Development of Solar Grade (SoG) Silicon

Joyce, David B ; Schmid, Frederick

Crystal Systems, Inc., 27 Congress Street, Salem, MA 01970, PH 978 745 0088, ext. 303

关键词: Solar Grade Silicon;    Shortages;    Liners;    Segregation;    Vacuum Refining;   

DOI  :  10.2172/922203
RP-ID  :  DOE/ER/83928-F
RP-ID  :  FG02-04ER83928
RP-ID  :  922203

美国|英语

来源: UNT Digital Library

【 摘 要 】

The rapid growth of the photovoltaics (PV) industry is threatened by the ongoing shortage of suitable solar grade (SoG) silicon. Until 2004, the PV industry relied on the off spec polysilicon from the electronics industry for feedstock. The rapid growth of PV meant that the demand for SoG silicon predictably surpassed this supply. The long-term prospects for PV are very bright as costs have come down, and efficiencies and economies of scale make PV generated electricity ever more competitive with grid electricity. However, the scalability of the current process for producing poly silicon again threatens the future. A less costly, higher volume production technique is needed to supply the long-term growth of the PV industry, and to reduce costs of PV even further. This long-term need was the motivation behind this SBIR proposal. Upgrading metallurgical grade (MG) silicon would fulfill the need for a low-cost, large-scale production. Past attempts to upgrade MG silicon have foundered/failed/had trouble reducing the low segregation coefficient elements, B, P, and Al. Most other elements in MG silicon can be purified very efficiently by directional solidification. Thus, in the Phase I program, Crystal Systems proposed a variety of techniques to reduce B, P, and Al in MG silicon to produce a low cost commercial technique for upgrading MG silicon. Of the variety of techniques tried, vacuum refining and some slagging and additions turned out to be the most promising. These were pursued in the Phase II study. By vacuum refining, the P was reduced from 14 to 0.22 ppmw and the Al was reduced from 370 ppmw to 0.065 ppmw. This process was scaled to 40 kg scale charges, and the results were expressed in terms of half-life, or time to reduce the impurity concentration in half. Best half-lives were 2 hours, typical were 4 hours. Scaling factors were developed to allow prediction of these results to larger scale melts. The vacuum refining required the development of new crucibles, as well as liners and coatings to allow the vacuum to be achieved. These developments also hold the promise of lower cost ingot growth, because several of these developments led to a reusable crucible. Liners and coatings were tested on 37 runs, under a variety of conditions. Although many of these did not fulfill the requirements of the program, several were very successful, particularly in allowing the crucible to be reused several times. The most interesting result was with slags and additives used to reduce P and Al. Although slags have been much studied with little success in removing P and B effectively, certain modeling suggested a particular type of slagging might be effective. This was tried, and found to be highly effective for P and surprisingly effective for B, as well. The best results indicate that > 99% of the P was removed, and > 75% of the B was removed by a slagging treatment. An operability issue involving separation of the slag and silicon was the final technical problem preventing the full-scale use of this technique, and there has been progress on this front. A slagging/additive technique is highly promising, because the rates of equilibration are very high, and this is a rapid technique that scales very well to large volumes with little increase in time. Materials of containment and slag/metal separation are issues that are continuing to be developed.