imec lifts the hood on its record-setting PERT solar PV cell

Belgian research center imec has produced a 21.5% efficient solar PV cell, using passivated emitter, rear totally diffused (PERT) PV technology. This is a new record for this type of cell on an industrial, full-sized wafer, and was verified by Fraunhofer ISE CalLab

pv magazine spoke with imec Scientific Director of Photovoltaics Jef Poortmans at Intersolar North America about this achievement, and why the institute is looking at PERT technology.

pv magazine: I’m familiar with the work that imec has done with passivated emitter rear contact (PERC) cells. Can you explain the difference between PERC and PERT?

Poortmans: imec has aso been working on PERC for 10 years, and we stopped development a half year ago. The first difference is that the PERC design that we used has always been on p-type silicon substrates. And the PERT is based on an n-type silicon substrate.

So why did we change from p-type to n-type? First of all because we are convinced that research will go to ever higher efficiencies to reach the theoretical limits of crystalline silicon, but it will be more straightforward to do that with n-type material. Most implementations by other groups put the emitter at the front using boron diffusion. That is pretty disruptive compared to what people do at this moment in their production lines, where they diffuse phosphorous at the front side, not boron. We moved the emitter to the rear side, and in this way we can keep many of the assets of the normal PERC process.

The boron emitter is made at the rear side, it is passivated by a dielectric, exactly like the PERC, and we open the dialectrical on both sides by means of laser ablation – excactly as we do for the PERC. We try for the rest to keep the process as much the same as what we do on the PERC.

There is another advantage to putting the emitter on the rear side. We are working on nickel-copper metallization, which we have proven to be reliable in our tests, but reliability is further enhanced by putting the emitter on the rear side and putting copper contacts on the front side.

All of these steps together allowed to reach 21.5% efficiency. But the story did not end there. We think we can reach 22% and maybe even slightly higher, by combining these steps with a selective emitter and a selective front surface field. But that is a different stack process.

pv magazine: What is the potential for commercializiation?

Poortmans: That is very strongly linked with the availability of n-type substrates. Many roadmaps including ITRPV predict that because we are moving to higher efficiencies, n-type silicon will be more and more preferred because it is easier to passivate and because it is more forgiving towards a number of contaminants like iron. n-type silicon is now around 5% of the total market, but we will gradually move to a situation where n-type silicon will be 30-40% of the market 5-10 years from now.

It is also for us, as an R&D institute, quite logical that we make this move towards n-type substrates. We want to be ahead of what the industry is doing.

pv magazine: How specialized is the machinery to make PERT cells as compared to PERC cells?

Poortmans: The systems that you use are essentially the same as the ones that you use for the PERC. You use phosphorous diffusion furnaces, we make copper contacts in the plating. The only difference is of course that you have to make a boron diffused emitter. And boron diffusion is a process which is characterized by a relatively large thermal budget and the removal of the BSG glass requires some adaptations of the process.

Therefore we are looking growing an epitaxial B-doped emitter on the rear side, and have already tried this approach. Our first attempts there, we achieved large-area efficiencies between 20 and 21%. But you see, the philosophy behind our move to this specific n-type structure was to keep things as much as possible the same. Because as this moment it is not easy to convince companies to make investments in new equipment. Because they first have to recover from the crisis which they have been suffering from in previous years.

pv magazine: Yes, everyone needs designs that are relatively simple and cheap.

Poortmans: There is one additional element I would like to add. In our press release we refer to the results which we have achieved by aluminum oxide surface passivation. We compared that with silicon oxide surface passivation, and we see that the aluminum oxide surface passivation gives us an initial boost of 0.3%, and the behavior under low illumination is also better.

But you could say of course that atomic layer deposition (ALD) of aluminum oxide is not yet implemented on a very large scale, but we are pretty sure that this will happen due to the spatial ALD system. ALD is equipment whereby substrates are coated very efficiently and rapidly by an atomic layer aluminum oxide. It should not be a big hurdle to commercialize this technology.

pv magazine: You mentioned earlier that you see more potential with PERT than with PERC. Is that because of the theoretical limits with p-type silicon or another factor?

Poortmans: The main thing which is driving us in the direction of n-type is the intrinsic property of n-type silicon surfaces to be more easily passivated than p-type silicon surfaces. Of course, at this moment an n-type silicon substrate is still somewhat more expensive than a p-type, but we expect that this difference will become smaller and smaller over time when more n-type silicon is being produced and the efficiency gain allows anyway some price difference.

We have made p-type PERC cells with efficiencies up to 21%, and we think that the n-PERT has an efficiency potential between 22 and 23%. So let’s say, roughly speaking, with n-type silicon you gain about 1% over p-type.