Your browser is not supported. Please update it.

7 June 2023

NexWafe wafer deposition slashes electricity when making solar

This week we spoke to Davor Sutija, CEO of NexWafe, a German company which uniquely manufactures photovoltaic silicon wafers in a vapor deposition process. In the context of the Inflation Reduction Act and equivalent European Union policies, the 60-employee company is now able to commercialize, with its first products to hit the market within 24 months. Shortly before our interview the company announced it had raised $32.1 million for construction of its first commercial facility at an industrial site in Bitterfeld south of Berlin – in the “Solar Valley” which has lain mostly abandoned since China took over the industry a decade ago.

That region is also a favoured location for other companies reviving Western manufacturing including Q Cells, Meyer Burger, and potentially Oxford PV. Funding for NexWafe’s facility came primarily from shareholders Reliance New Energy which is a subsidiary of India’s Reliance Industries, along with Saudi Aramco, and ATHOS, while another notable investor is former Australia Prime Minister Malcolm Turnbull.

Sutija has been involved in photovoltaics since the late 1990s, starting by founding a silicon ingot manufacturing company in north Norway. Three decades later, the industry uses the same techniques, just more efficiently with larger wafers. The Fluidized Bed Reactor (FBR), a form of polysilicon production which requires 1/3rd as much energy, but which can only ever play an auxiliary role in the industry, was also pioneered way back then – by a company which Sutija’s first manufacturer was eventually merged into.

NexWafe’s technique – Epitaxy – is completely different. In electronics, deposition of wafers is common and that is more or less what NexWafe has achieved for photovoltaic semiconductors on the back of years of R&D. NexWafe goes directly from the gas phase to a solid single-crystal silicon wafer, deposited on an identical substrate seed wafer. This involves introducing the right gases – the same chlorosilane gases ordinarily used in the Siemens Process – here used in a Chemical Vapor Deposition (CVD) reaction to create the wafer directly.

This replaces the three furthest upstream steps of the mainstream solar supply chain – polysilicon, ingot and wafer-slicing – with a single process. NexWafe claims to reduce energy consumption during wafer manufacturing by 70%. Those two processes operate at 1200°C and 1425°C respectively, forming around one-third of those steps’ total cost, which has been enough to push almost all of the global polysilicon and ingot-pulling solar industry in China’s inland rural provinces where electricity is cheapest.

The subsidies offered to each step of solar manufacturing in the US under the Inflation Reduction Act are generally around half of the cost of production observed in the mainstream Chinese industry. But in NexWafe’s case, Sutija claims, the subsidy ($0.62 per wafer) will be HIGHER than the production cost. NexWafe is setting up a subsidiary for America.

Low energy consumption also makes Nexwafe’s product more green than the competition. Today, with so much coal and gas still to be replaced worldwide, this is of questionable relevance in itself – deploying solar panels which were produced using Chinese coal-fired electricity is still a step in the right direction. But Sutija makes the interesting claim that solar manufacturing will account for 3% of all emissions from now through to 2050. “If we can reduce wafer energy consumption by 70%, then we can reduce overall solar industry emissions by a third, and reduce total global emissions from all sources by 1% through 2050.” We would’ve guessed that cutting out the Siemens Process and Czochralski Process would reduce total emissions by 50%, but Sutija only claims “a third, maybe more” to be on the safe side, considering the complexities of comparison.

The main appeal here is cost reduction, but there are also several qualitative improvements from switching to epitaxy. It’s worth describing the technology in detail. In NexWafe’s process the seed wafer is removed via a chemical step, anodic dissolution of silicon, which creates a porous structure sealed at the top to recreate the single-crystal temple, which becomes fragile enough that very minor shear force separates the seed from the produced wafer, allowing reuse of the wafer.

Asked how a single wafer is deposited at a time, Sutija explains “The trick is to make sure that the surface where the seed wafer lies is at the right temperature while the rest of the reactor is not at the temperature which allows for deposition. So you do have to control deposition, and it grows at a limited rate – it takes hours to form a wafer.”

Sutija continues to explain how NexWafe’s epitaxy differs from the mainstream Czochralski ingot-pulling method. “Czochralski takes hours to pull an ingot, but the point is we’re size-independent.” In the past couple of years the Czochralski process has had to adopt larger crucibles, growing from 26-inch to 28-inch to 32, 36 and finally 40-inch – constantly changing equipment. “They have to change the pull recipe, they have to specialize, they have to re-engineer their machine park with new tools and crystal pullers,” Sutija adds. “But our susceptors, our carriers, are far larger than a single wafer – we produce an array containing wafers. Whether the industry decides on M10 or G12 size, for us it doesn’t matter, we’re depositing on a much larger surface than either.”

The other major advantage of epitaxy, compared to pulling and slicing, is thickness. Mainstream wafer thickness has declined from 175μm to 155μm over the past eighteen months or so (Sutija explains that leading producers are heading to130μm on TOPCon and 110μm for heterojunction while the mainstream is still 160μm) and is continuing to decline, saving on silicon expenditure. The thinnest Si PV product we’ve seen announced is 90μm, with several at 110μm and the new normal for heterojunction being 130μm, 150μm for TOPCon right now. Mainstream techniques can serve this demand but pay a penalty – KERF loss. When cutting with a diamond wire, about 40μm is lost per cut.

There’s also saw damage etch removal which takes off a further 7.5μm from each side – so 55μm thickness is lost in total from cutting which for a 110μm wafer is an extra 50% on top of what makes it through to the finished product. And 110μm is the important benchmark, because “for heterojunction the performance improves thanks to a voltage improvement if you can go thinner, people expect 110μm by the end of the year – but saw damage and KERF removal will make them harder to handle (thinner means more fragile)” – and the losses dilute the polysilicon savings of going thinner. But for NexWafe, transitioning to thinner wafers simply means less deposition, which is cheaper. The thinnest wafer produced under laboratory conditions by NexWafe is 51μm, while the company regularly produces between 90μm and 130μm – it’s already comfortably at the lowest end of any announcement from the industry, able to serve the full spectrum of customers’ needs across TOPCon and heterojunction in terms of both wafer size and wafer thickness. Incidentally, Sutija believes that 90μm is for now too fragile for automated cell lines – breakage rates are too high already, only making it a target in future as stated, Sutija says, by several manufacturers at conferences he has attended.

Another qualitative advantage with epitaxy are lower impurities – there is no source of iron and little oxygen in the system, whereas the quartz crucible in the Czochralski process is silicon dioxide. Inevitably, when heated as much as it is, it does let off a small amount of impurities into the end product. “If you look at recent conferences, N-type – we’re talking specifically N-type monocrystalline wafers, these are phosphorous-doped – even these seem to be showing rings associated with oxygen levels which reduce cell efficiency by 0.5%. So NexWafe’s product is a premium product that makes higher-efficiency modules.

As for the market situation – Sutija says he is very excited about the Inflation Reduction Act and the European Solar Industry Alliance which are reshoring manufacturing back to the West. And this is understandable since wafers are the most concentrated supply chain segment, fully 97% according to an oft-cited statistic which Sutija also quotes (and which is probably now 98% after the past 12 months of Chinese expansions). Wafers will be the last supply chain segment to reshore since cells and module factories are more easily developed, while polysilicon never fully went away and still exports 7,000 tons (2.5 GW) a month to China. So NexWafe has a special, competitive offering and still plenty of time to seize market share. In particular with the Siemens and Czochralski processes removed, an epitaxy-based solar supply chain will have around half the CAPEX costs, compared to polysilicon-ingot-wafer-cell-module.

Low CAPEX is exactly what any business wants when it is getting started. Sutija predicts his company can reach 5% global market share in 2030 of a $30 billion global market for wafers, getting there via alliances with larger companies such as Reliance Industries.

Sutija also mentions the geopolitics – if the industry is left in the hands of China this is tens of billions of dollars, even a couple of hundred billion, being paid to the country each year for the next few decades for solar modules alone. “Our offering is technological not political, but we’re strategic on a global level to rebalance supply chains.”

NexWafe’s production facility is in Freiburg, Germany, near Fraunhofer ISE, and was spun out of their facilities eight years ago. “Fraunhofer developed this process for 15 years, we’ve licensed some background IP and developed our own patents, we have a total of 19 patent families and 30+ IPRs – trade secrets and things on their way to becoming patents. We are now planning, after completion of the prototype development, to receive planning permission for the full-scale (250 MW) facility in Bitterfeld in July and start construction and excavation immediately, with production to start in early 2025.”

Sutija expects that anyone else trying to muscle in and copy NexWafe with their own epitaxial wafer production will not be able to catch up quickly. For NexWafe, the main R&D challenge was adapting from the few microns thickness of electronics-oriented CVD to the 100 micron thickness of photovoltaics, while controlling thickness variation, parasitic deposition and crystal uniformity.

Asked what the cost structure is like for NexWafe, Sutija explains that the Chemical Vapor Deposition (CVD) reactors are the main CAPEX cost – overall there’s the cleaning step, electrochemical formation of porous layer, then CVD process, and mechanical process for separation (this one is cheap), and finally etching the seed wafer to ready it for reuse. Operational expenditure is mostly owed to chemicals for the CVD reaction and the wet electrochemical etching steps – chlorosilane, with NexWafe planning to use globally available tetrachlorosilane used in fibreoptics and silicones, rather than trichlorosilane which is mostly available in and for China’s polysilicon industry.

One last facet to NexWafe’s technology is that it could be used to make a Gallium Arsenide – Si tandem down the line. “I’m not a sceptic on perovskites, they will come to market, but impedance matching in a tandem is a challenge given different rates of degradation over the two semiconductors, and there’s bandgap overlap between perovskite and silicon.” says Sutija. “If you’re going to go all the way to the mid-30s for module efficiency, one way to do that will be with a very thin layer of gallium arsenide, perhaps using a type of epitaxy, on top of silicon – not any time soon, maybe in a decade’s time.”