Growing A Wafer From A Wafer

One way of really revolutionizing a silicon cell is to change the wafering process, where over half of the silicon is lost as kerf. The solution are so-called kerf-less technologies. A few start-ups have been working on different approaches for years with a varied degree of success – but recently there have been promising news. Next to US company 1366, which is currently building a commercial factory in New York State, German startup NexWafe GmbH recently completed its Series A funding.

NexWafe is a spin-off of renowned research institute Fraunhofer Institute of Solar Energy Systems (ISE) and has licensed ISE's epitaxy technology, on which the research institute has been working for over 15 years.

NextWafe's plans

In simple terms, a thick crystalline silicon layer is epitaxially deposited on a reusable template. Then the fully grown wafer is detached from the seed wafer. Most importantly, NexWafe is aiming to produce wafers of standard thickness, 150 to 180 µm, which means they would be a drop-in replacement for traditional CZ wafers. "Cell manufactures don't have to change their process when shifting to our wafers," says Stefan Reber, NexWafe's CEO and former head of Crystalline Silicon – Materials and Thin Film Solar Cells at ISE.

In the first step, the startup is aiming at n-type wafer segment, which Reber characterizes as premium and high efficiency market. These silicon slices of new class are comparable to the standard wafers in quality with encouraging first results. In September, 2015, NexWafe published that it reached an efficiency of 20% with solar cells based on epitaxially grown silicon substrates. The key element of commercializing the epitaxy approach is the production scale APCVD reactor, called ProConCVD, developed by ISE. The research center has also seed funded the project, while NexWafe announced on March 7, 2016 that it completed a €6 million series A financing round with private equity firm Lynwood (Schweiz) AG.

Processing details

NexWafe's technology is colloquially known as "lift-off" approach, which consists of four main steps – forming a porous silicon layer, reorganization, epitaxy and lift-off. In the first step, a thick Czochralski silicon substrate is electrochemically porosified in such a way that a low porosity layer is formed atop of two high porosity buried layers. These porous silicon wafers are subjected to a high temperature annealing step, called reorganization. This thermal treatment step, accomplished at about 1,100 to 1,300 °C, helps the low porosity top layer to restructure into a closed and continuous surface. The reorganization step has a significant effect on the crystal quality, because the restructured top layer of the substrate acts as a template for the subsequent epitaxial deposition of the silicon. If the layer is not sufficiently closed or when the surface is smooth, it leads to growth defects in the epitaxial layer. In epitaxy, the reorganized surface of the porous silicon, serving as seed layer, is grown to a required thickness by supplying silicon source gas into the deposition chamber. The deposited layer can be thin, in case of which it is bonded onto a low cost substrate. Alternately, it is thickened to a level that is high enough to be a self standing wafer, which is the case with NexWafe. When the desired thickness is reached, the buried layer with larger cavities, due to its weak mechanical properties, facilitates detachment of the epitaxial layer from the substrate. The bottom part of the substrate is again subjected to porosification. Choosing this path, the substrate repeatedly serves as a seeding wafer.

What kind of equipment is needed?

As for the implementation part, most of the research work accomplished at ISE has sourced the porous silicon samples from the Institute for Microelectronics in Stuttgart. However, Reber intends to have everything in house. "You have to have the whole value chain installed, if you want to do production," he says. The reorganization and epitaxy are accomplished in a highly modular ProConCVD system, which stands for Production Continuous Chemical Vapor Deposition. According to a technical paper by Reber (Advances In Equipment And Process Development For High-Throughput Continuous Silicon Epitaxy; Authors: Stefan Reber et al.) presented at the 27th European Photovoltaic Solar Energy Conference (EU PVSEC) in 2011, the ProCon is built based on an Atmospheric Pressure Chemical Vapor Deposition platform and the RTCVD160, a lab-type CVD processor for silicon  deposition on large areas, which has been used for research at ISE for many years. The ProCon has three tracks and each of the tracks is formed by means of substrates mounted to the two vertically parallel carriers. These carriers are designed to hold 9 wafers of 156 mm side length, and the tracks are guided through the reactors by top and bottom rails. A deposition chamber is formed by two separated graphite blocks – one with gas inlets to inject process gases and the other with gas outlet to extract. The loading and unloading ends of the reactor are provided with gas curtains to isolate the processing environment from the ambiance. According to an online spec sheet, the tool has 5 m2 total deposition area with a 2 m long stable temperature zone, accounting to a total footprint of 16 x 4 m. It has a rated power of 360 kW and supports maximum processing temperature of 1,300 °C realized with resistance heating. The substrates are transported continuously along the side walls of the chambers with a maximum speed of 12 m per hour, so that they are coated continuously in an inline fashion.

While the processing details of the reactor are not available, the particulars provided for the RTCVD160 furnace given in technical paper from ISE at the 31st EU PVSEC in 2015 (N-Type And P-Type Silicon Foils Fabricated In A Quasi-Inline Epi Reactor With Bulk Lifetimes Exceeding 500 µs; Authors: Stefan Jenz et al.) provides an indication. The reorganization processes was accomplished at 1,150 °C under 100% hydrogen atmosphere and epitaxy was also done at the same temperature, but in a trichlorosilane (TCS) / hydrogen atmosphere with a chlorine / hydrogen gas flow ratio of 10%. As for the throughput, Reber says, it always depends on the thickness. The online brochure puts it at about 1,200 wafers per hour, referring to a deposited layer thickness of 1 to 50 µm. For thicker wafers of 150 to 180 µm, such as the ones NexWafe intends to produce, the throughput decreases accordingly. However, Reber would not specify the exact number. On the other hand, the ProCon reactor is based on highly modular design, which enables scaling up the production capacity according to requirement. This is realized in several ways – by adding deposition tracks, increasing carrier height and/or deposition chamber length.  

Deposition and growth of epitaxial emitter

In addition to just depositing the silicon mass,  ProCOn as well as its previous version can also be employed for growing epitaxial emitter. For this the reactors equipped with two sequential chambers through which the substrate passes through. Even though the substrates are in one furnace, they are completely separated gas wise. This enables to run completely different processes, but at same temperature. That means, after growing the wafer, the same tool can also be used to grow epitaxial emitters. In simple terms, when the wafer comes out of the ProConCVD, a P-N junction is formed already. Diborane is used as the source of boron doping and phosphene is used as the source of  phosphorus impurity. However, NexWafe is mainly focusing on making wafers with base doping, like standard commercial wafers.

After the deposition, the epitaxial layer is separated from the base. Reber says his company is ready with a solution, but reveals only that the tool relies on "mechanical force". A research paper from Institute for Solar Energy Research Hamelin (ISFH) published in 2013 in the proceedings of the Silicon PV conference (Lift-Off Of Free-Standing Layers In  the Kerfless Porous Silicon Process; Authors: Sarah Kajari-Schröder et al.), describes a specially designed vacuum chuck successfully employed for the  lift-off process. After the epitaxial layer is separated, the base wafer is sent back to porosification and the cycle repeats, so the seed wafer can be reused several times. Reber confirms that the substrate can be reused several dozens of times, though the number mainly depends on the economic optimum.

Expansion targets

The economic optimum for NexWafe's technology, however, will only be exactly known when the company enters the mass production. The first round of funding is intended to build a pilot line with a capacity to supply the necessary quantity required for wafer qualification by its partners – cell and module makers. Without providing details, Reber says that the funds from the second round would be used to build a factory of around 250 MW, and then very quickly scale up to around 1 GW. If everything turns out well, a further growth is also possible.

About half the cost of traditional technology

As for the capital costs, Raber says that the capital expenditures required for this new factory are dramatically less than for today's incumbent process, taking into account production of polysilicon to crystallization and wafering. "It is much less than half," estimates Reber. On top comes the low-space requirement and ease in scalability. Exact calculations on manufacturing costs are sketchy at his stage, but Reber anticipates wafers production cost at NexWafe are about half of the costs of CZ. "Thus, we have the possibility to incentivize customers as well as save good margins for us to grow the business further," says Reber.

Next steps for NexWafe

At the beginning, NexWafe will focus on n-type, as Reber considers it as a fast growing premium market with high potential and larger opportunity. At a later stage, the company would also expand to p-type. The n-type wafers produced from the epi process showed mean minority carrier lifetimes are above 1,000 μs, indicating the same quality as n-type CZ wafers. The small 2  x 2 cm solar cells made from these wafers yielded 20% efficiency, corresponding to 39.6 mA/cm2 short circuit current density – the latter is a world record value for epitaxially grown silicon solar cells, according to the company. The good news is that NexWafe was able to attract financing for its Series A round despite today's low silicon prices. Getting enough money for the second round, which targets commercial manufacturing, will be very likely  much more difficult, especially in Europe. The old Continent, however, could take a look at how America is supporting its solar startups.  The state of New York in late 2015 offered another kerfless startup, 1366, a very attractive incentive package for its 250 MW factory – $56 million plus low-cost electricity supply.