August 13th, 2020
A major portion of the costs of the current c-Si photovoltaic technology are due to the wasteful usage of expensive high – purity silicon. One possibility for reducing wafer costs consists of simply using thinner and therefore more readily breakable wafers. In principle, it is already possible to reduce the wafer thickness to 150 pm and thus to obtain more wafers from a cast ingot and reduce the cost per wafer, if the efficiency and the production yield (breakage) remain constant. This is however currently not the case. Furthermore, more sawdust’ would be produced, so that the percentage of wasted silicon would increase.
Another, more elegant method is offered by the use of ribbon silicon. Silicon ribbons are crystalline wafers which are pulled directly from the melt with the required thickness of about 200 pm. Ribbon silicon wafers have the great advantage as compared to wafers cut from ingots that almost all of the material in the silicon melt can be used to produce the crystalline wafers. The lack of waste sawdust’ gives a significant reduction in the fraction of wafer costs within the overall cost of the modules. In addition, a second cost factor in the conventional process can be eliminated completely, namely the crystallization of the silicon ingots. With the condition that the efficiency of the cells made from ribbon-silicon wafers be just as high as that of cells made from wafers sawed out of ingots, up to 22 % of the module costs can be saved by using ribbon-silicon wafers (see Fig. 2). This would be the case for so-called RGS silicon, which we will describe below.
Another important factor is the superior usage of silicon. In the past, the photovoltaic industry was able to fill its needs by using the excess production of silicon for micro-electronics. The strong growth of photovoltaics in recent times has, however, had the result that solar-power applications now use notably more silicon than the integrated-circuit producers. Conserving silicon material is thus an additional advantage in the use of silicon ribbon, if one is
concerned about maintaining the high growth rates of the photovoltaic industry.
Over the years, many different production technologies for ribbon silicon have been tried out. These technologies can be reasonably classified in terms of the meniscus formed by the Si melt at the phase boundary between the liquid and the solid phase (see Fig. 3). In the case of meniscus type M1, the base area of the meniscus is defined by a mold into which the liquid Si climbs up above the surrounding surface due to capillary action. The wafer is pulled upwards out of the melt, at a pulling rate of around 1 to 2 cm/min; this is the main factor determining the thickness of the wafer. The heat of crystallization is carried off mainly by radiation, while convection hardly plays a role. Therefore, the rate of crystallization is relatively low and this, in turn, limits the pulling rates. During the pulling, the temperature gradient at the liquid-solid boundary must be controlled to within 1° C, i. e. with a very high precision from the technological point of view. This technique was commercialized as early as 1994 under the name Edge-defined Film-fed Growth’ (EFG) and was being further developed up to 2009 by the Wacker Schott Solar GmbH company in Alzenau (Franconia) . The silicon is pulled from the melt in the form of 5 m long tubes. A graphite mold gives it an octagonal shape with very thin walls. This closed shape avoids free edges which would have to be stabilized. A laser then cuts the 12.5 cm wide faces of the octagon into square or rectangular wafers.
The meniscus shape of type M2 has a broader base than type M1. It results when the wafer is pulled vertically upwards directly out of the melt. Owing to the longer meniscus, this type of silicon ribbon preparation can tolerate a greater fluctuation in the temperature gradient at the liquid-solid phase boundary, of around 10°C, which permits the use of more compact and less expensive processing equipment. An example of this type is the string-ribbon silicon developed by Evergreen Solar Inc. and commercialized since 2001. In this process, two fibers (strings’) made of a material which is kept secret by the producer are passed through the Si melt and pulled parallel and vertically upwards from the liquid surface (see in – fobox Internet’ on p. 45). This process makes use of the fact that silicon has a high surface tension, even greater than that of mercury. Thus a silicon film stretches between the two strings,
which are about 8 cm apart, like a soap-bubble film, and it solidifies to give a ribbon. A laser then cuts this ribbon into wafers of the desired size.
In this likewise vertical pulling method, the pulling velocity is limited for the same physical reasons as in the EFG process to 1-2 cm/min.
Considerably higher pulling velocities are possible if the wafer is pulled horizontally instead of vertically from the melt. This is done using the extended meniscus shape of type M3. For horizontal pulling, a substrate can be used on which the silicon solidifies. This is e. g. the case in the Ribbon Growth on Substrate (RGS) silicon process. This method was originated by Bayer AG, and is still in the developmental stage. A belt carrying substrate plates moves under the crucible containing the silicon melt. The silicon is deposited onto the plates. As soon as the wafers have crystallized, they detach themselves from the substrate due to the difference in thermal expansion coefficients, so that the substrate plates are again freed up for the next pass. This process is distinguished by rapid heat dissipation through the substrate plates. Furthermore, the direction of crystallization is decoupled from the horizontal pulling direction; it runs up vertically from the cooler face of the substrate to the upper surface of the wafers. Both these effects permit high pulling velocities up to 10 cm/s, i. e. a 30- to 60-fold higher throughput than in the vertical methods. This enormously reduces the cost of wafer production.