Long read: UNSW’s Martin Green sets out solar’s accelerating technological development

pv magazine: What do you make of the technological landscape of PV manufacturing in 2018?

Martin Green: I think things are changing quite quickly, compared to the history of PV where things have moved at a snail’s pace. 2018 has been an exciting year with manufacturers switching to more and more busbars and the PERC uptake continuing to be so strong and so rapid. I think there has been big changes over the last five years due to this transition.

Every trade show I go to, it is interesting to see what kind of products that companies are putting out and what type of cell technology they are using, and the details as to how they are putting their modules together.

Do you have a sense then that technology cycles are accelerating within solar?

Technology is accelerating compared to what we were seeing earlier in the industry. The industry has become much more competitive and manufacturers are looking for a technological edge – and that is accelerating the uptake of new technology. The transition to PERC has happened a lot more quickly than I could have imagined, and that is a good example.

And your involvement with PERC goes back quite some time doesn’t it?

I drew my first PERC cell in 1983, so that’s when it was invented I think.

There are some degradation issues that have been seen in PERC cells in particular, but also in a range of other cell technologies. UNSW has worked closely on these effects with a range of industry partners. Do you get the feeling that some of these degradation effects are being better understood and are on the way to being resolved?

I think the issues are well understood within our group and within companies that are collaborating with us on the technology. But there are some quite subtle effects that can occur with hydrogen in all types of silicon material – including n-type, as well as multi, and mono p-type.

I’m not sure all manufacturers are across the complete range of issues that can arise from hydrogenation. But certainly our team has a very good understanding of it.

Leading some of that work at UNSW was Stuart Wenham. His passing late last year was saddening to many within PV. What impact has his death had on the UNSW solar community and the work he was leading?

It was very sad with Stuart passing so suddenly. He had been planning to step back from his academic responsibilities and had been grooming his team so that they were able to operate with less involvement from him. That has stood his team in very good stead in being able to continue unabated. In fact, we had a lot of success in a funding round at the end of last year, which was supported by the Australian Government through the Australian Renewable Energy Agency (ARENA). This funding has really stimulated a lot of work with the hydrogenation technology and understanding of the complete range of issues that go with that technology.

The announcement of that funding round by ARENA was made at the UNSW Solar Industrial Research Facility (SIRS). What can you tell me about that facility?

It is a pilot line for cell production. There are two types of commercial activities on that line within the group. We have a number of manufacturers testing their hydrogenation equipment on that pilot line as well as being able to develop processes at commercial scale on that line.

To operate the line and utilize those facilities to their full extent requires substantial funding. This makes the recent round helpful in being able to push that technology along at a rate that will be of the most benefit to the industry.

Looking further ahead to next generation or step-change technologies. You have been working on tandem PV cells for some years. What is the latest status of that work?

I think a tandem on silicon is the logical way for silicon to progress. There has been some very good research results on III-V semiconductors on silicon, but I think that this approach is always going to be way too expensive for practical use. There is also a lot of interest in perovskites in silicon, but unfortunately it is looking like it is going to be very challenging to get them to the level of stability to which purchasers of solar modules have become accustomed with a silicon product.

We are earnestly looking for additional materials to stack onto silicon. Last year we filed patents on a whole new family of semiconductors. We invented 13 new families of inorganic compound semiconductors that had not been previously identified as semiconductors.

They are very similar to the CZTS material in which there is a lot of interest, but the CZTS technology has proven difficult to get to high efficiency. But we do have these 13 new families and many members of those families involve abundant, non-toxic elements. Our aim is to explore through those families as quickly as we can to see if there is a material there that has all the advantages of CZTS, in terms of non-toxicity and material availability, but is easier to work with. So, more like perovskites, in terms of being able to get beyond 20% efficiency quickly – which is what is needed for a tandem cell.

These are much more traditional semiconductors than the perovskites, but they are families have been overlooked in the past. With ab-initio computer simulations we are able to show that they are semiconductors, but now we have to synthesise a few to prove that they are viable. We will be publishing these results shortly and then other research groups can get involved to push the technology along more quickly than we would be able to do by ourselves.

You mention the 20% efficiency baseline. What kinds of efficiencies do you anticipate could be realized through tandem cell structures?

The 20% I mentioned is the efficiency that is needed for the top cell, to add to the silicon base. So, the top cell needs to be a thin film technology that can get over 20%. There are only seven or eight technologies that have gone over 20% for solar conversion efficiency so far – and most of those are III-V semiconductors.

The three best known thin-films are cadmium telluride (CdTe), CIGS and now perovskites that have achieved over 20% in thin film form. But, we will need to find more compounds that can do that.

Once there is the thin film that is over 20%, and that technology is tuned in, total efficiency equals this plus half of the efficiency of the silicon cell. If the silicon cell is 20%+, and half of that is added to the 20% of the top cell – this gives over 30% efficiency. This will represent a substantial jump over even the most expensive silicon technologies, such as back contact cells.

There is also the potential, particularly with these new families of semiconductor, for stacking more and more thin film cells on top of silicon to progressively improve the performance.

The PERC based tandem cell will be a little bit a like a heterojunction cell in that it will require contacting the thin film cells with such things as transparent conducting oxides. There will be a big boost in efficiency, but some associated costs related to these oxides. But when the second cell is added, then there is no need to add anything extra in that line. The second cell won’t give a particularly big efficiency boost, but it will be very inexpensive to add compared to adding the first cell – so things even out.

I can see that once the first cell on silicon is perfected, then there could be a very rapid transition to two cells on silicon, because the extra performance of that second cell would come at very low additional cost or associated technology development. I can therefore see a natural progression to two-cells on silicon, and then you are looking at efficiencies in the mid-30% range for that type of cell. So, I can see a natural evolution from where we are now to something well over 30% efficiency. But, first of all we have to find the right semiconductors to stack onto silicon.

Why is that you see CIGS or CdTe as being poor candidates for this application?

There are a couple of issues. One is the band gap: the CdTe band gap is incorrect for stacking on silicon. Then there are issues with the toxicity of cadmium. If the silicon manufacturers are going to get the lead out of their paste it doesn’t make much sense to add cadmium as one of the cell components.

The other issue is that all of the tellurium in the world comes from copper refining. Once that supply is exhausted, then tellurium becomes as scarce as gold – so if it was to mined, then the costs will be similar to gold. That makes it very difficult to increase the supply of tellurium beyond what can be delivered from copper processing. According to my calculations, this would limit CdTe production over the long term to less than 30 GW a year. I came to that conclusion about ten years ago and nothing has occurred in the interim to make me change my views on that.

There are similar problems with CIGS because of indium. Once pressure comes onto indium supply, then the price will go through the roof because the flat panel industry can afford to pay a lot more for it.

That is why, at the scale the industry is going to grow to, there is a need to look for materials in which there are no issues with supply. That is why CZTS, copper zinc tin sulphide, is attractive. Each of those elements are available in much larger quantities. The industry would have to grow to an enormous size before there were concerns about supply of those particular elements.

What kind of deposition do you envisage for the thin film cell?

What we have seen with evolution of solar technology is that the cost of processing the cell is now an insignificant fraction of the total cost of the panel. That wasn’t always the case. And adding the thin film layer is the kind of thing that is amendable to the type of cost reduction that we see in microelectronics – as an ongoing technology evolution with increased scale that will automatically cause those costs to reduce.

It would be a bit like experience with the PERC cell: Initially the additional processing costs would be significant but, as the process would become more integrated into production, the extra costs of doing something a bit more sophisticated would become relatively insignificant. I would be expecting that it would be possible to produce, on a cost-per-watt basis on the module level, at lower cost than a conventional product eventually – as a result of the improved efficiency and the difficulty in decreasing other module costs such as frames, glass, junction boxes, and the like.

The improved efficiency could come at lower cost even on the module level and then there would be the additional benefits at the system level.

And you remain unconvinced of perovskites, despite the continuing work from a big number of researchers and some young companies?

The stability issues are quite challenging. There are all kind of instabilities that must be addressed. Some of them relate to the chemical instability, sensitivity to moisture and oxygen and so on – which can probably be protected against, with additional encapsulation expense.

Along with the chemical instability, is the instability due to movement of the ions within the material. These ions are quite mobile: the components of the cell are actually moving around when the cell is operating under different conditions. The very mobile ions might represent an inherent problem with the technology.

Thermodynamic simulations additionally reveal that the compounds are on the borderline of instability – so they would rather be at a different state than as a perovskite compound. There is a fundamental issue with that kind of thermodynamic instability.

Finally, there is a range of light instabilities. It is possible to protect the device from moisture, to pass the moisture testing expected of modules. But there is instability in light exposure tests, because light actually can interfere with the stability of the material itself. This would appear to be the most difficult test for perovskites to pass, because light sensitivity is a problem for something that is going to be deployed in the field for 25 years. Indeed, at the moment 25 years is the expectation, but over time the expectations of PV module lifetimes would increase as the understanding of how to better encapsulate modules increases.

Silicon is a tough competitor for perovskites. It is very hard to see a silicon manufacturer adopting a product that is more efficient but doesn’t have the same stability as its baseline product – because no manufacturer wants to get a bad reputation regarding stability, in fact it could be fatal. In short, it is hard to see how a less stable product can make it to the mainstream market.