Gøran Bye is heading consultancy firm Advanced Materials Management Solutions, LLC (AMMS). He is a former CEO of REC Silicon, a leader in FBR technology, and LDK Silicon & Chemical Technologies, a once promising Chinese newcomer in the field of silicon production, which fell victim to its huge expansion plans. Today Bye is advising businesses and investors on PV and silicon materials industries. At the time of the interview he was interim Chief Project Officer for Qatar Solar Technologies, which announced to started production in March 2017.
FBR technology has been a proprietary technology a few companies are producing commercially using their own equipment, that's why only little know-how has spread to the outer world so far. Long-time seen by many as a challenger to incumbent Siemens-technology, FBR has remained mostly an unfulfilled promises – with big announcements that were not realized, joint ventures, bankruptcy and acquisitions. While it is difficult for an outsider to estimate in which direction the technology is moving, there is one expert in this segment who should know. For our Market Survey on Silicon CVD Reactors (see Market Survey Polysilicon CVD Reactors), TaiyangNews' Head of Technology Shravan K. Chunduri talked to silicon expert Gøran Bye
TaiyangNews: FBR technology has been a hot topic in the silicon space. What is your take on the technology?
Gøran Bye: The concept of using FBR is inviting people. There are two reasons. One is power usage in the reactor. That is immediately tantalizing. I am not entirely convinced that people have clearly realized that you need more energy for silane production. In principle OK, you use a little bit more energy for silane production, which is usually used for FBR, but if you look at entire FBR, one should save in cooling water and the off gas and vent gas recovery compared to TCS Siemens. At the end, it is large energy savings for silane FBR. Then it is possible to have close to 100% conversion of silane gas in the reactor in one run, while for TCS we are looking at 11%, 12% or maybe 16%.
TaiyangNews: Then why it is not really taking off? ITRPV has been consistently lowering its forecast for FBR?
Gøran Bye: One of the main reasons is that using silane as precursor results in the formation of dust, a fine particulate of silicon. The art of realizing full conversion also involves minimizing dust generation. This requires a very careful balance of temperatures and flows. Some people also find it difficult to achieve this balance. Then there are some limitations in terms of freedom to operate FBR; there are patents that are still valid.
TaiyangNews: What is so complicated about the technology?
Gøran Bye: As a matter fact, FBR itself is not very sophisticated. It is used in a lot of different processes around the world. You have this bed of seeds; you agitate by gas. When you have to make polysilicon with this, it has to deal with where do you heat it, how do you heat it. Because you have a hot wall technology – and if you heat from outside, there is a tendency of silane gas adhering to the inside of the reactor, leading to unwanted deposition on the wall. This results in higher metal-contamination. So one challenge is to keeping metals out of reach from the product. That is not so easy because if there is a hiccup in the production, the wall scales would fall down and end up in the product.
Then it is also a matter of how to extract the product – to suck the beads out or they would fallout at the bottom? In any case, the beads have to go through a series of filters, degassers and coolers. Here again, you have a possibility of contaminating the silicon with metallic impurities, which results in a large surface area with metallic impurities. The other challenge is the possibility of porosity in the beads, attract some oxygen.
So FBR is not as easy as it sounds. It might look bright from a financial and costs point of view, but there are a lot of things that are not necessarily so positive with FBR.
TaiyangNews: Are there issues with FBR beyond technology?
Gøran Bye: The last point I want to make is indeed not technical and my personal opinion. The Chinese industry seems to be very reluctant to use silane gas. They are somehow afraid of silane gas. True, silane is highly pyrophoric and we have examples of its explosion, but TCS and STC are not benign either. They all need to be treated with care. It seems like a couple of explosions and accidents going back quite a few years are still making the Chinese industry worry of using silane. I think they are more comfortable to sticking to the technology they already know. So I think that's the reason why many who have been looking at silane FBR, have decided not to do it. It even might have been difficult in getting financing. However, GCL is in the latter phases of taking over SunEdison's Solar Material Division with the IP for SunEdison's FBR – so let's see.
TaiyangNews: What about the limitations on the patents end?
Gøran Bye: I recall the old MEMC patents. One claim was that they had distribution plates with a lot of holes in it, and they seem to be using hydrogen in the outer holes in order to create some sort of hydrogen blanket around the wall to keep the beads away from hitting it. That is something you are not allowed to do.
There is also potentially a limitation to have the distribution plate with a large number of inlet-holes. REC Silicon also has patents around hydrogen and silane gas inlets into the FBR. Another area to maneuver around is putting some sort of liner. If you keep a liner to prevent beads hitting the walls, you may have other challenges. The thermal expansion coefficient of whatever material you are using might be different from the reactor itself. This might result in some heat stress. Then there is also the challenge of how to get heat in there and the options are getting heat from outside to heat through the liner or install internal heaters in and combinations of these. This adds to further complexities. For example, opting for internal heaters leads to the question of how to prevent silane depositing on these heaters. Here are also some limitations on what kind of liners you use, because there are some patents out there. In a nutshell the crucial questions are – How do you get the gas in, what's the distribution you have there? And how to keep it away from hitting the wall. These all have to be handled.
TaiyangNews: So what can you do despite the patent limitations?
Gøran Bye: To have a freedom to operate and proceed further, one must make some compromises – accept that the beads hit the wall and allow the gas to pass this film or layer of silicon. It is not the optimum, but in order to be legal, one has to let go some things.
You need to learn how to operate a FBR reactor. Even if you have fixed all the issues we talked about and accepted that they are as they are, operation is the key. How do you balance the velocity, concentrations, dust production, conversion, and not at least how do you keep the bed relatively stable when you extract the beads? There is an art to that, which is both art and science. And it takes time to learn. That is the big advantage of REC and MEMC. They had time to learn this. Others like GCL have been trying to catch-up, which could now go faster after the take-over of the SunEdison IP and technical resources. It is not unreasonable to think that it might take 5 to 6 years to learn to master a new technology.
TaiyanNews: What is your opinion about REC's latest venture in China?
Gøran Bye: One thing from REC is that they are very open and try to be very good in communication to the market. They have developed what they call FBR B, the second generation. It is larger than the ones they have now; it has some sort of liner in it. I think that's the key. In Moses Lake, REC has a product development facility, where they implement developments in pilot production. It can be done in pretty large scale in order to test and see how to operate. And I think FBR B is the technology that they want to build in China. Before they came up with the China plans, they announced an expansion project in Moses Lake that they since have abandoned. It was called reactor 25 and 26, which was going to be FBR B. Apparently they have had very good results on what they did on the pilot scale. Maybe it increases risk a little that they will go straight to that technology in China without having been able to test reactor 25 and 26 in Moses Lake on a commercial scale. But I am pretty sure they will succeed.
TaiyangNews: Do you think REC will finish the China project as announced in their Q3 report, they say that they are on track for startup in 2nd half 2017?
Gøran Bye: They were very open and specific – produce 7,300 tons in 2018, going up to 16,300 tons in 2019 and culminating it at 19,300 tons in 2020. They say that they are ready for startup in 2nd half 2017 and would be having half the production output in 2018. One should count backwards to estimate the feasibility. There will be a commissioning period. Wacker in Tennessee announced that their commissioning period was 8 months, though it depends how you define commissioning. Considering REC will have 6 months commissioning period, in order to start up in December 2017, construction should be finishing in June 2017. It might be a stretch to get there. However, as I said, they were very specific in their projections, so I assume they have a good plan.
TaiyangNews: What is your take on SMP, the joint venture of Samsung and SunEdison?
Gøran Bye: SMP is doing high pressure FBR in Korea. To increase production, you need to increase the size of the reactor, and/or you can increase the pressure within certain limits. SunEdison has been the leader in FBR technology. They have more than 30 years of experience with FBR. When they started this JV with Samsung, they were building up on a strong knowhow in FBR. I wouldn't be surprised, but I don't know for sure that they already had the design for this kind of reactor that they have installed in Korea in their R&D. So they have started with lot of knowledge, but not so much practical usage yet. They decided to go with large reactors, and increased the pressure in order to increase the capacity of the FBR. According to people who know a lot about fluidization, they all seem to think that increasing the pressure is not trivial but doable. I have reason to believe that they have been actually more successful than what people think on the outside.
TaiyangNews: Increasing the pressure seems to increase the dust, which is the biggest problem SMP has, right?
Gøran Bye: They report that they have dust and fines. Dust come in different sizes, they produce 12% or less. But look at REC's reports, they are not any better. Yes, there have been lots of teething problems in Korea. But I think the rumors out there have been more negative than a sober view on the realities would show. When someone starts something, we want it to be perfect in a couple of weeks. FBR simply needs time. Just look at REC, hence my comment to GCL; they are in a learning process.
TaiyangNews: Then why does their US plant have very high cost?
Gøran Bye: Their Pasadena plant is closed. There are two important aspects for FBR or any polysilicon production. Gas production is at the front-end and CVD on back-end. SunEdison was using its own silane technology based on silicon tetrafluoride, while REC was using the Union Carbide method. The reason for why costs were so high for SunEdison in Pasadena is allegedly that the costs of the raw material was prohibitively high. That's one thing, and then to get rid of the byproduct, so apparently the silane was the big problem cost-wise. However, in its Korean JV, the company is using the Union Carbide method for producing silane. Another of the center aspects of this method is using hydrochlorination reactors for reforming STC back to TCS ("cold conversion") rather than the hot converters, something that could be an advantage for the TCS- Siemens-based producers as well. Hydrochlorination was finally embraced in China. It is more or less an open technology. In fact, GTAT has delivered the front-end for SMP. Yingli had a small silane plant using silane for Siemens that was also using silicon tetrafluoride, similar to SunEdison's method. But they had significant problems because of the back- end. I think the SunEdison fluidized bed technology in Pasadena was probable better than it appears when you look at costs. Importantly, the plant was originally mainly supplying to the electronic industry.
TaiyangNews: Will Centrotherm's new development bring any solace to FBR technology?
Gøran Bye: It is based on vibration more than gas stream. I don't know it can solve the problem. You need to have pipes to extract the product, you need to keep it away from the wall, and you need to operate it well. Again, operation is the key to FBR. If you look at REC's earlier announcements, they have been operating their reactors form any days and many months at a time. That's the key to have uninterrupted smooth operation. I am not sure if the vibration-based product can do that, but I do not know any technical details of the reactor.
TaiyangNews: Are there particular issues for FBR to ramp up from R&D to production?
Gøran Bye: The more you scale up the worse it is. There is no doubt about it. But that is general and it is not specific with FBR. I don't think it is harder to scale up a FBR compared to Siemens technology. But there are so many more people out there that have experience with Siemens reactors, particularly TCS. And that's why there exists much more knowhow on Siemens technology. And: FBR is still a proprietary technology and not yet available commercially-off-the-shelf from third-party technology providers.