04513twfl.doc/006 修正日期92.7.9 玖、發明說明： 本發明是有關於一種光學接近度的校正方法（〇pticai Proximity Correction，〇PC)，且特別是有關於〜種利用形成 輔助圖案（Assist Feature)，或在原始圖案上加入邊角 或其尾端加上錘頭狀圖案（Hammerhead)，或者局1放大 縮小原始圖形的方法，而減少光學接近度效果（◦如丨⑼ Proximity Effect，ΟΡΕ)，以取得逼真精確之原始圖案的方 法。 Ρ 〇 ^ 在積體電路蓬勃發展的今日，元件縮小化與積集化是 必然之趨勢，也是各界積極發展的重要課題。而積體電路 的製造過程中，微影蝕刻步驟則爲一決定元件性能=重要 關鍵。隨著積集度（Integration)的逐漸提高，元件尺寸逐漸 縮小’元件與元件之間的距離也必須縮小，因此造成^微 影蝕刻步驟中，圖案轉移產生偏差（Deviatl〇n)，^如，^ 一個罩幕的圖案，利用影微蝕刻轉移到一晶片上時，常^ 發生圖案中，直角部分被鈍化，圖案之尾端收縮，以及線 寬的減小或增大等。這也就是所謂的光學接近度效應 (Optical Pr〇ximity Effect，ΟΡΕ)。這些偏差在元件尺寸較大， 或積集度較低的情形下，不致產生太大的負面效果，然而， 在高積集度的積體電路中，便嚴重影響到元件的效能。例 如，在高積集度的積體電路中，元件和元件的距離很小， 而當轉移到晶片上的圖案線寬脹膨時，極有可能產生局部 的圖案重疊，而造成短路現現。換言之，積體電路的效能 提昇’亦即，對應於尺寸的縮小可得到的速度之提昇，逐 045 13twfldoc/006 修正日期92.7.9 漸地受到一連串的微影蝕刻製程中，缺乏圖案的逼真度所 限制。 形成ΟΡΕ的因素，主要包括一些光學因素，例如光線 穿過光罩上不同圖案所產生的干涉，光阻製程，包括烘烤 溫度/時間，顯影等，不平整基底所產生之反射光，以及蝕 刻效果等。當可允許的尺寸誤差値隨著圖案尺寸而減小 時，圖案之轉移便隨著〇ΡΕ而產生更大的偏差。請參考第 1圖，其中虛線部分爲所欲轉移之圖案，而實線部分則爲 實際轉移之結果，如圖所示，所欲形成之矩形圖案之直角 部分被鈍化’而圖案之線寬脹膨。 習知爲解決上述之問題，發展了光學接近度之校正方 法(0PC)，例如在美國專利案件us Patent Νο.5,732,233以 及US Patent Νο.5,879,844中所提，在主要圖案（Main Pattern)，亦即所欲轉移之原始圖案上，加上一些輔助圖案 (Assist Feature)，或加寬或減小主要圖案之線寬，以加強 曝光時圖案之對比（Contrast)或提高製程裕度（pr〇cess Window) 〇 傳統的光學接近度校正方法，是對一測試圖案（Test Pattern)進行曝光，將曝光結果與所欲轉移之原始圖案相比 對，取得二者之間的偏移程度，以決定是否將進行光學接 近度的校正，並依據二者之間的偏差，決定校正的程度。 另外，參考第2圖，在上述之專利案件US Patent No.5,879,844中，以一導電層201(斜線部分）爲例，假設該 導電層形成於一擴散層202(點陣部分）之上，而在與擴散 045 13twfl.doc/006 修正日期92.7.9 層互相重疊的部分之導電層的平行邊緣（粗線部分），則爲 待校正的目標。當相鄰兩邊緣之間的距離小於一起始値， 或當邊緣之長度和距離小於另一起始値時，這些邊緣則將 進行光學接近度校正。 而有關於輔助圖案之部分，則取決於相鄰兩子圖案之 間的距離。 然而，由上述之方法，必需經過實際的實驗，以得知 是否將要進行光學接近度校正。又或者必需取決於待定義 層及其上下層之間的幾何關係。因此，在數據的處理上造 成相當大的困擾。 因此，本發明提供一種光學接近度校正方法。利用所 欲轉移之原始圖案的關鍵尺寸（Critical Dimension，CD)作爲 決定是否加入輔助圖案的依據。此外，先對原始圖形加入 角落襯線（Serif)或錘頭狀圖案（Hammerhead)，再加入一輔 助圖形，以避免校正圖案的重疊，或其中子圖案之間的距 離過小。 本發明提出一種光學接近度校正方法。首先，提供一 待轉移的原始圖案，然後，求得圖案的關鍵尺寸，當關鍵 尺寸的大小小於曝光光源的2.5倍波長時，對原始圖案加 入邊角襯線或錘頭狀圖案，而當關鍵尺寸的大小小於曝光 光源之波長時，則對原始圖案加上輔助圖案。再利用此一 校正圖案進行微影曝光，以將圖案轉移之偏移降至最低。 此外，本發明又提供了加入校正圖案的順序，亦即先形成 邊角襯線或錘頭狀圖案，再加入輔助圖案，如此，可以避 554259 修正日期92.7.9 04513twfl.doc/006 免圖案的重疊或圖案之間的距離過小，而導致其他的圖案 轉移偏移。 爲讓本發明之上述和其他目的、特徵、和優點能更明 顯易懂，下文特舉一較佳實施例，並配合所附圖式，作詳 細說明如下： 圖式之簡單說明： 第1圖繪示出習知利用微影步驟，進行圖案轉移的結 果。 第2圖繪示出習知之一待轉移圖案中，待校正之圖案 部分。 第3a圖繪示出本發明中，待轉移的主要圖案。 第3b圖繪示出本發明中，經由光學接近度校正後的圖 案。 第3c圖繪示出利用本發明，另一經由光學接近度校正 的圖案。 第4圖繪示出本發明中，光學接近度校正法的流程圖。 圖式之標記說明： 201 導電層 202 擴散層 300 主要圖案 300a :主要圖案之矩形部分 300b :主要圖案之狹長部分 302 :輔助圖案 304 :邊角襯線 045 13twf 1.doc/006 修正日期92.7.9 400 :提供一主要圖案 402 :檢查是否〇ϋ<2.5λ 404 :加入邊角襯線或錘頭狀圖案 406 :檢查是否CD<X 408 :加入輔助圖案 410 :進行曝光微影等步驟 實施例 一般而言’微影步驟所使用之儀器包括一線網 (Reticle)，亦即所謂的光罩，線網上具有一圖案，以對應 在一積體電路之某一層的圖形。通常，線網包括一玻璃板， 由一已定義的不透光，例如，鉻層（Chromium)所覆蓋。在 進行曝光時，線網放置於晶片和光源之間，而在線網和晶 片之間，又置放了一步進機（Stepper)。當光源入射至線網 時，光將穿越未被不透光層覆蓋的玻璃層，投射至晶片上 之一光阻層。如此，線網上的圖案便轉移至光阻層上。 如以上所述’由於曝光時，透過光罩的光線產生折射 現象或干涉現象，加上製程中的其他因素，轉移的圖案便 產生了形變。爲使轉移的圖案能夠逼真，減少其形變，本 發明提供了一種光學接近度的校正方法，利用直接取得待 轉移之原始圖案的關鍵尺寸（Cntical Dimension，CD)，做爲 是否進行校正之依據，然後，再利用習知的校正方法，以 求得最佳的逼真度（Fidelity)。 首先，提供一個待轉移的主要圖案3〇〇。主要圖案3〇〇 包括一矩形部分300a及一狹長部分3〇〇b。此一主要圖案 修正日期92.7.9 045 13twfl .doc/006 如第3a圖所示。接著，取得主要圖案300的關鍵尺寸（Cntical Dimension，CD)，當主要圖案300的關鍵尺寸CD小於約曝 光步驟中所使用之光源波長的2.5倍時，亦即CD<2.5X時， 則對主要圖案300加入一些邊角襯線（Senf)，或在其邊緣 加入一錘頭狀圖案（Hammerhead)，例如，在主要圖案300 之矩形圖案300a四個角落各自加上一個Serif。而當主要 圖案300之關鍵尺寸小於約曝光光源之波長，亦即CD<a 時，則對主要圖案300加入一些輔助圖案（Assist Feature)。 Serif或Hammerhead的加入，可以減少在轉移圖案上，邊 緣或角落的形變量，而輔助圖案的加入，則可以增加圖案 的對比，得到較高的解析度。 請參考第3b圖，其繪示出利用本發明所形成之圖案 校正。300a及300b代表各代表一主要圖案300的一部分， 而302爲輔助圖案，304爲邊角襯線。在此一例子中，校 正的圖案是先加入輔助圖案302,再加入邊角襯線304, 結果如圖所示，校正圖案中，邊角襯線和輔助圖案的距離 過於接近。 爲避免校正圖案具有如第3b圖所示之缺點，本發明 提出了一個方法，亦即先形成邊角襯線或錘頭狀圖案，再 加入輔助圖案，如此，輔助圖案的大小，可依據經由邊角 襯線或錘頭狀圖案校正的圖案，調整其尺寸大小，以進行 適當的校正。請參考第3c圖，提供一待轉移的主要圖案 300，其又包括了一矩形部分300a及一狹長部分300b。檢 查主要圖案300之關鍵尺寸，確認其是否小於約曝光光源 04513twfl.doc/006 修正日期92.7.9 之波長的2.5倍，當CD<2.5X時，則在矩形部分302a四個 角落，各自加上一個邊角襯線304。然後，再檢查主要圖 案300之關鍵尺寸是否小於約曝光光源之波長，當CDd 時，則在主要圖案300上加上輔助圖案302。結果，在第 3b圖中，輔助圖案302與邊角襯線304過於接近之部分， 在此處已做適當調整，不致引起其他曝光的副面影響。 本發明之流程，可參考第4圖，首先，在步驟400中， 提供一待轉移之主要圖案，然後在步驟402中，確認主要 圖案之關鍵尺寸是否小於約曝光波長之2.5倍，當其不小 於2.5λ時，則進入步驟406，而當其小於2.5λ時，則於步 驟404中，在主要圖案之各個角落加入一邊角襯線，或在 其邊緣加上錘頭狀圖案，接著，再進行步驟406，檢查主 要圖案之關鍵尺寸是否小於約曝光之波長，當其不小於λ 時，則進入步驟410，而當其小於λ時，則於步驟408中， 在主要圖案中加入輔助圖案，然後，進入步驟410，進行 影微製程之曝光等步驟。 雖然本發明已以一較佳實施例揭露如上，然其並非用 以限定本發明，任何熟習此技藝者，在不脫離本發明之精 神和範圍內，當可作各種之更動與潤飾，因此本發明之保 護範圍當視後附之申請專利範圍所界定者爲準。04513twfl.doc / 006 Date of revision 92.7.9 发明 Description of the invention: The present invention relates to a method for correcting optical proximity (〇pticai Proximity Correction, 〇PC), and in particular, it relates to ~ assisting the formation of auxiliary patterns (Assist Feature), or adding corners or tails with a Hammerhead pattern on the original pattern, or a method of enlarging and reducing the original pattern in round 1 to reduce the optical proximity effect (such as 丨 ⑼ Proximity Effect, ΟΡΕ ) To obtain a realistic and accurate original pattern. Ρ ^ In today's booming development of integrated circuits, component downsizing and integration are an inevitable trend and an important subject for active development in all walks of life. In the manufacturing process of integrated circuits, the lithography etching step is a key factor that determines the performance of the device. With the gradual increase of the integration, the size of the component is gradually reduced. The distance between the component and the component must also be reduced. Therefore, in the lithography etching step, the pattern transfer has a deviation (Deviatl0n). For example, ^ When a mask pattern is transferred to a wafer by shadow micro-etching, it often occurs that in the pattern, the right-angled portion is passivated, the end of the pattern shrinks, and the line width is reduced or increased. This is the so-called Optical Proximity Effect (OPE). These deviations do not cause much negative effects when the component size is large or the accumulation level is low. However, in a high-integration integrated circuit, the performance of the component is seriously affected. For example, in a high-integration integrated circuit, the distance between the component and the component is very small, and when the pattern line width transferred to the wafer expands, it is very likely that a partial pattern overlap will occur, causing a short circuit. In other words, the performance improvement of the integrated circuit 'that is, the increase in speed that can be obtained corresponding to the reduction in size is gradually improved by 045 13twfldoc / 006 92.7.9 gradually in a series of lithography etching processes, the lack of pattern fidelity Restricted. The factors that form ΟΡΕ mainly include some optical factors, such as interference caused by light passing through different patterns on the reticle, photoresistive processes, including baking temperature / time, development, etc., reflected light from uneven substrates, and etching Effect, etc. When the allowable dimensional error decreases with the size of the pattern, the shift of the pattern will cause a larger deviation with the OPE. Please refer to Figure 1, where the dotted line is the pattern to be transferred, and the solid line is the result of the actual transfer. As shown in the figure, the right-angled portion of the rectangular pattern to be formed is passivated, and the line of the pattern is expanded. Swell. Conventionally, in order to solve the above-mentioned problems, a method for correcting optical proximity (0PC) has been developed, for example, as mentioned in US Patent No. 5,732,233 and US Patent No. 5,879,844. In the Main Pattern, that is, Add some auxiliary patterns to the original pattern you want to transfer, or widen or reduce the line width of the main pattern to enhance the contrast of the pattern during exposure (Contrast) or increase the process margin (pr0cess Window) ) 〇 The traditional optical proximity correction method is to expose a test pattern, compare the exposure result with the original pattern to be transferred, and obtain the degree of deviation between the two to determine whether to Correct the optical proximity and determine the degree of correction based on the deviation between the two. In addition, referring to FIG. 2, in the above-mentioned patent case US Patent No. 5,879,844, a conductive layer 201 (slanted portion) is taken as an example, assuming that the conductive layer is formed on a diffusion layer 202 (lattice portion), and The parallel edge (thick line) of the conductive layer in the part overlapping with the diffusion 045 13twfl.doc / 006 correction date 92.7.9 layer is the target to be corrected. When the distance between two adjacent edges is less than one initial chirp, or when the length and distance of the edges are smaller than the other initial chirp, these edges are corrected for optical proximity. The part about the auxiliary pattern depends on the distance between two adjacent sub-patterns. However, from the above method, it is necessary to undergo actual experiments to know whether or not optical proximity correction is to be performed. Or it must depend on the geometric relationship between the layer to be defined and its upper and lower layers. Therefore, there is considerable confusion in data processing. Therefore, the present invention provides a method for correcting optical proximity. The critical dimension (CD) of the original pattern to be transferred is used as the basis for deciding whether to add the auxiliary pattern. In addition, first add a serif or Hammerhead pattern to the original pattern, and then add an auxiliary pattern to avoid overlapping of the correction patterns or the distance between the sub-patterns is too small. The invention provides a method for correcting optical proximity. First, provide an original pattern to be transferred. Then, find the key size of the pattern. When the size of the key size is less than 2.5 times the wavelength of the exposure light source, add a corner serif or hammerhead pattern to the original pattern. When the size is smaller than the wavelength of the exposure light source, an auxiliary pattern is added to the original pattern. This correction pattern is used for lithographic exposure to minimize the shift of the pattern transfer. In addition, the present invention also provides the order of adding correction patterns, that is, forming a corner serif or hammerhead pattern first, and then adding an auxiliary pattern. In this way, 554259 correction date 92.7.9 04513twfl.doc / 006 The overlap or the distance between the patterns is too small, causing other pattern shifts to shift. In order to make the above and other objects, features, and advantages of the present invention more comprehensible, a preferred embodiment is given below in conjunction with the accompanying drawings for detailed description as follows: Brief description of the drawings: FIG. 1 The results of conventional pattern transfer using lithography steps are shown. Fig. 2 illustrates a pattern portion to be corrected in a pattern to be transferred, which is a conventional one. Figure 3a illustrates the main pattern to be transferred in the present invention. Fig. 3b illustrates a pattern after optical proximity correction in the present invention. Fig. 3c shows another pattern which is corrected by the optical proximity using the present invention. FIG. 4 is a flowchart illustrating an optical proximity correction method in the present invention. Description of drawing marks: 201 conductive layer 202 diffusion layer 300 main pattern 300a: rectangular portion of main pattern 300b: elongated portion of main pattern 302: auxiliary pattern 304: corner serif 045 13twf 1.doc / 006 Date of revision 92.7. 9 400: Provide a main pattern 402: Check if 〇ϋ < 2.5λ 404: Add corner serif or hammerhead pattern 406: Check if CD < X 408: Add auxiliary pattern 410: Perform exposure lithography and other steps Generally speaking, the equipment used in the lithography step includes a reticle, also known as a photomask, which has a pattern corresponding to the pattern of a certain layer of an integrated circuit. Generally, the wire mesh includes a glass plate covered by a defined opaque, for example, Chromium layer. During the exposure, the wire net is placed between the wafer and the light source, and a stepper is placed between the wire net and the wafer. When the light source is incident on the wire mesh, the light will pass through the glass layer not covered by the opaque layer and be projected onto a photoresist layer on the wafer. In this way, the pattern on the wire net is transferred to the photoresist layer. As described above, 'the light transmitted through the mask causes refraction or interference when exposed, and other factors in the process cause the transferred pattern to deform. In order to make the transferred pattern realistic and reduce its deformation, the present invention provides a method for correcting optical proximity, using the Cntical Dimension (CD) of the original pattern to be transferred directly as a basis for whether to perform the correction. Then, the conventional correction method is used to obtain the best fidelity. First, a main pattern 300 to be transferred is provided. The main pattern 300 includes a rectangular portion 300a and a narrow portion 300b. This main pattern was revised at 92.7.9 045 13twfl.doc / 006 as shown in Figure 3a. Next, the key dimension (Cntical Dimension, CD) of the main pattern 300 is obtained. When the key dimension CD of the main pattern 300 is less than about 2.5 times the wavelength of the light source used in the exposure step, that is, when CD < 2.5X, The pattern 300 is added with some corner serifs, or a Hammerhead pattern is added to the edge thereof. For example, the four corners of the rectangular pattern 300a of the main pattern 300 are each added with a Serif. When the key size of the main pattern 300 is smaller than the wavelength of the exposure light source, that is, CD < a, some auxiliary patterns (Assist Feature) are added to the main pattern 300. The addition of Serif or Hammerhead can reduce the deformation of edges, corners or corners in the transfer pattern, and the addition of auxiliary patterns can increase the contrast of the pattern and obtain higher resolution. Please refer to FIG. 3b, which illustrates the pattern correction formed by the present invention. 300a and 300b each represent a part of a main pattern 300, 302 is an auxiliary pattern, and 304 is a corner serif. In this example, the corrected pattern is added with the auxiliary pattern 302 and then the corner serif 304. The result is shown in the figure. In the correction pattern, the distance between the corner serif and the auxiliary pattern is too close. In order to avoid the disadvantages of the correction pattern as shown in Fig. 3b, the present invention proposes a method, that is, forming a corner serif or hammerhead pattern first, and then adding an auxiliary pattern. In this way, the size of the auxiliary pattern can be determined by Adjust the size of the corner serif or hammerhead pattern for proper correction. Referring to FIG. 3c, a main pattern 300 to be transferred is provided, which further includes a rectangular portion 300a and a narrow portion 300b. Check the key dimensions of the main pattern 300 and confirm whether it is less than about 2.5 times the wavelength of the exposure light source 04513twfl.doc / 006 correction date 92.7.9. When CD < 2.5X, add the four corners of the rectangular portion 302a, respectively. One corner serif 304. Then, check whether the key dimension of the main pattern 300 is smaller than the wavelength of the exposure light source. When CDd, add the auxiliary pattern 302 to the main pattern 300. As a result, in Fig. 3b, the portion where the auxiliary pattern 302 and the corner serif 304 are too close to each other have been appropriately adjusted here, so as not to cause other side effects of exposure. The process of the present invention can refer to FIG. 4. First, in step 400, a main pattern to be transferred is provided, and then in step 402, it is confirmed whether the key dimension of the main pattern is less than about 2.5 times the exposure wavelength. When it is less than 2.5λ, it proceeds to step 406, and when it is less than 2.5λ, in step 404, a corner serif is added to each corner of the main pattern, or a hammer-shaped pattern is added to the edge, and then, Go to step 406 to check whether the key dimension of the main pattern is smaller than the exposure wavelength. When it is not smaller than λ, proceed to step 410, and when it is smaller than λ, add an auxiliary pattern to the main pattern in step 408. Then, the method proceeds to step 410 to perform steps such as exposure of a photomicrography process. Although the present invention has been disclosed as above with a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make various modifications and decorations without departing from the spirit and scope of the present invention. The scope of protection of the invention shall be determined by the scope of the attached patent application.