摘要
在製作太陽能晶片的過程中，矽晶棒需經過線切割的程序才能得到矽晶片，而線切割程序所使用的材料就包含有矽晶棒、不鏽鋼線鋸、含碳化矽粒子之聚乙二醇研磨漿液。晶棒經過切片後，有大約30-40%矽原料會以粉體的方式隨著研磨漿液流失，同時不鏽鋼金屬線也會因研磨而形成粉末流至漿液中，因此研磨後的漿液中包含的物質有矽、碳化矽、金屬粒子與聚乙二醇水溶液。以往此漿液多以廢棄物處理，目前已回收碳化矽及乙二醇，如能再回收其中之矽粉，勢必能降低太陽能晶片的製造成本。 故本研究的目的是在常溫下使用離心法與粒子相轉移法回收切割矽泥中的矽粉。起始矽泥分別經過丙酮清洗聚乙二醇水溶液與濃硝酸溶解金屬粒子，並氧化矽粉之表面而形成二氧化矽層，接著乾燥後的粉體即為實驗起始物。從初步的實驗分析得知矽與碳化矽粉的粒徑分佈範圍分別介於0.4-8μm與0.4-30μm之間，因此依據粉體的粒徑大小，本研究分別採用離心法與粒子相轉移法進行矽與碳化矽粉之分離。 離心法是藉由添加一密度介於矽(2.33 g/m3)與碳化矽(3.23 g/cm3)之間的液體至粉體中，經混合後再離心，離心所得的上層矽粉經清洗乾燥後即為產物。實驗探討的變數有固含量、液體密度、離心次數、離心時間、超音波攪拌時間以及界面活性劑濃度。從實驗的結果發現液體密度與超音波攪拌時間對於分離效果有顯著的影響，而且添加界面活性劑則有助於離心分離。然而由於離心力對於次微米碳化矽粒子的作用較小，所以分離的效果有限，最終產物的矽純度最高僅93 wt%且產率低於50%。 基於離心法在分離次微米粉體上的限制，吾人轉而研究並採用粒子相轉移法。粒子相轉移法的原理主要是利用粉體表面親疏水性質的作用力並配合重力與粒子界面之靜電力來進行分離。其方法是先將粉體分散於水中，藉由添加另一不互溶之油相至水相中，再經由混合攪拌使疏水性的粒子進入油相中而達到分離的目的。粒子相轉移法首先會與離心法配合，分離去除離心法所無法克服的次微米碳化矽粒子，以回收得到高純度矽粉，實驗曾探討油相種類、油/水體積比與六偏磷酸鈉濃度的影響。接著本研究更發展出兩階段的粒子相轉移法。第一階段所使用的油相密度大於水，目的在於分離數微米以上的碳化矽顆粒；第二階段則使用密度小於水的油相，目的是為了去除次微米至數微米之碳化矽粒子。實驗所探討的變數包括水相pH值、固含量、油/水體積比與分離次數。從結果發現以正丁醇作為油相的分離效果最佳，而且pH值會影響粉體與油滴間的界面電位，使pH值在等電位點附近有明顯的分離效果。最後經兩階段粒子相轉移法分離得到的矽粉產物純度可達99.0 wt%，且總產率仍有71.0%。本研究同時也對於不同含矽量的矽泥進行研究，結果發現粒子相轉移法並不受限於含矽量的多寡，因此粒子相轉移法的確可以從切割矽泥中回收高純度的矽粉。
並列摘要
The wire-saw process, in which the materials involved are silicon ingot, stainless wire, abrasive slurry containing silicon carbide particles and polyethylene glycol solution, is used to slice the silicon ingot into wafers for solar cell use. After the sawing stage, 30-40% silicon loss and metal fragments caused by attrition were carried by the used slurry, which usually was treated as the waste. At the present time, polyethylene glycol and silicon carbide have been recovered and recycled for abrasive slurry. If the loss silicon can be recovered either from the waste slurry, the shortage of feedstock and high cost of manufacturing wafers in the photovoltaic industry will be overcome. The objective of this research is to recover the high-purity silicon powder from the kerf loss slurry using the centrifugation and particle phase-transfer method at ambient temperature. The starting material, which consisted of silicon and silicon carbide with a particle size distribution of 0.4-8 μm and 0.4-30 μm respectively, was obtained by washing the waste slurry to remove the polyethylene glycol solution and metal fragments using acetone and nitric acid respectively. In the washing step using nitric acid, the surface of silicon particle was oxidized to form a silicon dioxide layer. In this work, the centrifugation was applied first by introducing a heavy-liquid with a density between that of silicon (2.33 g/cm3) and silicon carbide (3.23 g/cm3) into the starting material. After the mixing and centrifuging steps, the upper layer silicon cake was cleaned and dried to obtain the silicon-rich product. Several operating variables, including the solid volume concentration, heavy-liquid density, number of centrifugation, centrifugation time, mixing time of sonication, and the concentration of surfactant were investigated in this study. The results showed that the effects of heavy-liquid density, mixing time of sonication, and concentration of surfactant on the separation of particles were significant. However, the centrifugal force acted on the submicron SiC particles was insignificant so that the highest purity of Si-rich product was 93 wt% with a recovery lower than 50%. Due to the limitation of centrifugation for separating submicron particles, the particle phase-transfer (PPT) method, which is based on the difference in the surface properties between particles, i.e., hydrophobicity versus hydrophilicity, together with the gravity and electrostatic force on particle surface, was investigated and applied to recover the silicon powder. At the beginning, the starting material was dispersed into water, and then an immiscible oil was added into it. After shaking or stirring, the hydrophobic particles would transfer into the oil phase and thus the particles with different surface properties were separated. In this work, the particle phase-transfer method was first used to remove the submicron SiC particles remaining in the Si-rich product from the centrifugation separation so that a high-purity silicon powder was obtained. Several variables were investigated, including type of oil phase, oil/water volume ratio, and concentration of sodium hexametaphosphate. Moreover, a two-stage of particle phase-transfer method was developed: in the first stage an oil with a density higher than 1 g/cm3 was used to separate the micron-size SiC particles, and in the second stage another oil with a density lower than 1 g/cm3 was applied to remove the particles of submicron and a few microns. The investigated operating variables included pH value of water phase, solid volume concentration, oil/water volume ratio, and number of separating stage. As far as the oil phase was concerned, the n-butanol showed the best results among the solvents tested. As to the pH value, which affects the zeta potential, its effect on the purity and recovery of Si product was significant near the isoelectric point. The best purity of Si product was 99.0 wt% accompanied with a recovery of 71.0%. In addition, the wire-saw slurries with different silicon content were investigated by the PPT method and the results showed that the PPT method can be proceeded successfully to recover high-purity silicon powder from kerf loss slurry.