電子束微影鄰近效應參數測定與鄰近效應修正之研究

Abstract

為因應未來積體電路細線化之必然趨勢，達到準確的圖罩線寬控制，電子束鄰近效應修正之重要性正與日劇增。如何以既簡單、且又準確的實驗方法，進行電子束鄰近效應修正，是國際上近幾十年來熱門及重要的研究題 電子束鄰近效應參數和所使用之加速電壓、基材、阻劑性質、阻劑厚度、阻劑對比度及顯影過程等有關。國際上已有許多求取鄰近效應參數方法之文獻報導。在眾多方法中，以L.Stevens所提的圈餅法(Doughnut-Method)最易求得電子散射相關參數，但此法在求取前向散射參數時，易因電子束機台電流的不穩定，及SEM量測不易等問題，而不易求得。 本論文使用圈餅法搭配雙線法。除了以雙線法替代圈餅法，克服上述測求前向散射參數時所遇到之問題，更進一步地推導圈餅法，使其亦可成功地用於求取第三高斯項之鄰近效應參數。本論文並以連續增加劑量之線形法替代傳統大面積方形法，求得阻劑的臨界劑量，提高鄰近效應參數的準確度。本論文亦比較了雙高斯和三高斯函數的優缺點。在鄰近效應修正方面，以本論文所求得之鄰近效應參數搭配數學軟體，對電子束鄰近效應做初步的劑量修正，可得到很好的修正結果。對於線隙比為1之0.2微米圖形，修正前中心和兩旁圖案線寬的差異可達23.5％，劑量修正後，此線寬差異迅速降低至0.54％，且線寬符合關鍵尺寸±10％之製程要求。 由上述實驗驗證可知，圈餅法搭配雙線法為一快速且又準確之鄰近效應參數測求方法。

As the critical dimensions of integrated circuit technology continue to shrink, the E-beam proximity effect correction is indispensable for precise line width control of mask. Using simple and precise experiment method to carry out E-beam proximity effect correction is a hot and an important research topic in the past decades. The E-beam proximity parameters are related with the beam accelerating voltage, substrate material, the resist properties, resist thickness, resist contrast and the development process etc. Various methods for the determination of the proximity parameters have been reported. Among many experimental methods, the doughnut method reported by the L.Stenvens is the easiest way for the determination of the parameters of scattering, however, it is difficult to determine the forward scattering parameter because of the beam current instability of the E-beam writer and the uncertainty from scanning electron microscope (SEM) measurement. In this thesis, the doughnut method coupled with the double lines method are used. Besides using double lines method instead of doughnut method to overcome the problem of determination of the forward scattered parameter, further more, doughnut method has been extended and applied successfully to determine the proximity parameters of the third Gaussian. Using line method of increasing expose dose stepwise instead of large square pattern method to determine the critical dose of the resist, the precision of the proximity parameters can be improved. The advantage and the disadvantage of the double and triple Gaussian functions were also compared. In the Proximity effect correction, the application of the obtained optical parameters and Mathematical software to carry out the preliminary dose correction, resulted in good proximity effect correction. After dose correction, the line-width difference between the center and edge of the 0.2 μm L/S＝1 patterns reduced from 23.5% to 0.54% and kept critical dimension within ±10%. By our experiment, it is concluded that the doughnut method coupled with double lines method are a fast and precise method for the determination of the proximity parameters.