功率優化及色彩表現提升之分光光柵設計與製作

Abstract

現行的液晶顯示系統中利用彩色濾波片來達到全彩的效果是非常普遍且必要的，然而利用彩色濾波片，將會使光使用效率下降至24.89%。為了改善此問題，我們在現有的液晶顯示系統裡，加入分光光柵，使得光源先行分成紅色、綠色及藍色，且分別入射到相對應的子畫素，減少光源被彩色濾波片吸收的比例。 我們使用Dammann提出的純相位分光光柵結構，分光效率則是以Fraunhofer繞射理論作為計算基礎，其中可調變的參數為階數N、相位高度V與對應相位高度的波長，定義為中心波長 。 從計算的結果得知，加入分光光柵後，光的使用效率與小尺寸面板白光LED系統相比可以增加到47.3%，且色域大小方面可以增加到89.12%。 為了評估實際應用上將會遇到的問題對於效率的影響，我們利用傅立葉數學推導出Dammann光柵的解析解，在推導過程將元件寬度有限與斜向入射的問題一併考慮，使得我們能夠利用此解來計算分光後顏色在空間上的分佈，進一步使用來評估整體效率的影響。 接著，我們經由計算分析對位誤差、蝕刻誤差、線寬誤差三種製程誤差對於分光效率的影響。 最後，我們使用半導體製程製作測試片，驗證是否與理論計算吻合。

It is fundamental to use the color filter (CF) in LCD systems to achieve a full-color display. For this purpose, however, the total power efficiency will drop to 24.89% as the light passes through the CF. In order to solve this problem, we have developed a novel approach by inserting a diffractive color separation grating between the backlight and the CF to spatially separate the backlight into three bands (R,G,B) in advance. Then the three bands arrive at their corresponding sub-pixels to reduce the loss of backlight. The efficiency computation of color separation grating presented by Dammann in 1978 is based on Fraunhofer diffraction theory. We take white light LED as our computation example and tune the variables, such as the number of the steps and height of the grating and the central wavelength to get optimum improvement. The computation results demonstrate that we can get the best power efficiency up to 47.3% and the gamut area can be improved up to 89.12%. In order to evaluate the influence of efficiency in practical cases, including oblique incidence and finite element factor, we try to obtain closed-form formula for the efficiency of Dammann grating such that we can use the formula to compute space distribution of different color and furthermore to evaluate the total power efficiency. We also analyze the effect of the three main possible errors, including lateral misalignment, etching error, and line-width error, on the efficiency of 4-level phase color separation grating during the multiple binary mask fabrication. Finally, we use VLSI-like fabrication process to make the test device and verify whether efficiency of the practical device is coincident with the theoretical result.