201005564 九、發明說明： 【發明所屬之技術領域】 本發明係有關於一種光學鄰近效應修正法，特別是關 於一種光學鄰近效應修正法之修正步驟（〇PC correction recipe) ° 【先前技術】 為了防止在製作光罩圖案轉移時的關鍵尺寸偏差 ❺（Critical Dimension Variation，CD Variation)現象，通常在 製作光罩時’都會進行光學鄰近效應修正法(〇PC)，亦即是 將欲曝光在晶片之半導體基底上的原始圖案，利用資料電 腦和軟體運算加以計算修正，得到與原始圖案不同之結果 圖形，再將此結果圖形輸入電腦存檔《根據光學鄰近效應 修正(OPC)所得到的結果圖形製作於光罩上，光束透過此光 罩投影在半導體基底上的圖案可與原始圖案相當接近。 • 第1A圖和第1B圖顯示一種習知的光學鄰近效應修正 法之修正步驟，請參照第1A圖，在建立光學鄰近效應修 正之模型（OPC model)後’再以此〇pc model為基礎去執行 佈局圖案修正步驟。首先將光罩圖案上之邊緣分割成複數 個片段(segment) ’並在每個片段定義出該片段的目標點， 當作執行佈局圖案修正時的比對參考，根據上述比對之結 果決定該片段校正向外突出或向内退縮。在分割片段的過 程中，一般會於光罩圖案上之邊緣定義出目標點1〇1、 102、103(target point)和切點 1〇4、105(dissection point)， 96040/ 0516-A41604TW/Final 201005564 每個片段根據其目標點以OPC model為基礎去計算，並且 校正該片段需突出或内縮之大小。如第1B圖所示，上述 之步驟係使目標點101、102、103通過顯影後檢查（after development inspection ’ ADI)之圖案輪廓11〇。值得注意的 是’此習知的光學鄰近效應修正法之修正步驟，在建立危 險之區域1〇6(例如多邊形圖案中的邊角區或線段的尾端） 之第一目標點102和第一切點104時，係根據經驗法則決 參 定’且其著重的地方為佈局之精確度(fidelity)。易言之， 在定義片段時，頂點108至第一切點104之距離為一固定 值’且片段對應到的第一目標點1〇2的位置亦固定。因此， 爲了讓顯影後檢查之圖案輪廓11〇通過目標點1〇2，該校 正片^又（108至104之片段）向内縮或向外凸之大小亦為固 定。 此習知技術之方法雖然可以使曝出之圖案得到良好的 精確度’但其並沒有考慮到製程窗(pr〇cess wind〇w)，因此 • 不一定是最佳的光罩佈局。 【發明内容】 根據上述問題’本發明提出一種光學鄰近效應修正 法’包括以下步驟：建立光學鄰近效應修正法之模型，並 在執行接下來的佈局圖案片段修正步驟時，先設定複數個 預定目標點或預定切點，並依據哪一個預定目標點或預定 切點所產生出之修正圖案具有較佳的影像品質（image quahty)，於複數個預定目標點或預定切點中選定作為片段 96040/ 0516-A41604TW/Final 6 201005564 修正之目標點與切點。 【實施方式】 以下根據第2圖詳細說明本發明之流程’其中並會以 第3圖輔助說明之。首先，進行步驟S102，輸入佈局資料 庫（layout data base)及初版經〇PC光學鄰近效應校正之佈 局資料。接著，進行步驟S104 ’光學鄰近效應後確認 籲 (Post-OPC verification)，審視佈局並找出熱點區（hot-spot location)。熱點區即微_影成像之危險區域（dangerous layout) ’為微影製程窗（process window)之瓶頸。接著對 於熱點區進行步驟S106’以建立之光學鄰近效應修正模型 (OPC model)和本發明之光學鄰近效應校正修正程式 (OPC correction recipe)，校正佈局和模擬影像品質。進行 步驟S1G8，比較影像品f。進行步驟S11Q，輸 品質之佈局。 步驟S106至步驟SU〇係為本發明之重要 以第3圖輔助說明此部份。請參照第3 _ 卜下 多邊形圖案302(圖式中係顯示 尤皁上之 夕、息拟可盔乂工h廿 1一奉發明不限於此， 少邊1了為任何其它形狀），進行鮮鄰近致 正，根據多邊形圖案3〇2邊 4正法之修 音的县，：^日“+ 輯刀出複數個片段，值得注 意的疋，本發明特別有關熱點區324附 ^左 之目標點和切點位置之定義。在沿X軸方向=!正時 點3〇8時’可設定出複數個預定目標點 ^二切 -第-預定目標點304和一第二預定目如圖式=示 卞點306，而即使 96040/ 0516-A41604TW/Final 7 201005564 切點308相同，不同目標點也會使得校正片段突出或内縮 之距離不同，例如圖式中的第一預定目標點304造成第一 校正片段316，第二預定目標點306造成第二校正片段 318，而不同的校正片段會造成曝光後圖案之影像品質 (image quality)不同。又例如在沿Y軸方向，當選定一目標 點310時，可設定出複數個預定切點，例如圖式中顯示一 第一預定切點314和一第二預定切點312，而即使目標點 A 310相同，不同的切點也會使得校正片段突出或内縮之大 響 小不同，例如圖式中的第一預定切點314造成第三校正片 段322,第二預定切點312造成第四校正片段320,而此不 同的校正片段亦會造成曝光後圖案之影像品質不同。根據 上述，第一預定目標點304、第二預定目標點306、第一預 定切點314和第二預定切點312即會造成四種不同的校正 片段組合（第一校正片段316、第二校正片段318、第三校 正片段322和第四校正片段320)。本實施例於此四種不同 φ 的校正片段組合，找出哪一種具有最大的製程窗和最佳的 影像品質，進一步決定採用哪一個切點和哪一個目標點。 易言之，本發明在決定熱點區324鄰近頂點326之第一個 目標點和第一個切點之位置時，係依據最大的製程窗和最 佳的影像品質而決定。值得注意的是，本說明書雖然舉兩 個預定切點314、312和兩個預定目標點304、306組成之 四個校正片段排列組合為範例，但本發明不限於預定目標 點和/或預定切點的數目，預定目標點和/或預定切點的數目 越多，可得到越佳的影像品質，但所花的時間相對增加。 96040/ 0516-A41604TW/Final 8 201005564 例如，右S另一實施例有三個預定切點和三個 點，其將組成之九個校正片段排列組合，若當又實:: 例有四個預定切點和四個預定目標點，其將組成之十六個 校正片段排列組合，以此類推。 以1Γ詳細_有關影像品質之參數’影像對比(image contrast)係為影像品f之重要參數之一，而影像對比一般 係以;^準影像對數斜率(n〇rmalized ⑽y〇pe，nILS) ❹定義之。請參照第4圖，光罩開口間影像強度的斜率越陡 峭，則影像對比越佳，標準影像對數斜率(NILS0〖可以下 公式表示： = ’其中|係為光罩開口的寬度，1係為影像 強度’ X係為位置。標準影像對數斜率(NILS)越大，影像 對比越佳。 另一個定義影像品質之重要參數之一係為光罩誤差增 ❸ 進係數（mask error enhancement factor，MEEF)，光罩誤差 增進係數(MEEF)可用以下公式表示： ，3CDmask係為光罩之線寬尺寸，3CDresist 係為曝出之光阻的線寬尺寸，當光罩誤差增進係數(MEEF) 越小時，光阻的線寬尺寸較不會隨著光罩之線寬尺寸誤差 而變化，亦即光阻的線寬尺寸越穩定，影像品質越佳（1.0 為最佳）。 因此，本發明實施例在根據複數個的預定目標點和預 定切點決定目標點和切點之位置（特別是鄰近熱點區頂點 96040/ 0516-A41604TW/Final 9 201005564 ^第一個目標點和第一個切點）時，即可根據哪一個預定目 ‘點和/或預定切點得到較大的標準影像對數斜率 和/或較小的鮮誤差增進錄(MEEF)，決定目標點和切點 之位置。 ❹ 局 因此，在本發明一較佳實施例中，可根據上述實施例 ^最佳影像品質之佈局，亦即，具有最大的標準影像 局斜率(NILS)和/或最小的光罩誤差增進係數⑽ef)之佈 根據上述，本發明在進行光學鄰近 步驟時，所定義之目標點和切點並非固定=4= 段具有-定的彈性，不僅可考慮到設計饰局確片 考量到製程窗，因此可定義出較佳的影像‘質。又’亦 以上提供之實施例係用以描述本發明 參 概念，其可包括或運用：更廣泛ΐ: 置組成、製造和使用之特定方法，並 1裴 任何熟習此技藝者，在不脫離本發明 、發明’ 可作些許之更動與潤飾。因此，本發明=内’當 後附之申請專利範圍所界定者為準。’、濩靶圍，當視 96040/ 0516-Α41604TW/Final 201005564 【圖式簡單說明】 第1A圖和第1B圖顯示一種習知的光學鄰近效應修正 法之修正步驟。 第2圖說明本發明一實施例光學鄰近效應修正法之流 程。 第3圖顯示一光罩佈局之平面圖，用以說明本發明一實 施例光學鄰近效應修正法之修正步驟。 第4圖係顯示影像強度相對於位置之曲線圖。201005564 IX. Description of the Invention: [Technical Field] The present invention relates to an optical proximity effect correction method, and more particularly to an optical correction effect correction method (〇PC correction recipe) ° [Prior Art] In the process of making a reticle pattern transfer, the critical dimension variation (CD Variation) phenomenon is usually performed by the optical proximity effect correction method (〇PC) when the reticle is fabricated, that is, it is to be exposed on the wafer. The original pattern on the semiconductor substrate is calculated and corrected by using the data computer and the software operation to obtain a result pattern different from the original pattern, and then the result image is input into the computer archive. The result image obtained by the optical proximity effect correction (OPC) is On the reticle, the pattern of the light beam projected through the reticle on the semiconductor substrate can be quite close to the original pattern. • Figures 1A and 1B show a modification procedure of a conventional optical proximity correction method. Please refer to Figure 1A, after establishing the optical proximity effect correction model (OPC model), based on this 〇pc model. To perform the layout pattern correction step. First, the edge on the reticle pattern is divided into a plurality of segments and the target point of the segment is defined in each segment as a comparison reference when performing layout pattern correction, and the result is determined according to the result of the comparison. The segment correction protrudes outward or retracts inward. In the process of segmenting the segments, the target points 1〇1, 102, 103 (target point) and the tangent points 1〇4, 105 (dissection point) are generally defined at the edges on the mask pattern, 96040/ 0516-A41604TW/Final 201005564 Each segment is calculated based on its target point based on the OPC model, and the size of the segment needs to be highlighted or retracted. As shown in Fig. 1B, the above steps are such that the target points 101, 102, 103 pass through the pattern outline 11 of the after development inspection (ADI). It is worth noting that the correction step of the conventional optical proximity effect correction method is to establish a first target point 102 and a first point in a dangerous region 1〇6 (for example, a corner region in a polygonal pattern or a tail end of a line segment). When the point 104 is cut, it is determined according to the rule of thumb and the place where it is heavily weighted is the fidelity of the layout. In other words, when defining a segment, the distance from the vertex 108 to the first tangent point 104 is a fixed value 'and the position of the first target point 1〇2 corresponding to the segment is also fixed. Therefore, in order to pass the pattern outline 11〇 after the development inspection through the target point 1〇2, the size of the correction piece (the segment of 108 to 104) is also inwardly or outwardly convex. This prior art method allows the exposed pattern to be well-accurate' but does not take into account the process window (pr〇cess wind〇w), so it is not necessarily the optimum mask layout. SUMMARY OF THE INVENTION According to the above problem, the present invention provides an optical proximity effect correction method including the steps of: establishing a model of an optical proximity effect correction method, and setting a plurality of predetermined targets when performing a subsequent layout pattern segment correction step Point or predetermined tangent point, and according to which predetermined target point or predetermined tangent point, the correction pattern has better image quality (image quahty), and is selected as a segment 96040/ 0516-A41604TW among a plurality of predetermined target points or predetermined tangent points /Final 6 201005564 Corrected target points and cut points. [Embodiment] Hereinafter, the flow of the present invention will be described in detail with reference to Fig. 2, which will be explained with reference to Fig. 3. First, step S102 is performed to input the layout data base and the layout information of the first version of the PC optical proximity effect correction. Next, step S104 'Post-OPC verification' is performed, the layout is examined and a hot-spot location is found. The hotspot area is the dangerous layout of micro-image imaging, which is the bottleneck of the process window. Next, step S106' is performed on the hot spot to establish an optical proximity effect correction model (OPC model) and the optical proximity effect correction correction program (OPC correction recipe) of the present invention to correct layout and analog image quality. Step S1G8 is performed to compare the image products f. Perform step S11Q to transfer the layout of the quality. Steps S106 to SU are important to the present invention. This section is explained in detail in Fig. 3. Please refer to the 3rd_poly polygon pattern 302 (the figure shows the special soap on the eve, the interest can be helmeted, the h廿1 is not limited to the invention, the lesser side is any other shape), fresh Adjacent to the correction, according to the polygon pattern 3〇2 side 4 positive method of the sound of the county,: ^ day "+ knives out a number of fragments, notable 疋, the invention is particularly relevant to the hot spot 324 attached ^ left target point and The definition of the tangent point position. When the X-axis direction =! timing point 3〇8, 'a plurality of predetermined target points can be set ^two cut-first-predetermined target point 304 and a second predetermined order as the figure = indication point 306, and even if the 96040/ 0516-A41604TW/Final 7 201005564 tangent point 308 is the same, different target points may cause the corrected segment to protrude or retract differently, such as the first predetermined target point 304 in the drawing causing the first corrected segment 316. The second predetermined target point 306 causes the second corrected segment 318, and the different corrected segments may cause different image quality of the exposed pattern. For example, in the Y-axis direction, when a target point 310 is selected, Set a plurality of predetermined cut points, for example A first predetermined tangent point 314 and a second predetermined tangent point 312 are displayed in the figure, and even if the target point A 310 is the same, different tangent points may cause the correction fragment to protrude or the indentation to be small, such as in the drawing. The first predetermined tangent point 314 causes the third corrected segment 322, and the second predetermined tangent point 312 causes the fourth corrected segment 320, and the different corrected segments may also cause different image quality of the exposed pattern. According to the above, the first predetermined target point 304 The second predetermined target point 306, the first predetermined tangent point 314, and the second predetermined tangent point 312 cause four different combinations of correction segments (the first correction segment 316, the second correction segment 318, the third correction segment 322, and the fourth). Correction segment 320). In this embodiment, the four different φ correction segments are combined to find out which one has the largest process window and the best image quality, and further decide which cut point and which target point to use. The invention determines the position of the first target point and the first tangent point of the hot spot 324 adjacent to the vertex 326 according to the maximum process window and the best image quality. It should be noted that although the present specification exemplifies four calibration segment combinations consisting of two predetermined tangent points 314, 312 and two predetermined target points 304, 306, the present invention is not limited to a predetermined target point and/or a predetermined tangent point. The greater the number of predetermined target points and/or predetermined tangent points, the better the image quality can be obtained, but the time spent is relatively increased. 96040/ 0516-A41604TW/Final 8 201005564 For example, another embodiment of the right S has three a predetermined tangent point and three points, which are arranged in a combination of nine corrected segments, if the real:: the example has four predetermined tangent points and four predetermined target points, which arranges the sixteen corrected segments that are composed, And so on. The image contrast is one of the important parameters of the image f, and the image contrast is generally based on the image logarithm slope (n〇rmalized (10) y〇pe, nILS) ❹ Define it. Referring to Figure 4, the steeper the slope of the image intensity between the openings of the mask, the better the image contrast, the slope of the standard image logarithm (NILS0) can be expressed as follows: = 'where | is the width of the mask opening, 1 is The image intensity 'X is the position. The larger the standard image logarithmic slope (NILS), the better the image contrast. Another important parameter that defines the image quality is the mask error enhancement factor (MEEF). The mask error enhancement coefficient (MEEF) can be expressed by the following formula: 3CDmask is the line width dimension of the mask, 3CDresist is the line width dimension of the exposed photoresist, and the smaller the mask error enhancement coefficient (MEEF) is, The line width dimension of the photoresist does not change with the line width dimension error of the mask, that is, the more stable the line width dimension of the photoresist, the better the image quality (1.0 is optimal). Therefore, the embodiment of the present invention is Determining the position of the target point and the tangent point according to a plurality of predetermined target points and predetermined tangent points (especially when the neighboring hot spot vertex 96040/ 0516-A41604TW/Final 9 201005564 ^first target point and first tangent point) It is possible to obtain a larger standard image log slope and/or a smaller fresh error enhancement record (MEEF) depending on which predetermined target point and/or predetermined tangent point, and determine the position of the target point and the tangent point. In a preferred embodiment of the invention, the layout of the optimal image quality according to the above embodiment, that is, the layout having the largest standard image office slope (NILS) and/or the minimum mask error enhancement coefficient (10) ef, may be When the optical proximity step is performed in the present invention, the defined target point and the tangent point are not fixed = 4 = the segment has a certain elasticity, and not only the design decoration film can be considered to the process window, so that a better definition can be defined. The image 'quality. Further, the embodiments provided above are used to describe the concepts of the present invention, which may include or utilize: a broader method: a specific method of composition, manufacture, and use, and any one skilled in the art, without departing from the present invention. Inventions, inventions, can make some changes and refinements. Therefore, the present invention is defined by the scope of the appended claims. ’, 濩 target circumference, Vision 96040/ 0516-Α41604TW/Final 201005564 [Simple description of the drawings] Figures 1A and 1B show a modification procedure of a conventional optical proximity effect correction method. Fig. 2 is a view showing the flow of the optical proximity effect correction method of an embodiment of the present invention. Figure 3 is a plan view showing a mask layout for explaining the correcting steps of the optical proximity effect correction method of an embodiment of the present invention. Figure 4 shows a plot of image intensity versus position.
主要元件符號說明】 102〜第一目標點； 104〜第一切點； 106〜危險之區域； 110〜圖案輪廓； 3 04〜第一預定目標點； 308〜切點； 312〜第二預定切點； 316〜第一校正片段； 320〜第四校正片段； 3 24〜熱點區； 101〜線段的尾端目標點； 103〜標點； 105〜切點； 108〜頂點； 302〜多邊形圖案； 306〜第二預定目標點； 310〜目標點；Main component symbol description] 102~first target point; 104~first cut point; 106~ dangerous area; 110~ pattern outline; 3 04~first predetermined target point; 308~cut point; 312~second predetermined cut point; 316~first correction segment; 320~four correction segment; 324~hotspot region; 101~end segment target point; 103~punctuation; 105~cut point; 108~vertex; 302~polygon pattern; 306~second Predetermined target point; 310~target point;
314〜第一預定切點； 318〜第二校正片段； 322〜第三校正片段； 3 26〜頂點。 96040/ 0516-A41604TW/Final 11314~first predetermined tangent point; 318~second correction fragment; 322~third correction fragment; 3 26~ vertex. 96040/ 0516-A41604TW/Final 11