UMCD-2007-0260 24965twf.doc/n 九、發明說明： 【發明所屬之技術領域】 本發明是關於一種研磨製程，且特別是關於研磨製 程中一種研磨液的供給方法。 【先前技術】 在非常大尺寸積體電路(VLSI)製程或極大尺寸積體電 路(ULSI)的製程中，化學機械研磨(CMP)是唯一可以提供 全面平坦化的技術。基本上，化學機械研磨是利用與機 械研磨的磨砂輪(grinding wheel)相似的原理，配合化學物 質的輔助，將沈積於晶圓上之各層的粗糙表面變得平坦 光滑。晶圓的平坦化是影響對準的準確度的因素之一， 假如晶圓的平坦化步驟成效不彰，不僅是對準系統中的 罩幕無法準確的對準聚焦在晶圓上，而且在製程過程中 錯誤的可能性會相對地增加。 化學機械研磨（CMP)除了對晶圓表面輪廓提供全面 平坦化的效果外’亦可應用於垂直及水平金屬内連線 (Interconnects)之鑲嵌結構的製作、前段製程中元件淺溝 渠隔離製作及其他先進元件之平坦化與製作。而在多層 内連線的製程中，平坦化步驟更是不可或缺的。 圖1繪示一種習知的研磨設備。研磨設備的核 心單元是一個自動旋轉的研磨台102以及一個晶圓握柄 104’通常研磨台1〇2會跟安裝在研磨台ι〇2上的研磨塾 —起旋轉。而晶圓握把104的位置是可以調整的，提 供的晶圓110會被晶圓握把104固定抓住。一個研磨液 UMCD-2007-0260 24965twf.doc/n 供應器108係裝設在研磨台側，提供研磨用的研磨液 112。在研磨期間，晶圓握把104會確定晶_ ιι〇與研磨 墊106接觸，透過研磨墊106與研磨液112而對晶/圓ιι〇 進行的研磨。另外，在CMP設備動中通常會H一個 修整器114用來調整研磨墊狀況。 曰 八適當研赖或調整研磨液成 刀等因素，對於研磨表現及效用有極顯著之影塑。此外， 由於使用過之研磨液容易污染環境，—β :使高的今曰，謹慎評估研磨液之用量 /、1文用万八亦疋必要的。 【發明内容】 斷供法’可週期性地中 J液糟此’不但可以避免研磨液使用過度造成溢 良 :1研;!減少研磨液之使用’且提供穩定且均勻度 本發明提供—種三階段式研舰供應法，至 磨::層的一晶圓;⑻提供至少-種研磨液 運订研磨製程，起初研磨持續第一期間τι， ~ 磨ί間糾81上所具有之材料層會接_機^上的 磨持㈣_彳了研磨；⑻停止該研磨液之供應並繼續研 持碎第㈣Τ2;⑹恢復該研練之供應並繼續研磨 :Τ3:(權步·)增數次直至達： ·· 、十對該材料層所進行的整個研磨過程所需時 UMCD-2007-0260 24965t^f.d〇c/n 間為TO，而步驟(b)、（C) ' (d)所需的第一、第二與第三期 間分別是 ΤΙ、T2 與 T3 ’ 則 TO = ΤΙ + η* (T2+T3)，其 中η代表總共施行步驟(c)至⑷的總次數，且η為大於〇 的整數。 在本發明之一實施例中，上述之Τ1至少約為TO之 10%以上。而較佳之狀況，T1約為το之50%。το的範 圍例如約介於1-200秒之間。 在本發明之一實施例中，上述之T2約為το之 %至7.5%。而較佳之狀況，T2為1%丁0、2%T0或3% Τ0。 / 在本發明之一實施例中，上述之該材料層包括至少 一金屬層或至少一介電層，或金屬層或介電層之組合。 在本發明之一實施例中，上述之T2大小是與該研磨 墊之剩餘使用壽命成正相關。 在本發明之一實施例中，上述之步驟(b)至(d)中更包 括同時使用一修整器來調整該研磨墊的狀況。其中T2大 小是與該修整器之剩餘使用壽命成正相關。 在本發明之一實施例中，上述之Τ〇及Τ2的大小可 以線上監控方式來及時調整。 在本發明之一實施例中，上述方法甲機台具有大於 或等於20呎/秒平方(ft/sec2)的一向心加速度。 在進行研磨製程時，本發明所提供的研磨液供應方 法會間歇性地停止研磨液之供應，並且停歇週期與期間 可以視不同狀況來做調整，因此可以避免研磨液之浪費 並降低生產成本。 、 UMCD-2007-0260 24965twf.doc/i 為讓本發明之上述特徵和優點能更明顯易僅，下文 特舉較佳實施例，並配合所附圖式，作詳細說明如下。 【實施方式】 圖2A繪示根據本發明之一較佳實施例的研磨液供應 方法流程圖，而圖2B繪示根據本發明之一較佳實施例的 研磨製程剖面示意圖。較佳實施例中，係採用三階段式 研磨液供應法，請參照圖2A與圖2B，首先，於步驟20 中，提供具有一材料層202之一晶圓200。該材料層202 可為至少一金屬層、或至少一介電層、或金屬層與介電 層之組合。而該晶圓可為各種形式之半導體晶圓或非半 導體晶圓，如石夕晶圓、石夕鍺晶圓、絕緣層上覆石夕(s〇〗)晶 圓、玻璃基材、複合材料等。步驟22中，提供至少一種 研磨液208並針對晶圓200之材料層202來進行一研磨 製私’研磨持續第一期間(the first period)。其中，該至少 一種研磨液208可為含有研磨粒子之金屬或介電層所用 之研磨漿，亦可為去離子水；該研磨製程可為化學機械 研磨、機械研磨或電化學機械研磨。在研磨期間，晶圓 握把204會確定晶圓200上所具有之材料層202與研磨 塾206接觸’透過研磨塾206與研磨液208而對晶圓200 進行研磨。而步驟24中，則停止該研磨液之供應並繼續 研磨持續第二期間(the second period)。接著，於步驟26 中，恢復該研磨液之供應，仍舊繼續研磨持續第三期間 (the third period)。之後，步驟28中，重複步驟24至26 數次直至達到研磨終點或者達到所需之研磨效果，例如 UMCD-2007-0260 24965twf.doc/n 特定的膜厚。 該研磨製程可以是具有時間控制模式(time m〇de)或 具有終點偵測模式(end point mode)的研磨製程；而所項 間控制模式乃是指定整個或部分研磨過程所需時間之 研磨方式，而終點偵測模式則是依賴裝設在研磨塾中的 4貞測器來置測晶圓上之特定膜層的厚度來捉取終點，並 在量到終點後再過研磨(over-polish) —段時間。 若將針對此材料層202所進行的整個化學機械研磨 過程所需時間視為T0，而三階段所需的第一、第二與第 三期間分別訂為ΤΙ、T2與T3，則： TO = Tl+n*(T2+T3) 其中η代表總共施行步驟24至26之總次數；η為大於 〇的整數。 ' 根據本發明之較佳實施例，Τ1至少約為Τ0之1〇% 以上，較佳T1為T0之50%左右，以達到充分潤濕研磨0 墊的效果並避免晶圓乾磨。而T2約為T0之〇.5%至7 5 % °以上所述之步驟或範圍僅是本發明較佳實施例所舉 之例子’但本發明之範圍並僅僅不侷限在此。 + 舉例而言，當T2為1%丁0 ’重複4次(n = 4)且丁3 為 11%Τ0，Τ1 為 52%Τ0。 舉例而言，當Τ2為2%Τ0 ’重複3次(η = 3)且丁3 為 15%Τ0，ΤΙ 為 49%Τ0。 又舉例而言，當Τ2為3%Τ0，重複3次(11：=3)且丁3 為 14%丁〇，τΐ 為 49%Τ0。 又舉例而言，當Τ2為4%Τ0,重複3次(n==3)iT3 UMCD-2007-0260 24965twf.d〇c/n 為 12%T0，T1 為 52%丁〇。 重複3次(η = 3)且Τ3 又舉例而言，當T2為5%τ〇, 為 12%Τ0，Τ1 為 49%τ〇。 根據所進订之實驗數據 即-直持續供應研磨液），以丁2為2%:= 讀賊進行化學錢研㈣程來冊，在㈣= (_了:均勻度⑺與缺陷數㈣eit conns)表現上均表現與控制組相當， 佳，=。甚至在評估中央至邊緣指數(= • mdex)時’以T2為2%、视與5%丁〇之條件所進行 化學機械研磨結果柯滿足量紅製程需求。 此外’若以目前相容製程，針對Τ2為2%、3%與5 %Τ〇之條件來進行評估，估計每_^晶圓研磨成本分別 可減>、大約12%、18%或甚至3〇%之消耗，也就是降低 研磨;^使：量’而每年可節省高達數百萬之成本。 :般而言，針對特定被研磨之材料層，根據研磨前、 後於里測機台中非原地(ex_situ)地量測該材料膜層厚 度，配合所使用研磨墊種類與所使用之研磨液 ，而由研 磨前、後的膜厚差可計算得到膜層移除速率之外，更可 根據研磨前的材料層厚度作為決定總研磨時間(τ〇)的依 據（亦即決定TO之時間長度範圍）。本發明所使用之研 磨墊例如了以疋固疋磨粒式(行xed abrasive; fa)研磨塾或 表面具有不同圖案或溝槽的研磨墊。當然其他影響化學 機械研磨製程的其他變數如研磨頭所施的壓力、晶圓與 研磨塾的疋轉速度等等亦需納入考量，以調整τ〇之範圍 UMCD-2007-0260 24965twf.doc/i 大小。舉例而言，TO的範圍約為1-200秒之間。 而T2之時間長度範圍則可視研磨墊狀況或修整器 狀況’作彈性調整；甚至可即時或後續視該批研磨晶圓 之狀況或製程需要而配合調整。 更具體一點’當挑選T2之範圍時，需配合考量許多 不同因素。舉研磨墊為例子，所應選用之T2大小是與所 使用研磨塾之剩餘使用壽命成線性正相關。這是因為， 當研磨墊剛開始使用時，其研磨墊表面之狀況一般尚稱 理想而研磨移除效率較佳，在此時可考慮將停止供給研 磨液之時間延長（亦即使用較大數值的T2),而避免研 磨液使用之浪費。 同樣來說，研磨墊粗糙的表面有利於提高材料移除 率’而利祕整㈣研料進行適#的修整(⑶nditi〇ning) 可以增加研磨墊表面粗糙度，並使材料去除率趨於一 致。因此，若以修整器為例子，所應選用之T2大小是盥 該修整器之剩餘制壽命成正相關，亦即，當剛開始使 用較新的雜H來雜研縣時，其耗效果較好而可 使:磨錢持難移除料，故考慮將停止供給研磨液 之h間延長（亦即制較大數值的Τ2)，而避免浪費研 磨液巾&個研磨過&巾使用修整器來修整研磨塾的時 巧比二巧與T2大小相關’例如可在通研磨液進行研磨 =歸整11來崎轉4在未私研磨液進 行研磨時，則停止修整器之修整動作。所磨液 之讀，㈣對於研磨效料著決定性 之衫響因此’本發明之方料包括赠上㈣之方式， 丄/上IZr UMCD-2007-0260 24965twf.doc/n 之::狀況，並視研磨塾之表面狀況來及時 的大小。所_上即時之方式例如是於研磨 狀況。當铁貞U 11研磨墊之表面 本發明之方連續或分批進行研磨時， 來^判研祕^⑯照前—批晶圓研磨後檢測的狀況 饋控制下-㈣、狀似 ，之外#響研磨製程的其他變數包i研磨頭 慎:m圓與研磨墊的向心加速度等等，亦應審 ㈣估亚納人考量。舉例而言，在雜性cmp設備中， ^形研磨機台的向心加速度即是影響研磨移除速率的因 ^之- ’根據本發明之較佳實施例，研磨機台之向心加 逮度應例如至少大於或等於20呎/秒平方(ft/sec2)。 總之’本發明提供的研磨祕應方法更可包括所謂 先進製程㈣!(APQ的概念：將與TG相_參數值(例 如：研磨前材料層膜厚、研磨墊之種類或表面狀況等）、 與T1相關的參數值(例如：研磨前機台的空轉時間或每 批研磨晶圓相隔時間等）、或與丁2相關的參數值(例如： 研磨墊之剩餘使用壽命或修整器的剩餘使用壽命等)傳送 至製耘控制機台，由機台根據每種參數的權重或根據預 疋的公式來計算後，選擇Τ0、τ卜T2或T3之時間範圍， 並配合指定研磨機台之研磨配方。 圖3Α至圖3Ε為依照本發明一較佳實施例所繪示的 内連線的製作流程剖面圖。請參照圖3Α，首先，提供 一基底300至研磨機台31〇中，如其右上側放大圖所示， 11 1337112 LTMCD-2007-0260 24965twf.doc/n 基底300表面上具有至少一介電層302，而介電層302更 具有至少一開口 304。形成導體層306覆蓋於介雷屏3〇2 之開口 304中與介電層302之上。其中，導體層3a〇6例 如為金屬詹，其材質例如為銅或嫣。在形成導體層306 之刖’例如.更可以包括先形成—層阻障層305，共形覆 蓋於開口 304表面上。研磨機台310具有一研磨液供應 器316負責提供至少一種研磨液312至研磨墊314上， 便後續進行化學研磨製程。後續僅繪示出基底部分之剖 面放大圖’而為方便觀看特將上下顛倒。 然後，請參照圖3B，提供研磨液312至研磨墊314 上，以研磨液312配合研磨墊314對導體層3〇6進行初 步的研磨持續第一期間(τι)，而形成導體層3〇6a。 一接下來，請參照圖3C，停止供應研磨液31〇持續第 二期間(T2)’而配合研磨墊312繼續研磨導體層施， 以形成導體層306b β 之後，請參照圖3D，恢復供應研磨液31〇持續第三 期間(Τ3) ’而配合研磨塾312繼續對導體層鳩進行研 磨’以形成導體層306c。 如此重複圖3C至3D的步驟數次，直至去除介電層 上的導體層306c而形成金屬内連線观（圖3e)。 雖然本發明實施例乃是舉平坦化金㈣連線的化學 =械製程為例’但是並不得用以限定本發明，本發明所 展法可適縁前段製財元件淺溝渠隔離製作、 層或鑲嵌結構的製作，以及其他微機電系統平 化或平面顯示器之製作等。 12 1337112 UMCD-2007-0260 24965twf.doc/n 綜上所述，本發明在進行研磨製程時，間歇性停止 研磨液供應，並且停歇週期與期間可以視不同製程條 件、研磨墊狀況、修整器狀況等來做調整，不但可以避 免研磨液之浪費並降低生產成本，而且可以獲得令人滿 意的研磨均勻度與研磨效能。 雖然本發明已以較佳實施例揭露如上，然其並非用 以限定本發明，任何所屬技術領域中具有通常知識者， 鲁 在不脫離本發明之精神和範圍内，當可作些許之更動與 潤娜，因此本發明之保護範圍當視後附之申請專利範一圍 所界定者為準。 【圖式簡單說明】 圖1是習知之一種研磨設備的側面示意圖。 圖2A是依照本發明之一較佳實施例的研磨液供應 方法流程圖。 圖2B是依照本發明之一較佳實施例的研磨製程剖 • 面示意圖。 圖3A至圖3E為依照本發明—較佳實施例所繪示的 金屬内連線的製作流程剖面圖。 【主要元件符號說明】 100 : CMP 設備 102、310 :研磨機台 104、204 :晶圓握柄 106、206、314 :研磨塾 13 1337112 UMCD-2007-0260 24965twf.doc/n 108、316 :研磨液供應器 110、200 :晶圓 112、208、312 :研磨液 114 :修整器 202 :材料層 300 :基底 302 :介電層 304 :開口 305 :阻障層 306、306a、306b、306c :導體層 308 :金屬内連線UMCD-2007-0260 24965twf.doc/n IX. Description of the Invention: TECHNICAL FIELD The present invention relates to a polishing process, and more particularly to a method of supplying a polishing liquid in a polishing process. [Prior Art] Chemical mechanical polishing (CMP) is the only technology that provides full planarization in a very large size integrated circuit (VLSI) process or a very large integrated circuit (ULSI) process. Basically, chemical mechanical polishing uses a principle similar to that of mechanically ground grinding wheels. With the aid of chemical substances, the rough surface of each layer deposited on the wafer is flat and smooth. Wafer flattening is one of the factors that affect the accuracy of alignment. If the wafer flattening step is not effective, not only can the mask in the alignment system be accurately aligned on the wafer, but also The possibility of errors during the process will increase relatively. In addition to providing a comprehensive planarization effect on the surface profile of the wafer, chemical mechanical polishing (CMP) can also be applied to the fabrication of damascene structures for vertical and horizontal metal interconnects (Interconnects), shallow trench isolation for components in the front-end process, and others. The flattening and production of advanced components. In the process of multi-layer interconnects, the flattening step is even more indispensable. Figure 1 illustrates a conventional grinding apparatus. The core unit of the grinding apparatus is an automatically rotating polishing table 102 and a wafer holder 104'. Usually, the polishing table 1〇2 rotates with the grinding crucible mounted on the polishing table 〇2. The position of the wafer grip 104 is adjustable, and the supplied wafer 110 is held by the wafer grip 104. A polishing liquid UMCD-2007-0260 24965twf.doc/n The supplier 108 is mounted on the side of the polishing table to provide a polishing liquid 112 for polishing. During the polishing process, the wafer grip 104 determines that the crystal _ ιι is in contact with the polishing pad 106 and is polished by the polishing pad 106 and the polishing liquid 112 for the crystal/circle. In addition, a trimmer 114 is typically used to adjust the condition of the polishing pad during CMP device motion.曰 Eight appropriate research or adjustment of the slurry into a knife and other factors, for the performance and effectiveness of the grinding has a very significant shadow. In addition, since the used polishing liquid is easy to pollute the environment, -β: to make the high current, carefully evaluate the amount of the slurry / / 1 is also necessary. SUMMARY OF THE INVENTION The discontinuous supply method can be used to periodically remove the J liquid, which not only avoids excessive use of the polishing liquid, but also reduces the use of the polishing liquid and provides stability and uniformity. The three-stage ship-to-ship supply method, to the grinding:: one layer of the wafer; (8) to provide at least one type of polishing liquid to carry out the polishing process, initially grinding for the first period τι, ~ grinding the layer of material on the 81 Will be connected to the grinding on the machine ^ (4) _ 彳 grinding; (8) stop the supply of the slurry and continue to research (4) Τ 2; (6) restore the supply of the training and continue to grind: Τ 3: (quant step ·) increase Up to the time: ···, the entire grinding process performed on the material layer is required to be between UMCD-2007-0260 24965t^fd〇c/n and step (b), (C) ' (d) The first, second, and third periods required are ΤΙ, T2, and T3', respectively, then TO = ΤΙ + η* (T2+T3), where η represents the total number of times steps (c) through (4) are total, and η Is an integer greater than 〇. In an embodiment of the invention, the enthalpy 1 is at least about 10% or more of TO. In the preferred case, T1 is about 50% of το. The range of το is, for example, between about 1-200 seconds. In an embodiment of the invention, the above T2 is about ～5% of το. In the preferred case, T2 is 1% D, 2% T0 or 3% Τ0. / In an embodiment of the invention, the material layer comprises at least one metal layer or at least one dielectric layer, or a metal layer or a combination of dielectric layers. In one embodiment of the invention, the T2 size described above is positively correlated with the remaining useful life of the polishing pad. In an embodiment of the invention, the steps (b) to (d) above further include simultaneously using a trimmer to adjust the condition of the polishing pad. The T2 size is positively related to the remaining service life of the dresser. In an embodiment of the invention, the size of the above-mentioned Τ〇 and Τ2 can be adjusted in time by means of online monitoring. In one embodiment of the invention, the above method has a centripetal acceleration greater than or equal to 20 呎/sec squared (ft/sec 2 ). When the polishing process is carried out, the slurry supply method provided by the present invention intermittently stops the supply of the slurry, and the period and period of the suspension can be adjusted depending on the conditions, thereby avoiding waste of the slurry and reducing the production cost. The above features and advantages of the present invention will become more apparent from the following detailed description of the preferred embodiments of the invention. 2A is a flow chart of a polishing liquid supply method according to a preferred embodiment of the present invention, and FIG. 2B is a schematic cross-sectional view showing a polishing process according to a preferred embodiment of the present invention. In the preferred embodiment, a three-stage slurry supply method is used. Referring to Figures 2A and 2B, first, in step 20, a wafer 200 having a material layer 202 is provided. The material layer 202 can be at least one metal layer, or at least one dielectric layer, or a combination of a metal layer and a dielectric layer. The wafer can be various forms of semiconductor wafers or non-semiconductor wafers, such as Shixi wafers, Shixi 锗 wafers, insulating layer slabs, glass substrates, composite materials. Wait. In step 22, at least one slurry 208 is provided and a grinding process is performed for the material layer 202 of the wafer 200 for the first period. The at least one polishing liquid 208 may be a polishing slurry for a metal or a dielectric layer containing abrasive particles, or may be deionized water; the polishing process may be chemical mechanical polishing, mechanical grinding or electrochemical mechanical polishing. During polishing, the wafer grip 204 determines that the material layer 202 on the wafer 200 is in contact with the polishing crucible 206. The wafer 200 is ground through the polishing crucible 206 and the polishing fluid 208. In step 24, the supply of the slurry is stopped and the grinding is continued for the second period. Next, in step 26, the supply of the slurry is resumed and the grinding continues for the third period. Thereafter, in step 28, steps 24 through 26 are repeated several times until the end of the grinding is reached or the desired grinding effect is achieved, for example, UMCD-2007-0260 24965 twf.doc/n specific film thickness. The polishing process may be a polishing process having a time control mode (end time) or an end point mode; and the inter-item control mode is a grinding method that specifies the time required for the whole or part of the grinding process. The endpoint detection mode relies on the 4 detectors installed in the grinding bowl to measure the thickness of a specific layer on the wafer to capture the end point, and then grind after the amount reaches the end point (over-polish) ) -period of time. If the time required for the entire chemical mechanical polishing process performed for this material layer 202 is regarded as T0, and the first, second, and third periods required for the three stages are respectively designated as ΤΙ, T2, and T3, then: TO = Tl+n*(T2+T3) where η represents the total number of times steps 24 to 26 are performed in total; η is an integer greater than 〇. According to a preferred embodiment of the present invention, Τ1 is at least about 〇0% or more, preferably T1 is about 50% of T0, to achieve the effect of sufficiently wetting the 0 pad and avoiding dry grinding of the wafer. The step or range in which the T2 is about 0. 5% to 75 % ° or more is merely an example of the preferred embodiment of the present invention', but the scope of the present invention is not limited thereto. + For example, when T2 is 1% D = 0 ' repeat 4 times (n = 4) and D3 is 11% Τ 0, Τ 1 is 52% Τ 0. For example, when Τ2 is 2% Τ0 ′ repeated 3 times (η = 3) and D 3 is 15% Τ 0, ΤΙ is 49% Τ 0. For another example, when Τ2 is 3% Τ0, it is repeated 3 times (11:=3) and D3 is 14% 〇, and τΐ is 49% Τ0. For another example, when Τ2 is 4% Τ0, it is repeated 3 times (n==3) iT3 UMCD-2007-0260 24965twf.d〇c/n is 12%T0, and T1 is 52% 〇. Repeat 3 times (η = 3) and Τ3, for example, when T2 is 5%τ〇, it is 12%Τ0, and Τ1 is 49%τ〇. According to the experimental data that has been ordered, that is, the continuous supply of the slurry is as follows: 2% for D2: = thief for chemical money research (4), in (4) = (_: uniformity (7) and number of defects (four) eit conns The performance is consistent with the control group, good, =. Even when evaluating the center-to-edge index (= • mdex), the results of chemical mechanical polishing with a T2 of 2% and a condition of 5% are sufficient to meet the demand for the red process. In addition, if the current compatible process is used, the conditions for Τ2 are 2%, 3% and 5%, and it is estimated that the grinding cost per wafer can be reduced by >gt; about 12%, 18% or even The consumption of 3〇%, that is, the reduction of grinding; ^ makes: the amount 'and can save millions of dollars per year. Generally speaking, for a specific material layer to be ground, the thickness of the material layer is measured ex-situ in the machine before and after the grinding, and the type of the polishing pad used and the polishing liquid used are used. The film removal rate can be calculated from the difference in film thickness before and after grinding, and the thickness of the material layer before polishing can be used as the basis for determining the total polishing time (τ〇) (that is, determining the length of time of TO). range). The polishing pad used in the present invention is, for example, a polishing pad having a ruthenium abrasive granule or a surface having different patterns or grooves. Of course, other variables that affect the CMP process, such as the pressure applied by the grinding head, the tumbling speed of the wafer and the grinding burr, etc., should also be taken into consideration to adjust the range of τ〇. UMCD-2007-0260 24965twf.doc/i size. For example, the range of TO is between about 1-200 seconds. The length of the T2 can be adjusted flexibly depending on the condition of the polishing pad or the condition of the dresser; it can be adjusted even if the condition or process of the batch of wafers is needed immediately or in the future. More specifically, when choosing the scope of T2, there are many different factors to consider. Taking the polishing pad as an example, the size of T2 should be chosen to be linearly positively correlated with the remaining service life of the grinding crucible used. This is because, when the polishing pad is first used, the condition of the surface of the polishing pad is generally satisfactory, and the polishing removal efficiency is better. At this time, it is considered to extend the time for stopping the supply of the polishing liquid (that is, using a larger value) T2), while avoiding the waste of the use of the slurry. In the same way, the rough surface of the polishing pad is beneficial to improve the material removal rate, and the fineness of the polishing material is improved. ((3) nditi〇ning can increase the surface roughness of the polishing pad and make the material removal rate uniform. . Therefore, if the dresser is taken as an example, the size of T2 should be selected to be positively related to the remaining life of the dresser, that is, when the newer hybrid H is used to start in the county, the cost is better. However, it can be made that it is difficult to remove the material, so it is considered to stop the supply of the slurry (the larger value of Τ2), and avoid the waste of the abrasive cloth & a ground & The timing of dressing the grinding burrs is more than that of the T2 size. For example, it can be ground in the grinding liquid = rounded up to 11 to reverberate. 4 When the grinding is not carried out by the private grinding liquid, the dressing action of the dresser is stopped. The reading of the grinding fluid, (4) the decisive shirting of the grinding effect. Therefore, the method of the invention includes the method of giving (4), 丄/上IZr UMCD-2007-0260 24965twf.doc/n:: condition, and Depending on the surface condition of the grinding rakes, the size is timely. The immediate way is for the grinding condition. When the surface of the shovel U 11 polishing pad is continuously or batch-grinded by the method of the present invention, the condition of the inspection after the grinding of the batch wafer is controlled under the control - (4), like #响磨工艺Other variables package i grinding head caution: m circle and the centripetal acceleration of the polishing pad, etc., should also be judged (four) to estimate the Yana people consideration. For example, in a hybrid cmp device, the centripetal acceleration of the ^-shaped grinding machine is the factor that affects the rate of removal of the grinding - in accordance with a preferred embodiment of the present invention, the centripetal force of the grinding machine is increased. The degree should be, for example, at least greater than or equal to 20 呎/sec squared (ft/sec 2 ). In summary, the grinding secret method provided by the present invention may further include a so-called advanced process (4)! (APQ concept: _ phase value with TG (for example: film thickness of the material layer before polishing, type of polishing pad or surface condition, etc.), Parameter values associated with T1 (eg, idle time of the machine before grinding or time interval between batches of polished wafers, etc.), or parameter values associated with D2 (eg: remaining life of the pad or remaining use of the dresser) The life is transmitted to the control unit, and the machine selects the time range of Τ0, τ, T2 or T3 according to the weight of each parameter or according to the formula of the preset, and cooperates with the grinding of the specified grinding machine. Figure 3A to Figure 3 are cross-sectional views showing the manufacturing process of an interconnect according to a preferred embodiment of the present invention. Referring to Figure 3, first, a substrate 300 is provided to the grinder 31, as in the upper right. As shown in a side enlarged view, 11 1337112 LTMCD-2007-0260 24965 twf.doc/n The substrate 300 has at least one dielectric layer 302 on its surface, and the dielectric layer 302 further has at least one opening 304. The conductor layer 306 is formed to cover the dielectric layer. Opening of screen 3〇2 Above the dielectric layer 302, the conductor layer 3a 〇 6 is, for example, a metal material, such as copper or tantalum. After forming the conductor layer 306, for example, a layer may be formed first. 305, conformally covering the surface of the opening 304. The grinding machine 310 has a slurry supply 316 for providing at least one of the polishing liquid 312 to the polishing pad 314 for subsequent chemical polishing process. The cross-sectional enlarged view 'is turned upside down for convenience of viewing. Then, referring to FIG. 3B, the polishing liquid 312 is provided on the polishing pad 314, and the polishing liquid 312 is matched with the polishing pad 314 to perform preliminary grinding on the conductor layer 3〇6. A period (τι) is formed to form the conductor layer 3〇6a. Next, referring to FIG. 3C, the supply of the polishing liquid 31 is stopped for the second period (T2)', and the polishing pad 312 is continuously used to polish the conductor layer to form After the conductor layer 306b β, referring to FIG. 3D, the supply of the polishing liquid 31 is resumed for the third period (Τ3)', and the polishing layer 312 is continuously polished to form the conductor layer 306c. Thus repeating FIGS. 3C to 3D The steps are repeated several times until the conductor layer 306c on the dielectric layer is removed to form a metal interconnect view (Fig. 3e). Although the embodiment of the present invention is a chemical/mechanical process for flattening gold (4) wiring, It should not be used to limit the invention. The method of the present invention can be applied to the fabrication of shallow trenches of the front-end wealth-making components, the fabrication of layers or mosaic structures, and the fabrication of other MEMS flat-panel or flat-panel displays, etc. 12 1337112 UMCD-2007 -0260 24965twf.doc/n In summary, the present invention intermittently stops the supply of the slurry during the grinding process, and the period and period of the stop can be adjusted according to different process conditions, the condition of the polishing pad, the condition of the dresser, and the like. Not only can the waste of the slurry be avoided and the production cost can be reduced, but also satisfactory grinding uniformity and grinding efficiency can be obtained. Although the present invention has been disclosed in the above preferred embodiments, it is not intended to limit the scope of the present invention, and it is intended to be within the spirit and scope of the invention. Runna, therefore, the scope of protection of the present invention is subject to the definition of the attached patent application. BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a schematic side view of a conventional grinding apparatus. Fig. 2A is a flow chart showing a method of supplying a slurry according to a preferred embodiment of the present invention. Figure 2B is a schematic cross-sectional view of a polishing process in accordance with a preferred embodiment of the present invention. 3A through 3E are cross-sectional views showing a manufacturing process of a metal interconnect in accordance with a preferred embodiment of the present invention. [Description of main component symbols] 100: CMP equipment 102, 310: grinding machine table 104, 204: wafer grips 106, 206, 314: grinding crucible 13 1337112 UMCD-2007-0260 24965twf.doc/n 108, 316: grinding Liquid supply 110, 200: wafers 112, 208, 312: polishing liquid 114: trimmer 202: material layer 300: substrate 302: dielectric layer 304: opening 305: barrier layer 306, 306a, 306b, 306c: conductor Layer 308: Metal interconnect
1414