以奈米壓印技術製作大面積次微米孔洞陣列

Abstract

薄膜太陽能電池是值得期待的下世代太陽能電池之一，理因其低廉的價格和多方面的基板應用。多年研究下已研發出許多方法可以製作薄膜太陽能電池，empty-space in silicon (ESS) 是其中可期待的方法之一。 ESS的優勢在於製程上的低花費並能提供了良好的能量轉換率。然而，如果能夠導入奈米壓印技術取代原本的深紫外線的微影製程會是一個更大的突破。

＂軟性＂奈米壓印改良自原本的奈米壓印，以軟性的圖章取代傳統的硬式材質，因此有更好的延展和耐用性並能達到高解析度和良好的均勻度。除此之外，軟性奈米壓印並有機會在高低落差較大的試片上展現良好的性能。軟性奈米壓印可以解析為下列三種步驟：(1)製作軟性圖章 (2)壓印步驟 (3)圖形轉換。首先為了證明此技術在量產上有其優勢，我們比較了此技術和目前主要的下世代微影技術，並以cost of ownership (CoO)模型來證明軟性奈米壓印的可行性。

在製作軟性圖章上，一個適合的硬質罩幕層是影響表現的重要條件之一。不理想的幕層設計會導致圖形的變形和崩解。在此，我們考慮了變形的模型進而設計出一個適當的結構，並在測試中證明此結構的穩定及可利用性。另外在壓印的步驟中，抗黏層的塗佈有相對性的重大影響。實驗的過程中，我們並發現要達到均勻的表面並能保持較高的光阻高度差必須考量到適當的光阻塗佈厚度。由於奈米壓印的基本原理來自於光阻的流動和空洞的填補。在此概念下，成功發展出一個簡易的計算求出最恰當的光阻塗佈厚度，並且藉此達到良好的壓印成果。

在壓印的圖形上會有一層多餘的光阻層需要去除以達到後續良好的蝕刻圖形，在此我們以氧氣電漿蝕刻的方法與之去除，此法不僅簡單更有顯著的效果。而後續矽的蝕刻則以SF6 混合O2來進行研究。本文中討論了蝕刻製程中幾個重要的條件，例如壓力，氣體流量和能量功率。在氧氣的濃度增加的情況下，側壁的保護相對的提高但也因此減弱了蝕刻的能力。研究中也顯示在低功率和高壓力下，蝕刻的表現是較好的，無論是在側壁保護或是相對的蝕刻率上。

但也由於製程上的限制，單純以SF6 混合O2進行的蝕刻無法達到ESS對孔洞深度的要求。因此，我們刻意蝕刻出倒金字塔形狀的孔洞結合陽極氧化的步驟達到極深的孔洞。但又考量到此條件下的孔洞，其直徑的變異太大可能會增加ESS的困難度。因此，我們在矽晶片上沉積了一百奈米的二氧化矽當作額外的硬質罩幕層，而成功的製作出符合ESS要求的周期性孔洞。最後，實驗的結果客觀地顯示出軟性奈米壓印有著其多樣的發展性，且能在低花費且快速的情況下完成。

Thin-film solar cells are one of the promising approaches for the next generation of solar cells among various approaches due to their low price and wide applications of different substrates. There are many techniques to achieve thin-film solar cells. A layer-transfer process based on the empty-space in silicon (ESS) technique has a promising opportunity, because it presents potentially a better tradeoff between energy conversion efficiency and cost than other techniques. However, its biggest challenge is to reach competitive prices. The development of nano imprint lithography (NIL) as an alternative to optical lithography such as deep-ultra-violet (DUV) lithography is an important aspect of decreasing the cost.

Soft-NIL is an improved technique providing a high-resolution and good uniformity patterns by using a soft stamp instead of a rigid stamp. The soft-NIL process is briefly composed of the following three steps: (1) Fabrication of soft stamp, (2) imprint process and (3) pattern transfer. Here, we compare soft-NIL with other possible next generation lithography, and use the model called cost of ownership (CoO) to prove the feasibility of soft-NIL.

For the fabrication of the soft stamp, a suitable hard mask as a mold for soft stamp is the most important factor that affects the performance. Inappropriate design of the hard mask will result in deformations such as buckling, lateral collapse, and roof collapse. After considering deformation models, we fabricate soft stamps avoiding the above deformations. Besides, the soft stamp is proved sufficiently stable and reusable even over ten times imprint.

Concerning the imprint process, an anti stick layer coating strongly affect the performances, and the findings for imprinting a uniform structure and high resist contrast over a large area by calculating a suitable polymer thickness are demonstrated. Based on the optimized concerns including squeeze flow and cavity-filling mechanisms, we successfully imprint uniform structures with very thin residual layers over a large area.

Oxygen plasma is used to remove the residual layer, and anisotropic etching of silicon is performed by SF6 and O2 plasma. We prove the importance of accurately removing residual layers, and the essential of uniform imprinted structures. We report detailed discussions of the link between the etching profiles and etch parameters such as power, pressure, and gas flow. When higher O2 concentrations are added, it causes a better sidewall passivation but also decreases the etching rate of silicon. Therefore, improvements of the anisotropic etching are shown by applying low power density and high pressure since low power density strongly decreases the lateral etching and high pressure increases the silicon-etching rate.

Extremely deep pores achieved are achieved owning to a combination between dry-etch and anodization, because the soft-NIL patterning process allows to make regular pores. A better control of anodization is performed by an inverted-pyramid-shaped cavity intentionally etched providing specific points to etch. Based on this thought, we take advantage on a negative effect existing in dry etching process, which results in slopped sidewalls.

In the end, we deposit 100 nm silicon oxides before coating the thermal resist, and use this oxide as mask to etch silicon for the desired structures. The imprinted structures show the same property when imprinting above silicon oxide. Therefore, we could fabricate desire structures based on proper etching conditions in this process, and the results show many possibilities of soft-NIL.