201120586 六、發明說明： 【發明所屬之技術領域】 本發明係關於一種控制臨界尺寸的方法。特定言之，本 發明係關於一種使用特殊蝕刻氣體配方以控制蝕刻後臨界 尺寸的方法。 • 【先前技術】 微影製程可以說是整個半導體製程中最關鍵的步驟之 一。原因在於，優良的微影製程不僅能正面影響元件的集積 度（integration)及性能，同時也能正面影響產能及生產製 造成本。而光學微影設備更因其高產能（highthr〇ughput) 而成為目前最先進圖案轉印的主流技術。 基本的Μ衫裝程包含了光阻旋轉塗佈（spin c〇ating )、 •軟烤（soft baking )、曝光、曝後烤（p〇st exp〇雲 ba]dng )、 顯影及硬烤（hard baking) ...等等步驟。為了要能將光阻所 疋義的確實地轉移出去，.目前—種已知的方法是使用複 合光阻材料層。 第1-3圖例示先前技藝使用複合光阻材料層來轉移光阻 所定義圖案的方法。請參考第旧，例如，在基材谢上所 使用的光阻材料層110具有—種雙層結構，分別稱為底部抗 反射層（BarC) 111與光阻層112。光阻層.112已經預先經過 一適當之曝光與顯影步驟，而具有—駭之光阻圖案114。 201120586 光阻圖案114會具有特定之臨界尺寸。 •理論上’臨界尺寸不只會受到微影步驟參數的影響，飯 刻步驟的參數也會實質上影響後續所轉移圖案（圖未示）的 臨界尺寸’因此，使得臨界尺寸偏差（CD bias)，即對於固 定的顯影後微距量測（ADI CD)而言，不同蝕刻步驟的參 數會造成不同的蝕刻後微距量測（AEI CD)。因此對於固定 的顯影後微距量測（ADICD)而言，不同蝕刻步驟的參數 可能會造成蝕刻後微距量測（AEICD)保持不變，如第2 圖所示，或是不利地變大，如第3圖所示。此等改變的臨界 尺寸偏差代表蝕刻步驟沒有正確地將預定光阻圖案之臨界 尺^傳承下去，終究使得最後半導體元件之臨界尺寸過大而 不符合原始設計規格。 有鑑於此，仍然需要一種控制臨界尺寸的新穎方法，縮 】、顯影後與㈣後的臨界尺寸偏差，即，使顯影後微距量測 ^adicd)錢刻後微距量測（AEICD)儘量靠近，以盡 蓋維持微影步驟的臨界尺寸的正綠性。 【發明内容】 、本發明於是提出-雛制臨界尺相新穎方法。本發明 方法的結果可以使得_後之微距量測（AEICD)不合大於 顯影後之微距量測（ADICD)，以盡量維持微影步驟作界 尺寸的正確性。 本發明控制臨界尺寸的方法，首先，提供材料層。其次， 201120586 提供複合圖絲層。複合圖#化層會覆蓋材料層 義第-臨界尺寸之圖案。然後’使用_劑來崎钱列步疋 驟，以關複合圖案化層’再進行圖案轉移步驟 具有第二臨界尺寸之轉移圖案之材料層。蝕刻劑包；二氧化 碳。本發明控制臨界尺寸方法的特徵即在於：第二臨界尺寸 實質上小於第一臨界尺寸。 11 ' 本發明繼續又提出-種控制雙鑲嵌結構中臨界尺寸的201120586 VI. Description of the Invention: TECHNICAL FIELD OF THE INVENTION The present invention relates to a method of controlling a critical dimension. In particular, the present invention relates to a method of using a special etching gas recipe to control the critical dimension after etching. • [Prior Art] The lithography process is one of the most critical steps in the entire semiconductor process. The reason is that an excellent lithography process can not only positively affect the integration and performance of components, but also positively affect the production capacity and production cost. Optical lithography equipment has become the mainstream technology for the most advanced pattern transfer due to its high throughput. The basic shirting process includes spin c〇ating, soft baking, exposure, post-exposure (p〇st exp〇云) d, ) development and hard bake ( Hard baking) ...and so on. In order to be able to reliably remove the photoresist, a known method is to use a composite photoresist layer. 1-3 illustrate a prior art method of transferring a pattern defined by a photoresist using a layer of a composite photoresist material. Please refer to the old one. For example, the photoresist material layer 110 used on the substrate has a two-layer structure called a bottom anti-reflection layer (BarC) 111 and a photoresist layer 112, respectively. The photoresist layer 112 has been previously subjected to a suitable exposure and development step to have a photoresist pattern 114. The 201120586 photoresist pattern 114 will have a specific critical dimension. • Theoretically, the critical dimension is not only affected by the parameters of the lithography step, but the parameters of the rice cooking step will also substantially affect the critical dimension of the subsequent transferred pattern (not shown). Therefore, the critical dimension deviation (CD bias), That is, for a fixed post-development macro measurement (ADI CD), the parameters of the different etching steps result in different post-etch macro measurements (AEI CD). Therefore, for fixed post-development macro measurement (ADICD), the parameters of different etching steps may cause the post-etch macro measurement (AEICD) to remain unchanged, as shown in Figure 2, or disadvantageously large. As shown in Figure 3. The critical dimension deviation of these changes represents that the etching step does not properly pass the critical dimension of the predetermined photoresist pattern, and eventually the critical dimension of the final semiconductor component is too large to conform to the original design specifications. In view of this, there is still a need for a novel method for controlling the critical dimension, shrinking, and the critical dimension deviation after development and (4), that is, making the macro measurement after development (adicd) and the macro measurement (AEICD) as much as possible. Close to cover the green color of the critical dimension of the lithography step. SUMMARY OF THE INVENTION The present invention thus proposes a novel method of cutting off the critical dimension. The result of the method of the present invention can be such that the post-magnitude measurement (AEICD) is not greater than the post-development macro measurement (ADICD) to maintain the correctness of the lithography step boundary. The method of the present invention for controlling critical dimensions first provides a layer of material. Secondly, 201120586 provides a composite layer of silk. The composite layer will cover the pattern of the first critical dimension of the material layer. Then, the material layer having the transfer pattern of the second critical dimension is then subjected to a step of patterning to close the composite patterned layer. Etchant pack; carbon dioxide. The method of controlling the critical dimension of the present invention is characterized in that the second critical dimension is substantially smaller than the first critical dimension. 11 ' The present invention continues to propose a type of control for critical dimensions in a dual damascene structure
方法。首先，提供多層材料層與位於多層材料層上之硬遮罩 層。硬遮罩㈣定義雙鑲嵌結射之溝渠尺寸。其次，提供 複合圖案化層。複合圖案化層會覆蓋硬遮罩層，並具有定義 第-臨界尺寸之圖案。然後’使用钱刻氣體來進行敍刻步 驟，以關複合圖案化層，再進行圖轉移步驟，藉此使得 多層材料層形成用於雙鑲嵌結構中之通孔。此通孔具有第二 £»界尺寸祕職體包含二氧化碳。本發明控制雙镶寂結 構中臨界尺寸方㈣特徵即在於：第二臨界尺寸實f上小於 第一臨界尺寸。 【實施方式】 本發月首先提供-種控制臨界尺寸的方法。本發明控制 L界尺寸的方法可以應用於㈣單鑲嵌結構、雙鑲嵌结構、 淺溝槽隔離狀騎之臨界尺寸。本發日㈣特徵之-在於， 使用包含兩層先阻中失置有切抗反射層（麵）的圖案化 複口層來疋義材料層之圖案或材料層中的開口，其中定義圖 201120586 聿化複合層時會使用到二氧化碳作為蝕刻氣體。第4A_6圖 例示本發明控制臨界尺寸之方法用於形成閘極之示意圖。首 先，如第4A圖所示，提供一材料層2〇1，例如閘極材料層， 與複合圖案化層202。材料層2〇1可以包含用於形成閘極之 多種材料層。如果是用於建立多晶矽閘極，如第4A圖所示， 材料層201可以包含閘極氧化物層211、閘極多晶矽材料層 212、與視憒況需要之硬遮罩層213。如果是用於建立金屬閘 極，如第4B圖所示，材料層2〇1可以包含高介電係數材料 層211’ ’例如包含銓（Hf)之高介電係數材料層，位於高介 電係數材料層211’上之功函數（w〇rkfuncti〇n)金屬層an， 用來調整閘極的功函數、以及位於功函數金屬層212，上方之 閘電極層213’，其可為任何導電材料，例如多晶矽，與對應 之硬遮罩。 視情況需要’高介電係數材料層211，的下方可選擇性地 設有二氧化矽層（圖未示）。另外，高介電係數材料層211， 與金屬層212’間還可以選擇性地設有帽蓋介電層（圖未示）。 複合圖案化層202則可以包含不同性質之正型光阻或是 負型光阻’與含石夕抗反射層（silicon_containinghard_mask bottom anti-reflection coating，Shb)。例如，複合圖案化層 202可以包含對深紫外光敏感之光阻層221，例如KrF光阻 層，抗反射層222與I-line光阻層223。如本技藝人士所共 知，抗反射層222可以是單一或多層抗反射層，而成分可以 為含石夕之有機尚为子聚合物（organosilicon polymer)或聚石夕 201120586 物（polysilane) ’ 至少具有一發色基團（chr〇m〇ph〇re gr〇Up)以 及一交聯基團((^0881丨111^1^81'0叩），而1_1丨1^光阻材料對於 365nm波長之光源特別敏感。較佳者光阻層221已預先定 義有所需之佈局圖案。第4A圖與第4B圖例示光阻層221 已預先定義有閘極圖案224。 複合圖案化層202中之光阻層221即可以作為後續蝕刻 步驟之遮罩，因此預先建立有定義第一臨界尺寸之圖案 • 224。例如，使用光阻旋轉塗佈、教烤、曝光、曝後烤、顯 影及硬烤…等步驟，結合掃描機及/或步進機之操作，使得光 阻層221建立有定義第一臨界尺寸之圖案224。之後，即使 用定義有第一臨界尺寸之圖案224，來將預定之閘極形狀轉 移至下方之材料層中。 接下來，如第5圖所示，以圖案化光阻層221作為蝕刻 遮罩進行第一次之蝕刻步驟’以將定義有第一臨界尺寸之圖 鲁案224向下轉移至抗反射層222，而形成具有第一臨界尺寸 之第一轉移圖案225。如果抗反射層222之成分為含石夕之有 機高分子聚合物或聚矽物，則可以使用三氟曱烷與四氟曱烷 之混合氣體作為蝕刻劑。此時，I-line光阻層.223可以作為 蝕刻停止層。或是，視情況需要，亦可以對I-line光阻層223 進行過餘刻.。 再來，如第6圖所示，以具有第一臨界尺寸之第一轉移 圖案225作為蝕刻遮罩，進行第二次之蝕刻步驟，於是蝕刻 I-line光阻層223以形成一圖案226。第二次之蝕刻步驟可能 201120586 二^移除圖案化光阻層221。可以使用一特殊調配之鞋刻 而彤、二此第二次之蝕刻步驟，以便蝕刻I_Hne光阻層223， 例^成:有第二臨界尺寸之圖案226。姓刻劑可以為氣體， 八—氧化碳。視情況需要，本發明之蝕刻氣體還可以額外 匕3 —輔助氣體，例如一氧化碳。 可以^所有的複合圖案化層观皆被定義之後，接下來，4就 ,用剩下來的圖案化層2〇2對材料層2〇1中之硬遮罩層 取不同>蝕刻來繼續定義圖案。冬後，視情況需要，可以採 5的步騍。例如，可以再使用剩下來的圖案化層2〇2與 =案=^更遮罩層213 ’對材料層2〇1進行一次或多次的姓 逝或^ **進行一灰化製程’先去除剩下來的圖案化層 Γ、’ς'後才以圖案化的硬遮罩層213對下方的材料層2〇1 進行一次或多次的蝕刻。 在、1過本發明第二次之㈣步驟後，可以在〗光阻 1 223中得到纖刻減少的結果。一般咸信，藏刻會負面 =響轉移随之臨界尺寸大小。換言之’顯刻會改變轉 圖案之臨界尺寸，使得在進行較小料尺寸的圖案轉移之 時’所得的結果會有比簡不同之臨界尺寸偏差。 與傳統使用氧氣之勒刻配方相較，本發明使用二氧化碳 作為㈣Hine光阻層223的氣體，就有效抑制了不欲之側 餘刻現象，使得_後之微距量測（Am CD )實f上近似顯 影後之微距量測（ADICD)，以盡量_微影步驟中臨界尺 寸的正確傳遞。接下來，即可使用習知之餘刻方法來繼續建 201120586 立閘極結構。此等方法為本技藝人士所共知，故在此不多作 贅述。承上所述，本發明方法的特徵之一即在於：使用二氧 化石反作為餘刻I-line光阻層223的氣體，以有效地抑制不欲 之側姓刻現象。此等不可預期的功效，即反應在蝕刻結束 後’第二臨界尺寸實質上近似第一臨界尺寸之成果，而不改 變預期之臨界尺寸。 另一方面，本發明控制臨界尺寸方法可以應用於控制雙 擊鑲肷結構之臨界尺寸。本發明的特徵之一在於’使用包含兩 層光阻中間夾置有含矽抗反射層（SHB)的圖案化複合層來 定義材料層之圖案或材料層中的開口，其中定義圖案化複合 層時會使用到二氧化碳作為钱刻氣體。換言之，本發明方法 可以進一步應用在基材中，形成用於單鑲嵌結構或雙鑲嵌結 構中之通孔。第7-12圖例示本發明用於控制雙鑲嵌結構中 臨界尺寸方法之示意圖。第7圖所示，提供多層材料層3〇1 •與複合圖案化層302。多層材料層301可以包含多種材料 層，例如包含硬遮罩層303、包含氧化物之塾層311、超低 介電材料層（ultra low-k material layer ) 312與四乙基石夕氧烧 (tetraethoxysilane，TEOS )層 313。硬遮罩層 3〇3 可以為一 複合硬遮罩層，例如包含氮化鈦層314、氮氧化石夕層315與 帽蓋層316等等。硬遮罩層303中各層間之順序僅為例示之 用，而不以此為限。 複合圖案化層302可ά包含不同性質之正型光阻或是負 型光阻、含矽抗反射層（SHB)與位於含矽抗反射層上之單 201120586 層或多層抗反射層。例如，複合圖案化層3〇2可以包含對深 紫外光敏感之光阻層321，例如KrF光阻層，抗反射層322 與I-line光阻層323。複合圖案化層302中還可以選擇性地 有位於含矽抗反射層（SHB)上的單層或多層抗反射層，圖 中則一起以抗反射層322作為代表。如本技藝人士所共知， 抗反射層322成分可以為含矽之有機高分子聚合物或聚矽 物’至少具有一發色基團以及一交聯基團，而—光阻材 料對於365nm波長冬光源特別敏感。較佳者，光阻層321 已預先定義有所需之佈局圖案。如第7圖例示，光阻層321 已預先定義有用於雙鑲嵌結構中之通孔圖案3〇4。另一方 面，硬遮罩層303則預先定義—雙鑲嵌結構中之一溝渠尺寸。 接下來’如第8圖所示，以圖案化光阻層321作為姓刻 遮罩進行第-次之烟步驟’以將定義有第—臨界尺寸之圖 案扣4向下轉移至抗反射層322，而形成具有第一臨界尺寸 之第-轉移圖案305 °如果抗反射層322之成分為含石夕之有 機高分子聚合物或聚矽物’則可以使用三氟曱烷與四氟曱烷 之混合氣體作為姓刻劑。此時，Mine光阻層奶可以作為 飾刻停止層。或是’視情兄需要，亦可以對光阻層奶 進行過钱刻。 再來如第9圖所示，以具有第一臨界尺寸之第一轉移 圖案305之抗反射層322作為#刻遮罩，進行第二次之侧 步驟，於是触刻！-iine光阻層323以形成圖案遍。此等蝕 刻步驟可能會完全移除圖案化光阻層321。可以使用一特殊 201120586 調配之蝕刻劑來進行此蝕刻步驟，以便蝕刻I-line光阻層 323，而形成具有第二臨界尺寸之圖案3〇6。蝕刻劑可以為氣 體，例如二氧化碳。視情況需要，蝕刻氣體還可以額外包含 一輔助氣體’例如一氧化碳。 之後，如第10圖所示，進行後續之蝕刻步驟，將圖案 3〇6，再次向下轉移至四乙基矽氧烷層313與超低介電材料 層312，以形成雙鑲嵌結構中之通孔3〇7。接下來，就可以 • 先藉由灰化製程將I-Hne光阻層323剝除，如第11圖所示， 再使用預先定義有雙鑲嵌結構中溝渠尺寸之硬遮罩層3〇3作 為蝕刻遮罩，再次進行蝕刻步驟，於是就可以在四乙基矽氧 烷層313與超低介電材料層312中形成雙鑲嵌結構中之溝渠 308 ’同時還在墊層3u中形成雙鑲嵌結構中之通孔3〇7，如 第12圖所示。由於此時-溝渠308與通孔307皆已一併定義 在四乙基矽氧烷層313與超低介電材料層312中，即可視為 ^在四乙基矽氧烷層313與超低介電材料層312中形成同時具 有溝渠308與通孔307之雙鑲嵌結構。 由於在先前I-line光阻層323之蝕刻時，本發明係使用 二氧化碳作為蝕刻劑來避免側蝕現象，進而能精確地控制 I-line光阻層323中產生具有第二臨界尺寸之圖案306,因此 所得雙鑲嵌結構中之通孔307當然會具有所預期之臨界尺 寸。 與傳統使用氧氣之蝕刻配方相較，本發明使用二氧化碳 作為蝕刻Ι-Hne光阻層323的蝕刻配方，可以有效抑制負面 11 201120586 的側仙衫，並使得仙狀_量測（A D 於顯影後之微距量測（ADICD)，以 不曰大 界尺寸的正確傳遞。本發明方法的特徵之中臨 用二氧化碳作為邮—323的二吏 效地抑制不欲之側侧現象。此等不可職的功效，= 於具有較小臨界尺寸之圖案之則轉移時，可 之微距量測（规CD)不會大於顯影後之微距免:後 ⑼。換言之，第二臨界尺寸實f上會小於第—臨界尺寸。 以上所述僅為本發明之較佳實施例，凡依本發明 所做之均等變化與_ 1應屬本發明之涵蓋範圍。 轨圍 【圖式簡單說明】 第1-3圖例不先前㈣使用複合光阻材 所定義圖㈣料。 ㈣移先阻 之示=6圖例示本發糊臨界尺寸㈣ 之一 f ^12圖例示本發明控制雙鑲嵌結構中臨界尺寸方法 【主要元件符號說明】 101基材 110複合光阻材料層 12 201120586 111底部抗反射層 112光阻層 114光阻圖案 201材料層 202複合圖案化層 211閘極氧化物層 211’高介電係數材料層 φ 212閘極多晶矽材料層 212’功函數金屬層 213硬遮罩層 213’閘電極層 221光阻層 222抗反射層 2231-line光阻層 224閘極圖案 胃225第一轉移圖案 226圖案 301多層材料層 302複合圖案化層 303硬遮罩層 304通孔圖案 305第一轉移圖案 306圖案 201120586 307通孔 308溝渠 311墊層 312超低介電材料層 313四乙基矽氧烷層 314氮化鈦層 315氮氧化矽層 316帽蓋層 321光阻層 322抗反射層 3231-line光阻層method. First, a layer of multilayer material and a hard mask layer on the layer of multilayer material are provided. Hard mask (4) defines the size of the ditch of the double inlaid junction. Second, a composite patterned layer is provided. The composite patterned layer covers the hard mask layer and has a pattern defining a first critical dimension. The engraving step is then performed using a gas engraving gas to close the composite patterned layer and then performing a pattern transfer step whereby the multilayer material layer is formed into vias for use in the dual damascene structure. This through hole has a second £» boundary size secret body containing carbon dioxide. The feature of the present invention for controlling the critical dimension of the double-inserted structure is that the second critical dimension is smaller than the first critical dimension. [Embodiment] This month first provides a method of controlling the critical size. The method for controlling the size of the L boundary of the present invention can be applied to (4) the critical dimension of a single damascene structure, a double damascene structure, and a shallow trench isolation ride. The feature of the present invention is that the patterning layer of the material layer or the opening of the material layer is defined by using a patterned reticular layer containing two layers of the first resist and having the anti-reflective layer (face) missing, wherein the definition map 201120586 Carbon dioxide is used as an etching gas when the composite layer is deuterated. 4A-6 is a schematic view showing a method of controlling a critical dimension of the present invention for forming a gate. First, as shown in Fig. 4A, a material layer 2?1, such as a gate material layer, and a composite patterned layer 202 are provided. The material layer 2〇1 may comprise a plurality of material layers for forming a gate. If used to create a polysilicon gate, as shown in FIG. 4A, the material layer 201 can include a gate oxide layer 211, a gate polysilicon material layer 212, and a hard mask layer 213 as needed. If it is used to establish a metal gate, as shown in FIG. 4B, the material layer 2〇1 may comprise a high-k material layer 211′′ such as a layer of high-k material containing hafnium (Hf), located in a high dielectric. A work function (w〇rkfuncti〇n) metal layer an on the coefficient material layer 211' is used to adjust the work function of the gate, and the gate electrode layer 213' located above the work function metal layer 212, which may be any conductive Materials such as polysilicon, with corresponding hard masks. Optionally, a layer of high dielectric constant material layer 211 may be provided with a layer of germanium dioxide (not shown). In addition, a cap dielectric layer (not shown) may be selectively disposed between the high-k material layer 211 and the metal layer 212'. The composite patterned layer 202 may comprise a positive-type photoresist or a negative-type photoresist and a silicon-containing hard_mask bottom anti-reflection coating (Shb). For example, the composite patterned layer 202 can comprise a photoresist layer 221 that is sensitive to deep ultraviolet light, such as a KrF photoresist layer, an anti-reflective layer 222, and an I-line photoresist layer 223. As is known to those skilled in the art, the anti-reflective layer 222 can be a single or multi-layer anti-reflective layer, and the composition can be an organosilicon polymer or a polysilicon of at least 201120586 (polysilane). Has a chromophore group (chr〇m〇ph〇re gr〇Up) and a cross-linking group ((^0881丨111^1^81'0叩), while 1_1丨1^ photoresist material for 365nm wavelength The light source is particularly sensitive. Preferably, the photoresist layer 221 has a predetermined layout pattern defined in advance. FIGS. 4A and 4B illustrate that the photoresist layer 221 has a gate pattern 224 defined in advance. The photoresist layer 221 can be used as a mask for the subsequent etching step, so that a pattern defining the first critical dimension 224 is established in advance. For example, using a photoresist spin coating, teaching baking, exposure, exposure baking, development, and hard baking Steps, in conjunction with the operation of the scanner and/or stepper, cause the photoresist layer 221 to have a pattern 224 defining a first critical dimension. Thereafter, a pattern 224 defining a first critical dimension is used to pre-determine The gate shape is transferred to the underlying material layer. Next, as shown in FIG. 5, the first etching step is performed by using the patterned photoresist layer 221 as an etch mask to transfer the Tulu case 224 defining the first critical dimension downward to the anti-reflection layer 222. And forming a first transfer pattern 225 having a first critical dimension. If the composition of the anti-reflective layer 222 is an organic high molecular polymer or a polyfluorene containing a diarrhea, trifluorodecane and tetrafluorodecane may be used. The mixed gas is used as an etchant. At this time, the I-line photoresist layer .223 can be used as an etch stop layer. Alternatively, the I-line photoresist layer 223 can be left over as needed. As shown in FIG. 6, the first transfer pattern 225 having the first critical dimension is used as an etch mask, and a second etching step is performed, thereby etching the I-line photoresist layer 223 to form a pattern 226. The second time The etching step may remove the patterned photoresist layer 221 by using a special matching shoe, and the second etching step may be used to etch the I_Hne photoresist layer 223, for example: having a second Critical dimension pattern 226. Surname can be gas, eight - Oxidized carbon. The etching gas of the present invention may additionally contain an auxiliary gas such as carbon monoxide as needed. After all the composite patterned layers are defined, then, the remaining pattern is used. The layer 2〇2 takes a different <etching of the hard mask layer in the material layer 2〇1 to continue defining the pattern. After the winter, as needed, a step of 5 can be taken. For example, the remaining pattern can be reused. Layer 2 〇 2 and = case = ^ more mask layer 213 'One or more times of the material layer 2 〇 1 is killed or ^ ** an ashing process 'first remove the remaining patterned layer Γ, After the 'ς', the underlying material layer 2〇1 is etched one or more times with the patterned hard mask layer 213. After the second (fourth) step of the present invention, the result of the fiber reduction can be obtained in the photoresist 1 223. Generally speaking, the engraving will be negative = the transfer will follow the critical size. In other words, 'significantly changes the critical dimension of the transition pattern so that the resulting result at the time of pattern transfer with a smaller material size will have a different critical dimension deviation than the simple one. Compared with the conventional oxygen engraving formula, the present invention uses carbon dioxide as the gas of the (four) Hine photoresist layer 223, thereby effectively suppressing the unwanted side remnant phenomenon, so that the macro measurement (Am CD) after the _ Approximate development of the macro measurement (ADICD) to try to pass the correct size of the critical dimension in the lithography step. Next, you can use the conventional method to continue to build the 201120586 vertical gate structure. These methods are well known to those skilled in the art and are therefore not described here. As described above, one of the features of the method of the present invention is that the sulfur dioxide is used as the gas of the residual I-line photoresist layer 223 to effectively suppress the side phenomenon of the unwanted side. These unpredictable effects, i.e., the reaction at the end of the etch, the second critical dimension substantially approximates the outcome of the first critical dimension without changing the expected critical dimension. Alternatively, the method of controlling critical dimensions of the present invention can be applied to control the critical dimension of a double-click stud structure. One of the features of the present invention is to define a pattern in a material layer or an opening in a material layer using a patterned composite layer comprising a tantalum-containing anti-reflective layer (SHB) sandwiched between two layers of photoresist, wherein a patterned composite layer is defined Carbon dioxide is used as a gas for engraving. In other words, the method of the present invention can be further applied to a substrate to form vias for use in a single damascene structure or a dual damascene structure. Figures 7-12 illustrate schematic views of a method for controlling critical dimensions in a dual damascene structure of the present invention. As shown in Fig. 7, a multilayer material layer 3〇1 and a composite patterned layer 302 are provided. The multilayer material layer 301 may comprise a plurality of material layers, for example comprising a hard mask layer 303, an oxide containing layer 311, an ultra low-k material layer 312 and tetraethoxysilane. , TEOS) layer 313. The hard mask layer 3〇3 may be a composite hard mask layer, for example comprising a titanium nitride layer 314, a nitrous oxide layer 315 and a cap layer 316, and the like. The order of the layers in the hard mask layer 303 is for illustrative purposes only and is not limited thereto. The composite patterned layer 302 may comprise a positive or negative photoresist of different properties, a ytterbium-containing anti-reflective layer (SHB), and a single 201120586 layer or a plurality of anti-reflective layers on the anti-reflective layer. For example, the composite patterned layer 3〇2 may comprise a photoresist layer 321 that is sensitive to deep ultraviolet light, such as a KrF photoresist layer, an anti-reflective layer 322 and an I-line photoresist layer 323. The single-layer or multi-layer anti-reflection layer on the antimony-containing anti-reflective layer (SHB) may also be selectively provided in the composite patterning layer 302, and the anti-reflection layer 322 is collectively represented in the figure. As is known to those skilled in the art, the anti-reflective layer 322 component may be a cerium-containing organic high molecular polymer or polythene material having at least one chromophoric group and a crosslinking group, and the photoresist material has a wavelength of 365 nm. Winter light sources are particularly sensitive. Preferably, the photoresist layer 321 has been previously defined with a desired layout pattern. As illustrated in Fig. 7, the photoresist layer 321 has been previously defined with a via pattern 3〇4 for use in a dual damascene structure. On the other hand, the hard mask layer 303 is pre-defined as one of the trench dimensions in the dual damascene structure. Next, as shown in FIG. 8, the patterned photoresist layer 321 is used as a surname mask to perform the first-time smoke step' to transfer the pattern buckle 4 defining the first critical dimension downward to the anti-reflection layer 322. And forming a first-transition pattern having a first critical dimension of 305 °. If the composition of the anti-reflective layer 322 is an organic polymer or a polythene containing a diarrhea, trifluorodecane and tetrafluorodecane may be used. The mixed gas is used as a surname. At this time, Mine photoresist layer milk can be used as a finishing stop layer. Or, as the brothers need, they can also make money for the photoresist. Further, as shown in Fig. 9, the anti-reflection layer 322 having the first transfer pattern 305 having the first critical dimension is used as the #刻遮, and the second side step is performed, so that it is engraved! The -iine photoresist layer 323 is patterned to form a pattern. These etching steps may completely remove the patterned photoresist layer 321 . This etching step can be performed using a special 201120586 formulated etchant to etch the I-line photoresist layer 323 to form a pattern 3〇6 having a second critical dimension. The etchant can be a gas such as carbon dioxide. The etching gas may additionally contain an auxiliary gas such as carbon monoxide, as the case requires. Thereafter, as shown in FIG. 10, a subsequent etching step is performed to transfer the pattern 3〇6 again down to the tetraethylphosphonium layer 313 and the ultra-low dielectric material layer 312 to form a dual damascene structure. Through hole 3〇7. Next, it is possible to first strip the I-Hne photoresist layer 323 by a ashing process, as shown in Fig. 11, and then use a hard mask layer 3〇3 pre-defined as the size of the trench in the dual damascene structure. The mask is etched and the etching step is performed again, so that the trench 308' in the dual damascene structure can be formed in the tetraethyl siloxane layer 313 and the ultra-low dielectric material layer 312 while forming a dual damascene structure in the underlayer 3u. Through hole 3〇7, as shown in Figure 12. Since the trench 308 and the via 307 are both defined in the tetraethyl siloxane layer 313 and the ultra-low dielectric material layer 312, it can be regarded as a tetraethyl siloxane layer 313 and ultra low. A dual damascene structure having a trench 308 and a via 307 is formed in the dielectric material layer 312. Due to the etching of the previous I-line photoresist layer 323, the present invention uses carbon dioxide as an etchant to avoid side etching, thereby enabling precise control of the pattern 306 having a second critical dimension in the I-line photoresist layer 323. Thus, the vias 307 in the resulting dual damascene structure will of course have the expected critical dimensions. Compared with the conventional etching scheme using oxygen, the present invention uses carbon dioxide as an etching recipe for etching the yttrium-Hne photoresist layer 323, which can effectively suppress the negative side of the 2011 11586, and make the sensation _ measurement (AD after development) The macro measurement (ADICD), in order to ensure the correct transmission of large-size dimensions. Among the features of the method of the present invention, carbon dioxide is used as the second side of the post-323 to suppress the side phenomenon of the undesired side. Efficacy, = when transferring to a pattern with a smaller critical dimension, the macro measurement (CD) will not be greater than the macro after development: (9). In other words, the second critical dimension will be It is less than the first critical dimension. The above description is only the preferred embodiment of the present invention, and the equal variation and _1 according to the present invention should be covered by the present invention. Track circumference [simple description of the figure] 3 The illustration is not the previous (4) The material defined by the composite photoresist (4). (4) The shift of the first resistance = 6 shows the critical dimension of the hair paste (4) One of the f ^ 12 diagram illustrates the method of controlling the critical dimension of the dual damascene structure of the present invention [ Main component symbol description] 101 substrate 110 composite photoresist material layer 12 201120586 111 bottom anti-reflection layer 112 photoresist layer 114 photoresist pattern 201 material layer 202 composite patterned layer 211 gate oxide layer 211 'high dielectric constant material layer φ 212 gate Polycrystalline germanium material layer 212' work function metal layer 213 hard mask layer 213' gate electrode layer 221 photoresist layer 222 anti-reflection layer 2231-line photoresist layer 224 gate pattern stomach 225 first transfer pattern 226 pattern 301 multi-layer material layer 302 Composite patterned layer 303 hard mask layer 304 via pattern 305 first transfer pattern 306 pattern 201120586 307 via 308 trench 311 pad 312 ultra low dielectric material layer 313 tetraethyl decane layer 314 titanium nitride layer 315 Niobium oxynitride layer 316 cap layer 321 photoresist layer 322 anti-reflection layer 3231-line photoresist layer