TW201112302A - Methods for constructing an optimal endpoint algorithm - Google Patents

Methods for constructing an optimal endpoint algorithm Download PDF

Info

Publication number
TW201112302A
TW201112302A TW099121511A TW99121511A TW201112302A TW 201112302 A TW201112302 A TW 201112302A TW 099121511 A TW099121511 A TW 099121511A TW 99121511 A TW99121511 A TW 99121511A TW 201112302 A TW201112302 A TW 201112302A
Authority
TW
Taiwan
Prior art keywords
end point
group
algorithm
endpoint
delay
Prior art date
Application number
TW099121511A
Other languages
Chinese (zh)
Other versions
TWI480917B (en
Inventor
Jiang-Xin Wang
Andrew James Perry
Vijayakumar C Venugopal
Original Assignee
Lam Res Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Lam Res Corp filed Critical Lam Res Corp
Publication of TW201112302A publication Critical patent/TW201112302A/en
Application granted granted Critical
Publication of TWI480917B publication Critical patent/TWI480917B/en

Links

Classifications

    • H—ELECTRICITY
    • H10—SEMICONDUCTOR DEVICES; ELECTRIC SOLID-STATE DEVICES NOT OTHERWISE PROVIDED FOR
    • H10P—GENERIC PROCESSES OR APPARATUS FOR THE MANUFACTURE OR TREATMENT OF DEVICES COVERED BY CLASS H10
    • H10P50/00—Etching of wafers, substrates or parts of devices
    • H10P50/20—Dry etching; Plasma etching; Reactive-ion etching
    • H10P50/24—Dry etching; Plasma etching; Reactive-ion etching of semiconductor materials
    • H10P50/242—Dry etching; Plasma etching; Reactive-ion etching of semiconductor materials of Group IV materials
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01J—ELECTRIC DISCHARGE TUBES OR DISCHARGE LAMPS
    • H01J37/00—Discharge tubes with provision for introducing objects or material to be exposed to the discharge, e.g. for the purpose of examination or processing thereof
    • H01J37/32—Gas-filled discharge tubes
    • H01J37/32917—Plasma diagnostics
    • H01J37/32935—Monitoring and controlling tubes by information coming from the object and/or discharge
    • H—ELECTRICITY
    • H01—ELECTRIC ELEMENTS
    • H01J—ELECTRIC DISCHARGE TUBES OR DISCHARGE LAMPS
    • H01J37/00—Discharge tubes with provision for introducing objects or material to be exposed to the discharge, e.g. for the purpose of examination or processing thereof
    • H01J37/32—Gas-filled discharge tubes
    • H01J37/32917—Plasma diagnostics
    • H01J37/3299—Feedback systems
    • H—ELECTRICITY
    • H05—ELECTRIC TECHNIQUES NOT OTHERWISE PROVIDED FOR
    • H05H—PLASMA TECHNIQUE; PRODUCTION OF ACCELERATED ELECTRICALLY-CHARGED PARTICLES OR OF NEUTRONS; PRODUCTION OR ACCELERATION OF NEUTRAL MOLECULAR OR ATOMIC BEAMS
    • H05H1/00—Generating plasma; Handling plasma
    • H05H1/24—Generating plasma
    • H05H1/46—Generating plasma using applied electromagnetic fields, e.g. high frequency or microwave energy
    • H—ELECTRICITY
    • H10—SEMICONDUCTOR DEVICES; ELECTRIC SOLID-STATE DEVICES NOT OTHERWISE PROVIDED FOR
    • H10P—GENERIC PROCESSES OR APPARATUS FOR THE MANUFACTURE OR TREATMENT OF DEVICES COVERED BY CLASS H10
    • H10P14/00—Formation of materials, e.g. in the shape of layers or pillars
    • H10P14/20—Formation of materials, e.g. in the shape of layers or pillars of semiconductor materials
    • H10P14/24—Formation of materials, e.g. in the shape of layers or pillars of semiconductor materials using chemical vapour deposition [CVD]
    • H—ELECTRICITY
    • H10—SEMICONDUCTOR DEVICES; ELECTRIC SOLID-STATE DEVICES NOT OTHERWISE PROVIDED FOR
    • H10P—GENERIC PROCESSES OR APPARATUS FOR THE MANUFACTURE OR TREATMENT OF DEVICES COVERED BY CLASS H10
    • H10P50/00—Etching of wafers, substrates or parts of devices
    • H10P50/20—Dry etching; Plasma etching; Reactive-ion etching
    • H10P50/24—Dry etching; Plasma etching; Reactive-ion etching of semiconductor materials
    • H10P50/246—Dry etching; Plasma etching; Reactive-ion etching of semiconductor materials of Group III-V materials
    • H—ELECTRICITY
    • H10—SEMICONDUCTOR DEVICES; ELECTRIC SOLID-STATE DEVICES NOT OTHERWISE PROVIDED FOR
    • H10P—GENERIC PROCESSES OR APPARATUS FOR THE MANUFACTURE OR TREATMENT OF DEVICES COVERED BY CLASS H10
    • H10P50/00—Etching of wafers, substrates or parts of devices
    • H10P50/20—Dry etching; Plasma etching; Reactive-ion etching
    • H10P50/28—Dry etching; Plasma etching; Reactive-ion etching of insulating materials
    • H10P50/282—Dry etching; Plasma etching; Reactive-ion etching of insulating materials of inorganic materials
    • H10P50/283—Dry etching; Plasma etching; Reactive-ion etching of insulating materials of inorganic materials by chemical means
    • H—ELECTRICITY
    • H10—SEMICONDUCTOR DEVICES; ELECTRIC SOLID-STATE DEVICES NOT OTHERWISE PROVIDED FOR
    • H10W—GENERIC PACKAGES, INTERCONNECTIONS, CONNECTORS OR OTHER CONSTRUCTIONAL DETAILS OF DEVICES COVERED BY CLASS H10
    • H10W20/00—Interconnections in chips, wafers or substrates
    • H10W20/01—Manufacture or treatment
    • H10W20/071—Manufacture or treatment of dielectric parts thereof
    • H10W20/093—Manufacture or treatment of dielectric parts thereof by modifying materials of the dielectric parts
    • H10W20/095—Manufacture or treatment of dielectric parts thereof by modifying materials of the dielectric parts by irradiating with electromagnetic or particle radiation
    • H—ELECTRICITY
    • H10—SEMICONDUCTOR DEVICES; ELECTRIC SOLID-STATE DEVICES NOT OTHERWISE PROVIDED FOR
    • H10W—GENERIC PACKAGES, INTERCONNECTIONS, CONNECTORS OR OTHER CONSTRUCTIONAL DETAILS OF DEVICES COVERED BY CLASS H10
    • H10W20/00—Interconnections in chips, wafers or substrates
    • H10W20/01—Manufacture or treatment
    • H10W20/071—Manufacture or treatment of dielectric parts thereof
    • H10W20/093—Manufacture or treatment of dielectric parts thereof by modifying materials of the dielectric parts
    • H10W20/096—Manufacture or treatment of dielectric parts thereof by modifying materials of the dielectric parts by contacting with gases, liquids or plasmas

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Plasma & Fusion (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Electromagnetism (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Drying Of Semiconductors (AREA)
  • Testing And Monitoring For Control Systems (AREA)
  • General Factory Administration (AREA)
  • Chemical Vapour Deposition (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Container, Conveyance, Adherence, Positioning, Of Wafer (AREA)
  • Indication And Recording Devices For Special Purposes And Tariff Metering Devices (AREA)
  • Complex Calculations (AREA)
  • Combined Controls Of Internal Combustion Engines (AREA)
  • Arrangements For Transmission Of Measured Signals (AREA)
  • Plasma Technology (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

A method for automatically identifying an optimal endpoint algorithm for qualifying a process endpoint during substrate processing within a plasma processing system is provided. The method includes receiving sensor data from a plurality of sensors during substrate processing of at least one substrate within the plasma processing system, wherein the sensor data includes a plurality of signal streams from a plurality of sensor channels. The method also includes identifying an endpoint domain, wherein the endpoint domain is an approximate period within which the process endpoint is expected to occur. The method further includes analyzing the sensor data to generate a set of potential endpoint signatures. The method yet also includes converting the set of potential endpoint signatures into a set of optimal endpoint algorithms. The method yet further includes importing one optimal endpoint algorithm of the set of optimal endpoint algorithms into production environment.

Description

201112302 六、發明說明： 【發明所屬之技術領域】 本發明係關於最佳終點演算法的構建方法。 本申請案錄共同擁有賊時專辦請案「電祕理機 先進設備控制/先進製程控制之方法與系統」之優先權，盆美國 利、申請案第61/222，1〇2號、代理人案號第咖狐職:州則 唬，由發明人Venugopal等人於2009年6月30日提出申 内容併入本文以供參考。 τ明王又 本部分連續案主張共同受讓之專利申請案「 :，未受控制事件之佈置與其方法」之優先權 1唬弟12/555,674號、代理人案號第P2〇〇2/LMRX_p 明人Huang等人於2009年9月8日裎屮由咬由七 $丘η兵谨々s“士車w“月「曰如出申§月，该案係相關於並主 ^同又讓之叫專利申μ案「辨識在製程模組階段的未受控 f件之佈置與其方法」之優先權，其申請序號第6ι肪⑽4 ^、 代理人錢第P2G02P/L職_ρΐ79ρι 聊年Μ 3〇日謝請，全文峨入本文^1^。寻人於 【先前技術】 為易於論述，定義數個詞彙如下。 資料組·測量記錄，其為處理機具參數 變化點.轩變化發生時之時間序列上的函數。 ，點：-製程(如石夕層姓刻）已達到或接成 ,終點域-在資料組中預期終點要發生期間之^之I間^ 常相當寬且係基於使用者預估。 ' °° 4。、，’；點域通 部分最小平方鑑別分析(PLS_DA par Discriminant Analysis)-找出二组資料之 =ast Squares ====多個可能應變數(在心= 用PLS-DA在PLS_DA中，γ變數並不連續 ）寸J便 離散值或集合組成。PLS_DA會嘗試如χ變數 可用以將輸入資料分類至離散集合之其中—者中、、’复性、，且&，其 201112302 =點域—在終點域之前的資料組部分。 後κ點域-在終點域之後的資料組八 點二合數合的推:中之-特殊的變化 本質通物成。够肋讀該變化 來) 消耗最少並生產高品;處理湖嚴密掌控，俾使 論，識—製靡。如本文所 圖總與終關時。其他《(如管道_會導致:號 f利於討論，圖！呈現建構終 型相同。舉例而言，若在生產 ；;用於生產处中的基板類 似圖形的基板作為測試ϋ產训使用特定圖形的基板，使用相 板時在=以=^。在-實例中，當處理基201112302 VI. Description of the Invention: [Technical Field to Which the Invention Is Ascribed] The present invention relates to a method of constructing an optimal end point algorithm. The application case has the privilege of the privilege. The priority of the "Electricity Control Machine Advanced Equipment Control / Advanced Process Control Method and System" is the priority of the application, and the application of the US Patent, No. 61/222, No. 1 and No. 2 The singularity of the singularity of the singularity of the singularity of the singularity of the singularity of the singularity of the singularity of the singularity of the syllabus. τ 明王 and this part of the continuation case claim the priority of the patent application ":, the arrangement of uncontrolled incidents and its methods" 1 brother 12/555, 674, agent case number P2〇〇2/LMRX_p On September 8th, 2009, Huang et al. were bitten by the seven-thousand-nine squad, and the sergeant’s "sports car" was "successful", and the case was related to the main and the same. Called the patent application case "Recognition of the arrangement and method of uncontrolled f-pieces in the process module stage", the application number is 6ppm (10)4 ^, agent Qian Di P2G02P/L job _ρΐ79ρι 聊年Μ 3 Thank you for your next day, the full text is included in this article ^1^. Looking for people [Prior Art] For ease of discussion, define several words as follows. Data set · Measurement record, which is a function of the time series on the time when the processing tool parameter change point. , point: - process (such as Shi Xi layer surname) has been reached or completed, the end field - in the data set in the expected end point of the occurrence of ^ ^ ^ is often quite wide and based on user estimates. ' °° 4. ,,;; PLS_DA par Discriminant Analysis - Find two groups of data = ast Squares ==== Multiple possible strain numbers (in heart = PLS-DA in PLS_DA, γ variable It is not continuous. It is composed of discrete values or sets. PLS_DA will try to classify the input data into one of the discrete sets—the sufficiency, and the &, 201112302 = dotfield—the portion of the dataset before the endpoint field. After the κ dot domain - after the endpoint domain, the octet of the dataset is combined: the medium-specific change is essential. Enough to read the change to) consume the least and produce high-quality products; handle the lake with strict control, and make sense, know---------- As shown in this article, the total and final levels. Others (such as the pipe _ will lead to: the number f is conducive to discussion, the figure! presents the same final construction. For example, if it is in production;; the substrate used in the production area is similar to the graphic substrate as the test ϋ production training using specific graphics The substrate, when using the phase plate, is ===^. In the instance, when the processing base

Spectro喊r)、溫度感測器等emissi〇n 以千計)的感測器管道荒集資料。"+。可仗數以百汁(若非數 可用，號的資料。因為有過多資料 通常負===；魏方崎度域ϋ家使用者 在下一步驟刚，專家使用者會針對一個以上的信號檢視信 201112302 號圖形的變化。專家使用者利用 析。在一實例中，軟體程式為一簡 的叙肢程式來協助分 分析。在另一實例中，軟體程式為“行簡單計算與 而言， 係用以_化騎信號歷程n㈣視覺化程式，舉例 且可用以分析器取得 記號的任務可.能是令人卻^任目此，辨識終點特徵 管道中有超過2，_個波長測量結果。=感測器 測器管道(如提供關於溫度、壓力 t等可在其他感 中找到，若是需要分析每個信號鮮H寻貝^的感測器管道） 無法克服的任務。 、° & ’專豕使用者會面臨 如可預期的，根據應用，若干 點資料。舉例而言，二信號A與B 其;信號的終 信號B具有少於信號A的雜訊 白二點貝枓。然而，因為 號。假設有幾十或幾百個作 ^ 較佳的終點特徵記 任献別提最佳終點組的 爾為終點的徵兆。信號圖形的變 的蜂值便代表變化。雖然在“手斜=號斜率 任務，然而近年來，Pif# _化—直為乏味的 加困難。對於用以#唬艾艾付較不明顯，此任務變得更 此：-反上的小型開放區域之配方而言更是如 減i處理(如钱刻)的開放區域小到(如q%的基 &成)彳。唬茭化極為細微，以致人眼幾乎無法察覺。 料信為’專家使用者會刪除其認為與辨識終點無關的資 種方法包括辨識並刪除專家使用者不』 的標的區域’其係經常位於預終點域與後終點域 =間=為找出並琢磨終點特徵記號的成本很高(在專家時間上）， =以目^為令預終點與後終點域盡可能地大，以限制剩餘要 終點的區域。 201112302 選專家細者製程，專紐用封如僅分胖 料組:精選信號包括基於專家使用 如可預期的，僅由一過濾資料組運作合婵Τ β Μ田 ίί:ΐ:除之風險。換句話說，藉由過，’專家 然是最佳終點特徵記號。4中所辨識的終點特徵記號不必 在已辨識信號變化之後，專家 ，匕作為終點候選值之穩健,ί=^|=析，判定信 ί號歷程，狀該信號變化，‘“。1^ 會分析 用者接著便重回便仗貧料組中刪除。專家使 的終點。口，、乏未的任務’在其他信號中辨識「難以捉摸」 組中組數位過軸用在資料 制性地包括時序過遽器、、二=5 ’可用的過遽器實例非限 會減少資料組的雜‘，但因然對資料組使用過濾、器 所以通常會審慎使器：、亦會增加信號的即時延遲， 用多變量分析，專家使用者 ^一乂縮小資料組。為了使 终點的形狀。藉由預定炊丨2^々專豕使用者被指望要預設 預期形狀的信號。在變量分析本質上刪除未呈現 呈現此形狀的信號便被疋終點形狀係定義為波華，未 有「期望」形狀，就會點特徵記號不具 7 201112302 如由前述可知，從過多資 可為令人卻步的任務且執行終點特徵記號之任務 辨識出終點特徵記號，僅會週計算）。另外，-旦 為終點特徵記號的合適性之定旦心5，執二信號或信號組合作 終點特徵記號的信號變化，專ί二者ί::例中’為驗證作為 同的時間帶中找尋相似的信號變u二析:，信號，在約略相 費大量時間辨識第—終點特徵啡，=專家使用者已花 時間、資源以及/或是意願驗證結^。專碰用者可能不會總是有 在下一步驟108,專家使用者基 的型態。通常，終點演算法的型質挑選終點演算法 然而乂號之導數。 佳的終點演算法。不僅能辨識終";反曲點)反而可提供較 號所相關的最佳終點演算法之能力辨識終點特徵記 用者)擁有的專業知識。 b力南要少數使用者(即使是專家使 旦已出ί:、=法取t化广及/或是測試演算法的設定值。一 因為測試環境與生產環境之間生產終點演算法。 產前需要調整終點演算法的設終點演算法移至生 非限制性地包括平順魏器、延牛例整的設定值 值等等。 逛守間、凟异法型恶的特定設定 在一實例中，在測試環境中用以 ^ f中可能導致無法接受的即時延遲。如本貝==過 i」】二巧。若是_演細咖ώίί定 起使用，在終點料法觸終點之前基板可能钱刻過度。 201112302 為了使即Βψ延遲最小，必須調整過遽哭。 在將終點演算法移向生產之前二 。在—實例中，使用終定是否已最 料的貧料紐上。若是終點鮮法使用用來建構終點演 點’就視該設定值為最佳。然而 ^的設定值正確辨識終 =，就賴_奴值。在奴奴無紅確辨識 試(透過試誤法）。 值马取4之丽必須執行多次測 在下—步驟112，針對在铢點、、當曾4 定。若是已執行穩健性測試（步驟^ ^亍穩健性夠試進行判 ，其他基板相關的資料組。在一 使用該終點演算法於 藍集資料。接著就使用終縣算法於:._γ處理第二測試基板並 演，能_識終點，就視該终點首組中。若是該終點 就視該終_算料獨績、而2終點演算法無_識終畔 考讀健性測試需要時^ 、、、法之任務。 被沿用到生產環境中而未分析，許多終點演算法係 經常被視為建構終點演算換句話說，步驟lu 疒$山由囷1了知，建構終點渾管法的太本士夕迕车 、、^由具專業知識與經驗ϋ j方法大夕為手動程序，敌 制，移至生產上的終點雜的分析。考量資源的限 合理時間内，單—人員不、〕缺乏量化支持。另外’因為右 組合，所建構的終點演月^有^法分析所有信號以及/或是信 算法。 _製程而言可料總是最佳終點海 因此’建構穩健的終點演算法之簡化方法備受期待。。 【發明内容】 【實施方式】 本發明現將參照如__所補數個實關詳細描述之。 201112302 =列對本發明崎徹了解，提出大量具體細節-仍可施行本不^若干或全部該具體細節下， 並未詳述熟知的製程步驟ίί；是=免不亦要的干擾本發明， 明可=:②⑦與，之各式實施例。應當謹 記在心本發 可讀指令可;: 半導體、磁性、光磁性、而言，電腦可讀媒體包括 他形式之電腦可病姐雕。ίι式：或用以儲存電腦可讀編碼的其 之設備。此類設備包括電路(相實$ ===備置的=括:適士= 各式任務之電腦/運算裝置與專用/可•施例有關的 之方ϊ照ir月ΐ施例，提供用以自動發覺並最佳“點渾笞法 之方法。本發明實施例包括建構終點演管法之太本ϋί开法 程的最佳終點。本發明實施例亦句衽、I方法’其判疋一 法之就地方法。包括在生產環境中使用終點演 本發=ί=終例來討論巧^ 因=些論述係意圖作為 自動分析資料、自動判概tr期間. 佳終點演算法至生產上。 、忑唬、共自動叼入^ 在先前技術中，單一人員純粹是因為眘 時間期間内能分析所有信號。不像先前$術；合理白 用演算法引擎來執行分析。因為資料係自動，可發 10 2〇1112302 外，因為現由演特=己號的相關性為特徵。此 終點演算法。 尺夕貝枓以建構—組穩健的最佳 點域)相==函if^终點之標的區域(如终 便可__算糾料 在一實施例中，演算法引擎雜 , 量分析中的可能終_徵記可能形狀，其代表多變 於每個可祕轉徵記號的形°=技^’使用者不需要對 距等等）。取而代之的是：：識(如波峰、波谷、間 號，演算法引擎會產生可能大、擎已辨識可能終點特徵記 口識的可能終闕徵記·秘於=算糾擎所 中’演算法引擎係配置為 形狀(如曲線）。在一實施例 試，以辨識一製程的最佳終點選值之資料調節與測 變異性可藉由執行逐步迴歸導出^為時間函數之各參數 限時間區間中，判定各資料輸 個製程歷程的-串有 在計算斜率的時間區間係可設定在—實施财，用 併去除與終點不相 ^資^2^接射射去除雜訊，且— 在—貧施例申，〇ES信號合 之變化程度(即斜率)分植。在衣程推展而可見的變異性. 續波砂在-起。藉由依斜率具相似斜率變異的連 主、，號中的雜訊會大幅減少。此合。需要分析的信號量 表呈，，^最可能包含與終點相關的。曰邮號與信號群組的列 貝轭例中，執行楝選 的測試基板中之特徵，則該終^it ί亚非在所有或極大部分 而’若是一終點特徵記號出現在不穩健而可刪除。然 "亲基板上，因為控制基板為並 201112302 未接受蝕刻而因此不應產生終點特徵記號之基板，所以該铁點 徵記號亦可刪除。 在-實施例中執行多變量分析。在一實例中，由該分析所 結果係用作為部分最小平方鑑別分析(PLS_DA，Partial U诚 Squares Discriminant Analysis)的輪入，俾使在每個依斜率群 各個別信號之權重為最佳。在—實施例中，並非求入 終點，的預期形狀(如先前技術所要求），而是pLS视會 點之標的區域以及由演算法引擎所提供之形狀。 、 f =實施例中’由OES信號而來經犯抛的結果可與 聯合並結合。在—實施例中，可重複pls_da於i的 以產生精實的最佳可能終闕徵記餘合， 问對比與即時終點計算的低運算負荷。 〜、有 實關巾，可祕崎徵記縣轉料具Spectro calls r), temperature sensors, etc. emissi〇n thousands of sensors pipeline waste data. "+. Can count the number of hundred (if not a number of available, number information. Because there is too much information is usually negative ===; Wei Fangqi domain users in the next step, expert users will be more than one signal inspection letter Changes in the 201112302 graphic. Expert users use the analysis. In one example, the software program is a simple narrative program to assist in the analysis. In another example, the software program is "for simple calculations, Taking the _ riding signal history n (four) visualization program, for example, the task that can be used to obtain the mark by the analyzer can be used to make it possible, and there are more than 2, _ wavelength measurement results in the identification end point feature pipeline. The detector pipe (such as providing temperature, pressure t, etc. can be found in other senses, if it is necessary to analyze the sensor pipeline of each signal), ° & 'Special The user will face as expected, depending on the application, several points of information. For example, the two signals A and B; the final signal B of the signal has less than the noise of the signal A. Suppose there are a few Ten or a few hundred better endpoint characteristics are noted as the indication of the end point of the best end point group. The change in the bee value of the signal pattern represents the change. Although in the "hand skew = number slope task, in recent years Come, Pif# _ _ is straightforward and difficult to add. It is less obvious for #唬艾艾付, this task becomes even more: - the formula of the small open area on the reverse is more The open area (such as money engraving) is small (such as q% base & 彳). The sputum is so subtle that the human eye can hardly detect it. The letter is that 'expert users will delete it and think it has nothing to do with the end of identification. The method of hiring includes identifying and deleting the target area of the expert user. The system is often located in the pre-final field and the post-end field = between = the cost of finding and honing the end point feature is high (in expert time), = Use the target to make the pre-final and post-end fields as large as possible to limit the remaining end points. 201112302 Select the expert process, the special button is only for the fat group: the selected signals include based on expert use. Expected to be filtered by only one data set婵ΤβΜ田ίίί:ΐ: In addition to the risk. In other words, by the expert, it is the best end point feature mark. The end point feature mark identified in 4 does not have to be after the identified signal change, the expert , 匕 as the end point candidate value of the robust, ί = ^ | = analysis, determine the letter of the history, the signal changes, '". 1 ^ will analyze the user and then return to the sticky group to delete. Experts make End point. Port, lack of tasks 'identify in other signals' "unpredictable" group group digits over-axis used in data system including timing filter, two = 5 'available over-the-counter instance non Limitation will reduce the miscellaneous of the data set, but it is usually prudent to use the filter for the data set: it will also increase the immediate delay of the signal. With multivariate analysis, the expert user will narrow down the data set. In order to make the shape of the end point. By prescribing the user, the user is expected to pre-set the desired shape of the signal. In the nature of the variable analysis, the signal that does not present this shape is deleted, and the end point shape is defined as a wave. If there is no "expected" shape, the characteristic mark will not be 7 201112302. As can be seen from the above, from too much capital, The task of deterrence and the task of performing the end point feature mark recognizes the end point feature mark, which is only calculated weekly. In addition, if it is the appropriateness of the end point feature mark, the signal change of the two signal or the signal group cooperation end point feature mark, in the case of 'in the case of verification as the same time zone search A similar signal changes: the signal, in a considerable amount of time to identify the first-end feature morphine, = expert users have spent time, resources and / or willing to verify the conclusion ^. A dedicated user may not always have the type of expert user base in the next step 108. Usually, the model of the endpoint algorithm selects the endpoint algorithm but the derivative of the nickname. Good endpoint algorithm. Not only can the end of the "reflex point" but the ability to provide the best end-point algorithm associated with the ability to identify the end-point feature record) has the expertise. b Linan wants a small number of users (even if the experts have already issued ί:, = method to take the wide and / or test algorithm settings. First, because of the production end point algorithm between the test environment and the production environment. Before the end point algorithm needs to adjust the end point algorithm to move to non-limiting, including the smoothing of the Wei, the extension of the set value of the value, etc. The specific settings of the staggering, strange and evil in an instance In the test environment, it may cause unacceptable immediate delay in ^f. For example, if this is the case, if the _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 201112302 In order to minimize the immediate delay, you must adjust the crying. Before moving the end point algorithm to production, in the case of the example, use the final material that has been determined to be the most important. Use to construct the end point of the show 'just consider the set value is the best. However, the set value of ^ correctly identify the final =, it depends on the _ slave value. In the slaves no red identification test (through trial and error). Take 4 丽 must perform multiple measurements below - step 1 12, for the point of the point, when it has been determined. If the robustness test has been performed (step ^ ^ 亍 robust enough to try to judge, other substrate-related data sets. In the use of the end point algorithm in the blue set data. Then use the final county algorithm to: . γ process the second test substrate and play, can identify the end point, it will be regarded as the end of the first group. If the end point is regarded as the final _ calculation material alone, and 2 end point algorithm The task of ^, , and method is required to be tested in the life test. It is used in the production environment without analysis. Many end-point algorithms are often regarded as constructing the end point calculus. In other words, the step lu 疒 $ The mountain is known by the 囷1, and the construction of the end of the 浑 法 的 太 太 迕 、 、 , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , Considering the limited time of resources, the single-person does not have quantitative support. In addition, because of the right combination, the constructed end-point performance has a method to analyze all signals and/or a letter algorithm. Always the best end point of the sea thus 'building a solid end The simplified method of the algorithm is expected. [Embodiment] The present invention will now be described in detail with reference to the details of the number of __. 201112302 = column for the understanding of the present invention, a lot of specific details - The specific process details may not be described in detail, and the well-known process steps are not described in detail; the present invention is indispensable for the present invention, and the various embodiments may be used. The readable instructions can be recorded in the heart;: semiconductor, magnetic, photomagnetic, and computer readable media include computer sorcerer carvings of his form. ίι: or a device for storing computer readable codes. Such equipment includes circuits (consistent $ ===prepared = including: 士士 = computer / computing device for various tasks and special / can be used in the case of the application) Automatically detect and optimize the method of “pointing method”. Embodiments of the present invention include the best end point for constructing the end-of-pipe method. The embodiment of the present invention is also a method of singularity and method of determining the method. Including the use of the end-point performance in the production environment = ί = the final case to discuss clever ^ = some discussion system is intended as an automatic analysis of data, automatic judgment period tr. Good end-point algorithm to production.忑唬, 共, automatic input ^ In the prior art, a single person is purely because of the careful analysis of all signals during the time period. Unlike the previous $ surgery; reasonable white to use the algorithm engine to perform the analysis. Because the data is automatic, it can be sent out of 10 2〇1112302, because it is characterized by the correlation of the performance of the special number. This endpoint algorithm.尺 枓 枓 枓 枓 枓 枓 枓 枓 枓 枓 ) ) ) ) ) if = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = The possible end_signature may be a shape that represents a change to the shape of each dexterous sign. The user does not need a distance, etc.). Instead, it is:: knowledge (such as peaks, troughs, and numbers), the algorithm engine will produce a possible event that may be large, the engine has identified the possible end points, and the secrets of the engine. It is configured as a shape (such as a curve). In an embodiment, the data adjustment and the measurement variability of identifying the optimal end point selection value of a process can be derived by performing a stepwise regression to obtain a time interval of each parameter of the time function. Judging that each data is transferred to a process history - the string has a time interval in which the slope is calculated. It can be set in - the implementation of the money, and the removal and the end point are not compatible with the ^ 2 ^ shot to remove the noise, and - in - poor For example, the degree of change (ie slope) of the 〇ES signal is divided. The variability seen in the clothing process is extended. The continuation of the sand is in the beginning. The information will be greatly reduced. This is the semaphore that needs to be analyzed, and ^ is most likely to be related to the end point. In the column yoke example of the postal number and signal group, the characteristics of the selected test substrate are performed. Then the end ^it 亚 Asia is not at all or Most of the 'if the end point feature mark appears to be unstable and can be deleted. On the parent substrate, because the control substrate is and the 201112302 is not subjected to etching, the substrate of the end point feature mark should not be generated, so the iron point mark It can also be deleted. In the embodiment, multivariate analysis is performed. In one example, the results of the analysis are used as rounds of partial least squares discriminant analysis (PLS_DA, Partial U Squares Discriminant Analysis). The weights of the individual signals according to the slope group are optimal. In the embodiment, the expected shape is not the end point, as required by the prior art, but the target area of the pLS point of view and by the algorithm engine. The shape provided. f = in the embodiment, the result of the omission by the OES signal can be combined with and combined. In the embodiment, the pls_da can be repeated to generate the best possible final sign. Hey, ask for the low computational load of the comparison and the instantaneous end point calculation. ~, there is a real off towel, can be the secrets of the county

遲超過最大可容許的即時延遲 &，則該即時ί冬點演算法J 保畫ir貫施例巾’會基於有㈣訊對無職訊之轉(此播η ί 將m綠法铸。在—實ίί =保真率與低即時延遲之演算法係視為較穩健的例—中， 已執仃分寺，即時終點演算法之其中一者 :^ —旦 =圖口式^下列論述可更加了解本發；的特徵鱼“產中。 演算t方ΐ發喻_—簡物圖：轉建構终點 組感測器策集。 勤貧料、電漿資料等等)係藉由- 止’_板。藉能= 12 201112302 中’資料製程變異性有關的雜訊。在—實施例 腔室的資料，亦理的測試基板。藉由併入來自』 在下―步^^與腔室之間的差異相關的雜訊。不° 換句話說，“ 會發生抛_之概略時間期間。 對寬廣的時間J間：二點域為-概略又 最佳引f以執行資料分析與產生—組 所以可分析來自不止一個=亚非手動執行資料分析， 涉及較大量的資料，因為本技術者知悉即使 j被刪除，所以由多個基板而二:尋得之終點特徵 向較為穩健。 貧卄祜所建構之終點演算法傾 圖3A與3B呈現本發明實施例 料組分析及產生最佳終點演算 ^圖，描奢執行資 於討論，將偕_ 5 一併討論圖3A: ‘、=弄法引擎。為利 描1 會在二實施例中資料組推展成為最佳終點J曾二現主一方5圖， 在第一步驟302,演算法引擎對可、、咨’ '、去列表之貫例。 執行線性配適。換句話說，各信號献於初始資料群組502) 料群組5〇4)。為使雜訊最少並使辨^間分成均勻片段(資 段的長度相當重要。若>{段長度太之可祕最大，片 終點。若片段太短，斜率(如^在^驟、^占=平均掉而錯過該 在-實施财，可預以段長度的最小 >)會_訊影響。 最小片段長度較1/10秒長。在另—容Γ仓ί夏。在一實施例中， 集的資料，最大片段長度較2秒短。貝也’中’針對在10Hz所蒐 的段計算斜率及其相應If the delay exceeds the maximum allowable immediate delay &, then the instant ί winter point algorithm J will be based on the (four) message to the no-communication turn (this broadcast η ί will m green method cast. - The actual ίί = fidelity rate and low real-time delay algorithm is considered to be a more robust example - in the case of the sub-division, one of the instant end-point algorithms: ^ 旦 = 图口^ The following discussion can be More understanding of this issue; the characteristics of the fish "production. calculus t square ΐ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ '_板. Borrowing energy = 12 201112302 in the 'data process variability related noise. In the case of the example chamber, also the test substrate. By incorporating from the next step ^^ and the chamber The difference between the noises is not. In other words, "there will be a brief time period of throwing _. For a wide time J: two-point domain is - rough and best cited f to perform data analysis and generation - group can be analyzed from more than one = Asian and African manual data analysis, involving a larger amount of data, because the technology It is known that even if j is deleted, it is composed of a plurality of substrates and two: the end point feature found is relatively stable. The endpoint algorithm 3A and 3B constructed by the inferiority presents the analysis of the data set of the embodiment of the present invention and the generation of the best end point calculation ^ map, and the execution of the luxury calculation is discussed, and the discussion of FIG. 3A is discussed. = The engine is used. In the second embodiment, in the second embodiment, the data group is promoted to become the best end point. J is the second party. In the first step 302, the algorithm engine is in the right, and the list is taken. Perform linear adaptation. In other words, each signal is presented to the initial data group 502) group 5〇4). In order to minimize the noise and divide the discrimination into even segments (the length of the segment is quite important. If the length of the segment is too large, the end point of the segment. If the segment is too short, the slope (such as ^ in ^^, ^ Occupation = average miss and miss the in-implementation, which can be pre-emitted by the minimum length of the segment.) The minimum segment length is longer than 1/10 seconds. In another case, it is in the case of another. In the set of data, the maximum fragment length is shorter than 2 seconds. Bei also 'zhong' calculates the slope and its corresponding for the segment searched at 10Hz.

在一貫例令，若是信號A 13 201112302 已被分成十個片段，便會判定信 料群組506A)。在—實施例中，:十個斜率與斜率雜訊值(資 (資料群組506B)。 斜率雜矾值可用以使該斜率標準化 此外或另外，演算法弓|擎合 作為輸入來執行多變量分析;率雜訊值所縮放的斜率 器管道組合的麵，產生斜率平方分析)，基於來自感測 資料群組5_)。在一實= 中與^訊值的額外列表(亦包括於 (亦包括於資料群組5〇6B)。 斜率雜況值可用以使斜率標準化 506A)，在;建:‘3片率雜訊值列表(資料群組 候選值。在-實例中，演算法會辨二識帶有終點資料的信號 信號斜率的變異量。量化^率變/旦^與其片段），量化各 斜率的標準差。在—實例ϋ的一種方法包括計算標準化 ^此實例t.，*標，额表帶的信號。 尚斜率變異(相對於斜率 、二貧枓之^唬。因此，具 群組508)。 "訊)之彳5唬會被辨識為信號候選值(資料 在下==①至少购信號)， 長至信號波長頻帶(資料群異的連續波 中，若是在255奈米與夺米之門/ =號！。在一實例 會被結合至一個單—信號異且測量結果 物件！已大幅減少，所以可減輕運算負荷。因為而要刀析的 506Β^，"1合^ = 引擎辨識標準化信號列表(資料群组 _ ) s在下面衣矛王中捕捉偏移與雜訊。換句話嘮， 號’因其具有高斜率上變異:玆斜ΐ 模式茭化(如偏移、雜訊等等)之可能候選值。 ’、般 14 201112302 驟312’演算法引擎藉由結合具相似斜率變旦的連嘖 ίί 頻ΐ(铜組514)來減少 在標準化OES信Ϊ 上與步驟删類似，除步驟312係用 伸(在#算法觸财信鮮道產生高對比感測器 ί 组5121ί標準化波長頻帶(資料群組514)之列 貝方&例中，骑各貿料組中的信許分算β 旦 丄 號t的終點資料之可能性’所以可將。料組_义2 相較具低斜率變異的信號，具高斜率變 θ植^下/驟316 ’ 异法引擎搜尋高對比感測器信號以及/或 點•記號(資料群組‘2 敢触賴舰。可在不同的 對稱過^。時間對稱過濾器在—特定點 =的點_异平均值。此過濾n僅·於後處 製程執行期間。不像時間不對稱過濾器 g s 起最小的時間扭曲以及/或是振幅扭曲 會經歷最小的即時延遲。 過濾貢枓 如由前述可知，各資料群、组包括過量的信號。在 而ί5將各資料群 5分等’所以資料分析時間可藉*減少搜尋值 备降低。在一實例中，並非搜尋資料群組508中的所 =疋^析前1G個高對比感測器信號。可搜尋的物不 冋。執行回收遞減分析以判定最佳數量。干里了月匕不 ·+_在下—步驟318 ’演算糾擎鱗高對比感測11辞MiΆ 508 i，510)f^，m/,M t(t 5；；^^ 15 201112302 ^率士尋找終點域中的可能終點特徵記號(資料群組5降夢由 ^用各@對比感:廳錢/頻帶對各標準化感測器信號/頻帶二匕 。’可辨識的可能終點特徵記號會具有較高的保直率。 在下-步驟320，演算法引擎搜尋資料結果(資料群组516斑 合(資料群M 52G)分等。換句話說 ‘ 間細之終雜徵記號，叫輯_信號^訊 勃/羊Slg nolse rati0)。在一實施例中，在相同導數上 iHtii °換句話說’即使發生在第—導數轉值盘發生在 ϋ Ϊ值在相同的時間區間内發生，喊 重複ί:::3己，ί”糊于穩健性測試以移除可能的非 間並不-致，舉二」特徵記在多個基板 結果’所《會捨棄該^能===徵記號可能為雜訊，偏移的 分為裸露石夕區的光阻遮罩之基板為具有一部 ::者tT=_'全由光阻“二蓋 ..因此’控制基板應;不i有=基=②==象。 =終點特徵記號之其中一者，則會捨棄該匹、 被測Ϊ2可;=特獨以-實例中’ 終點特徵記號之特徵記號。藉性以刪除非真正的 _步較紅败徵 16 201112302 於相^性㈡2 i 法引擎執行多變®相關性分析’例如基 取以使可能終點特徵 細通論巧 析需要知道特徵記號曲_ ，句話# ’ ms 提供終點特徵記號形狀 先前技術，由演算法引聲所雜钟夕、交1分析的輸人。不像 形狀特徵。因此，可輸人至多i量相徵記號f有不同的 被辨識的3能終點特徵記號的形狀。刀析的輸入量係依照已 決定)與各信號相關終點特徵記號列表所 中信號之間的相關矩陣。記號與各感測器管道 權重以及/或是單位，用以#夂 可應用至每個信號的最佳 然多變量分析可幫助最佳化°^^=徵記號的對比最佳。雖 且，即使相關性的PLS分析= 斤二佳終點演算法列表。並 相關性的PLS分析，而可為杯付刑At例中，本發明並非限於 在下-步驟324，演算法^擎的多變量分析。 (資料群組522)成具最小即時可能終點特徵記號 ，。換句話說，演算法引擎係配資料群組 隶小即時延遲於生產期間執行的終點渾點i寺徵記號成以 自動計算各終點演算法所需的奴值貫施例中，會 上召喚終點。即時過遽:遲在每個處理測試基板 以最小化發生於無限^衝回應過滹哭，屺憶體組件的初始值 點接2料歷程開端之終點“法^變脈波。這對於終 肩异法引擎會針對各可能終點特徵 法。在—實施例中，若是演算法引擎益即時終點演算 …、凌建構即時終點演算法， 17 201112302 則不會提供终點演算法。在一實例中， $每個處理測試基板上召喚/辨識終點之；寺:=構 會提供終點演算法。. 才、、”·、占6、开法，則不 終點演算法。二實例^_卩時延遲的 因為該即時延遲會導致生產期間的^度^^^超^5門檻， 終點演算法。 沒蚀到基板’所以會刪除該 之即未通過一組穩健性評斷桿準 所有測試基二二包括以最小即時 制基板上辨識終點。換句話說，例包括不在控 遲的終點演算法之分等會較t。^率，則具有較小即時延 异法具有相同的即時延 冑=巾’若是二個終點.演 有較高等級。 j具有較焉保真率的終點演算法會具 回頭參照圖2,在下—牛輙ΟΛ。 a /' 用者=田列中，移至生產上的即時動移至生 ：：^ 5 :=_穩健的終點演算法，若以供 貝S且經驗顯示藉由白勒在|丨 3以分鐘内執行。另;程3:ίξ的即時終點演算法. 非專家使用輪。 7野匕知終點域產出 18 201112302 演算法列表’使用者可迅速重新定義終點域並回傳 貝异法引擎，以在數分鐘内產生新的終點演算法列表。 點演產=巾魏綱—辟_，其_行即時終 在第一步驟402執行一配方。 ，下-步驟4G4 ’藉由—組感·在基板處理期間取得 欲鞔J — ί驟406 ’就地利用終點演算法，分析資料以辨識勢程 _運算引擎分析㈣。因為可能策ί ί ^ "m异引擎配置為處理大量資料的高速處理模电。^料 曰仗感測“接送出’科需先經過製造主控制 = 模組控制器。由Huang等人於2_年9月8日提出卜主程 ^請案第_，674號描述適合肋執行該分析之^電： 士了^驟408，系統作出關於是否辨識終點之判定。 右疋尚未辨識終點，則系統會回到步驟4〇4。 .=前=辨止配方。 時終點演算法之方法《» Ι&由自動化^例提供辨識最佳的即 ，使用者之$求。軸如本文所述的 U對專 算法移至生產中。並且，因為構建 演 雖然已用數個較佳實施例描述本、，^^ | t壬務。 =的變更、替換、與均等物。雖然本文提;在本發， 係思圖為作為例證而非限制本發明 纟：實例，該實例 皆使用終點作為實例，本發财可用化使==自始至終 期間的信號變化事件。 匕其為發生在處理 並且，為求方便，本文提供標題盥 建構本文的申請專利範圍之範#。另。^要二t不應用以 的形式，且為求方便而提供於此，因而卜高為高度簡化 發明，其係陳述於申請專利範圍中。若是；使^^= 19 201112302 詞彙係意®具有-般理解的數學意義— »應當注意實行本發明的方法與‘代g於;的 變更、替換= 【圖式簡單說明】 類似喊私，且其中 圖1呈現建構終點演算法之一種簡單方法。 法的圖=本發明實施例的一簡單流程圖’描綠建構終點演算 圖3A與3B呈現本發明實施例的一簡單流程圖， 見攻佳終點演算法的步驟與演算法引擎。 、a執仃發 圖4呈現本發明實施例的一簡單流程圖，1用 一 終點演算法於生產環境中。 、只仃邊敢佳 圓5呈現本發明實施例的一方塊圖，描繪資料纟且 曰 終點演算法列表之實例。 ' $展马取佳 【主要元件符號說明】 102-116 步驟 202-208 步驟 302-330 步驟 402-410 步驟 5〇2 初始資料群組 504 信號均勻片段 506A、506B、508-524 資料群組In the usual case, if the signal A 13 201112302 has been divided into ten segments, the message group 506A) is determined. In the embodiment, there are ten slope and slope noise values (data group 506B). The slope noise value can be used to normalize the slope. Additionally or alternatively, the algorithm bows the engine to perform multivariate input. Analysis; rate noise values are scaled by the surface of the sloper pipe combination to produce a slope squared analysis) based on the sense data group 5_). An additional list of values in a real = (also included in the data group 5〇6B). The slope hysteresis value can be used to normalize the slope 506A), in: build: '3 chip rate noise List of values (data group candidate values. In the example, the algorithm will identify the variance of the slope of the signal signal with the endpoint data. Quantize the rate/deny^ and its fragments) and quantify the standard deviation of each slope. One method of -instance 包括 includes calculating the normalized ^this example t.,* mark, the signal of the forehead band. The slope is also mutated (relative to the slope, the second stagnation. Therefore, with group 508). "Xunzhi) 彳5唬 will be recognized as the signal candidate value (data ===1 at least buy signal), and grow to the signal wavelength band (in the continuous wave of data group, if it is at 255 nm and the gate of the rice / =!! In an instance will be combined into a single-signal difference and the measurement object! Has been greatly reduced, so the computational load can be reduced. Because of the 506Β^,"1 combined ^ = engine identification standardization The signal list (data group _) s captures the offset and noise in the following spears. In other words, the number 'because it has a high slope variability: ΐ oblique mode 茭 (such as offset, noise And so on) possible candidate values. ', General 14 201112302 Step 312' algorithm engine reduces the number of steps on the standardized OES signal by combining the 啧ίί frequency (copper group 514) with similar slope-denier In addition to the step 312, the extension (in the # algorithm touches the fresh channel produces a high contrast sensor ί group 5121 ί normalized wavelength band (data group 514) in the list of Bayesian & The possibility of calculating the end point data of β 丄 丄 t ' 所以 所以 所以 所以 所以 所以 料 料 料 料 料Signals with lower slope variability, with high slope change θ plant ^ ^ / 316 'Different engine search for high contrast sensor signals and / or points • mark (data group '2 dare to touch the ship. Can be different The symmetry passes. The time-symmetric filter is at the point-specific point = the same value. This filter n is only during the post-process execution. Unlike the time-asymmetric filter gs, it has minimal time warping and/or Amplitude distortion will experience the minimum instantaneous delay. Filtering Gongga As can be seen from the above, each data group and group includes excessive signals. In addition, ί5 divides each data group by 5' so the data analysis time can be borrowed to reduce the search value. In an example, the first 1G high-contrast sensor signals are not searched in the search data group 508. The searchable objects are not defective. The recovery and decrement analysis is performed to determine the optimal number.匕不·+_下下-Step 318 'calculations and corrections scales high contrast sensing 11 words MiΆ 508 i,510)f^,m/,M t(t 5;;^^ 15 201112302 ^ rate looking for the end point domain Possible end point feature mark (data group 5 dreams by ^ use each @ contrast: hall money / With the pair of standardized sensor signals/bands 匕. 'The identifiable possible endpoint feature markers will have a higher rate of salvage. In the next step 320, the algorithm engine searches for data results (data group 516 spotted (data group) M 52G) is equal. In other words, 'the end of the fine symbol, called the album _ signal ^ ze Bo / sheep Slg nolse rati0). In an embodiment, iHtii ° in the same derivative, in other words 'even occurs In the first derivative of the value disc occurs in the ϋ Ϊ value occurs in the same time interval, shouting repeated ί::: 3, ί" paste in the robustness test to remove the possible non-inferior, not to mention The feature is recorded in the results of multiple substrates, which will discard the ^ energy === the symbol may be noise, and the offset of the substrate of the photoresist mask divided into the bare stone area has one part:: tT =_'All by the photoresist "two covers.. Therefore" the control substrate should; not i have = base = 2 = = like. = One of the end point signatures will discard the one, the measured Ϊ 2 can be; = uniquely - in the instance of the 'end point feature token characteristic token. Borrowing to remove non-real _steps vs. red defeats 16 201112302 in the phase (2) 2 i method engine to perform multivariate ® correlation analysis 'for example, to make the possible end point feature fine-grained analysis need to know the feature token _, The sentence # ' ms provides the end feature signature shape prior art, the algorithm is used to introduce the sound of the clock, the intersection of the analysis of the input. Unlike shape features. Therefore, it is possible to input at most the i-signal sign f with different shapes of the recognized 3-energy end point signature. The input of the knife is based on the correlation matrix between the signals in the list of characteristic signatures associated with each signal. Marks and sensor pipeline weights and/or units are used to #夂 The best multivariate analysis that can be applied to each signal helps optimize the optimization of the °^^= sign. Even though, even the correlation of PLS analysis = the second best algorithm list. And the related PLS analysis, but can be used for the cup. In the example of At, the invention is not limited to the multivariate analysis of the algorithm in the next step 324. (data group 522) has the minimum immediate possible end point feature mark. In other words, the algorithm engine is equipped with a data group that is small and has an immediate delay in the execution of the end point of the production process. The sign is automatically calculated for each end point algorithm. . Immediate over-the-clock: late in each processing test substrate to minimize the occurrence of infinite ^ rush response to crying, the initial value of the memory component is connected to the end of the 2 process history "method change pulse. This is for the end shoulder The heterogeneous engine will target each possible endpoint feature. In the example, if the algorithm engine benefits the immediate endpoint calculus..., Ling constructs the instantaneous endpoint algorithm, 17 201112302 will not provide the endpoint algorithm. In an example, $ Summon/identify the end point on each processing test substrate; Temple: = Construction will provide the end point algorithm.. Only, , "·, account for 6, open method, then no end algorithm. The second instance is delayed by ^_卩 because the immediate delay will result in a ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ The substrate is not etched, so it will be deleted, that is, it will not pass a set of robustness determination bars. All test bases 22 include identifying the end points on the substrate with minimum real-time. In other words, the example includes the fact that the score of the endpoint algorithm that is not late is later than t. ^ rate, then has a smaller immediate extension method with the same immediate delay = towel 'if there are two end points. The performance has a higher level. j. The endpoint algorithm with a higher fidelity rate will be referred back to Figure 2, in the lower-burdock. a / 'user = field column, move to production on the immediate move to life: : ^ 5 : = _ robust end point algorithm, if the shell S and experience shows by Bai Le in | 丨 3 Execute in minutes. Another; Cheng 3: ίξ's instant endpoint algorithm. Non-experts use the wheel. 7 野 knows the end-of-field output 18 201112302 Algorithm List 'The user can quickly redefine the end-point field and pass back the Bifle Engine to generate a new list of end-point algorithms in a matter of minutes. Point production = towel Wei Gang - _, its _ line immediately in the first step 402 to execute a recipe. , Down - Step 4G4 ' By means of group sense - Obtained during substrate processing — — 406 406 ‘In situ use the end point algorithm to analyze the data to identify the potential range _ calculation engine analysis (4). Because it is possible to apply a high-speed processing mode to handle large amounts of data. ^ 曰仗 曰仗 曰仗 接送 接送 接送 接送 接送 接送 接送 科 科 科 科 科 科 科 科 科 科 科 科 科 科 科 科 科 科 科 科 科 科 科 科 制造 制造 制造 制造 制造 制造 制造 制造 制造 制造 制造 制造 制造 制造Performing the analysis of the ^ electric power: 士 ^ 408, the system makes a determination as to whether to identify the end point. If the right 疋 has not recognized the end point, the system will return to step 4 〇 4. . = before = discriminate the formula. The method "» Ι & is provided by the automation method to provide the best identification, that is, the user's demand. The axis is moved to the production of the U-specific algorithm as described herein. And, because the construction performance has been used several times better. The embodiment describes the changes, substitutions, and equivalents of the present invention. Although the present invention is described herein, the present invention is illustrated by way of example and not limitation. As an example, the end point can be used to make == signal change events from start to finish. 匕 为 发生 发生 处理 处理 处理 并且 并且 并且 并且 并且 并且 本文 本文 本文 本文 本文 本文 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号t is not applied in the form and is provided for convenience Therefore, Bu Gao is a highly simplified invention, which is stated in the scope of the patent application. If so; ^^= 19 201112302 vocabulary meanings have a general understanding of the mathematical meaning - » should pay attention to the implementation of the method of the present invention and 'g Change; Replacement = [Simplified description of the schema] Similar to shouting, and Figure 1 shows a simple method of constructing the end point algorithm. The diagram of the method = a simple flow chart of the embodiment of the invention 'the green construction end point calculus 3A and 3B show a simple flow chart of the embodiment of the present invention, see the steps and algorithm engine of the attacking end point algorithm. A, Figure 4 shows a simple flow chart of the embodiment of the present invention, and 1 uses an end point. The algorithm is in the production environment. Only a block diagram of the embodiment of the present invention is presented, and an example of the data list and the list of end-point algorithms is depicted. '$展马取佳[Main component symbol description] 102 -116 Steps 202-208 Steps 302-330 Steps 402-410 Step 5〇2 Initial Data Group 504 Signal Uniform Segments 506A, 506B, 508-524 Data Group

Claims (1)

2〇1112302 七、申請專利範圍： -2碑:===，在， 概略終點域，其中該終點域為預計會發生該製程終點的一 生「組可能終點特徵記號； 弓丨入該組最二巧:組最佳終點演算法;以及 、…、I秀异法的—最佳終點法至生產環境中。 2.如申請專利範圍第1項之 b 法，其中該感測器資料係從多於—個的基板演算法之方 法，其中自^辨識:最佳終點演算法之方 ，段以將從該數個信號流； 法’^=識-最佳終點演算法之方 1：^自動辨識一最佳終點演算法之方 記號，該感測器資ί S3:徵記號之第一組可能終點特徵 針對該感測器資料，呌瞀t ::其中針對該數個片段⑶ 5十异該斜率的斜率變異， 以從該數個信號流中辨識出一組高 21 201112302 對比，其中該組高對比信號具有高斜率變異， 茲的連毅*成為—組信號波長頻帶’ 信 ，該組信號波長頻帶分等，以及 號波㈣糾雜耻錄與該組 6· 法，H 之自動辨識一最佳終點演算法之方 記號二該感測器資二:::：徵記㈣ 及4==;=:率值放之斜率以 化的^^-組標準化的相應斜率雜斤’以產生―組標準 識有=ΐ^辨 帶。，合具她纖異的輪峨,波長頻 把該標準化信號分等， * ==:=¾:二及 ==標準化信號波 範圍第5私自動辨識-..._—〜 鉍。己號成為該組最佳終點演算法包 若是該組可能終點特徵記萝 能終點特徵記號具有相似b :可能終轉徵記號與第二 、冬點特徵記號與該第二可能^期間，結合該第-可能 f，其中祕可能終點辨、識—最佳終點演算法之方 括 .可 22 201112302 非重從該組可能終點特徵記號中移除可能的 組最以關性分析以辨識該組可能終點特徵記號的- 轉化該組最佳終點特徵記號成為一 終點ΐίΐ:?，時延遲係基於過渡器:遲〜延遲之即時 算法 至少—者產生最佳終點演管法 以^有大於-__目應，‘時終點演 即時終點演S即=點’’、异法無法通過-穩健性測試，刪除該 等係分等，㈣分 8·如申請專利範圍第6 其中轉化該_祕轉徵絲法之方法， 可能終點特徵記號具有相似匕可力終點特徵記號與第二 終點可能終^期間，結合該第—可能 非重複性終，:特該乡且可能終點特徵記號中移除可能的 組最關性分析以辨識物能終點特徵記號的一 終點组;;具最小即時延遲之即時 藉由勃;^遲係基於過遽器延遲， #除具有且,佳終點演算法 算法，以及 __相應即時延遲之即時終點演 即時終點演^算法無法通過-穩健性戦，删除讀 201112302 箄传^點演具法之各最佳終點演算法分等盆 寻係基於保真率_即時延遲之其中至少—者。4，其中讀分 9· 法 條件之其中至少 如申請專利範圍第i項之 ，ί 2最佳_演算法之之方 …、者。 組使用者定義 10. 含 幾定 終點之方法， 一種在—處理腔室中處理基板的期間 包 在一基板上執行一配方； 一組感測器接收處理資料； 辨識-製程終點；以及#去刀析捕理貧料； 停止該基板處理。 一圍第1G項之在-處理腔室中處理知 其中該分析係'藉由-運”ιίΓ基板的期間遍 ，引冬為崎錢理Α量龍之—高速處輪，其中制 12.如申請專利範圍第 定一終點之方法.，其令該最佳终點令處理基細期間鑑 器接收感測器資料，其中^ 5 的期間’從數個感測 來的數個信號流; 貝科包括從數個_器管道而 概略=Γ終闕，射祕_為_會發生_轉點的一 轉化該、ij能生;:可能終點特徵記號； W入节細异一寸试5己竦成為—組最佳終里占、、畲嘗 5亥'、且取佳終點演算法的一最佳终點法、二肩异法；以及 一 /ίΓ主生產環境宁。 24 201112302 13. 定一終; 來 -t:二=範在-處理腔室中處理基板的期間鑑 。〜點之方法，其中該感測器資料係從多於一個的基板蒐集而 n申請專利範圍第13項之在一處理腔室中處理基板的期 料執其中Ϊ感測器資料之該分析包括對該感測器i 區間分成數個片段 。 _ ' 日：， 賴_喻的期間鏗 可能=特徵記號之第一組 :其中針對該數罐之各片段計Ϊί斜;與== 對比丄;Iί ίIt二:ίΓ數個信號流中辨識出-組高 社人Β，、甲这同對比“唬具有高斜率變異， 的連續綠成為—組信號波長頻帶， 組信 ，該組信號波長頻帶分等，以及 號波用辨於至少部分的該高對比信號與該 徵包括波峰、波谷其愉分類特 定'上it弟中理腔室中處理基板的期間鑑 可能第該感測器資^ST記號之第二組 及該“==率„雜訊值所縮放之斜率以 化的斜率與一组標準化的相應斜分析’以產生-組標準 2〇Π 12302 怎ίίϊϊ準i匕^ίί，率變異，以從該數個信號流中辨 其巾該標準化錢具有高斜率與低變里， —具相靖率«的連續《縣—組_^號波長頻 把該標準化信號分等， 把该組標準化信號波長頻帶分等，以及 可能終點特徵記號 標準對比信號與—組信號波長頻帶對該 凝Μ化㈣波長鮮之比率，以產生該第二組 17 圍其第中= 在，室中處理基极__ 終點演算法包括 μ、、且可此終點特被§己號成為該組最佳 可能終點“ 虎之第-可能終點特徵記號與第二 非重複性终^ 以從該組可能終:特徵記 號中移除可能的 組最關性分析以辨識該組可能終點特徵記號的一 轉化該組最佳終點特徵纪轳,a s ^, 終點=t，其中該即時延即時延遲之即時 算法，以及. 貝足門松的相應即時延遲之即時終點演 即時終點 二;=;點演算法無法通過-穩健性測試 ’刪除該 把該組最佳終點演算法 +斤 等係基於保真率與師法分等’其中該分 26 201112302 ' 18·如中請專利範圍第16項之在—處理 其中轉化該組可能終點特徵 可能終點特徵以具與第二 終點特徵記號與該第二可能終點特徵記號，。忒弟一可能 非重觸組可能終點特徵記號中移除可能的 上=關性分析以辨識該組可 終，3，=該即即時延遲之即時 算法，以及 、予1疋門㈣相應即時延遲之即時終點演 即時終點點演算法無法通過—穩健性測試，刪除該 等係=;====法分等，該分 、、且使用者疋義條件之其中至少一者。 Λ係基於分等與 20.如申請專利範圍 定—終點之方法，其中該板的期間鐘 八、圖式： 272〇1112302 VII. Patent application scope: -2 monument: ===, in, the summary end point field, where the end point field is the lifetime end of the process that is expected to occur at the end of the process; the bow enters the group's second Qiao: the best end point algorithm for the group; and, ..., I show the same method - the best end point method to the production environment. 2. As in the patent application, the b method of the first item, wherein the sensor data is from more than In the method of substrate algorithm, where the identification is: the best end point algorithm, the segment will be from the number of signal streams; the method '^=--the best end point algorithm is 1:^ automatic Identifying the square sign of the best endpoint algorithm, the sensor ί S3: the first set of possible endpoint characteristics of the token is for the sensor data, 呌瞀t:: where the number of fragments (3) is different The slope of the slope is mutated to identify a set of high 21 201112302 comparisons from the plurality of signal streams, wherein the set of high contrast signals has a high slope variation, and the Lianyi* becomes a group signal wavelength band' letter, the group Signal wavelength band grading, and number wave (four) rectification With the group 6· method, the automatic identification of H is the best end point algorithm. The sensor is 2::::sign (4) and 4==;=: the slope of the rate value is reduced. ^-Group normalized corresponding slope ' 以 to produce a group standard 有 ΐ 辨 辨 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 And == normalized signal wave range 5th private automatic identification -..._-~ 铋. The number becomes the best end point algorithm package of the group. If the group is the end point feature, the end point feature symbol has similar b: possible end The sign of the transfer and the second, winter point feature mark and the second possible ^ period, combined with the first-possible f, wherein the secret end point discrimination, the knowledge - the best end point algorithm is included. 22 201112302 The group may end the feature mark to remove the possible group most critically to identify the group of possible end point feature tokens - convert the group's best end point feature token to an end point ΐίΐ:?, the time delay is based on the transition: late ~ The immediate algorithm of delay is at least - the best end-of-life performance method is generated by ^ greater than -_ _ eye should be, 'the end point of the performance of the end point performance S is = point '', the law can not pass - the robustness test, delete the department points, etc., (four) points 8 · such as the scope of patent application, which is the conversion of the _ secret In the method of the smear method, it is possible that the end point feature mark has a similar 匕 终点 end point feature mark and the second end point may end the period, combined with the first-possible non-repetitive end: the special township and possibly the end point feature mark may be removed The group's most critical analysis to identify the end point group of the end point feature of the object; the instant with the minimum immediate delay; the late delay based on the delay, the #divide and the best endpoint algorithm, and __The corresponding instant delay of the instantaneous end point performance of the immediate end point performance ^ algorithm can not pass - robustness 删除, delete read 201112302 箄 ^ 点 演 演 演 演 演 演 盆 盆 盆 盆 盆 基于 基于 基于 基于 基于 基于 基于 基于At least one of them. 4, wherein the reading points 9 · law conditions at least as claimed in the scope of the patent i, ί 2 best _ algorithm of ... .... Group user definition 10. A method comprising several endpoints, one performing a recipe on a substrate during processing of the substrate in the processing chamber; a set of sensors receiving processing data; an identification-process endpoint; and # going Knife out the trapping material; stop the substrate processing. A section of the 1G item is processed in the processing chamber, where the analysis is performed by the "transport" of the ιίΓ substrate, and the winter is the Sakizaki Α Α 龙 — 高速 高速 高速 高速 , , , , , , , , , , , , , The method of claiming a first end point of the patent scope, which causes the optimal end point to process the sensor to receive the sensor data, wherein the period of ^ 5 'several signal streams from several senses; Section includes a number of _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ - the best end of the group, the taste of 5 Hai', and the best end point method of the best end point algorithm, the two shoulders of the same method; and one / Γ Γ main production environment 宁. 24 201112302 13. ; -t: two = in the process chamber processing the substrate period. ~ point method, wherein the sensor data is collected from more than one substrate and n patent application scope 13 The analysis of the processing of the substrate in the processing chamber, wherein the analysis of the sensor data comprises dividing the sensor i interval into Fragment. _ ' Day:, Lai _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Identifying - group Gaoshe people, A, this comparison with "唬 has a high slope variation, continuous green becomes - group signal wavelength band, group letter, the group signal wavelength band is equal, and the number wave is discriminated at least in part The high-contrast signal and the sign include a peak, a trough, and a classification of the particular 'in the middle of the processing chamber in the middle chamber. The second group of the sensor's ^ST mark and the "== rate „The slope of the noise value is scaled to the slope of the set and the corresponding set of normalized oblique analysis' to generate a set of standard 2〇Π 12302 how to 匕I匕^ίί, rate variation, to distinguish from the number of signal streams The standardized money of the towel has a high slope and a low variation, and the continuous "County-Group _^ wavelength frequency of the gradual rate «equalizes the normalized signal, classifies the set of standardized signal wavelength bands, and possibly the end point. Feature mark standard comparison signal and group letter The ratio of the wavelength band to the condensed (four) wavelength is fresh to produce the second group 17 around the middle = in, the processing base in the chamber __ the end point algorithm includes μ, and the end point is specifically § The number becomes the best possible end point for the group "The Tiger's No. - Possible Endpoint Feature Sign and the Second Non-Repetitive End ^ to remove the possible group most criticality analysis from the set of possible end points: to identify the possible end point of the group A conversion of the feature mark to the best end point feature of the group, as ^, end point = t, wherein the immediate delay of the immediate delay algorithm, and the corresponding instantaneous delay of the corresponding instant delay of the foot of the foot of the foot end of the instant end two; =; The dot algorithm cannot pass the - robustness test 'delete the best end point algorithm of the group + jin, etc. based on the fidelity rate and the division of the teacher's method, which is 26 201112302 ' 18 · In the process of processing, the set of possible end point features may be end point features with a second end point feature mark and the second possible end point feature mark. A possible non-repeating group may remove the possible up=off-sense analysis to identify the group end-of-end feature, 3, = the immediate algorithm of immediate delay, and the 1st (4) corresponding immediate delay The immediate end point performance instant end point algorithm cannot pass the - robustness test, deletes the system =; ==== method points, etc., and at least one of the user and the user's ambiguous conditions. The system is based on the method of grading and 20. For example, the scope of the patent application, the period of the board, the pattern: 27
TW099121511A 2009-06-30 2010-06-30 Construction method of optimal endpoint algorithm TWI480917B (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US22210209P 2009-06-30 2009-06-30
US22202409P 2009-06-30 2009-06-30

Publications (2)

Publication Number Publication Date
TW201112302A true TW201112302A (en) 2011-04-01
TWI480917B TWI480917B (en) 2015-04-11

Family

ID=43411705

Family Applications (5)

Application Number Title Priority Date Filing Date
TW099121515A TWI484435B (en) 2009-06-30 2010-06-30 Method and apparatus for verifying etch rate uniformity to verify plasma chamber
TW099121516A TWI509375B (en) 2009-06-30 2010-06-30 Method and apparatus for implementing an automated in-line process control scheme during recipe execution
TW099121519A TWI536193B (en) 2009-06-30 2010-06-30 Method and apparatus for predictive preventive maintenance of processing chamber
TW099121511A TWI480917B (en) 2009-06-30 2010-06-30 Construction method of optimal endpoint algorithm
TW099121513A TWI495970B (en) 2009-06-30 2010-06-30 Method and device for detecting in-situ fast transient event

Family Applications Before (3)

Application Number Title Priority Date Filing Date
TW099121515A TWI484435B (en) 2009-06-30 2010-06-30 Method and apparatus for verifying etch rate uniformity to verify plasma chamber
TW099121516A TWI509375B (en) 2009-06-30 2010-06-30 Method and apparatus for implementing an automated in-line process control scheme during recipe execution
TW099121519A TWI536193B (en) 2009-06-30 2010-06-30 Method and apparatus for predictive preventive maintenance of processing chamber

Family Applications After (1)

Application Number Title Priority Date Filing Date
TW099121513A TWI495970B (en) 2009-06-30 2010-06-30 Method and device for detecting in-situ fast transient event

Country Status (6)

Country Link
JP (5) JP5629770B2 (en)
KR (5) KR101741274B1 (en)
CN (5) CN102473590B (en)
SG (5) SG176564A1 (en)
TW (5) TWI484435B (en)
WO (5) WO2011002811A2 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI485758B (en) * 2011-09-23 2015-05-21

Families Citing this family (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10128090B2 (en) 2012-02-22 2018-11-13 Lam Research Corporation RF impedance model based fault detection
US9502221B2 (en) * 2013-07-26 2016-11-22 Lam Research Corporation Etch rate modeling and use thereof with multiple parameters for in-chamber and chamber-to-chamber matching
CN104730372B (en) * 2013-12-13 2018-08-10 朗姆研究公司 Fault detect based on RF impedance models
US10192763B2 (en) * 2015-10-05 2019-01-29 Applied Materials, Inc. Methodology for chamber performance matching for semiconductor equipment
US10269545B2 (en) * 2016-08-03 2019-04-23 Lam Research Corporation Methods for monitoring plasma processing systems for advanced process and tool control
US9972478B2 (en) * 2016-09-16 2018-05-15 Lam Research Corporation Method and process of implementing machine learning in complex multivariate wafer processing equipment
US11067515B2 (en) * 2017-11-28 2021-07-20 Taiwan Semiconductor Manufacturing Co., Ltd. Apparatus and method for inspecting a wafer process chamber
CN108847381A (en) * 2018-05-25 2018-11-20 深圳市华星光电半导体显示技术有限公司 The method for testing substrate and extended testing system substrate service life
US10651097B2 (en) * 2018-08-30 2020-05-12 Lam Research Corporation Using identifiers to map edge ring part numbers onto slot numbers
US20200266037A1 (en) * 2019-02-14 2020-08-20 Advanced Energy Industries, Inc. Maintenance for remote plasma sources
DE102019209110A1 (en) * 2019-06-24 2020-12-24 Sms Group Gmbh Industrial plant, in particular plant in the metal-producing industry or the aluminum or steel industry, and method for operating an industrial plant, in particular a plant in the metal-producing industry or the aluminum or steel industry
GB202010471D0 (en) * 2020-07-08 2020-08-19 Univ Exeter Control of processing equipment
KR102882466B1 (en) * 2020-10-15 2025-11-05 에이에스엠 아이피 홀딩 비.브이. Predictive maintenance method, and predictive maintenance device
JP7289992B1 (en) * 2021-07-13 2023-06-12 株式会社日立ハイテク Diagnostic apparatus and diagnostic method, plasma processing apparatus and semiconductor device manufacturing system
US12147212B2 (en) * 2021-12-21 2024-11-19 Applied Materials, Inc. Diagnostic methods for substrate manufacturing chambers using physics-based models
WO2023148967A1 (en) * 2022-02-07 2023-08-10 株式会社日立ハイテク Diagnostic device, diagnostic method, semiconductor manufacturing device system, and semiconductor device manufacturing system
US12191126B2 (en) 2022-02-15 2025-01-07 Applied Materials, Inc. Process control knob estimation
US20240021450A1 (en) * 2022-07-15 2024-01-18 Verity Instruments, Inc. Control for semiconductor processing systems
CN119013763A (en) * 2023-03-17 2024-11-22 株式会社日立高新技术 State prediction device, state prediction method and state prediction system

Family Cites Families (59)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5272872A (en) * 1992-11-25 1993-12-28 Ford Motor Company Method and apparatus of on-board catalytic converter efficiency monitoring
JP3301238B2 (en) * 1994-10-25 2002-07-15 三菱電機株式会社 Etching method
JPH08148474A (en) * 1994-11-16 1996-06-07 Sony Corp Dry etching end point detection method and apparatus
JPH09306894A (en) * 1996-05-17 1997-11-28 Sony Corp Optimal emission spectrum automatic detection system
US6197116B1 (en) * 1996-08-29 2001-03-06 Fujitsu Limited Plasma processing system
JP3630931B2 (en) * 1996-08-29 2005-03-23 富士通株式会社 Plasma processing apparatus, process monitoring method, and semiconductor device manufacturing method
US5993615A (en) * 1997-06-19 1999-11-30 International Business Machines Corporation Method and apparatus for detecting arcs
WO1999014394A1 (en) * 1997-09-17 1999-03-25 Tokyo Electron Limited Device and method for detecting and preventing arcing in rf plasma systems
US5986747A (en) 1998-09-24 1999-11-16 Applied Materials, Inc. Apparatus and method for endpoint detection in non-ionizing gaseous reactor environments
US8617351B2 (en) * 2002-07-09 2013-12-31 Applied Materials, Inc. Plasma reactor with minimal D.C. coils for cusp, solenoid and mirror fields for plasma uniformity and device damage reduction
JP2001338856A (en) * 2000-05-30 2001-12-07 Tokyo Seimitsu Co Ltd Process controller for semiconductor manufacturing system
JP4554037B2 (en) * 2000-07-04 2010-09-29 東京エレクトロン株式会社 Consumable consumption level prediction method and deposited film thickness prediction method
US6567718B1 (en) * 2000-07-28 2003-05-20 Advanced Micro Devices, Inc. Method and apparatus for monitoring consumable performance
US6391787B1 (en) * 2000-10-13 2002-05-21 Lam Research Corporation Stepped upper electrode for plasma processing uniformity
US6821794B2 (en) 2001-10-04 2004-11-23 Novellus Systems, Inc. Flexible snapshot in endpoint detection
JP2003151955A (en) * 2001-11-19 2003-05-23 Nec Kansai Ltd Plasma etching method
WO2003102724A2 (en) * 2002-05-29 2003-12-11 Tokyo Electron Limited Method and system for data handling, storage and manipulation
US6825050B2 (en) * 2002-06-07 2004-11-30 Lam Research Corporation Integrated stepwise statistical process control in a plasma processing system
US20040031052A1 (en) * 2002-08-12 2004-02-12 Liberate Technologies Information platform
US6781383B2 (en) * 2002-09-24 2004-08-24 Scientific System Research Limited Method for fault detection in a plasma process
CN100407215C (en) * 2002-09-30 2008-07-30 东京毅力科创株式会社 Method and apparatus for monitoring and controlling semiconductor production process
BRPI0314881B1 (en) * 2002-10-25 2019-01-08 S&C Electric Company system and method for controlling distribution of electricity through a grid
JP4365109B2 (en) * 2003-01-29 2009-11-18 株式会社日立ハイテクノロジーズ Plasma processing equipment
US6969619B1 (en) * 2003-02-18 2005-11-29 Novellus Systems, Inc. Full spectrum endpoint detection
JP2004295348A (en) * 2003-03-26 2004-10-21 Mori Seiki Co Ltd Machine tool maintenance management system
US20060006139A1 (en) * 2003-05-09 2006-01-12 David Johnson Selection of wavelengths for end point in a time division multiplexed process
US7101805B2 (en) * 2003-05-09 2006-09-05 Unaxis Usa Inc. Envelope follower end point detection in time division multiplexed processes
JP2004335841A (en) * 2003-05-09 2004-11-25 Tokyo Electron Ltd Prediction apparatus and prediction method for plasma processing apparatus
US7062411B2 (en) * 2003-06-11 2006-06-13 Scientific Systems Research Limited Method for process control of semiconductor manufacturing equipment
JP4043408B2 (en) * 2003-06-16 2008-02-06 東京エレクトロン株式会社 Substrate processing apparatus and substrate processing method
US6902646B2 (en) * 2003-08-14 2005-06-07 Advanced Energy Industries, Inc. Sensor array for measuring plasma characteristics in plasma processing environments
KR100567745B1 (en) * 2003-09-25 2006-04-05 동부아남반도체 주식회사 Life Prediction Device and Life Prediction Method of Sputtering Target
US8036869B2 (en) * 2003-09-30 2011-10-11 Tokyo Electron Limited System and method for using first-principles simulation to control a semiconductor manufacturing process via a simulation result or a derived empirical model
US7930053B2 (en) * 2003-12-23 2011-04-19 Beacons Pharmaceuticals Pte Ltd Virtual platform to facilitate automated production
US7233878B2 (en) * 2004-01-30 2007-06-19 Tokyo Electron Limited Method and system for monitoring component consumption
US7146237B2 (en) * 2004-04-07 2006-12-05 Mks Instruments, Inc. Controller and method to mediate data collection from smart sensors for fab applications
JP2006004992A (en) * 2004-06-15 2006-01-05 Seiko Epson Corp Polishing apparatus management system, management apparatus, management apparatus control program, and management apparatus control method
TWI336823B (en) * 2004-07-10 2011-02-01 Onwafer Technologies Inc Methods of and apparatuses for maintenance, diagnosis, and optimization of processes
US7292045B2 (en) * 2004-09-04 2007-11-06 Applied Materials, Inc. Detection and suppression of electrical arcing
JP4972277B2 (en) * 2004-11-10 2012-07-11 東京エレクトロン株式会社 Substrate processing apparatus recovery method, apparatus recovery program, and substrate processing apparatus
US7828929B2 (en) * 2004-12-30 2010-11-09 Research Electro-Optics, Inc. Methods and devices for monitoring and controlling thin film processing
JP4707421B2 (en) * 2005-03-14 2011-06-22 東京エレクトロン株式会社 Processing apparatus, consumable part management method for processing apparatus, processing system, and consumable part management method for processing system
JP2006328510A (en) * 2005-05-30 2006-12-07 Ulvac Japan Ltd Plasma treatment method and device
TWI338321B (en) * 2005-06-16 2011-03-01 Unaxis Usa Inc Process change detection through the use of evolutionary algorithms
US7409260B2 (en) * 2005-08-22 2008-08-05 Applied Materials, Inc. Substrate thickness measuring during polishing
US7302363B2 (en) * 2006-03-31 2007-11-27 Tokyo Electron Limited Monitoring a system during low-pressure processes
US7413672B1 (en) * 2006-04-04 2008-08-19 Lam Research Corporation Controlling plasma processing using parameters derived through the use of a planar ion flux probing arrangement
US7829468B2 (en) * 2006-06-07 2010-11-09 Lam Research Corporation Method and apparatus to detect fault conditions of plasma processing reactor
KR20080006750A (en) * 2006-07-13 2008-01-17 삼성전자주식회사 Plasma Doping System for Semiconductor Device Manufacturing
US20080063810A1 (en) * 2006-08-23 2008-03-13 Applied Materials, Inc. In-situ process state monitoring of chamber
CN100587902C (en) * 2006-09-15 2010-02-03 北京北方微电子基地设备工艺研究中心有限责任公司 A method for online prediction of etching equipment maintenance
JP2008158769A (en) * 2006-12-22 2008-07-10 Tokyo Electron Ltd Substrate processing system, control device, setting information monitoring method, and storage medium storing setting information monitoring program
US7548830B2 (en) * 2007-02-23 2009-06-16 General Electric Company System and method for equipment remaining life estimation
US7674636B2 (en) * 2007-03-12 2010-03-09 Tokyo Electron Limited Dynamic temperature backside gas control for improved within-substrate process uniformity
US8055203B2 (en) * 2007-03-14 2011-11-08 Mks Instruments, Inc. Multipoint voltage and current probe system
JP2008311338A (en) * 2007-06-13 2008-12-25 Harada Sangyo Kk Vacuum treatment apparatus and abnormal discharge precognition device used therefor, and control method of vacuum treatment apparatus
KR100892248B1 (en) * 2007-07-24 2009-04-09 주식회사 디엠에스 Endpoint detection device for real time control of plasma reactor, plasma reactor including the same, and endpoint detection method thereof
US20090106290A1 (en) * 2007-10-17 2009-04-23 Rivard James P Method of analyzing manufacturing process data
JP4983575B2 (en) * 2007-11-30 2012-07-25 パナソニック株式会社 Plasma processing apparatus and plasma processing method

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI485758B (en) * 2011-09-23 2015-05-21

Also Published As

Publication number Publication date
WO2011002800A3 (en) 2011-04-07
KR101708077B1 (en) 2017-02-17
CN102474968B (en) 2015-09-02
WO2011002803A3 (en) 2011-03-03
WO2011002804A2 (en) 2011-01-06
JP2012532462A (en) 2012-12-13
KR20120037420A (en) 2012-04-19
JP5624618B2 (en) 2014-11-12
WO2011002811A2 (en) 2011-01-06
WO2011002811A3 (en) 2011-02-24
KR20120101293A (en) 2012-09-13
CN102804353A (en) 2012-11-28
JP2012532463A (en) 2012-12-13
JP2012532464A (en) 2012-12-13
TWI495970B (en) 2015-08-11
KR101741271B1 (en) 2017-05-29
WO2011002810A2 (en) 2011-01-06
WO2011002810A4 (en) 2011-06-03
KR20120037421A (en) 2012-04-19
SG176147A1 (en) 2011-12-29
CN102473631A (en) 2012-05-23
KR20120037419A (en) 2012-04-19
CN102473590B (en) 2014-11-26
CN102804353B (en) 2015-04-15
TWI509375B (en) 2015-11-21
TW201129936A (en) 2011-09-01
TWI536193B (en) 2016-06-01
CN102804929A (en) 2012-11-28
JP5629770B2 (en) 2014-11-26
KR101708078B1 (en) 2017-02-17
TWI484435B (en) 2015-05-11
CN102474968A (en) 2012-05-23
JP5599882B2 (en) 2014-10-01
TW201108022A (en) 2011-03-01
JP2012532460A (en) 2012-12-13
WO2011002803A2 (en) 2011-01-06
WO2011002800A2 (en) 2011-01-06
CN102473590A (en) 2012-05-23
JP2012532461A (en) 2012-12-13
SG176565A1 (en) 2012-01-30
TWI480917B (en) 2015-04-11
WO2011002810A3 (en) 2011-04-14
KR20120047871A (en) 2012-05-14
SG176564A1 (en) 2012-01-30
CN102804929B (en) 2015-11-25
JP5693573B2 (en) 2015-04-01
KR101741272B1 (en) 2017-05-29
SG176566A1 (en) 2012-01-30
TW201129884A (en) 2011-09-01
SG176567A1 (en) 2012-01-30
CN102473631B (en) 2014-11-26
KR101741274B1 (en) 2017-05-29
WO2011002804A3 (en) 2011-03-03
TW201115288A (en) 2011-05-01

Similar Documents

Publication Publication Date Title
TW201112302A (en) Methods for constructing an optimal endpoint algorithm
CN107527337B (en) A kind of the video object removal altering detecting method based on deep learning
CN110428876B (en) Steel material design method based on machine learning algorithm of physical guidance
CN108416775B (en) An ore particle size detection method based on deep learning
CN103456013B (en) A kind of method representing similarity between super-pixel and tolerance super-pixel
CN107133628A (en) A kind of method and device for setting up data identification model
CN101197131A (en) Random voiceprint password verification system, random voiceprint password lock and generation method thereof
CN103150498A (en) Hardware trojan identification method based on single classification supporting vector machine
CN107328868A (en) A kind of Analysis of Acoustic Emission Signal method of quick identification ceramic coating failure type
CN120410558B (en) Machine Learning-Based Method for Detecting Forming Characteristics and Assessing Quality of Titanium Alloy Forgings
CN102194106B (en) Human face recognition method used in door access system
CN119207581B (en) Method for predicting 2OM locus in RNA based on ensemble learning
CN104915626A (en) Face identification method and apparatus
CN106295478A (en) A kind of image characteristic extracting method and device
TW200928843A (en) Virtual metrology system applied on chemical vapor deposition process
CN106980774A (en) A kind of extended method of DNA methylation chip data
CN106548212A (en) A kind of secondary weighted KNN musical genre classification methods
CN110927478B (en) Method and system for determining state of transformer equipment of power system
Ushmaev et al. Biometric fusion: Robust approach
CN112734722A (en) Flame endpoint carbon content prediction method based on improved complete local binary pattern
CN111563553A (en) Method for predicting alkalinity of pellet ore phase based on PCA-CNN classification recognition
CN101609502B (en) Human face detecting method based on sequence simplifying support vector
JP2004046734A (en) Web system performance prediction device, web system performance prediction method, and program
CN109767430A (en) The quality determining method and quality detecting system of valuable bills
CN120410564B (en) A knowledge graph-based intelligent method for distinguishing genuine from fake Southern Red Agate