1286961 九、發明說明： 【發明所屬之技術領域】 本發明是有關料導體製造技術，且制是細於__湖於一化 學機械研磨製程機台中之化學機械研磨製程之移除率的方法。 【先前技術】 化學機械研磨(chemical mechanical polishing ;於下文中簡稱為CMp)勢 程為當今用以平坦化如形成於半導體基底上絕緣層及傳導層之常用製程： 如石夕、二氧切、_、銅雜特料之研制通常藉㈣磨塾及適當 漿之結合所完成。 ^ 此外，於研磨過程中需藉由一晶圓載具以把持晶圓並使之面向散饰有 研漿之研磨墊上，此研磨墊則設置於一旋轉型或線性移動型之平台上。 一般而s，CMP製程係牽涉如研磨墊之線性或旋轉速度、晶圓載具之 下壓力、研磨時間、CMP研漿流速與化學狀況等不同的控制參數，藉由把 持及旋轉一半導體晶圓而機械性地研磨之。然而，此些參數通常隨著不同 CMP製程而有所改變，以致於所得到之移除率有所不同。對於一 〇娜製 私而§，上述所得到移除率係為必要且需應用於此CMp製程中以決定 所須之研磨時間。 、 壬 通常，於一 CMP製程機台中僅施行如一内層介電層CMp製程、金屬 層間介電層CMP冑程錢溝槽隔雜CMP餘之單-CMP餘。經由製 程轉換可將此CMP製雜纟雛細磨其働雌_蚊材料(例如介 電材料或金屬材料)之特定CMP製程。如此，當有需要時，CMp製程機台 可轉換其所施行之CMp製程以適度消除產射其他特定製程之待處理晶圓 (WIP)。故轉換上述CMP製程機台所施行之CMP製程時，便需要一新c總 製程之移除率。 第1圖為一流程圖，用以說明依據習知技術轉換一 CMP製程機台中所 0503-9683TWF(5.0) 5 1286961 製狀流程。於步驟Si中，首先利用一 cMp製程機台施行具有 CMP :磨二【二^製程,研磨至少—控制晶圓’上述控制晶圓於 響則爰之私可經由一量測裝置測得叫旱到一厚度差。於步驟幻 除率程時間及上述控制晶圓厚度差以得到此第—製程之移 製姆，此娜除率 农往铖口中以鈿仃此第一製程，如步驟S4所示。如果t 述移除率經判斷不合乎製程要求，操作者(例如 :::::::: cmp ------ 複步驟°胁_5巾撤完畢後，重 並解決之。H 除率以確任機台問題是否經確認 .CMP ,, 產口 π πΜ、 減CMP餘機台所研磨之後續 則繼之狀_為閒置 續產口日圓料/ 可為此⑽製程機台所研磨之後 *虐4此曰π 屬相同之CMP製程，此CMP製程機台將可繼 :數之心;二步驟S9中所示。若後續產品晶圓係屬於具有不同控制1286961 IX. Description of the Invention: [Technical Field of the Invention] The present invention relates to a material conductor manufacturing technique, and is a method which is finer than the removal rate of a chemical mechanical polishing process in a chemical mechanical polishing process machine. [Prior Art] Chemical mechanical polishing (hereinafter referred to as CMp for short) is a common process for planarizing, for example, an insulating layer and a conductive layer formed on a semiconductor substrate: such as Shi Xi, dioxotomy, _, the development of copper special materials is usually done by (4) the combination of grinding and proper pulp. ^ In addition, in the grinding process, a wafer carrier is used to hold the wafer and face the polishing pad with the slurry, which is disposed on a rotating or linear moving platform. Generally, the CMP process involves different control parameters such as the linearity or rotational speed of the polishing pad, the pressure under the wafer carrier, the polishing time, the CMP slurry flow rate, and the chemical condition, by holding and rotating a semiconductor wafer. Mechanically ground. However, these parameters typically vary with different CMP processes such that the resulting removal rates vary. For a private system, §, the above removal rate is necessary and needs to be applied to this CMp process to determine the required grinding time.壬 Generally, in a CMP process machine, only a single-CMP residue such as an inner dielectric layer CMp process, a metal interlayer dielectric layer CMP process, a trench isolation CMP, and the like are performed. This CMP mash can be used to process a specific CMP process for 働 female-mosquito materials (e.g., dielectric materials or metallic materials). Thus, when needed, the CMp process machine can convert its CMp process to moderately eliminate wafers to be processed (WIP) for other specific processes. Therefore, when converting the CMP process performed by the above CMP process machine, a new c total process removal rate is required. Figure 1 is a flow chart for explaining the process of converting 0503-9683TWF(5.0) 5 1286961 in a CMP process machine according to the prior art. In step Si, first, using a cMp process machine to perform CMP: grinding two [two processes, grinding at least - controlling the wafer], the control wafer is in the sound of the sound can be measured by a measuring device To a difference in thickness. In the step of eliminating the lapse time and the above-mentioned control wafer thickness difference to obtain the shift of the first process, the grading rate is the first process in the farm, as shown in step S4. If the removal rate is judged to be inconsistent with the process requirements, the operator (for example:::::::: cmp ------ repeat step ° threat _5 towel after the completion of the withdrawal, and solve it. H divide The rate is determined whether the problem of the machine is confirmed. CMP, the production port π π Μ, the subsequent grinding of the CMP machine is followed by the _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Abuse of this 曰 π is the same CMP process, the CMP process machine will be able to follow: number of hearts; two steps S9 shown. If the subsequent product wafers belong to different control
台内而峨㈣概⑽製程機 製程之隸拖 、 70成此CMP製程機台内所施行CMP ⑽製程'，於Γ製中，需再次於此⑽製程機台内施行一新 測得上述姉晶圓^磨^數^研磨至少—控制M，經由—量測裝置 著計瞀制尹拉…、、後之厚度以得到一厚度差。於步驟SU中，接 步驟^ ^判·厚度差以得到此新CMP製程之移除率。於 之_。除物合乎新⑽製程娜(如製程規範 移除#於CMP新製程之要求，則可將此新移除率應用於 0503-9683TWF(5.0) 1286961 此CMP f程機纟中並開始研磨產線令等待新⑽製程之 驟阳所示。若新移除率不合乎晴新製程 J曰曰囫，如步 程師或製程工程師)將被告知且此CMp製程機 呆:(例如設備工 尹止直到確認影響移除率之問題所在，如步/似二示。=程將暫時 之檢查結束後’將再次步驟sl〇〜sl2以得到此新製_之另j S14中 細刪峨顧蝴 =’於-CMP製程機台中轉換CMp製程時，往往需 ，〜阳所顯示之控制晶__程序。上述缝複之步驟中如= 曰曰囡的賴、其於CMP餘雜麟與晶齡面微師a,之 制晶圓的處雖往佔據了可觀的機台製程時間，進而減少此⑽製程機: 於研磨產品晶圓之魏製程時間。此外，由於再—次的測機程序所使^ 控制晶圓數量亦提升了整體製程之成本。 •於美國第6,5_號專利針，Yang #人教導了—種藉由改變權數 (weightmg factor)而有效控制製程時間以製造至少一晶圓之半導體製程。1 可精確控制研磨時間之CMP之移除率的测錢藉由—製程機台提供一預 先預測之製錢相及-滅侧之餘辭。縣，當今㈣卿㈣預測 之製程速率可藉由具有第-可_數之第—方程式，制上述預先預 測之製程辭及碱制之製程鱗偶變_制。麟，依據輸入於 此製程機台内之上述當今預·程速率及—財目標值即可得到一製程時 間。最後’晶圓可藉由此製程咖並採用此製程機台而製造而成。然而， 於美國第6,514,861號專利案中僅針對一相同之CMp製程提供了精確控制 對於後縯晶圓的研磨時間的方法。 有紐此，吾等實需要可減少CMp機台之非製程時間、其控制晶圓用 U整體成权酬CMP触絲率及麵CMp製减纟雌行CMp 製程的方法。 0503-9683TWF(5.0) 7 1286961 £發明内容】 有鑑於此，本發明的主要目的 移除率的方奴克服前述習知系統及餘之缺預_學機械研磨(CMP〕 本發明的另—目的就找供—種無二外瑜 製程機台内轉換化學機械研磨製程之方法。"製程而於化學機械研磨 因此，本發明之預測化學機械研磨移 一化學機械研磨製程機台中施行 万法’包括下列步驟··於 除率；提供-轉換係數’·以及藉由該轉換^械乘·製程以得到-量測移 一預測移除率。 、’、：上該量測移除率以產生 此外，本發明之轉換化學機械研磨製程 化學機械研磨製程機台中施行 法，包括下列步驟··於一 圓以得到-量測移除率，·施行不同於研磨製程研磨至少-控制晶 化學機械研磨製程，其施行方式包括 =學機械研磨製程之-第二 以產生-酬移除率；於上述化 ^ 移除率乘上-轉換係數 以決定-所需之研磨時間；以及製程機台中施行該酬移除率 於本發明中之轉換係數可‘下二^^機^磨製程。 研磨製程中之-控制參數做為變數且 k擇上述第-化學機械 數值且研磨至少-控_以得_===，改變上述變數之 式；選擇上㈣-化學機械研麵程控=之移轉度間之-關係 步驟⑻直到得到第-化學機械研磨製程令之::數做為-新變數並重複 移除厚度狀__、式；將上軸_乘^=錄與該控制晶圓之 有該第一化學機械製程相同種類控制參數于一1—工作函數；提供具 程；分別將上述第一轉機械研磨製“ 二化學機械研磨製 個已知控财錄獻卫作自數t卩射彳—細研賴財之複數 果，·以及以上述第-計算結果除 算結果及-第二計算結 十异結私得到-轉換係數。 0503-9683TWF(5.0) 1286961 再者，本發明亦提供了適用於旋轉型化學機械研磨製程機台之一工作 函數’用以計算其轉換化學機械研磨製程中所需之轉換係數之用 作函數為： *223-0.605e^x'3,4^/2,117-*]*[l 〇85-〇 3026~^_80/98*39^] 以雇施(Z韻，；其中χ為第—類控制參數、γ為第二類控制她 以及Z為第三類控制參數。 上述第-類控制參數可為-晶圓载具所施加之下麗力（d〇wn醫d force) ’社述第二酿制參數可為—化學顧研磨研漿之錢，而上述第 二類控制參數則可為一研磨墊之旋轉速度。 為了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，並配合所附圖示，作詳細說明如下： 【實施方式】 本發明之實補將配合第2圖至第4 詳細敘述如下。於第2圖 中’係顯不-CMP製程機台之部分圖示。此CMp製程機台1〇包含一晶圓 載具12、-研漿供應器14以及一研磨塾16。晶圓載具12係設置於研磨墊 16上並於一 CMp製程中實行下壓力18。此外，於一 CMP製程中上述研漿 供應器14則於適當流速下供應適當之CMp研漿。研雜⑹系設置於可線 性地或旋轉地移動之一平台(未顯示)上。於晶圓a上之一預定之研磨厚度 d可藉由CMP製私機台1〇中之控制器(未顯示)採用藉由於製程機台 10中研磨至少一控制晶圓所得到之一已知移除率以決定移除此厚度d所須 之研磨時間。 幾台中隸換CMPM鋥之方法： 第3圖為一流程圖，用以說明依據本發明之一實施例中於一 cmp製程 機台中轉換CMP製程之流程。於步驟S1G1中，首㈣用_ CMp製程機台 〇503-9683TWF(5.0) 9 1286961 刪參數之第—CMp製程以研㈣—㈣晶圓。此迪控 _之移罐性或補)速度、晶圓载具之施二 =、研漿之化學條件等。而上述第—CMp製程可為用以塗二 乳切、硼磷魏璃(BPSG)或低介電錄峨⑽则之⑽介電層 、^}^r dielectoc，ILD)、金屬層間介電層(inter_metal 祕咖c ; 夂溝槽隔離物(shallow french iS0lati0n 程亦可為研磨塗佈有如銅、鎢或銘等金屬之—金屬cMp製程。此外， 控制晶圓於CMP «前後之厚度可經由一量測裝置測得以得到一厚度差。 於步驟S102中，接著藉由計算製程時間及上述控制晶圓厚度差以得到此第 一製程之量_除率。接著，步驟S1G3侧斷上述制移除率是否合乎此 第:CMP製程之要求(如製程規範之規格）。如果上述量测移除率經判斷為 合乎製程要求，此量測移除率將應此CMp製程機台中以施行此第一製 程，如步驟S舰所示。如果上述量測移除率經判斷不合乎製程要求，操作 者(例如設備工程師或製程工程師)將被告知且此⑽製程機台所施行之 ㈣！程將暫時中止直到確認影響移除率之問題所在，如步驟娜中所 不。並於步驟S105巾之檢查完畢後，重複步驟sl〇1〜sl〇3以得到此第一掣 程之另-移除率以確任機台問題是否經確認並解決之。接著，於步驟㈣ ^判斷產線上技有其他可為此CMPt程機台所研磨之後續產品晶圓。 若無其他可為此CMP製程機台所研磨之後續產品晶圓，將結束上述CMP 製程且此CMP製程機台之狀態將轉為閒置_)，如步驟_中所顯示。 若仍有其他可為此CMP _機台所研磨之後續產品晶圓，於步驟_中 則繼續判斷後續產品晶圓是否為相同製程之晶圓。若此後續產品晶圓仍屬 於相同之CMP製程，此CMP製程機台將可繼續研磨此些晶圓，如步驟sl〇9 中所不。若後續產品晶圓係屬於具有不同控制參數之其他製程的產品晶 ，’便需要將—新移除率以應用於此CMP製程機台内而決定新製輯需之 時間’進而完成此CMP製程機台内所施行CMP製程之轉換。所以，於步 0503-9683TWF(5.0) 1286961 騍S110中，藉由將上述量測移除率乘上一轉換係數以產生一預測移除率。 =步驟S111中，此預測移除率接著代入此CMP製程機台中並開始;磨後 績製程晶®之㈣程進而完成此CMP製程機台中之CMP製轉換。因此， =CMP製程機台可於較先前技術為短之閒置(idle)時間内執行新製程之後 績產品晶圓的研磨。於步驟S112中則接著判斷產線中是否有其他後續產品 晶圓。若判斷為無他晶圓存在，上述CMP製程便可結束，如挪S、L : 不。若判斷仍有後續晶圓到來，則可重複如步驟麵〜S112或s⑽〜謂 中所不之步驟。 通迴CMP移險率之太法： 、於第3 ®所補換CMP製程方法巾之瞻步驟測將透過第 之流程圖詳細加以解說。 ° 於步驟S2〇l中，·g*先提供如前述第一 CMp製程之一量測 者於步驟中則藉由下列步驟以得到控制晶圓之移與 = 間之關係式，其步驟包括： 參數 數；⑻L CMP餘中之—變數w其餘控制參 制晶r之=:r=r⑽—控侧剛此變數與控 (c).選擇第-CMP製程中另—控制參數做為—新變數並重複步驟In the Taiwanese (4) general (10) process mechanism, the CMP (10) process is carried out in 70% of the CMP process machines. In the system, it is necessary to perform a new measurement in the (10) process machine. The wafer is polished to at least the control M, and the thickness of the yoke is measured by the measuring device to obtain a thickness difference. In step SU, the difference in thickness is determined to obtain the removal rate of the new CMP process. In _. In addition to the new (10) process Na (if the process specification removes # in the new CMP process requirements, this new removal rate can be applied to 0503-9683TWF (5.0) 1286961 this CMP f machine 纟 and start grinding line Wait for the new (10) process to show the sudden yang. If the new removal rate does not meet the sunny process J曰曰囫, such as the stepper or process engineer) will be informed and the CMp process machine stays: (for example, equipment workers Yin Until the problem that affects the removal rate is confirmed, such as step/like two indications. = The process will be temporarily checked after the end of the process. 'The steps will be sl1~sl2 again to get this new system_ another j S14 in the fine deletion 峨 蝴 = = 'When converting the CMp process in the CMP process machine, it is often necessary to control the crystal __ program shown by ~Yang. The above steps of the stitching are as follows: 曰曰囡 曰曰囡 、 其 其 其 其 其 CMP CMP CMP 晶 晶Micro-a division, the wafer fabrication area occupies considerable machine processing time, and thus reduces the (10) process: the process time of grinding the product wafer. In addition, due to the re-testing procedure ^ Controlling the number of wafers also increases the cost of the overall process. • US Patent No. 6, 5_, Yang #人教A semiconductor process that can control the process time to manufacture at least one wafer by changing the weight mg factor. 1 The measurement of the CMP removal rate that can precisely control the polishing time is provided by the process machine. The pre-predicted system of money and the remnant of the side. County, the current (four) Qing (four) predicted process rate can be made by the first-coordinated equation - the above-mentioned pre-predicted process and alkali process The scale change _ system. Lin, according to the above-mentioned current pre-process rate and the target value entered in the process machine can get a process time. Finally, the wafer can be used to make the process and use the process machine. It is manufactured in Taiwan. However, in the US Patent No. 6,514,861, only a method for accurately controlling the grinding time of the wafer for the same CMp process is provided. In view of this, we really need to reduce the CMp machine. The non-process time of the Taiwan, the method of controlling the wafer U, the weight of the CMP, and the method of reducing the CMp process by the CMp system. 0503-9683TWF(5.0) 7 1286961 £Inventive content] In view of this, this Main purpose of the invention In addition to the rate of slaves to overcome the aforementioned conventional system and the lack of pre-study mechanical polishing (CMP) Another object of the present invention is to find a way to convert the chemical mechanical polishing process in the machine. Process and chemical mechanical polishing Therefore, the predictive chemical mechanical polishing shifting process of the present invention is carried out in a chemical mechanical polishing process machine, including the following steps: · removal rate; supply-conversion coefficient'· and by means of the conversion Multiply and process to obtain - the measured shift and the predicted removal rate. , ',: the measurement removal rate is generated to generate the addition, the conversion chemical mechanical polishing process of the present invention is performed in the chemical mechanical polishing process machine, including the following steps ································································································ Multiplying the above-mentioned removal rate by the conversion factor to determine the required grinding time; and the conversion factor in the process machine for performing the removal rate in the present invention can be '2' . In the grinding process, the control parameter is used as a variable and k selects the above-mentioned first-chemical mechanical value and grinds at least - control_ to obtain _===, changing the formula of the above variables; selecting the upper (four)-chemical mechanical surface program control = shift Between the rotation-relationship step (8) until the first CMP process is obtained: the number is - the new variable and the thickness __, the formula is repeatedly removed; the upper axis _ multiplication ^ = recorded with the control wafer The first type of chemical mechanical process has the same type of control parameter in a 1-work function; the process is provided; respectively, the first mechanical mechanical polishing system is used to make a known chemical control record.卩射彳—Study the plural fruit of Lai Cai, and divide the result by the above-mentioned calculation result and - the second calculation result is the same - the conversion coefficient. 0503-9683TWF(5.0) 1286961 Furthermore, the present invention also A function function for calculating the conversion factor required for the conversion CMP process is provided for: *223-0.605e^x'3,4^ /2,117-*]*[l 〇85-〇3026~^_80/98*39^] to hire (Z Yun,; For the first-class control parameter, γ is the second type to control her and Z is the third type of control parameter. The above-mentioned first-class control parameter can be - the wafer carrier is applied under the force of d The second brewing parameter of the society can be - the chemical money of the grinding slurry, and the second type of control parameter can be the rotational speed of a polishing pad. The above and other objects, features and advantages of the present invention can be made. The following is a detailed description of the preferred embodiment, and the following description will be given in detail: [Embodiment] The present invention will be described in detail with reference to FIGS. 2 to 4. In the figure, a part of the CMP process machine is shown. The CMp process machine 1 includes a wafer carrier 12, a slurry supply 14 and a polishing pad 16. The wafer carrier 12 is arranged for grinding. The lower pressure 18 is applied to the pad 16 and in a CMp process. Further, in the CMP process, the slurry supply 14 supplies the appropriate CMp slurry at a suitable flow rate. The grinding (6) system is arranged to be linear or rotatable. Ground moving on one of the platforms (not shown). One of the scheduled studies on wafer a The thickness d can be determined by a controller (not shown) in the CMP machine to determine the removal of the thickness by polishing one of the known wafers in the process machine 10 to at least one of the control wafers. The grinding time required for d. Several methods for replacing CMPM鋥: Figure 3 is a flow chart for explaining the process of converting the CMP process in a cmp process machine according to an embodiment of the present invention. In step S1G1 In the first, the first (four) with _ CMp process machine 〇 503-9683TWF (5.0) 9 1286961 delete the parameters of the first - CMp process to research (four) - (four) wafer. The speed of the _ control tank, or the speed of the wafer carrier, the chemical conditions of the slurry. The first CMp process may be a (10) dielectric layer, a dielectric layer, a metal interlayer dielectric layer for coating a nipple, a borophosphorus (BPSG) or a low dielectric recording (10). (inter_metal secret coffee c; 夂 trench spacers (shallow French iS0lati0n can also be coated with metal such as copper, tungsten or metal-metal cMp process. In addition, the thickness of the control wafer before and after CMP « can be passed through The measuring device determines a thickness difference. In step S102, the amount of the first process is removed by calculating the process time and the difference in the wafer thickness. The step S1G3 is then removed. Whether the rate meets the requirements of this: CMP process (such as the specification of the process specification). If the above measurement removal rate is judged to be in compliance with the process requirements, the measurement removal rate will be applied to the CMp process machine to perform this first The process, as shown in step S. If the above measurement removal rate is judged to be inconsistent with the process requirements, the operator (such as the equipment engineer or process engineer) will be informed that the (4) process performed by the (10) process machine will be temporarily suspended. Until the impact of removal rate is confirmed The problem lies in the step Na, and after the inspection of the step S105 is completed, the steps sl1〇1~sl〇3 are repeated to obtain the other-removal rate of the first process to determine whether the machine problem is After confirming and resolving. Next, in step (4) ^ judge the production line has other subsequent product wafers that can be ground for this CMP machine. If there are no other product wafers that can be polished for this CMP process machine, End the above CMP process and the status of the CMP process machine will be turned to idle_), as shown in step _. If there are other subsequent product wafers that can be ground for this CMP machine, continue in step _ Determining whether the subsequent product wafers are wafers of the same process. If the subsequent product wafers are still in the same CMP process, the CMP process machine will continue to grind the wafers, as in step sl9. Subsequent product wafers are product crystals of other processes with different control parameters, 'there will be a new removal rate to be applied to the CMP process machine to determine the time required for the new process' to complete the CMP process The CMP process is implemented in the station. Therefore, in step 0503-9683TWF (5.0) 1286961 骒S110, a predicted removal rate is generated by multiplying the above-mentioned measurement removal rate by a conversion coefficient. = In step S111, the predicted removal rate is then substituted. The CMP process machine starts in the same stage; the (4) process of the process technology is completed to complete the CMP conversion in the CMP process machine. Therefore, the =CMP process machine can be shorter than the prior art for idle time. After the execution of the new process, the polishing of the product wafer is performed. In step S112, it is determined whether there are other subsequent product wafers in the production line. If it is determined that no wafer exists, the CMP process can be ended, such as moving S, L. : Do not. If it is determined that there is still a subsequent wafer arrival, the steps as in step S~112 or s(10)~ can be repeated. The method of returning to the CMP migration rate: The step-by-step test of the replacement of the CMP process method in Section 3 will be explained in detail through the flow chart. ° In step S2〇1, ·g* first provides one of the first CMp processes as described above, and in the step, the following steps are used to obtain a relationship between the control wafer shift and the =, the steps of which include: Number of parameters; (8) L CMP rest - variable w rest control crystallization crystal = = r = r (10) - control side just this variable and control (c). Select the first - CMP process - control parameters as - new variables And repeat the steps
CMPCMP
CMP -新⑽製程及其控制參數以預測其CMP移除率 1於^中=提供 將第- CMP製程及此新㈣財之複數個已知控繼值代 0503-9683TWF(5.0) 11 1286961 fr計算結果及—第二計算結果。於步驟s施’接著用 d 5 Pt程移除率之—轉換雜便可藉由上述第—計算址果除上 第—計算結果而得到而無須進行額外之研磨操作。 “ 書機触或 1力狐度以及CMP研漿德學躲及流速。 莖 移動(線性或旋轉)速度、晶圓載具之下壓力、及CMP研漿 “專二類控制參數係為影響量測移除率之最大因素。因此，依據本發明 ^娜鮮移除率之方法可選擇包含前述三類之不同總類控制參數作為決 疋上述關係式之主要變數以得到上述工作函數。 、依據柄月之’施例之於—CMp製程機台内轉換c辦製程之方法 適用於轉換研磨塗佈有相似物理性質材料(如介電材料或金屬)之⑽製 f 口此無顧外之峨程序便可完成—CMP製程機台之製程轉換。所 以上述CMP f程機台之製程時間__可大幅提昇並且可以減健制晶 圓之數量以及生產成本。 遞轉型中轉換CMP ,葙夕蛙早， 首先提供複數台執行介電材料CMp製程之旋轉型CMp製程機台，此 些機台例如為顧材料公司所製造之Mirxa⑽設備…般而言，在此每 - Mnra機台係執行如ILD、MD或奶# CMp製程之一且當有需要時可 經由額外峨程序雜至其他製程。主要辟前述CMp雜巾移除率之控 制參數種難騎經之旋觀度、晶關具之下壓相及⑽研裝之流 速。此些控制參數應用於各個製程中之數值則如表！中所示。 表1· 度(rpm) 下壓力(psi) 研漿流(ml/min) ILD CMP —_63 4 150 STI CMP ^_63 4.2 200 IMD CMP ——_108 4.6 100 0503-9683TWF(5.0) 12 1286961 以執行ILD CMP製程之-MilTa CMP設傷為例，並於下文中以驗& A 稱呼之，上述ILD CMP製程之量測移除率可藉由週期性(每天、每半天或已 研磨晶圓之數量)測機程序所制。而欲將上述MiitaA所執行之iLDcMp 製程轉換成IMD或STI CMP製程時，便需要—轉換係數。此轉換係數則可 依據前述本發明之預測CMP移除率方法所獲得。 百先，選定如晶圓載具之下遷力的一第一類控制參數作為變數且固定 二他控制錄錄概賴辦63鱗麵及CMP補錢為15〇毫升 /母分鐘)且於改變不同之下動數值(3〜6碎坪方英吸)之狀態下研磨至少一 控制晶圓以分別得到-移除厚度。如此，如第5圖所示，藉由如指數衰退 之-數學模組的迴歸，便可得到此驗a A中移除率(_vai論；跋)與晶 圓載/、下壓力X的關係式’並表現出跋正比於(χ34),2』 係。 · <關 接著’選定如CMP研驗速的—第二魅财數料魏且固定其他 控，參數值(旋轉速賴定於63轉/每分鐘及晶_具之下動為4.2碎坪 方英吸)且於改變不同之研漿流速(5〇〜25〇毫升/每分鐘)之狀態下研磨至少 ^控^推分_卜移除厚度。如此，如第6 _示，藉由如指數衰 ^之一數學模組的迴歸，便可得到此Mirm Α中移除率(職_咖㈣ ” MP研水々丨[速γ的關係式，並表現出反^正比於1⑽胤吻 之關係。 再來’選定如研磨墊旋轉速度的—第三類控制參數作為變數且固定其 =(|參數值(CMP研漿流速為15G毫升/每分鐘及晶_具之下壓力為4.2 二磁方魏)膽改變不同之研磨墊旋轉速度⑼〜⑽轉海分鐘)之狀態下 控制晶圓以分別得到—移除厚度。如此，如第7圖所示，藉由 如祕^退之-數學模組的迴歸，便可得到此驗& A㈣除抑㈣㈣ rae’RR)與研磨塾旋轉速度z的關係式，並 U87-0.719e-^3〇4)^M# 〇 比於 0503-9683TWF(5.0) 13 1286961 —因此，嫌率娜__之主雜够朗之— 可猎由上述三棚赋_麵制並如第〗式所絲。 ’ ’CMP - new (10) process and its control parameters to predict its CMP removal rate 1 in ^ = provide the first CMP process and this new (four) financial multiple known control value generation 0503-9683TWF (5.0) 11 1286961 fr Calculate the result and - the second calculation result. In the step s, then the d 5 Pt removal rate-conversion can be obtained by dividing the above-mentioned calculation result by the first calculation result without performing an additional grinding operation. “Book touch or 1 force fox and CMP slurry learning and flow rate. Stem movement (linear or rotary) speed, under-wafer carrier pressure, and CMP slurry “specialized control parameter system for impact measurement The biggest factor in removal rate. Therefore, according to the method of the present invention, the different total class control parameters of the above three types can be selected as the main variables of the above relationship to obtain the above work function. According to the method of “handling month”, the method of converting the c-process in the CMp process is suitable for converting and coating the material with similar physical properties (such as dielectric material or metal). After that, the program can be completed - the process conversion of the CMP process machine. Therefore, the process time of the above CMP f-machine can be greatly improved and the number of crystals and the production cost can be reduced. In the process of transformation, CMP, 葙夕蛙, first provided a plurality of rotary CMp process machines for performing dielectric material CMp processes, such as Mirxa (10) devices manufactured by Gus Materials Co., Ltd. Each - Mnra machine performs one of the processes such as ILD, MD or Milk # CMp and can be mixed to other processes via additional tampering when needed. The control parameters of the aforementioned CMp rag removal rate are mainly the degree of difficulty in riding, the pressure phase under the crystal cutting device, and (10) the flow rate of the research and installation. The values of these control parameters applied to each process are as shown in the table! Shown in . Table 1. Degrees (rpm) Pressure (psi) Slurry flow (ml/min) ILD CMP —_63 4 150 STI CMP ^_63 4.2 200 IMD CMP ——_108 4.6 100 0503-9683TWF(5.0) 12 1286961 to perform ILD The CMP process-MilTa CMP set-up is taken as an example, and is referred to below as the test & A, the measurement removal rate of the above ILD CMP process can be cyclical (daily, half-day or the number of wafers that have been ground) ) Tested by the machine program. To convert the iLDcMp process performed by MiitaA to an IMD or STI CMP process, a conversion factor is required. This conversion factor can then be obtained in accordance with the predictive CMP removal rate method of the present invention described above. Hundreds of first, select a first type of control parameters such as the relocation of the wafer carrier as a variable and fixed the second control record, the average of 63 scales and CMP money is 15 〇 ml / mother minutes) and change At least one control wafer is ground in the state of the lower value (3 to 6 slabs) to obtain - remove the thickness, respectively. Thus, as shown in Fig. 5, by the regression of the mathematical module such as exponential decay, the relationship between the removal rate (_vai theory; 跋) and the wafer load/down pressure X in this test can be obtained. 'And it shows that it is proportional to (χ34), 2′′. · < off then 'selected as CMP research speed' - the second charm money material Wei and fixed other control, parameter value (rotation speed is set at 63 rev / min and crystal _ under the movement is 4.2 laps Fang Ying suction) and grinding at least the state of the slurry flow rate (5 〇 ~ 25 〇 / min) at least ^ control ^ push points _ remove the thickness. Thus, as shown in the sixth example, by the regression of a mathematical module such as exponential decay, the relationship of the removal rate of the Mirm(Α) can be obtained. And it shows that the inverse is proportional to the relationship between 1(10) and kiss. Then 'select the rotation speed of the polishing pad—the third type of control parameter as the variable and fix it = (| parameter value (CMP slurry flow rate is 15G ml / min And the pressure under the crystal _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ It can be seen that, by the regression of the mathematical module, the relationship between the test & A (four) divide (4) (four) rae 'RR) and the rotational speed z of the grinding crucible, and U87-0.719e-^3〇 4)^M# 〇 于 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 050 '
Fs!^,ZH^^ 然彳_先ILD CMP製財之已練錄从欲賴c 中(在此例如為STI或細CMP)之已知控制參數值代入此工作函數以得 到一第-5丨減果以及-第二計算結果，上述已知控制參數剌如表 戶斤示。 然後，用以預測不同CMP製程移除率之轉換係數便可藉由上述原先 CMP製程(如ILD CMP)之第一計算結果除上轉換後cMp製程(如阳或 IMD CMP)之第二計异結果而得到。表2中係顯示了由肋c·轉換至如 CMP或IMDCMP之轉換係數的結果。一般而言，此轉換係數較佳地為大 體介於0.5〜2之間。 表2. 轉換係數 STI CMP 1.12 IMD CMP 1.41 —因此，於Mirm A内轉換CMP製程之移除率可藉由於MhTa A内週期 性(每天、每半天或已研磨晶圓之數量)測機程序所得到其所施行之製 私(如ILD CMP製程)之量測移除率乘上表2中所示之轉換係數而得到。然 後，便可得到一預測移除率並將之代入MilTa A内已轉換其所施行之CMp 製程。因此，Mirra A内所施行之CMP製程可由原先的ILD CMp直接轉換 為STI或IMD CMP製程而無須額外之晶圓研磨程序。故可提升此a 應用於製程之妥善時間(uptime)並減少其所使用控制晶圓之數量以及整體 Mirra A之生產成本。 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 何熟習此技藝者，在不脫離本發明之精神和範圍内，當可作各種之更動與 0503-9683TWF(5.0) 14 ^86961 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為準 【圖式簡單說明】 第1圖為-流程圖，係用以說明依據習知技術之方法於—c 台中轉換㈣製程之流程； 表％機 要結=圖為-CMP製程機台之部分圖示，用以解說CMP製程機台之主 c•製程機台中轉程據本發明之-實施例中之方法於- Γ5 ^ ，係職明第3财步驟sug之詳細施行流程； 係式。7圖賴示各㈣參數魅制晶圓之移除厚賴之個別關 【主要元件符號說明】 14〜研漿供應器； 18〜下壓力； 22〜晶圓； 12〜晶圓栽具； 16〜研磨塾； 20〜CMP研襞· d〜研磨厚度。Fs!^,ZH^^ Then _First ILD CMP has been programmed to substitute the value of the known control parameter from the c (in this case, STI or fine CMP) into this work function to get a -5丨 丨 以及 and - the second calculation result, the above known control parameters are as shown in the table. Then, the conversion coefficient used to predict the removal rate of different CMP processes can be divided by the first calculation result of the above-mentioned original CMP process (such as ILD CMP), except for the second calculation of the up-conversion cMp process (such as yang or IMD CMP). The result is obtained. Table 2 shows the results of conversion from rib c· to conversion coefficients such as CMP or IMDCMP. In general, the conversion factor is preferably between 0.5 and 2 in general. Table 2. Conversion Factor STI CMP 1.12 IMD CMP 1.41—Thus, the removal rate of the CMP process in the Mirm A can be determined by the periodicity of the MhTa A (daily, half-day or number of wafers that have been ground). The measurement removal rate obtained by the manufacturing process (such as the ILD CMP process) obtained by multiplying it by the conversion factor shown in Table 2 is obtained. Then, a predicted removal rate is obtained and substituted into the CMp process that has been converted in MilTa A. Therefore, the CMP process performed in Mirra A can be directly converted from the original ILD CMp to an STI or IMD CMP process without additional wafer grinding procedures. This improves the uptime of the application process and reduces the number of control wafers used and the overall Mirra A production cost. Although the present invention has been disclosed in the above preferred embodiments, it is not intended to limit the invention, and it is possible to make various modifications with 0503-9683TWF (5.0) without departing from the spirit and scope of the present invention. 14 ^86961 Retouching, therefore, the scope of protection of the present invention is defined by the scope of the appended patent application [Simplified description of the drawings] Figure 1 is a flow chart for explaining the method according to the prior art. -c Taichung conversion (four) process flow; table % machine junction = picture is - part of the CMP process machine diagram, used to illustrate the main process of the CMP process machine in the process of the machine according to the invention - embodiment The method of the method is in - Γ 5 ^, the detailed implementation process of the third step of the sug. 7图赖示 each (4) parameter charm wafer removal of the individual depends on [main component symbol description] 14 ~ slurry supply; 18 ~ lower pressure; 22 ~ wafer; 12 ~ wafer planting; 16 ~ Grinding 塾; 20~CMP mortar · d ~ grinding thickness.
0503-9683TWF(5.〇) 150503-9683TWF(5.〇) 15