a scaling of devices. Thus, aluminium used conventionally in chips industry is to be replaced with copper. But now another problem arises: copper is very reactive with silicon. To prevent this reaction and also to prevent diffusion of copper into working areas of a chip (which results in a failure of the chip), it is necessary to have some kind of barrier.

What is your thesis about?

Our aim was to make a material, which can serve as a diffusion barrier against Cu. Such a material should be suitable for a fabrication of chips with Cu metallisation. The copper metallisation can be used to make chips faster. Also dimensions of interconnects can be reduced and this is important for a scaling of devices. Thus, aluminium used conventionally in chips industry is to be replaced with copper. But now another problem arises: copper is very reactive with silicon. To prevent this reaction and also to prevent diffusion of copper into working areas of a chip (which results in a failure of the chip), it is necessary to have some kind of barrier. We made a barrier of tungsten nitride using Atomic Layer Deposition. In order to estimate how the diffusion barrier works we tested this material in different devices (diodes, capacitors etc.)

I take it that the outcome of your research is very important for industry?

Yes, it is.

How did you start your research?

At first I tried Chemical Vapour Deposition (CVD). The reactive chamber was very big and the deposition rate very high. In general we could deposit compounds, these are tungsten nitride containing silicon and tungsten silicide containing nitrogen. But further tests showed that these materials are not suitable to be diffusion barriers. Tungsten nitride had a very high resistance. Tungsten silicide was reactive with copper. So it turned out that this line of approach was not going to work.

And how did you feel about that?

I was very frustrated. I did so many experiments and all the time a piece of equipment broke down. It took up a lot of time. Once we even got a fire in the machine, which was really scary.

Everything seemed to happen at the same time, the smoke, the fire and the alarm, which went on immediately. So we ran. I looked back and it was really a big fire, I was absolutely shocked. Fortunately my colleague had closed off the SiH4 supply – which is the biggest cause of fire because it can ignite by itself – and the fire died down.

So you have been working with dangerous stuff?

Yes, but that applies to everybody who works in the clean room. You should always take the proper precautions. And also the clean room is full of detectors, for fire and gases and so on. When the alarm switches on there is one instruction only that everybody must follow immediately. That is ‘run’

When the (CVD) process did not work as you expected, what made you think of another line of research?

It was clear that the studied process did not work, but our aim was a working diffusion barrier. Of course we discussed everything on a regular basis. That time I had two years more to finish my project. So we decided this period was enough to start with something new. In literature there were some indications how to reduce the deposition rate. Our choice was a new process; that is Atomic Layer Deposition. It had become clear that with the old CVD we would not reach our goal, that is why my group had to build a new machine for Atomic Layer Deposition. In the mean time I continued doing a research in Philips testing other materials. We heard something about the promising properties of tungsten carbide- nitride, and tried to deposit this material with different precursors. We made our fist try with a very cheap precursor and got a satisfactory result.

So they are very pleased at Philips.

They are. No one did this before, we are really the first. And we made excellent barriers.

What are your future plans?

I am looking for a job in this region. My husband works at this university and his parents leave nearby. So we would like to stay here. Who knows, perhaps with one of the Mesa spin off companies…

For the summary of the thesis, click here.