鎳誘發/鎳誘發側向結晶矽太陽能電池-成長機制、鎳捉聚與太陽電池效能

Abstract

在本論文研究中，我們探討了利用鎳金屬誘發結晶與鎳金屬誘發側向結晶來製作矽基太陽電池元件，在材料上的選擇，鎳可以得到較好的特性是因為鎳的矽化物與矽材料的晶格不匹配只有0.44%以及製程溫度較低等因素考量下，而NIC與NILC在先前之研究大多是應用於低溫多晶矽薄膜電晶體(LTPS-TFT)，但利用此方法來製作多晶矽薄膜會造成鎳金屬或是鎳的矽化物殘留在矽薄膜或是元件內部，導致元件效率變差，因此，在本研究將利用鎳金屬捉聚的技術來降低鎳金屬在元件內的殘留量，並且可以有效提升太陽電池元件之效率

另外，除了針對NIC與NILC兩種結晶方式之研究探討之外，本論文還針對沒有金屬污染疑慮之固相結晶法(SPC)所製作的太陽電池來比較在電性上的差異，針對NIC、NILC與SPC三種方法研究其結晶成長機制，我們利用低掠角X光繞射分析其結晶大小與方向，發現在結晶晶粒大小上，NIC可以得到最大之晶粒，其次是SPC，最後才是NILC，因為Ni在550℃下於單晶矽下的擴散係數為遠遠高於Ni在非晶矽，造成側向結晶的速度遠遠低於Ni在單晶矽中的擴散，因此在進行側向結晶時，會因為Ni從矽晶圓中擴散至a-Si中阻礙了側向結晶；也發現在我們的元件結構中NIC、NILC與SPC主要之結晶方向分別為(111)、(111)與(103)。針對元件效率，我們發現NIC的效率最高，其次是SPC，最後才是NILC，其原因有兩個，第一個原因是以結晶晶粒大小來看，晶粒越大則晶界就越少，因此造成光電導越大以及效率就越高；而第二個原因是NIC的結晶是以縱向往下結晶其方向與電子和電洞移動方向相同，因此所經過的晶界很少使得载子移動速度較快；但NILC結晶是以側向針狀的方式結晶，因此與電子和電洞移動方向是垂直的，因此所經過的晶界最多使得载子移動速度較慢而降低了元件之效率。

本論文也針對鎳金屬殘留對於太陽電池元件之影響性做探討，我們比較了NIC與NILC兩種太陽電池元件在經過鎳捉聚步驟之後在材料上以及電性上之差異，在材料分析上，可以發現經過鎳捉聚步驟的試片的確有效降低了鎳金屬原子在矽薄膜中的殘留量；而在電性上，發現經過鎳捉聚步驟的試片不管是NIC或是NILC都可以得到比較好之效率，另外，不管有沒有經過鎳捉聚步驟，NIC之效率還是比 NILC來的高，其結果也符合我們前一個實驗的結果。

在本論文之研究中，我們成功的利用NIC與NILC這兩種結晶技術製作出其效率在9%以上之太陽電池元件；而對於鎳金屬殘留對太陽電池元件的影響性，也利用鎳捉聚步驟來證明鎳金屬殘留的確會導致元件效率下降。

In this thesis, we fabricated the silicon base solar cells using Ni-metal induced crystallization and Ni-metal induced lateral crystallization. In material selection, nickel had good qualities in this mechanism because of nickel silicide had 0.44% of lattice mismatch with silicon and lower process temperature of crystallization. In general the NIC and NILC technologies were applied on LTPS-TFT previously, but it would lead to poor device performance that nickel atoms remained in the silicon films or devices. Therefore, in this study the Ni-gettering technology was be used to decrease the amounts of residual Ni atoms and improve the performance of solar cells efficiently.

In addition, we fabricated the solar cells by a method of solid phase crystallization (SPC) of residual metal free and investigated the difference of electric properties besides NIC and NILC. We used Grazing Incident X-ray Diffraction (GIXRD) to investigate the grain size and texture of NIC, NILC and SPC poly-silicon. We compared grain size in the three methods. NIC could get the biggest grain size, SPC was secondary and NILC was smallest. Because the diffusion coefficients of nickel atoms in single crystal silicon is higher than in amorphous silicon under 550℃ annealing. This result would lead the velocity of diffusion in lateral crystallization is lower than in single crystal silicon. Moreover nickel atoms would diffuse from silicon wafer to amorphous silicon to hinder lateral crystallization. We also detected by GIXRD the orientation of NIC, NILC and SPC were (111), (111) and (103), respectively. In the efficiency performance NIC could get the best, SPC was secondary and NILC was lowest. The two phenomenons could explain why NIC was best or NILC was lowest. First, when poly-silicon have bigger grain size lead to less grain boundary. This result could make higher photoconductivity to increase efficiency. Second, the growth direction of NIC was vertical that was the same with electrons and holes moving direction, and therefore crossing less grain boundaries. This result could get higher carrier mobility. However, the growth direction of NILC was horizontal (needle line) that would cross a lot of grain boundaries to decrease carrier mobility and efficiency.

We investigated the effect of residual nickel atoms in solar cell devices. For NIC and NILC samples, the difference of electric and material properties will be study after Ni-gettering process. After Ni-gettering process we observed nickel concentration drop by SIMS analysis. However we also observed better efficiency after Ni-gettering process whether NIC sample or NILC. In addition, the efficiency of NIC is better than NILC whether use Ni-gettering or not. The result is identical to previous experiment. In our research, we successful fabricated over 9% efficiency solar cells using NIC and NILC methods Furthermore, we verified the residual nickel atoms will reduce solar cell efficiency by Ni-gettering process.