矽晶切割廢泥的回收與多晶矽生長在太陽光電之應用研究

Abstract

矽晶型太陽能電池占整個太陽能電池市場的90%以上，而太陽能級矽原料主要來源於西門子法製程(Siemens process)，然而近幾年太陽能產業的迅速發展，導致原物矽突然間的大量缺乏，因而衝擊到太陽能產業的發展。為了產業的正面發展與降低太陽能發電成本，低成本的原物料以及耗損物料的回收，或是高效能太陽能電池的生產，將扮演重要的角色。 以西門子法製程生產太陽能級矽原料，成本約在40 $/kg左右，而矽晶棒在晶片切割步驟時，大約有40%的矽損失(kerf loss)，這些矽都以污泥的方式被丟棄或是經過回收碳化矽粒子之後丟棄，在矽成本如此高之下，將切割矽泥丟棄而不設法回收是非常可惜的。過去未能回收切割矽泥中的矽，主要原因是因為碳化矽與矽的分離十分困難。本論文中針對攙硼之太陽能單晶矽切割時所產生之矽污泥進行矽原料之分離提純，吾人發展出一個新的製程方法—高溫熱處理，可以解決過去碳化矽與矽分離困難的問題，成功的將矽由切割矽泥中分離出來，並提純至接近太陽能級矽原料的規格標準。研究中使用回收矽生長出來的多晶矽，雜質總含量低於5 ppmw，電阻率(0.5~1.6 Ω-cm)可控制在市售晶片標準範圍內(0.5~10 Ω-cm)，平均載子壽命超過1 μs已達到太陽能電池製作所需的規格要求。晶體經過切片製作成太陽能電池後光電轉換效率最高可達12.6%，接近吾人使用乾淨矽原料來製作的太陽能電池效率(13.2%)，證實切割矽泥回收再利用的構想是可行的。本研究的矽泥回收程序的矽總回收率可達20 %，估計成本約32 $/kg，不過目前仍有磷汙染的問題尚待解決，假使未來可以將切割矽泥生產端的磷汙染源排除，矽泥回收程序的總回收率將可提升到40%，且矽泥回收的成本可以降低到16 $/kg，相較起西門子法的製造成本來說，深具發展的潛力。 對於多晶矽太陽能電池之應用，由於晶界對於太陽能電池的能量轉換效率有一定的影響，因此過去有文獻報導利用改變降溫速度或坩堝下降速度，來得到具有Σ3鈍性晶界之多晶矽，來降低多晶晶界對太陽能轉換效率的影響，然而這些方法由於受限於生長設備的尺寸，目前只成功運用在3吋以下的小型生長系統上。本論文首次發展出cooling spot技術，來進行多晶矽生長時之晶向控制。利用cooling spot晶向控制生長所得到之多晶矽，晶體底部的晶粒晶向可以受到控制，晶體底部不但晶粒大小可以增加之外，載子壽命也有明顯的提升。此外將cooling spot配合上坩堝速度的控制，更可以大幅度增加生長晶體的品質，晶體底部的載子壽命可由原先未控制的0.41 μs，在生長控制後增加到1.08 μs，製作成太陽能電池後的量子效率更是可以大幅度增加。本研究將cooling spot實際運用在大型多晶生長設備上，初步實驗證實可得到與小型實驗相同的趨勢，生長出來的晶體底部平均晶粒大小可由1.25 cm增加到1.75 cm，載子壽命也可從0.44 μs增加到0.63 μs。

Solar photovoltaic (PV) energy will shortly be in great demand, since it is inexhaustible and cleaner than any conventional energy resources. The global PV production was over 2.6 GW in 2006 alone, out of which, the majority was the silicon wafer-based solar cells. Because the fast growing PV market is mainly based on crystalline silicon, the lack of silicon raw materials in recent years is becoming a critical issue. The exponential growth of the solar cell industry has driven up the price of silicon several times and the shortage of silicon has been a serious issue since 2003. So far, most of solar grade silicon (SoG-Si) is from the Siemens process, which is energy intensive and high cost. Therefore, cheaper routes for producing SoG-Si are being developed in the PV industries, but the progress in slow. The price of raw silicon has increased almost ten times for the past five years. Therefore, seeking a feasible and low-cost route for producing SoG-Si is important for the PV industry. On the other hand, because there is about 40% kerf silicon loss during slicing processing, the recycle of the silicon from the cutting slurry could save the raw material used significantly. However, the attempt has not yet been succeeded so far. In this thesis, we developed a novel approach to recover SoG-Si from the Si/SiC mixture obtained from the cutting slurry waste. Herein a detailed discussion of the processes involved in the whole recycle research is reported. The average resistivity and minority carrier lifetime of the grown crystals from recycled silicon were found to be about 0.7 Ω-cm and 1.02 μs, respectively, which were close to the original sawing silicon ingots. Solar cells using multi-crystalline wafers of recovered silicon were fabricated and the best energy conversion efficiency (12.6%) was comparable to the ones from the high-purity silicon (13.2%). The yield of recycle process is about 20% and the cost (32 $/kg) is much lower than the right now price of solar grade silicon. Multi-crystalline silicon (mc-Si) grown by the directional solidification method has attracted world wide attention as a solar cell material because of its low production cost and high throughput. However, the efficiency of mc-Si solar cells is usually lower than single-crystalline silicon solar cells because of the presence of variety of defects such as randomly oriented grain boundaries, dislocations, inclusions and oxides, in large concentration. These defects act as recombination center for light generated electrons and holes and therefore are harmful for the solar cell performance. Recently, grain boundaries have been shown to have different characteristics under different orientations. For example, the Σ3 boundary is a well-known inactive boundary as it does not act as recombination center. Therefore, in order to improve the efficiency of mc-Si based solar cells, efforts have been directed towards growth of mc-Si with highly oriented grains. In fact, controlling the cooling speed during initial stage of solidification induces dendrite growth along the crucible bottom wall which is responsible for growth of large grains with Σ3 grain boundary. In this thesis, we report a novel method to control the grain orientation by using cooling spots at the bottom of the crucible. The cooling spots induce locally larger radial thermal gradients to enhance dendritic growth, but not limited to the size of the crucible. Studies have been carried out for laboratory scale, and some preliminary results have been obtained from an industrial scale experiment. The minority carrier lifetime measurements and electron back scattering pattern analysis in the grown crystals have been carried out to assess the effect of using cooling spot.